From 8308c54d7e312f7a03e2ce2057d0837e6fe3843f Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Thu, 11 Sep 2008 01:31:50 -0700 Subject: generic: redefine resource_size_t as phys_addr_t There's no good reason why a resource_size_t shouldn't just be a physical address, so simply redefine it in terms of phys_addr_t. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/e820.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 66e48aa2dd1b..477f4bb7e552 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -1276,12 +1276,10 @@ void __init e820_reserve_resources(void) res = alloc_bootmem_low(sizeof(struct resource) * e820.nr_map); for (i = 0; i < e820.nr_map; i++) { end = e820.map[i].addr + e820.map[i].size - 1; -#ifndef CONFIG_RESOURCES_64BIT - if (end > 0x100000000ULL) { + if (end != (resource_size_t)end) { res++; continue; } -#endif res->name = e820_type_to_string(e820.map[i].type); res->start = e820.map[i].addr; res->end = end; -- cgit v1.2.3 From 2fd47094f92fa2bdbf99be33294a7b6b97785a70 Mon Sep 17 00:00:00 2001 From: Thomas Renninger Date: Mon, 1 Sep 2008 14:27:03 +0200 Subject: CPUFREQ: powernow-k8: Try to detect old BIOS, not supporting CPU freq on a recent AMD CPUs. Make use of FW_BUG interface to give vendors and users the ability to automatically check for powernow-k8 related BIOS bugs by: dmesg |grep "Firmware Bug" Signed-off-by: Thomas Renninger Signed-off-by: Andi Kleen Signed-off-by: Len Brown --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 42 ++++++++++++++++++------------- 1 file changed, 24 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 84bb395038d8..4e0c6abd7ca4 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -45,7 +45,6 @@ #endif #define PFX "powernow-k8: " -#define BFX PFX "BIOS error: " #define VERSION "version 2.20.00" #include "powernow-k8.h" @@ -536,35 +535,40 @@ static int check_pst_table(struct powernow_k8_data *data, struct pst_s *pst, u8 for (j = 0; j < data->numps; j++) { if (pst[j].vid > LEAST_VID) { - printk(KERN_ERR PFX "vid %d invalid : 0x%x\n", j, pst[j].vid); + printk(KERN_ERR FW_BUG PFX "vid %d invalid : 0x%x\n", + j, pst[j].vid); return -EINVAL; } if (pst[j].vid < data->rvo) { /* vid + rvo >= 0 */ - printk(KERN_ERR BFX "0 vid exceeded with pstate %d\n", j); + printk(KERN_ERR FW_BUG PFX "0 vid exceeded with pstate" + " %d\n", j); return -ENODEV; } if (pst[j].vid < maxvid + data->rvo) { /* vid + rvo >= maxvid */ - printk(KERN_ERR BFX "maxvid exceeded with pstate %d\n", j); + printk(KERN_ERR FW_BUG PFX "maxvid exceeded with pstate" + " %d\n", j); return -ENODEV; } if (pst[j].fid > MAX_FID) { - printk(KERN_ERR BFX "maxfid exceeded with pstate %d\n", j); + printk(KERN_ERR FW_BUG PFX "maxfid exceeded with pstate" + " %d\n", j); return -ENODEV; } if (j && (pst[j].fid < HI_FID_TABLE_BOTTOM)) { /* Only first fid is allowed to be in "low" range */ - printk(KERN_ERR BFX "two low fids - %d : 0x%x\n", j, pst[j].fid); + printk(KERN_ERR FW_BUG PFX "two low fids - %d : " + "0x%x\n", j, pst[j].fid); return -EINVAL; } if (pst[j].fid < lastfid) lastfid = pst[j].fid; } if (lastfid & 1) { - printk(KERN_ERR BFX "lastfid invalid\n"); + printk(KERN_ERR FW_BUG PFX "lastfid invalid\n"); return -EINVAL; } if (lastfid > LO_FID_TABLE_TOP) - printk(KERN_INFO BFX "first fid not from lo freq table\n"); + printk(KERN_INFO FW_BUG PFX "first fid not from lo freq table\n"); return 0; } @@ -672,13 +676,13 @@ static int find_psb_table(struct powernow_k8_data *data) dprintk("table vers: 0x%x\n", psb->tableversion); if (psb->tableversion != PSB_VERSION_1_4) { - printk(KERN_ERR BFX "PSB table is not v1.4\n"); + printk(KERN_ERR FW_BUG PFX "PSB table is not v1.4\n"); return -ENODEV; } dprintk("flags: 0x%x\n", psb->flags1); if (psb->flags1) { - printk(KERN_ERR BFX "unknown flags\n"); + printk(KERN_ERR FW_BUG PFX "unknown flags\n"); return -ENODEV; } @@ -705,7 +709,7 @@ static int find_psb_table(struct powernow_k8_data *data) } } if (cpst != 1) { - printk(KERN_ERR BFX "numpst must be 1\n"); + printk(KERN_ERR FW_BUG PFX "numpst must be 1\n"); return -ENODEV; } @@ -1130,17 +1134,19 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) "ACPI Processor module before starting this " "driver.\n"); #else - printk(KERN_ERR PFX "Your BIOS does not provide ACPI " - "_PSS objects in a way that Linux understands. " - "Please report this to the Linux ACPI maintainers" - " and complain to your BIOS vendor.\n"); + printk(KERN_ERR FW_BUG PFX "Your BIOS does not provide" + " ACPI _PSS objects in a way that Linux " + "understands. Please report this to the Linux " + "ACPI maintainers and complain to your BIOS " + "vendor.\n"); #endif kfree(data); return -ENODEV; } if (pol->cpu != 0) { - printk(KERN_ERR PFX "No ACPI _PSS objects for CPU other than " - "CPU0. Complain to your BIOS vendor.\n"); + printk(KERN_ERR FW_BUG PFX "No ACPI _PSS objects for " + "CPU other than CPU0. Complain to your BIOS " + "vendor.\n"); kfree(data); return -ENODEV; } @@ -1193,7 +1199,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) /* min/max the cpu is capable of */ if (cpufreq_frequency_table_cpuinfo(pol, data->powernow_table)) { - printk(KERN_ERR PFX "invalid powernow_table\n"); + printk(KERN_ERR FW_BUG PFX "invalid powernow_table\n"); powernow_k8_cpu_exit_acpi(data); kfree(data->powernow_table); kfree(data); -- cgit v1.2.3 From d0d0f7432c9cbd52cb2f31d499f8292b13a7ecac Mon Sep 17 00:00:00 2001 From: Matt Mackall Date: Thu, 9 Oct 2008 12:41:50 -0500 Subject: x86: remove magic number from ACPI sleep stack buffer x86_64 SMP suspend to RAM uses a 10k temporary stack for saving the kernel state, but only 4k of it is used. Shrink it to 4k. Signed-off-by: Matt Mackall Acked-by: Pavel Machek Signed-off-by: Len Brown --- arch/x86/kernel/acpi/sleep.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 426e5d91b63a..29cf3403abef 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -97,7 +97,7 @@ int acpi_save_state_mem(void) #else /* CONFIG_64BIT */ header->trampoline_segment = setup_trampoline() >> 4; #ifdef CONFIG_SMP - stack_start.sp = temp_stack + 4096; + stack_start.sp = temp_stack + sizeof(temp_stack); #endif initial_code = (unsigned long)wakeup_long64; saved_magic = 0x123456789abcdef0; -- cgit v1.2.3 From 5000cadcf3188e935dae28c4fc7e24639704ea55 Mon Sep 17 00:00:00 2001 From: Matt Mackall Date: Thu, 9 Oct 2008 11:56:21 -0500 Subject: x86: trim ACPI sleep stack buffer x86_64 SMP suspend to RAM uses a 10k temporary stack for saving the kernel state, but only 4k of it is used. Shrink it to 4k. Signed-off-by: Matt Mackall Signed-off-by: Len Brown --- arch/x86/kernel/acpi/sleep.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 29cf3403abef..55d10cbe65b1 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -21,7 +21,7 @@ unsigned long acpi_realmode_flags; static unsigned long acpi_realmode; #if defined(CONFIG_SMP) && defined(CONFIG_64BIT) -static char temp_stack[10240]; +static char temp_stack[4096]; #endif /** -- cgit v1.2.3 From ee297533279a802eac8b1cbea8e65b24b36a1aac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 24 Sep 2008 19:04:31 -0700 Subject: ACPI: don't load acpi_cpufreq if acpi=off Signed-off-by: Yinghai Lu Signed-off-by: Len Brown --- arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index dd097b835839..9943b4c87746 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -779,6 +779,9 @@ static int __init acpi_cpufreq_init(void) { int ret; + if (acpi_disabled) + return 0; + dprintk("acpi_cpufreq_init\n"); ret = acpi_cpufreq_early_init(); -- cgit v1.2.3 From 8b1fa1d7b22f386747c7b78b918d4c680c16066f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 29 Jul 2008 12:36:02 +0200 Subject: ftrace: mark lapic_wd_event() notrace it can be called in the NMI path: [ 0.645999] calling ftrace_dynamic_init+0x0/0xd6 [ 0.647521] ------------[ cut here ]------------ [ 0.647521] WARNING: at kernel/trace/ftrace.c:348 ftrace_record_ip+0x4e/0x252() [ 0.647521] Modules linked in: [ 0.647521] Pid: 15, comm: kstop1 Not tainted 2.6.27-rc1-tip #22686 [ 0.647521] [ 0.647521] Call Trace: [ 0.647521] [] warn_on_slowpath+0x5d/0x84 [ 0.647521] [] ? lapic_wd_event+0xb/0x5c [ 0.647521] [] ftrace_record_ip+0x4e/0x252 [ 0.647521] [] mcount_call+0x5/0x31 [ 0.647521] [] ? lapic_wd_event+0x10/0x5c [ 0.647521] [] nmi_watchdog_tick+0x19d/0x1ad [ 0.647521] [] default_do_nmi+0x75/0x1e3 [ 0.647521] [] do_nmi+0x5d/0x94 [ 0.647521] [] nmi+0xa2/0xc2 [ 0.647521] [] ? check_bytes_and_report+0x11/0xcc [ 0.647521] <> [] ? mcount_call+0x5/0x31 [ 0.647521] [] check_object+0x61/0x1b0 [ 0.647521] [] __slab_free+0x169/0x2ae [ 0.647521] [] ? __cleanup_sighand+0x25/0x27 [ 0.647521] [] ? __cleanup_sighand+0x25/0x27 [ 0.647521] [] kmem_cache_free+0x85/0xb9 [ 0.647521] [] __cleanup_sighand+0x25/0x27 [ 0.647521] [] release_task+0x256/0x339 [ 0.647521] [] do_exit+0x764/0x7ef [ 0.647521] [] __xchg+0x0/0x38 [ 0.647521] [] ? stop_cpu+0x0/0xb2 [ 0.647521] [] ? stop_cpu+0x0/0xb2 [ 0.647521] [] kthread+0x4e/0x7b [ 0.647521] [] child_rip+0xa/0x11 [ 0.647521] [] ? restore_args+0x0/0x30 [ 0.647521] [] ? native_load_tls+0x14/0x2e [ 0.647521] [] ? kthread+0x0/0x7b [ 0.647521] [] ? child_rip+0x0/0x11 [ 0.647521] [ 0.647521] ---[ end trace 4eaa2a86a8e2da22 ]--- [ 0.672032] initcall ftrace_dynamic_init+0x0/0xd6 returned 0 after 19 msecs also mark it no-kprobes while at it. Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perfctr-watchdog.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index 6bff382094f5..9abd48b22674 100644 --- a/arch/x86/kernel/cpu/perfctr-watchdog.c +++ b/arch/x86/kernel/cpu/perfctr-watchdog.c @@ -17,6 +17,8 @@ #include #include #include +#include + #include #include @@ -336,7 +338,8 @@ static void single_msr_unreserve(void) release_perfctr_nmi(wd_ops->perfctr); } -static void single_msr_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes +single_msr_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { /* start the cycle over again */ write_watchdog_counter(wd->perfctr_msr, NULL, nmi_hz); @@ -401,7 +404,7 @@ static int setup_p6_watchdog(unsigned nmi_hz) return 1; } -static void p6_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes p6_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { /* * P6 based Pentium M need to re-unmask @@ -605,7 +608,7 @@ static void p4_unreserve(void) release_perfctr_nmi(MSR_P4_IQ_PERFCTR0); } -static void p4_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes p4_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { unsigned dummy; /* @@ -784,7 +787,7 @@ unsigned lapic_adjust_nmi_hz(unsigned hz) return hz; } -int lapic_wd_event(unsigned nmi_hz) +int __kprobes lapic_wd_event(unsigned nmi_hz) { struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk); u64 ctr; -- cgit v1.2.3 From 0a37605c2261a06d8cafc62dee11374ad676c8c4 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Aug 2008 15:45:12 -0400 Subject: ftrace: x86 mcount stub x86 now sets up the mcount locations through the build and no longer needs to record the ip when the function is executed. This patch changes the initial mcount to simply return. There's no need to do any other work. If the ftrace start up test fails, the original mcount will be what everything will use, so having this as fast as possible is a good thing. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 14 -------------- arch/x86/kernel/entry_64.S | 26 -------------------------- arch/x86/kernel/ftrace.c | 14 ++------------ 3 files changed, 2 insertions(+), 52 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index b21fbfaffe39..4e4269c73bb7 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1153,20 +1153,6 @@ ENDPROC(xen_failsafe_callback) #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) - pushl %eax - pushl %ecx - pushl %edx - movl 0xc(%esp), %eax - subl $MCOUNT_INSN_SIZE, %eax - -.globl mcount_call -mcount_call: - call ftrace_stub - - popl %edx - popl %ecx - popl %eax - ret END(mcount) diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 1db6ce4314e1..09e7145484c5 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -64,32 +64,6 @@ #ifdef CONFIG_FTRACE #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) - - subq $0x38, %rsp - movq %rax, (%rsp) - movq %rcx, 8(%rsp) - movq %rdx, 16(%rsp) - movq %rsi, 24(%rsp) - movq %rdi, 32(%rsp) - movq %r8, 40(%rsp) - movq %r9, 48(%rsp) - - movq 0x38(%rsp), %rdi - subq $MCOUNT_INSN_SIZE, %rdi - -.globl mcount_call -mcount_call: - call ftrace_stub - - movq 48(%rsp), %r9 - movq 40(%rsp), %r8 - movq 32(%rsp), %rdi - movq 24(%rsp), %rsi - movq 16(%rsp), %rdx - movq 8(%rsp), %rcx - movq (%rsp), %rax - addq $0x38, %rsp - retq END(mcount) diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index ab115cd15fdf..96aadbfedcc6 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -112,18 +112,8 @@ notrace int ftrace_update_ftrace_func(ftrace_func_t func) notrace int ftrace_mcount_set(unsigned long *data) { - unsigned long ip = (long)(&mcount_call); - unsigned long *addr = data; - unsigned char old[MCOUNT_INSN_SIZE], *new; - - /* - * Replace the mcount stub with a pointer to the - * ip recorder function. - */ - memcpy(old, &mcount_call, MCOUNT_INSN_SIZE); - new = ftrace_call_replace(ip, *addr); - *addr = ftrace_modify_code(ip, old, new); - + /* mcount is initialized as a nop */ + *data = 0; return 0; } -- cgit v1.2.3 From 732f3ca7d4ba3c1be8d051d52302ef441ee7748b Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Aug 2008 18:05:05 -0400 Subject: ftrace: use only 5 byte nops for x86 Mathieu Desnoyers revealed a bug in the original code. The nop that is used to relpace the mcount caller can be a two part nop. This runs the risk where a process can be preempted after executing the first nop, but before the second part of the nop. The ftrace code calls kstop_machine to keep multiple CPUs from executing code that is being modified, but it does not protect against a task preempting in the middle of a two part nop. If the above preemption happens and the tracer is enabled, after the kstop_machine runs, all those nops will be calls to the trace function. If the preempted process that was preempted between the two nops is executed again, it will execute half of the call to the trace function, and this might crash the system. This patch instead uses what both the latest Intel and AMD spec suggests. That is the P6_NOP5 sequence of "0x0f 0x1f 0x44 0x00 0x00". Note, some older CPUs and QEMU might fault on this nop, so this nop is executed with fault handling first. If it detects a fault, it will then use the code "0x66 0x66 0x66 0x66 0x90". If that faults, it will then default to a simple "jmp 1f; .byte 0x00 0x00 0x00; 1:". The jmp is not optimal but will do if the first two can not be executed. TODO: Examine the cpuid to determine the nop to use. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 68 +++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 61 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 96aadbfedcc6..4151c91254e8 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -16,8 +16,8 @@ #include #include -#include #include +#include /* Long is fine, even if it is only 4 bytes ;-) */ @@ -119,13 +119,67 @@ notrace int ftrace_mcount_set(unsigned long *data) int __init ftrace_dyn_arch_init(void *data) { - const unsigned char *const *noptable = find_nop_table(); - - /* This is running in kstop_machine */ - - ftrace_mcount_set(data); + extern const unsigned char ftrace_test_p6nop[]; + extern const unsigned char ftrace_test_nop5[]; + extern const unsigned char ftrace_test_jmp[]; + int faulted = 0; - ftrace_nop = (unsigned long *)noptable[MCOUNT_INSN_SIZE]; + /* + * There is no good nop for all x86 archs. + * We will default to using the P6_NOP5, but first we + * will test to make sure that the nop will actually + * work on this CPU. If it faults, we will then + * go to a lesser efficient 5 byte nop. If that fails + * we then just use a jmp as our nop. This isn't the most + * efficient nop, but we can not use a multi part nop + * since we would then risk being preempted in the middle + * of that nop, and if we enabled tracing then, it might + * cause a system crash. + * + * TODO: check the cpuid to determine the best nop. + */ + asm volatile ( + "jmp ftrace_test_jmp\n" + /* This code needs to stay around */ + ".section .text, \"ax\"\n" + "ftrace_test_jmp:" + "jmp ftrace_test_p6nop\n" + ".byte 0x00,0x00,0x00\n" /* 2 byte jmp + 3 bytes */ + "ftrace_test_p6nop:" + P6_NOP5 + "jmp 1f\n" + "ftrace_test_nop5:" + ".byte 0x66,0x66,0x66,0x66,0x90\n" + "jmp 1f\n" + ".previous\n" + "1:" + ".section .fixup, \"ax\"\n" + "2: movl $1, %0\n" + " jmp ftrace_test_nop5\n" + "3: movl $2, %0\n" + " jmp 1b\n" + ".previous\n" + _ASM_EXTABLE(ftrace_test_p6nop, 2b) + _ASM_EXTABLE(ftrace_test_nop5, 3b) + : "=r"(faulted) : "0" (faulted)); + + switch (faulted) { + case 0: + pr_info("ftrace: converting mcount calls to 0f 1f 44 00 00\n"); + ftrace_nop = (unsigned long *)ftrace_test_p6nop; + break; + case 1: + pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n"); + ftrace_nop = (unsigned long *)ftrace_test_nop5; + break; + case 2: + pr_info("ftrace: converting mcount calls to jmp 1f\n"); + ftrace_nop = (unsigned long *)ftrace_test_jmp; + break; + } + + /* The return code is retured via data */ + *(unsigned long *)data = 0; return 0; } -- cgit v1.2.3 From 6f93fc076a464bfe24e8d4c5fea3f6ca5bdb264d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 20 Aug 2008 12:55:07 -0400 Subject: ftrace: x86 use copy to and from user functions The modification of code is performed either by kstop_machine, before SMP starts, or on module code before the module is executed. There is no reason to do the modifications from assembly. The copy to and from user functions are sufficient and produces cleaner and easier to read code. Thanks to Benjamin Herrenschmidt for suggesting the idea. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 38 +++++++++++++------------------------- 1 file changed, 13 insertions(+), 25 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 4151c91254e8..082d99642622 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -11,6 +11,7 @@ #include #include +#include #include #include #include @@ -60,11 +61,7 @@ notrace int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { - unsigned replaced; - unsigned old = *(unsigned *)old_code; /* 4 bytes */ - unsigned new = *(unsigned *)new_code; /* 4 bytes */ - unsigned char newch = new_code[4]; - int faulted = 0; + unsigned char replaced[MCOUNT_INSN_SIZE]; /* * Note: Due to modules and __init, code can @@ -72,29 +69,20 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * as well as code changing. * * No real locking needed, this code is run through - * kstop_machine. + * kstop_machine, or before SMP starts. */ - asm volatile ( - "1: lock\n" - " cmpxchg %3, (%2)\n" - " jnz 2f\n" - " movb %b4, 4(%2)\n" - "2:\n" - ".section .fixup, \"ax\"\n" - "3: movl $1, %0\n" - " jmp 2b\n" - ".previous\n" - _ASM_EXTABLE(1b, 3b) - : "=r"(faulted), "=a"(replaced) - : "r"(ip), "r"(new), "c"(newch), - "0"(faulted), "a"(old) - : "memory"); - sync_core(); + if (__copy_from_user(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) + return 1; - if (replaced != old && replaced != new) - faulted = 2; + if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) + return 2; - return faulted; + WARN_ON_ONCE(__copy_to_user((char __user *)ip, new_code, + MCOUNT_INSN_SIZE)); + + sync_core(); + + return 0; } notrace int ftrace_update_ftrace_func(ftrace_func_t func) -- cgit v1.2.3 From 37a52f5ef120b93734bb2461744512b55695f69c Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Tue, 23 Sep 2008 15:05:46 -0700 Subject: x86: suppress trivial sparse signedness warnings Could just as easily change the three casts to cast to the correct type...this patch changes the type of ftrace_nop instead. Supresses sparse warnings: arch/x86/kernel/ftrace.c:157:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:157:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:157:14: got unsigned long * arch/x86/kernel/ftrace.c:161:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:161:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:161:14: got unsigned long * arch/x86/kernel/ftrace.c:165:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:165:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:165:14: got unsigned long * Signed-off-by: Harvey Harrison Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 082d99642622..66d900248fc2 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -22,7 +22,7 @@ /* Long is fine, even if it is only 4 bytes ;-) */ -static long *ftrace_nop; +static unsigned long *ftrace_nop; union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; -- cgit v1.2.3 From ac2b86fdef5b44f194eefaa6b7b6aea9423d1bc2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Fr=C3=A9d=C3=A9ric=20Weisbecker?= Date: Wed, 24 Sep 2008 16:31:56 +0100 Subject: x86/ftrace: use uaccess in atomic context With latest -tip I get this bug: [ 49.439988] in_atomic():0, irqs_disabled():1 [ 49.440118] INFO: lockdep is turned off. [ 49.440118] Pid: 2814, comm: modprobe Tainted: G W 2.6.27-rc7 #4 [ 49.440118] [] __might_sleep+0xe1/0x120 [ 49.440118] [] ftrace_modify_code+0x2a/0xd0 [ 49.440118] [] ? ftrace_test_p6nop+0x0/0xa [ 49.440118] [] __ftrace_update_code+0xfe/0x2f0 [ 49.440118] [] ? ftrace_test_p6nop+0x0/0xa [ 49.440118] [] ftrace_convert_nops+0x50/0x80 [ 49.440118] [] ftrace_init_module+0x16/0x20 [ 49.440118] [] load_module+0x185b/0x1d30 [ 49.440118] [] ? find_get_page+0x0/0xf0 [ 49.440118] [] ? sprintf+0x0/0x30 [ 49.440118] [] ? mutex_lock_interruptible_nested+0x1f2/0x350 [ 49.440118] [] sys_init_module+0x53/0x1b0 [ 49.440118] [] ? do_page_fault+0x0/0x740 [ 49.440118] [] syscall_call+0x7/0xb [ 49.440118] ======================= It is because ftrace_modify_code() calls copy_to_user and copy_from_user. These functions have been inserted after guessing that there couldn't be any race condition but copy_[to/from]_user might sleep and __ftrace_update_code is called with local_irq_saved. These function have been inserted since this commit: d5e92e8978fd2574e415dc2792c5eb592978243d: "ftrace: x86 use copy from user function" Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 66d900248fc2..222507e8157b 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -71,13 +71,13 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * No real locking needed, this code is run through * kstop_machine, or before SMP starts. */ - if (__copy_from_user(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) + if (__copy_from_user_inatomic(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) return 1; if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) return 2; - WARN_ON_ONCE(__copy_to_user((char __user *)ip, new_code, + WARN_ON_ONCE(__copy_to_user_inatomic((char __user *)ip, new_code, MCOUNT_INSN_SIZE)); sync_core(); -- cgit v1.2.3 From 8b27386a9ce9c7f0f8702cff7565a46802ad57d1 Mon Sep 17 00:00:00 2001 From: Anders Kaseorg Date: Thu, 9 Oct 2008 22:19:08 -0400 Subject: ftrace: make ftrace_test_p6nop disassembler-friendly Commit 4c3dc21b136f8cb4b72afee16c3ba7e961656c0b in tip introduced the 5-byte NOP ftrace_test_p6nop: jmp . + 5 .byte 0x00, 0x00, 0x00 This is not friendly to disassemblers because an odd number of 0x00s ends in the middle of an instruction boundary. This changes the 0x00s to 1-byte NOPs (0x90). Signed-off-by: Anders Kaseorg Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 222507e8157b..d073d981a730 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -132,7 +132,9 @@ int __init ftrace_dyn_arch_init(void *data) ".section .text, \"ax\"\n" "ftrace_test_jmp:" "jmp ftrace_test_p6nop\n" - ".byte 0x00,0x00,0x00\n" /* 2 byte jmp + 3 bytes */ + "nop\n" + "nop\n" + "nop\n" /* 2 byte jmp + 3 bytes */ "ftrace_test_p6nop:" P6_NOP5 "jmp 1f\n" @@ -161,7 +163,7 @@ int __init ftrace_dyn_arch_init(void *data) ftrace_nop = (unsigned long *)ftrace_test_nop5; break; case 2: - pr_info("ftrace: converting mcount calls to jmp 1f\n"); + pr_info("ftrace: converting mcount calls to jmp . + 5\n"); ftrace_nop = (unsigned long *)ftrace_test_jmp; break; } -- cgit v1.2.3 From 3807f345b2c610336c17c7624a0d496a38df75a0 Mon Sep 17 00:00:00 2001 From: Glauber Costa Date: Mon, 28 Jul 2008 11:47:52 -0300 Subject: x86: paravirt: factor out cpu_khz to common code KVM intends to use paravirt code to calibrate khz. Xen current code will do just fine. So as a first step, factor out code to pvclock.c. Signed-off-by: Glauber Costa Signed-off-by: Avi Kivity --- arch/x86/kernel/pvclock.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c index 05fbe9a0325a..1c54b5fb7aed 100644 --- a/arch/x86/kernel/pvclock.c +++ b/arch/x86/kernel/pvclock.c @@ -97,6 +97,18 @@ static unsigned pvclock_get_time_values(struct pvclock_shadow_time *dst, return dst->version; } +unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src) +{ + u64 tsc_khz = 1000000ULL << 32; + + do_div(tsc_khz, src->tsc_to_system_mul); + if (src->tsc_shift < 0) + tsc_khz <<= -src->tsc_shift; + else + tsc_khz >>= src->tsc_shift; + return tsc_khz; +} + cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src) { struct pvclock_shadow_time shadow; -- cgit v1.2.3 From 0293615f3fb9886b6b23800c121be293bb7483e9 Mon Sep 17 00:00:00 2001 From: Glauber Costa Date: Mon, 28 Jul 2008 11:47:53 -0300 Subject: x86: KVM guest: use paravirt function to calculate cpu khz We're currently facing timing problems in guests that do calibration under heavy load, and then the load vanishes. This means we'll have a much lower lpj than we actually should, and delays end up taking less time than they should, which is a nasty bug. Solution is to pass on the lpj value from host to guest, and have it preset. Signed-off-by: Glauber Costa Signed-off-by: Avi Kivity --- arch/x86/kernel/kvmclock.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c index d02def06ca91..774ac4991568 100644 --- a/arch/x86/kernel/kvmclock.c +++ b/arch/x86/kernel/kvmclock.c @@ -78,6 +78,34 @@ static cycle_t kvm_clock_read(void) return ret; } +/* + * If we don't do that, there is the possibility that the guest + * will calibrate under heavy load - thus, getting a lower lpj - + * and execute the delays themselves without load. This is wrong, + * because no delay loop can finish beforehand. + * Any heuristics is subject to fail, because ultimately, a large + * poll of guests can be running and trouble each other. So we preset + * lpj here + */ +static unsigned long kvm_get_tsc_khz(void) +{ + return preset_lpj; +} + +static void kvm_get_preset_lpj(void) +{ + struct pvclock_vcpu_time_info *src; + unsigned long khz; + u64 lpj; + + src = &per_cpu(hv_clock, 0); + khz = pvclock_tsc_khz(src); + + lpj = ((u64)khz * 1000); + do_div(lpj, HZ); + preset_lpj = lpj; +} + static struct clocksource kvm_clock = { .name = "kvm-clock", .read = kvm_clock_read, @@ -153,6 +181,7 @@ void __init kvmclock_init(void) pv_time_ops.get_wallclock = kvm_get_wallclock; pv_time_ops.set_wallclock = kvm_set_wallclock; pv_time_ops.sched_clock = kvm_clock_read; + pv_time_ops.get_tsc_khz = kvm_get_tsc_khz; #ifdef CONFIG_X86_LOCAL_APIC pv_apic_ops.setup_secondary_clock = kvm_setup_secondary_clock; #endif @@ -163,6 +192,7 @@ void __init kvmclock_init(void) #ifdef CONFIG_KEXEC machine_ops.crash_shutdown = kvm_crash_shutdown; #endif + kvm_get_preset_lpj(); clocksource_register(&kvm_clock); } } -- cgit v1.2.3 From a08546001c2b0f584ffc81987340943a7d6d6acb Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Tue, 23 Sep 2008 11:01:45 -0700 Subject: x86: pvclock: fix shadowed variable warning arch/x86/kernel/pvclock.c:102:6: warning: symbol 'tsc_khz' shadows an earlier one include/asm/tsc.h:18:21: originally declared here Signed-off-by: Harvey Harrison Signed-off-by: Andrew Morton Signed-off-by: Avi Kivity --- arch/x86/kernel/pvclock.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c index 1c54b5fb7aed..4f9c55f3a7c0 100644 --- a/arch/x86/kernel/pvclock.c +++ b/arch/x86/kernel/pvclock.c @@ -99,14 +99,14 @@ static unsigned pvclock_get_time_values(struct pvclock_shadow_time *dst, unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src) { - u64 tsc_khz = 1000000ULL << 32; + u64 pv_tsc_khz = 1000000ULL << 32; - do_div(tsc_khz, src->tsc_to_system_mul); + do_div(pv_tsc_khz, src->tsc_to_system_mul); if (src->tsc_shift < 0) - tsc_khz <<= -src->tsc_shift; + pv_tsc_khz <<= -src->tsc_shift; else - tsc_khz >>= src->tsc_shift; - return tsc_khz; + pv_tsc_khz >>= src->tsc_shift; + return pv_tsc_khz; } cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src) -- cgit v1.2.3 From 1339c367a842a9fc34b8fcfa1abadb07b11848ad Mon Sep 17 00:00:00 2001 From: Pavel Vasilyev Date: Wed, 15 Oct 2008 17:33:48 -0400 Subject: fix CONFIG_MMCONFIG=n build warning arch/x86/kernel/acpi/boot.c:100: warning: 'acpi_mcfg_64bit_base_addr' defined but not used http://bugzilla.kernel.org/show_bug.cgi?id=11743 Signed-off-by: Pavel Vasilyev Signed-off-by: Len Brown --- arch/x86/kernel/acpi/boot.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index c102af85df9c..0c2742f8c4da 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -97,7 +97,6 @@ static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE; #warning ACPI uses CMPXCHG, i486 and later hardware #endif -static int acpi_mcfg_64bit_base_addr __initdata = FALSE; /* -------------------------------------------------------------------------- Boot-time Configuration @@ -156,6 +155,9 @@ char *__init __acpi_map_table(unsigned long phys, unsigned long size) } #ifdef CONFIG_PCI_MMCONFIG + +static int acpi_mcfg_64bit_base_addr __initdata = FALSE; + /* The physical address of the MMCONFIG aperture. Set from ACPI tables. */ struct acpi_mcfg_allocation *pci_mmcfg_config; int pci_mmcfg_config_num; -- cgit v1.2.3 From 1f3fcd4b1adc972d5c6a34cfed98931c46575b49 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:49:44 -0700 Subject: add per_cpu_dyn_array support allow dyn-array in per_cpu area, allocated dynamically. usage: | /* in .h */ | struct kernel_stat { | struct cpu_usage_stat cpustat; | unsigned int *irqs; | }; | | /* in .c */ | DEFINE_PER_CPU(struct kernel_stat, kstat); | | DEFINE_PER_CPU_DYN_ARRAY_ADDR(per_cpu__kstat_irqs, per_cpu__kstat.irqs, sizeof(unsigned int), nr_irqs, sizeof(unsigned long), NULL); after setup_percpu()/per_cpu_alloc_dyn_array(), the dyn_array in per_cpu area is ready to use. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup_percpu.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 0e67f72d9316..13ba7a83808d 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -140,7 +140,7 @@ static void __init setup_cpu_pda_map(void) */ void __init setup_per_cpu_areas(void) { - ssize_t size = PERCPU_ENOUGH_ROOM; + ssize_t size, old_size; char *ptr; int cpu; @@ -148,7 +148,8 @@ void __init setup_per_cpu_areas(void) setup_cpu_pda_map(); /* Copy section for each CPU (we discard the original) */ - size = PERCPU_ENOUGH_ROOM; + old_size = PERCPU_ENOUGH_ROOM; + size = old_size + per_cpu_dyn_array_size(); printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n", size); @@ -176,6 +177,8 @@ void __init setup_per_cpu_areas(void) per_cpu_offset(cpu) = ptr - __per_cpu_start; memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); + per_cpu_alloc_dyn_array(cpu, ptr + old_size); + } printk(KERN_DEBUG "NR_CPUS: %d, nr_cpu_ids: %d, nr_node_ids %d\n", -- cgit v1.2.3 From 1f8ff037a871690c762d267d8a052529d3102fc9 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:49:45 -0700 Subject: x86: alloc dyn_array all together so could spare some memory with small alignment in bootmem also tighten the alignment checking, and make print out less debug info. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup_percpu.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 13ba7a83808d..2b7dab699e83 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -140,26 +140,31 @@ static void __init setup_cpu_pda_map(void) */ void __init setup_per_cpu_areas(void) { - ssize_t size, old_size; + ssize_t size, old_size, da_size; char *ptr; int cpu; + unsigned long align = 1; /* Setup cpu_pda map */ setup_cpu_pda_map(); /* Copy section for each CPU (we discard the original) */ old_size = PERCPU_ENOUGH_ROOM; - size = old_size + per_cpu_dyn_array_size(); + da_size = per_cpu_dyn_array_size(&align); + align = max_t(unsigned long, PAGE_SIZE, align); + size = roundup(old_size + da_size, align); printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n", size); for_each_possible_cpu(cpu) { #ifndef CONFIG_NEED_MULTIPLE_NODES - ptr = alloc_bootmem_pages(size); + ptr = __alloc_bootmem(size, align, + __pa(MAX_DMA_ADDRESS)); #else int node = early_cpu_to_node(cpu); if (!node_online(node) || !NODE_DATA(node)) { - ptr = alloc_bootmem_pages(size); + ptr = __alloc_bootmem(size, align, + __pa(MAX_DMA_ADDRESS)); printk(KERN_INFO "cpu %d has no node %d or node-local memory\n", cpu, node); @@ -168,7 +173,8 @@ void __init setup_per_cpu_areas(void) cpu, __pa(ptr)); } else { - ptr = alloc_bootmem_pages_node(NODE_DATA(node), size); + ptr = __alloc_bootmem_node(NODE_DATA(node), size, align, + __pa(MAX_DMA_ADDRESS)); if (ptr) printk(KERN_DEBUG "per cpu data for cpu%d on node%d at %016lx\n", cpu, node, __pa(ptr)); -- cgit v1.2.3 From 6da55c3e8da88e8a7cb6452160776ad6706798ad Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:49:46 -0700 Subject: x86: enable dyn_array support Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/vmlinux_32.lds.S | 1 + arch/x86/kernel/vmlinux_64.lds.S | 3 +++ 2 files changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index a9b8560adbc2..c36007ab3940 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -145,6 +145,7 @@ SECTIONS *(.x86_cpu_dev.init) __x86_cpu_dev_end = .; } + DYN_ARRAY_INIT(8) SECURITY_INIT . = ALIGN(4); .altinstructions : AT(ADDR(.altinstructions) - LOAD_OFFSET) { diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 46e05447405b..30973dbac8c2 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -173,6 +173,9 @@ SECTIONS *(.x86_cpu_dev.init) } __x86_cpu_dev_end = .; + + DYN_ARRAY_INIT(8) + SECURITY_INIT . = ALIGN(8); -- cgit v1.2.3 From 0799e432acfda879eaeef9622426bfa1434f3786 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:49:48 -0700 Subject: x86: use nr_irqs also add first_free_entry and pin_map_size, which were NR_IRQS derived constants. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 26 ++++++++++++++------------ arch/x86/kernel/io_apic_64.c | 33 +++++++++++++++++---------------- arch/x86/kernel/irq_32.c | 8 ++++---- arch/x86/kernel/irq_64.c | 8 ++++---- arch/x86/kernel/irqinit_32.c | 2 +- arch/x86/kernel/irqinit_64.c | 2 +- 6 files changed, 41 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index e710289f673e..d382990244f0 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -70,6 +70,7 @@ int timer_through_8259 __initdata; */ int sis_apic_bug = -1; +int first_free_entry = NR_IRQS; /* * # of IRQ routing registers */ @@ -100,6 +101,8 @@ static int disable_timer_pin_1 __initdata; #define MAX_PLUS_SHARED_IRQS NR_IRQS #define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) +int pin_map_size = PIN_MAP_SIZE; + /* * This is performance-critical, we want to do it O(1) * @@ -213,7 +216,6 @@ static void ioapic_mask_entry(int apic, int pin) */ static void add_pin_to_irq(unsigned int irq, int apic, int pin) { - static int first_free_entry = NR_IRQS; struct irq_pin_list *entry = irq_2_pin + irq; while (entry->next) @@ -222,7 +224,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) if (entry->pin != -1) { entry->next = first_free_entry; entry = irq_2_pin + entry->next; - if (++first_free_entry >= PIN_MAP_SIZE) + if (++first_free_entry >= pin_map_size) panic("io_apic.c: whoops"); } entry->apic = apic; @@ -457,7 +459,7 @@ static inline void rotate_irqs_among_cpus(unsigned long useful_load_threshold) int i, j; for_each_online_cpu(i) { - for (j = 0; j < NR_IRQS; j++) { + for (j = 0; j < nr_irqs; j++) { if (!irq_desc[j].action) continue; /* Is it a significant load ? */ @@ -492,7 +494,7 @@ static void do_irq_balance(void) if (!cpu_online(i)) continue; package_index = CPU_TO_PACKAGEINDEX(i); - for (j = 0; j < NR_IRQS; j++) { + for (j = 0; j < nr_irqs; j++) { unsigned long value_now, delta; /* Is this an active IRQ or balancing disabled ? */ if (!irq_desc[j].action || irq_balancing_disabled(j)) @@ -587,7 +589,7 @@ tryanotherirq: */ move_this_load = 0; selected_irq = -1; - for (j = 0; j < NR_IRQS; j++) { + for (j = 0; j < nr_irqs; j++) { /* Is this an active IRQ? */ if (!irq_desc[j].action) continue; @@ -664,7 +666,7 @@ static int balanced_irq(void *unused) long time_remaining = balanced_irq_interval; /* push everything to CPU 0 to give us a starting point. */ - for (i = 0 ; i < NR_IRQS ; i++) { + for (i = 0 ; i < nr_irqs ; i++) { irq_desc[i].pending_mask = cpumask_of_cpu(0); set_pending_irq(i, cpumask_of_cpu(0)); } @@ -712,8 +714,8 @@ static int __init balanced_irq_init(void) physical_balance = 1; for_each_online_cpu(i) { - irq_cpu_data[i].irq_delta = kzalloc(sizeof(unsigned long) * NR_IRQS, GFP_KERNEL); - irq_cpu_data[i].last_irq = kzalloc(sizeof(unsigned long) * NR_IRQS, GFP_KERNEL); + irq_cpu_data[i].irq_delta = kzalloc(sizeof(unsigned long) * nr_irqs, GFP_KERNEL); + irq_cpu_data[i].last_irq = kzalloc(sizeof(unsigned long) * nr_irqs, GFP_KERNEL); if (irq_cpu_data[i].irq_delta == NULL || irq_cpu_data[i].last_irq == NULL) { printk(KERN_ERR "balanced_irq_init: out of memory"); goto failed; @@ -1441,7 +1443,7 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for (i = 0; i < NR_IRQS; i++) { + for (i = 0; i < nr_irqs; i++) { struct irq_pin_list *entry = irq_2_pin + i; if (entry->pin < 0) continue; @@ -1621,7 +1623,7 @@ static void __init enable_IO_APIC(void) int i, apic; unsigned long flags; - for (i = 0; i < PIN_MAP_SIZE; i++) { + for (i = 0; i < pin_map_size; i++) { irq_2_pin[i].pin = -1; irq_2_pin[i].next = 0; } @@ -2005,7 +2007,7 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for (irq = 0; irq < NR_IRQS ; irq++) { + for (irq = 0; irq < nr_irqs ; irq++) { if (IO_APIC_IRQ(irq) && !irq_vector[irq]) { /* * Hmm.. We don't have an entry for this, @@ -2449,7 +2451,7 @@ int create_irq(void) irq = -ENOSPC; spin_lock_irqsave(&vector_lock, flags); - for (new = (NR_IRQS - 1); new >= 0; new--) { + for (new = (nr_irqs - 1); new >= 0; new--) { if (platform_legacy_irq(new)) continue; if (irq_vector[new] != 0) diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 02063ae042f7..448384c7c1e8 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -132,6 +132,7 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); #define MAX_PLUS_SHARED_IRQS NR_IRQS #define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) +int pin_map_size = PIN_MAP_SIZE; /* * This is performance-critical, we want to do it O(1) * @@ -224,7 +225,7 @@ static inline void io_apic_sync(unsigned int apic) int pin; \ struct irq_pin_list *entry = irq_2_pin + irq; \ \ - BUG_ON(irq >= NR_IRQS); \ + BUG_ON(irq >= nr_irqs); \ for (;;) { \ unsigned int reg; \ pin = entry->pin; \ @@ -301,7 +302,7 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) int apic, pin; struct irq_pin_list *entry = irq_2_pin + irq; - BUG_ON(irq >= NR_IRQS); + BUG_ON(irq >= nr_irqs); for (;;) { unsigned int reg; apic = entry->apic; @@ -358,19 +359,19 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ +int first_free_entry = NR_IRQS; static void add_pin_to_irq(unsigned int irq, int apic, int pin) { - static int first_free_entry = NR_IRQS; struct irq_pin_list *entry = irq_2_pin + irq; - BUG_ON(irq >= NR_IRQS); + BUG_ON(irq >= nr_irqs); while (entry->next) entry = irq_2_pin + entry->next; if (entry->pin != -1) { entry->next = first_free_entry; entry = irq_2_pin + entry->next; - if (++first_free_entry >= PIN_MAP_SIZE) + if (++first_free_entry >= pin_map_size) panic("io_apic.c: ran out of irq_2_pin entries!"); } entry->apic = apic; @@ -634,7 +635,7 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) best_guess = irq; } } - BUG_ON(best_guess >= NR_IRQS); + BUG_ON(best_guess >= nr_irqs); return best_guess; } @@ -766,7 +767,7 @@ static int pin_2_irq(int idx, int apic, int pin) irq += nr_ioapic_registers[i++]; irq += pin; } - BUG_ON(irq >= NR_IRQS); + BUG_ON(irq >= nr_irqs); return irq; } @@ -801,7 +802,7 @@ static int __assign_irq_vector(int irq, cpumask_t mask) int cpu; struct irq_cfg *cfg; - BUG_ON((unsigned)irq >= NR_IRQS); + BUG_ON((unsigned)irq >= nr_irqs); cfg = &irq_cfg[irq]; /* Only try and allocate irqs on cpus that are present */ @@ -875,7 +876,7 @@ static void __clear_irq_vector(int irq) cpumask_t mask; int cpu, vector; - BUG_ON((unsigned)irq >= NR_IRQS); + BUG_ON((unsigned)irq >= nr_irqs); cfg = &irq_cfg[irq]; BUG_ON(!cfg->vector); @@ -895,7 +896,7 @@ void __setup_vector_irq(int cpu) int irq, vector; /* Mark the inuse vectors */ - for (irq = 0; irq < NR_IRQS; ++irq) { + for (irq = 0; irq < nr_irqs; ++irq) { if (!cpu_isset(cpu, irq_cfg[irq].domain)) continue; vector = irq_cfg[irq].vector; @@ -1193,7 +1194,7 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for (i = 0; i < NR_IRQS; i++) { + for (i = 0; i < nr_irqs; i++) { struct irq_pin_list *entry = irq_2_pin + i; if (entry->pin < 0) continue; @@ -1366,7 +1367,7 @@ void __init enable_IO_APIC(void) int i, apic; unsigned long flags; - for (i = 0; i < PIN_MAP_SIZE; i++) { + for (i = 0; i < pin_map_size; i++) { irq_2_pin[i].pin = -1; irq_2_pin[i].next = 0; } @@ -1658,7 +1659,7 @@ static void ir_irq_migration(struct work_struct *work) { int irq; - for (irq = 0; irq < NR_IRQS; irq++) { + for (irq = 0; irq < nr_irqs; irq++) { struct irq_desc *desc = irq_desc + irq; if (desc->status & IRQ_MOVE_PENDING) { unsigned long flags; @@ -1707,7 +1708,7 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) struct irq_desc *desc; struct irq_cfg *cfg; irq = __get_cpu_var(vector_irq)[vector]; - if (irq >= NR_IRQS) + if (irq >= nr_irqs) continue; desc = irq_desc + irq; @@ -1865,7 +1866,7 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for (irq = 0; irq < NR_IRQS ; irq++) { + for (irq = 0; irq < nr_irqs ; irq++) { if (IO_APIC_IRQ(irq) && !irq_cfg[irq].vector) { /* * Hmm.. We don't have an entry for this, @@ -2279,7 +2280,7 @@ int create_irq(void) irq = -ENOSPC; spin_lock_irqsave(&vector_lock, flags); - for (new = (NR_IRQS - 1); new >= 0; new--) { + for (new = (nr_irqs - 1); new >= 0; new--) { if (platform_legacy_irq(new)) continue; if (irq_cfg[new].vector != 0) diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index b71e02d42f4f..4c7ffb32854c 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -226,7 +226,7 @@ unsigned int do_IRQ(struct pt_regs *regs) int overflow, irq = ~regs->orig_ax; struct irq_desc *desc = irq_desc + irq; - if (unlikely((unsigned)irq >= NR_IRQS)) { + if (unlikely((unsigned)irq >= nr_irqs)) { printk(KERN_EMERG "%s: cannot handle IRQ %d\n", __func__, irq); BUG(); @@ -271,7 +271,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i < NR_IRQS) { + if (i < nr_irqs) { unsigned any_count = 0; spin_lock_irqsave(&irq_desc[i].lock, flags); @@ -303,7 +303,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); skip: spin_unlock_irqrestore(&irq_desc[i].lock, flags); - } else if (i == NR_IRQS) { + } else if (i == nr_irqs) { seq_printf(p, "NMI: "); for_each_online_cpu(j) seq_printf(p, "%10u ", nmi_count(j)); @@ -396,7 +396,7 @@ void fixup_irqs(cpumask_t map) unsigned int irq; static int warned; - for (irq = 0; irq < NR_IRQS; irq++) { + for (irq = 0; irq < nr_irqs; irq++) { cpumask_t mask; if (irq == 2) continue; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index f065fe9071b9..e1f0839430d2 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -81,7 +81,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i < NR_IRQS) { + if (i < nr_irqs) { unsigned any_count = 0; spin_lock_irqsave(&irq_desc[i].lock, flags); @@ -112,7 +112,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); skip: spin_unlock_irqrestore(&irq_desc[i].lock, flags); - } else if (i == NR_IRQS) { + } else if (i == nr_irqs) { seq_printf(p, "NMI: "); for_each_online_cpu(j) seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count); @@ -201,7 +201,7 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) stack_overflow_check(regs); #endif - if (likely(irq < NR_IRQS)) + if (likely(irq < nr_irqs)) generic_handle_irq(irq); else { if (!disable_apic) @@ -224,7 +224,7 @@ void fixup_irqs(cpumask_t map) unsigned int irq; static int warned; - for (irq = 0; irq < NR_IRQS; irq++) { + for (irq = 0; irq < nr_irqs; irq++) { cpumask_t mask; int break_affinity = 0; int set_affinity = 1; diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 9200a1e2752d..65c1c9507707 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -100,7 +100,7 @@ void __init native_init_IRQ(void) */ for (i = 0; i < (NR_VECTORS - FIRST_EXTERNAL_VECTOR); i++) { int vector = FIRST_EXTERNAL_VECTOR + i; - if (i >= NR_IRQS) + if (i >= nr_irqs) break; /* SYSCALL_VECTOR was reserved in trap_init. */ if (!test_bit(vector, used_vectors)) diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index 5b5be9d43c2a..165c5d9b0d1a 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -142,7 +142,7 @@ void __init init_ISA_irqs(void) init_bsp_APIC(); init_8259A(0); - for (i = 0; i < NR_IRQS; i++) { + for (i = 0; i < nr_irqs; i++) { irq_desc[i].status = IRQ_DISABLED; irq_desc[i].action = NULL; irq_desc[i].depth = 1; -- cgit v1.2.3 From 301e619020dd67bde7e7e64bb9ffb7f30d26c979 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:02 -0700 Subject: x86: use dyn_array in io_apic_xx.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 54 +++++++++++++++++++++++++++++++++++--------- arch/x86/kernel/io_apic_64.c | 28 +++++++++++++++++------ arch/x86/kernel/setup.c | 6 +++++ 3 files changed, 70 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index d382990244f0..7f2bcc3dad82 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -70,7 +70,7 @@ int timer_through_8259 __initdata; */ int sis_apic_bug = -1; -int first_free_entry = NR_IRQS; +int first_free_entry; /* * # of IRQ routing registers */ @@ -98,10 +98,7 @@ static int disable_timer_pin_1 __initdata; * Rough estimation of how many shared IRQs there are, can * be changed anytime. */ -#define MAX_PLUS_SHARED_IRQS NR_IRQS -#define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) - -int pin_map_size = PIN_MAP_SIZE; +int pin_map_size; /* * This is performance-critical, we want to do it O(1) @@ -112,7 +109,9 @@ int pin_map_size = PIN_MAP_SIZE; static struct irq_pin_list { int apic, pin, next; -} irq_2_pin[PIN_MAP_SIZE]; +} *irq_2_pin; + +DEFINE_DYN_ARRAY(irq_2_pin, sizeof(struct irq_pin_list), pin_map_size, 16, NULL); struct io_apic { unsigned int index; @@ -403,9 +402,28 @@ static struct irq_cpu_info { #define CPU_TO_PACKAGEINDEX(i) (first_cpu(per_cpu(cpu_sibling_map, i))) -static cpumask_t balance_irq_affinity[NR_IRQS] = { - [0 ... NR_IRQS-1] = CPU_MASK_ALL -}; +static cpumask_t balance_irq_affinity_init __initdata = CPU_MASK_ALL; + +static cpumask_t *balance_irq_affinity; + + +static void __init irq_affinity_init_work(void *data) +{ + struct dyn_array *da = data; + + int i; + struct balance_irq_affinity *affinity; + + affinity = *da->name; + + for (i = 0; i < *da->nr; i++) + memcpy(&affinity[i], &balance_irq_affinity_init, + sizeof(struct balance_irq_affinity)); + +} + +DEFINE_DYN_ARRAY(balance_irq_affinity, sizeof(struct balance_irq_affinity), nr_irqs, PAGE_SIZE, irq_affinity_init_work); + void set_balance_irq_affinity(unsigned int irq, cpumask_t mask) { @@ -1170,14 +1188,28 @@ static inline int IO_APIC_irq_trigger(int irq) } /* irq_vectors is indexed by the sum of all RTEs in all I/O APICs. */ -static u8 irq_vector[NR_IRQ_VECTORS] __read_mostly = { FIRST_DEVICE_VECTOR , 0 }; +static u8 irq_vector_init_first __initdata = FIRST_DEVICE_VECTOR; +static u8 *irq_vector; + +static void __init irq_vector_init_work(void *data) +{ + struct dyn_array *da = data; + + u8 *irq_vec; + + irq_vec = *da->name; + + irq_vec[0] = irq_vector_init_first; +} + +DEFINE_DYN_ARRAY(irq_vector, sizeof(u8), nr_irqs, PAGE_SIZE, irq_vector_init_work); static int __assign_irq_vector(int irq) { static int current_vector = FIRST_DEVICE_VECTOR, current_offset; int vector, offset; - BUG_ON((unsigned)irq >= NR_IRQ_VECTORS); + BUG_ON((unsigned)irq >= nr_irqs); if (irq_vector[irq] > 0) return irq_vector[irq]; diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 448384c7c1e8..93a3ffabfe6a 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -66,7 +66,7 @@ struct irq_cfg { }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ -static struct irq_cfg irq_cfg[NR_IRQS] __read_mostly = { +static struct irq_cfg irq_cfg_legacy[] __initdata = { [0] = { .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, [1] = { .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, [2] = { .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, @@ -85,6 +85,17 @@ static struct irq_cfg irq_cfg[NR_IRQS] __read_mostly = { [15] = { .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, }; +static struct irq_cfg *irq_cfg; + +static void __init init_work(void *data) +{ + struct dyn_array *da = data; + + memcpy(*da->name, irq_cfg_legacy, sizeof(irq_cfg_legacy)); +} + +DEFINE_DYN_ARRAY(irq_cfg, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); + static int assign_irq_vector(int irq, cpumask_t mask); int first_system_vector = 0xfe; @@ -129,10 +140,9 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); * Rough estimation of how many shared IRQs there are, can * be changed anytime. */ -#define MAX_PLUS_SHARED_IRQS NR_IRQS -#define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) -int pin_map_size = PIN_MAP_SIZE; +int pin_map_size; + /* * This is performance-critical, we want to do it O(1) * @@ -141,8 +151,12 @@ int pin_map_size = PIN_MAP_SIZE; */ static struct irq_pin_list { - short apic, pin, next; -} irq_2_pin[PIN_MAP_SIZE]; + short apic, pin; + int next; +} *irq_2_pin; + +DEFINE_DYN_ARRAY(irq_2_pin, sizeof(struct irq_pin_list), pin_map_size, sizeof(struct irq_pin_list), NULL); + struct io_apic { unsigned int index; @@ -359,7 +373,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -int first_free_entry = NR_IRQS; +int first_free_entry; static void add_pin_to_irq(unsigned int irq, int apic, int pin) { struct irq_pin_list *entry = irq_2_pin + irq; diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 2255782e8d4b..558ec26b08e2 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1067,9 +1067,15 @@ void __init setup_arch(char **cmdline_p) #endif prefill_possible_map(); + #ifdef CONFIG_X86_64 + /* need to wait for nr_cpu_ids settle down */ + if (nr_irqs == NR_IRQS) + nr_irqs = 32 * nr_cpu_ids + 224; init_cpu_to_node(); #endif + pin_map_size = nr_irqs * 2; + first_free_entry = nr_irqs; init_apic_mappings(); ioapic_init_mappings(); -- cgit v1.2.3 From a84488c213a8cfc29200344a6fb6357d48c8ed85 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 19 Aug 2008 20:50:31 -0700 Subject: irq: sparse irqs, fix #3 fix non-APIC UP build: arch/x86/kernel/built-in.o: In function `setup_arch': : undefined reference to `pin_map_size' arch/x86/kernel/built-in.o: In function `setup_arch': : undefined reference to `first_free_entry' Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 558ec26b08e2..02e3a6697977 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1074,8 +1074,10 @@ void __init setup_arch(char **cmdline_p) nr_irqs = 32 * nr_cpu_ids + 224; init_cpu_to_node(); #endif +#ifdef CONFIG_X86_IO_APIC pin_map_size = nr_irqs * 2; first_free_entry = nr_irqs; +#endif init_apic_mappings(); ioapic_init_mappings(); -- cgit v1.2.3 From 71f521bbaf375b685aeea20c6b0ed8600cd6edfe Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:03 -0700 Subject: x86, irq: get nr_irqs from madt Until now, NR_IRQS was derived from black magic defines that had to be "large enough" to both accomodate NR_CPUS and MAX_NR_IO_APICs. This resulted in a way too large irq_desc[] array on most x86 systems. Especially with larger CPU masks, the size of irq_desc can spiral out of control quickly. So be smarter about it and use precise allocation instead: determine the default maximum possible IRQ number from the ACPI MADT. Use a minimum limit of at least 32 IRQs for broken BIOSes. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index eb875cdc7367..3e9d163fd92f 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -957,6 +957,29 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) nr_ioapics++; } +int get_nr_irqs_via_madt(void) +{ + int idx; + int nr = 0; + + for (idx = 0; idx < nr_ioapics; idx++) { + if (mp_ioapic_routing[idx].gsi_end > nr) + nr = mp_ioapic_routing[idx].gsi_end; + } + + nr++; + + /* double it for hotplug and msi and nmi */ + nr <<= 1; + + /* something wrong ? */ + if (nr < 32) + nr = 32; + + return nr; + +} + static void assign_to_mp_irq(struct mp_config_intsrc *m, struct mp_config_intsrc *mp_irq) { @@ -1254,9 +1277,12 @@ static int __init acpi_parse_madt_ioapic_entries(void) return count; } + + nr_irqs = get_nr_irqs_via_madt(); + count = acpi_table_parse_madt(ACPI_MADT_TYPE_INTERRUPT_OVERRIDE, acpi_parse_int_src_ovr, - NR_IRQ_VECTORS); + nr_irqs); if (count < 0) { printk(KERN_ERR PREFIX "Error parsing interrupt source overrides entry\n"); @@ -1276,7 +1302,7 @@ static int __init acpi_parse_madt_ioapic_entries(void) count = acpi_table_parse_madt(ACPI_MADT_TYPE_NMI_SOURCE, acpi_parse_nmi_src, - NR_IRQ_VECTORS); + nr_irqs); if (count < 0) { printk(KERN_ERR PREFIX "Error parsing NMI SRC entry\n"); /* TBD: Cleanup to allow fallback to MPS */ -- cgit v1.2.3 From 08678b0841267c1d00d771fe01548d86043d065e Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:05 -0700 Subject: generic: sparse irqs: use irq_desc() together with dyn_array, instead of irq_desc[] add CONFIG_HAVE_SPARSE_IRQ to for use condensed array. Get rid of irq_desc[] array assumptions. Preallocate 32 irq_desc, and irq_desc() will try to get more. ( No change in functionality is expected anywhere, except the odd build failure where we missed a code site or where a crossing commit itroduces new irq_desc[] usage. ) v2: according to Eric, change get_irq_desc() to irq_desc() Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 46 +++++++++++++++++++------- arch/x86/kernel/io_apic_64.c | 75 +++++++++++++++++++++++++++--------------- arch/x86/kernel/irq_32.c | 24 ++++++++------ arch/x86/kernel/irq_64.c | 35 +++++++++++--------- arch/x86/kernel/irqinit_64.c | 10 +++--- arch/x86/kernel/visws_quirks.c | 30 +++++++++-------- 6 files changed, 139 insertions(+), 81 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 7f2bcc3dad82..c2160cfdec9b 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -345,6 +345,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) struct irq_pin_list *entry = irq_2_pin + irq; unsigned int apicid_value; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, cpumask, cpu_online_map); if (cpus_empty(tmp)) @@ -365,7 +366,8 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) break; entry = irq_2_pin + entry->next; } - irq_desc[irq].affinity = cpumask; + desc = irq_to_desc(irq); + desc->affinity = cpumask; spin_unlock_irqrestore(&ioapic_lock, flags); } @@ -475,10 +477,12 @@ static inline void balance_irq(int cpu, int irq) static inline void rotate_irqs_among_cpus(unsigned long useful_load_threshold) { int i, j; + struct irq_desc *desc; for_each_online_cpu(i) { for (j = 0; j < nr_irqs; j++) { - if (!irq_desc[j].action) + desc = irq_to_desc(j); + if (!desc->action) continue; /* Is it a significant load ? */ if (IRQ_DELTA(CPU_TO_PACKAGEINDEX(i), j) < @@ -505,6 +509,7 @@ static void do_irq_balance(void) unsigned long tmp_cpu_irq; unsigned long imbalance = 0; cpumask_t allowed_mask, target_cpu_mask, tmp; + struct irq_desc *desc; for_each_possible_cpu(i) { int package_index; @@ -515,7 +520,8 @@ static void do_irq_balance(void) for (j = 0; j < nr_irqs; j++) { unsigned long value_now, delta; /* Is this an active IRQ or balancing disabled ? */ - if (!irq_desc[j].action || irq_balancing_disabled(j)) + desc = irq_to_desc(j); + if (!desc->action || irq_balancing_disabled(j)) continue; if (package_index == i) IRQ_DELTA(package_index, j) = 0; @@ -609,7 +615,8 @@ tryanotherirq: selected_irq = -1; for (j = 0; j < nr_irqs; j++) { /* Is this an active IRQ? */ - if (!irq_desc[j].action) + desc = irq_to_desc(j); + if (!desc->action) continue; if (imbalance <= IRQ_DELTA(max_loaded, j)) continue; @@ -682,10 +689,12 @@ static int balanced_irq(void *unused) int i; unsigned long prev_balance_time = jiffies; long time_remaining = balanced_irq_interval; + struct irq_desc *desc; /* push everything to CPU 0 to give us a starting point. */ for (i = 0 ; i < nr_irqs ; i++) { - irq_desc[i].pending_mask = cpumask_of_cpu(0); + desc = irq_to_desc(i); + desc->pending_mask = cpumask_of_cpu(0); set_pending_irq(i, cpumask_of_cpu(0)); } @@ -1254,13 +1263,16 @@ static struct irq_chip ioapic_chip; static void ioapic_register_intr(int irq, int vector, unsigned long trigger) { + struct irq_desc *desc; + + desc = irq_to_desc(irq); if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) { - irq_desc[irq].status |= IRQ_LEVEL; + desc->status |= IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_fasteoi_irq, "fasteoi"); } else { - irq_desc[irq].status &= ~IRQ_LEVEL; + desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_edge_irq, "edge"); } @@ -2027,6 +2039,7 @@ static struct irq_chip ioapic_chip __read_mostly = { static inline void init_IO_APIC_traps(void) { int irq; + struct irq_desc *desc; /* * NOTE! The local APIC isn't very good at handling @@ -2048,9 +2061,11 @@ static inline void init_IO_APIC_traps(void) */ if (irq < 16) make_8259A_irq(irq); - else + else { + desc = irq_to_desc(irq); /* Strange. Oh, well.. */ - irq_desc[irq].chip = &no_irq_chip; + desc->chip = &no_irq_chip; + } } } } @@ -2089,7 +2104,10 @@ static struct irq_chip lapic_chip __read_mostly = { static void lapic_register_intr(int irq, int vector) { - irq_desc[irq].status &= ~IRQ_LEVEL; + struct irq_desc *desc; + + desc = irq_to_desc(irq); + desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, "edge"); set_intr_gate(vector, interrupt[irq]); @@ -2556,6 +2574,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) unsigned int dest; cpumask_t tmp; int vector; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2575,7 +2594,8 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); write_msi_msg(irq, &msg); - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif /* CONFIG_SMP */ @@ -2649,6 +2669,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2659,7 +2680,8 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) dest = cpu_mask_to_apicid(mask); target_ht_irq(irq, dest); - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 93a3ffabfe6a..cab5a25d81b1 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -345,6 +345,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) unsigned long flags; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -361,9 +362,10 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) */ dest = SET_APIC_LOGICAL_ID(dest); + desc = irq_to_desc(irq); spin_lock_irqsave(&ioapic_lock, flags); __target_IO_APIC_irq(irq, dest, cfg->vector); - irq_desc[irq].affinity = mask; + desc->affinity = mask; spin_unlock_irqrestore(&ioapic_lock, flags); } #endif @@ -933,14 +935,17 @@ static struct irq_chip ir_ioapic_chip; static void ioapic_register_intr(int irq, unsigned long trigger) { + struct irq_desc *desc; + + desc = irq_to_desc(irq); if (trigger) - irq_desc[irq].status |= IRQ_LEVEL; + desc->status |= IRQ_LEVEL; else - irq_desc[irq].status &= ~IRQ_LEVEL; + desc->status &= ~IRQ_LEVEL; #ifdef CONFIG_INTR_REMAP if (irq_remapped(irq)) { - irq_desc[irq].status |= IRQ_MOVE_PCNTXT; + desc->status |= IRQ_MOVE_PCNTXT; if (trigger) set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, handle_fasteoi_irq, @@ -1596,10 +1601,10 @@ static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); static void migrate_ioapic_irq(int irq, cpumask_t mask) { struct irq_cfg *cfg = irq_cfg + irq; - struct irq_desc *desc = irq_desc + irq; + struct irq_desc *desc; cpumask_t tmp, cleanup_mask; struct irte irte; - int modify_ioapic_rte = desc->status & IRQ_LEVEL; + int modify_ioapic_rte; unsigned int dest; unsigned long flags; @@ -1616,6 +1621,8 @@ static void migrate_ioapic_irq(int irq, cpumask_t mask) cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); + desc = irq_to_desc(irq); + modify_ioapic_rte = desc->status & IRQ_LEVEL; if (modify_ioapic_rte) { spin_lock_irqsave(&ioapic_lock, flags); __target_IO_APIC_irq(irq, dest, cfg->vector); @@ -1637,12 +1644,13 @@ static void migrate_ioapic_irq(int irq, cpumask_t mask) cfg->move_in_progress = 0; } - irq_desc[irq].affinity = mask; + desc->affinity = mask; } static int migrate_irq_remapped_level(int irq) { int ret = -1; + struct irq_desc *desc = irq_to_desc(irq); mask_IO_APIC_irq(irq); @@ -1658,11 +1666,11 @@ static int migrate_irq_remapped_level(int irq) } /* everthing is clear. we have right of way */ - migrate_ioapic_irq(irq, irq_desc[irq].pending_mask); + migrate_ioapic_irq(irq, desc->pending_mask); ret = 0; - irq_desc[irq].status &= ~IRQ_MOVE_PENDING; - cpus_clear(irq_desc[irq].pending_mask); + desc->status &= ~IRQ_MOVE_PENDING; + cpus_clear(desc->pending_mask); unmask: unmask_IO_APIC_irq(irq); @@ -1674,7 +1682,7 @@ static void ir_irq_migration(struct work_struct *work) int irq; for (irq = 0; irq < nr_irqs; irq++) { - struct irq_desc *desc = irq_desc + irq; + struct irq_desc *desc = irq_to_desc(irq); if (desc->status & IRQ_MOVE_PENDING) { unsigned long flags; @@ -1686,8 +1694,7 @@ static void ir_irq_migration(struct work_struct *work) continue; } - desc->chip->set_affinity(irq, - irq_desc[irq].pending_mask); + desc->chip->set_affinity(irq, desc->pending_mask); spin_unlock_irqrestore(&desc->lock, flags); } } @@ -1698,9 +1705,11 @@ static void ir_irq_migration(struct work_struct *work) */ static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) { - if (irq_desc[irq].status & IRQ_LEVEL) { - irq_desc[irq].status |= IRQ_MOVE_PENDING; - irq_desc[irq].pending_mask = mask; + struct irq_desc *desc = irq_to_desc(irq); + + if (desc->status & IRQ_LEVEL) { + desc->status |= IRQ_MOVE_PENDING; + desc->pending_mask = mask; migrate_irq_remapped_level(irq); return; } @@ -1725,7 +1734,7 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) if (irq >= nr_irqs) continue; - desc = irq_desc + irq; + desc = irq_to_desc(irq); cfg = irq_cfg + irq; spin_lock(&desc->lock); if (!cfg->move_cleanup_count) @@ -1791,7 +1800,7 @@ static void ack_apic_level(unsigned int irq) irq_complete_move(irq); #ifdef CONFIG_GENERIC_PENDING_IRQ /* If we are moving the irq we need to mask it */ - if (unlikely(irq_desc[irq].status & IRQ_MOVE_PENDING)) { + if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { do_unmask_irq = 1; mask_IO_APIC_irq(irq); } @@ -1868,6 +1877,7 @@ static struct irq_chip ir_ioapic_chip __read_mostly = { static inline void init_IO_APIC_traps(void) { int irq; + struct irq_desc *desc; /* * NOTE! The local APIC isn't very good at handling @@ -1889,9 +1899,11 @@ static inline void init_IO_APIC_traps(void) */ if (irq < 16) make_8259A_irq(irq); - else + else { + desc = irq_to_desc(irq); /* Strange. Oh, well.. */ - irq_desc[irq].chip = &no_irq_chip; + desc->chip = &no_irq_chip; + } } } } @@ -1926,7 +1938,10 @@ static struct irq_chip lapic_chip __read_mostly = { static void lapic_register_intr(int irq) { - irq_desc[irq].status &= ~IRQ_LEVEL; + struct irq_desc *desc; + + desc = irq_to_desc(irq); + desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, "edge"); } @@ -2402,6 +2417,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) struct msi_msg msg; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2421,7 +2437,8 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); write_msi_msg(irq, &msg); - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #ifdef CONFIG_INTR_REMAP @@ -2435,6 +2452,7 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) unsigned int dest; cpumask_t tmp, cleanup_mask; struct irte irte; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2469,7 +2487,8 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) cfg->move_in_progress = 0; } - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif #endif /* CONFIG_SMP */ @@ -2543,7 +2562,7 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) #ifdef CONFIG_INTR_REMAP if (irq_remapped(irq)) { - struct irq_desc *desc = irq_desc + irq; + struct irq_desc *desc = irq_to_desc(irq); /* * irq migration in process context */ @@ -2655,6 +2674,7 @@ static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) struct msi_msg msg; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2674,7 +2694,8 @@ static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); dmar_msi_write(irq, &msg); - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif /* CONFIG_SMP */ @@ -2731,6 +2752,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) struct irq_cfg *cfg = irq_cfg + irq; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2743,7 +2765,8 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) dest = cpu_mask_to_apicid(tmp); target_ht_irq(irq, dest, cfg->vector); - irq_desc[irq].affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 4c7ffb32854c..ede513be517d 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -224,7 +224,7 @@ unsigned int do_IRQ(struct pt_regs *regs) struct pt_regs *old_regs; /* high bit used in ret_from_ code */ int overflow, irq = ~regs->orig_ax; - struct irq_desc *desc = irq_desc + irq; + struct irq_desc *desc = irq_to_desc(irq); if (unlikely((unsigned)irq >= nr_irqs)) { printk(KERN_EMERG "%s: cannot handle IRQ %d\n", @@ -273,15 +273,16 @@ int show_interrupts(struct seq_file *p, void *v) if (i < nr_irqs) { unsigned any_count = 0; + struct irq_desc *desc = irq_to_desc(i); - spin_lock_irqsave(&irq_desc[i].lock, flags); + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP any_count = kstat_irqs(i); #else for_each_online_cpu(j) any_count |= kstat_cpu(j).irqs[i]; #endif - action = irq_desc[i].action; + action = desc->action; if (!action && !any_count) goto skip; seq_printf(p, "%3d: ",i); @@ -291,8 +292,8 @@ int show_interrupts(struct seq_file *p, void *v) for_each_online_cpu(j) seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]); #endif - seq_printf(p, " %8s", irq_desc[i].chip->name); - seq_printf(p, "-%-8s", irq_desc[i].name); + seq_printf(p, " %8s", desc->chip->name); + seq_printf(p, "-%-8s", desc->name); if (action) { seq_printf(p, " %s", action->name); @@ -302,7 +303,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); skip: - spin_unlock_irqrestore(&irq_desc[i].lock, flags); + spin_unlock_irqrestore(&desc->lock, flags); } else if (i == nr_irqs) { seq_printf(p, "NMI: "); for_each_online_cpu(j) @@ -398,17 +399,20 @@ void fixup_irqs(cpumask_t map) for (irq = 0; irq < nr_irqs; irq++) { cpumask_t mask; + struct irq_desc *desc; + if (irq == 2) continue; - cpus_and(mask, irq_desc[irq].affinity, map); + desc = irq_to_desc(irq); + cpus_and(mask, desc->affinity, map); if (any_online_cpu(mask) == NR_CPUS) { printk("Breaking affinity for irq %i\n", irq); mask = map; } - if (irq_desc[irq].chip->set_affinity) - irq_desc[irq].chip->set_affinity(irq, mask); - else if (irq_desc[irq].action && !(warned++)) + if (desc->chip->set_affinity) + desc->chip->set_affinity(irq, mask); + else if (desc->action && !(warned++)) printk("Cannot set affinity for irq %i\n", irq); } diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index e1f0839430d2..738eb65a924e 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -83,15 +83,16 @@ int show_interrupts(struct seq_file *p, void *v) if (i < nr_irqs) { unsigned any_count = 0; + struct irq_desc *desc = irq_to_desc(i); - spin_lock_irqsave(&irq_desc[i].lock, flags); + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP any_count = kstat_irqs(i); #else for_each_online_cpu(j) any_count |= kstat_cpu(j).irqs[i]; #endif - action = irq_desc[i].action; + action = desc->action; if (!action && !any_count) goto skip; seq_printf(p, "%3d: ",i); @@ -101,8 +102,8 @@ int show_interrupts(struct seq_file *p, void *v) for_each_online_cpu(j) seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]); #endif - seq_printf(p, " %8s", irq_desc[i].chip->name); - seq_printf(p, "-%-8s", irq_desc[i].name); + seq_printf(p, " %8s", desc->chip->name); + seq_printf(p, "-%-8s", desc->name); if (action) { seq_printf(p, " %s", action->name); @@ -111,7 +112,7 @@ int show_interrupts(struct seq_file *p, void *v) } seq_putc(p, '\n'); skip: - spin_unlock_irqrestore(&irq_desc[i].lock, flags); + spin_unlock_irqrestore(&desc->lock, flags); } else if (i == nr_irqs) { seq_printf(p, "NMI: "); for_each_online_cpu(j) @@ -228,37 +229,39 @@ void fixup_irqs(cpumask_t map) cpumask_t mask; int break_affinity = 0; int set_affinity = 1; + struct irq_desc *desc; if (irq == 2) continue; + desc = irq_to_desc(irq); /* interrupt's are disabled at this point */ - spin_lock(&irq_desc[irq].lock); + spin_lock(&desc->lock); if (!irq_has_action(irq) || - cpus_equal(irq_desc[irq].affinity, map)) { - spin_unlock(&irq_desc[irq].lock); + cpus_equal(desc->affinity, map)) { + spin_unlock(&desc->lock); continue; } - cpus_and(mask, irq_desc[irq].affinity, map); + cpus_and(mask, desc->affinity, map); if (cpus_empty(mask)) { break_affinity = 1; mask = map; } - if (irq_desc[irq].chip->mask) - irq_desc[irq].chip->mask(irq); + if (desc->chip->mask) + desc->chip->mask(irq); - if (irq_desc[irq].chip->set_affinity) - irq_desc[irq].chip->set_affinity(irq, mask); + if (desc->chip->set_affinity) + desc->chip->set_affinity(irq, mask); else if (!(warned++)) set_affinity = 0; - if (irq_desc[irq].chip->unmask) - irq_desc[irq].chip->unmask(irq); + if (desc->chip->unmask) + desc->chip->unmask(irq); - spin_unlock(&irq_desc[irq].lock); + spin_unlock(&desc->lock); if (break_affinity && set_affinity) printk("Broke affinity for irq %i\n", irq); diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index 165c5d9b0d1a..0744b49b4d12 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -143,9 +143,11 @@ void __init init_ISA_irqs(void) init_8259A(0); for (i = 0; i < nr_irqs; i++) { - irq_desc[i].status = IRQ_DISABLED; - irq_desc[i].action = NULL; - irq_desc[i].depth = 1; + struct irq_desc *desc = irq_to_desc(i); + + desc->status = IRQ_DISABLED; + desc->action = NULL; + desc->depth = 1; if (i < 16) { /* @@ -157,7 +159,7 @@ void __init init_ISA_irqs(void) /* * 'high' PCI IRQs filled in on demand */ - irq_desc[i].chip = &no_irq_chip; + desc->chip = &no_irq_chip; } } } diff --git a/arch/x86/kernel/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 61a97e616f70..9d85ab384435 100644 --- a/arch/x86/kernel/visws_quirks.c +++ b/arch/x86/kernel/visws_quirks.c @@ -484,10 +484,11 @@ static void disable_cobalt_irq(unsigned int irq) static unsigned int startup_cobalt_irq(unsigned int irq) { unsigned long flags; + struct irq_desc *desc = irq_to_desc(irq); spin_lock_irqsave(&cobalt_lock, flags); - if ((irq_desc[irq].status & (IRQ_DISABLED | IRQ_INPROGRESS | IRQ_WAITING))) - irq_desc[irq].status &= ~(IRQ_DISABLED | IRQ_INPROGRESS | IRQ_WAITING); + if ((desc->status & (IRQ_DISABLED | IRQ_INPROGRESS | IRQ_WAITING))) + desc->status &= ~(IRQ_DISABLED | IRQ_INPROGRESS | IRQ_WAITING); enable_cobalt_irq(irq); spin_unlock_irqrestore(&cobalt_lock, flags); return 0; @@ -506,9 +507,10 @@ static void ack_cobalt_irq(unsigned int irq) static void end_cobalt_irq(unsigned int irq) { unsigned long flags; + struct irq_desc *desc = irq_to_desc(irq); spin_lock_irqsave(&cobalt_lock, flags); - if (!(irq_desc[irq].status & (IRQ_DISABLED | IRQ_INPROGRESS))) + if (!(desc->status & (IRQ_DISABLED | IRQ_INPROGRESS))) enable_cobalt_irq(irq); spin_unlock_irqrestore(&cobalt_lock, flags); } @@ -626,7 +628,7 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) spin_unlock_irqrestore(&i8259A_lock, flags); - desc = irq_desc + realirq; + desc = irq_to_desc(realirq); /* * handle this 'virtual interrupt' as a Cobalt one now. @@ -662,27 +664,29 @@ void init_VISWS_APIC_irqs(void) int i; for (i = 0; i < CO_IRQ_APIC0 + CO_APIC_LAST + 1; i++) { - irq_desc[i].status = IRQ_DISABLED; - irq_desc[i].action = 0; - irq_desc[i].depth = 1; + struct irq_desc *desc = irq_to_desc(i); + + desc->status = IRQ_DISABLED; + desc->action = 0; + desc->depth = 1; if (i == 0) { - irq_desc[i].chip = &cobalt_irq_type; + desc->chip = &cobalt_irq_type; } else if (i == CO_IRQ_IDE0) { - irq_desc[i].chip = &cobalt_irq_type; + desc->chip = &cobalt_irq_type; } else if (i == CO_IRQ_IDE1) { - irq_desc[i].chip = &cobalt_irq_type; + desc->chip = &cobalt_irq_type; } else if (i == CO_IRQ_8259) { - irq_desc[i].chip = &piix4_master_irq_type; + desc->chip = &piix4_master_irq_type; } else if (i < CO_IRQ_APIC0) { - irq_desc[i].chip = &piix4_virtual_irq_type; + desc->chip = &piix4_virtual_irq_type; } else if (IS_CO_APIC(i)) { - irq_desc[i].chip = &cobalt_irq_type; + desc->chip = &cobalt_irq_type; } } -- cgit v1.2.3 From 3ac2de48ed3c998df7f366e039c97eedb27e7c3d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:06 -0700 Subject: x86: add irq_cfg in io_apic_64.c preallocate size is 32, and if it is not enough, irq_cfg will more via alloc_bootmem() or kzalloc(). (depending on how early we are in system setup) v2: fix typo about size of init_one_irq_cfg ... should use sizeof(struct irq_cfg) v3: according to Eric, change get_irq_cfg() to irq_cfg() v4: squash add irq_cfg_alloc in Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 209 ++++++++++++++++++++++++++++++++++--------- 1 file changed, 169 insertions(+), 40 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index cab5a25d81b1..858c37a31a2f 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -57,7 +57,11 @@ #define __apicdebuginit(type) static type __init +struct irq_cfg; + struct irq_cfg { + unsigned int irq; + struct irq_cfg *next; cpumask_t domain; cpumask_t old_domain; unsigned move_cleanup_count; @@ -67,34 +71,132 @@ struct irq_cfg { /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ static struct irq_cfg irq_cfg_legacy[] __initdata = { - [0] = { .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, + [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, + [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, + [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, + [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, + [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, + [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, + [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, + [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, + [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, + [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, + [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, + [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, + [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, + [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, + [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, + [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, }; -static struct irq_cfg *irq_cfg; +static struct irq_cfg irq_cfg_init = { .irq = -1U, }; +/* need to be biger than size of irq_cfg_legacy */ +static int nr_irq_cfg = 32; + +static int __init parse_nr_irq_cfg(char *arg) +{ + if (arg) { + nr_irq_cfg = simple_strtoul(arg, NULL, 0); + if (nr_irq_cfg < 32) + nr_irq_cfg = 32; + } + return 0; +} + +early_param("nr_irq_cfg", parse_nr_irq_cfg); + +static void init_one_irq_cfg(struct irq_cfg *cfg) +{ + memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); +} static void __init init_work(void *data) { struct dyn_array *da = data; + struct irq_cfg *cfg; + int i; - memcpy(*da->name, irq_cfg_legacy, sizeof(irq_cfg_legacy)); + cfg = *da->name; + + memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); + + i = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + for (; i < *da->nr; i++) + init_one_irq_cfg(&cfg[i]); + + for (i = 1; i < *da->nr; i++) + cfg[i-1].next = &cfg[i]; } -DEFINE_DYN_ARRAY(irq_cfg, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); +static struct irq_cfg *irq_cfgx; +DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); + +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + struct irq_cfg *cfg; + + BUG_ON(irq == -1U); + + cfg = &irq_cfgx[0]; + while (cfg) { + if (cfg->irq == irq) + return cfg; + + if (cfg->irq == -1U) + return NULL; + + cfg = cfg->next; + } + + return NULL; +} + +static struct irq_cfg *irq_cfg_alloc(unsigned int irq) +{ + struct irq_cfg *cfg, *cfg_pri; + int i; + int count = 0; + + BUG_ON(irq == -1U); + + cfg_pri = cfg = &irq_cfgx[0]; + while (cfg) { + if (cfg->irq == irq) + return cfg; + + if (cfg->irq == -1U) { + cfg->irq = irq; + return cfg; + } + cfg_pri = cfg; + cfg = cfg->next; + count++; + } + + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); + + if (after_bootmem) + cfg = kzalloc(sizeof(struct irq_cfg)*nr_irq_cfg, GFP_ATOMIC); + else + cfg = __alloc_bootmem_nopanic(sizeof(struct irq_cfg)*nr_irq_cfg, PAGE_SIZE, 0); + + if (!cfg) + panic("please boot with nr_irq_cfg= %d\n", count * 2); + + for (i = 0; i < nr_irq_cfg; i++) + init_one_irq_cfg(&cfg[i]); + + for (i = 1; i < nr_irq_cfg; i++) + cfg[i-1].next = &cfg[i]; + + cfg->irq = irq; + cfg_pri->next = cfg; + + return cfg; +} static int assign_irq_vector(int irq, cpumask_t mask); @@ -341,7 +443,7 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg = irq_cfg(irq); unsigned long flags; unsigned int dest; cpumask_t tmp; @@ -381,6 +483,8 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) struct irq_pin_list *entry = irq_2_pin + irq; BUG_ON(irq >= nr_irqs); + irq_cfg_alloc(irq); + while (entry->next) entry = irq_2_pin + entry->next; @@ -819,7 +923,7 @@ static int __assign_irq_vector(int irq, cpumask_t mask) struct irq_cfg *cfg; BUG_ON((unsigned)irq >= nr_irqs); - cfg = &irq_cfg[irq]; + cfg = irq_cfg(irq); /* Only try and allocate irqs on cpus that are present */ cpus_and(mask, mask, cpu_online_map); @@ -893,7 +997,7 @@ static void __clear_irq_vector(int irq) int cpu, vector; BUG_ON((unsigned)irq >= nr_irqs); - cfg = &irq_cfg[irq]; + cfg = irq_cfg(irq); BUG_ON(!cfg->vector); vector = cfg->vector; @@ -913,17 +1017,23 @@ void __setup_vector_irq(int cpu) /* Mark the inuse vectors */ for (irq = 0; irq < nr_irqs; ++irq) { - if (!cpu_isset(cpu, irq_cfg[irq].domain)) + struct irq_cfg *cfg = irq_cfg(irq); + + if (!cpu_isset(cpu, cfg->domain)) continue; - vector = irq_cfg[irq].vector; + vector = cfg->vector; per_cpu(vector_irq, cpu)[vector] = irq; } /* Mark the free vectors */ for (vector = 0; vector < NR_VECTORS; ++vector) { + struct irq_cfg *cfg; + irq = per_cpu(vector_irq, cpu)[vector]; if (irq < 0) continue; - if (!cpu_isset(cpu, irq_cfg[irq].domain)) + + cfg = irq_cfg(irq); + if (!cpu_isset(cpu, cfg->domain)) per_cpu(vector_irq, cpu)[vector] = -1; } } @@ -1029,13 +1139,15 @@ static int setup_ioapic_entry(int apic, int irq, static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, int trigger, int polarity) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; struct IO_APIC_route_entry entry; cpumask_t mask; if (!IO_APIC_IRQ(irq)) return; + cfg = irq_cfg(irq); + mask = TARGET_CPUS; if (assign_irq_vector(irq, mask)) return; @@ -1553,7 +1665,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) static int ioapic_retrigger_irq(unsigned int irq) { - struct irq_cfg *cfg = &irq_cfg[irq]; + struct irq_cfg *cfg = irq_cfg(irq); unsigned long flags; spin_lock_irqsave(&vector_lock, flags); @@ -1600,7 +1712,7 @@ static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); */ static void migrate_ioapic_irq(int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; struct irq_desc *desc; cpumask_t tmp, cleanup_mask; struct irte irte; @@ -1618,6 +1730,7 @@ static void migrate_ioapic_irq(int irq, cpumask_t mask) if (assign_irq_vector(irq, mask)) return; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -1735,7 +1848,7 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) continue; desc = irq_to_desc(irq); - cfg = irq_cfg + irq; + cfg = irq_cfg(irq); spin_lock(&desc->lock); if (!cfg->move_cleanup_count) goto unlock; @@ -1754,7 +1867,7 @@ unlock: static void irq_complete_move(unsigned int irq) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg = irq_cfg(irq); unsigned vector, me; if (likely(!cfg->move_in_progress)) @@ -1891,7 +2004,10 @@ static inline void init_IO_APIC_traps(void) * 0x80, because int 0x80 is hm, kind of importantish. ;) */ for (irq = 0; irq < nr_irqs ; irq++) { - if (IO_APIC_IRQ(irq) && !irq_cfg[irq].vector) { + struct irq_cfg *cfg; + + cfg = irq_cfg(irq); + if (IO_APIC_IRQ(irq) && !cfg->vector) { /* * Hmm.. We don't have an entry for this, * so default to an old-fashioned 8259 @@ -2028,7 +2144,7 @@ static inline void __init unlock_ExtINT_logic(void) */ static inline void __init check_timer(void) { - struct irq_cfg *cfg = irq_cfg + 0; + struct irq_cfg *cfg = irq_cfg(0); int apic1, pin1, apic2, pin2; unsigned long flags; int no_pin1 = 0; @@ -2306,14 +2422,19 @@ int create_irq(void) int irq; int new; unsigned long flags; + struct irq_cfg *cfg_new; irq = -ENOSPC; spin_lock_irqsave(&vector_lock, flags); for (new = (nr_irqs - 1); new >= 0; new--) { if (platform_legacy_irq(new)) continue; - if (irq_cfg[new].vector != 0) + cfg_new = irq_cfg(new); + if (cfg_new && cfg_new->vector != 0) continue; + /* check if need to create one */ + if (!cfg_new) + cfg_new = irq_cfg_alloc(new); if (__assign_irq_vector(new, TARGET_CPUS) == 0) irq = new; break; @@ -2346,7 +2467,7 @@ void destroy_irq(unsigned int irq) #ifdef CONFIG_PCI_MSI static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; int err; unsigned dest; cpumask_t tmp; @@ -2356,6 +2477,7 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms if (err) return err; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -2413,7 +2535,7 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms #ifdef CONFIG_SMP static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; @@ -2426,6 +2548,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) if (assign_irq_vector(irq, mask)) return; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -2448,7 +2571,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) */ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp, cleanup_mask; struct irte irte; @@ -2464,6 +2587,7 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) if (assign_irq_vector(irq, mask)) return; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -2670,7 +2794,7 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_SMP static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; @@ -2683,6 +2807,7 @@ static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) if (assign_irq_vector(irq, mask)) return; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -2749,7 +2874,7 @@ static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp; struct irq_desc *desc; @@ -2761,6 +2886,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) if (assign_irq_vector(irq, mask)) return; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -2783,7 +2909,7 @@ static struct irq_chip ht_irq_chip = { int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) { - struct irq_cfg *cfg = irq_cfg + irq; + struct irq_cfg *cfg; int err; cpumask_t tmp; @@ -2793,6 +2919,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) struct ht_irq_msg msg; unsigned dest; + cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -2891,6 +3018,7 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) void __init setup_ioapic_dest(void) { int pin, ioapic, irq, irq_entry; + struct irq_cfg *cfg; if (skip_ioapic_setup == 1) return; @@ -2906,7 +3034,8 @@ void __init setup_ioapic_dest(void) * when you have too many devices, because at that time only boot * cpu is online. */ - if (!irq_cfg[irq].vector) + cfg = irq_cfg(irq); + if (!cfg->vector) setup_IO_APIC_irq(ioapic, pin, irq, irq_trigger(irq_entry), irq_polarity(irq_entry)); -- cgit v1.2.3 From e5a53714acfc7b5f868d07d27c5f02cb00b118db Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:07 -0700 Subject: x86: put irq_2_pin pointer into irq_cfg preallocate 32 irq_2_pin entries, and use get_one_free_irq_2_pin() to get one more and link to irq_cfg if needed. so don't waste one where no irq is enabled. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 161 ++++++++++++++++++++++++++++++++----------- 1 file changed, 120 insertions(+), 41 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 858c37a31a2f..51ef7eb75f2e 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -58,10 +58,11 @@ #define __apicdebuginit(type) static type __init struct irq_cfg; - +struct irq_pin_list; struct irq_cfg { unsigned int irq; struct irq_cfg *next; + struct irq_pin_list *irq_2_pin; cpumask_t domain; cpumask_t old_domain; unsigned move_cleanup_count; @@ -252,13 +253,66 @@ int pin_map_size; * between pins and IRQs. */ -static struct irq_pin_list { - short apic, pin; - int next; -} *irq_2_pin; +struct irq_pin_list { + int apic, pin; + struct irq_pin_list *next; +}; + +static struct irq_pin_list *irq_2_pin_head; +/* fill one page ? */ +static int nr_irq_2_pin = 0x100; +static struct irq_pin_list *irq_2_pin_ptr; +static void __init irq_2_pin_init_work(void *data) +{ + struct dyn_array *da = data; + struct irq_pin_list *pin; + int i; + + pin = *da->name; + + for (i = 1; i < *da->nr; i++) + pin[i-1].next = &pin[i]; + + irq_2_pin_ptr = &pin[0]; +} +DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); + +static struct irq_pin_list *get_one_free_irq_2_pin(void) +{ + struct irq_pin_list *pin; + int i; + + pin = irq_2_pin_ptr; + + if (pin) { + irq_2_pin_ptr = pin->next; + pin->next = NULL; + return pin; + } + + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); + + if (after_bootmem) + pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, + GFP_ATOMIC); + else + pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * + nr_irq_2_pin, PAGE_SIZE, 0); + + if (!pin) + panic("can not get more irq_2_pin\n"); -DEFINE_DYN_ARRAY(irq_2_pin, sizeof(struct irq_pin_list), pin_map_size, sizeof(struct irq_pin_list), NULL); + for (i = 1; i < nr_irq_2_pin; i++) + pin[i-1].next = &pin[i]; + irq_2_pin_ptr = pin->next; + pin->next = NULL; + + return pin; +} struct io_apic { unsigned int index; @@ -300,16 +354,17 @@ static bool io_apic_level_ack_pending(unsigned int irq) { struct irq_pin_list *entry; unsigned long flags; + struct irq_cfg *cfg = irq_cfg(irq); spin_lock_irqsave(&ioapic_lock, flags); - entry = irq_2_pin + irq; + entry = cfg->irq_2_pin; for (;;) { unsigned int reg; int pin; - pin = entry->pin; - if (pin == -1) + if (!entry) break; + pin = entry->pin; reg = io_apic_read(entry->apic, 0x10 + pin*2); /* Is the remote IRR bit set? */ if (reg & IO_APIC_REDIR_REMOTE_IRR) { @@ -318,7 +373,7 @@ static bool io_apic_level_ack_pending(unsigned int irq) } if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } spin_unlock_irqrestore(&ioapic_lock, flags); @@ -339,21 +394,24 @@ static inline void io_apic_sync(unsigned int apic) \ { \ int pin; \ - struct irq_pin_list *entry = irq_2_pin + irq; \ + struct irq_cfg *cfg; \ + struct irq_pin_list *entry; \ \ BUG_ON(irq >= nr_irqs); \ + cfg = irq_cfg(irq); \ + entry = cfg->irq_2_pin; \ for (;;) { \ unsigned int reg; \ - pin = entry->pin; \ - if (pin == -1) \ + if (!entry) \ break; \ + pin = entry->pin; \ reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ reg ACTION; \ io_apic_modify(entry->apic, reg); \ FINAL; \ if (!entry->next) \ break; \ - entry = irq_2_pin + entry->next; \ + entry = entry->next; \ } \ } @@ -416,15 +474,20 @@ static void ioapic_mask_entry(int apic, int pin) static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) { int apic, pin; - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg; + struct irq_pin_list *entry; BUG_ON(irq >= nr_irqs); + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; for (;;) { unsigned int reg; + + if (!entry) + break; + apic = entry->apic; pin = entry->pin; - if (pin == -1) - break; /* * With interrupt-remapping, destination information comes * from interrupt-remapping table entry. @@ -437,7 +500,7 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) io_apic_modify(apic, reg); if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } } @@ -480,22 +543,35 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) int first_free_entry; static void add_pin_to_irq(unsigned int irq, int apic, int pin) { - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg; + struct irq_pin_list *entry; BUG_ON(irq >= nr_irqs); - irq_cfg_alloc(irq); + /* first time to refer irq_cfg, so with new */ + cfg = irq_cfg_alloc(irq); + entry = cfg->irq_2_pin; + if (!entry) { + entry = get_one_free_irq_2_pin(); + cfg->irq_2_pin = entry; + entry->apic = apic; + entry->pin = pin; + printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); + return; + } - while (entry->next) - entry = irq_2_pin + entry->next; + while (entry->next) { + /* not again, please */ + if (entry->apic == apic && entry->pin == pin) + return; - if (entry->pin != -1) { - entry->next = first_free_entry; - entry = irq_2_pin + entry->next; - if (++first_free_entry >= pin_map_size) - panic("io_apic.c: ran out of irq_2_pin entries!"); + entry = entry->next; } + + entry->next = get_one_free_irq_2_pin(); + entry = entry->next; entry->apic = apic; entry->pin = pin; + printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); } /* @@ -505,17 +581,24 @@ static void __init replace_pin_at_irq(unsigned int irq, int oldapic, int oldpin, int newapic, int newpin) { - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg = irq_cfg(irq); + struct irq_pin_list *entry = cfg->irq_2_pin; + int replaced = 0; - while (1) { + while (entry) { if (entry->apic == oldapic && entry->pin == oldpin) { entry->apic = newapic; entry->pin = newpin; - } - if (!entry->next) + replaced = 1; + /* every one is different, right? */ break; - entry = irq_2_pin + entry->next; + } + entry = entry->next; } + + /* why? call replace before add? */ + if (!replaced) + add_pin_to_irq(irq, newapic, newpin); } @@ -1326,15 +1409,16 @@ __apicdebuginit(void) print_IO_APIC(void) } printk(KERN_DEBUG "IRQ to pin mappings:\n"); for (i = 0; i < nr_irqs; i++) { - struct irq_pin_list *entry = irq_2_pin + i; - if (entry->pin < 0) + struct irq_cfg *cfg = irq_cfg(i); + struct irq_pin_list *entry = cfg->irq_2_pin; + if (!entry) continue; printk(KERN_DEBUG "IRQ%d ", i); for (;;) { printk("-> %d:%d", entry->apic, entry->pin); if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } printk("\n"); } @@ -1495,14 +1579,9 @@ void __init enable_IO_APIC(void) { union IO_APIC_reg_01 reg_01; int i8259_apic, i8259_pin; - int i, apic; + int apic; unsigned long flags; - for (i = 0; i < pin_map_size; i++) { - irq_2_pin[i].pin = -1; - irq_2_pin[i].next = 0; - } - /* * The number of IO-APIC IRQ registers (== #pins): */ -- cgit v1.2.3 From 7f95ec9e4c12fd067febfd57532da1166d75d858 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:09 -0700 Subject: x86: move kstat_irqs from kstat to irq_desc based on Eric's patch ... together mold it with dyn_array for irq_desc, will allcate kstat_irqs for nr_irq_desc alltogether if needed. -- at that point nr_cpus is known already. v2: make sure system without generic_hardirqs works they don't have irq_desc v3: fix merging v4: [mingo@elte.hu] fix typo [ mingo@elte.hu ] irq: build fix fix: arch/x86/xen/spinlock.c: In function 'xen_spin_lock_slow': arch/x86/xen/spinlock.c:90: error: 'struct kernel_stat' has no member named 'irqs' Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 2 +- arch/x86/kernel/irq_32.c | 4 ++-- arch/x86/kernel/irq_64.c | 4 ++-- arch/x86/kernel/visws_quirks.c | 2 +- 4 files changed, 6 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index c2160cfdec9b..204884b1415a 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -526,7 +526,7 @@ static void do_irq_balance(void) if (package_index == i) IRQ_DELTA(package_index, j) = 0; /* Determine the total count per processor per IRQ */ - value_now = (unsigned long) kstat_cpu(i).irqs[j]; + value_now = (unsigned long) kstat_irqs_cpu(j, i); /* Determine the activity per processor per IRQ */ delta = value_now - LAST_CPU_IRQ(i, j); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index ede513be517d..576c5df6cad8 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -280,7 +280,7 @@ int show_interrupts(struct seq_file *p, void *v) any_count = kstat_irqs(i); #else for_each_online_cpu(j) - any_count |= kstat_cpu(j).irqs[i]; + any_count |= kstat_irqs_cpu(i, j); #endif action = desc->action; if (!action && !any_count) @@ -290,7 +290,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_printf(p, "%10u ", kstat_irqs(i)); #else for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]); + seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); #endif seq_printf(p, " %8s", desc->chip->name); seq_printf(p, "-%-8s", desc->name); diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 738eb65a924e..4a0a4eb44dcb 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -90,7 +90,7 @@ int show_interrupts(struct seq_file *p, void *v) any_count = kstat_irqs(i); #else for_each_online_cpu(j) - any_count |= kstat_cpu(j).irqs[i]; + any_count |= kstat_irqs_cpu(i, j); #endif action = desc->action; if (!action && !any_count) @@ -100,7 +100,7 @@ int show_interrupts(struct seq_file *p, void *v) seq_printf(p, "%10u ", kstat_irqs(i)); #else for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_cpu(j).irqs[i]); + seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); #endif seq_printf(p, " %8s", desc->chip->name); seq_printf(p, "-%-8s", desc->name); diff --git a/arch/x86/kernel/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 9d85ab384435..817aa55a1209 100644 --- a/arch/x86/kernel/visws_quirks.c +++ b/arch/x86/kernel/visws_quirks.c @@ -633,7 +633,7 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) /* * handle this 'virtual interrupt' as a Cobalt one now. */ - kstat_cpu(smp_processor_id()).irqs[realirq]++; + kstat_irqs_this_cpu(desc)++; if (likely(desc->action != NULL)) handle_IRQ_event(realirq, desc->action); -- cgit v1.2.3 From 2c6927a38f65b53b62f86158fba29a068c4e8b6a Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:11 -0700 Subject: irq: replace loop with nr_irqs with for_each_irq_desc There are a handful of loops that go from 0 to nr_irqs and use get_irq_desc() on them. These would allocate all the irq_desc entries, regardless of the need for them. Use the smarter for_each_irq_desc() iterator that will only iterate over the present ones. v2: make sure arch without GENERIC_HARDIRQS work too Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 6 +++--- arch/x86/kernel/irq_64.c | 5 ++--- arch/x86/kernel/irqinit_64.c | 17 +++++------------ 3 files changed, 10 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 51ef7eb75f2e..708be9724daf 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -1871,10 +1871,10 @@ unmask: static void ir_irq_migration(struct work_struct *work) { - int irq; + unsigned int irq; + struct irq_desc *desc; - for (irq = 0; irq < nr_irqs; irq++) { - struct irq_desc *desc = irq_to_desc(irq); + for_each_irq_desc(irq, desc) { if (desc->status & IRQ_MOVE_PENDING) { unsigned long flags; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 4a0a4eb44dcb..b3cf55e325f5 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -224,17 +224,16 @@ void fixup_irqs(cpumask_t map) { unsigned int irq; static int warned; + struct irq_desc *desc; - for (irq = 0; irq < nr_irqs; irq++) { + for_each_irq_desc(irq, desc) { cpumask_t mask; int break_affinity = 0; int set_affinity = 1; - struct irq_desc *desc; if (irq == 2) continue; - desc = irq_to_desc(irq); /* interrupt's are disabled at this point */ spin_lock(&desc->lock); diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index 0744b49b4d12..cd9f42d028d9 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -142,25 +142,18 @@ void __init init_ISA_irqs(void) init_bsp_APIC(); init_8259A(0); - for (i = 0; i < nr_irqs; i++) { + for (i = 0; i < 16; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; desc->action = NULL; desc->depth = 1; - if (i < 16) { - /* - * 16 old-style INTA-cycle interrupts: - */ - set_irq_chip_and_handler_name(i, &i8259A_chip, + /* + * 16 old-style INTA-cycle interrupts: + */ + set_irq_chip_and_handler_name(i, &i8259A_chip, handle_level_irq, "XT"); - } else { - /* - * 'high' PCI IRQs filled in on demand - */ - desc->chip = &no_irq_chip; - } } } -- cgit v1.2.3 From 46b8214d12c274bd4265aae482ab7ffe69d94818 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:13 -0700 Subject: x86, ioapic: replace loop with nr_irqs with for_each_irq_icfg Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 708be9724daf..60d60061659c 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -129,6 +129,9 @@ static void __init init_work(void *data) cfg[i-1].next = &cfg[i]; } +#define for_each_irq_cfg(cfg) \ + for (cfg = irq_cfgx; cfg && cfg->irq != -1U; cfg = cfg->next) + static struct irq_cfg *irq_cfgx; DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); @@ -1097,20 +1100,18 @@ void __setup_vector_irq(int cpu) /* Initialize vector_irq on a new cpu */ /* This function must be called with vector_lock held */ int irq, vector; + struct irq_cfg *cfg; /* Mark the inuse vectors */ - for (irq = 0; irq < nr_irqs; ++irq) { - struct irq_cfg *cfg = irq_cfg(irq); - + for_each_irq_cfg(cfg) { if (!cpu_isset(cpu, cfg->domain)) continue; vector = cfg->vector; + irq = cfg->irq; per_cpu(vector_irq, cpu)[vector] = irq; } /* Mark the free vectors */ for (vector = 0; vector < NR_VECTORS; ++vector) { - struct irq_cfg *cfg; - irq = per_cpu(vector_irq, cpu)[vector]; if (irq < 0) continue; @@ -1340,6 +1341,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_01 reg_01; union IO_APIC_reg_02 reg_02; unsigned long flags; + struct irq_cfg *cfg; if (apic_verbosity == APIC_QUIET) return; @@ -1408,12 +1410,11 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for (i = 0; i < nr_irqs; i++) { - struct irq_cfg *cfg = irq_cfg(i); + for_each_irq_cfg(cfg) { struct irq_pin_list *entry = cfg->irq_2_pin; if (!entry) continue; - printk(KERN_DEBUG "IRQ%d ", i); + printk(KERN_DEBUG "IRQ%d ", cfg->irq); for (;;) { printk("-> %d:%d", entry->apic, entry->pin); if (!entry->next) @@ -2070,6 +2071,7 @@ static inline void init_IO_APIC_traps(void) { int irq; struct irq_desc *desc; + struct irq_cfg *cfg; /* * NOTE! The local APIC isn't very good at handling @@ -2082,10 +2084,8 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for (irq = 0; irq < nr_irqs ; irq++) { - struct irq_cfg *cfg; - - cfg = irq_cfg(irq); + for_each_irq_cfg(cfg) { + irq = cfg->irq; if (IO_APIC_IRQ(irq) && !cfg->vector) { /* * Hmm.. We don't have an entry for this, -- cgit v1.2.3 From 7d94f7ca401dd7f445fda9a971a48aa5427b3e55 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:14 -0700 Subject: irq: remove >= nr_irqs checking with config_have_sparse_irq remove irq limit checks - nr_irqs is dynamic and we expand anytime. v2: fix checking about result irq_cfg_without_new, so could use msi again v3: use irq_desc_without_new to check irq is valid Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 9 --------- arch/x86/kernel/irq_64.c | 2 +- 2 files changed, 1 insertion(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 60d60061659c..1b8cccb5ba25 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -400,7 +400,6 @@ static inline void io_apic_sync(unsigned int apic) struct irq_cfg *cfg; \ struct irq_pin_list *entry; \ \ - BUG_ON(irq >= nr_irqs); \ cfg = irq_cfg(irq); \ entry = cfg->irq_2_pin; \ for (;;) { \ @@ -480,7 +479,6 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) struct irq_cfg *cfg; struct irq_pin_list *entry; - BUG_ON(irq >= nr_irqs); cfg = irq_cfg(irq); entry = cfg->irq_2_pin; for (;;) { @@ -549,7 +547,6 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) struct irq_cfg *cfg; struct irq_pin_list *entry; - BUG_ON(irq >= nr_irqs); /* first time to refer irq_cfg, so with new */ cfg = irq_cfg_alloc(irq); entry = cfg->irq_2_pin; @@ -841,7 +838,6 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) best_guess = irq; } } - BUG_ON(best_guess >= nr_irqs); return best_guess; } @@ -973,7 +969,6 @@ static int pin_2_irq(int idx, int apic, int pin) irq += nr_ioapic_registers[i++]; irq += pin; } - BUG_ON(irq >= nr_irqs); return irq; } @@ -1008,7 +1003,6 @@ static int __assign_irq_vector(int irq, cpumask_t mask) int cpu; struct irq_cfg *cfg; - BUG_ON((unsigned)irq >= nr_irqs); cfg = irq_cfg(irq); /* Only try and allocate irqs on cpus that are present */ @@ -1082,7 +1076,6 @@ static void __clear_irq_vector(int irq) cpumask_t mask; int cpu, vector; - BUG_ON((unsigned)irq >= nr_irqs); cfg = irq_cfg(irq); BUG_ON(!cfg->vector); @@ -1924,8 +1917,6 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) struct irq_desc *desc; struct irq_cfg *cfg; irq = __get_cpu_var(vector_irq)[vector]; - if (irq >= nr_irqs) - continue; desc = irq_to_desc(irq); cfg = irq_cfg(irq); diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index b3cf55e325f5..a3e36336d914 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -202,7 +202,7 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) stack_overflow_check(regs); #endif - if (likely(irq < nr_irqs)) + if (likely(__irq_to_desc(irq))) generic_handle_irq(irq); else { if (!disable_apic) -- cgit v1.2.3 From 46926b67fc663d357a1a8174328998a9e49da0b8 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:15 -0700 Subject: generic: add irq_desc in function in parameter So we could remove some duplicated calling to irq_desc v2: make sure irq_desc in init/main.c is not used without generic_hardirqs Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_64.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index a3e36336d914..f58b995b30ee 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -189,6 +189,7 @@ u64 arch_irq_stat(void) asmlinkage unsigned int do_IRQ(struct pt_regs *regs) { struct pt_regs *old_regs = set_irq_regs(regs); + struct irq_desc *desc; /* high bit used in ret_from_ code */ unsigned vector = ~regs->orig_ax; @@ -202,8 +203,9 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) stack_overflow_check(regs); #endif - if (likely(__irq_to_desc(irq))) - generic_handle_irq(irq); + desc = __irq_to_desc(irq); + if (likely(desc)) + generic_handle_irq_desc(irq, desc); else { if (!disable_apic) ack_APIC_irq(); -- cgit v1.2.3 From 1d5f6b36c4736af1dac396d6267eb53dcc8c0021 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:16 -0700 Subject: x86: check with without_new in show_interrupts so we don't get new one that we don't need it. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_64.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index f58b995b30ee..f337f87c1e16 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -83,7 +83,10 @@ int show_interrupts(struct seq_file *p, void *v) if (i < nr_irqs) { unsigned any_count = 0; - struct irq_desc *desc = irq_to_desc(i); + struct irq_desc *desc = __irq_to_desc(i); + + if (!desc) + return 0; spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP -- cgit v1.2.3 From cb5bc83225a86ca53bbb889ed8439e4fd6cf44ac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:17 -0700 Subject: x86_64: rename irq_desc/irq_desc_alloc change names: irq_desc() ==> irq_desc_alloc __irq_desc() ==> irq_desc Also split a few of the uses in lowlevel x86 code. v2: need to check if desc is null in smp_irq_move_cleanup Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 10 +++++++++- arch/x86/kernel/irq_64.c | 4 ++-- arch/x86/kernel/irqinit_64.c | 3 ++- 3 files changed, 13 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 1b8cccb5ba25..a054db9ef190 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -1124,7 +1124,12 @@ static void ioapic_register_intr(int irq, unsigned long trigger) { struct irq_desc *desc; - desc = irq_to_desc(irq); + /* first time to use this irq_desc */ + if (irq < 16) + desc = irq_to_desc(irq); + else + desc = irq_to_desc_alloc(irq); + if (trigger) desc->status |= IRQ_LEVEL; else @@ -1919,6 +1924,9 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) irq = __get_cpu_var(vector_irq)[vector]; desc = irq_to_desc(irq); + if (!desc) + continue; + cfg = irq_cfg(irq); spin_lock(&desc->lock); if (!cfg->move_cleanup_count) diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index f337f87c1e16..5d5976e0311a 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -83,7 +83,7 @@ int show_interrupts(struct seq_file *p, void *v) if (i < nr_irqs) { unsigned any_count = 0; - struct irq_desc *desc = __irq_to_desc(i); + struct irq_desc *desc = irq_to_desc(i); if (!desc) return 0; @@ -206,7 +206,7 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) stack_overflow_check(regs); #endif - desc = __irq_to_desc(irq); + desc = irq_to_desc(irq); if (likely(desc)) generic_handle_irq_desc(irq, desc); else { diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index cd9f42d028d9..d17fbc26d96f 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -143,7 +143,8 @@ void __init init_ISA_irqs(void) init_8259A(0); for (i = 0; i < 16; i++) { - struct irq_desc *desc = irq_to_desc(i); + /* first time call this irq_desc */ + struct irq_desc *desc = irq_to_desc_alloc(i); desc->status = IRQ_DISABLED; desc->action = NULL; -- cgit v1.2.3 From a2f9f43858db64cb8b45c4f6746d7a52b80d4dcb Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:19 -0700 Subject: x86_64: separate irq_cfgx from irq_cfgx_free so later don't need to compare with -1U Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 92 ++++++++++++++++++++++++++++---------------- 1 file changed, 59 insertions(+), 33 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index a054db9ef190..8ab7ae01773f 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -111,44 +111,44 @@ static void init_one_irq_cfg(struct irq_cfg *cfg) memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); } +static struct irq_cfg *irq_cfgx; +static struct irq_cfg *irq_cfgx_free; static void __init init_work(void *data) { struct dyn_array *da = data; struct irq_cfg *cfg; + int legacy_count; int i; cfg = *da->name; memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - i = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); - for (; i < *da->nr; i++) + legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + for (i = legacy_count; i < *da->nr; i++) init_one_irq_cfg(&cfg[i]); for (i = 1; i < *da->nr; i++) cfg[i-1].next = &cfg[i]; + + irq_cfgx_free = &irq_cfgx[legacy_count]; + irq_cfgx[legacy_count - 1].next = NULL; } #define for_each_irq_cfg(cfg) \ - for (cfg = irq_cfgx; cfg && cfg->irq != -1U; cfg = cfg->next) + for (cfg = irq_cfgx; cfg; cfg = cfg->next) -static struct irq_cfg *irq_cfgx; DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); static struct irq_cfg *irq_cfg(unsigned int irq) { struct irq_cfg *cfg; - BUG_ON(irq == -1U); - - cfg = &irq_cfgx[0]; + cfg = irq_cfgx; while (cfg) { if (cfg->irq == irq) return cfg; - if (cfg->irq == -1U) - return NULL; - cfg = cfg->next; } @@ -161,44 +161,70 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) int i; int count = 0; - BUG_ON(irq == -1U); - - cfg_pri = cfg = &irq_cfgx[0]; + cfg_pri = cfg = irq_cfgx; while (cfg) { if (cfg->irq == irq) return cfg; - if (cfg->irq == -1U) { - cfg->irq = irq; - return cfg; - } cfg_pri = cfg; cfg = cfg->next; count++; } - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); + if (!irq_cfgx_free) { + unsigned long phys; + unsigned long total_bytes; + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); - if (after_bootmem) - cfg = kzalloc(sizeof(struct irq_cfg)*nr_irq_cfg, GFP_ATOMIC); - else - cfg = __alloc_bootmem_nopanic(sizeof(struct irq_cfg)*nr_irq_cfg, PAGE_SIZE, 0); + total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; + if (after_bootmem) + cfg = kzalloc(total_bytes, GFP_ATOMIC); + else + cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); - if (!cfg) - panic("please boot with nr_irq_cfg= %d\n", count * 2); + if (!cfg) + panic("please boot with nr_irq_cfg= %d\n", count * 2); - for (i = 0; i < nr_irq_cfg; i++) - init_one_irq_cfg(&cfg[i]); + phys = __pa(cfg); + printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); - for (i = 1; i < nr_irq_cfg; i++) - cfg[i-1].next = &cfg[i]; + for (i = 0; i < nr_irq_cfg; i++) + init_one_irq_cfg(&cfg[i]); - cfg->irq = irq; - cfg_pri->next = cfg; + for (i = 1; i < nr_irq_cfg; i++) + cfg[i-1].next = &cfg[i]; + + irq_cfgx_free = cfg; + } + cfg = irq_cfgx_free; + irq_cfgx_free = irq_cfgx_free->next; + cfg->next = NULL; + if (cfg_pri) + cfg_pri->next = cfg; + else + irq_cfgx = cfg; + cfg->irq = irq; + printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); +#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG + { + /* dump the results */ + struct irq_cfg *cfg; + unsigned long phys; + unsigned long bytes = sizeof(struct irq_cfg); + + printk(KERN_DEBUG "=========================== %d\n", irq); + printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); + for_each_irq_cfg(cfg) { + phys = __pa(cfg); + printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); + } + printk(KERN_DEBUG "===========================\n"); + } +#endif return cfg; } -- cgit v1.2.3 From 52b17329d6d0a4824b89206803a430915031ff23 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:20 -0700 Subject: x86_64: make /proc/interrupts work with dyn irq_desc loop with irq_desc list Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_64.c | 31 ++++++++++++++++++++++++------- 1 file changed, 24 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 5d5976e0311a..7bd841a9c640 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -70,9 +70,27 @@ static inline void stack_overflow_check(struct pt_regs *regs) int show_interrupts(struct seq_file *p, void *v) { - int i = *(loff_t *) v, j; + int i, j; struct irqaction * action; unsigned long flags; + unsigned int entries; + struct irq_desc *desc; + int tail = 0; + +#ifdef CONFIG_HAVE_SPARSE_IRQ + desc = (struct irq_desc *)v; + entries = -1U; + i = desc->irq; + if (!desc->next) + tail = 1; +#else + entries = nr_irqs - 1; + i = *(loff_t *) v; + if (i == nr_irqs) + tail = 1; + else + desc = irq_to_desc(i); +#endif if (i == 0) { seq_printf(p, " "); @@ -81,12 +99,8 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i < nr_irqs) { + if (i <= entries) { unsigned any_count = 0; - struct irq_desc *desc = irq_to_desc(i); - - if (!desc) - return 0; spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP @@ -116,7 +130,9 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); skip: spin_unlock_irqrestore(&desc->lock, flags); - } else if (i == nr_irqs) { + } + + if (tail) { seq_printf(p, "NMI: "); for_each_online_cpu(j) seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count); @@ -155,6 +171,7 @@ skip: seq_printf(p, " Spurious interrupts\n"); seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); } + return 0; } -- cgit v1.2.3 From 6d50bc26836e16a9589e0b128d527c29e30d722a Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:22 -0700 Subject: x86: use 28 bits irq NR for pci msi/msix and ht also print out irq no in /proc/interrups and /proc/stat in hex, so could tell bus/dev/func. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 64 ++++++++++++++++++++++++++++++++++---------- arch/x86/kernel/irq_64.c | 2 +- 2 files changed, 51 insertions(+), 15 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 8ab7ae01773f..b0d4abc55a11 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -2520,17 +2520,21 @@ device_initcall(ioapic_init_sysfs); /* * Dynamic irq allocate and deallocation */ -int create_irq(void) +unsigned int create_irq_nr(unsigned int irq_want) { /* Allocate an unused irq */ - int irq; - int new; + unsigned int irq; + unsigned int new; unsigned long flags; struct irq_cfg *cfg_new; - irq = -ENOSPC; +#ifndef CONFIG_HAVE_SPARSE_IRQ + irq_want = nr_irqs - 1; +#endif + + irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = (nr_irqs - 1); new >= 0; new--) { + for (new = irq_want; new > 0; new--) { if (platform_legacy_irq(new)) continue; cfg_new = irq_cfg(new); @@ -2545,12 +2549,24 @@ int create_irq(void) } spin_unlock_irqrestore(&vector_lock, flags); - if (irq >= 0) { + if (irq > 0) { dynamic_irq_init(irq); } return irq; } +int create_irq(void) +{ + int irq; + + irq = create_irq_nr(nr_irqs - 1); + + if (irq == 0) + irq = -1; + + return irq; +} + void destroy_irq(unsigned int irq) { unsigned long flags; @@ -2803,13 +2819,29 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) return 0; } +static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) +{ + unsigned int irq; + + irq = dev->bus->number; + irq <<= 8; + irq |= dev->devfn; + irq <<= 12; + + return irq; +} + int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) { - int irq, ret; + unsigned int irq; + int ret; + unsigned int irq_want; - irq = create_irq(); - if (irq < 0) - return irq; + irq_want = build_irq_for_pci_dev(dev) + 0x100; + + irq = create_irq_nr(irq_want); + if (irq == 0) + return -1; #ifdef CONFIG_INTR_REMAP if (!intr_remapping_enabled) @@ -2836,18 +2868,22 @@ error: int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) { - int irq, ret, sub_handle; + unsigned int irq; + int ret, sub_handle; struct msi_desc *desc; + unsigned int irq_want; + #ifdef CONFIG_INTR_REMAP struct intel_iommu *iommu = 0; int index = 0; #endif + irq_want = build_irq_for_pci_dev(dev) + 0x100; sub_handle = 0; list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq(); - if (irq < 0) - return irq; + irq = create_irq_nr(irq_want--); + if (irq == 0) + return -1; #ifdef CONFIG_INTR_REMAP if (!intr_remapping_enabled) goto no_ir; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 7bd841a9c640..348a11168c2b 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -112,7 +112,7 @@ int show_interrupts(struct seq_file *p, void *v) action = desc->action; if (!action && !any_count) goto skip; - seq_printf(p, "%3d: ",i); + seq_printf(p, "%#x: ",i); #ifndef CONFIG_SMP seq_printf(p, "%10u ", kstat_irqs(i)); #else -- cgit v1.2.3 From 8b8e8c1bf7275eca859fe551dfa484134eaf013b Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:23 -0700 Subject: x86: remove irqbalance in kernel for 32 bit This has been deprecated for years, the user space irqbalanced utility works better with numa, has configurable policies, etc... Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 402 ------------------------------------------- arch/x86/kernel/quirks.c | 3 - 2 files changed, 405 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 204884b1415a..668edf226067 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -371,408 +371,6 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) spin_unlock_irqrestore(&ioapic_lock, flags); } -#if defined(CONFIG_IRQBALANCE) -# include /* kernel_thread() */ -# include /* kstat */ -# include /* kmalloc() */ -# include - -#define IRQBALANCE_CHECK_ARCH -999 -#define MAX_BALANCED_IRQ_INTERVAL (5*HZ) -#define MIN_BALANCED_IRQ_INTERVAL (HZ/2) -#define BALANCED_IRQ_MORE_DELTA (HZ/10) -#define BALANCED_IRQ_LESS_DELTA (HZ) - -static int irqbalance_disabled __read_mostly = IRQBALANCE_CHECK_ARCH; -static int physical_balance __read_mostly; -static long balanced_irq_interval __read_mostly = MAX_BALANCED_IRQ_INTERVAL; - -static struct irq_cpu_info { - unsigned long *last_irq; - unsigned long *irq_delta; - unsigned long irq; -} irq_cpu_data[NR_CPUS]; - -#define CPU_IRQ(cpu) (irq_cpu_data[cpu].irq) -#define LAST_CPU_IRQ(cpu, irq) (irq_cpu_data[cpu].last_irq[irq]) -#define IRQ_DELTA(cpu, irq) (irq_cpu_data[cpu].irq_delta[irq]) - -#define IDLE_ENOUGH(cpu,now) \ - (idle_cpu(cpu) && ((now) - per_cpu(irq_stat, (cpu)).idle_timestamp > 1)) - -#define IRQ_ALLOWED(cpu, allowed_mask) cpu_isset(cpu, allowed_mask) - -#define CPU_TO_PACKAGEINDEX(i) (first_cpu(per_cpu(cpu_sibling_map, i))) - -static cpumask_t balance_irq_affinity_init __initdata = CPU_MASK_ALL; - -static cpumask_t *balance_irq_affinity; - - -static void __init irq_affinity_init_work(void *data) -{ - struct dyn_array *da = data; - - int i; - struct balance_irq_affinity *affinity; - - affinity = *da->name; - - for (i = 0; i < *da->nr; i++) - memcpy(&affinity[i], &balance_irq_affinity_init, - sizeof(struct balance_irq_affinity)); - -} - -DEFINE_DYN_ARRAY(balance_irq_affinity, sizeof(struct balance_irq_affinity), nr_irqs, PAGE_SIZE, irq_affinity_init_work); - - -void set_balance_irq_affinity(unsigned int irq, cpumask_t mask) -{ - balance_irq_affinity[irq] = mask; -} - -static unsigned long move(int curr_cpu, cpumask_t allowed_mask, - unsigned long now, int direction) -{ - int search_idle = 1; - int cpu = curr_cpu; - - goto inside; - - do { - if (unlikely(cpu == curr_cpu)) - search_idle = 0; -inside: - if (direction == 1) { - cpu++; - if (cpu >= NR_CPUS) - cpu = 0; - } else { - cpu--; - if (cpu == -1) - cpu = NR_CPUS-1; - } - } while (!cpu_online(cpu) || !IRQ_ALLOWED(cpu, allowed_mask) || - (search_idle && !IDLE_ENOUGH(cpu, now))); - - return cpu; -} - -static inline void balance_irq(int cpu, int irq) -{ - unsigned long now = jiffies; - cpumask_t allowed_mask; - unsigned int new_cpu; - - if (irqbalance_disabled) - return; - - cpus_and(allowed_mask, cpu_online_map, balance_irq_affinity[irq]); - new_cpu = move(cpu, allowed_mask, now, 1); - if (cpu != new_cpu) - set_pending_irq(irq, cpumask_of_cpu(new_cpu)); -} - -static inline void rotate_irqs_among_cpus(unsigned long useful_load_threshold) -{ - int i, j; - struct irq_desc *desc; - - for_each_online_cpu(i) { - for (j = 0; j < nr_irqs; j++) { - desc = irq_to_desc(j); - if (!desc->action) - continue; - /* Is it a significant load ? */ - if (IRQ_DELTA(CPU_TO_PACKAGEINDEX(i), j) < - useful_load_threshold) - continue; - balance_irq(i, j); - } - } - balanced_irq_interval = max((long)MIN_BALANCED_IRQ_INTERVAL, - balanced_irq_interval - BALANCED_IRQ_LESS_DELTA); - return; -} - -static void do_irq_balance(void) -{ - int i, j; - unsigned long max_cpu_irq = 0, min_cpu_irq = (~0); - unsigned long move_this_load = 0; - int max_loaded = 0, min_loaded = 0; - int load; - unsigned long useful_load_threshold = balanced_irq_interval + 10; - int selected_irq; - int tmp_loaded, first_attempt = 1; - unsigned long tmp_cpu_irq; - unsigned long imbalance = 0; - cpumask_t allowed_mask, target_cpu_mask, tmp; - struct irq_desc *desc; - - for_each_possible_cpu(i) { - int package_index; - CPU_IRQ(i) = 0; - if (!cpu_online(i)) - continue; - package_index = CPU_TO_PACKAGEINDEX(i); - for (j = 0; j < nr_irqs; j++) { - unsigned long value_now, delta; - /* Is this an active IRQ or balancing disabled ? */ - desc = irq_to_desc(j); - if (!desc->action || irq_balancing_disabled(j)) - continue; - if (package_index == i) - IRQ_DELTA(package_index, j) = 0; - /* Determine the total count per processor per IRQ */ - value_now = (unsigned long) kstat_irqs_cpu(j, i); - - /* Determine the activity per processor per IRQ */ - delta = value_now - LAST_CPU_IRQ(i, j); - - /* Update last_cpu_irq[][] for the next time */ - LAST_CPU_IRQ(i, j) = value_now; - - /* Ignore IRQs whose rate is less than the clock */ - if (delta < useful_load_threshold) - continue; - /* update the load for the processor or package total */ - IRQ_DELTA(package_index, j) += delta; - - /* Keep track of the higher numbered sibling as well */ - if (i != package_index) - CPU_IRQ(i) += delta; - /* - * We have sibling A and sibling B in the package - * - * cpu_irq[A] = load for cpu A + load for cpu B - * cpu_irq[B] = load for cpu B - */ - CPU_IRQ(package_index) += delta; - } - } - /* Find the least loaded processor package */ - for_each_online_cpu(i) { - if (i != CPU_TO_PACKAGEINDEX(i)) - continue; - if (min_cpu_irq > CPU_IRQ(i)) { - min_cpu_irq = CPU_IRQ(i); - min_loaded = i; - } - } - max_cpu_irq = ULONG_MAX; - -tryanothercpu: - /* - * Look for heaviest loaded processor. - * We may come back to get the next heaviest loaded processor. - * Skip processors with trivial loads. - */ - tmp_cpu_irq = 0; - tmp_loaded = -1; - for_each_online_cpu(i) { - if (i != CPU_TO_PACKAGEINDEX(i)) - continue; - if (max_cpu_irq <= CPU_IRQ(i)) - continue; - if (tmp_cpu_irq < CPU_IRQ(i)) { - tmp_cpu_irq = CPU_IRQ(i); - tmp_loaded = i; - } - } - - if (tmp_loaded == -1) { - /* - * In the case of small number of heavy interrupt sources, - * loading some of the cpus too much. We use Ingo's original - * approach to rotate them around. - */ - if (!first_attempt && imbalance >= useful_load_threshold) { - rotate_irqs_among_cpus(useful_load_threshold); - return; - } - goto not_worth_the_effort; - } - - first_attempt = 0; /* heaviest search */ - max_cpu_irq = tmp_cpu_irq; /* load */ - max_loaded = tmp_loaded; /* processor */ - imbalance = (max_cpu_irq - min_cpu_irq) / 2; - - /* - * if imbalance is less than approx 10% of max load, then - * observe diminishing returns action. - quit - */ - if (imbalance < (max_cpu_irq >> 3)) - goto not_worth_the_effort; - -tryanotherirq: - /* if we select an IRQ to move that can't go where we want, then - * see if there is another one to try. - */ - move_this_load = 0; - selected_irq = -1; - for (j = 0; j < nr_irqs; j++) { - /* Is this an active IRQ? */ - desc = irq_to_desc(j); - if (!desc->action) - continue; - if (imbalance <= IRQ_DELTA(max_loaded, j)) - continue; - /* Try to find the IRQ that is closest to the imbalance - * without going over. - */ - if (move_this_load < IRQ_DELTA(max_loaded, j)) { - move_this_load = IRQ_DELTA(max_loaded, j); - selected_irq = j; - } - } - if (selected_irq == -1) - goto tryanothercpu; - - imbalance = move_this_load; - - /* For physical_balance case, we accumulated both load - * values in the one of the siblings cpu_irq[], - * to use the same code for physical and logical processors - * as much as possible. - * - * NOTE: the cpu_irq[] array holds the sum of the load for - * sibling A and sibling B in the slot for the lowest numbered - * sibling (A), _AND_ the load for sibling B in the slot for - * the higher numbered sibling. - * - * We seek the least loaded sibling by making the comparison - * (A+B)/2 vs B - */ - load = CPU_IRQ(min_loaded) >> 1; - for_each_cpu_mask(j, per_cpu(cpu_sibling_map, min_loaded)) { - if (load > CPU_IRQ(j)) { - /* This won't change cpu_sibling_map[min_loaded] */ - load = CPU_IRQ(j); - min_loaded = j; - } - } - - cpus_and(allowed_mask, - cpu_online_map, - balance_irq_affinity[selected_irq]); - target_cpu_mask = cpumask_of_cpu(min_loaded); - cpus_and(tmp, target_cpu_mask, allowed_mask); - - if (!cpus_empty(tmp)) { - /* mark for change destination */ - set_pending_irq(selected_irq, cpumask_of_cpu(min_loaded)); - - /* Since we made a change, come back sooner to - * check for more variation. - */ - balanced_irq_interval = max((long)MIN_BALANCED_IRQ_INTERVAL, - balanced_irq_interval - BALANCED_IRQ_LESS_DELTA); - return; - } - goto tryanotherirq; - -not_worth_the_effort: - /* - * if we did not find an IRQ to move, then adjust the time interval - * upward - */ - balanced_irq_interval = min((long)MAX_BALANCED_IRQ_INTERVAL, - balanced_irq_interval + BALANCED_IRQ_MORE_DELTA); - return; -} - -static int balanced_irq(void *unused) -{ - int i; - unsigned long prev_balance_time = jiffies; - long time_remaining = balanced_irq_interval; - struct irq_desc *desc; - - /* push everything to CPU 0 to give us a starting point. */ - for (i = 0 ; i < nr_irqs ; i++) { - desc = irq_to_desc(i); - desc->pending_mask = cpumask_of_cpu(0); - set_pending_irq(i, cpumask_of_cpu(0)); - } - - set_freezable(); - for ( ; ; ) { - time_remaining = schedule_timeout_interruptible(time_remaining); - try_to_freeze(); - if (time_after(jiffies, - prev_balance_time+balanced_irq_interval)) { - preempt_disable(); - do_irq_balance(); - prev_balance_time = jiffies; - time_remaining = balanced_irq_interval; - preempt_enable(); - } - } - return 0; -} - -static int __init balanced_irq_init(void) -{ - int i; - struct cpuinfo_x86 *c; - cpumask_t tmp; - - cpus_shift_right(tmp, cpu_online_map, 2); - c = &boot_cpu_data; - /* When not overwritten by the command line ask subarchitecture. */ - if (irqbalance_disabled == IRQBALANCE_CHECK_ARCH) - irqbalance_disabled = NO_BALANCE_IRQ; - if (irqbalance_disabled) - return 0; - - /* disable irqbalance completely if there is only one processor online */ - if (num_online_cpus() < 2) { - irqbalance_disabled = 1; - return 0; - } - /* - * Enable physical balance only if more than 1 physical processor - * is present - */ - if (smp_num_siblings > 1 && !cpus_empty(tmp)) - physical_balance = 1; - - for_each_online_cpu(i) { - irq_cpu_data[i].irq_delta = kzalloc(sizeof(unsigned long) * nr_irqs, GFP_KERNEL); - irq_cpu_data[i].last_irq = kzalloc(sizeof(unsigned long) * nr_irqs, GFP_KERNEL); - if (irq_cpu_data[i].irq_delta == NULL || irq_cpu_data[i].last_irq == NULL) { - printk(KERN_ERR "balanced_irq_init: out of memory"); - goto failed; - } - } - - printk(KERN_INFO "Starting balanced_irq\n"); - if (!IS_ERR(kthread_run(balanced_irq, NULL, "kirqd"))) - return 0; - printk(KERN_ERR "balanced_irq_init: failed to spawn balanced_irq"); -failed: - for_each_possible_cpu(i) { - kfree(irq_cpu_data[i].irq_delta); - irq_cpu_data[i].irq_delta = NULL; - kfree(irq_cpu_data[i].last_irq); - irq_cpu_data[i].last_irq = NULL; - } - return 0; -} - -int __devinit irqbalance_disable(char *str) -{ - irqbalance_disabled = 1; - return 1; -} - -__setup("noirqbalance", irqbalance_disable); - -late_initcall(balanced_irq_init); -#endif /* CONFIG_IRQBALANCE */ #endif /* CONFIG_SMP */ #ifndef CONFIG_SMP diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c index f6a11b9b1f98..67465ed89310 100644 --- a/arch/x86/kernel/quirks.c +++ b/arch/x86/kernel/quirks.c @@ -35,9 +35,6 @@ static void __devinit quirk_intel_irqbalance(struct pci_dev *dev) if (!(word & (1 << 13))) { dev_info(&dev->dev, "Intel E7520/7320/7525 detected; " "disabling irq balancing and affinity\n"); -#ifdef CONFIG_IRQBALANCE - irqbalance_disable(""); -#endif noirqdebug_setup(""); #ifdef CONFIG_PROC_FS no_irq_affinity = 1; -- cgit v1.2.3 From a1420f395d7721895c05ba3dedf150294d2c0e4d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:24 -0700 Subject: x86: add irq_cfg for 32bit it only contains vector ... Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 71 ++++++++++++++++++++++++++++---------------- 1 file changed, 46 insertions(+), 25 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 668edf226067..033ad953bee3 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -94,6 +94,43 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); static int disable_timer_pin_1 __initdata; +struct irq_cfg { + u8 vector; +}; + + +/* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ +static struct irq_cfg irq_cfg_legacy[] __initdata = { + [0] = { .vector = FIRST_DEVICE_VECTOR, }, +}; + +static void __init init_work(void *data) +{ + struct dyn_array *da = data; + struct irq_cfg *cfg; + int legacy_count; + int i; + + cfg = *da->name; + + legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + + BUG_ON(legacy_count > nr_irqs); + + memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); +} + +static struct irq_cfg *irq_cfgx; +DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); + +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + if (irq >= nr_irqs) + return NULL; + + return &irq_cfgx[irq]; +} + /* * Rough estimation of how many shared IRQs there are, can * be changed anytime. @@ -794,22 +831,6 @@ static inline int IO_APIC_irq_trigger(int irq) return 0; } -/* irq_vectors is indexed by the sum of all RTEs in all I/O APICs. */ -static u8 irq_vector_init_first __initdata = FIRST_DEVICE_VECTOR; -static u8 *irq_vector; - -static void __init irq_vector_init_work(void *data) -{ - struct dyn_array *da = data; - - u8 *irq_vec; - - irq_vec = *da->name; - - irq_vec[0] = irq_vector_init_first; -} - -DEFINE_DYN_ARRAY(irq_vector, sizeof(u8), nr_irqs, PAGE_SIZE, irq_vector_init_work); static int __assign_irq_vector(int irq) { @@ -818,8 +839,8 @@ static int __assign_irq_vector(int irq) BUG_ON((unsigned)irq >= nr_irqs); - if (irq_vector[irq] > 0) - return irq_vector[irq]; + if (irq_cfg(irq)->vector > 0) + return irq_cfg(irq)->vector; vector = current_vector; offset = current_offset; @@ -836,7 +857,7 @@ next: current_vector = vector; current_offset = offset; - irq_vector[irq] = vector; + irq_cfg(irq)->vector = vector; return vector; } @@ -1598,7 +1619,7 @@ static void ack_ioapic_quirk_irq(unsigned int irq) * operation to prevent an edge-triggered interrupt escaping meanwhile. * The idea is from Manfred Spraul. --macro */ - i = irq_vector[irq]; + i = irq_cfg(irq)->vector; v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); @@ -1615,7 +1636,7 @@ static void ack_ioapic_quirk_irq(unsigned int irq) static int ioapic_retrigger_irq(unsigned int irq) { - send_IPI_self(irq_vector[irq]); + send_IPI_self(irq_cfg(irq)->vector); return 1; } @@ -1651,7 +1672,7 @@ static inline void init_IO_APIC_traps(void) * 0x80, because int 0x80 is hm, kind of importantish. ;) */ for (irq = 0; irq < nr_irqs ; irq++) { - if (IO_APIC_IRQ(irq) && !irq_vector[irq]) { + if (IO_APIC_IRQ(irq) && !irq_cfg(irq)->vector) { /* * Hmm.. We don't have an entry for this, * so default to an old-fashioned 8259 @@ -2102,7 +2123,7 @@ int create_irq(void) for (new = (nr_irqs - 1); new >= 0; new--) { if (platform_legacy_irq(new)) continue; - if (irq_vector[new] != 0) + if (irq_cfg(new)->vector != 0) continue; vector = __assign_irq_vector(new); if (likely(vector > 0)) @@ -2125,8 +2146,8 @@ void destroy_irq(unsigned int irq) dynamic_irq_cleanup(irq); spin_lock_irqsave(&vector_lock, flags); - clear_bit(irq_vector[irq], used_vectors); - irq_vector[irq] = 0; + clear_bit(irq_cfg(irq)->vector, used_vectors); + irq_cfg(irq)->vector = 0; spin_unlock_irqrestore(&vector_lock, flags); } -- cgit v1.2.3 From da51a821314dd0ec7126f2bf9a62117146fbb6b9 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:25 -0700 Subject: x86: make 32bit use irq_cfg_alloc, etc Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 180 ++++++++++++++++++++++++++++++++++++++----- 1 file changed, 160 insertions(+), 20 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 033ad953bee3..5a83d7f5b147 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -94,41 +94,172 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); static int disable_timer_pin_1 __initdata; +struct irq_cfg; + struct irq_cfg { + unsigned int irq; + struct irq_cfg *next; u8 vector; }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ static struct irq_cfg irq_cfg_legacy[] __initdata = { - [0] = { .vector = FIRST_DEVICE_VECTOR, }, + [0] = { .irq = 0, .vector = IRQ0_VECTOR, }, + [1] = { .irq = 1, .vector = IRQ1_VECTOR, }, + [2] = { .irq = 2, .vector = IRQ2_VECTOR, }, + [3] = { .irq = 3, .vector = IRQ3_VECTOR, }, + [4] = { .irq = 4, .vector = IRQ4_VECTOR, }, + [5] = { .irq = 5, .vector = IRQ5_VECTOR, }, + [6] = { .irq = 6, .vector = IRQ6_VECTOR, }, + [7] = { .irq = 7, .vector = IRQ7_VECTOR, }, + [8] = { .irq = 8, .vector = IRQ8_VECTOR, }, + [9] = { .irq = 9, .vector = IRQ9_VECTOR, }, + [10] = { .irq = 10, .vector = IRQ10_VECTOR, }, + [11] = { .irq = 11, .vector = IRQ11_VECTOR, }, + [12] = { .irq = 12, .vector = IRQ12_VECTOR, }, + [13] = { .irq = 13, .vector = IRQ13_VECTOR, }, + [14] = { .irq = 14, .vector = IRQ14_VECTOR, }, + [15] = { .irq = 15, .vector = IRQ15_VECTOR, }, }; +static struct irq_cfg irq_cfg_init = { .irq = -1U, }; +/* need to be biger than size of irq_cfg_legacy */ +static int nr_irq_cfg = 32; + +static int __init parse_nr_irq_cfg(char *arg) +{ + if (arg) { + nr_irq_cfg = simple_strtoul(arg, NULL, 0); + if (nr_irq_cfg < 32) + nr_irq_cfg = 32; + } + return 0; +} + +early_param("nr_irq_cfg", parse_nr_irq_cfg); + +static void init_one_irq_cfg(struct irq_cfg *cfg) +{ + memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); +} + +static struct irq_cfg *irq_cfgx; +static struct irq_cfg *irq_cfgx_free; static void __init init_work(void *data) { - struct dyn_array *da = data; - struct irq_cfg *cfg; - int legacy_count; - int i; + struct dyn_array *da = data; + struct irq_cfg *cfg; + int legacy_count; + int i; + + cfg = *da->name; - cfg = *da->name; + memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + for (i = legacy_count; i < *da->nr; i++) + init_one_irq_cfg(&cfg[i]); - BUG_ON(legacy_count > nr_irqs); + for (i = 1; i < *da->nr; i++) + cfg[i-1].next = &cfg[i]; - memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); + irq_cfgx_free = &irq_cfgx[legacy_count]; + irq_cfgx[legacy_count - 1].next = NULL; } -static struct irq_cfg *irq_cfgx; -DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); +#define for_each_irq_cfg(cfg) \ + for (cfg = irq_cfgx; cfg; cfg = cfg->next) + +DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); static struct irq_cfg *irq_cfg(unsigned int irq) { - if (irq >= nr_irqs) - return NULL; + struct irq_cfg *cfg; + + cfg = irq_cfgx; + while (cfg) { + if (cfg->irq == irq) + return cfg; - return &irq_cfgx[irq]; + cfg = cfg->next; + } + + return NULL; +} + +static struct irq_cfg *irq_cfg_alloc(unsigned int irq) +{ + struct irq_cfg *cfg, *cfg_pri; + int i; + int count = 0; + + cfg_pri = cfg = irq_cfgx; + while (cfg) { + if (cfg->irq == irq) + return cfg; + + cfg_pri = cfg; + cfg = cfg->next; + count++; + } + + if (!irq_cfgx_free) { + unsigned long phys; + unsigned long total_bytes; + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); + + total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; + if (after_bootmem) + cfg = kzalloc(total_bytes, GFP_ATOMIC); + else + cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); + + if (!cfg) + panic("please boot with nr_irq_cfg= %d\n", count * 2); + + phys = __pa(cfg); + printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); + + for (i = 0; i < nr_irq_cfg; i++) + init_one_irq_cfg(&cfg[i]); + + for (i = 1; i < nr_irq_cfg; i++) + cfg[i-1].next = &cfg[i]; + + irq_cfgx_free = cfg; + } + + cfg = irq_cfgx_free; + irq_cfgx_free = irq_cfgx_free->next; + cfg->next = NULL; + if (cfg_pri) + cfg_pri->next = cfg; + else + irq_cfgx = cfg; + cfg->irq = irq; + printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); + +#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG + { + /* dump the results */ + struct irq_cfg *cfg; + unsigned long phys; + unsigned long bytes = sizeof(struct irq_cfg); + + printk(KERN_DEBUG "=========================== %d\n", irq); + printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); + for_each_irq_cfg(cfg) { + phys = __pa(cfg); + printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); + } + printk(KERN_DEBUG "===========================\n"); + } +#endif + return cfg; } /* @@ -254,6 +385,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) { struct irq_pin_list *entry = irq_2_pin + irq; + irq_cfg_alloc(irq); while (entry->next) entry = irq_2_pin + entry->next; @@ -836,11 +968,13 @@ static int __assign_irq_vector(int irq) { static int current_vector = FIRST_DEVICE_VECTOR, current_offset; int vector, offset; + struct irq_cfg *cfg; BUG_ON((unsigned)irq >= nr_irqs); - if (irq_cfg(irq)->vector > 0) - return irq_cfg(irq)->vector; + cfg = irq_cfg(irq); + if (cfg->vector > 0) + return cfg->vector; vector = current_vector; offset = current_offset; @@ -857,7 +991,7 @@ next: current_vector = vector; current_offset = offset; - irq_cfg(irq)->vector = vector; + cfg->vector = vector; return vector; } @@ -1659,6 +1793,7 @@ static inline void init_IO_APIC_traps(void) { int irq; struct irq_desc *desc; + struct irq_cfg *cfg; /* * NOTE! The local APIC isn't very good at handling @@ -1671,8 +1806,9 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for (irq = 0; irq < nr_irqs ; irq++) { - if (IO_APIC_IRQ(irq) && !irq_cfg(irq)->vector) { + for_each_irq_cfg(cfg) { + irq = cfg->irq; + if (IO_APIC_IRQ(irq) && !cfg->vector) { /* * Hmm.. We don't have an entry for this, * so default to an old-fashioned 8259 @@ -2117,14 +2253,18 @@ int create_irq(void) /* Allocate an unused irq */ int irq, new, vector = 0; unsigned long flags; + struct irq_cfg *cfg_new; irq = -ENOSPC; spin_lock_irqsave(&vector_lock, flags); for (new = (nr_irqs - 1); new >= 0; new--) { if (platform_legacy_irq(new)) continue; - if (irq_cfg(new)->vector != 0) + cfg_new = irq_cfg(new); + if (cfg_new && cfg_new->vector != 0) continue; + if (!cfg_new) + cfg_new = irq_cfg_alloc(new); vector = __assign_irq_vector(new); if (likely(vector > 0)) irq = new; -- cgit v1.2.3 From 0f978f4505e96227a89b3c9447552aca983c6b57 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:26 -0700 Subject: x86: make 32bit to use irq_2_pin in irq_cfg so it is more like 64 bit. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 152 +++++++++++++++++++++++++++++++++---------- 1 file changed, 117 insertions(+), 35 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 5a83d7f5b147..59d2e8a273e3 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -95,10 +95,11 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); static int disable_timer_pin_1 __initdata; struct irq_cfg; - +struct irq_pin_list; struct irq_cfg { unsigned int irq; struct irq_cfg *next; + struct irq_pin_list *irq_2_pin; u8 vector; }; @@ -275,11 +276,66 @@ int pin_map_size; * between pins and IRQs. */ -static struct irq_pin_list { - int apic, pin, next; -} *irq_2_pin; +struct irq_pin_list { + int apic, pin; + struct irq_pin_list *next; +}; + +static struct irq_pin_list *irq_2_pin_head; +/* fill one page ? */ +static int nr_irq_2_pin = 0x100; +static struct irq_pin_list *irq_2_pin_ptr; +static void __init irq_2_pin_init_work(void *data) +{ + struct dyn_array *da = data; + struct irq_pin_list *pin; + int i; + + pin = *da->name; + + for (i = 1; i < *da->nr; i++) + pin[i-1].next = &pin[i]; + + irq_2_pin_ptr = &pin[0]; +} +DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); + +static struct irq_pin_list *get_one_free_irq_2_pin(void) +{ + struct irq_pin_list *pin; + int i; + + pin = irq_2_pin_ptr; + + if (pin) { + irq_2_pin_ptr = pin->next; + pin->next = NULL; + return pin; + } + + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); + + if (after_bootmem) + pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, + GFP_ATOMIC); + else + pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * + nr_irq_2_pin, PAGE_SIZE, 0); + + if (!pin) + panic("can not get more irq_2_pin\n"); -DEFINE_DYN_ARRAY(irq_2_pin, sizeof(struct irq_pin_list), pin_map_size, 16, NULL); + for (i = 1; i < nr_irq_2_pin; i++) + pin[i-1].next = &pin[i]; + + irq_2_pin_ptr = pin->next; + pin->next = NULL; + + return pin; +} struct io_apic { unsigned int index; @@ -383,20 +439,34 @@ static void ioapic_mask_entry(int apic, int pin) */ static void add_pin_to_irq(unsigned int irq, int apic, int pin) { - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg; + struct irq_pin_list *entry; + + /* first time to refer irq_cfg, so with new */ + cfg = irq_cfg_alloc(irq); + entry = cfg->irq_2_pin; + if (!entry) { + entry = get_one_free_irq_2_pin(); + cfg->irq_2_pin = entry; + entry->apic = apic; + entry->pin = pin; + printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); + return; + } - irq_cfg_alloc(irq); - while (entry->next) - entry = irq_2_pin + entry->next; + while (entry->next) { + /* not again, please */ + if (entry->apic == apic && entry->pin == pin) + return; - if (entry->pin != -1) { - entry->next = first_free_entry; - entry = irq_2_pin + entry->next; - if (++first_free_entry >= pin_map_size) - panic("io_apic.c: whoops"); + entry = entry->next; } + + entry->next = get_one_free_irq_2_pin(); + entry = entry->next; entry->apic = apic; entry->pin = pin; + printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); } /* @@ -406,35 +476,45 @@ static void __init replace_pin_at_irq(unsigned int irq, int oldapic, int oldpin, int newapic, int newpin) { - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg = irq_cfg(irq); + struct irq_pin_list *entry = cfg->irq_2_pin; + int replaced = 0; - while (1) { + while (entry) { if (entry->apic == oldapic && entry->pin == oldpin) { entry->apic = newapic; entry->pin = newpin; - } - if (!entry->next) + replaced = 1; + /* every one is different, right? */ break; - entry = irq_2_pin + entry->next; + } + entry = entry->next; } + + /* why? call replace before add? */ + if (!replaced) + add_pin_to_irq(irq, newapic, newpin); } static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) { - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_cfg *cfg; + struct irq_pin_list *entry; unsigned int pin, reg; + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; for (;;) { - pin = entry->pin; - if (pin == -1) + if (!entry) break; + pin = entry->pin; reg = io_apic_read(entry->apic, 0x10 + pin*2); reg &= ~disable; reg |= enable; io_apic_modify(entry->apic, 0x10 + pin*2, reg); if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } } @@ -509,13 +589,18 @@ static void clear_IO_APIC(void) #ifdef CONFIG_SMP static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) { + struct irq_cfg *cfg; unsigned long flags; int pin; - struct irq_pin_list *entry = irq_2_pin + irq; + struct irq_pin_list *entry; unsigned int apicid_value; cpumask_t tmp; struct irq_desc *desc; + + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; + cpus_and(tmp, cpumask, cpu_online_map); if (cpus_empty(tmp)) tmp = TARGET_CPUS; @@ -527,13 +612,13 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) apicid_value = apicid_value << 24; spin_lock_irqsave(&ioapic_lock, flags); for (;;) { - pin = entry->pin; - if (pin == -1) + if (!entry) break; + pin = entry->pin; io_apic_write(entry->apic, 0x10 + 1 + pin*2, apicid_value); if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } desc = irq_to_desc(irq); desc->affinity = cpumask; @@ -1152,6 +1237,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_02 reg_02; union IO_APIC_reg_03 reg_03; unsigned long flags; + struct irq_cfg *cfg; if (apic_verbosity == APIC_QUIET) return; @@ -1240,16 +1326,16 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for (i = 0; i < nr_irqs; i++) { - struct irq_pin_list *entry = irq_2_pin + i; - if (entry->pin < 0) + for_each_irq_cfg(cfg) { + struct irq_pin_list *entry = cfg->irq_2_pin; + if (!entry) continue; printk(KERN_DEBUG "IRQ%d ", i); for (;;) { printk("-> %d:%d", entry->apic, entry->pin); if (!entry->next) break; - entry = irq_2_pin + entry->next; + entry = entry->next; } printk("\n"); } @@ -1420,10 +1506,6 @@ static void __init enable_IO_APIC(void) int i, apic; unsigned long flags; - for (i = 0; i < pin_map_size; i++) { - irq_2_pin[i].pin = -1; - irq_2_pin[i].next = 0; - } if (!pirqs_enabled) for (i = 0; i < MAX_PIRQS; i++) pirq_entries[i] = -1; -- cgit v1.2.3 From 199751d715bba5b469ea22adadc68a4166bfa4f5 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:27 -0700 Subject: x86: make 32 bit to use sparse_irq but actually irq still needs to be less than NR_IRQS, because interrupt[NR_IRQS] in entry.S. need to enable per_cpu vector... Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 63 ++++++++++++++++++++++++++++++-------------- arch/x86/kernel/irq_32.c | 37 +++++++++++++++++++------- arch/x86/kernel/irqinit_32.c | 7 +++++ 3 files changed, 78 insertions(+), 29 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 59d2e8a273e3..66c0a91362a7 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -595,7 +595,6 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) struct irq_pin_list *entry; unsigned int apicid_value; cpumask_t tmp; - struct irq_desc *desc; cfg = irq_cfg(irq); @@ -620,8 +619,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) break; entry = entry->next; } - desc = irq_to_desc(irq); - desc->affinity = cpumask; + irq_to_desc(irq)->affinity = cpumask; spin_unlock_irqrestore(&ioapic_lock, flags); } @@ -1055,8 +1053,6 @@ static int __assign_irq_vector(int irq) int vector, offset; struct irq_cfg *cfg; - BUG_ON((unsigned)irq >= nr_irqs); - cfg = irq_cfg(irq); if (cfg->vector > 0) return cfg->vector; @@ -1103,7 +1099,12 @@ static void ioapic_register_intr(int irq, int vector, unsigned long trigger) { struct irq_desc *desc; - desc = irq_to_desc(irq); + /* first time to use this irq_desc */ + if (irq < 16) + desc = irq_to_desc(irq); + else + desc = irq_to_desc_alloc(irq); + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) { desc->status |= IRQ_LEVEL; @@ -2330,16 +2331,19 @@ device_initcall(ioapic_init_sysfs); /* * Dynamic irq allocate and deallocation */ -int create_irq(void) +unsigned int create_irq_nr(unsigned int irq_want) { /* Allocate an unused irq */ - int irq, new, vector = 0; + unsigned int irq, new, vector = 0; unsigned long flags; struct irq_cfg *cfg_new; - irq = -ENOSPC; + /* only can use bus/dev/fn.. when per_cpu vector is used */ + irq_want = nr_irqs - 1; + + irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = (nr_irqs - 1); new >= 0; new--) { + for (new = (nr_irqs - 1); new > 0; new--) { if (platform_legacy_irq(new)) continue; cfg_new = irq_cfg(new); @@ -2354,13 +2358,18 @@ int create_irq(void) } spin_unlock_irqrestore(&vector_lock, flags); - if (irq >= 0) { + if (irq > 0) { set_intr_gate(vector, interrupt[irq]); dynamic_irq_init(irq); } return irq; } +int create_irq(void) +{ + return create_irq_nr(nr_irqs - 1); +} + void destroy_irq(unsigned int irq) { unsigned long flags; @@ -2415,7 +2424,6 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) unsigned int dest; cpumask_t tmp; int vector; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2435,8 +2443,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); write_msi_msg(irq, &msg); - desc = irq_to_desc(irq); - desc->affinity = mask; + irq_to_desc(irq)->affinity = mask; } #endif /* CONFIG_SMP */ @@ -2455,13 +2462,31 @@ static struct irq_chip msi_chip = { .retrigger = ioapic_retrigger_irq, }; +static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) +{ + unsigned int irq; + + irq = dev->bus->number; + irq <<= 8; + irq |= dev->devfn; + irq <<= 12; + + return irq; +} + int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) { struct msi_msg msg; int irq, ret; - irq = create_irq(); - if (irq < 0) - return irq; + + unsigned int irq_want; + + irq_want = build_irq_for_pci_dev(dev) + 0x100; + + irq = create_irq_nr(irq_want); + + if (irq == 0) + return -1; ret = msi_compose_msg(dev, irq, &msg); if (ret < 0) { @@ -2510,7 +2535,6 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2521,8 +2545,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) dest = cpu_mask_to_apicid(mask); target_ht_irq(irq, dest); - desc = irq_to_desc(irq); - desc->affinity = mask; + irq_to_desc(irq)->affinity = mask; } #endif diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 576c5df6cad8..0a57e39159a8 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -224,9 +224,10 @@ unsigned int do_IRQ(struct pt_regs *regs) struct pt_regs *old_regs; /* high bit used in ret_from_ code */ int overflow, irq = ~regs->orig_ax; - struct irq_desc *desc = irq_to_desc(irq); + struct irq_desc *desc; - if (unlikely((unsigned)irq >= nr_irqs)) { + desc = irq_to_desc(irq); + if (unlikely(!desc)) { printk(KERN_EMERG "%s: cannot handle IRQ %d\n", __func__, irq); BUG(); @@ -263,6 +264,24 @@ int show_interrupts(struct seq_file *p, void *v) int i = *(loff_t *) v, j; struct irqaction * action; unsigned long flags; + unsigned int entries; + struct irq_desc *desc; + int tail = 0; + +#ifdef CONFIG_HAVE_SPARSE_IRQ + desc = (struct irq_desc *)v; + entries = -1U; + i = desc->irq; + if (!desc->next) + tail = 1; +#else + entries = nr_irqs - 1; + i = *(loff_t *) v; + if (i == nr_irqs) + tail = 1; + else + desc = irq_to_desc(i); +#endif if (i == 0) { seq_printf(p, " "); @@ -271,9 +290,8 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i < nr_irqs) { + if (i <= entries) { unsigned any_count = 0; - struct irq_desc *desc = irq_to_desc(i); spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP @@ -285,7 +303,7 @@ int show_interrupts(struct seq_file *p, void *v) action = desc->action; if (!action && !any_count) goto skip; - seq_printf(p, "%3d: ",i); + seq_printf(p, "%#x: ",i); #ifndef CONFIG_SMP seq_printf(p, "%10u ", kstat_irqs(i)); #else @@ -304,7 +322,9 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); skip: spin_unlock_irqrestore(&desc->lock, flags); - } else if (i == nr_irqs) { + } + + if (tail) { seq_printf(p, "NMI: "); for_each_online_cpu(j) seq_printf(p, "%10u ", nmi_count(j)); @@ -396,15 +416,14 @@ void fixup_irqs(cpumask_t map) { unsigned int irq; static int warned; + struct irq_desc *desc; - for (irq = 0; irq < nr_irqs; irq++) { + for_each_irq_desc(irq, desc) { cpumask_t mask; - struct irq_desc *desc; if (irq == 2) continue; - desc = irq_to_desc(irq); cpus_and(mask, desc->affinity, map); if (any_online_cpu(mask) == NR_CPUS) { printk("Breaking affinity for irq %i\n", irq); diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 65c1c9507707..ded09ac2642e 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -69,6 +69,13 @@ void __init init_ISA_irqs (void) * 16 old-style INTA-cycle interrupts: */ for (i = 0; i < 16; i++) { + /* first time call this irq_desc */ + struct irq_desc *desc = irq_to_desc_alloc(i); + + desc->status = IRQ_DISABLED; + desc->action = NULL; + desc->depth = 1; + set_irq_chip_and_handler_name(i, &i8259A_chip, handle_level_irq, "XT"); } -- cgit v1.2.3 From 497c9a195db918d3f035e8cb3021e5d4d035516e Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:28 -0700 Subject: x86: make 32bit support per_cpu vector so we can merge io_apic_32.c and io_apic_64.c v2: Use cpu_online_map as target cpus for bigsmp, just like 64-bit is doing. Also remove some unused TARGET_CPUS macro. v3: need to check if desc is null in smp_irq_move_cleanup also migration needs to reset vector too, so copy __target_IO_APIC_irq from 64bit. (the duplication will go away once the two files are unified.) Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 2 +- arch/x86/kernel/io_apic_32.c | 719 +++++++++++++++++++++++++++---------------- arch/x86/kernel/irq_32.c | 18 +- arch/x86/kernel/irqinit_32.c | 41 ++- 4 files changed, 500 insertions(+), 280 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index b21fbfaffe39..4d82171d0f9c 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -629,7 +629,7 @@ ENTRY(interrupt) ENTRY(irq_entries_start) RING0_INT_FRAME vector=0 -.rept NR_IRQS +.rept NR_VECTORS ALIGN .if vector CFI_ADJUST_CFA_OFFSET -4 diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 66c0a91362a7..ea33d3c74970 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -48,6 +48,7 @@ #include #include +#include #include #include @@ -60,7 +61,7 @@ atomic_t irq_mis_count; static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; static DEFINE_SPINLOCK(ioapic_lock); -DEFINE_SPINLOCK(vector_lock); +static DEFINE_SPINLOCK(vector_lock); int timer_through_8259 __initdata; @@ -100,28 +101,32 @@ struct irq_cfg { unsigned int irq; struct irq_cfg *next; struct irq_pin_list *irq_2_pin; + cpumask_t domain; + cpumask_t old_domain; + unsigned move_cleanup_count; u8 vector; + u8 move_in_progress : 1; }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ static struct irq_cfg irq_cfg_legacy[] __initdata = { - [0] = { .irq = 0, .vector = IRQ0_VECTOR, }, - [1] = { .irq = 1, .vector = IRQ1_VECTOR, }, - [2] = { .irq = 2, .vector = IRQ2_VECTOR, }, - [3] = { .irq = 3, .vector = IRQ3_VECTOR, }, - [4] = { .irq = 4, .vector = IRQ4_VECTOR, }, - [5] = { .irq = 5, .vector = IRQ5_VECTOR, }, - [6] = { .irq = 6, .vector = IRQ6_VECTOR, }, - [7] = { .irq = 7, .vector = IRQ7_VECTOR, }, - [8] = { .irq = 8, .vector = IRQ8_VECTOR, }, - [9] = { .irq = 9, .vector = IRQ9_VECTOR, }, - [10] = { .irq = 10, .vector = IRQ10_VECTOR, }, - [11] = { .irq = 11, .vector = IRQ11_VECTOR, }, - [12] = { .irq = 12, .vector = IRQ12_VECTOR, }, - [13] = { .irq = 13, .vector = IRQ13_VECTOR, }, - [14] = { .irq = 14, .vector = IRQ14_VECTOR, }, - [15] = { .irq = 15, .vector = IRQ15_VECTOR, }, + [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, + [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, + [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, + [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, + [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, + [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, + [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, + [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, + [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, + [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, + [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, + [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, + [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, + [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, + [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, + [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, }; static struct irq_cfg irq_cfg_init = { .irq = -1U, }; @@ -263,6 +268,7 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) return cfg; } +static int assign_irq_vector(int irq, cpumask_t mask); /* * Rough estimation of how many shared IRQs there are, can * be changed anytime. @@ -432,6 +438,65 @@ static void ioapic_mask_entry(int apic, int pin) spin_unlock_irqrestore(&ioapic_lock, flags); } +#ifdef CONFIG_SMP +static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) +{ + int apic, pin; + struct irq_cfg *cfg; + struct irq_pin_list *entry; + + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; + for (;;) { + unsigned int reg; + + if (!entry) + break; + + apic = entry->apic; + pin = entry->pin; + io_apic_write(apic, 0x11 + pin*2, dest); + reg = io_apic_read(apic, 0x10 + pin*2); + reg &= ~IO_APIC_REDIR_VECTOR_MASK; + reg |= vector; + io_apic_modify(apic, 0x10 + pin *2, reg); + if (!entry->next) + break; + entry = entry->next; + } +} +static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + unsigned long flags; + unsigned int dest; + cpumask_t tmp; + + cfg = irq_cfg(irq); + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cpus_and(tmp, cfg->domain, mask); + + dest = cpu_mask_to_apicid(tmp); + /* + * Only the high 8 bits are valid. + */ + dest = SET_APIC_LOGICAL_ID(dest); + + spin_lock_irqsave(&ioapic_lock, flags); + __target_IO_APIC_irq(irq, dest, cfg->vector); + irq_to_desc(irq)->affinity = mask; + spin_unlock_irqrestore(&ioapic_lock, flags); +} + +#endif /* CONFIG_SMP */ + /* * The common case is 1:1 IRQ<->pin mappings. Sometimes there are * shared ISA-space IRQs, so we have to support them. We are super @@ -586,45 +651,6 @@ static void clear_IO_APIC(void) clear_IO_APIC_pin(apic, pin); } -#ifdef CONFIG_SMP -static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t cpumask) -{ - struct irq_cfg *cfg; - unsigned long flags; - int pin; - struct irq_pin_list *entry; - unsigned int apicid_value; - cpumask_t tmp; - - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - - cpus_and(tmp, cpumask, cpu_online_map); - if (cpus_empty(tmp)) - tmp = TARGET_CPUS; - - cpus_and(cpumask, tmp, CPU_MASK_ALL); - - apicid_value = cpu_mask_to_apicid(cpumask); - /* Prepare to do the io_apic_write */ - apicid_value = apicid_value << 24; - spin_lock_irqsave(&ioapic_lock, flags); - for (;;) { - if (!entry) - break; - pin = entry->pin; - io_apic_write(entry->apic, 0x10 + 1 + pin*2, apicid_value); - if (!entry->next) - break; - entry = entry->next; - } - irq_to_desc(irq)->affinity = cpumask; - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -#endif /* CONFIG_SMP */ - #ifndef CONFIG_SMP void send_IPI_self(int vector) { @@ -789,32 +815,6 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) } EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); -/* - * This function currently is only a helper for the i386 smp boot process where - * we need to reprogram the ioredtbls to cater for the cpus which have come online - * so mask in all cases should simply be TARGET_CPUS - */ -#ifdef CONFIG_SMP -void __init setup_ioapic_dest(void) -{ - int pin, ioapic, irq, irq_entry; - - if (skip_ioapic_setup == 1) - return; - - for (ioapic = 0; ioapic < nr_ioapics; ioapic++) { - for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { - irq_entry = find_irq_entry(ioapic, pin, mp_INT); - if (irq_entry == -1) - continue; - irq = pin_2_irq(irq_entry, ioapic, pin); - set_ioapic_affinity_irq(irq, TARGET_CPUS); - } - - } -} -#endif - #if defined(CONFIG_EISA) || defined(CONFIG_MCA) /* * EISA Edge/Level control register, ELCR @@ -1046,47 +1046,138 @@ static inline int IO_APIC_irq_trigger(int irq) return 0; } +void lock_vector_lock(void) +{ + /* Used to the online set of cpus does not change + * during assign_irq_vector. + */ + spin_lock(&vector_lock); +} -static int __assign_irq_vector(int irq) +void unlock_vector_lock(void) { - static int current_vector = FIRST_DEVICE_VECTOR, current_offset; - int vector, offset; - struct irq_cfg *cfg; + spin_unlock(&vector_lock); +} - cfg = irq_cfg(irq); - if (cfg->vector > 0) - return cfg->vector; +static int __assign_irq_vector(int irq, cpumask_t mask) +{ + static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; + unsigned int old_vector; + int cpu; + struct irq_cfg *cfg; - vector = current_vector; - offset = current_offset; -next: - vector += 8; - if (vector >= first_system_vector) { - offset = (offset + 1) % 8; - vector = FIRST_DEVICE_VECTOR + offset; - } - if (vector == current_vector) - return -ENOSPC; - if (test_and_set_bit(vector, used_vectors)) - goto next; + cfg = irq_cfg(irq); - current_vector = vector; - current_offset = offset; - cfg->vector = vector; + /* Only try and allocate irqs on cpus that are present */ + cpus_and(mask, mask, cpu_online_map); - return vector; -} + if ((cfg->move_in_progress) || cfg->move_cleanup_count) + return -EBUSY; -static int assign_irq_vector(int irq) -{ + old_vector = cfg->vector; + if (old_vector) { + cpumask_t tmp; + cpus_and(tmp, cfg->domain, mask); + if (!cpus_empty(tmp)) + return 0; + } + + for_each_cpu_mask_nr(cpu, mask) { + cpumask_t domain, new_mask; + int new_cpu; + int vector, offset; + + domain = vector_allocation_domain(cpu); + cpus_and(new_mask, domain, cpu_online_map); + + vector = current_vector; + offset = current_offset; +next: + vector += 8; + if (vector >= first_system_vector) { + /* If we run out of vectors on large boxen, must share them. */ + offset = (offset + 1) % 8; + vector = FIRST_DEVICE_VECTOR + offset; + } + if (unlikely(current_vector == vector)) + continue; + if (vector == SYSCALL_VECTOR) + goto next; + + for_each_cpu_mask_nr(new_cpu, new_mask) + if (per_cpu(vector_irq, new_cpu)[vector] != -1) + goto next; + /* Found one! */ + current_vector = vector; + current_offset = offset; + if (old_vector) { + cfg->move_in_progress = 1; + cfg->old_domain = cfg->domain; + } + for_each_cpu_mask_nr(new_cpu, new_mask) + per_cpu(vector_irq, new_cpu)[vector] = irq; + cfg->vector = vector; + cfg->domain = domain; + return 0; + } + return -ENOSPC; +} + +static int assign_irq_vector(int irq, cpumask_t mask) +{ + int err; unsigned long flags; - int vector; spin_lock_irqsave(&vector_lock, flags); - vector = __assign_irq_vector(irq); + err = __assign_irq_vector(irq, mask); spin_unlock_irqrestore(&vector_lock, flags); - return vector; + return err; +} + +static void __clear_irq_vector(int irq) +{ + struct irq_cfg *cfg; + cpumask_t mask; + int cpu, vector; + + cfg = irq_cfg(irq); + BUG_ON(!cfg->vector); + + vector = cfg->vector; + cpus_and(mask, cfg->domain, cpu_online_map); + for_each_cpu_mask_nr(cpu, mask) + per_cpu(vector_irq, cpu)[vector] = -1; + + cfg->vector = 0; + cpus_clear(cfg->domain); +} + +void __setup_vector_irq(int cpu) +{ + /* Initialize vector_irq on a new cpu */ + /* This function must be called with vector_lock held */ + int irq, vector; + struct irq_cfg *cfg; + + /* Mark the inuse vectors */ + for_each_irq_cfg(cfg) { + if (!cpu_isset(cpu, cfg->domain)) + continue; + vector = cfg->vector; + irq = cfg->irq; + per_cpu(vector_irq, cpu)[vector] = irq; + } + /* Mark the free vectors */ + for (vector = 0; vector < NR_VECTORS; ++vector) { + irq = per_cpu(vector_irq, cpu)[vector]; + if (irq < 0) + continue; + + cfg = irq_cfg(irq); + if (!cpu_isset(cpu, cfg->domain)) + per_cpu(vector_irq, cpu)[vector] = -1; + } } static struct irq_chip ioapic_chip; @@ -1095,7 +1186,7 @@ static struct irq_chip ioapic_chip; #define IOAPIC_EDGE 0 #define IOAPIC_LEVEL 1 -static void ioapic_register_intr(int irq, int vector, unsigned long trigger) +static void ioapic_register_intr(int irq, unsigned long trigger) { struct irq_desc *desc; @@ -1115,79 +1206,109 @@ static void ioapic_register_intr(int irq, int vector, unsigned long trigger) set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_edge_irq, "edge"); } - set_intr_gate(vector, interrupt[irq]); } -static void __init setup_IO_APIC_irqs(void) +static int setup_ioapic_entry(int apic, int irq, + struct IO_APIC_route_entry *entry, + unsigned int destination, int trigger, + int polarity, int vector) { + /* + * add it to the IO-APIC irq-routing table: + */ + memset(entry,0,sizeof(*entry)); + + entry->delivery_mode = INT_DELIVERY_MODE; + entry->dest_mode = INT_DEST_MODE; + entry->dest.logical.logical_dest = destination; + + entry->mask = 0; /* enable IRQ */ + entry->trigger = trigger; + entry->polarity = polarity; + entry->vector = vector; + + /* Mask level triggered irqs. + * Use IRQ_DELAYED_DISABLE for edge triggered irqs. + */ + if (trigger) + entry->mask = 1; + + return 0; +} + +static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, + int trigger, int polarity) +{ + struct irq_cfg *cfg; struct IO_APIC_route_entry entry; - int apic, pin, idx, irq, first_notcon = 1, vector; + cpumask_t mask; + + if (!IO_APIC_IRQ(irq)) + return; + + cfg = irq_cfg(irq); + + mask = TARGET_CPUS; + if (assign_irq_vector(irq, mask)) + return; + + cpus_and(mask, cfg->domain, mask); + + apic_printk(APIC_VERBOSE,KERN_DEBUG + "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " + "IRQ %d Mode:%i Active:%i)\n", + apic, mp_ioapics[apic].mp_apicid, pin, cfg->vector, + irq, trigger, polarity); + + + if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, + cpu_mask_to_apicid(mask), trigger, polarity, + cfg->vector)) { + printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", + mp_ioapics[apic].mp_apicid, pin); + __clear_irq_vector(irq); + return; + } + + ioapic_register_intr(irq, trigger); + if (irq < 16) + disable_8259A_irq(irq); + + ioapic_write_entry(apic, pin, entry); +} + +static void __init setup_IO_APIC_irqs(void) +{ + int apic, pin, idx, irq, first_notcon = 1; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); for (apic = 0; apic < nr_ioapics; apic++) { for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - /* - * add it to the IO-APIC irq-routing table: - */ - memset(&entry, 0, sizeof(entry)); - - entry.delivery_mode = INT_DELIVERY_MODE; - entry.dest_mode = INT_DEST_MODE; - entry.mask = 0; /* enable IRQ */ - entry.dest.logical.logical_dest = - cpu_mask_to_apicid(TARGET_CPUS); - - idx = find_irq_entry(apic, pin, mp_INT); + idx = find_irq_entry(apic,pin,mp_INT); if (idx == -1) { if (first_notcon) { - apic_printk(APIC_VERBOSE, KERN_DEBUG - " IO-APIC (apicid-pin) %d-%d", - mp_ioapics[apic].mp_apicid, - pin); + apic_printk(APIC_VERBOSE, KERN_DEBUG " IO-APIC (apicid-pin) %d-%d", mp_ioapics[apic].mp_apicid, pin); first_notcon = 0; } else - apic_printk(APIC_VERBOSE, ", %d-%d", - mp_ioapics[apic].mp_apicid, pin); + apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mp_apicid, pin); continue; } - if (!first_notcon) { apic_printk(APIC_VERBOSE, " not connected.\n"); first_notcon = 1; } - entry.trigger = irq_trigger(idx); - entry.polarity = irq_polarity(idx); - - if (irq_trigger(idx)) { - entry.trigger = 1; - entry.mask = 1; - } - irq = pin_2_irq(idx, apic, pin); - /* - * skip adding the timer int on secondary nodes, which causes - * a small but painful rift in the time-space continuum - */ - if (multi_timer_check(apic, irq)) - continue; - else - add_pin_to_irq(irq, apic, pin); - if (!apic && !IO_APIC_IRQ(irq)) - continue; + if (multi_timer_check(apic, irq)) + continue; - if (IO_APIC_IRQ(irq)) { - vector = assign_irq_vector(irq); - entry.vector = vector; - ioapic_register_intr(irq, vector, IOAPIC_AUTO); + add_pin_to_irq(irq, apic, pin); - if (!apic && (irq < 16)) - disable_8259A_irq(irq); - } - ioapic_write_entry(apic, pin, entry); + setup_IO_APIC_irq(apic, pin, irq, + irq_trigger(idx), irq_polarity(idx)); } } @@ -1221,7 +1342,7 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, * The timer IRQ doesn't have to know that behind the * scene we may have a 8259A-master in AEOI mode ... */ - ioapic_register_intr(0, vector, IOAPIC_EDGE); + ioapic_register_intr(0, IOAPIC_EDGE); /* * Add it to the IO-APIC irq-routing table: @@ -1805,8 +1926,10 @@ static unsigned int startup_ioapic_irq(unsigned int irq) return was_pending; } +static void irq_complete_move(unsigned int irq); static void ack_ioapic_irq(unsigned int irq) { + irq_complete_move(irq); move_native_irq(irq); ack_APIC_irq(); } @@ -1816,6 +1939,7 @@ static void ack_ioapic_quirk_irq(unsigned int irq) unsigned long v; int i; + irq_complete_move(irq); move_native_irq(irq); /* * It appears there is an erratum which affects at least version 0x11 @@ -1858,6 +1982,64 @@ static int ioapic_retrigger_irq(unsigned int irq) return 1; } +#ifdef CONFIG_SMP +asmlinkage void smp_irq_move_cleanup_interrupt(void) +{ + unsigned vector, me; + ack_APIC_irq(); + irq_enter(); + + me = smp_processor_id(); + for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { + unsigned int irq; + struct irq_desc *desc; + struct irq_cfg *cfg; + irq = __get_cpu_var(vector_irq)[vector]; + + desc = irq_to_desc(irq); + if (!desc) + continue; + + cfg = irq_cfg(irq); + spin_lock(&desc->lock); + if (!cfg->move_cleanup_count) + goto unlock; + + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) + goto unlock; + + __get_cpu_var(vector_irq)[vector] = -1; + cfg->move_cleanup_count--; +unlock: + spin_unlock(&desc->lock); + } + + irq_exit(); +} + +static void irq_complete_move(unsigned int irq) +{ + struct irq_cfg *cfg = irq_cfg(irq); + unsigned vector, me; + + if (likely(!cfg->move_in_progress)) + return; + + vector = ~get_irq_regs()->orig_ax; + me = smp_processor_id(); + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { + cpumask_t cleanup_mask; + + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } +} +#else +static inline void irq_complete_move(unsigned int irq) {} +#endif + static struct irq_chip ioapic_chip __read_mostly = { .name = "IO-APIC", .startup = startup_ioapic_irq, @@ -1940,7 +2122,7 @@ static struct irq_chip lapic_chip __read_mostly = { .ack = ack_lapic_irq, }; -static void lapic_register_intr(int irq, int vector) +static void lapic_register_intr(int irq) { struct irq_desc *desc; @@ -1948,7 +2130,6 @@ static void lapic_register_intr(int irq, int vector) desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, "edge"); - set_intr_gate(vector, interrupt[irq]); } static void __init setup_nmi(void) @@ -2036,9 +2217,9 @@ static inline void __init unlock_ExtINT_logic(void) */ static inline void __init check_timer(void) { + struct irq_cfg *cfg = irq_cfg(0); int apic1, pin1, apic2, pin2; int no_pin1 = 0; - int vector; unsigned int ver; unsigned long flags; @@ -2051,8 +2232,7 @@ static inline void __init check_timer(void) * get/set the timer IRQ vector: */ disable_8259A_irq(0); - vector = assign_irq_vector(0); - set_intr_gate(vector, interrupt[0]); + assign_irq_vector(0, TARGET_CPUS); /* * As IRQ0 is to be enabled in the 8259A, the virtual @@ -2074,7 +2254,7 @@ static inline void __init check_timer(void) apic_printk(APIC_QUIET, KERN_INFO "..TIMER: vector=0x%02X " "apic1=%d pin1=%d apic2=%d pin2=%d\n", - vector, apic1, pin1, apic2, pin2); + cfg->vector, apic1, pin1, apic2, pin2); /* * Some BIOS writers are clueless and report the ExtINTA @@ -2098,7 +2278,7 @@ static inline void __init check_timer(void) */ if (no_pin1) { add_pin_to_irq(0, apic1, pin1); - setup_timer_IRQ0_pin(apic1, pin1, vector); + setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } unmask_IO_APIC_irq(0); if (timer_irq_works()) { @@ -2123,7 +2303,7 @@ static inline void __init check_timer(void) * legacy devices should be connected to IO APIC #0 */ replace_pin_at_irq(0, apic1, pin1, apic2, pin2); - setup_timer_IRQ0_pin(apic2, pin2, vector); + setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); unmask_IO_APIC_irq(0); enable_8259A_irq(0); if (timer_irq_works()) { @@ -2154,8 +2334,8 @@ static inline void __init check_timer(void) apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as Virtual Wire IRQ...\n"); - lapic_register_intr(0, vector); - apic_write(APIC_LVT0, APIC_DM_FIXED | vector); /* Fixed mode */ + lapic_register_intr(0); + apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ enable_8259A_irq(0); if (timer_irq_works()) { @@ -2163,7 +2343,7 @@ static inline void __init check_timer(void) goto out; } disable_8259A_irq(0); - apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | vector); + apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector); apic_printk(APIC_QUIET, KERN_INFO "..... failed.\n"); apic_printk(APIC_QUIET, KERN_INFO @@ -2207,12 +2387,6 @@ out: void __init setup_IO_APIC(void) { - int i; - - /* Reserve all the system vectors. */ - for (i = first_system_vector; i < NR_VECTORS; i++) - set_bit(i, used_vectors); - enable_IO_APIC(); io_apic_irqs = ~PIC_IRQS; @@ -2334,12 +2508,14 @@ device_initcall(ioapic_init_sysfs); unsigned int create_irq_nr(unsigned int irq_want) { /* Allocate an unused irq */ - unsigned int irq, new, vector = 0; + unsigned int irq, new; unsigned long flags; struct irq_cfg *cfg_new; +#ifndef CONFIG_HAVE_SPARSE_IRQ /* only can use bus/dev/fn.. when per_cpu vector is used */ irq_want = nr_irqs - 1; +#endif irq = 0; spin_lock_irqsave(&vector_lock, flags); @@ -2351,15 +2527,13 @@ unsigned int create_irq_nr(unsigned int irq_want) continue; if (!cfg_new) cfg_new = irq_cfg_alloc(new); - vector = __assign_irq_vector(new); - if (likely(vector > 0)) + if (__assign_irq_vector(new, TARGET_CPUS) == 0) irq = new; break; } spin_unlock_irqrestore(&vector_lock, flags); if (irq > 0) { - set_intr_gate(vector, interrupt[irq]); dynamic_irq_init(irq); } return irq; @@ -2377,8 +2551,7 @@ void destroy_irq(unsigned int irq) dynamic_irq_cleanup(irq); spin_lock_irqsave(&vector_lock, flags); - clear_bit(irq_cfg(irq)->vector, used_vectors); - irq_cfg(irq)->vector = 0; + __clear_irq_vector(irq); spin_unlock_irqrestore(&vector_lock, flags); } @@ -2388,57 +2561,65 @@ void destroy_irq(unsigned int irq) #ifdef CONFIG_PCI_MSI static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg) { - int vector; + struct irq_cfg *cfg; + int err; unsigned dest; + cpumask_t tmp; - vector = assign_irq_vector(irq); - if (vector >= 0) { - dest = cpu_mask_to_apicid(TARGET_CPUS); - - msg->address_hi = MSI_ADDR_BASE_HI; - msg->address_lo = - MSI_ADDR_BASE_LO | - ((INT_DEST_MODE == 0) ? -MSI_ADDR_DEST_MODE_PHYSICAL: - MSI_ADDR_DEST_MODE_LOGICAL) | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_ADDR_REDIRECTION_CPU: - MSI_ADDR_REDIRECTION_LOWPRI) | - MSI_ADDR_DEST_ID(dest); + tmp = TARGET_CPUS; + err = assign_irq_vector(irq, tmp); + if (err) + return err; - msg->data = - MSI_DATA_TRIGGER_EDGE | - MSI_DATA_LEVEL_ASSERT | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? -MSI_DATA_DELIVERY_FIXED: - MSI_DATA_DELIVERY_LOWPRI) | - MSI_DATA_VECTOR(vector); - } - return vector; + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, tmp); + dest = cpu_mask_to_apicid(tmp); + + msg->address_hi = MSI_ADDR_BASE_HI; + msg->address_lo = + MSI_ADDR_BASE_LO | + ((INT_DEST_MODE == 0) ? + MSI_ADDR_DEST_MODE_PHYSICAL: + MSI_ADDR_DEST_MODE_LOGICAL) | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_ADDR_REDIRECTION_CPU: + MSI_ADDR_REDIRECTION_LOWPRI) | + MSI_ADDR_DEST_ID(dest); + + msg->data = + MSI_DATA_TRIGGER_EDGE | + MSI_DATA_LEVEL_ASSERT | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_DATA_DELIVERY_FIXED: + MSI_DATA_DELIVERY_LOWPRI) | + MSI_DATA_VECTOR(cfg->vector); + + return err; } #ifdef CONFIG_SMP static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; - int vector; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) - tmp = TARGET_CPUS; + return; - vector = assign_irq_vector(irq); - if (vector < 0) + if (assign_irq_vector(irq, mask)) return; - dest = cpu_mask_to_apicid(mask); + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); read_msi_msg(irq, &msg); msg.data &= ~MSI_DATA_VECTOR_MASK; - msg.data |= MSI_DATA_VECTOR(vector); + msg.data |= MSI_DATA_VECTOR(cfg->vector); msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; msg.address_lo |= MSI_ADDR_DEST_ID(dest); @@ -2517,15 +2698,15 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_SMP -static void target_ht_irq(unsigned int irq, unsigned int dest) +static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) { struct ht_irq_msg msg; fetch_ht_irq_msg(irq, &msg); - msg.address_lo &= ~(HT_IRQ_LOW_DEST_ID_MASK); + msg.address_lo &= ~(HT_IRQ_LOW_VECTOR_MASK | HT_IRQ_LOW_DEST_ID_MASK); msg.address_hi &= ~(HT_IRQ_HIGH_DEST_ID_MASK); - msg.address_lo |= HT_IRQ_LOW_DEST_ID(dest); + msg.address_lo |= HT_IRQ_LOW_VECTOR(vector) | HT_IRQ_LOW_DEST_ID(dest); msg.address_hi |= HT_IRQ_HIGH_DEST_ID(dest); write_ht_irq_msg(irq, &msg); @@ -2533,18 +2714,22 @@ static void target_ht_irq(unsigned int irq, unsigned int dest) static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) - tmp = TARGET_CPUS; + return; - cpus_and(mask, tmp, CPU_MASK_ALL); + if (assign_irq_vector(irq, mask)) + return; - dest = cpu_mask_to_apicid(mask); + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); - target_ht_irq(irq, dest); + target_ht_irq(irq, dest, cfg->vector); irq_to_desc(irq)->affinity = mask; } #endif @@ -2562,16 +2747,18 @@ static struct irq_chip ht_irq_chip = { int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) { - int vector; + struct irq_cfg *cfg; + int err; + cpumask_t tmp; - vector = assign_irq_vector(irq); - if (vector >= 0) { + tmp = TARGET_CPUS; + err = assign_irq_vector(irq, tmp); + if ( !err) { struct ht_irq_msg msg; unsigned dest; - cpumask_t tmp; - cpus_clear(tmp); - cpu_set(vector >> 8, tmp); + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); @@ -2579,7 +2766,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) msg.address_lo = HT_IRQ_LOW_BASE | HT_IRQ_LOW_DEST_ID(dest) | - HT_IRQ_LOW_VECTOR(vector) | + HT_IRQ_LOW_VECTOR(cfg->vector) | ((INT_DEST_MODE == 0) ? HT_IRQ_LOW_DM_PHYSICAL : HT_IRQ_LOW_DM_LOGICAL) | @@ -2594,7 +2781,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) set_irq_chip_and_handler_name(irq, &ht_irq_chip, handle_edge_irq, "edge"); } - return vector; + return err; } #endif /* CONFIG_HT_IRQ */ @@ -2705,50 +2892,21 @@ int __init io_apic_get_redir_entries(int ioapic) } -int io_apic_set_pci_routing(int ioapic, int pin, int irq, int edge_level, int active_high_low) +int io_apic_set_pci_routing(int ioapic, int pin, int irq, int triggering, int polarity) { - struct IO_APIC_route_entry entry; - if (!IO_APIC_IRQ(irq)) { printk(KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", ioapic); return -EINVAL; } - /* - * Generate a PCI IRQ routing entry and program the IOAPIC accordingly. - * Note that we mask (disable) IRQs now -- these get enabled when the - * corresponding device driver registers for this IRQ. - */ - - memset(&entry, 0, sizeof(entry)); - - entry.delivery_mode = INT_DELIVERY_MODE; - entry.dest_mode = INT_DEST_MODE; - entry.dest.logical.logical_dest = cpu_mask_to_apicid(TARGET_CPUS); - entry.trigger = edge_level; - entry.polarity = active_high_low; - entry.mask = 1; - /* * IRQs < 16 are already in the irq_2_pin[] map */ if (irq >= 16) add_pin_to_irq(irq, ioapic, pin); - entry.vector = assign_irq_vector(irq); - - apic_printk(APIC_DEBUG, KERN_DEBUG "IOAPIC[%d]: Set PCI routing entry " - "(%d-%d -> 0x%x -> IRQ %d Mode:%i Active:%i)\n", ioapic, - mp_ioapics[ioapic].mp_apicid, pin, entry.vector, irq, - edge_level, active_high_low); - - ioapic_register_intr(irq, entry.vector, edge_level); - - if (!ioapic && (irq < 16)) - disable_8259A_irq(irq); - - ioapic_write_entry(ioapic, pin, entry); + setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); return 0; } @@ -2774,6 +2932,47 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) #endif /* CONFIG_ACPI */ +/* + * This function currently is only a helper for the i386 smp boot process where + * we need to reprogram the ioredtbls to cater for the cpus which have come online + * so mask in all cases should simply be TARGET_CPUS + */ +#ifdef CONFIG_SMP +void __init setup_ioapic_dest(void) +{ + int pin, ioapic, irq, irq_entry; + struct irq_cfg *cfg; + struct irq_desc *desc; + + if (skip_ioapic_setup == 1) + return; + + for (ioapic = 0; ioapic < nr_ioapics; ioapic++) { + for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { + irq_entry = find_irq_entry(ioapic, pin, mp_INT); + if (irq_entry == -1) + continue; + irq = pin_2_irq(irq_entry, ioapic, pin); + + /* setup_IO_APIC_irqs could fail to get vector for some device + * when you have too many devices, because at that time only boot + * cpu is online. + */ + cfg = irq_cfg(irq); + if (!cfg->vector) + setup_IO_APIC_irq(ioapic, pin, irq, + irq_trigger(irq_entry), + irq_polarity(irq_entry)); + else { + desc = irq_to_desc(irq); + set_ioapic_affinity_irq(irq, TARGET_CPUS); + } + } + + } +} +#endif + static int __init parse_disable_timer_pin_1(char *arg) { disable_timer_pin_1 = 1; diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 0a57e39159a8..b51ffdcfa31a 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -223,21 +223,25 @@ unsigned int do_IRQ(struct pt_regs *regs) { struct pt_regs *old_regs; /* high bit used in ret_from_ code */ - int overflow, irq = ~regs->orig_ax; + int overflow; + unsigned vector = ~regs->orig_ax; struct irq_desc *desc; + unsigned irq; - desc = irq_to_desc(irq); - if (unlikely(!desc)) { - printk(KERN_EMERG "%s: cannot handle IRQ %d\n", - __func__, irq); - BUG(); - } old_regs = set_irq_regs(regs); irq_enter(); + irq = __get_cpu_var(vector_irq)[vector]; overflow = check_stack_overflow(); + desc = irq_to_desc(irq); + if (unlikely(!desc)) { + printk(KERN_EMERG "%s: cannot handle IRQ %d vector %#x\n", + __func__, irq, vector); + BUG(); + } + if (!execute_on_irq_stack(overflow, desc, irq)) { if (unlikely(overflow)) print_stack_overflow(); diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index ded09ac2642e..9092103a18eb 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -90,6 +90,27 @@ static struct irqaction irq2 = { .name = "cascade", }; +DEFINE_PER_CPU(vector_irq_t, vector_irq) = { + [0 ... IRQ0_VECTOR - 1] = -1, + [IRQ0_VECTOR] = 0, + [IRQ1_VECTOR] = 1, + [IRQ2_VECTOR] = 2, + [IRQ3_VECTOR] = 3, + [IRQ4_VECTOR] = 4, + [IRQ5_VECTOR] = 5, + [IRQ6_VECTOR] = 6, + [IRQ7_VECTOR] = 7, + [IRQ8_VECTOR] = 8, + [IRQ9_VECTOR] = 9, + [IRQ10_VECTOR] = 10, + [IRQ11_VECTOR] = 11, + [IRQ12_VECTOR] = 12, + [IRQ13_VECTOR] = 13, + [IRQ14_VECTOR] = 14, + [IRQ15_VECTOR] = 15, + [IRQ15_VECTOR + 1 ... NR_VECTORS - 1] = -1 +}; + /* Overridden in paravirt.c */ void init_IRQ(void) __attribute__((weak, alias("native_init_IRQ"))); @@ -105,22 +126,14 @@ void __init native_init_IRQ(void) * us. (some of these will be overridden and become * 'special' SMP interrupts) */ - for (i = 0; i < (NR_VECTORS - FIRST_EXTERNAL_VECTOR); i++) { - int vector = FIRST_EXTERNAL_VECTOR + i; - if (i >= nr_irqs) - break; + for (i = FIRST_EXTERNAL_VECTOR; i < NR_VECTORS; i++) { /* SYSCALL_VECTOR was reserved in trap_init. */ - if (!test_bit(vector, used_vectors)) - set_intr_gate(vector, interrupt[i]); + if (i != SYSCALL_VECTOR) + set_intr_gate(i, interrupt[i]); } -#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_SMP) - /* - * IRQ0 must be given a fixed assignment and initialized, - * because it's used before the IO-APIC is set up. - */ - set_intr_gate(FIRST_DEVICE_VECTOR, interrupt[0]); +#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_SMP) /* * The reschedule interrupt is a CPU-to-CPU reschedule-helper * IPI, driven by wakeup. @@ -135,6 +148,9 @@ void __init native_init_IRQ(void) /* IPI for single call function */ set_intr_gate(CALL_FUNCTION_SINGLE_VECTOR, call_function_single_interrupt); + + /* Low priority IPI to cleanup after moving an irq */ + set_intr_gate(IRQ_MOVE_CLEANUP_VECTOR, irq_move_cleanup_interrupt); #endif #ifdef CONFIG_X86_LOCAL_APIC @@ -168,3 +184,4 @@ void __init native_init_IRQ(void) irq_ctx_init(smp_processor_id()); } + -- cgit v1.2.3 From 7a959cff725872ce9c3a534f10724d7bb2cb3c4a Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:32 -0700 Subject: x86: add debug info for 32bit sparse_irq so could figure out bugs where we get an interrupt, but vector_irq is not initialized yet. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 8 ++++++-- arch/x86/kernel/io_apic_64.c | 2 ++ arch/x86/kernel/irq_32.c | 4 ++-- 3 files changed, 10 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index ea33d3c74970..3001924bdd36 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -1114,8 +1114,12 @@ next: cfg->move_in_progress = 1; cfg->old_domain = cfg->domain; } - for_each_cpu_mask_nr(new_cpu, new_mask) - per_cpu(vector_irq, new_cpu)[vector] = irq; + printk(KERN_DEBUG "assign_irq_vector: irq %d vector %#x cpu ", irq, vector); + for_each_cpu_mask_nr(new_cpu, new_mask) { + per_cpu(vector_irq, new_cpu)[vector] = irq; + printk(KERN_CONT " %d ", new_cpu); + } + printk(KERN_CONT "\n"); cfg->vector = vector; cfg->domain = domain; return 0; diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index b0d4abc55a11..30d2e3811313 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -1394,6 +1394,8 @@ __apicdebuginit(void) print_IO_APIC(void) printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); + printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); + printk(KERN_DEBUG "....... : LTS : %X\n", reg_00.bits.LTS); printk(KERN_DEBUG ".... register #01: %08X\n", *(int *)®_01); printk(KERN_DEBUG "....... : max redirection entries: %04X\n", reg_01.bits.entries); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index b51ffdcfa31a..cc929f2f84f1 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -237,8 +237,8 @@ unsigned int do_IRQ(struct pt_regs *regs) desc = irq_to_desc(irq); if (unlikely(!desc)) { - printk(KERN_EMERG "%s: cannot handle IRQ %d vector %#x\n", - __func__, irq, vector); + printk(KERN_EMERG "%s: cannot handle IRQ %d vector %#x cpu %d\n", + __func__, irq, vector, smp_processor_id()); BUG(); } -- cgit v1.2.3 From d83e94acd95789829804fd9e442bd18975f4dc89 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:33 -0700 Subject: x86, io-apic: remove union about dest for log/phy let user decide the meaning of the bits. This unifies the 32-bit and 64-bit io-apic code a bit. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 18 +++++++----------- 1 file changed, 7 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 3001924bdd36..353e586822af 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -1224,7 +1224,7 @@ static int setup_ioapic_entry(int apic, int irq, entry->delivery_mode = INT_DELIVERY_MODE; entry->dest_mode = INT_DEST_MODE; - entry->dest.logical.logical_dest = destination; + entry->dest = destination; entry->mask = 0; /* enable IRQ */ entry->trigger = trigger; @@ -1336,7 +1336,7 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, */ entry.dest_mode = INT_DEST_MODE; entry.mask = 1; /* mask IRQ now */ - entry.dest.logical.logical_dest = cpu_mask_to_apicid(TARGET_CPUS); + entry.dest = cpu_mask_to_apicid(TARGET_CPUS); entry.delivery_mode = INT_DELIVERY_MODE; entry.polarity = 0; entry.trigger = 0; @@ -1425,19 +1425,15 @@ __apicdebuginit(void) print_IO_APIC(void) printk(KERN_DEBUG ".... IRQ redirection table:\n"); - printk(KERN_DEBUG " NR Log Phy Mask Trig IRR Pol" - " Stat Dest Deli Vect: \n"); + printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" + " Stat Dmod Deli Vect: \n"); for (i = 0; i <= reg_01.bits.entries; i++) { struct IO_APIC_route_entry entry; entry = ioapic_read_entry(apic, i); - printk(KERN_DEBUG " %02x %03X %02X ", - i, - entry.dest.logical.logical_dest, - entry.dest.physical.physical_dest - ); + printk(KERN_DEBUG " %02x %02X ", i, entry.dest); printk("%1d %1d %1d %1d %1d %1d %1d %02X\n", entry.mask, @@ -1717,7 +1713,7 @@ void disable_IO_APIC(void) entry.dest_mode = 0; /* Physical */ entry.delivery_mode = dest_ExtINT; /* ExtInt */ entry.vector = 0; - entry.dest.physical.physical_dest = read_apic_id(); + entry.dest = read_apic_id(); /* * Add it to the IO-APIC irq-routing table: @@ -2185,7 +2181,7 @@ static inline void __init unlock_ExtINT_logic(void) entry1.dest_mode = 0; /* physical delivery */ entry1.mask = 0; /* unmask IRQ now */ - entry1.dest.physical.physical_dest = hard_smp_processor_id(); + entry1.dest = hard_smp_processor_id(); entry1.delivery_mode = dest_ExtINT; entry1.polarity = entry0.polarity; entry1.trigger = 0; -- cgit v1.2.3 From 1d02519242c23450b043e5e8a9e3cb84a8666fe3 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:34 -0700 Subject: x86: ordering functions in io_apic_32.c prepare for unification: try to make functions be of the same order to io_apic_64.c. v2: add calling setup_msi_irq back to arch_setup_msi_irq Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 178 +++++++++++++++++++++++-------------------- 1 file changed, 94 insertions(+), 84 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 353e586822af..9531ef33362b 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -1029,23 +1029,6 @@ static int pin_2_irq(int idx, int apic, int pin) return irq; } -static inline int IO_APIC_irq_trigger(int irq) -{ - int apic, idx, pin; - - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - idx = find_irq_entry(apic, pin, mp_INT); - if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) - return irq_trigger(idx); - } - } - /* - * nonexistent IRQs are edge default - */ - return 0; -} - void lock_vector_lock(void) { /* Used to the online set of cpus does not change @@ -1190,6 +1173,23 @@ static struct irq_chip ioapic_chip; #define IOAPIC_EDGE 0 #define IOAPIC_LEVEL 1 +static inline int IO_APIC_irq_trigger(int irq) +{ + int apic, idx, pin; + + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + idx = find_irq_entry(apic, pin, mp_INT); + if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) + return irq_trigger(idx); + } + } + /* + * nonexistent IRQs are edge default + */ + return 0; +} + static void ioapic_register_intr(int irq, unsigned long trigger) { struct irq_desc *desc; @@ -1926,55 +1926,6 @@ static unsigned int startup_ioapic_irq(unsigned int irq) return was_pending; } -static void irq_complete_move(unsigned int irq); -static void ack_ioapic_irq(unsigned int irq) -{ - irq_complete_move(irq); - move_native_irq(irq); - ack_APIC_irq(); -} - -static void ack_ioapic_quirk_irq(unsigned int irq) -{ - unsigned long v; - int i; - - irq_complete_move(irq); - move_native_irq(irq); -/* - * It appears there is an erratum which affects at least version 0x11 - * of I/O APIC (that's the 82093AA and cores integrated into various - * chipsets). Under certain conditions a level-triggered interrupt is - * erroneously delivered as edge-triggered one but the respective IRR - * bit gets set nevertheless. As a result the I/O unit expects an EOI - * message but it will never arrive and further interrupts are blocked - * from the source. The exact reason is so far unknown, but the - * phenomenon was observed when two consecutive interrupt requests - * from a given source get delivered to the same CPU and the source is - * temporarily disabled in between. - * - * A workaround is to simulate an EOI message manually. We achieve it - * by setting the trigger mode to edge and then to level when the edge - * trigger mode gets detected in the TMR of a local APIC for a - * level-triggered interrupt. We mask the source for the time of the - * operation to prevent an edge-triggered interrupt escaping meanwhile. - * The idea is from Manfred Spraul. --macro - */ - i = irq_cfg(irq)->vector; - - v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); - - ack_APIC_irq(); - - if (!(v & (1 << (i & 0x1f)))) { - atomic_inc(&irq_mis_count); - spin_lock(&ioapic_lock); - __mask_and_edge_IO_APIC_irq(irq); - __unmask_and_level_IO_APIC_irq(irq); - spin_unlock(&ioapic_lock); - } -} - static int ioapic_retrigger_irq(unsigned int irq) { send_IPI_self(irq_cfg(irq)->vector); @@ -2040,13 +1991,61 @@ static void irq_complete_move(unsigned int irq) static inline void irq_complete_move(unsigned int irq) {} #endif +static void ack_apic_edge(unsigned int irq) +{ + irq_complete_move(irq); + move_native_irq(irq); + ack_APIC_irq(); +} + +static void ack_apic_level(unsigned int irq) +{ + unsigned long v; + int i; + + irq_complete_move(irq); + move_native_irq(irq); +/* + * It appears there is an erratum which affects at least version 0x11 + * of I/O APIC (that's the 82093AA and cores integrated into various + * chipsets). Under certain conditions a level-triggered interrupt is + * erroneously delivered as edge-triggered one but the respective IRR + * bit gets set nevertheless. As a result the I/O unit expects an EOI + * message but it will never arrive and further interrupts are blocked + * from the source. The exact reason is so far unknown, but the + * phenomenon was observed when two consecutive interrupt requests + * from a given source get delivered to the same CPU and the source is + * temporarily disabled in between. + * + * A workaround is to simulate an EOI message manually. We achieve it + * by setting the trigger mode to edge and then to level when the edge + * trigger mode gets detected in the TMR of a local APIC for a + * level-triggered interrupt. We mask the source for the time of the + * operation to prevent an edge-triggered interrupt escaping meanwhile. + * The idea is from Manfred Spraul. --macro + */ + i = irq_cfg(irq)->vector; + + v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); + + ack_APIC_irq(); + + if (!(v & (1 << (i & 0x1f)))) { + atomic_inc(&irq_mis_count); + spin_lock(&ioapic_lock); + __mask_and_edge_IO_APIC_irq(irq); + __unmask_and_level_IO_APIC_irq(irq); + spin_unlock(&ioapic_lock); + } +} + static struct irq_chip ioapic_chip __read_mostly = { .name = "IO-APIC", .startup = startup_ioapic_irq, .mask = mask_IO_APIC_irq, .unmask = unmask_IO_APIC_irq, - .ack = ack_ioapic_irq, - .eoi = ack_ioapic_quirk_irq, + .ack = ack_apic_edge, + .eoi = ack_apic_level, #ifdef CONFIG_SMP .set_affinity = set_ioapic_affinity_irq, #endif @@ -2094,11 +2093,6 @@ static inline void init_IO_APIC_traps(void) * The local APIC irq-chip implementation: */ -static void ack_lapic_irq(unsigned int irq) -{ - ack_APIC_irq(); -} - static void mask_lapic_irq(unsigned int irq) { unsigned long v; @@ -2115,6 +2109,11 @@ static void unmask_lapic_irq(unsigned int irq) apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); } +static void ack_lapic_irq(unsigned int irq) +{ + ack_APIC_irq(); +} + static struct irq_chip lapic_chip __read_mostly = { .name = "local-APIC", .mask = mask_lapic_irq, @@ -2636,13 +2635,31 @@ static struct irq_chip msi_chip = { .name = "PCI-MSI", .unmask = unmask_msi_irq, .mask = mask_msi_irq, - .ack = ack_ioapic_irq, + .ack = ack_apic_edge, #ifdef CONFIG_SMP .set_affinity = set_msi_irq_affinity, #endif .retrigger = ioapic_retrigger_irq, }; + +static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) +{ + int ret; + struct msi_msg msg; + + ret = msi_compose_msg(dev, irq, &msg); + if (ret < 0) + return ret; + + set_irq_msi(irq, desc); + write_msi_msg(irq, &msg); + + set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); + + return 0; +} + static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) { unsigned int irq; @@ -2657,7 +2674,6 @@ static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) { - struct msi_msg msg; int irq, ret; unsigned int irq_want; @@ -2669,17 +2685,11 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) if (irq == 0) return -1; - ret = msi_compose_msg(dev, irq, &msg); + ret = setup_msi_irq(dev, desc, irq); if (ret < 0) { destroy_irq(irq); return ret; - } - - set_irq_msi(irq, desc); - write_msi_msg(irq, &msg); - - set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, - "edge"); + } return 0; } @@ -2738,7 +2748,7 @@ static struct irq_chip ht_irq_chip = { .name = "PCI-HT", .mask = mask_ht_irq, .unmask = unmask_ht_irq, - .ack = ack_ioapic_irq, + .ack = ack_apic_edge, #ifdef CONFIG_SMP .set_affinity = set_ht_irq_affinity, #endif -- cgit v1.2.3 From 8ea5371baa82db452a8d93e9977b418d30944e32 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:35 -0700 Subject: x86: ordering functions in io_apic_64.c try to make functions have the same order between 32-bit and 64-bit. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 67 ++++++++++++++++++++++---------------------- 1 file changed, 33 insertions(+), 34 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 30d2e3811313..07e1e45ee026 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -409,40 +409,6 @@ static bool io_apic_level_ack_pending(unsigned int irq) return false; } -/* - * Synchronize the IO-APIC and the CPU by doing - * a dummy read from the IO-APIC - */ -static inline void io_apic_sync(unsigned int apic) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - readl(&io_apic->data); -} - -#define __DO_ACTION(R, ACTION, FINAL) \ - \ -{ \ - int pin; \ - struct irq_cfg *cfg; \ - struct irq_pin_list *entry; \ - \ - cfg = irq_cfg(irq); \ - entry = cfg->irq_2_pin; \ - for (;;) { \ - unsigned int reg; \ - if (!entry) \ - break; \ - pin = entry->pin; \ - reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ - reg ACTION; \ - io_apic_modify(entry->apic, reg); \ - FINAL; \ - if (!entry->next) \ - break; \ - entry = entry->next; \ - } \ -} - union entry_union { struct { u32 w1, w2; }; struct IO_APIC_route_entry entry; @@ -627,6 +593,39 @@ static void __init replace_pin_at_irq(unsigned int irq, add_pin_to_irq(irq, newapic, newpin); } +/* + * Synchronize the IO-APIC and the CPU by doing + * a dummy read from the IO-APIC + */ +static inline void io_apic_sync(unsigned int apic) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + readl(&io_apic->data); +} + +#define __DO_ACTION(R, ACTION, FINAL) \ + \ +{ \ + int pin; \ + struct irq_cfg *cfg; \ + struct irq_pin_list *entry; \ + \ + cfg = irq_cfg(irq); \ + entry = cfg->irq_2_pin; \ + for (;;) { \ + unsigned int reg; \ + if (!entry) \ + break; \ + pin = entry->pin; \ + reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ + reg ACTION; \ + io_apic_modify(entry->apic, reg); \ + FINAL; \ + if (!entry->next) \ + break; \ + entry = entry->next; \ + } \ +} #define DO_ACTION(name,R,ACTION, FINAL) \ \ -- cgit v1.2.3 From efa2559f65167989f1893cb065e3126d4f13ba60 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:36 -0700 Subject: x86: order variables in io_apic_xx.c move first_system_vector to apic_64.c. also add #ifdef CONFIG_INTR_REMAP to prepare 32 bit to use same file. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 5 ++ arch/x86/kernel/io_apic_32.c | 79 ++++++++++----------- arch/x86/kernel/io_apic_64.c | 160 +++++++++++++++++++++++-------------------- 3 files changed, 127 insertions(+), 117 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 94ddb69ae15e..4d7a188025b3 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -33,6 +33,7 @@ #include #include #include +#include #include #include #include @@ -59,6 +60,10 @@ int x2apic_preenabled; int local_apic_timer_c2_ok; EXPORT_SYMBOL_GPL(local_apic_timer_c2_ok); +int first_system_vector = 0xfe; + +char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; + /* * Debug level, exported for io_apic.c */ diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 9531ef33362b..3010bdd3352d 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -54,24 +54,22 @@ #define __apicdebuginit(type) static type __init -int (*ioapic_renumber_irq)(int ioapic, int irq); -atomic_t irq_mis_count; - -/* Where if anywhere is the i8259 connect in external int mode */ -static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; - -static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); - -int timer_through_8259 __initdata; - /* * Is the SiS APIC rmw bug present ? * -1 = don't know, 0 = no, 1 = yes */ int sis_apic_bug = -1; +static DEFINE_SPINLOCK(ioapic_lock); +static DEFINE_SPINLOCK(vector_lock); + int first_free_entry; +/* + * Rough estimation of how many shared IRQs there are, can + * be changed anytime. + */ +int pin_map_size; + /* * # of IRQ routing registers */ @@ -93,7 +91,15 @@ int mp_bus_id_to_type[MAX_MP_BUSSES]; DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); -static int disable_timer_pin_1 __initdata; +int skip_ioapic_setup; + +static int __init parse_noapic(char *arg) +{ + /* disable IO-APIC */ + disable_ioapic_setup(); + return 0; +} +early_param("noapic", parse_noapic); struct irq_cfg; struct irq_pin_list; @@ -268,13 +274,6 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) return cfg; } -static int assign_irq_vector(int irq, cpumask_t mask); -/* - * Rough estimation of how many shared IRQs there are, can - * be changed anytime. - */ -int pin_map_size; - /* * This is performance-critical, we want to do it O(1) * @@ -465,6 +464,9 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) entry = entry->next; } } + +static int assign_irq_vector(int irq, cpumask_t mask); + static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) { struct irq_cfg *cfg; @@ -677,7 +679,6 @@ void send_IPI_self(int vector) #define MAX_PIRQS 8 static int pirq_entries [MAX_PIRQS]; static int pirqs_enabled; -int skip_ioapic_setup; static int __init ioapic_pirq_setup(char *str) { @@ -981,6 +982,7 @@ static inline int irq_trigger(int idx) return MPBIOS_trigger(idx); } +int (*ioapic_renumber_irq)(int ioapic, int irq); static int pin_2_irq(int idx, int apic, int pin) { int irq, i; @@ -1621,6 +1623,9 @@ __apicdebuginit(int) print_all_ICs(void) fs_initcall(print_all_ICs); +/* Where if anywhere is the i8259 connect in external int mode */ +static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; + static void __init enable_IO_APIC(void) { union IO_APIC_reg_01 reg_01; @@ -1998,6 +2003,7 @@ static void ack_apic_edge(unsigned int irq) ack_APIC_irq(); } +atomic_t irq_mis_count; static void ack_apic_level(unsigned int irq) { unsigned long v; @@ -2208,6 +2214,17 @@ static inline void __init unlock_ExtINT_logic(void) ioapic_write_entry(apic, pin, entry0); } +static int disable_timer_pin_1 __initdata; + +static int __init parse_disable_timer_pin_1(char *arg) +{ + disable_timer_pin_1 = 1; + return 0; +} +early_param("disable_timer_pin_1", parse_disable_timer_pin_1); + +int timer_through_8259 __initdata; + /* * This code may look a bit paranoid, but it's supposed to cooperate with * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ @@ -2983,28 +3000,6 @@ void __init setup_ioapic_dest(void) } #endif -static int __init parse_disable_timer_pin_1(char *arg) -{ - disable_timer_pin_1 = 1; - return 0; -} -early_param("disable_timer_pin_1", parse_disable_timer_pin_1); - -static int __init parse_enable_timer_pin_1(char *arg) -{ - disable_timer_pin_1 = -1; - return 0; -} -early_param("enable_timer_pin_1", parse_enable_timer_pin_1); - -static int __init parse_noapic(char *arg) -{ - /* disable IO-APIC */ - disable_ioapic_setup(); - return 0; -} -early_param("noapic", parse_noapic); - void __init ioapic_init_mappings(void) { unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 07e1e45ee026..847aa7987645 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -57,6 +57,47 @@ #define __apicdebuginit(type) static type __init +int ioapic_force; + +int sis_apic_bug; /* not actually supported, dummy for compile */ + +static DEFINE_SPINLOCK(ioapic_lock); +static DEFINE_SPINLOCK(vector_lock); + +int first_free_entry; +/* + * Rough estimation of how many shared IRQs there are, can + * be changed anytime. + */ +int pin_map_size; + +/* + * # of IRQ routing registers + */ +int nr_ioapic_registers[MAX_IO_APICS]; + +/* I/O APIC entries */ +struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; +int nr_ioapics; + +/* MP IRQ source entries */ +struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; + +/* # of MP IRQ source entries */ +int mp_irq_entries; + +DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); + +int skip_ioapic_setup; + +static int __init parse_noapic(char *str) +{ + disable_ioapic_setup(); + return 0; +} +early_param("noapic", parse_noapic); + + struct irq_cfg; struct irq_pin_list; struct irq_cfg { @@ -228,53 +269,6 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) return cfg; } -static int assign_irq_vector(int irq, cpumask_t mask); - -int first_system_vector = 0xfe; - -char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; - -int sis_apic_bug; /* not actually supported, dummy for compile */ - -static int no_timer_check; - -static int disable_timer_pin_1 __initdata; - -int timer_through_8259 __initdata; - -/* Where if anywhere is the i8259 connect in external int mode */ -static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; - -static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); - -/* - * # of IRQ routing registers - */ -int nr_ioapic_registers[MAX_IO_APICS]; - -/* I/O APIC RTE contents at the OS boot up */ -struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; - -/* I/O APIC entries */ -struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; -int nr_ioapics; - -/* MP IRQ source entries */ -struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; - -/* # of MP IRQ source entries */ -int mp_irq_entries; - -DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); - -/* - * Rough estimation of how many shared IRQs there are, can - * be changed anytime. - */ - -int pin_map_size; - /* * This is performance-critical, we want to do it O(1) * @@ -481,12 +475,16 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) apic = entry->apic; pin = entry->pin; +#ifdef CONFIG_INTR_REMAP /* * With interrupt-remapping, destination information comes * from interrupt-remapping table entry. */ if (!irq_remapped(irq)) io_apic_write(apic, 0x11 + pin*2, dest); +#else + io_apic_write(apic, 0x11 + pin*2, dest); +#endif reg = io_apic_read(apic, 0x10 + pin*2); reg &= ~IO_APIC_REDIR_VECTOR_MASK; reg |= vector; @@ -497,6 +495,8 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) } } +static int assign_irq_vector(int irq, cpumask_t mask); + static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) { struct irq_cfg *cfg = irq_cfg(irq); @@ -533,7 +533,6 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -int first_free_entry; static void add_pin_to_irq(unsigned int irq, int apic, int pin) { struct irq_cfg *cfg; @@ -679,6 +678,10 @@ static void clear_IO_APIC (void) clear_IO_APIC_pin(apic, pin); } +#ifdef CONFIG_INTR_REMAP +/* I/O APIC RTE contents at the OS boot up */ +static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; + /* * Saves and masks all the unmasked IO-APIC RTE's */ @@ -741,25 +744,7 @@ void reinit_intr_remapped_IO_APIC(int intr_remapping) */ restore_IO_APIC_setup(); } - -int skip_ioapic_setup; -int ioapic_force; - -static int __init parse_noapic(char *str) -{ - disable_ioapic_setup(); - return 0; -} -early_param("noapic", parse_noapic); - -/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ -static int __init disable_timer_pin_setup(char *arg) -{ - disable_timer_pin_1 = 1; - return 1; -} -__setup("disable_timer_pin_1", disable_timer_pin_setup); - +#endif /* * Find the IRQ entry number of a certain pin. @@ -1327,8 +1312,10 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, { struct IO_APIC_route_entry entry; +#ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) return; +#endif memset(&entry, 0, sizeof(entry)); @@ -1601,6 +1588,9 @@ __apicdebuginit(int) print_all_ICs(void) fs_initcall(print_all_ICs); +/* Where if anywhere is the i8259 connect in external int mode */ +static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; + void __init enable_IO_APIC(void) { union IO_APIC_reg_01 reg_01; @@ -1695,6 +1685,15 @@ void disable_IO_APIC(void) disconnect_bsp_APIC(ioapic_i8259.pin != -1); } +static int no_timer_check; + +static int __init notimercheck(char *s) +{ + no_timer_check = 1; + return 1; +} +__setup("no_timer_check", notimercheck); + /* * There is a nasty bug in some older SMP boards, their mptable lies * about the timer IRQ. We do the following to work around the situation: @@ -1708,6 +1707,9 @@ static int __init timer_irq_works(void) unsigned long t1 = jiffies; unsigned long flags; + if (no_timer_check) + return 1; + local_save_flags(flags); local_irq_enable(); /* Let ten ticks pass... */ @@ -2239,6 +2241,17 @@ static inline void __init unlock_ExtINT_logic(void) ioapic_write_entry(apic, pin, entry0); } +static int disable_timer_pin_1 __initdata; +/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ +static int __init disable_timer_pin_setup(char *arg) +{ + disable_timer_pin_1 = 1; + return 0; +} +early_param("disable_timer_pin_1", disable_timer_pin_setup); + +int timer_through_8259 __initdata; + /* * This code may look a bit paranoid, but it's supposed to cooperate with * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ @@ -2286,8 +2299,10 @@ static inline void __init check_timer(void) * 8259A. */ if (pin1 == -1) { +#ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) panic("BIOS bug: timer not connected to IO-APIC"); +#endif pin1 = pin2; apic1 = apic2; no_pin1 = 1; @@ -2305,7 +2320,7 @@ static inline void __init check_timer(void) setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } unmask_IO_APIC_irq(0); - if (!no_timer_check && timer_irq_works()) { + if (timer_irq_works()) { if (nmi_watchdog == NMI_IO_APIC) { setup_nmi(); enable_8259A_irq(0); @@ -2314,8 +2329,10 @@ static inline void __init check_timer(void) clear_IO_APIC_pin(0, pin1); goto out; } +#ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) panic("timer doesn't work through Interrupt-remapped IO-APIC"); +#endif clear_IO_APIC_pin(apic1, pin1); if (!no_pin1) apic_printk(APIC_QUIET, KERN_ERR "..MP-BIOS bug: " @@ -2391,13 +2408,6 @@ out: local_irq_restore(flags); } -static int __init notimercheck(char *s) -{ - no_timer_check = 1; - return 1; -} -__setup("no_timer_check", notimercheck); - /* * Traditionally ISA IRQ2 is the cascade IRQ, and is not available * to devices. However there may be an I/O APIC pin available for -- cgit v1.2.3 From d4057bdb6a3bb85dd44f9f39f41eac53696fd637 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:38 -0700 Subject: x86: make headers files the same in io_apic_xx.c also make no_timer_check to be global on 64 bit, because vmi_32 is using that. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 15 ++++++++++++--- arch/x86/kernel/io_apic_64.c | 12 +++++++++--- 2 files changed, 21 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 3010bdd3352d..48184e126f2c 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -25,28 +25,37 @@ #include #include #include -#include +#include #include #include #include #include #include -#include #include #include #include #include -#include /* time_after() */ +#include /* time_after() */ +#ifdef CONFIG_ACPI +#include +#endif +#include +#include +#include #include #include #include +#include +#include +#include #include #include #include #include #include #include +#include #include #include diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 847aa7987645..3c66e5a8e899 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -27,12 +27,15 @@ #include #include #include +#include #include +#include #include #include #include -#include -#include +#include +#include +#include /* time_after() */ #ifdef CONFIG_ACPI #include #endif @@ -46,14 +49,17 @@ #include #include #include +#include #include #include #include #include +#include #include #include #include +#include #define __apicdebuginit(type) static type __init @@ -1685,7 +1691,7 @@ void disable_IO_APIC(void) disconnect_bsp_APIC(ioapic_i8259.pin != -1); } -static int no_timer_check; +int no_timer_check __initdata; static int __init notimercheck(char *s) { -- cgit v1.2.3 From f876d213a59c363d2492e399cc6c24edd6f3c368 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:39 -0700 Subject: x86: make 64 handle sis_apic_bug like the 32 bit do we have 64bit system with sis chipset? [ mingo@elte.hu: nope, the problem chipset was 32-bit only. The code symmetry is good nevertheless. ] Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_64.c | 28 ++++++++++++++++++++++++---- 1 file changed, 24 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 3c66e5a8e899..915af9b87aa4 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -65,7 +65,11 @@ int ioapic_force; -int sis_apic_bug; /* not actually supported, dummy for compile */ +/* + * Is the SiS APIC rmw bug present ? + * -1 = don't know, 0 = no, 1 = yes + */ +int sis_apic_bug = -1; static DEFINE_SPINLOCK(ioapic_lock); static DEFINE_SPINLOCK(vector_lock); @@ -373,9 +377,11 @@ static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned i * Re-write a value: to be used for read-modify-write * cycles where the read already set up the index register. */ -static inline void io_apic_modify(unsigned int apic, unsigned int value) +static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) { struct io_apic __iomem *io_apic = io_apic_base(apic); + if (sis_apic_bug) + writel(reg, &io_apic->index); writel(value, &io_apic->data); } @@ -494,7 +500,7 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) reg = io_apic_read(apic, 0x10 + pin*2); reg &= ~IO_APIC_REDIR_VECTOR_MASK; reg |= vector; - io_apic_modify(apic, reg); + io_apic_modify(apic, 0x10 + pin*2, reg); if (!entry->next) break; entry = entry->next; @@ -624,7 +630,7 @@ static inline void io_apic_sync(unsigned int apic) pin = entry->pin; \ reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ reg ACTION; \ - io_apic_modify(entry->apic, reg); \ + io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ FINAL; \ if (!entry->next) \ break; \ @@ -2450,6 +2456,20 @@ void __init setup_IO_APIC(void) check_timer(); } +/* + * Called after all the initialization is done. If we didnt find any + * APIC bugs then we can allow the modify fast path + */ + +static int __init io_apic_bug_finalize(void) +{ + if (sis_apic_bug == -1) + sis_apic_bug = 0; + return 0; +} + +late_initcall(io_apic_bug_finalize); + struct sysfs_ioapic_data { struct sys_device dev; struct IO_APIC_route_entry entry[0]; -- cgit v1.2.3 From aa45f97b1bb40adae1288669e73350907ffae85e Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:40 -0700 Subject: x86: remove ioapic_force no user left. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 1 - arch/x86/kernel/io_apic_64.c | 2 -- 2 files changed, 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 4d7a188025b3..cd860856a0b7 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -1813,7 +1813,6 @@ static int __init apic_set_verbosity(char *arg) if (!arg) { #ifdef CONFIG_X86_64 skip_ioapic_setup = 0; - ioapic_force = 1; return 0; #endif return -EINVAL; diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 915af9b87aa4..b70fd8232185 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -63,8 +63,6 @@ #define __apicdebuginit(type) static type __init -int ioapic_force; - /* * Is the SiS APIC rmw bug present ? * -1 = don't know, 0 = no, 1 = yes -- cgit v1.2.3 From 047c8fdb8718890e3340073b178d0859d0c7f91f Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:41 -0700 Subject: x86: make io_apic_64.c and io_apic_32.c the same all the same except INTR_REMAPPING related and ioapic io resource. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 213 +++++++++++++- arch/x86/kernel/io_apic_64.c | 663 ++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 829 insertions(+), 47 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 48184e126f2c..3ed36041c81e 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -123,7 +123,6 @@ struct irq_cfg { u8 move_in_progress : 1; }; - /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ static struct irq_cfg irq_cfg_legacy[] __initdata = { [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, @@ -391,6 +390,38 @@ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned writel(value, &io_apic->data); } +#ifdef CONFIG_X86_64 +static bool io_apic_level_ack_pending(unsigned int irq) +{ + struct irq_pin_list *entry; + unsigned long flags; + struct irq_cfg *cfg = irq_cfg(irq); + + spin_lock_irqsave(&ioapic_lock, flags); + entry = cfg->irq_2_pin; + for (;;) { + unsigned int reg; + int pin; + + if (!entry) + break; + pin = entry->pin; + reg = io_apic_read(entry->apic, 0x10 + pin*2); + /* Is the remote IRR bit set? */ + if (reg & IO_APIC_REDIR_REMOTE_IRR) { + spin_unlock_irqrestore(&ioapic_lock, flags); + return true; + } + if (!entry->next) + break; + entry = entry->next; + } + spin_unlock_irqrestore(&ioapic_lock, flags); + + return false; +} +#endif + union entry_union { struct { u32 w1, w2; }; struct IO_APIC_route_entry entry; @@ -483,17 +514,15 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) unsigned int dest; cpumask_t tmp; - cfg = irq_cfg(irq); - cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; + cfg = irq_cfg(irq); if (assign_irq_vector(irq, mask)) return; cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); /* * Only the high 8 bits are valid. @@ -572,6 +601,54 @@ static void __init replace_pin_at_irq(unsigned int irq, add_pin_to_irq(irq, newapic, newpin); } +#ifdef CONFIG_X86_64 +/* + * Synchronize the IO-APIC and the CPU by doing + * a dummy read from the IO-APIC + */ +static inline void io_apic_sync(unsigned int apic) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + readl(&io_apic->data); +} + +#define __DO_ACTION(R, ACTION, FINAL) \ + \ +{ \ + int pin; \ + struct irq_cfg *cfg; \ + struct irq_pin_list *entry; \ + \ + cfg = irq_cfg(irq); \ + entry = cfg->irq_2_pin; \ + for (;;) { \ + unsigned int reg; \ + if (!entry) \ + break; \ + pin = entry->pin; \ + reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ + reg ACTION; \ + io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ + FINAL; \ + if (!entry->next) \ + break; \ + entry = entry->next; \ + } \ +} + +#define DO_ACTION(name,R,ACTION, FINAL) \ + \ + static void name##_IO_APIC_irq (unsigned int irq) \ + __DO_ACTION(R, ACTION, FINAL) + +/* mask = 1 */ +DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) + +/* mask = 0 */ +DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) + +#else + static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) { struct irq_cfg *cfg; @@ -620,6 +697,8 @@ static void __unmask_and_level_IO_APIC_irq(unsigned int irq) IO_APIC_REDIR_MASKED); } +#endif + static void mask_IO_APIC_irq(unsigned int irq) { unsigned long flags; @@ -1055,6 +1134,17 @@ void unlock_vector_lock(void) static int __assign_irq_vector(int irq, cpumask_t mask) { + /* + * NOTE! The local APIC isn't very good at handling + * multiple interrupts at the same interrupt level. + * As the interrupt level is determined by taking the + * vector number and shifting that right by 4, we + * want to spread these out a bit so that they don't + * all fall in the same interrupt level. + * + * Also, we've got to be careful not to trash gate + * 0x80, because int 0x80 is hm, kind of importantish. ;) + */ static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; unsigned int old_vector; int cpu; @@ -1095,9 +1185,13 @@ next: } if (unlikely(current_vector == vector)) continue; - if (vector == SYSCALL_VECTOR) +#ifdef CONFIG_X86_64 + if (vector == IA32_SYSCALL_VECTOR) goto next; - +#else + if (vector == SYSCALL_VECTOR) + goto next; +#endif for_each_cpu_mask_nr(new_cpu, new_mask) if (per_cpu(vector_irq, new_cpu)[vector] != -1) goto next; @@ -1184,6 +1278,7 @@ static struct irq_chip ioapic_chip; #define IOAPIC_EDGE 0 #define IOAPIC_LEVEL 1 +#ifdef CONFIG_X86_32 static inline int IO_APIC_irq_trigger(int irq) { int apic, idx, pin; @@ -1200,6 +1295,12 @@ static inline int IO_APIC_irq_trigger(int irq) */ return 0; } +#else +static inline int IO_APIC_irq_trigger(int irq) +{ + return 1; +} +#endif static void ioapic_register_intr(int irq, unsigned long trigger) { @@ -1212,15 +1313,18 @@ static void ioapic_register_intr(int irq, unsigned long trigger) desc = irq_to_desc_alloc(irq); if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || - trigger == IOAPIC_LEVEL) { + trigger == IOAPIC_LEVEL) desc->status |= IRQ_LEVEL; + else + desc->status &= ~IRQ_LEVEL; + + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || + trigger == IOAPIC_LEVEL) set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_fasteoi_irq, "fasteoi"); - } else { - desc->status &= ~IRQ_LEVEL; + else set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_edge_irq, "edge"); - } } static int setup_ioapic_entry(int apic, int irq, @@ -1662,7 +1766,6 @@ static void __init enable_IO_APIC(void) struct IO_APIC_route_entry entry; entry = ioapic_read_entry(apic, pin); - /* If the interrupt line is enabled and in ExtInt mode * I have found the pin where the i8259 is connected. */ @@ -2012,6 +2115,60 @@ static void ack_apic_edge(unsigned int irq) ack_APIC_irq(); } +#ifdef CONFIG_X86_64 +static void ack_apic_level(unsigned int irq) +{ + int do_unmask_irq = 0; + + irq_complete_move(irq); +#ifdef CONFIG_GENERIC_PENDING_IRQ + /* If we are moving the irq we need to mask it */ + if (unlikely(desc->status & IRQ_MOVE_PENDING)) { + do_unmask_irq = 1; + mask_IO_APIC_irq(irq); + } +#endif + + /* + * We must acknowledge the irq before we move it or the acknowledge will + * not propagate properly. + */ + ack_APIC_irq(); + + /* Now we can move and renable the irq */ + if (unlikely(do_unmask_irq)) { + /* Only migrate the irq if the ack has been received. + * + * On rare occasions the broadcast level triggered ack gets + * delayed going to ioapics, and if we reprogram the + * vector while Remote IRR is still set the irq will never + * fire again. + * + * To prevent this scenario we read the Remote IRR bit + * of the ioapic. This has two effects. + * - On any sane system the read of the ioapic will + * flush writes (and acks) going to the ioapic from + * this cpu. + * - We get to see if the ACK has actually been delivered. + * + * Based on failed experiments of reprogramming the + * ioapic entry from outside of irq context starting + * with masking the ioapic entry and then polling until + * Remote IRR was clear before reprogramming the + * ioapic I don't trust the Remote IRR bit to be + * completey accurate. + * + * However there appears to be no other way to plug + * this race, so if the Remote IRR bit is not + * accurate and is causing problems then it is a hardware bug + * and you can go talk to the chipset vendor about it. + */ + if (!io_apic_level_ack_pending(irq)) + move_masked_irq(irq, desc); + unmask_IO_APIC_irq(irq); + } +} +#else atomic_t irq_mis_count; static void ack_apic_level(unsigned int irq) { @@ -2053,6 +2210,7 @@ static void ack_apic_level(unsigned int irq) spin_unlock(&ioapic_lock); } } +#endif static struct irq_chip ioapic_chip __read_mostly = { .name = "IO-APIC", @@ -2224,7 +2382,7 @@ static inline void __init unlock_ExtINT_logic(void) } static int disable_timer_pin_1 __initdata; - +/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ static int __init parse_disable_timer_pin_1(char *arg) { disable_timer_pin_1 = 1; @@ -2244,9 +2402,9 @@ static inline void __init check_timer(void) { struct irq_cfg *cfg = irq_cfg(0); int apic1, pin1, apic2, pin2; - int no_pin1 = 0; - unsigned int ver; unsigned long flags; + unsigned int ver; + int no_pin1 = 0; local_irq_save(flags); @@ -2550,6 +2708,7 @@ unsigned int create_irq_nr(unsigned int irq_want) cfg_new = irq_cfg(new); if (cfg_new && cfg_new->vector != 0) continue; + /* check if need to create one */ if (!cfg_new) cfg_new = irq_cfg_alloc(new); if (__assign_irq_vector(new, TARGET_CPUS) == 0) @@ -2720,6 +2879,32 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) return 0; } +int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) +{ + unsigned int irq; + int ret, sub_handle; + struct msi_desc *desc; + unsigned int irq_want; + + irq_want = build_irq_for_pci_dev(dev) + 0x100; + sub_handle = 0; + list_for_each_entry(desc, &dev->msi_list, list) { + irq = create_irq_nr(irq_want--); + if (irq == 0) + return -1; + ret = setup_msi_irq(dev, desc, irq); + if (ret < 0) + goto error; + sub_handle++; + } + return 0; + +error: + destroy_irq(irq); + return ret; +} + + void arch_teardown_msi_irq(unsigned int irq) { destroy_irq(irq); diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index b70fd8232185..940c4167b325 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -94,18 +94,22 @@ struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; /* # of MP IRQ source entries */ int mp_irq_entries; +#if defined (CONFIG_MCA) || defined (CONFIG_EISA) +int mp_bus_id_to_type[MAX_MP_BUSSES]; +#endif + DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); int skip_ioapic_setup; static int __init parse_noapic(char *str) { + /* disable IO-APIC */ disable_ioapic_setup(); return 0; } early_param("noapic", parse_noapic); - struct irq_cfg; struct irq_pin_list; struct irq_cfg { @@ -374,6 +378,8 @@ static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned i /* * Re-write a value: to be used for read-modify-write * cycles where the read already set up the index register. + * + * Older SiS APIC requires we rewrite the index register */ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) { @@ -383,6 +389,7 @@ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned writel(value, &io_apic->data); } +#ifdef CONFIG_X86_64 static bool io_apic_level_ack_pending(unsigned int irq) { struct irq_pin_list *entry; @@ -412,6 +419,7 @@ static bool io_apic_level_ack_pending(unsigned int irq) return false; } +#endif union entry_union { struct { u32 w1, w2; }; @@ -509,7 +517,7 @@ static int assign_irq_vector(int irq, cpumask_t mask); static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg(irq); + struct irq_cfg *cfg; unsigned long flags; unsigned int dest; cpumask_t tmp; @@ -519,12 +527,12 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) if (cpus_empty(tmp)) return; + cfg = irq_cfg(irq); if (assign_irq_vector(irq, mask)) return; cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); - /* * Only the high 8 bits are valid. */ @@ -536,7 +544,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) desc->affinity = mask; spin_unlock_irqrestore(&ioapic_lock, flags); } -#endif +#endif /* CONFIG_SMP */ /* * The common case is 1:1 IRQ<->pin mappings. Sometimes there are @@ -602,6 +610,7 @@ static void __init replace_pin_at_irq(unsigned int irq, add_pin_to_irq(irq, newapic, newpin); } +#ifdef CONFIG_X86_64 /* * Synchronize the IO-APIC and the CPU by doing * a dummy read from the IO-APIC @@ -647,6 +656,58 @@ DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) /* mask = 0 */ DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) +#else + +static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) +{ + struct irq_cfg *cfg; + struct irq_pin_list *entry; + unsigned int pin, reg; + + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; + for (;;) { + if (!entry) + break; + pin = entry->pin; + reg = io_apic_read(entry->apic, 0x10 + pin*2); + reg &= ~disable; + reg |= enable; + io_apic_modify(entry->apic, 0x10 + pin*2, reg); + if (!entry->next) + break; + entry = entry->next; + } +} + +/* mask = 1 */ +static void __mask_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, 0); +} + +/* mask = 0 */ +static void __unmask_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, 0, IO_APIC_REDIR_MASKED); +} + +/* mask = 1, trigger = 0 */ +static void __mask_and_edge_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, + IO_APIC_REDIR_LEVEL_TRIGGER); +} + +/* mask = 0, trigger = 1 */ +static void __unmask_and_level_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_LEVEL_TRIGGER, + IO_APIC_REDIR_MASKED); +} + +#endif + static void mask_IO_APIC_irq (unsigned int irq) { unsigned long flags; @@ -688,6 +749,64 @@ static void clear_IO_APIC (void) clear_IO_APIC_pin(apic, pin); } +#if !defined(CONFIG_SMP) && defined(CONFIG_X86_32) +void send_IPI_self(int vector) +{ + unsigned int cfg; + + /* + * Wait for idle. + */ + apic_wait_icr_idle(); + cfg = APIC_DM_FIXED | APIC_DEST_SELF | vector | APIC_DEST_LOGICAL; + /* + * Send the IPI. The write to APIC_ICR fires this off. + */ + apic_write(APIC_ICR, cfg); +} +#endif /* !CONFIG_SMP && CONFIG_X86_32*/ + +#ifdef CONFIG_X86_32 +/* + * support for broken MP BIOSs, enables hand-redirection of PIRQ0-7 to + * specific CPU-side IRQs. + */ + +#define MAX_PIRQS 8 +static int pirq_entries [MAX_PIRQS]; +static int pirqs_enabled; + +static int __init ioapic_pirq_setup(char *str) +{ + int i, max; + int ints[MAX_PIRQS+1]; + + get_options(str, ARRAY_SIZE(ints), ints); + + for (i = 0; i < MAX_PIRQS; i++) + pirq_entries[i] = -1; + + pirqs_enabled = 1; + apic_printk(APIC_VERBOSE, KERN_INFO + "PIRQ redirection, working around broken MP-BIOS.\n"); + max = MAX_PIRQS; + if (ints[0] < MAX_PIRQS) + max = ints[0]; + + for (i = 0; i < max; i++) { + apic_printk(APIC_VERBOSE, KERN_DEBUG + "... PIRQ%d -> IRQ %d\n", i, ints[i+1]); + /* + * PIRQs are mapped upside down, usually. + */ + pirq_entries[MAX_PIRQS-i-1] = ints[i+1]; + } + return 1; +} + +__setup("pirq=", ioapic_pirq_setup); +#endif /* CONFIG_X86_32 */ + #ifdef CONFIG_INTR_REMAP /* I/O APIC RTE contents at the OS boot up */ static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; @@ -861,18 +980,51 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) return best_guess; } +EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); + +#if defined(CONFIG_EISA) || defined(CONFIG_MCA) +/* + * EISA Edge/Level control register, ELCR + */ +static int EISA_ELCR(unsigned int irq) +{ + if (irq < 16) { + unsigned int port = 0x4d0 + (irq >> 3); + return (inb(port) >> (irq & 7)) & 1; + } + apic_printk(APIC_VERBOSE, KERN_INFO + "Broken MPtable reports ISA irq %d\n", irq); + return 0; +} + +#endif + /* ISA interrupts are always polarity zero edge triggered, * when listed as conforming in the MP table. */ #define default_ISA_trigger(idx) (0) #define default_ISA_polarity(idx) (0) +/* EISA interrupts are always polarity zero and can be edge or level + * trigger depending on the ELCR value. If an interrupt is listed as + * EISA conforming in the MP table, that means its trigger type must + * be read in from the ELCR */ + +#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].mp_srcbusirq)) +#define default_EISA_polarity(idx) default_ISA_polarity(idx) + /* PCI interrupts are always polarity one level triggered, * when listed as conforming in the MP table. */ #define default_PCI_trigger(idx) (1) #define default_PCI_polarity(idx) (1) +/* MCA interrupts are always polarity zero level triggered, + * when listed as conforming in the MP table. */ + +#define default_MCA_trigger(idx) (1) +#define default_MCA_polarity(idx) default_ISA_polarity(idx) + static int MPBIOS_polarity(int idx) { int bus = mp_irqs[idx].mp_srcbus; @@ -930,6 +1082,36 @@ static int MPBIOS_trigger(int idx) trigger = default_ISA_trigger(idx); else trigger = default_PCI_trigger(idx); +#if defined(CONFIG_EISA) || defined(CONFIG_MCA) + switch (mp_bus_id_to_type[bus]) { + case MP_BUS_ISA: /* ISA pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_EISA: /* EISA pin */ + { + trigger = default_EISA_trigger(idx); + break; + } + case MP_BUS_PCI: /* PCI pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_MCA: /* MCA pin */ + { + trigger = default_MCA_trigger(idx); + break; + } + default: + { + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 1; + break; + } + } +#endif break; case 1: /* edge */ { @@ -967,6 +1149,7 @@ static inline int irq_trigger(int idx) return MPBIOS_trigger(idx); } +int (*ioapic_renumber_irq)(int ioapic, int irq); static int pin_2_irq(int idx, int apic, int pin) { int irq, i; @@ -988,7 +1171,32 @@ static int pin_2_irq(int idx, int apic, int pin) while (i < apic) irq += nr_ioapic_registers[i++]; irq += pin; + /* + * For MPS mode, so far only needed by ES7000 platform + */ + if (ioapic_renumber_irq) + irq = ioapic_renumber_irq(apic, irq); } + +#ifdef CONFIG_X86_32 + /* + * PCI IRQ command line redirection. Yes, limits are hardcoded. + */ + if ((pin >= 16) && (pin <= 23)) { + if (pirq_entries[pin-16] != -1) { + if (!pirq_entries[pin-16]) { + apic_printk(APIC_VERBOSE, KERN_DEBUG + "disabling PIRQ%d\n", pin-16); + } else { + irq = pirq_entries[pin-16]; + apic_printk(APIC_VERBOSE, KERN_DEBUG + "using PIRQ%d -> IRQ %d\n", + pin-16, irq); + } + } + } +#endif + return irq; } @@ -1058,8 +1266,13 @@ next: } if (unlikely(current_vector == vector)) continue; +#ifdef CONFIG_X86_64 if (vector == IA32_SYSCALL_VECTOR) goto next; +#else + if (vector == SYSCALL_VECTOR) + goto next; +#endif for_each_cpu_mask_nr(new_cpu, new_mask) if (per_cpu(vector_irq, new_cpu)[vector] != -1) goto next; @@ -1140,6 +1353,34 @@ static struct irq_chip ioapic_chip; static struct irq_chip ir_ioapic_chip; #endif +#define IOAPIC_AUTO -1 +#define IOAPIC_EDGE 0 +#define IOAPIC_LEVEL 1 + +#ifdef CONFIG_X86_32 +static inline int IO_APIC_irq_trigger(int irq) +{ + int apic, idx, pin; + + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + idx = find_irq_entry(apic, pin, mp_INT); + if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) + return irq_trigger(idx); + } + } + /* + * nonexistent IRQs are edge default + */ + return 0; +} +#else +static inline int IO_APIC_irq_trigger(int irq) +{ + return 1; +} +#endif + static void ioapic_register_intr(int irq, unsigned long trigger) { struct irq_desc *desc; @@ -1150,7 +1391,8 @@ static void ioapic_register_intr(int irq, unsigned long trigger) else desc = irq_to_desc_alloc(irq); - if (trigger) + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || + trigger == IOAPIC_LEVEL) desc->status |= IRQ_LEVEL; else desc->status &= ~IRQ_LEVEL; @@ -1168,7 +1410,8 @@ static void ioapic_register_intr(int irq, unsigned long trigger) return; } #endif - if (trigger) + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || + trigger == IOAPIC_LEVEL) set_irq_chip_and_handler_name(irq, &ioapic_chip, handle_fasteoi_irq, "fasteoi"); @@ -1303,6 +1546,10 @@ static void __init setup_IO_APIC_irqs(void) } irq = pin_2_irq(idx, apic, pin); +#ifdef CONFIG_X86_32 + if (multi_timer_check(apic, irq)) + continue; +#endif add_pin_to_irq(irq, apic, pin); setup_IO_APIC_irq(apic, pin, irq, @@ -1360,6 +1607,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_00 reg_00; union IO_APIC_reg_01 reg_01; union IO_APIC_reg_02 reg_02; + union IO_APIC_reg_03 reg_03; unsigned long flags; struct irq_cfg *cfg; @@ -1384,6 +1632,8 @@ __apicdebuginit(void) print_IO_APIC(void) reg_01.raw = io_apic_read(apic, 1); if (reg_01.bits.version >= 0x10) reg_02.raw = io_apic_read(apic, 2); + if (reg_01.bits.version >= 0x20) + reg_03.raw = io_apic_read(apic, 3); spin_unlock_irqrestore(&ioapic_lock, flags); printk("\n"); @@ -1399,11 +1649,27 @@ __apicdebuginit(void) print_IO_APIC(void) printk(KERN_DEBUG "....... : PRQ implemented: %X\n", reg_01.bits.PRQ); printk(KERN_DEBUG "....... : IO APIC version: %04X\n", reg_01.bits.version); - if (reg_01.bits.version >= 0x10) { + /* + * Some Intel chipsets with IO APIC VERSION of 0x1? don't have reg_02, + * but the value of reg_02 is read as the previous read register + * value, so ignore it if reg_02 == reg_01. + */ + if (reg_01.bits.version >= 0x10 && reg_02.raw != reg_01.raw) { printk(KERN_DEBUG ".... register #02: %08X\n", reg_02.raw); printk(KERN_DEBUG "....... : arbitration: %02X\n", reg_02.bits.arbitration); } + /* + * Some Intel chipsets with IO APIC VERSION of 0x2? don't have reg_02 + * or reg_03, but the value of reg_0[23] is read as the previous read + * register value, so ignore it if reg_03 == reg_0[12]. + */ + if (reg_01.bits.version >= 0x20 && reg_03.raw != reg_02.raw && + reg_03.raw != reg_01.raw) { + printk(KERN_DEBUG ".... register #03: %08X\n", reg_03.raw); + printk(KERN_DEBUG "....... : Boot DT : %X\n", reg_03.bits.boot_DT); + } + printk(KERN_DEBUG ".... IRQ redirection table:\n"); printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" @@ -1475,7 +1741,7 @@ __apicdebuginit(void) print_APIC_bitfield(int base) __apicdebuginit(void) print_local_APIC(void *dummy) { unsigned int v, ver, maxlvt; - unsigned long icr; + u64 icr; if (apic_verbosity == APIC_QUIET) return; @@ -1492,11 +1758,13 @@ __apicdebuginit(void) print_local_APIC(void *dummy) v = apic_read(APIC_TASKPRI); printk(KERN_DEBUG "... APIC TASKPRI: %08x (%02x)\n", v, v & APIC_TPRI_MASK); - v = apic_read(APIC_ARBPRI); - printk(KERN_DEBUG "... APIC ARBPRI: %08x (%02x)\n", v, - v & APIC_ARBPRI_MASK); - v = apic_read(APIC_PROCPRI); - printk(KERN_DEBUG "... APIC PROCPRI: %08x\n", v); + if (APIC_INTEGRATED(ver)) { /* !82489DX */ + v = apic_read(APIC_ARBPRI); + printk(KERN_DEBUG "... APIC ARBPRI: %08x (%02x)\n", v, + v & APIC_ARBPRI_MASK); + v = apic_read(APIC_PROCPRI); + printk(KERN_DEBUG "... APIC PROCPRI: %08x\n", v); + } v = apic_read(APIC_EOI); printk(KERN_DEBUG "... APIC EOI: %08x\n", v); @@ -1516,8 +1784,13 @@ __apicdebuginit(void) print_local_APIC(void *dummy) printk(KERN_DEBUG "... APIC IRR field:\n"); print_APIC_bitfield(APIC_IRR); - v = apic_read(APIC_ESR); - printk(KERN_DEBUG "... APIC ESR: %08x\n", v); + if (APIC_INTEGRATED(ver)) { /* !82489DX */ + if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ + apic_write(APIC_ESR, 0); + + v = apic_read(APIC_ESR); + printk(KERN_DEBUG "... APIC ESR: %08x\n", v); + } icr = apic_icr_read(); printk(KERN_DEBUG "... APIC ICR: %08x\n", (u32)icr); @@ -1608,6 +1881,13 @@ void __init enable_IO_APIC(void) int apic; unsigned long flags; +#ifdef CONFIG_X86_32 + int i; + if (!pirqs_enabled) + for (i = 0; i < MAX_PIRQS; i++) + pirq_entries[i] = -1; +#endif + /* * The number of IO-APIC IRQ registers (== #pins): */ @@ -1636,6 +1916,10 @@ void __init enable_IO_APIC(void) } found_i8259: /* Look to see what if the MP table has reported the ExtINT */ + /* If we could not find the appropriate pin by looking at the ioapic + * the i8259 probably is not connected the ioapic but give the + * mptable a chance anyway. + */ i8259_pin = find_isa_irq_pin(0, mp_ExtINT); i8259_apic = find_isa_irq_apic(0, mp_ExtINT); /* Trust the MP table if nothing is setup in the hardware */ @@ -1695,6 +1979,122 @@ void disable_IO_APIC(void) disconnect_bsp_APIC(ioapic_i8259.pin != -1); } +#ifdef CONFIG_X86_32 +/* + * function to set the IO-APIC physical IDs based on the + * values stored in the MPC table. + * + * by Matt Domsch Tue Dec 21 12:25:05 CST 1999 + */ + +static void __init setup_ioapic_ids_from_mpc(void) +{ + union IO_APIC_reg_00 reg_00; + physid_mask_t phys_id_present_map; + int apic; + int i; + unsigned char old_id; + unsigned long flags; + + if (x86_quirks->setup_ioapic_ids && x86_quirks->setup_ioapic_ids()) + return; + + /* + * Don't check I/O APIC IDs for xAPIC systems. They have + * no meaning without the serial APIC bus. + */ + if (!(boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) + || APIC_XAPIC(apic_version[boot_cpu_physical_apicid])) + return; + /* + * This is broken; anything with a real cpu count has to + * circumvent this idiocy regardless. + */ + phys_id_present_map = ioapic_phys_id_map(phys_cpu_present_map); + + /* + * Set the IOAPIC ID to the value stored in the MPC table. + */ + for (apic = 0; apic < nr_ioapics; apic++) { + + /* Read the register 0 value */ + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(apic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + old_id = mp_ioapics[apic].mp_apicid; + + if (mp_ioapics[apic].mp_apicid >= get_physical_broadcast()) { + printk(KERN_ERR "BIOS bug, IO-APIC#%d ID is %d in the MPC table!...\n", + apic, mp_ioapics[apic].mp_apicid); + printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", + reg_00.bits.ID); + mp_ioapics[apic].mp_apicid = reg_00.bits.ID; + } + + /* + * Sanity check, is the ID really free? Every APIC in a + * system must have a unique ID or we get lots of nice + * 'stuck on smp_invalidate_needed IPI wait' messages. + */ + if (check_apicid_used(phys_id_present_map, + mp_ioapics[apic].mp_apicid)) { + printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n", + apic, mp_ioapics[apic].mp_apicid); + for (i = 0; i < get_physical_broadcast(); i++) + if (!physid_isset(i, phys_id_present_map)) + break; + if (i >= get_physical_broadcast()) + panic("Max APIC ID exceeded!\n"); + printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", + i); + physid_set(i, phys_id_present_map); + mp_ioapics[apic].mp_apicid = i; + } else { + physid_mask_t tmp; + tmp = apicid_to_cpu_present(mp_ioapics[apic].mp_apicid); + apic_printk(APIC_VERBOSE, "Setting %d in the " + "phys_id_present_map\n", + mp_ioapics[apic].mp_apicid); + physids_or(phys_id_present_map, phys_id_present_map, tmp); + } + + + /* + * We need to adjust the IRQ routing table + * if the ID changed. + */ + if (old_id != mp_ioapics[apic].mp_apicid) + for (i = 0; i < mp_irq_entries; i++) + if (mp_irqs[i].mp_dstapic == old_id) + mp_irqs[i].mp_dstapic + = mp_ioapics[apic].mp_apicid; + + /* + * Read the right value from the MPC table and + * write it into the ID register. + */ + apic_printk(APIC_VERBOSE, KERN_INFO + "...changing IO-APIC physical APIC ID to %d ...", + mp_ioapics[apic].mp_apicid); + + reg_00.bits.ID = mp_ioapics[apic].mp_apicid; + spin_lock_irqsave(&ioapic_lock, flags); + + /* + * Sanity check + */ + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(apic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + if (reg_00.bits.ID != mp_ioapics[apic].mp_apicid) + printk("could not set ID!\n"); + else + apic_printk(APIC_VERBOSE, " ok.\n"); + } +} +#endif + int no_timer_check __initdata; static int __init notimercheck(char *s) @@ -1780,8 +2180,10 @@ static unsigned int startup_ioapic_irq(unsigned int irq) return was_pending; } +#ifdef CONFIG_X86_64 static int ioapic_retrigger_irq(unsigned int irq) { + struct irq_cfg *cfg = irq_cfg(irq); unsigned long flags; @@ -1791,6 +2193,14 @@ static int ioapic_retrigger_irq(unsigned int irq) return 1; } +#else +static int ioapic_retrigger_irq(unsigned int irq) +{ + send_IPI_self(irq_cfg(irq)->vector); + + return 1; +} +#endif /* * Level and edge triggered IO-APIC interrupts need different handling, @@ -1952,7 +2362,9 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) { unsigned vector, me; ack_APIC_irq(); +#ifdef CONFIG_X86_64 exit_idle(); +#endif irq_enter(); me = smp_processor_id(); @@ -2024,6 +2436,7 @@ static void ack_apic_edge(unsigned int irq) ack_APIC_irq(); } +#ifdef CONFIG_X86_64 static void ack_apic_level(unsigned int irq) { int do_unmask_irq = 0; @@ -2076,6 +2489,49 @@ static void ack_apic_level(unsigned int irq) unmask_IO_APIC_irq(irq); } } +#else +atomic_t irq_mis_count; +static void ack_apic_level(unsigned int irq) +{ + unsigned long v; + int i; + + irq_complete_move(irq); + move_native_irq(irq); + /* + * It appears there is an erratum which affects at least version 0x11 + * of I/O APIC (that's the 82093AA and cores integrated into various + * chipsets). Under certain conditions a level-triggered interrupt is + * erroneously delivered as edge-triggered one but the respective IRR + * bit gets set nevertheless. As a result the I/O unit expects an EOI + * message but it will never arrive and further interrupts are blocked + * from the source. The exact reason is so far unknown, but the + * phenomenon was observed when two consecutive interrupt requests + * from a given source get delivered to the same CPU and the source is + * temporarily disabled in between. + * + * A workaround is to simulate an EOI message manually. We achieve it + * by setting the trigger mode to edge and then to level when the edge + * trigger mode gets detected in the TMR of a local APIC for a + * level-triggered interrupt. We mask the source for the time of the + * operation to prevent an edge-triggered interrupt escaping meanwhile. + * The idea is from Manfred Spraul. --macro + */ + i = irq_cfg(irq)->vector; + + v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); + + ack_APIC_irq(); + + if (!(v & (1 << (i & 0x1f)))) { + atomic_inc(&irq_mis_count); + spin_lock(&ioapic_lock); + __mask_and_edge_IO_APIC_irq(irq); + __unmask_and_level_IO_APIC_irq(irq); + spin_unlock(&ioapic_lock); + } +} +#endif static struct irq_chip ioapic_chip __read_mostly = { .name = "IO-APIC", @@ -2141,20 +2597,24 @@ static inline void init_IO_APIC_traps(void) } } -static void unmask_lapic_irq(unsigned int irq) +/* + * The local APIC irq-chip implementation: + */ + +static void mask_lapic_irq(unsigned int irq) { unsigned long v; v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); + apic_write(APIC_LVT0, v | APIC_LVT_MASKED); } -static void mask_lapic_irq(unsigned int irq) +static void unmask_lapic_irq(unsigned int irq) { unsigned long v; v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v | APIC_LVT_MASKED); + apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); } static void ack_lapic_irq (unsigned int irq) @@ -2182,19 +2642,19 @@ static void lapic_register_intr(int irq) static void __init setup_nmi(void) { /* - * Dirty trick to enable the NMI watchdog ... + * Dirty trick to enable the NMI watchdog ... * We put the 8259A master into AEOI mode and * unmask on all local APICs LVT0 as NMI. * * The idea to use the 8259A in AEOI mode ('8259A Virtual Wire') * is from Maciej W. Rozycki - so we do not have to EOI from * the NMI handler or the timer interrupt. - */ - printk(KERN_INFO "activating NMI Watchdog ..."); + */ + apic_printk(APIC_VERBOSE, KERN_INFO "activating NMI Watchdog ..."); enable_NMI_through_LVT0(); - printk(" done.\n"); + apic_printk(APIC_VERBOSE, " done.\n"); } /* @@ -2211,12 +2671,17 @@ static inline void __init unlock_ExtINT_logic(void) unsigned char save_control, save_freq_select; pin = find_isa_irq_pin(8, mp_INT); + if (pin == -1) { + WARN_ON_ONCE(1); + return; + } apic = find_isa_irq_apic(8, mp_INT); - if (pin == -1) + if (apic == -1) { + WARN_ON_ONCE(1); return; + } entry0 = ioapic_read_entry(apic, pin); - clear_IO_APIC_pin(apic, pin); memset(&entry1, 0, sizeof(entry1)); @@ -2268,17 +2733,21 @@ int timer_through_8259 __initdata; * is so screwy. Thanks to Brian Perkins for testing/hacking this beast * fanatically on his truly buggy board. * - * FIXME: really need to revamp this for modern platforms only. + * FIXME: really need to revamp this for all platforms. */ static inline void __init check_timer(void) { struct irq_cfg *cfg = irq_cfg(0); int apic1, pin1, apic2, pin2; unsigned long flags; + unsigned int ver; int no_pin1 = 0; local_irq_save(flags); + ver = apic_read(APIC_LVR); + ver = GET_APIC_VERSION(ver); + /* * get/set the timer IRQ vector: */ @@ -2287,10 +2756,18 @@ static inline void __init check_timer(void) /* * As IRQ0 is to be enabled in the 8259A, the virtual - * wire has to be disabled in the local APIC. + * wire has to be disabled in the local APIC. Also + * timer interrupts need to be acknowledged manually in + * the 8259A for the i82489DX when using the NMI + * watchdog as that APIC treats NMIs as level-triggered. + * The AEOI mode will finish them in the 8259A + * automatically. */ apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); init_8259A(1); +#ifdef CONFIG_X86_32 + timer_ack = (nmi_watchdog == NMI_IO_APIC && !APIC_INTEGRATED(ver)); +#endif pin1 = find_isa_irq_pin(0, mp_INT); apic1 = find_isa_irq_apic(0, mp_INT); @@ -2382,6 +2859,9 @@ static inline void __init check_timer(void) "through the IO-APIC - disabling NMI Watchdog!\n"); nmi_watchdog = NMI_NONE; } +#ifdef CONFIG_X86_32 + timer_ack = 0; +#endif apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as Virtual Wire IRQ...\n"); @@ -2435,19 +2915,29 @@ out: * the I/O APIC in all cases now. No actual device should request * it anyway. --macro */ -#define PIC_IRQS (1<<2) +#define PIC_IRQS (1 << PIC_CASCADE_IR) void __init setup_IO_APIC(void) { +#ifdef CONFIG_X86_32 + enable_IO_APIC(); +#else /* * calling enable_IO_APIC() is moved to setup_local_APIC for BP */ +#endif io_apic_irqs = ~PIC_IRQS; apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); - + /* + * Set up IO-APIC IRQ routing. + */ +#ifdef CONFIG_X86_32 + if (!acpi_ioapic) + setup_ioapic_ids_from_mpc(); +#endif sync_Arb_IDs(); setup_IO_APIC_irqs(); init_IO_APIC_traps(); @@ -3128,7 +3618,93 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) #ifdef CONFIG_ACPI -#define IO_APIC_MAX_ID 0xFE +#ifdef CONFIG_X86_32 +int __init io_apic_get_unique_id(int ioapic, int apic_id) +{ + union IO_APIC_reg_00 reg_00; + static physid_mask_t apic_id_map = PHYSID_MASK_NONE; + physid_mask_t tmp; + unsigned long flags; + int i = 0; + + /* + * The P4 platform supports up to 256 APIC IDs on two separate APIC + * buses (one for LAPICs, one for IOAPICs), where predecessors only + * supports up to 16 on one shared APIC bus. + * + * TBD: Expand LAPIC/IOAPIC support on P4-class systems to take full + * advantage of new APIC bus architecture. + */ + + if (physids_empty(apic_id_map)) + apic_id_map = ioapic_phys_id_map(phys_cpu_present_map); + + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(ioapic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + if (apic_id >= get_physical_broadcast()) { + printk(KERN_WARNING "IOAPIC[%d]: Invalid apic_id %d, trying " + "%d\n", ioapic, apic_id, reg_00.bits.ID); + apic_id = reg_00.bits.ID; + } + + /* + * Every APIC in a system must have a unique ID or we get lots of nice + * 'stuck on smp_invalidate_needed IPI wait' messages. + */ + if (check_apicid_used(apic_id_map, apic_id)) { + + for (i = 0; i < get_physical_broadcast(); i++) { + if (!check_apicid_used(apic_id_map, i)) + break; + } + + if (i == get_physical_broadcast()) + panic("Max apic_id exceeded!\n"); + + printk(KERN_WARNING "IOAPIC[%d]: apic_id %d already used, " + "trying %d\n", ioapic, apic_id, i); + + apic_id = i; + } + + tmp = apicid_to_cpu_present(apic_id); + physids_or(apic_id_map, apic_id_map, tmp); + + if (reg_00.bits.ID != apic_id) { + reg_00.bits.ID = apic_id; + + spin_lock_irqsave(&ioapic_lock, flags); + io_apic_write(ioapic, 0, reg_00.raw); + reg_00.raw = io_apic_read(ioapic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + /* Sanity check */ + if (reg_00.bits.ID != apic_id) { + printk("IOAPIC[%d]: Unable to change apic_id!\n", ioapic); + return -1; + } + } + + apic_printk(APIC_VERBOSE, KERN_INFO + "IOAPIC[%d]: Assigned apic_id %d\n", ioapic, apic_id); + + return apic_id; +} + +int __init io_apic_get_version(int ioapic) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(ioapic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + + return reg_01.bits.version; +} +#endif int __init io_apic_get_redir_entries (int ioapic) { @@ -3226,6 +3802,7 @@ void __init setup_ioapic_dest(void) } #endif +#ifdef CONFIG_X86_64 #define IOAPIC_RESOURCE_NAME_SIZE 11 static struct resource *ioapic_resources; @@ -3261,36 +3838,56 @@ static struct resource * __init ioapic_setup_resources(void) return res; } +#endif void __init ioapic_init_mappings(void) { unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; - struct resource *ioapic_res; int i; +#ifdef CONFIG_X86_64 + struct resource *ioapic_res; ioapic_res = ioapic_setup_resources(); +#endif for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { ioapic_phys = mp_ioapics[i].mp_apicaddr; +#ifdef CONFIG_X86_32 + if (!ioapic_phys) { + printk(KERN_ERR + "WARNING: bogus zero IO-APIC " + "address found in MPTABLE, " + "disabling IO/APIC support!\n"); + smp_found_config = 0; + skip_ioapic_setup = 1; + goto fake_ioapic_page; + } +#endif } else { +#ifdef CONFIG_X86_32 +fake_ioapic_page: +#endif ioapic_phys = (unsigned long) alloc_bootmem_pages(PAGE_SIZE); ioapic_phys = __pa(ioapic_phys); } set_fixmap_nocache(idx, ioapic_phys); apic_printk(APIC_VERBOSE, - "mapped IOAPIC to %016lx (%016lx)\n", + "mapped IOAPIC to %08lx (%08lx)\n", __fix_to_virt(idx), ioapic_phys); idx++; +#ifdef CONFIG_X86_64 if (ioapic_res != NULL) { ioapic_res->start = ioapic_phys; ioapic_res->end = ioapic_phys + (4 * 1024) - 1; ioapic_res++; } +#endif } } +#ifdef CONFIG_X86_64 static int __init ioapic_insert_resources(void) { int i; @@ -3313,4 +3910,4 @@ static int __init ioapic_insert_resources(void) /* Insert the IO APIC resources after PCI initialization has occured to handle * IO APICS that are mapped in on a BAR in PCI space. */ late_initcall(ioapic_insert_resources); - +#endif -- cgit v1.2.3 From 54168ed7f2a4f3fc2780e645124ae952598da601 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 20 Aug 2008 09:07:45 +0200 Subject: x86: make io_apic_32.c the same as io_apic_64.c Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic_32.c | 1521 ++++++++++++++++++++++++++++++------------ 1 file changed, 1104 insertions(+), 417 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index 3ed36041c81e..fba6d6ee3480 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -35,7 +35,7 @@ #include #include #include -#include /* time_after() */ +#include /* time_after() */ #ifdef CONFIG_ACPI #include #endif @@ -64,8 +64,8 @@ #define __apicdebuginit(type) static type __init /* - * Is the SiS APIC rmw bug present ? - * -1 = don't know, 0 = no, 1 = yes + * Is the SiS APIC rmw bug present ? + * -1 = don't know, 0 = no, 1 = yes */ int sis_apic_bug = -1; @@ -102,7 +102,7 @@ DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); int skip_ioapic_setup; -static int __init parse_noapic(char *arg) +static int __init parse_noapic(char *str) { /* disable IO-APIC */ disable_ioapic_setup(); @@ -188,7 +188,7 @@ static void __init init_work(void *data) irq_cfgx[legacy_count - 1].next = NULL; } -#define for_each_irq_cfg(cfg) \ +#define for_each_irq_cfg(cfg) \ for (cfg = irq_cfgx; cfg; cfg = cfg->next) DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); @@ -262,7 +262,6 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) irq_cfgx = cfg; cfg->irq = irq; printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); - #ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG { /* dump the results */ @@ -384,9 +383,9 @@ static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned i */ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) { - volatile struct io_apic __iomem *io_apic = io_apic_base(apic); - if (sis_apic_bug) - writel(reg, &io_apic->index); + struct io_apic __iomem *io_apic = io_apic_base(apic); + if (sis_apic_bug) + writel(reg, &io_apic->index); writel(value, &io_apic->data); } @@ -494,11 +493,20 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) apic = entry->apic; pin = entry->pin; +#ifdef CONFIG_INTR_REMAP + /* + * With interrupt-remapping, destination information comes + * from interrupt-remapping table entry. + */ + if (!irq_remapped(irq)) + io_apic_write(apic, 0x11 + pin*2, dest); +#else io_apic_write(apic, 0x11 + pin*2, dest); +#endif reg = io_apic_read(apic, 0x10 + pin*2); reg &= ~IO_APIC_REDIR_VECTOR_MASK; reg |= vector; - io_apic_modify(apic, 0x10 + pin *2, reg); + io_apic_modify(apic, 0x10 + pin*2, reg); if (!entry->next) break; entry = entry->next; @@ -513,6 +521,7 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) unsigned long flags; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -529,12 +538,12 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) */ dest = SET_APIC_LOGICAL_ID(dest); + desc = irq_to_desc(irq); spin_lock_irqsave(&ioapic_lock, flags); __target_IO_APIC_irq(irq, dest, cfg->vector); - irq_to_desc(irq)->affinity = mask; + desc->affinity = mask; spin_unlock_irqrestore(&ioapic_lock, flags); } - #endif /* CONFIG_SMP */ /* @@ -699,7 +708,7 @@ static void __unmask_and_level_IO_APIC_irq(unsigned int irq) #endif -static void mask_IO_APIC_irq(unsigned int irq) +static void mask_IO_APIC_irq (unsigned int irq) { unsigned long flags; @@ -708,7 +717,7 @@ static void mask_IO_APIC_irq(unsigned int irq) spin_unlock_irqrestore(&ioapic_lock, flags); } -static void unmask_IO_APIC_irq(unsigned int irq) +static void unmask_IO_APIC_irq (unsigned int irq) { unsigned long flags; @@ -725,14 +734,13 @@ static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) entry = ioapic_read_entry(apic, pin); if (entry.delivery_mode == dest_SMI) return; - /* * Disable it in the IO-APIC irq-routing table: */ ioapic_mask_entry(apic, pin); } -static void clear_IO_APIC(void) +static void clear_IO_APIC (void) { int apic, pin; @@ -741,7 +749,7 @@ static void clear_IO_APIC(void) clear_IO_APIC_pin(apic, pin); } -#ifndef CONFIG_SMP +#if !defined(CONFIG_SMP) && defined(CONFIG_X86_32) void send_IPI_self(int vector) { unsigned int cfg; @@ -756,9 +764,9 @@ void send_IPI_self(int vector) */ apic_write(APIC_ICR, cfg); } -#endif /* !CONFIG_SMP */ - +#endif /* !CONFIG_SMP && CONFIG_X86_32*/ +#ifdef CONFIG_X86_32 /* * support for broken MP BIOSs, enables hand-redirection of PIRQ0-7 to * specific CPU-side IRQs. @@ -797,6 +805,75 @@ static int __init ioapic_pirq_setup(char *str) } __setup("pirq=", ioapic_pirq_setup); +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_INTR_REMAP +/* I/O APIC RTE contents at the OS boot up */ +static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; + +/* + * Saves and masks all the unmasked IO-APIC RTE's + */ +int save_mask_IO_APIC_setup(void) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + int apic, pin; + + /* + * The number of IO-APIC IRQ registers (== #pins): + */ + for (apic = 0; apic < nr_ioapics; apic++) { + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(apic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + nr_ioapic_registers[apic] = reg_01.bits.entries+1; + } + + for (apic = 0; apic < nr_ioapics; apic++) { + early_ioapic_entries[apic] = + kzalloc(sizeof(struct IO_APIC_route_entry) * + nr_ioapic_registers[apic], GFP_KERNEL); + if (!early_ioapic_entries[apic]) + return -ENOMEM; + } + + for (apic = 0; apic < nr_ioapics; apic++) + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + struct IO_APIC_route_entry entry; + + entry = early_ioapic_entries[apic][pin] = + ioapic_read_entry(apic, pin); + if (!entry.mask) { + entry.mask = 1; + ioapic_write_entry(apic, pin, entry); + } + } + return 0; +} + +void restore_IO_APIC_setup(void) +{ + int apic, pin; + + for (apic = 0; apic < nr_ioapics; apic++) + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) + ioapic_write_entry(apic, pin, + early_ioapic_entries[apic][pin]); +} + +void reinit_intr_remapped_IO_APIC(int intr_remapping) +{ + /* + * for now plain restore of previous settings. + * TBD: In the case of OS enabling interrupt-remapping, + * IO-APIC RTE's need to be setup to point to interrupt-remapping + * table entries. for now, do a plain restore, and wait for + * the setup_IO_APIC_irqs() to do proper initialization. + */ + restore_IO_APIC_setup(); +} +#endif /* * Find the IRQ entry number of a certain pin. @@ -848,7 +925,7 @@ static int __init find_isa_irq_apic(int irq, int type) } if (i < mp_irq_entries) { int apic; - for (apic = 0; apic < nr_ioapics; apic++) { + for(apic = 0; apic < nr_ioapics; apic++) { if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic) return apic; } @@ -867,10 +944,10 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) { int apic, i, best_guess = -1; - apic_printk(APIC_DEBUG, "querying PCI -> IRQ mapping bus:%d, " - "slot:%d, pin:%d.\n", bus, slot, pin); + apic_printk(APIC_DEBUG, "querying PCI -> IRQ mapping bus:%d, slot:%d, pin:%d.\n", + bus, slot, pin); if (test_bit(bus, mp_bus_not_pci)) { - printk(KERN_WARNING "PCI BIOS passed nonexistent PCI bus %d!\n", bus); + apic_printk(APIC_VERBOSE, "PCI BIOS passed nonexistent PCI bus %d!\n", bus); return -1; } for (i = 0; i < mp_irq_entries; i++) { @@ -885,7 +962,7 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) !mp_irqs[i].mp_irqtype && (bus == lbus) && (slot == ((mp_irqs[i].mp_srcbusirq >> 2) & 0x1f))) { - int irq = pin_2_irq(i, apic, mp_irqs[i].mp_dstirq); + int irq = pin_2_irq(i,apic,mp_irqs[i].mp_dstirq); if (!(apic || IO_APIC_IRQ(irq))) continue; @@ -902,6 +979,7 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) } return best_guess; } + EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); #if defined(CONFIG_EISA) || defined(CONFIG_MCA) @@ -918,6 +996,7 @@ static int EISA_ELCR(unsigned int irq) "Broken MPtable reports ISA irq %d\n", irq); return 0; } + #endif /* ISA interrupts are always polarity zero edge triggered, @@ -954,36 +1033,36 @@ static int MPBIOS_polarity(int idx) /* * Determine IRQ line polarity (high active or low active): */ - switch (mp_irqs[idx].mp_irqflag & 3) { - case 0: /* conforms, ie. bus-type dependent polarity */ - { - polarity = test_bit(bus, mp_bus_not_pci)? - default_ISA_polarity(idx): - default_PCI_polarity(idx); - break; - } - case 1: /* high active */ - { - polarity = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } - case 3: /* low active */ - { - polarity = 1; - break; - } - default: /* invalid */ + switch (mp_irqs[idx].mp_irqflag & 3) { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } + case 0: /* conforms, ie. bus-type dependent polarity */ + if (test_bit(bus, mp_bus_not_pci)) + polarity = default_ISA_polarity(idx); + else + polarity = default_PCI_polarity(idx); + break; + case 1: /* high active */ + { + polarity = 0; + break; + } + case 2: /* reserved */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + polarity = 1; + break; + } + case 3: /* low active */ + { + polarity = 1; + break; + } + default: /* invalid */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + polarity = 1; + break; + } } return polarity; } @@ -996,67 +1075,67 @@ static int MPBIOS_trigger(int idx) /* * Determine IRQ trigger mode (edge or level sensitive): */ - switch ((mp_irqs[idx].mp_irqflag>>2) & 3) { - case 0: /* conforms, ie. bus-type dependent */ + switch ((mp_irqs[idx].mp_irqflag>>2) & 3) { - trigger = test_bit(bus, mp_bus_not_pci)? - default_ISA_trigger(idx): - default_PCI_trigger(idx); + case 0: /* conforms, ie. bus-type dependent */ + if (test_bit(bus, mp_bus_not_pci)) + trigger = default_ISA_trigger(idx); + else + trigger = default_PCI_trigger(idx); #if defined(CONFIG_EISA) || defined(CONFIG_MCA) - switch (mp_bus_id_to_type[bus]) { - case MP_BUS_ISA: /* ISA pin */ - { - /* set before the switch */ + switch (mp_bus_id_to_type[bus]) { + case MP_BUS_ISA: /* ISA pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_EISA: /* EISA pin */ + { + trigger = default_EISA_trigger(idx); + break; + } + case MP_BUS_PCI: /* PCI pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_MCA: /* MCA pin */ + { + trigger = default_MCA_trigger(idx); + break; + } + default: + { + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 1; + break; + } + } +#endif break; - } - case MP_BUS_EISA: /* EISA pin */ + case 1: /* edge */ { - trigger = default_EISA_trigger(idx); + trigger = 0; break; } - case MP_BUS_PCI: /* PCI pin */ + case 2: /* reserved */ { - /* set before the switch */ + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 1; break; } - case MP_BUS_MCA: /* MCA pin */ + case 3: /* level */ { - trigger = default_MCA_trigger(idx); + trigger = 1; break; } - default: + default: /* invalid */ { printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; + trigger = 0; break; } } -#endif - break; - } - case 1: /* edge */ - { - trigger = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; - break; - } - case 3: /* level */ - { - trigger = 1; - break; - } - default: /* invalid */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 0; - break; - } - } return trigger; } @@ -1082,9 +1161,9 @@ static int pin_2_irq(int idx, int apic, int pin) if (mp_irqs[idx].mp_dstirq != pin) printk(KERN_ERR "broken BIOS or MPTABLE parser, ayiee!!\n"); - if (test_bit(bus, mp_bus_not_pci)) + if (test_bit(bus, mp_bus_not_pci)) { irq = mp_irqs[idx].mp_srcbusirq; - else { + } else { /* * PCI IRQs are mapped in order */ @@ -1092,14 +1171,14 @@ static int pin_2_irq(int idx, int apic, int pin) while (i < apic) irq += nr_ioapic_registers[i++]; irq += pin; - - /* - * For MPS mode, so far only needed by ES7000 platform - */ - if (ioapic_renumber_irq) - irq = ioapic_renumber_irq(apic, irq); + /* + * For MPS mode, so far only needed by ES7000 platform + */ + if (ioapic_renumber_irq) + irq = ioapic_renumber_irq(apic, irq); } +#ifdef CONFIG_X86_32 /* * PCI IRQ command line redirection. Yes, limits are hardcoded. */ @@ -1116,6 +1195,8 @@ static int pin_2_irq(int idx, int apic, int pin) } } } +#endif + return irq; } @@ -1145,74 +1226,70 @@ static int __assign_irq_vector(int irq, cpumask_t mask) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; - unsigned int old_vector; - int cpu; - struct irq_cfg *cfg; + static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; + unsigned int old_vector; + int cpu; + struct irq_cfg *cfg; - cfg = irq_cfg(irq); + cfg = irq_cfg(irq); - /* Only try and allocate irqs on cpus that are present */ - cpus_and(mask, mask, cpu_online_map); + /* Only try and allocate irqs on cpus that are present */ + cpus_and(mask, mask, cpu_online_map); - if ((cfg->move_in_progress) || cfg->move_cleanup_count) - return -EBUSY; + if ((cfg->move_in_progress) || cfg->move_cleanup_count) + return -EBUSY; - old_vector = cfg->vector; - if (old_vector) { - cpumask_t tmp; - cpus_and(tmp, cfg->domain, mask); - if (!cpus_empty(tmp)) - return 0; - } + old_vector = cfg->vector; + if (old_vector) { + cpumask_t tmp; + cpus_and(tmp, cfg->domain, mask); + if (!cpus_empty(tmp)) + return 0; + } - for_each_cpu_mask_nr(cpu, mask) { - cpumask_t domain, new_mask; - int new_cpu; - int vector, offset; + for_each_cpu_mask_nr(cpu, mask) { + cpumask_t domain, new_mask; + int new_cpu; + int vector, offset; - domain = vector_allocation_domain(cpu); - cpus_and(new_mask, domain, cpu_online_map); + domain = vector_allocation_domain(cpu); + cpus_and(new_mask, domain, cpu_online_map); - vector = current_vector; - offset = current_offset; + vector = current_vector; + offset = current_offset; next: - vector += 8; - if (vector >= first_system_vector) { - /* If we run out of vectors on large boxen, must share them. */ - offset = (offset + 1) % 8; - vector = FIRST_DEVICE_VECTOR + offset; - } - if (unlikely(current_vector == vector)) - continue; + vector += 8; + if (vector >= first_system_vector) { + /* If we run out of vectors on large boxen, must share them. */ + offset = (offset + 1) % 8; + vector = FIRST_DEVICE_VECTOR + offset; + } + if (unlikely(current_vector == vector)) + continue; #ifdef CONFIG_X86_64 - if (vector == IA32_SYSCALL_VECTOR) - goto next; + if (vector == IA32_SYSCALL_VECTOR) + goto next; #else - if (vector == SYSCALL_VECTOR) - goto next; + if (vector == SYSCALL_VECTOR) + goto next; #endif - for_each_cpu_mask_nr(new_cpu, new_mask) - if (per_cpu(vector_irq, new_cpu)[vector] != -1) - goto next; - /* Found one! */ - current_vector = vector; - current_offset = offset; - if (old_vector) { - cfg->move_in_progress = 1; - cfg->old_domain = cfg->domain; - } - printk(KERN_DEBUG "assign_irq_vector: irq %d vector %#x cpu ", irq, vector); - for_each_cpu_mask_nr(new_cpu, new_mask) { - per_cpu(vector_irq, new_cpu)[vector] = irq; - printk(KERN_CONT " %d ", new_cpu); + for_each_cpu_mask_nr(new_cpu, new_mask) + if (per_cpu(vector_irq, new_cpu)[vector] != -1) + goto next; + /* Found one! */ + current_vector = vector; + current_offset = offset; + if (old_vector) { + cfg->move_in_progress = 1; + cfg->old_domain = cfg->domain; } - printk(KERN_CONT "\n"); - cfg->vector = vector; - cfg->domain = domain; - return 0; - } - return -ENOSPC; + for_each_cpu_mask_nr(new_cpu, new_mask) + per_cpu(vector_irq, new_cpu)[vector] = irq; + cfg->vector = vector; + cfg->domain = domain; + return 0; + } + return -ENOSPC; } static int assign_irq_vector(int irq, cpumask_t mask) @@ -1223,7 +1300,6 @@ static int assign_irq_vector(int irq, cpumask_t mask) spin_lock_irqsave(&vector_lock, flags); err = __assign_irq_vector(irq, mask); spin_unlock_irqrestore(&vector_lock, flags); - return err; } @@ -1269,36 +1345,39 @@ void __setup_vector_irq(int cpu) cfg = irq_cfg(irq); if (!cpu_isset(cpu, cfg->domain)) per_cpu(vector_irq, cpu)[vector] = -1; - } + } } static struct irq_chip ioapic_chip; +#ifdef CONFIG_INTR_REMAP +static struct irq_chip ir_ioapic_chip; +#endif -#define IOAPIC_AUTO -1 -#define IOAPIC_EDGE 0 -#define IOAPIC_LEVEL 1 +#define IOAPIC_AUTO -1 +#define IOAPIC_EDGE 0 +#define IOAPIC_LEVEL 1 #ifdef CONFIG_X86_32 static inline int IO_APIC_irq_trigger(int irq) { - int apic, idx, pin; + int apic, idx, pin; - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - idx = find_irq_entry(apic, pin, mp_INT); - if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) - return irq_trigger(idx); - } - } - /* - * nonexistent IRQs are edge default - */ - return 0; + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + idx = find_irq_entry(apic, pin, mp_INT); + if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) + return irq_trigger(idx); + } + } + /* + * nonexistent IRQs are edge default + */ + return 0; } #else static inline int IO_APIC_irq_trigger(int irq) { - return 1; + return 1; } #endif @@ -1318,13 +1397,27 @@ static void ioapic_register_intr(int irq, unsigned long trigger) else desc->status &= ~IRQ_LEVEL; +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + desc->status |= IRQ_MOVE_PCNTXT; + if (trigger) + set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, + handle_fasteoi_irq, + "fasteoi"); + else + set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, + handle_edge_irq, "edge"); + return; + } +#endif if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_fasteoi_irq, "fasteoi"); + handle_fasteoi_irq, + "fasteoi"); else set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_edge_irq, "edge"); + handle_edge_irq, "edge"); } static int setup_ioapic_entry(int apic, int irq, @@ -1337,11 +1430,45 @@ static int setup_ioapic_entry(int apic, int irq, */ memset(entry,0,sizeof(*entry)); - entry->delivery_mode = INT_DELIVERY_MODE; - entry->dest_mode = INT_DEST_MODE; - entry->dest = destination; +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) { + struct intel_iommu *iommu = map_ioapic_to_ir(apic); + struct irte irte; + struct IR_IO_APIC_route_entry *ir_entry = + (struct IR_IO_APIC_route_entry *) entry; + int index; + + if (!iommu) + panic("No mapping iommu for ioapic %d\n", apic); + + index = alloc_irte(iommu, irq, 1); + if (index < 0) + panic("Failed to allocate IRTE for ioapic %d\n", apic); + + memset(&irte, 0, sizeof(irte)); + + irte.present = 1; + irte.dst_mode = INT_DEST_MODE; + irte.trigger_mode = trigger; + irte.dlvry_mode = INT_DELIVERY_MODE; + irte.vector = vector; + irte.dest_id = IRTE_DEST(destination); + + modify_irte(irq, &irte); + + ir_entry->index2 = (index >> 15) & 0x1; + ir_entry->zero = 0; + ir_entry->format = 1; + ir_entry->index = (index & 0x7fff); + } else +#endif + { + entry->delivery_mode = INT_DELIVERY_MODE; + entry->dest_mode = INT_DEST_MODE; + entry->dest = destination; + } - entry->mask = 0; /* enable IRQ */ + entry->mask = 0; /* enable IRQ */ entry->trigger = trigger; entry->polarity = polarity; entry->vector = vector; @@ -1351,12 +1478,11 @@ static int setup_ioapic_entry(int apic, int irq, */ if (trigger) entry->mask = 1; - return 0; } static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, - int trigger, int polarity) + int trigger, int polarity) { struct irq_cfg *cfg; struct IO_APIC_route_entry entry; @@ -1420,10 +1546,10 @@ static void __init setup_IO_APIC_irqs(void) } irq = pin_2_irq(idx, apic, pin); - +#ifdef CONFIG_X86_32 if (multi_timer_check(apic, irq)) continue; - +#endif add_pin_to_irq(irq, apic, pin); setup_IO_APIC_irq(apic, pin, irq, @@ -1443,6 +1569,11 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, { struct IO_APIC_route_entry entry; +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + return; +#endif + memset(&entry, 0, sizeof(entry)); /* @@ -1461,7 +1592,7 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, * The timer IRQ doesn't have to know that behind the * scene we may have a 8259A-master in AEOI mode ... */ - ioapic_register_intr(0, IOAPIC_EDGE); + set_irq_chip_and_handler_name(0, &ioapic_chip, handle_edge_irq, "edge"); /* * Add it to the IO-APIC irq-routing table: @@ -1501,17 +1632,18 @@ __apicdebuginit(void) print_IO_APIC(void) reg_01.raw = io_apic_read(apic, 1); if (reg_01.bits.version >= 0x10) reg_02.raw = io_apic_read(apic, 2); - if (reg_01.bits.version >= 0x20) - reg_03.raw = io_apic_read(apic, 3); + if (reg_01.bits.version >= 0x20) + reg_03.raw = io_apic_read(apic, 3); spin_unlock_irqrestore(&ioapic_lock, flags); + printk("\n"); printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); printk(KERN_DEBUG "....... : LTS : %X\n", reg_00.bits.LTS); - printk(KERN_DEBUG ".... register #01: %08X\n", reg_01.raw); + printk(KERN_DEBUG ".... register #01: %08X\n", *(int *)®_01); printk(KERN_DEBUG "....... : max redirection entries: %04X\n", reg_01.bits.entries); printk(KERN_DEBUG "....... : PRQ implemented: %X\n", reg_01.bits.PRQ); @@ -1548,7 +1680,10 @@ __apicdebuginit(void) print_IO_APIC(void) entry = ioapic_read_entry(apic, i); - printk(KERN_DEBUG " %02x %02X ", i, entry.dest); + printk(KERN_DEBUG " %02x %03X ", + i, + entry.dest + ); printk("%1d %1d %1d %1d %1d %1d %1d %02X\n", entry.mask, @@ -1567,7 +1702,7 @@ __apicdebuginit(void) print_IO_APIC(void) struct irq_pin_list *entry = cfg->irq_2_pin; if (!entry) continue; - printk(KERN_DEBUG "IRQ%d ", i); + printk(KERN_DEBUG "IRQ%d ", cfg->irq); for (;;) { printk("-> %d:%d", entry->apic, entry->pin); if (!entry->next) @@ -1614,8 +1749,7 @@ __apicdebuginit(void) print_local_APIC(void *dummy) printk("\n" KERN_DEBUG "printing local APIC contents on CPU#%d/%d:\n", smp_processor_id(), hard_smp_processor_id()); v = apic_read(APIC_ID); - printk(KERN_INFO "... APIC ID: %08x (%01x)\n", v, - GET_APIC_ID(v)); + printk(KERN_INFO "... APIC ID: %08x (%01x)\n", v, read_apic_id()); v = apic_read(APIC_LVR); printk(KERN_INFO "... APIC VERSION: %08x\n", v); ver = GET_APIC_VERSION(v); @@ -1624,7 +1758,7 @@ __apicdebuginit(void) print_local_APIC(void *dummy) v = apic_read(APIC_TASKPRI); printk(KERN_DEBUG "... APIC TASKPRI: %08x (%02x)\n", v, v & APIC_TPRI_MASK); - if (APIC_INTEGRATED(ver)) { /* !82489DX */ + if (APIC_INTEGRATED(ver)) { /* !82489DX */ v = apic_read(APIC_ARBPRI); printk(KERN_DEBUG "... APIC ARBPRI: %08x (%02x)\n", v, v & APIC_ARBPRI_MASK); @@ -1650,9 +1784,10 @@ __apicdebuginit(void) print_local_APIC(void *dummy) printk(KERN_DEBUG "... APIC IRR field:\n"); print_APIC_bitfield(APIC_IRR); - if (APIC_INTEGRATED(ver)) { /* !82489DX */ - if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ + if (APIC_INTEGRATED(ver)) { /* !82489DX */ + if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ apic_write(APIC_ESR, 0); + v = apic_read(APIC_ESR); printk(KERN_DEBUG "... APIC ESR: %08x\n", v); } @@ -1710,11 +1845,11 @@ __apicdebuginit(void) print_PIC(void) v = inb(0xa0) << 8 | inb(0x20); printk(KERN_DEBUG "... PIC IRR: %04x\n", v); - outb(0x0b, 0xa0); - outb(0x0b, 0x20); + outb(0x0b,0xa0); + outb(0x0b,0x20); v = inb(0xa0) << 8 | inb(0x20); - outb(0x0a, 0xa0); - outb(0x0a, 0x20); + outb(0x0a,0xa0); + outb(0x0a,0x20); spin_unlock_irqrestore(&i8259A_lock, flags); @@ -1739,16 +1874,19 @@ fs_initcall(print_all_ICs); /* Where if anywhere is the i8259 connect in external int mode */ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; -static void __init enable_IO_APIC(void) +void __init enable_IO_APIC(void) { union IO_APIC_reg_01 reg_01; int i8259_apic, i8259_pin; - int i, apic; + int apic; unsigned long flags; +#ifdef CONFIG_X86_32 + int i; if (!pirqs_enabled) for (i = 0; i < MAX_PIRQS; i++) pirq_entries[i] = -1; +#endif /* * The number of IO-APIC IRQ registers (== #pins): @@ -1759,7 +1897,7 @@ static void __init enable_IO_APIC(void) spin_unlock_irqrestore(&ioapic_lock, flags); nr_ioapic_registers[apic] = reg_01.bits.entries+1; } - for (apic = 0; apic < nr_ioapics; apic++) { + for(apic = 0; apic < nr_ioapics; apic++) { int pin; /* See if any of the pins is in ExtINT mode */ for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { @@ -1830,16 +1968,18 @@ void disable_IO_APIC(void) entry.dest_mode = 0; /* Physical */ entry.delivery_mode = dest_ExtINT; /* ExtInt */ entry.vector = 0; - entry.dest = read_apic_id(); + entry.dest = read_apic_id(); /* * Add it to the IO-APIC irq-routing table: */ ioapic_write_entry(ioapic_i8259.apic, ioapic_i8259.pin, entry); } + disconnect_bsp_APIC(ioapic_i8259.pin != -1); } +#ifdef CONFIG_X86_32 /* * function to set the IO-APIC physical IDs based on the * values stored in the MPC table. @@ -1940,8 +2080,6 @@ static void __init setup_ioapic_ids_from_mpc(void) reg_00.bits.ID = mp_ioapics[apic].mp_apicid; spin_lock_irqsave(&ioapic_lock, flags); - io_apic_write(apic, 0, reg_00.raw); - spin_unlock_irqrestore(&ioapic_lock, flags); /* * Sanity check @@ -1955,6 +2093,7 @@ static void __init setup_ioapic_ids_from_mpc(void) apic_printk(APIC_VERBOSE, " ok.\n"); } } +#endif int no_timer_check __initdata; @@ -1994,9 +2133,10 @@ static int __init timer_irq_works(void) * might have cached one ExtINT interrupt. Finally, at * least one tick may be lost due to delays. */ + + /* jiffies wrap? */ if (time_after(jiffies, t1 + 4)) return 1; - return 0; } @@ -2014,8 +2154,6 @@ static int __init timer_irq_works(void) */ /* - * Startup quirk: - * * Starting up a edge-triggered IO-APIC interrupt is * nasty - we need to make sure that we get the edge. * If it is already asserted for some reason, we need @@ -2023,9 +2161,8 @@ static int __init timer_irq_works(void) * * This is not complete - we should be able to fake * an edge even if it isn't on the 8259A... - * - * (We do this for level-triggered IRQs too - it cannot hurt.) */ + static unsigned int startup_ioapic_irq(unsigned int irq) { int was_pending = 0; @@ -2043,70 +2180,254 @@ static unsigned int startup_ioapic_irq(unsigned int irq) return was_pending; } +#ifdef CONFIG_X86_64 static int ioapic_retrigger_irq(unsigned int irq) { - send_IPI_self(irq_cfg(irq)->vector); + + struct irq_cfg *cfg = irq_cfg(irq); + unsigned long flags; + + spin_lock_irqsave(&vector_lock, flags); + send_IPI_mask(cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); + spin_unlock_irqrestore(&vector_lock, flags); return 1; } - -#ifdef CONFIG_SMP -asmlinkage void smp_irq_move_cleanup_interrupt(void) +#else +static int ioapic_retrigger_irq(unsigned int irq) { - unsigned vector, me; - ack_APIC_irq(); - irq_enter(); - - me = smp_processor_id(); - for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { - unsigned int irq; - struct irq_desc *desc; - struct irq_cfg *cfg; - irq = __get_cpu_var(vector_irq)[vector]; + send_IPI_self(irq_cfg(irq)->vector); - desc = irq_to_desc(irq); - if (!desc) - continue; + return 1; +} +#endif - cfg = irq_cfg(irq); - spin_lock(&desc->lock); - if (!cfg->move_cleanup_count) - goto unlock; +/* + * Level and edge triggered IO-APIC interrupts need different handling, + * so we use two separate IRQ descriptors. Edge triggered IRQs can be + * handled with the level-triggered descriptor, but that one has slightly + * more overhead. Level-triggered interrupts cannot be handled with the + * edge-triggered handler, without risking IRQ storms and other ugly + * races. + */ - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) - goto unlock; +#ifdef CONFIG_SMP - __get_cpu_var(vector_irq)[vector] = -1; - cfg->move_cleanup_count--; -unlock: - spin_unlock(&desc->lock); - } +#ifdef CONFIG_INTR_REMAP +static void ir_irq_migration(struct work_struct *work); - irq_exit(); -} +static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); -static void irq_complete_move(unsigned int irq) +/* + * Migrate the IO-APIC irq in the presence of intr-remapping. + * + * For edge triggered, irq migration is a simple atomic update(of vector + * and cpu destination) of IRTE and flush the hardware cache. + * + * For level triggered, we need to modify the io-apic RTE aswell with the update + * vector information, along with modifying IRTE with vector and destination. + * So irq migration for level triggered is little bit more complex compared to + * edge triggered migration. But the good news is, we use the same algorithm + * for level triggered migration as we have today, only difference being, + * we now initiate the irq migration from process context instead of the + * interrupt context. + * + * In future, when we do a directed EOI (combined with cpu EOI broadcast + * suppression) to the IO-APIC, level triggered irq migration will also be + * as simple as edge triggered migration and we can do the irq migration + * with a simple atomic update to IO-APIC RTE. + */ +static void migrate_ioapic_irq(int irq, cpumask_t mask) { - struct irq_cfg *cfg = irq_cfg(irq); - unsigned vector, me; + struct irq_cfg *cfg; + struct irq_desc *desc; + cpumask_t tmp, cleanup_mask; + struct irte irte; + int modify_ioapic_rte; + unsigned int dest; + unsigned long flags; - if (likely(!cfg->move_in_progress)) + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) return; - vector = ~get_irq_regs()->orig_ax; - me = smp_processor_id(); - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { - cpumask_t cleanup_mask; + if (get_irte(irq, &irte)) + return; - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + desc = irq_to_desc(irq); + modify_ioapic_rte = desc->status & IRQ_LEVEL; + if (modify_ioapic_rte) { + spin_lock_irqsave(&ioapic_lock, flags); + __target_IO_APIC_irq(irq, dest, cfg->vector); + spin_unlock_irqrestore(&ioapic_lock, flags); + } + + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + /* + * Modified the IRTE and flushes the Interrupt entry cache. + */ + modify_irte(irq, &irte); + + if (cfg->move_in_progress) { + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } + + desc->affinity = mask; +} + +static int migrate_irq_remapped_level(int irq) +{ + int ret = -1; + struct irq_desc *desc = irq_to_desc(irq); + + mask_IO_APIC_irq(irq); + + if (io_apic_level_ack_pending(irq)) { + /* + * Interrupt in progress. Migrating irq now will change the + * vector information in the IO-APIC RTE and that will confuse + * the EOI broadcast performed by cpu. + * So, delay the irq migration to the next instance. + */ + schedule_delayed_work(&ir_migration_work, 1); + goto unmask; + } + + /* everthing is clear. we have right of way */ + migrate_ioapic_irq(irq, desc->pending_mask); + + ret = 0; + desc->status &= ~IRQ_MOVE_PENDING; + cpus_clear(desc->pending_mask); + +unmask: + unmask_IO_APIC_irq(irq); + return ret; +} + +static void ir_irq_migration(struct work_struct *work) +{ + unsigned int irq; + struct irq_desc *desc; + + for_each_irq_desc(irq, desc) { + if (desc->status & IRQ_MOVE_PENDING) { + unsigned long flags; + + spin_lock_irqsave(&desc->lock, flags); + if (!desc->chip->set_affinity || + !(desc->status & IRQ_MOVE_PENDING)) { + desc->status &= ~IRQ_MOVE_PENDING; + spin_unlock_irqrestore(&desc->lock, flags); + continue; + } + + desc->chip->set_affinity(irq, desc->pending_mask); + spin_unlock_irqrestore(&desc->lock, flags); + } + } +} + +/* + * Migrates the IRQ destination in the process context. + */ +static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc = irq_to_desc(irq); + + if (desc->status & IRQ_LEVEL) { + desc->status |= IRQ_MOVE_PENDING; + desc->pending_mask = mask; + migrate_irq_remapped_level(irq); + return; + } + + migrate_ioapic_irq(irq, mask); +} +#endif + +asmlinkage void smp_irq_move_cleanup_interrupt(void) +{ + unsigned vector, me; + ack_APIC_irq(); +#ifdef CONFIG_X86_64 + exit_idle(); +#endif + irq_enter(); + + me = smp_processor_id(); + for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { + unsigned int irq; + struct irq_desc *desc; + struct irq_cfg *cfg; + irq = __get_cpu_var(vector_irq)[vector]; + + desc = irq_to_desc(irq); + if (!desc) + continue; + + cfg = irq_cfg(irq); + spin_lock(&desc->lock); + if (!cfg->move_cleanup_count) + goto unlock; + + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) + goto unlock; + + __get_cpu_var(vector_irq)[vector] = -1; + cfg->move_cleanup_count--; +unlock: + spin_unlock(&desc->lock); + } + + irq_exit(); +} + +static void irq_complete_move(unsigned int irq) +{ + struct irq_cfg *cfg = irq_cfg(irq); + unsigned vector, me; + + if (likely(!cfg->move_in_progress)) + return; + + vector = ~get_irq_regs()->orig_ax; + me = smp_processor_id(); + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { + cpumask_t cleanup_mask; + + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); cfg->move_in_progress = 0; } } #else static inline void irq_complete_move(unsigned int irq) {} #endif +#ifdef CONFIG_INTR_REMAP +static void ack_x2apic_level(unsigned int irq) +{ + ack_x2APIC_irq(); +} + +static void ack_x2apic_edge(unsigned int irq) +{ + ack_x2APIC_irq(); +} +#endif static void ack_apic_edge(unsigned int irq) { @@ -2118,55 +2439,55 @@ static void ack_apic_edge(unsigned int irq) #ifdef CONFIG_X86_64 static void ack_apic_level(unsigned int irq) { - int do_unmask_irq = 0; + int do_unmask_irq = 0; - irq_complete_move(irq); + irq_complete_move(irq); #ifdef CONFIG_GENERIC_PENDING_IRQ - /* If we are moving the irq we need to mask it */ - if (unlikely(desc->status & IRQ_MOVE_PENDING)) { - do_unmask_irq = 1; - mask_IO_APIC_irq(irq); - } + /* If we are moving the irq we need to mask it */ + if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { + do_unmask_irq = 1; + mask_IO_APIC_irq(irq); + } #endif - /* - * We must acknowledge the irq before we move it or the acknowledge will - * not propagate properly. - */ - ack_APIC_irq(); - - /* Now we can move and renable the irq */ - if (unlikely(do_unmask_irq)) { - /* Only migrate the irq if the ack has been received. - * - * On rare occasions the broadcast level triggered ack gets - * delayed going to ioapics, and if we reprogram the - * vector while Remote IRR is still set the irq will never - * fire again. - * - * To prevent this scenario we read the Remote IRR bit - * of the ioapic. This has two effects. - * - On any sane system the read of the ioapic will - * flush writes (and acks) going to the ioapic from - * this cpu. - * - We get to see if the ACK has actually been delivered. - * - * Based on failed experiments of reprogramming the - * ioapic entry from outside of irq context starting - * with masking the ioapic entry and then polling until - * Remote IRR was clear before reprogramming the - * ioapic I don't trust the Remote IRR bit to be - * completey accurate. - * - * However there appears to be no other way to plug - * this race, so if the Remote IRR bit is not - * accurate and is causing problems then it is a hardware bug - * and you can go talk to the chipset vendor about it. - */ - if (!io_apic_level_ack_pending(irq)) - move_masked_irq(irq, desc); - unmask_IO_APIC_irq(irq); - } + /* + * We must acknowledge the irq before we move it or the acknowledge will + * not propagate properly. + */ + ack_APIC_irq(); + + /* Now we can move and renable the irq */ + if (unlikely(do_unmask_irq)) { + /* Only migrate the irq if the ack has been received. + * + * On rare occasions the broadcast level triggered ack gets + * delayed going to ioapics, and if we reprogram the + * vector while Remote IRR is still set the irq will never + * fire again. + * + * To prevent this scenario we read the Remote IRR bit + * of the ioapic. This has two effects. + * - On any sane system the read of the ioapic will + * flush writes (and acks) going to the ioapic from + * this cpu. + * - We get to see if the ACK has actually been delivered. + * + * Based on failed experiments of reprogramming the + * ioapic entry from outside of irq context starting + * with masking the ioapic entry and then polling until + * Remote IRR was clear before reprogramming the + * ioapic I don't trust the Remote IRR bit to be + * completey accurate. + * + * However there appears to be no other way to plug + * this race, so if the Remote IRR bit is not + * accurate and is causing problems then it is a hardware bug + * and you can go talk to the chipset vendor about it. + */ + if (!io_apic_level_ack_pending(irq)) + move_masked_irq(irq); + unmask_IO_APIC_irq(irq); + } } #else atomic_t irq_mis_count; @@ -2177,25 +2498,25 @@ static void ack_apic_level(unsigned int irq) irq_complete_move(irq); move_native_irq(irq); -/* - * It appears there is an erratum which affects at least version 0x11 - * of I/O APIC (that's the 82093AA and cores integrated into various - * chipsets). Under certain conditions a level-triggered interrupt is - * erroneously delivered as edge-triggered one but the respective IRR - * bit gets set nevertheless. As a result the I/O unit expects an EOI - * message but it will never arrive and further interrupts are blocked - * from the source. The exact reason is so far unknown, but the - * phenomenon was observed when two consecutive interrupt requests - * from a given source get delivered to the same CPU and the source is - * temporarily disabled in between. - * - * A workaround is to simulate an EOI message manually. We achieve it - * by setting the trigger mode to edge and then to level when the edge - * trigger mode gets detected in the TMR of a local APIC for a - * level-triggered interrupt. We mask the source for the time of the - * operation to prevent an edge-triggered interrupt escaping meanwhile. - * The idea is from Manfred Spraul. --macro - */ + /* + * It appears there is an erratum which affects at least version 0x11 + * of I/O APIC (that's the 82093AA and cores integrated into various + * chipsets). Under certain conditions a level-triggered interrupt is + * erroneously delivered as edge-triggered one but the respective IRR + * bit gets set nevertheless. As a result the I/O unit expects an EOI + * message but it will never arrive and further interrupts are blocked + * from the source. The exact reason is so far unknown, but the + * phenomenon was observed when two consecutive interrupt requests + * from a given source get delivered to the same CPU and the source is + * temporarily disabled in between. + * + * A workaround is to simulate an EOI message manually. We achieve it + * by setting the trigger mode to edge and then to level when the edge + * trigger mode gets detected in the TMR of a local APIC for a + * level-triggered interrupt. We mask the source for the time of the + * operation to prevent an edge-triggered interrupt escaping meanwhile. + * The idea is from Manfred Spraul. --macro + */ i = irq_cfg(irq)->vector; v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); @@ -2225,6 +2546,20 @@ static struct irq_chip ioapic_chip __read_mostly = { .retrigger = ioapic_retrigger_irq, }; +#ifdef CONFIG_INTR_REMAP +static struct irq_chip ir_ioapic_chip __read_mostly = { + .name = "IR-IO-APIC", + .startup = startup_ioapic_irq, + .mask = mask_IO_APIC_irq, + .unmask = unmask_IO_APIC_irq, + .ack = ack_x2apic_edge, + .eoi = ack_x2apic_level, +#ifdef CONFIG_SMP + .set_affinity = set_ir_ioapic_affinity_irq, +#endif + .retrigger = ioapic_retrigger_irq, +}; +#endif static inline void init_IO_APIC_traps(void) { @@ -2282,7 +2617,7 @@ static void unmask_lapic_irq(unsigned int irq) apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); } -static void ack_lapic_irq(unsigned int irq) +static void ack_lapic_irq (unsigned int irq) { ack_APIC_irq(); } @@ -2383,12 +2718,12 @@ static inline void __init unlock_ExtINT_logic(void) static int disable_timer_pin_1 __initdata; /* Actually the next is obsolete, but keep it for paranoid reasons -AK */ -static int __init parse_disable_timer_pin_1(char *arg) +static int __init disable_timer_pin_setup(char *arg) { disable_timer_pin_1 = 1; return 0; } -early_param("disable_timer_pin_1", parse_disable_timer_pin_1); +early_param("disable_timer_pin_1", disable_timer_pin_setup); int timer_through_8259 __initdata; @@ -2397,6 +2732,8 @@ int timer_through_8259 __initdata; * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ * is so screwy. Thanks to Brian Perkins for testing/hacking this beast * fanatically on his truly buggy board. + * + * FIXME: really need to revamp this for all platforms. */ static inline void __init check_timer(void) { @@ -2408,8 +2745,8 @@ static inline void __init check_timer(void) local_irq_save(flags); - ver = apic_read(APIC_LVR); - ver = GET_APIC_VERSION(ver); + ver = apic_read(APIC_LVR); + ver = GET_APIC_VERSION(ver); /* * get/set the timer IRQ vector: @@ -2428,7 +2765,9 @@ static inline void __init check_timer(void) */ apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); init_8259A(1); +#ifdef CONFIG_X86_32 timer_ack = (nmi_watchdog == NMI_IO_APIC && !APIC_INTEGRATED(ver)); +#endif pin1 = find_isa_irq_pin(0, mp_INT); apic1 = find_isa_irq_apic(0, mp_INT); @@ -2447,6 +2786,10 @@ static inline void __init check_timer(void) * 8259A. */ if (pin1 == -1) { +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + panic("BIOS bug: timer not connected to IO-APIC"); +#endif pin1 = pin2; apic1 = apic2; no_pin1 = 1; @@ -2473,6 +2816,10 @@ static inline void __init check_timer(void) clear_IO_APIC_pin(0, pin1); goto out; } +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + panic("timer doesn't work through Interrupt-remapped IO-APIC"); +#endif clear_IO_APIC_pin(apic1, pin1); if (!no_pin1) apic_printk(APIC_QUIET, KERN_ERR "..MP-BIOS bug: " @@ -2512,7 +2859,9 @@ static inline void __init check_timer(void) "through the IO-APIC - disabling NMI Watchdog!\n"); nmi_watchdog = NMI_NONE; } +#ifdef CONFIG_X86_32 timer_ack = 0; +#endif apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as Virtual Wire IRQ...\n"); @@ -2570,17 +2919,25 @@ out: void __init setup_IO_APIC(void) { + +#ifdef CONFIG_X86_32 enable_IO_APIC(); +#else + /* + * calling enable_IO_APIC() is moved to setup_local_APIC for BP + */ +#endif io_apic_irqs = ~PIC_IRQS; - printk("ENABLING IO-APIC IRQs\n"); - - /* - * Set up IO-APIC IRQ routing. - */ - if (!acpi_ioapic) - setup_ioapic_ids_from_mpc(); + apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); + /* + * Set up IO-APIC IRQ routing. + */ +#ifdef CONFIG_X86_32 + if (!acpi_ioapic) + setup_ioapic_ids_from_mpc(); +#endif sync_Arb_IDs(); setup_IO_APIC_irqs(); init_IO_APIC_traps(); @@ -2588,15 +2945,15 @@ void __init setup_IO_APIC(void) } /* - * Called after all the initialization is done. If we didnt find any - * APIC bugs then we can allow the modify fast path + * Called after all the initialization is done. If we didnt find any + * APIC bugs then we can allow the modify fast path */ static int __init io_apic_bug_finalize(void) { - if (sis_apic_bug == -1) - sis_apic_bug = 0; - return 0; + if (sis_apic_bug == -1) + sis_apic_bug = 0; + return 0; } late_initcall(io_apic_bug_finalize); @@ -2605,7 +2962,7 @@ struct sysfs_ioapic_data { struct sys_device dev; struct IO_APIC_route_entry entry[0]; }; -static struct sysfs_ioapic_data *mp_ioapic_data[MAX_IO_APICS]; +static struct sysfs_ioapic_data * mp_ioapic_data[MAX_IO_APICS]; static int ioapic_suspend(struct sys_device *dev, pm_message_t state) { @@ -2615,8 +2972,8 @@ static int ioapic_suspend(struct sys_device *dev, pm_message_t state) data = container_of(dev, struct sysfs_ioapic_data, dev); entry = data->entry; - for (i = 0; i < nr_ioapic_registers[dev->id]; i++) - entry[i] = ioapic_read_entry(dev->id, i); + for (i = 0; i < nr_ioapic_registers[dev->id]; i ++, entry ++ ) + *entry = ioapic_read_entry(dev->id, i); return 0; } @@ -2653,14 +3010,14 @@ static struct sysdev_class ioapic_sysdev_class = { static int __init ioapic_init_sysfs(void) { - struct sys_device *dev; - int i, size, error = 0; + struct sys_device * dev; + int i, size, error; error = sysdev_class_register(&ioapic_sysdev_class); if (error) return error; - for (i = 0; i < nr_ioapics; i++) { + for (i = 0; i < nr_ioapics; i++ ) { size = sizeof(struct sys_device) + nr_ioapic_registers[i] * sizeof(struct IO_APIC_route_entry); mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL); @@ -2691,18 +3048,18 @@ device_initcall(ioapic_init_sysfs); unsigned int create_irq_nr(unsigned int irq_want) { /* Allocate an unused irq */ - unsigned int irq, new; + unsigned int irq; + unsigned int new; unsigned long flags; struct irq_cfg *cfg_new; #ifndef CONFIG_HAVE_SPARSE_IRQ - /* only can use bus/dev/fn.. when per_cpu vector is used */ irq_want = nr_irqs - 1; #endif irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = (nr_irqs - 1); new > 0; new--) { + for (new = irq_want; new > 0; new--) { if (platform_legacy_irq(new)) continue; cfg_new = irq_cfg(new); @@ -2725,7 +3082,14 @@ unsigned int create_irq_nr(unsigned int irq_want) int create_irq(void) { - return create_irq_nr(nr_irqs - 1); + int irq; + + irq = create_irq_nr(nr_irqs - 1); + + if (irq == 0) + irq = -1; + + return irq; } void destroy_irq(unsigned int irq) @@ -2734,6 +3098,9 @@ void destroy_irq(unsigned int irq) dynamic_irq_cleanup(irq); +#ifdef CONFIG_INTR_REMAP + free_irte(irq); +#endif spin_lock_irqsave(&vector_lock, flags); __clear_irq_vector(irq); spin_unlock_irqrestore(&vector_lock, flags); @@ -2759,25 +3126,54 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); - msg->address_hi = MSI_ADDR_BASE_HI; - msg->address_lo = - MSI_ADDR_BASE_LO | - ((INT_DEST_MODE == 0) ? - MSI_ADDR_DEST_MODE_PHYSICAL: - MSI_ADDR_DEST_MODE_LOGICAL) | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_ADDR_REDIRECTION_CPU: - MSI_ADDR_REDIRECTION_LOWPRI) | - MSI_ADDR_DEST_ID(dest); - - msg->data = - MSI_DATA_TRIGGER_EDGE | - MSI_DATA_LEVEL_ASSERT | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_DATA_DELIVERY_FIXED: - MSI_DATA_DELIVERY_LOWPRI) | - MSI_DATA_VECTOR(cfg->vector); +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + struct irte irte; + int ir_index; + u16 sub_handle; + + ir_index = map_irq_to_irte_handle(irq, &sub_handle); + BUG_ON(ir_index == -1); + + memset (&irte, 0, sizeof(irte)); + + irte.present = 1; + irte.dst_mode = INT_DEST_MODE; + irte.trigger_mode = 0; /* edge */ + irte.dlvry_mode = INT_DELIVERY_MODE; + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + modify_irte(irq, &irte); + + msg->address_hi = MSI_ADDR_BASE_HI; + msg->data = sub_handle; + msg->address_lo = MSI_ADDR_BASE_LO | MSI_ADDR_IR_EXT_INT | + MSI_ADDR_IR_SHV | + MSI_ADDR_IR_INDEX1(ir_index) | + MSI_ADDR_IR_INDEX2(ir_index); + } else +#endif + { + msg->address_hi = MSI_ADDR_BASE_HI; + msg->address_lo = + MSI_ADDR_BASE_LO | + ((INT_DEST_MODE == 0) ? + MSI_ADDR_DEST_MODE_PHYSICAL: + MSI_ADDR_DEST_MODE_LOGICAL) | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_ADDR_REDIRECTION_CPU: + MSI_ADDR_REDIRECTION_LOWPRI) | + MSI_ADDR_DEST_ID(dest); + msg->data = + MSI_DATA_TRIGGER_EDGE | + MSI_DATA_LEVEL_ASSERT | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_DATA_DELIVERY_FIXED: + MSI_DATA_DELIVERY_LOWPRI) | + MSI_DATA_VECTOR(cfg->vector); + } return err; } @@ -2788,6 +3184,7 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) struct msi_msg msg; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2808,8 +3205,61 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); write_msi_msg(irq, &msg); - irq_to_desc(irq)->affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } + +#ifdef CONFIG_INTR_REMAP +/* + * Migrate the MSI irq to another cpumask. This migration is + * done in the process context using interrupt-remapping hardware. + */ +static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + unsigned int dest; + cpumask_t tmp, cleanup_mask; + struct irte irte; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (get_irte(irq, &irte)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + /* + * atomically update the IRTE with the new destination and vector. + */ + modify_irte(irq, &irte); + + /* + * After this point, all the interrupts will start arriving + * at the new destination. So, time to cleanup the previous + * vector allocation. + */ + if (cfg->move_in_progress) { + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } + + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif #endif /* CONFIG_SMP */ /* @@ -2827,6 +3277,45 @@ static struct irq_chip msi_chip = { .retrigger = ioapic_retrigger_irq, }; +#ifdef CONFIG_INTR_REMAP +static struct irq_chip msi_ir_chip = { + .name = "IR-PCI-MSI", + .unmask = unmask_msi_irq, + .mask = mask_msi_irq, + .ack = ack_x2apic_edge, +#ifdef CONFIG_SMP + .set_affinity = ir_set_msi_irq_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +/* + * Map the PCI dev to the corresponding remapping hardware unit + * and allocate 'nvec' consecutive interrupt-remapping table entries + * in it. + */ +static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) +{ + struct intel_iommu *iommu; + int index; + + iommu = map_dev_to_ir(dev); + if (!iommu) { + printk(KERN_ERR + "Unable to map PCI %s to iommu\n", pci_name(dev)); + return -ENOENT; + } + + index = alloc_irte(iommu, irq, nvec); + if (index < 0) { + printk(KERN_ERR + "Unable to allocate %d IRTE for PCI %s\n", nvec, + pci_name(dev)); + return -ENOSPC; + } + return index; +} +#endif static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) { @@ -2840,7 +3329,17 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) set_irq_msi(irq, desc); write_msi_msg(irq, &msg); - set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + struct irq_desc *desc = irq_to_desc(irq); + /* + * irq migration in process context + */ + desc->status |= IRQ_MOVE_PCNTXT; + set_irq_chip_and_handler_name(irq, &msi_ir_chip, handle_edge_irq, "edge"); + } else +#endif + set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); return 0; } @@ -2859,59 +3358,164 @@ static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) { - int irq, ret; - + unsigned int irq; + int ret; unsigned int irq_want; irq_want = build_irq_for_pci_dev(dev) + 0x100; irq = create_irq_nr(irq_want); - if (irq == 0) return -1; +#ifdef CONFIG_INTR_REMAP + if (!intr_remapping_enabled) + goto no_ir; + + ret = msi_alloc_irte(dev, irq, 1); + if (ret < 0) + goto error; +no_ir: +#endif ret = setup_msi_irq(dev, desc, irq); if (ret < 0) { destroy_irq(irq); return ret; - } - + } return 0; + +#ifdef CONFIG_INTR_REMAP +error: + destroy_irq(irq); + return ret; +#endif } int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) { - unsigned int irq; - int ret, sub_handle; - struct msi_desc *desc; - unsigned int irq_want; - - irq_want = build_irq_for_pci_dev(dev) + 0x100; - sub_handle = 0; - list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq_nr(irq_want--); - if (irq == 0) - return -1; - ret = setup_msi_irq(dev, desc, irq); - if (ret < 0) - goto error; - sub_handle++; - } - return 0; + unsigned int irq; + int ret, sub_handle; + struct msi_desc *desc; + unsigned int irq_want; + +#ifdef CONFIG_INTR_REMAP + struct intel_iommu *iommu = 0; + int index = 0; +#endif + + irq_want = build_irq_for_pci_dev(dev) + 0x100; + sub_handle = 0; + list_for_each_entry(desc, &dev->msi_list, list) { + irq = create_irq_nr(irq_want--); + if (irq == 0) + return -1; +#ifdef CONFIG_INTR_REMAP + if (!intr_remapping_enabled) + goto no_ir; + + if (!sub_handle) { + /* + * allocate the consecutive block of IRTE's + * for 'nvec' + */ + index = msi_alloc_irte(dev, irq, nvec); + if (index < 0) { + ret = index; + goto error; + } + } else { + iommu = map_dev_to_ir(dev); + if (!iommu) { + ret = -ENOENT; + goto error; + } + /* + * setup the mapping between the irq and the IRTE + * base index, the sub_handle pointing to the + * appropriate interrupt remap table entry. + */ + set_irte_irq(irq, iommu, index, sub_handle); + } +no_ir: +#endif + ret = setup_msi_irq(dev, desc, irq); + if (ret < 0) + goto error; + sub_handle++; + } + return 0; error: - destroy_irq(irq); - return ret; + destroy_irq(irq); + return ret; } - void arch_teardown_msi_irq(unsigned int irq) { destroy_irq(irq); } -#endif /* CONFIG_PCI_MSI */ +#ifdef CONFIG_DMAR +#ifdef CONFIG_SMP +static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + struct msi_msg msg; + unsigned int dest; + cpumask_t tmp; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + dmar_msi_read(irq, &msg); + + msg.data &= ~MSI_DATA_VECTOR_MASK; + msg.data |= MSI_DATA_VECTOR(cfg->vector); + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; + msg.address_lo |= MSI_ADDR_DEST_ID(dest); + + dmar_msi_write(irq, &msg); + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif /* CONFIG_SMP */ + +struct irq_chip dmar_msi_type = { + .name = "DMAR_MSI", + .unmask = dmar_msi_unmask, + .mask = dmar_msi_mask, + .ack = ack_apic_edge, +#ifdef CONFIG_SMP + .set_affinity = dmar_msi_set_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +int arch_setup_dmar_msi(unsigned int irq) +{ + int ret; + struct msi_msg msg; + ret = msi_compose_msg(NULL, irq, &msg); + if (ret < 0) + return ret; + dmar_msi_write(irq, &msg); + set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq, + "edge"); + return 0; +} +#endif + +#endif /* CONFIG_PCI_MSI */ /* * Hypertransport interrupt support */ @@ -2938,6 +3542,7 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp; + struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -2951,7 +3556,8 @@ static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) dest = cpu_mask_to_apicid(tmp); target_ht_irq(irq, dest, cfg->vector); - irq_to_desc(irq)->affinity = mask; + desc = irq_to_desc(irq); + desc->affinity = mask; } #endif @@ -2974,7 +3580,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) tmp = TARGET_CPUS; err = assign_irq_vector(irq, tmp); - if ( !err) { + if (!err) { struct ht_irq_msg msg; unsigned dest; @@ -3007,11 +3613,12 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) #endif /* CONFIG_HT_IRQ */ /* -------------------------------------------------------------------------- - ACPI-based IOAPIC Configuration + ACPI-based IOAPIC Configuration -------------------------------------------------------------------------- */ #ifdef CONFIG_ACPI +#ifdef CONFIG_X86_32 int __init io_apic_get_unique_id(int ioapic, int apic_id) { union IO_APIC_reg_00 reg_00; @@ -3086,7 +3693,6 @@ int __init io_apic_get_unique_id(int ioapic, int apic_id) return apic_id; } - int __init io_apic_get_version(int ioapic) { union IO_APIC_reg_01 reg_01; @@ -3098,9 +3704,9 @@ int __init io_apic_get_version(int ioapic) return reg_01.bits.version; } +#endif - -int __init io_apic_get_redir_entries(int ioapic) +int __init io_apic_get_redir_entries (int ioapic) { union IO_APIC_reg_01 reg_01; unsigned long flags; @@ -3113,10 +3719,10 @@ int __init io_apic_get_redir_entries(int ioapic) } -int io_apic_set_pci_routing(int ioapic, int pin, int irq, int triggering, int polarity) +int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) { if (!IO_APIC_IRQ(irq)) { - printk(KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", + apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", ioapic); return -EINVAL; } @@ -3132,6 +3738,7 @@ int io_apic_set_pci_routing(int ioapic, int pin, int irq, int triggering, int po return 0; } + int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) { int i; @@ -3163,7 +3770,6 @@ void __init setup_ioapic_dest(void) { int pin, ioapic, irq, irq_entry; struct irq_cfg *cfg; - struct irq_desc *desc; if (skip_ioapic_setup == 1) return; @@ -3184,43 +3790,124 @@ void __init setup_ioapic_dest(void) setup_IO_APIC_irq(ioapic, pin, irq, irq_trigger(irq_entry), irq_polarity(irq_entry)); - else { - desc = irq_to_desc(irq); +#ifdef CONFIG_INTR_REMAP + else if (intr_remapping_enabled) + set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); +#endif + else set_ioapic_affinity_irq(irq, TARGET_CPUS); - } } } } #endif +#ifdef CONFIG_X86_64 +#define IOAPIC_RESOURCE_NAME_SIZE 11 + +static struct resource *ioapic_resources; + +static struct resource * __init ioapic_setup_resources(void) +{ + unsigned long n; + struct resource *res; + char *mem; + int i; + + if (nr_ioapics <= 0) + return NULL; + + n = IOAPIC_RESOURCE_NAME_SIZE + sizeof(struct resource); + n *= nr_ioapics; + + mem = alloc_bootmem(n); + res = (void *)mem; + + if (mem != NULL) { + mem += sizeof(struct resource) * nr_ioapics; + + for (i = 0; i < nr_ioapics; i++) { + res[i].name = mem; + res[i].flags = IORESOURCE_MEM | IORESOURCE_BUSY; + sprintf(mem, "IOAPIC %u", i); + mem += IOAPIC_RESOURCE_NAME_SIZE; + } + } + + ioapic_resources = res; + + return res; +} +#endif + void __init ioapic_init_mappings(void) { unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; int i; +#ifdef CONFIG_X86_64 + struct resource *ioapic_res; + ioapic_res = ioapic_setup_resources(); +#endif for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { ioapic_phys = mp_ioapics[i].mp_apicaddr; - if (!ioapic_phys) { - printk(KERN_ERR - "WARNING: bogus zero IO-APIC " - "address found in MPTABLE, " - "disabling IO/APIC support!\n"); - smp_found_config = 0; - skip_ioapic_setup = 1; - goto fake_ioapic_page; - } +#ifdef CONFIG_X86_32 + if (!ioapic_phys) { + printk(KERN_ERR + "WARNING: bogus zero IO-APIC " + "address found in MPTABLE, " + "disabling IO/APIC support!\n"); + smp_found_config = 0; + skip_ioapic_setup = 1; + goto fake_ioapic_page; + } +#endif } else { +#ifdef CONFIG_X86_32 fake_ioapic_page: +#endif ioapic_phys = (unsigned long) - alloc_bootmem_pages(PAGE_SIZE); + alloc_bootmem_pages(PAGE_SIZE); ioapic_phys = __pa(ioapic_phys); } set_fixmap_nocache(idx, ioapic_phys); - printk(KERN_DEBUG "mapped IOAPIC to %08lx (%08lx)\n", - __fix_to_virt(idx), ioapic_phys); + apic_printk(APIC_VERBOSE, + "mapped IOAPIC to %08lx (%08lx)\n", + __fix_to_virt(idx), ioapic_phys); idx++; + +#ifdef CONFIG_X86_64 + if (ioapic_res != NULL) { + ioapic_res->start = ioapic_phys; + ioapic_res->end = ioapic_phys + (4 * 1024) - 1; + ioapic_res++; + } +#endif } } +#ifdef CONFIG_X86_64 +static int __init ioapic_insert_resources(void) +{ + int i; + struct resource *r = ioapic_resources; + + if (!r) { + printk(KERN_ERR + "IO APIC resources could be not be allocated.\n"); + return -1; + } + + for (i = 0; i < nr_ioapics; i++) { + insert_resource(&iomem_resource, r); + r++; + } + + return 0; +} + +/* Insert the IO APIC resources after PCI initialization has occured to handle + * IO APICS that are mapped in on a BAR in PCI space. */ +late_initcall(ioapic_insert_resources); +#endif -- cgit v1.2.3 From 26d347c2c035b1f4c5b3c5094f3046db9ec920f5 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:42 -0700 Subject: rename io_apic_64.c and io_apic_32.c to io_apic.c The two files are now line by line equal. (sans a printk) Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/io_apic.c | 3913 ++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/io_apic_32.c | 3913 ------------------------------------------ arch/x86/kernel/io_apic_64.c | 3913 ------------------------------------------ 4 files changed, 3914 insertions(+), 7827 deletions(-) create mode 100644 arch/x86/kernel/io_apic.c delete mode 100644 arch/x86/kernel/io_apic_32.c delete mode 100644 arch/x86/kernel/io_apic_64.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 0d41f0343dc0..7264f9c3af8f 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -61,7 +61,7 @@ obj-$(CONFIG_X86_64_SMP) += tsc_sync.o smpcommon.o obj-$(CONFIG_X86_TRAMPOLINE) += trampoline_$(BITS).o obj-$(CONFIG_X86_MPPARSE) += mpparse.o obj-$(CONFIG_X86_LOCAL_APIC) += apic_$(BITS).o nmi.o -obj-$(CONFIG_X86_IO_APIC) += io_apic_$(BITS).o +obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c new file mode 100644 index 000000000000..fba6d6ee3480 --- /dev/null +++ b/arch/x86/kernel/io_apic.c @@ -0,0 +1,3913 @@ +/* + * Intel IO-APIC support for multi-Pentium hosts. + * + * Copyright (C) 1997, 1998, 1999, 2000 Ingo Molnar, Hajnalka Szabo + * + * Many thanks to Stig Venaas for trying out countless experimental + * patches and reporting/debugging problems patiently! + * + * (c) 1999, Multiple IO-APIC support, developed by + * Ken-ichi Yaku and + * Hidemi Kishimoto , + * further tested and cleaned up by Zach Brown + * and Ingo Molnar + * + * Fixes + * Maciej W. Rozycki : Bits for genuine 82489DX APICs; + * thanks to Eric Gilmore + * and Rolf G. Tews + * for testing these extensively + * Paul Diefenbaugh : Added full ACPI support + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include /* time_after() */ +#ifdef CONFIG_ACPI +#include +#endif +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#define __apicdebuginit(type) static type __init + +/* + * Is the SiS APIC rmw bug present ? + * -1 = don't know, 0 = no, 1 = yes + */ +int sis_apic_bug = -1; + +static DEFINE_SPINLOCK(ioapic_lock); +static DEFINE_SPINLOCK(vector_lock); + +int first_free_entry; +/* + * Rough estimation of how many shared IRQs there are, can + * be changed anytime. + */ +int pin_map_size; + +/* + * # of IRQ routing registers + */ +int nr_ioapic_registers[MAX_IO_APICS]; + +/* I/O APIC entries */ +struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; +int nr_ioapics; + +/* MP IRQ source entries */ +struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; + +/* # of MP IRQ source entries */ +int mp_irq_entries; + +#if defined (CONFIG_MCA) || defined (CONFIG_EISA) +int mp_bus_id_to_type[MAX_MP_BUSSES]; +#endif + +DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); + +int skip_ioapic_setup; + +static int __init parse_noapic(char *str) +{ + /* disable IO-APIC */ + disable_ioapic_setup(); + return 0; +} +early_param("noapic", parse_noapic); + +struct irq_cfg; +struct irq_pin_list; +struct irq_cfg { + unsigned int irq; + struct irq_cfg *next; + struct irq_pin_list *irq_2_pin; + cpumask_t domain; + cpumask_t old_domain; + unsigned move_cleanup_count; + u8 vector; + u8 move_in_progress : 1; +}; + +/* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ +static struct irq_cfg irq_cfg_legacy[] __initdata = { + [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, + [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, + [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, + [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, + [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, + [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, + [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, + [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, + [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, + [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, + [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, + [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, + [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, + [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, + [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, + [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, +}; + +static struct irq_cfg irq_cfg_init = { .irq = -1U, }; +/* need to be biger than size of irq_cfg_legacy */ +static int nr_irq_cfg = 32; + +static int __init parse_nr_irq_cfg(char *arg) +{ + if (arg) { + nr_irq_cfg = simple_strtoul(arg, NULL, 0); + if (nr_irq_cfg < 32) + nr_irq_cfg = 32; + } + return 0; +} + +early_param("nr_irq_cfg", parse_nr_irq_cfg); + +static void init_one_irq_cfg(struct irq_cfg *cfg) +{ + memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); +} + +static struct irq_cfg *irq_cfgx; +static struct irq_cfg *irq_cfgx_free; +static void __init init_work(void *data) +{ + struct dyn_array *da = data; + struct irq_cfg *cfg; + int legacy_count; + int i; + + cfg = *da->name; + + memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); + + legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + for (i = legacy_count; i < *da->nr; i++) + init_one_irq_cfg(&cfg[i]); + + for (i = 1; i < *da->nr; i++) + cfg[i-1].next = &cfg[i]; + + irq_cfgx_free = &irq_cfgx[legacy_count]; + irq_cfgx[legacy_count - 1].next = NULL; +} + +#define for_each_irq_cfg(cfg) \ + for (cfg = irq_cfgx; cfg; cfg = cfg->next) + +DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); + +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + struct irq_cfg *cfg; + + cfg = irq_cfgx; + while (cfg) { + if (cfg->irq == irq) + return cfg; + + cfg = cfg->next; + } + + return NULL; +} + +static struct irq_cfg *irq_cfg_alloc(unsigned int irq) +{ + struct irq_cfg *cfg, *cfg_pri; + int i; + int count = 0; + + cfg_pri = cfg = irq_cfgx; + while (cfg) { + if (cfg->irq == irq) + return cfg; + + cfg_pri = cfg; + cfg = cfg->next; + count++; + } + + if (!irq_cfgx_free) { + unsigned long phys; + unsigned long total_bytes; + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); + + total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; + if (after_bootmem) + cfg = kzalloc(total_bytes, GFP_ATOMIC); + else + cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); + + if (!cfg) + panic("please boot with nr_irq_cfg= %d\n", count * 2); + + phys = __pa(cfg); + printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); + + for (i = 0; i < nr_irq_cfg; i++) + init_one_irq_cfg(&cfg[i]); + + for (i = 1; i < nr_irq_cfg; i++) + cfg[i-1].next = &cfg[i]; + + irq_cfgx_free = cfg; + } + + cfg = irq_cfgx_free; + irq_cfgx_free = irq_cfgx_free->next; + cfg->next = NULL; + if (cfg_pri) + cfg_pri->next = cfg; + else + irq_cfgx = cfg; + cfg->irq = irq; + printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); +#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG + { + /* dump the results */ + struct irq_cfg *cfg; + unsigned long phys; + unsigned long bytes = sizeof(struct irq_cfg); + + printk(KERN_DEBUG "=========================== %d\n", irq); + printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); + for_each_irq_cfg(cfg) { + phys = __pa(cfg); + printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); + } + printk(KERN_DEBUG "===========================\n"); + } +#endif + return cfg; +} + +/* + * This is performance-critical, we want to do it O(1) + * + * the indexing order of this array favors 1:1 mappings + * between pins and IRQs. + */ + +struct irq_pin_list { + int apic, pin; + struct irq_pin_list *next; +}; + +static struct irq_pin_list *irq_2_pin_head; +/* fill one page ? */ +static int nr_irq_2_pin = 0x100; +static struct irq_pin_list *irq_2_pin_ptr; +static void __init irq_2_pin_init_work(void *data) +{ + struct dyn_array *da = data; + struct irq_pin_list *pin; + int i; + + pin = *da->name; + + for (i = 1; i < *da->nr; i++) + pin[i-1].next = &pin[i]; + + irq_2_pin_ptr = &pin[0]; +} +DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); + +static struct irq_pin_list *get_one_free_irq_2_pin(void) +{ + struct irq_pin_list *pin; + int i; + + pin = irq_2_pin_ptr; + + if (pin) { + irq_2_pin_ptr = pin->next; + pin->next = NULL; + return pin; + } + + /* + * we run out of pre-allocate ones, allocate more + */ + printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); + + if (after_bootmem) + pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, + GFP_ATOMIC); + else + pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * + nr_irq_2_pin, PAGE_SIZE, 0); + + if (!pin) + panic("can not get more irq_2_pin\n"); + + for (i = 1; i < nr_irq_2_pin; i++) + pin[i-1].next = &pin[i]; + + irq_2_pin_ptr = pin->next; + pin->next = NULL; + + return pin; +} + +struct io_apic { + unsigned int index; + unsigned int unused[3]; + unsigned int data; +}; + +static __attribute_const__ struct io_apic __iomem *io_apic_base(int idx) +{ + return (void __iomem *) __fix_to_virt(FIX_IO_APIC_BASE_0 + idx) + + (mp_ioapics[idx].mp_apicaddr & ~PAGE_MASK); +} + +static inline unsigned int io_apic_read(unsigned int apic, unsigned int reg) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + writel(reg, &io_apic->index); + return readl(&io_apic->data); +} + +static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned int value) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + writel(reg, &io_apic->index); + writel(value, &io_apic->data); +} + +/* + * Re-write a value: to be used for read-modify-write + * cycles where the read already set up the index register. + * + * Older SiS APIC requires we rewrite the index register + */ +static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + if (sis_apic_bug) + writel(reg, &io_apic->index); + writel(value, &io_apic->data); +} + +#ifdef CONFIG_X86_64 +static bool io_apic_level_ack_pending(unsigned int irq) +{ + struct irq_pin_list *entry; + unsigned long flags; + struct irq_cfg *cfg = irq_cfg(irq); + + spin_lock_irqsave(&ioapic_lock, flags); + entry = cfg->irq_2_pin; + for (;;) { + unsigned int reg; + int pin; + + if (!entry) + break; + pin = entry->pin; + reg = io_apic_read(entry->apic, 0x10 + pin*2); + /* Is the remote IRR bit set? */ + if (reg & IO_APIC_REDIR_REMOTE_IRR) { + spin_unlock_irqrestore(&ioapic_lock, flags); + return true; + } + if (!entry->next) + break; + entry = entry->next; + } + spin_unlock_irqrestore(&ioapic_lock, flags); + + return false; +} +#endif + +union entry_union { + struct { u32 w1, w2; }; + struct IO_APIC_route_entry entry; +}; + +static struct IO_APIC_route_entry ioapic_read_entry(int apic, int pin) +{ + union entry_union eu; + unsigned long flags; + spin_lock_irqsave(&ioapic_lock, flags); + eu.w1 = io_apic_read(apic, 0x10 + 2 * pin); + eu.w2 = io_apic_read(apic, 0x11 + 2 * pin); + spin_unlock_irqrestore(&ioapic_lock, flags); + return eu.entry; +} + +/* + * When we write a new IO APIC routing entry, we need to write the high + * word first! If the mask bit in the low word is clear, we will enable + * the interrupt, and we need to make sure the entry is fully populated + * before that happens. + */ +static void +__ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) +{ + union entry_union eu; + eu.entry = e; + io_apic_write(apic, 0x11 + 2*pin, eu.w2); + io_apic_write(apic, 0x10 + 2*pin, eu.w1); +} + +static void ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) +{ + unsigned long flags; + spin_lock_irqsave(&ioapic_lock, flags); + __ioapic_write_entry(apic, pin, e); + spin_unlock_irqrestore(&ioapic_lock, flags); +} + +/* + * When we mask an IO APIC routing entry, we need to write the low + * word first, in order to set the mask bit before we change the + * high bits! + */ +static void ioapic_mask_entry(int apic, int pin) +{ + unsigned long flags; + union entry_union eu = { .entry.mask = 1 }; + + spin_lock_irqsave(&ioapic_lock, flags); + io_apic_write(apic, 0x10 + 2*pin, eu.w1); + io_apic_write(apic, 0x11 + 2*pin, eu.w2); + spin_unlock_irqrestore(&ioapic_lock, flags); +} + +#ifdef CONFIG_SMP +static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) +{ + int apic, pin; + struct irq_cfg *cfg; + struct irq_pin_list *entry; + + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; + for (;;) { + unsigned int reg; + + if (!entry) + break; + + apic = entry->apic; + pin = entry->pin; +#ifdef CONFIG_INTR_REMAP + /* + * With interrupt-remapping, destination information comes + * from interrupt-remapping table entry. + */ + if (!irq_remapped(irq)) + io_apic_write(apic, 0x11 + pin*2, dest); +#else + io_apic_write(apic, 0x11 + pin*2, dest); +#endif + reg = io_apic_read(apic, 0x10 + pin*2); + reg &= ~IO_APIC_REDIR_VECTOR_MASK; + reg |= vector; + io_apic_modify(apic, 0x10 + pin*2, reg); + if (!entry->next) + break; + entry = entry->next; + } +} + +static int assign_irq_vector(int irq, cpumask_t mask); + +static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + unsigned long flags; + unsigned int dest; + cpumask_t tmp; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + cfg = irq_cfg(irq); + if (assign_irq_vector(irq, mask)) + return; + + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + /* + * Only the high 8 bits are valid. + */ + dest = SET_APIC_LOGICAL_ID(dest); + + desc = irq_to_desc(irq); + spin_lock_irqsave(&ioapic_lock, flags); + __target_IO_APIC_irq(irq, dest, cfg->vector); + desc->affinity = mask; + spin_unlock_irqrestore(&ioapic_lock, flags); +} +#endif /* CONFIG_SMP */ + +/* + * The common case is 1:1 IRQ<->pin mappings. Sometimes there are + * shared ISA-space IRQs, so we have to support them. We are super + * fast in the common case, and fast for shared ISA-space IRQs. + */ +static void add_pin_to_irq(unsigned int irq, int apic, int pin) +{ + struct irq_cfg *cfg; + struct irq_pin_list *entry; + + /* first time to refer irq_cfg, so with new */ + cfg = irq_cfg_alloc(irq); + entry = cfg->irq_2_pin; + if (!entry) { + entry = get_one_free_irq_2_pin(); + cfg->irq_2_pin = entry; + entry->apic = apic; + entry->pin = pin; + printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); + return; + } + + while (entry->next) { + /* not again, please */ + if (entry->apic == apic && entry->pin == pin) + return; + + entry = entry->next; + } + + entry->next = get_one_free_irq_2_pin(); + entry = entry->next; + entry->apic = apic; + entry->pin = pin; + printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); +} + +/* + * Reroute an IRQ to a different pin. + */ +static void __init replace_pin_at_irq(unsigned int irq, + int oldapic, int oldpin, + int newapic, int newpin) +{ + struct irq_cfg *cfg = irq_cfg(irq); + struct irq_pin_list *entry = cfg->irq_2_pin; + int replaced = 0; + + while (entry) { + if (entry->apic == oldapic && entry->pin == oldpin) { + entry->apic = newapic; + entry->pin = newpin; + replaced = 1; + /* every one is different, right? */ + break; + } + entry = entry->next; + } + + /* why? call replace before add? */ + if (!replaced) + add_pin_to_irq(irq, newapic, newpin); +} + +#ifdef CONFIG_X86_64 +/* + * Synchronize the IO-APIC and the CPU by doing + * a dummy read from the IO-APIC + */ +static inline void io_apic_sync(unsigned int apic) +{ + struct io_apic __iomem *io_apic = io_apic_base(apic); + readl(&io_apic->data); +} + +#define __DO_ACTION(R, ACTION, FINAL) \ + \ +{ \ + int pin; \ + struct irq_cfg *cfg; \ + struct irq_pin_list *entry; \ + \ + cfg = irq_cfg(irq); \ + entry = cfg->irq_2_pin; \ + for (;;) { \ + unsigned int reg; \ + if (!entry) \ + break; \ + pin = entry->pin; \ + reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ + reg ACTION; \ + io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ + FINAL; \ + if (!entry->next) \ + break; \ + entry = entry->next; \ + } \ +} + +#define DO_ACTION(name,R,ACTION, FINAL) \ + \ + static void name##_IO_APIC_irq (unsigned int irq) \ + __DO_ACTION(R, ACTION, FINAL) + +/* mask = 1 */ +DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) + +/* mask = 0 */ +DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) + +#else + +static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) +{ + struct irq_cfg *cfg; + struct irq_pin_list *entry; + unsigned int pin, reg; + + cfg = irq_cfg(irq); + entry = cfg->irq_2_pin; + for (;;) { + if (!entry) + break; + pin = entry->pin; + reg = io_apic_read(entry->apic, 0x10 + pin*2); + reg &= ~disable; + reg |= enable; + io_apic_modify(entry->apic, 0x10 + pin*2, reg); + if (!entry->next) + break; + entry = entry->next; + } +} + +/* mask = 1 */ +static void __mask_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, 0); +} + +/* mask = 0 */ +static void __unmask_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, 0, IO_APIC_REDIR_MASKED); +} + +/* mask = 1, trigger = 0 */ +static void __mask_and_edge_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, + IO_APIC_REDIR_LEVEL_TRIGGER); +} + +/* mask = 0, trigger = 1 */ +static void __unmask_and_level_IO_APIC_irq(unsigned int irq) +{ + __modify_IO_APIC_irq(irq, IO_APIC_REDIR_LEVEL_TRIGGER, + IO_APIC_REDIR_MASKED); +} + +#endif + +static void mask_IO_APIC_irq (unsigned int irq) +{ + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + __mask_IO_APIC_irq(irq); + spin_unlock_irqrestore(&ioapic_lock, flags); +} + +static void unmask_IO_APIC_irq (unsigned int irq) +{ + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + __unmask_IO_APIC_irq(irq); + spin_unlock_irqrestore(&ioapic_lock, flags); +} + +static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) +{ + struct IO_APIC_route_entry entry; + + /* Check delivery_mode to be sure we're not clearing an SMI pin */ + entry = ioapic_read_entry(apic, pin); + if (entry.delivery_mode == dest_SMI) + return; + /* + * Disable it in the IO-APIC irq-routing table: + */ + ioapic_mask_entry(apic, pin); +} + +static void clear_IO_APIC (void) +{ + int apic, pin; + + for (apic = 0; apic < nr_ioapics; apic++) + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) + clear_IO_APIC_pin(apic, pin); +} + +#if !defined(CONFIG_SMP) && defined(CONFIG_X86_32) +void send_IPI_self(int vector) +{ + unsigned int cfg; + + /* + * Wait for idle. + */ + apic_wait_icr_idle(); + cfg = APIC_DM_FIXED | APIC_DEST_SELF | vector | APIC_DEST_LOGICAL; + /* + * Send the IPI. The write to APIC_ICR fires this off. + */ + apic_write(APIC_ICR, cfg); +} +#endif /* !CONFIG_SMP && CONFIG_X86_32*/ + +#ifdef CONFIG_X86_32 +/* + * support for broken MP BIOSs, enables hand-redirection of PIRQ0-7 to + * specific CPU-side IRQs. + */ + +#define MAX_PIRQS 8 +static int pirq_entries [MAX_PIRQS]; +static int pirqs_enabled; + +static int __init ioapic_pirq_setup(char *str) +{ + int i, max; + int ints[MAX_PIRQS+1]; + + get_options(str, ARRAY_SIZE(ints), ints); + + for (i = 0; i < MAX_PIRQS; i++) + pirq_entries[i] = -1; + + pirqs_enabled = 1; + apic_printk(APIC_VERBOSE, KERN_INFO + "PIRQ redirection, working around broken MP-BIOS.\n"); + max = MAX_PIRQS; + if (ints[0] < MAX_PIRQS) + max = ints[0]; + + for (i = 0; i < max; i++) { + apic_printk(APIC_VERBOSE, KERN_DEBUG + "... PIRQ%d -> IRQ %d\n", i, ints[i+1]); + /* + * PIRQs are mapped upside down, usually. + */ + pirq_entries[MAX_PIRQS-i-1] = ints[i+1]; + } + return 1; +} + +__setup("pirq=", ioapic_pirq_setup); +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_INTR_REMAP +/* I/O APIC RTE contents at the OS boot up */ +static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; + +/* + * Saves and masks all the unmasked IO-APIC RTE's + */ +int save_mask_IO_APIC_setup(void) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + int apic, pin; + + /* + * The number of IO-APIC IRQ registers (== #pins): + */ + for (apic = 0; apic < nr_ioapics; apic++) { + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(apic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + nr_ioapic_registers[apic] = reg_01.bits.entries+1; + } + + for (apic = 0; apic < nr_ioapics; apic++) { + early_ioapic_entries[apic] = + kzalloc(sizeof(struct IO_APIC_route_entry) * + nr_ioapic_registers[apic], GFP_KERNEL); + if (!early_ioapic_entries[apic]) + return -ENOMEM; + } + + for (apic = 0; apic < nr_ioapics; apic++) + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + struct IO_APIC_route_entry entry; + + entry = early_ioapic_entries[apic][pin] = + ioapic_read_entry(apic, pin); + if (!entry.mask) { + entry.mask = 1; + ioapic_write_entry(apic, pin, entry); + } + } + return 0; +} + +void restore_IO_APIC_setup(void) +{ + int apic, pin; + + for (apic = 0; apic < nr_ioapics; apic++) + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) + ioapic_write_entry(apic, pin, + early_ioapic_entries[apic][pin]); +} + +void reinit_intr_remapped_IO_APIC(int intr_remapping) +{ + /* + * for now plain restore of previous settings. + * TBD: In the case of OS enabling interrupt-remapping, + * IO-APIC RTE's need to be setup to point to interrupt-remapping + * table entries. for now, do a plain restore, and wait for + * the setup_IO_APIC_irqs() to do proper initialization. + */ + restore_IO_APIC_setup(); +} +#endif + +/* + * Find the IRQ entry number of a certain pin. + */ +static int find_irq_entry(int apic, int pin, int type) +{ + int i; + + for (i = 0; i < mp_irq_entries; i++) + if (mp_irqs[i].mp_irqtype == type && + (mp_irqs[i].mp_dstapic == mp_ioapics[apic].mp_apicid || + mp_irqs[i].mp_dstapic == MP_APIC_ALL) && + mp_irqs[i].mp_dstirq == pin) + return i; + + return -1; +} + +/* + * Find the pin to which IRQ[irq] (ISA) is connected + */ +static int __init find_isa_irq_pin(int irq, int type) +{ + int i; + + for (i = 0; i < mp_irq_entries; i++) { + int lbus = mp_irqs[i].mp_srcbus; + + if (test_bit(lbus, mp_bus_not_pci) && + (mp_irqs[i].mp_irqtype == type) && + (mp_irqs[i].mp_srcbusirq == irq)) + + return mp_irqs[i].mp_dstirq; + } + return -1; +} + +static int __init find_isa_irq_apic(int irq, int type) +{ + int i; + + for (i = 0; i < mp_irq_entries; i++) { + int lbus = mp_irqs[i].mp_srcbus; + + if (test_bit(lbus, mp_bus_not_pci) && + (mp_irqs[i].mp_irqtype == type) && + (mp_irqs[i].mp_srcbusirq == irq)) + break; + } + if (i < mp_irq_entries) { + int apic; + for(apic = 0; apic < nr_ioapics; apic++) { + if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic) + return apic; + } + } + + return -1; +} + +/* + * Find a specific PCI IRQ entry. + * Not an __init, possibly needed by modules + */ +static int pin_2_irq(int idx, int apic, int pin); + +int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) +{ + int apic, i, best_guess = -1; + + apic_printk(APIC_DEBUG, "querying PCI -> IRQ mapping bus:%d, slot:%d, pin:%d.\n", + bus, slot, pin); + if (test_bit(bus, mp_bus_not_pci)) { + apic_printk(APIC_VERBOSE, "PCI BIOS passed nonexistent PCI bus %d!\n", bus); + return -1; + } + for (i = 0; i < mp_irq_entries; i++) { + int lbus = mp_irqs[i].mp_srcbus; + + for (apic = 0; apic < nr_ioapics; apic++) + if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic || + mp_irqs[i].mp_dstapic == MP_APIC_ALL) + break; + + if (!test_bit(lbus, mp_bus_not_pci) && + !mp_irqs[i].mp_irqtype && + (bus == lbus) && + (slot == ((mp_irqs[i].mp_srcbusirq >> 2) & 0x1f))) { + int irq = pin_2_irq(i,apic,mp_irqs[i].mp_dstirq); + + if (!(apic || IO_APIC_IRQ(irq))) + continue; + + if (pin == (mp_irqs[i].mp_srcbusirq & 3)) + return irq; + /* + * Use the first all-but-pin matching entry as a + * best-guess fuzzy result for broken mptables. + */ + if (best_guess < 0) + best_guess = irq; + } + } + return best_guess; +} + +EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); + +#if defined(CONFIG_EISA) || defined(CONFIG_MCA) +/* + * EISA Edge/Level control register, ELCR + */ +static int EISA_ELCR(unsigned int irq) +{ + if (irq < 16) { + unsigned int port = 0x4d0 + (irq >> 3); + return (inb(port) >> (irq & 7)) & 1; + } + apic_printk(APIC_VERBOSE, KERN_INFO + "Broken MPtable reports ISA irq %d\n", irq); + return 0; +} + +#endif + +/* ISA interrupts are always polarity zero edge triggered, + * when listed as conforming in the MP table. */ + +#define default_ISA_trigger(idx) (0) +#define default_ISA_polarity(idx) (0) + +/* EISA interrupts are always polarity zero and can be edge or level + * trigger depending on the ELCR value. If an interrupt is listed as + * EISA conforming in the MP table, that means its trigger type must + * be read in from the ELCR */ + +#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].mp_srcbusirq)) +#define default_EISA_polarity(idx) default_ISA_polarity(idx) + +/* PCI interrupts are always polarity one level triggered, + * when listed as conforming in the MP table. */ + +#define default_PCI_trigger(idx) (1) +#define default_PCI_polarity(idx) (1) + +/* MCA interrupts are always polarity zero level triggered, + * when listed as conforming in the MP table. */ + +#define default_MCA_trigger(idx) (1) +#define default_MCA_polarity(idx) default_ISA_polarity(idx) + +static int MPBIOS_polarity(int idx) +{ + int bus = mp_irqs[idx].mp_srcbus; + int polarity; + + /* + * Determine IRQ line polarity (high active or low active): + */ + switch (mp_irqs[idx].mp_irqflag & 3) + { + case 0: /* conforms, ie. bus-type dependent polarity */ + if (test_bit(bus, mp_bus_not_pci)) + polarity = default_ISA_polarity(idx); + else + polarity = default_PCI_polarity(idx); + break; + case 1: /* high active */ + { + polarity = 0; + break; + } + case 2: /* reserved */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + polarity = 1; + break; + } + case 3: /* low active */ + { + polarity = 1; + break; + } + default: /* invalid */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + polarity = 1; + break; + } + } + return polarity; +} + +static int MPBIOS_trigger(int idx) +{ + int bus = mp_irqs[idx].mp_srcbus; + int trigger; + + /* + * Determine IRQ trigger mode (edge or level sensitive): + */ + switch ((mp_irqs[idx].mp_irqflag>>2) & 3) + { + case 0: /* conforms, ie. bus-type dependent */ + if (test_bit(bus, mp_bus_not_pci)) + trigger = default_ISA_trigger(idx); + else + trigger = default_PCI_trigger(idx); +#if defined(CONFIG_EISA) || defined(CONFIG_MCA) + switch (mp_bus_id_to_type[bus]) { + case MP_BUS_ISA: /* ISA pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_EISA: /* EISA pin */ + { + trigger = default_EISA_trigger(idx); + break; + } + case MP_BUS_PCI: /* PCI pin */ + { + /* set before the switch */ + break; + } + case MP_BUS_MCA: /* MCA pin */ + { + trigger = default_MCA_trigger(idx); + break; + } + default: + { + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 1; + break; + } + } +#endif + break; + case 1: /* edge */ + { + trigger = 0; + break; + } + case 2: /* reserved */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 1; + break; + } + case 3: /* level */ + { + trigger = 1; + break; + } + default: /* invalid */ + { + printk(KERN_WARNING "broken BIOS!!\n"); + trigger = 0; + break; + } + } + return trigger; +} + +static inline int irq_polarity(int idx) +{ + return MPBIOS_polarity(idx); +} + +static inline int irq_trigger(int idx) +{ + return MPBIOS_trigger(idx); +} + +int (*ioapic_renumber_irq)(int ioapic, int irq); +static int pin_2_irq(int idx, int apic, int pin) +{ + int irq, i; + int bus = mp_irqs[idx].mp_srcbus; + + /* + * Debugging check, we are in big trouble if this message pops up! + */ + if (mp_irqs[idx].mp_dstirq != pin) + printk(KERN_ERR "broken BIOS or MPTABLE parser, ayiee!!\n"); + + if (test_bit(bus, mp_bus_not_pci)) { + irq = mp_irqs[idx].mp_srcbusirq; + } else { + /* + * PCI IRQs are mapped in order + */ + i = irq = 0; + while (i < apic) + irq += nr_ioapic_registers[i++]; + irq += pin; + /* + * For MPS mode, so far only needed by ES7000 platform + */ + if (ioapic_renumber_irq) + irq = ioapic_renumber_irq(apic, irq); + } + +#ifdef CONFIG_X86_32 + /* + * PCI IRQ command line redirection. Yes, limits are hardcoded. + */ + if ((pin >= 16) && (pin <= 23)) { + if (pirq_entries[pin-16] != -1) { + if (!pirq_entries[pin-16]) { + apic_printk(APIC_VERBOSE, KERN_DEBUG + "disabling PIRQ%d\n", pin-16); + } else { + irq = pirq_entries[pin-16]; + apic_printk(APIC_VERBOSE, KERN_DEBUG + "using PIRQ%d -> IRQ %d\n", + pin-16, irq); + } + } + } +#endif + + return irq; +} + +void lock_vector_lock(void) +{ + /* Used to the online set of cpus does not change + * during assign_irq_vector. + */ + spin_lock(&vector_lock); +} + +void unlock_vector_lock(void) +{ + spin_unlock(&vector_lock); +} + +static int __assign_irq_vector(int irq, cpumask_t mask) +{ + /* + * NOTE! The local APIC isn't very good at handling + * multiple interrupts at the same interrupt level. + * As the interrupt level is determined by taking the + * vector number and shifting that right by 4, we + * want to spread these out a bit so that they don't + * all fall in the same interrupt level. + * + * Also, we've got to be careful not to trash gate + * 0x80, because int 0x80 is hm, kind of importantish. ;) + */ + static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; + unsigned int old_vector; + int cpu; + struct irq_cfg *cfg; + + cfg = irq_cfg(irq); + + /* Only try and allocate irqs on cpus that are present */ + cpus_and(mask, mask, cpu_online_map); + + if ((cfg->move_in_progress) || cfg->move_cleanup_count) + return -EBUSY; + + old_vector = cfg->vector; + if (old_vector) { + cpumask_t tmp; + cpus_and(tmp, cfg->domain, mask); + if (!cpus_empty(tmp)) + return 0; + } + + for_each_cpu_mask_nr(cpu, mask) { + cpumask_t domain, new_mask; + int new_cpu; + int vector, offset; + + domain = vector_allocation_domain(cpu); + cpus_and(new_mask, domain, cpu_online_map); + + vector = current_vector; + offset = current_offset; +next: + vector += 8; + if (vector >= first_system_vector) { + /* If we run out of vectors on large boxen, must share them. */ + offset = (offset + 1) % 8; + vector = FIRST_DEVICE_VECTOR + offset; + } + if (unlikely(current_vector == vector)) + continue; +#ifdef CONFIG_X86_64 + if (vector == IA32_SYSCALL_VECTOR) + goto next; +#else + if (vector == SYSCALL_VECTOR) + goto next; +#endif + for_each_cpu_mask_nr(new_cpu, new_mask) + if (per_cpu(vector_irq, new_cpu)[vector] != -1) + goto next; + /* Found one! */ + current_vector = vector; + current_offset = offset; + if (old_vector) { + cfg->move_in_progress = 1; + cfg->old_domain = cfg->domain; + } + for_each_cpu_mask_nr(new_cpu, new_mask) + per_cpu(vector_irq, new_cpu)[vector] = irq; + cfg->vector = vector; + cfg->domain = domain; + return 0; + } + return -ENOSPC; +} + +static int assign_irq_vector(int irq, cpumask_t mask) +{ + int err; + unsigned long flags; + + spin_lock_irqsave(&vector_lock, flags); + err = __assign_irq_vector(irq, mask); + spin_unlock_irqrestore(&vector_lock, flags); + return err; +} + +static void __clear_irq_vector(int irq) +{ + struct irq_cfg *cfg; + cpumask_t mask; + int cpu, vector; + + cfg = irq_cfg(irq); + BUG_ON(!cfg->vector); + + vector = cfg->vector; + cpus_and(mask, cfg->domain, cpu_online_map); + for_each_cpu_mask_nr(cpu, mask) + per_cpu(vector_irq, cpu)[vector] = -1; + + cfg->vector = 0; + cpus_clear(cfg->domain); +} + +void __setup_vector_irq(int cpu) +{ + /* Initialize vector_irq on a new cpu */ + /* This function must be called with vector_lock held */ + int irq, vector; + struct irq_cfg *cfg; + + /* Mark the inuse vectors */ + for_each_irq_cfg(cfg) { + if (!cpu_isset(cpu, cfg->domain)) + continue; + vector = cfg->vector; + irq = cfg->irq; + per_cpu(vector_irq, cpu)[vector] = irq; + } + /* Mark the free vectors */ + for (vector = 0; vector < NR_VECTORS; ++vector) { + irq = per_cpu(vector_irq, cpu)[vector]; + if (irq < 0) + continue; + + cfg = irq_cfg(irq); + if (!cpu_isset(cpu, cfg->domain)) + per_cpu(vector_irq, cpu)[vector] = -1; + } +} + +static struct irq_chip ioapic_chip; +#ifdef CONFIG_INTR_REMAP +static struct irq_chip ir_ioapic_chip; +#endif + +#define IOAPIC_AUTO -1 +#define IOAPIC_EDGE 0 +#define IOAPIC_LEVEL 1 + +#ifdef CONFIG_X86_32 +static inline int IO_APIC_irq_trigger(int irq) +{ + int apic, idx, pin; + + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + idx = find_irq_entry(apic, pin, mp_INT); + if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) + return irq_trigger(idx); + } + } + /* + * nonexistent IRQs are edge default + */ + return 0; +} +#else +static inline int IO_APIC_irq_trigger(int irq) +{ + return 1; +} +#endif + +static void ioapic_register_intr(int irq, unsigned long trigger) +{ + struct irq_desc *desc; + + /* first time to use this irq_desc */ + if (irq < 16) + desc = irq_to_desc(irq); + else + desc = irq_to_desc_alloc(irq); + + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || + trigger == IOAPIC_LEVEL) + desc->status |= IRQ_LEVEL; + else + desc->status &= ~IRQ_LEVEL; + +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + desc->status |= IRQ_MOVE_PCNTXT; + if (trigger) + set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, + handle_fasteoi_irq, + "fasteoi"); + else + set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, + handle_edge_irq, "edge"); + return; + } +#endif + if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || + trigger == IOAPIC_LEVEL) + set_irq_chip_and_handler_name(irq, &ioapic_chip, + handle_fasteoi_irq, + "fasteoi"); + else + set_irq_chip_and_handler_name(irq, &ioapic_chip, + handle_edge_irq, "edge"); +} + +static int setup_ioapic_entry(int apic, int irq, + struct IO_APIC_route_entry *entry, + unsigned int destination, int trigger, + int polarity, int vector) +{ + /* + * add it to the IO-APIC irq-routing table: + */ + memset(entry,0,sizeof(*entry)); + +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) { + struct intel_iommu *iommu = map_ioapic_to_ir(apic); + struct irte irte; + struct IR_IO_APIC_route_entry *ir_entry = + (struct IR_IO_APIC_route_entry *) entry; + int index; + + if (!iommu) + panic("No mapping iommu for ioapic %d\n", apic); + + index = alloc_irte(iommu, irq, 1); + if (index < 0) + panic("Failed to allocate IRTE for ioapic %d\n", apic); + + memset(&irte, 0, sizeof(irte)); + + irte.present = 1; + irte.dst_mode = INT_DEST_MODE; + irte.trigger_mode = trigger; + irte.dlvry_mode = INT_DELIVERY_MODE; + irte.vector = vector; + irte.dest_id = IRTE_DEST(destination); + + modify_irte(irq, &irte); + + ir_entry->index2 = (index >> 15) & 0x1; + ir_entry->zero = 0; + ir_entry->format = 1; + ir_entry->index = (index & 0x7fff); + } else +#endif + { + entry->delivery_mode = INT_DELIVERY_MODE; + entry->dest_mode = INT_DEST_MODE; + entry->dest = destination; + } + + entry->mask = 0; /* enable IRQ */ + entry->trigger = trigger; + entry->polarity = polarity; + entry->vector = vector; + + /* Mask level triggered irqs. + * Use IRQ_DELAYED_DISABLE for edge triggered irqs. + */ + if (trigger) + entry->mask = 1; + return 0; +} + +static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, + int trigger, int polarity) +{ + struct irq_cfg *cfg; + struct IO_APIC_route_entry entry; + cpumask_t mask; + + if (!IO_APIC_IRQ(irq)) + return; + + cfg = irq_cfg(irq); + + mask = TARGET_CPUS; + if (assign_irq_vector(irq, mask)) + return; + + cpus_and(mask, cfg->domain, mask); + + apic_printk(APIC_VERBOSE,KERN_DEBUG + "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " + "IRQ %d Mode:%i Active:%i)\n", + apic, mp_ioapics[apic].mp_apicid, pin, cfg->vector, + irq, trigger, polarity); + + + if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, + cpu_mask_to_apicid(mask), trigger, polarity, + cfg->vector)) { + printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", + mp_ioapics[apic].mp_apicid, pin); + __clear_irq_vector(irq); + return; + } + + ioapic_register_intr(irq, trigger); + if (irq < 16) + disable_8259A_irq(irq); + + ioapic_write_entry(apic, pin, entry); +} + +static void __init setup_IO_APIC_irqs(void) +{ + int apic, pin, idx, irq, first_notcon = 1; + + apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); + + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + + idx = find_irq_entry(apic,pin,mp_INT); + if (idx == -1) { + if (first_notcon) { + apic_printk(APIC_VERBOSE, KERN_DEBUG " IO-APIC (apicid-pin) %d-%d", mp_ioapics[apic].mp_apicid, pin); + first_notcon = 0; + } else + apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mp_apicid, pin); + continue; + } + if (!first_notcon) { + apic_printk(APIC_VERBOSE, " not connected.\n"); + first_notcon = 1; + } + + irq = pin_2_irq(idx, apic, pin); +#ifdef CONFIG_X86_32 + if (multi_timer_check(apic, irq)) + continue; +#endif + add_pin_to_irq(irq, apic, pin); + + setup_IO_APIC_irq(apic, pin, irq, + irq_trigger(idx), irq_polarity(idx)); + } + } + + if (!first_notcon) + apic_printk(APIC_VERBOSE, " not connected.\n"); +} + +/* + * Set up the timer pin, possibly with the 8259A-master behind. + */ +static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, + int vector) +{ + struct IO_APIC_route_entry entry; + +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + return; +#endif + + memset(&entry, 0, sizeof(entry)); + + /* + * We use logical delivery to get the timer IRQ + * to the first CPU. + */ + entry.dest_mode = INT_DEST_MODE; + entry.mask = 1; /* mask IRQ now */ + entry.dest = cpu_mask_to_apicid(TARGET_CPUS); + entry.delivery_mode = INT_DELIVERY_MODE; + entry.polarity = 0; + entry.trigger = 0; + entry.vector = vector; + + /* + * The timer IRQ doesn't have to know that behind the + * scene we may have a 8259A-master in AEOI mode ... + */ + set_irq_chip_and_handler_name(0, &ioapic_chip, handle_edge_irq, "edge"); + + /* + * Add it to the IO-APIC irq-routing table: + */ + ioapic_write_entry(apic, pin, entry); +} + + +__apicdebuginit(void) print_IO_APIC(void) +{ + int apic, i; + union IO_APIC_reg_00 reg_00; + union IO_APIC_reg_01 reg_01; + union IO_APIC_reg_02 reg_02; + union IO_APIC_reg_03 reg_03; + unsigned long flags; + struct irq_cfg *cfg; + + if (apic_verbosity == APIC_QUIET) + return; + + printk(KERN_DEBUG "number of MP IRQ sources: %d.\n", mp_irq_entries); + for (i = 0; i < nr_ioapics; i++) + printk(KERN_DEBUG "number of IO-APIC #%d registers: %d.\n", + mp_ioapics[i].mp_apicid, nr_ioapic_registers[i]); + + /* + * We are a bit conservative about what we expect. We have to + * know about every hardware change ASAP. + */ + printk(KERN_INFO "testing the IO APIC.......................\n"); + + for (apic = 0; apic < nr_ioapics; apic++) { + + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(apic, 0); + reg_01.raw = io_apic_read(apic, 1); + if (reg_01.bits.version >= 0x10) + reg_02.raw = io_apic_read(apic, 2); + if (reg_01.bits.version >= 0x20) + reg_03.raw = io_apic_read(apic, 3); + spin_unlock_irqrestore(&ioapic_lock, flags); + + printk("\n"); + printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); + printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); + printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); + printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); + printk(KERN_DEBUG "....... : LTS : %X\n", reg_00.bits.LTS); + + printk(KERN_DEBUG ".... register #01: %08X\n", *(int *)®_01); + printk(KERN_DEBUG "....... : max redirection entries: %04X\n", reg_01.bits.entries); + + printk(KERN_DEBUG "....... : PRQ implemented: %X\n", reg_01.bits.PRQ); + printk(KERN_DEBUG "....... : IO APIC version: %04X\n", reg_01.bits.version); + + /* + * Some Intel chipsets with IO APIC VERSION of 0x1? don't have reg_02, + * but the value of reg_02 is read as the previous read register + * value, so ignore it if reg_02 == reg_01. + */ + if (reg_01.bits.version >= 0x10 && reg_02.raw != reg_01.raw) { + printk(KERN_DEBUG ".... register #02: %08X\n", reg_02.raw); + printk(KERN_DEBUG "....... : arbitration: %02X\n", reg_02.bits.arbitration); + } + + /* + * Some Intel chipsets with IO APIC VERSION of 0x2? don't have reg_02 + * or reg_03, but the value of reg_0[23] is read as the previous read + * register value, so ignore it if reg_03 == reg_0[12]. + */ + if (reg_01.bits.version >= 0x20 && reg_03.raw != reg_02.raw && + reg_03.raw != reg_01.raw) { + printk(KERN_DEBUG ".... register #03: %08X\n", reg_03.raw); + printk(KERN_DEBUG "....... : Boot DT : %X\n", reg_03.bits.boot_DT); + } + + printk(KERN_DEBUG ".... IRQ redirection table:\n"); + + printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" + " Stat Dmod Deli Vect: \n"); + + for (i = 0; i <= reg_01.bits.entries; i++) { + struct IO_APIC_route_entry entry; + + entry = ioapic_read_entry(apic, i); + + printk(KERN_DEBUG " %02x %03X ", + i, + entry.dest + ); + + printk("%1d %1d %1d %1d %1d %1d %1d %02X\n", + entry.mask, + entry.trigger, + entry.irr, + entry.polarity, + entry.delivery_status, + entry.dest_mode, + entry.delivery_mode, + entry.vector + ); + } + } + printk(KERN_DEBUG "IRQ to pin mappings:\n"); + for_each_irq_cfg(cfg) { + struct irq_pin_list *entry = cfg->irq_2_pin; + if (!entry) + continue; + printk(KERN_DEBUG "IRQ%d ", cfg->irq); + for (;;) { + printk("-> %d:%d", entry->apic, entry->pin); + if (!entry->next) + break; + entry = entry->next; + } + printk("\n"); + } + + printk(KERN_INFO ".................................... done.\n"); + + return; +} + +__apicdebuginit(void) print_APIC_bitfield(int base) +{ + unsigned int v; + int i, j; + + if (apic_verbosity == APIC_QUIET) + return; + + printk(KERN_DEBUG "0123456789abcdef0123456789abcdef\n" KERN_DEBUG); + for (i = 0; i < 8; i++) { + v = apic_read(base + i*0x10); + for (j = 0; j < 32; j++) { + if (v & (1< 3) /* Due to the Pentium erratum 3AP. */ + apic_write(APIC_ESR, 0); + + v = apic_read(APIC_ESR); + printk(KERN_DEBUG "... APIC ESR: %08x\n", v); + } + + icr = apic_icr_read(); + printk(KERN_DEBUG "... APIC ICR: %08x\n", icr); + printk(KERN_DEBUG "... APIC ICR2: %08x\n", icr >> 32); + + v = apic_read(APIC_LVTT); + printk(KERN_DEBUG "... APIC LVTT: %08x\n", v); + + if (maxlvt > 3) { /* PC is LVT#4. */ + v = apic_read(APIC_LVTPC); + printk(KERN_DEBUG "... APIC LVTPC: %08x\n", v); + } + v = apic_read(APIC_LVT0); + printk(KERN_DEBUG "... APIC LVT0: %08x\n", v); + v = apic_read(APIC_LVT1); + printk(KERN_DEBUG "... APIC LVT1: %08x\n", v); + + if (maxlvt > 2) { /* ERR is LVT#3. */ + v = apic_read(APIC_LVTERR); + printk(KERN_DEBUG "... APIC LVTERR: %08x\n", v); + } + + v = apic_read(APIC_TMICT); + printk(KERN_DEBUG "... APIC TMICT: %08x\n", v); + v = apic_read(APIC_TMCCT); + printk(KERN_DEBUG "... APIC TMCCT: %08x\n", v); + v = apic_read(APIC_TDCR); + printk(KERN_DEBUG "... APIC TDCR: %08x\n", v); + printk("\n"); +} + +__apicdebuginit(void) print_all_local_APICs(void) +{ + on_each_cpu(print_local_APIC, NULL, 1); +} + +__apicdebuginit(void) print_PIC(void) +{ + unsigned int v; + unsigned long flags; + + if (apic_verbosity == APIC_QUIET) + return; + + printk(KERN_DEBUG "\nprinting PIC contents\n"); + + spin_lock_irqsave(&i8259A_lock, flags); + + v = inb(0xa1) << 8 | inb(0x21); + printk(KERN_DEBUG "... PIC IMR: %04x\n", v); + + v = inb(0xa0) << 8 | inb(0x20); + printk(KERN_DEBUG "... PIC IRR: %04x\n", v); + + outb(0x0b,0xa0); + outb(0x0b,0x20); + v = inb(0xa0) << 8 | inb(0x20); + outb(0x0a,0xa0); + outb(0x0a,0x20); + + spin_unlock_irqrestore(&i8259A_lock, flags); + + printk(KERN_DEBUG "... PIC ISR: %04x\n", v); + + v = inb(0x4d1) << 8 | inb(0x4d0); + printk(KERN_DEBUG "... PIC ELCR: %04x\n", v); +} + +__apicdebuginit(int) print_all_ICs(void) +{ + print_PIC(); + print_all_local_APICs(); + print_IO_APIC(); + + return 0; +} + +fs_initcall(print_all_ICs); + + +/* Where if anywhere is the i8259 connect in external int mode */ +static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; + +void __init enable_IO_APIC(void) +{ + union IO_APIC_reg_01 reg_01; + int i8259_apic, i8259_pin; + int apic; + unsigned long flags; + +#ifdef CONFIG_X86_32 + int i; + if (!pirqs_enabled) + for (i = 0; i < MAX_PIRQS; i++) + pirq_entries[i] = -1; +#endif + + /* + * The number of IO-APIC IRQ registers (== #pins): + */ + for (apic = 0; apic < nr_ioapics; apic++) { + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(apic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + nr_ioapic_registers[apic] = reg_01.bits.entries+1; + } + for(apic = 0; apic < nr_ioapics; apic++) { + int pin; + /* See if any of the pins is in ExtINT mode */ + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + struct IO_APIC_route_entry entry; + entry = ioapic_read_entry(apic, pin); + + /* If the interrupt line is enabled and in ExtInt mode + * I have found the pin where the i8259 is connected. + */ + if ((entry.mask == 0) && (entry.delivery_mode == dest_ExtINT)) { + ioapic_i8259.apic = apic; + ioapic_i8259.pin = pin; + goto found_i8259; + } + } + } + found_i8259: + /* Look to see what if the MP table has reported the ExtINT */ + /* If we could not find the appropriate pin by looking at the ioapic + * the i8259 probably is not connected the ioapic but give the + * mptable a chance anyway. + */ + i8259_pin = find_isa_irq_pin(0, mp_ExtINT); + i8259_apic = find_isa_irq_apic(0, mp_ExtINT); + /* Trust the MP table if nothing is setup in the hardware */ + if ((ioapic_i8259.pin == -1) && (i8259_pin >= 0)) { + printk(KERN_WARNING "ExtINT not setup in hardware but reported by MP table\n"); + ioapic_i8259.pin = i8259_pin; + ioapic_i8259.apic = i8259_apic; + } + /* Complain if the MP table and the hardware disagree */ + if (((ioapic_i8259.apic != i8259_apic) || (ioapic_i8259.pin != i8259_pin)) && + (i8259_pin >= 0) && (ioapic_i8259.pin >= 0)) + { + printk(KERN_WARNING "ExtINT in hardware and MP table differ\n"); + } + + /* + * Do not trust the IO-APIC being empty at bootup + */ + clear_IO_APIC(); +} + +/* + * Not an __init, needed by the reboot code + */ +void disable_IO_APIC(void) +{ + /* + * Clear the IO-APIC before rebooting: + */ + clear_IO_APIC(); + + /* + * If the i8259 is routed through an IOAPIC + * Put that IOAPIC in virtual wire mode + * so legacy interrupts can be delivered. + */ + if (ioapic_i8259.pin != -1) { + struct IO_APIC_route_entry entry; + + memset(&entry, 0, sizeof(entry)); + entry.mask = 0; /* Enabled */ + entry.trigger = 0; /* Edge */ + entry.irr = 0; + entry.polarity = 0; /* High */ + entry.delivery_status = 0; + entry.dest_mode = 0; /* Physical */ + entry.delivery_mode = dest_ExtINT; /* ExtInt */ + entry.vector = 0; + entry.dest = read_apic_id(); + + /* + * Add it to the IO-APIC irq-routing table: + */ + ioapic_write_entry(ioapic_i8259.apic, ioapic_i8259.pin, entry); + } + + disconnect_bsp_APIC(ioapic_i8259.pin != -1); +} + +#ifdef CONFIG_X86_32 +/* + * function to set the IO-APIC physical IDs based on the + * values stored in the MPC table. + * + * by Matt Domsch Tue Dec 21 12:25:05 CST 1999 + */ + +static void __init setup_ioapic_ids_from_mpc(void) +{ + union IO_APIC_reg_00 reg_00; + physid_mask_t phys_id_present_map; + int apic; + int i; + unsigned char old_id; + unsigned long flags; + + if (x86_quirks->setup_ioapic_ids && x86_quirks->setup_ioapic_ids()) + return; + + /* + * Don't check I/O APIC IDs for xAPIC systems. They have + * no meaning without the serial APIC bus. + */ + if (!(boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) + || APIC_XAPIC(apic_version[boot_cpu_physical_apicid])) + return; + /* + * This is broken; anything with a real cpu count has to + * circumvent this idiocy regardless. + */ + phys_id_present_map = ioapic_phys_id_map(phys_cpu_present_map); + + /* + * Set the IOAPIC ID to the value stored in the MPC table. + */ + for (apic = 0; apic < nr_ioapics; apic++) { + + /* Read the register 0 value */ + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(apic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + old_id = mp_ioapics[apic].mp_apicid; + + if (mp_ioapics[apic].mp_apicid >= get_physical_broadcast()) { + printk(KERN_ERR "BIOS bug, IO-APIC#%d ID is %d in the MPC table!...\n", + apic, mp_ioapics[apic].mp_apicid); + printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", + reg_00.bits.ID); + mp_ioapics[apic].mp_apicid = reg_00.bits.ID; + } + + /* + * Sanity check, is the ID really free? Every APIC in a + * system must have a unique ID or we get lots of nice + * 'stuck on smp_invalidate_needed IPI wait' messages. + */ + if (check_apicid_used(phys_id_present_map, + mp_ioapics[apic].mp_apicid)) { + printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n", + apic, mp_ioapics[apic].mp_apicid); + for (i = 0; i < get_physical_broadcast(); i++) + if (!physid_isset(i, phys_id_present_map)) + break; + if (i >= get_physical_broadcast()) + panic("Max APIC ID exceeded!\n"); + printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", + i); + physid_set(i, phys_id_present_map); + mp_ioapics[apic].mp_apicid = i; + } else { + physid_mask_t tmp; + tmp = apicid_to_cpu_present(mp_ioapics[apic].mp_apicid); + apic_printk(APIC_VERBOSE, "Setting %d in the " + "phys_id_present_map\n", + mp_ioapics[apic].mp_apicid); + physids_or(phys_id_present_map, phys_id_present_map, tmp); + } + + + /* + * We need to adjust the IRQ routing table + * if the ID changed. + */ + if (old_id != mp_ioapics[apic].mp_apicid) + for (i = 0; i < mp_irq_entries; i++) + if (mp_irqs[i].mp_dstapic == old_id) + mp_irqs[i].mp_dstapic + = mp_ioapics[apic].mp_apicid; + + /* + * Read the right value from the MPC table and + * write it into the ID register. + */ + apic_printk(APIC_VERBOSE, KERN_INFO + "...changing IO-APIC physical APIC ID to %d ...", + mp_ioapics[apic].mp_apicid); + + reg_00.bits.ID = mp_ioapics[apic].mp_apicid; + spin_lock_irqsave(&ioapic_lock, flags); + + /* + * Sanity check + */ + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(apic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + if (reg_00.bits.ID != mp_ioapics[apic].mp_apicid) + printk("could not set ID!\n"); + else + apic_printk(APIC_VERBOSE, " ok.\n"); + } +} +#endif + +int no_timer_check __initdata; + +static int __init notimercheck(char *s) +{ + no_timer_check = 1; + return 1; +} +__setup("no_timer_check", notimercheck); + +/* + * There is a nasty bug in some older SMP boards, their mptable lies + * about the timer IRQ. We do the following to work around the situation: + * + * - timer IRQ defaults to IO-APIC IRQ + * - if this function detects that timer IRQs are defunct, then we fall + * back to ISA timer IRQs + */ +static int __init timer_irq_works(void) +{ + unsigned long t1 = jiffies; + unsigned long flags; + + if (no_timer_check) + return 1; + + local_save_flags(flags); + local_irq_enable(); + /* Let ten ticks pass... */ + mdelay((10 * 1000) / HZ); + local_irq_restore(flags); + + /* + * Expect a few ticks at least, to be sure some possible + * glue logic does not lock up after one or two first + * ticks in a non-ExtINT mode. Also the local APIC + * might have cached one ExtINT interrupt. Finally, at + * least one tick may be lost due to delays. + */ + + /* jiffies wrap? */ + if (time_after(jiffies, t1 + 4)) + return 1; + return 0; +} + +/* + * In the SMP+IOAPIC case it might happen that there are an unspecified + * number of pending IRQ events unhandled. These cases are very rare, + * so we 'resend' these IRQs via IPIs, to the same CPU. It's much + * better to do it this way as thus we do not have to be aware of + * 'pending' interrupts in the IRQ path, except at this point. + */ +/* + * Edge triggered needs to resend any interrupt + * that was delayed but this is now handled in the device + * independent code. + */ + +/* + * Starting up a edge-triggered IO-APIC interrupt is + * nasty - we need to make sure that we get the edge. + * If it is already asserted for some reason, we need + * return 1 to indicate that is was pending. + * + * This is not complete - we should be able to fake + * an edge even if it isn't on the 8259A... + */ + +static unsigned int startup_ioapic_irq(unsigned int irq) +{ + int was_pending = 0; + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + if (irq < 16) { + disable_8259A_irq(irq); + if (i8259A_irq_pending(irq)) + was_pending = 1; + } + __unmask_IO_APIC_irq(irq); + spin_unlock_irqrestore(&ioapic_lock, flags); + + return was_pending; +} + +#ifdef CONFIG_X86_64 +static int ioapic_retrigger_irq(unsigned int irq) +{ + + struct irq_cfg *cfg = irq_cfg(irq); + unsigned long flags; + + spin_lock_irqsave(&vector_lock, flags); + send_IPI_mask(cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); + spin_unlock_irqrestore(&vector_lock, flags); + + return 1; +} +#else +static int ioapic_retrigger_irq(unsigned int irq) +{ + send_IPI_self(irq_cfg(irq)->vector); + + return 1; +} +#endif + +/* + * Level and edge triggered IO-APIC interrupts need different handling, + * so we use two separate IRQ descriptors. Edge triggered IRQs can be + * handled with the level-triggered descriptor, but that one has slightly + * more overhead. Level-triggered interrupts cannot be handled with the + * edge-triggered handler, without risking IRQ storms and other ugly + * races. + */ + +#ifdef CONFIG_SMP + +#ifdef CONFIG_INTR_REMAP +static void ir_irq_migration(struct work_struct *work); + +static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); + +/* + * Migrate the IO-APIC irq in the presence of intr-remapping. + * + * For edge triggered, irq migration is a simple atomic update(of vector + * and cpu destination) of IRTE and flush the hardware cache. + * + * For level triggered, we need to modify the io-apic RTE aswell with the update + * vector information, along with modifying IRTE with vector and destination. + * So irq migration for level triggered is little bit more complex compared to + * edge triggered migration. But the good news is, we use the same algorithm + * for level triggered migration as we have today, only difference being, + * we now initiate the irq migration from process context instead of the + * interrupt context. + * + * In future, when we do a directed EOI (combined with cpu EOI broadcast + * suppression) to the IO-APIC, level triggered irq migration will also be + * as simple as edge triggered migration and we can do the irq migration + * with a simple atomic update to IO-APIC RTE. + */ +static void migrate_ioapic_irq(int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + struct irq_desc *desc; + cpumask_t tmp, cleanup_mask; + struct irte irte; + int modify_ioapic_rte; + unsigned int dest; + unsigned long flags; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (get_irte(irq, &irte)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + desc = irq_to_desc(irq); + modify_ioapic_rte = desc->status & IRQ_LEVEL; + if (modify_ioapic_rte) { + spin_lock_irqsave(&ioapic_lock, flags); + __target_IO_APIC_irq(irq, dest, cfg->vector); + spin_unlock_irqrestore(&ioapic_lock, flags); + } + + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + /* + * Modified the IRTE and flushes the Interrupt entry cache. + */ + modify_irte(irq, &irte); + + if (cfg->move_in_progress) { + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } + + desc->affinity = mask; +} + +static int migrate_irq_remapped_level(int irq) +{ + int ret = -1; + struct irq_desc *desc = irq_to_desc(irq); + + mask_IO_APIC_irq(irq); + + if (io_apic_level_ack_pending(irq)) { + /* + * Interrupt in progress. Migrating irq now will change the + * vector information in the IO-APIC RTE and that will confuse + * the EOI broadcast performed by cpu. + * So, delay the irq migration to the next instance. + */ + schedule_delayed_work(&ir_migration_work, 1); + goto unmask; + } + + /* everthing is clear. we have right of way */ + migrate_ioapic_irq(irq, desc->pending_mask); + + ret = 0; + desc->status &= ~IRQ_MOVE_PENDING; + cpus_clear(desc->pending_mask); + +unmask: + unmask_IO_APIC_irq(irq); + return ret; +} + +static void ir_irq_migration(struct work_struct *work) +{ + unsigned int irq; + struct irq_desc *desc; + + for_each_irq_desc(irq, desc) { + if (desc->status & IRQ_MOVE_PENDING) { + unsigned long flags; + + spin_lock_irqsave(&desc->lock, flags); + if (!desc->chip->set_affinity || + !(desc->status & IRQ_MOVE_PENDING)) { + desc->status &= ~IRQ_MOVE_PENDING; + spin_unlock_irqrestore(&desc->lock, flags); + continue; + } + + desc->chip->set_affinity(irq, desc->pending_mask); + spin_unlock_irqrestore(&desc->lock, flags); + } + } +} + +/* + * Migrates the IRQ destination in the process context. + */ +static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc = irq_to_desc(irq); + + if (desc->status & IRQ_LEVEL) { + desc->status |= IRQ_MOVE_PENDING; + desc->pending_mask = mask; + migrate_irq_remapped_level(irq); + return; + } + + migrate_ioapic_irq(irq, mask); +} +#endif + +asmlinkage void smp_irq_move_cleanup_interrupt(void) +{ + unsigned vector, me; + ack_APIC_irq(); +#ifdef CONFIG_X86_64 + exit_idle(); +#endif + irq_enter(); + + me = smp_processor_id(); + for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { + unsigned int irq; + struct irq_desc *desc; + struct irq_cfg *cfg; + irq = __get_cpu_var(vector_irq)[vector]; + + desc = irq_to_desc(irq); + if (!desc) + continue; + + cfg = irq_cfg(irq); + spin_lock(&desc->lock); + if (!cfg->move_cleanup_count) + goto unlock; + + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) + goto unlock; + + __get_cpu_var(vector_irq)[vector] = -1; + cfg->move_cleanup_count--; +unlock: + spin_unlock(&desc->lock); + } + + irq_exit(); +} + +static void irq_complete_move(unsigned int irq) +{ + struct irq_cfg *cfg = irq_cfg(irq); + unsigned vector, me; + + if (likely(!cfg->move_in_progress)) + return; + + vector = ~get_irq_regs()->orig_ax; + me = smp_processor_id(); + if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { + cpumask_t cleanup_mask; + + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } +} +#else +static inline void irq_complete_move(unsigned int irq) {} +#endif +#ifdef CONFIG_INTR_REMAP +static void ack_x2apic_level(unsigned int irq) +{ + ack_x2APIC_irq(); +} + +static void ack_x2apic_edge(unsigned int irq) +{ + ack_x2APIC_irq(); +} +#endif + +static void ack_apic_edge(unsigned int irq) +{ + irq_complete_move(irq); + move_native_irq(irq); + ack_APIC_irq(); +} + +#ifdef CONFIG_X86_64 +static void ack_apic_level(unsigned int irq) +{ + int do_unmask_irq = 0; + + irq_complete_move(irq); +#ifdef CONFIG_GENERIC_PENDING_IRQ + /* If we are moving the irq we need to mask it */ + if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { + do_unmask_irq = 1; + mask_IO_APIC_irq(irq); + } +#endif + + /* + * We must acknowledge the irq before we move it or the acknowledge will + * not propagate properly. + */ + ack_APIC_irq(); + + /* Now we can move and renable the irq */ + if (unlikely(do_unmask_irq)) { + /* Only migrate the irq if the ack has been received. + * + * On rare occasions the broadcast level triggered ack gets + * delayed going to ioapics, and if we reprogram the + * vector while Remote IRR is still set the irq will never + * fire again. + * + * To prevent this scenario we read the Remote IRR bit + * of the ioapic. This has two effects. + * - On any sane system the read of the ioapic will + * flush writes (and acks) going to the ioapic from + * this cpu. + * - We get to see if the ACK has actually been delivered. + * + * Based on failed experiments of reprogramming the + * ioapic entry from outside of irq context starting + * with masking the ioapic entry and then polling until + * Remote IRR was clear before reprogramming the + * ioapic I don't trust the Remote IRR bit to be + * completey accurate. + * + * However there appears to be no other way to plug + * this race, so if the Remote IRR bit is not + * accurate and is causing problems then it is a hardware bug + * and you can go talk to the chipset vendor about it. + */ + if (!io_apic_level_ack_pending(irq)) + move_masked_irq(irq); + unmask_IO_APIC_irq(irq); + } +} +#else +atomic_t irq_mis_count; +static void ack_apic_level(unsigned int irq) +{ + unsigned long v; + int i; + + irq_complete_move(irq); + move_native_irq(irq); + /* + * It appears there is an erratum which affects at least version 0x11 + * of I/O APIC (that's the 82093AA and cores integrated into various + * chipsets). Under certain conditions a level-triggered interrupt is + * erroneously delivered as edge-triggered one but the respective IRR + * bit gets set nevertheless. As a result the I/O unit expects an EOI + * message but it will never arrive and further interrupts are blocked + * from the source. The exact reason is so far unknown, but the + * phenomenon was observed when two consecutive interrupt requests + * from a given source get delivered to the same CPU and the source is + * temporarily disabled in between. + * + * A workaround is to simulate an EOI message manually. We achieve it + * by setting the trigger mode to edge and then to level when the edge + * trigger mode gets detected in the TMR of a local APIC for a + * level-triggered interrupt. We mask the source for the time of the + * operation to prevent an edge-triggered interrupt escaping meanwhile. + * The idea is from Manfred Spraul. --macro + */ + i = irq_cfg(irq)->vector; + + v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); + + ack_APIC_irq(); + + if (!(v & (1 << (i & 0x1f)))) { + atomic_inc(&irq_mis_count); + spin_lock(&ioapic_lock); + __mask_and_edge_IO_APIC_irq(irq); + __unmask_and_level_IO_APIC_irq(irq); + spin_unlock(&ioapic_lock); + } +} +#endif + +static struct irq_chip ioapic_chip __read_mostly = { + .name = "IO-APIC", + .startup = startup_ioapic_irq, + .mask = mask_IO_APIC_irq, + .unmask = unmask_IO_APIC_irq, + .ack = ack_apic_edge, + .eoi = ack_apic_level, +#ifdef CONFIG_SMP + .set_affinity = set_ioapic_affinity_irq, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +#ifdef CONFIG_INTR_REMAP +static struct irq_chip ir_ioapic_chip __read_mostly = { + .name = "IR-IO-APIC", + .startup = startup_ioapic_irq, + .mask = mask_IO_APIC_irq, + .unmask = unmask_IO_APIC_irq, + .ack = ack_x2apic_edge, + .eoi = ack_x2apic_level, +#ifdef CONFIG_SMP + .set_affinity = set_ir_ioapic_affinity_irq, +#endif + .retrigger = ioapic_retrigger_irq, +}; +#endif + +static inline void init_IO_APIC_traps(void) +{ + int irq; + struct irq_desc *desc; + struct irq_cfg *cfg; + + /* + * NOTE! The local APIC isn't very good at handling + * multiple interrupts at the same interrupt level. + * As the interrupt level is determined by taking the + * vector number and shifting that right by 4, we + * want to spread these out a bit so that they don't + * all fall in the same interrupt level. + * + * Also, we've got to be careful not to trash gate + * 0x80, because int 0x80 is hm, kind of importantish. ;) + */ + for_each_irq_cfg(cfg) { + irq = cfg->irq; + if (IO_APIC_IRQ(irq) && !cfg->vector) { + /* + * Hmm.. We don't have an entry for this, + * so default to an old-fashioned 8259 + * interrupt if we can.. + */ + if (irq < 16) + make_8259A_irq(irq); + else { + desc = irq_to_desc(irq); + /* Strange. Oh, well.. */ + desc->chip = &no_irq_chip; + } + } + } +} + +/* + * The local APIC irq-chip implementation: + */ + +static void mask_lapic_irq(unsigned int irq) +{ + unsigned long v; + + v = apic_read(APIC_LVT0); + apic_write(APIC_LVT0, v | APIC_LVT_MASKED); +} + +static void unmask_lapic_irq(unsigned int irq) +{ + unsigned long v; + + v = apic_read(APIC_LVT0); + apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); +} + +static void ack_lapic_irq (unsigned int irq) +{ + ack_APIC_irq(); +} + +static struct irq_chip lapic_chip __read_mostly = { + .name = "local-APIC", + .mask = mask_lapic_irq, + .unmask = unmask_lapic_irq, + .ack = ack_lapic_irq, +}; + +static void lapic_register_intr(int irq) +{ + struct irq_desc *desc; + + desc = irq_to_desc(irq); + desc->status &= ~IRQ_LEVEL; + set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, + "edge"); +} + +static void __init setup_nmi(void) +{ + /* + * Dirty trick to enable the NMI watchdog ... + * We put the 8259A master into AEOI mode and + * unmask on all local APICs LVT0 as NMI. + * + * The idea to use the 8259A in AEOI mode ('8259A Virtual Wire') + * is from Maciej W. Rozycki - so we do not have to EOI from + * the NMI handler or the timer interrupt. + */ + apic_printk(APIC_VERBOSE, KERN_INFO "activating NMI Watchdog ..."); + + enable_NMI_through_LVT0(); + + apic_printk(APIC_VERBOSE, " done.\n"); +} + +/* + * This looks a bit hackish but it's about the only one way of sending + * a few INTA cycles to 8259As and any associated glue logic. ICR does + * not support the ExtINT mode, unfortunately. We need to send these + * cycles as some i82489DX-based boards have glue logic that keeps the + * 8259A interrupt line asserted until INTA. --macro + */ +static inline void __init unlock_ExtINT_logic(void) +{ + int apic, pin, i; + struct IO_APIC_route_entry entry0, entry1; + unsigned char save_control, save_freq_select; + + pin = find_isa_irq_pin(8, mp_INT); + if (pin == -1) { + WARN_ON_ONCE(1); + return; + } + apic = find_isa_irq_apic(8, mp_INT); + if (apic == -1) { + WARN_ON_ONCE(1); + return; + } + + entry0 = ioapic_read_entry(apic, pin); + clear_IO_APIC_pin(apic, pin); + + memset(&entry1, 0, sizeof(entry1)); + + entry1.dest_mode = 0; /* physical delivery */ + entry1.mask = 0; /* unmask IRQ now */ + entry1.dest = hard_smp_processor_id(); + entry1.delivery_mode = dest_ExtINT; + entry1.polarity = entry0.polarity; + entry1.trigger = 0; + entry1.vector = 0; + + ioapic_write_entry(apic, pin, entry1); + + save_control = CMOS_READ(RTC_CONTROL); + save_freq_select = CMOS_READ(RTC_FREQ_SELECT); + CMOS_WRITE((save_freq_select & ~RTC_RATE_SELECT) | 0x6, + RTC_FREQ_SELECT); + CMOS_WRITE(save_control | RTC_PIE, RTC_CONTROL); + + i = 100; + while (i-- > 0) { + mdelay(10); + if ((CMOS_READ(RTC_INTR_FLAGS) & RTC_PF) == RTC_PF) + i -= 10; + } + + CMOS_WRITE(save_control, RTC_CONTROL); + CMOS_WRITE(save_freq_select, RTC_FREQ_SELECT); + clear_IO_APIC_pin(apic, pin); + + ioapic_write_entry(apic, pin, entry0); +} + +static int disable_timer_pin_1 __initdata; +/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ +static int __init disable_timer_pin_setup(char *arg) +{ + disable_timer_pin_1 = 1; + return 0; +} +early_param("disable_timer_pin_1", disable_timer_pin_setup); + +int timer_through_8259 __initdata; + +/* + * This code may look a bit paranoid, but it's supposed to cooperate with + * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ + * is so screwy. Thanks to Brian Perkins for testing/hacking this beast + * fanatically on his truly buggy board. + * + * FIXME: really need to revamp this for all platforms. + */ +static inline void __init check_timer(void) +{ + struct irq_cfg *cfg = irq_cfg(0); + int apic1, pin1, apic2, pin2; + unsigned long flags; + unsigned int ver; + int no_pin1 = 0; + + local_irq_save(flags); + + ver = apic_read(APIC_LVR); + ver = GET_APIC_VERSION(ver); + + /* + * get/set the timer IRQ vector: + */ + disable_8259A_irq(0); + assign_irq_vector(0, TARGET_CPUS); + + /* + * As IRQ0 is to be enabled in the 8259A, the virtual + * wire has to be disabled in the local APIC. Also + * timer interrupts need to be acknowledged manually in + * the 8259A for the i82489DX when using the NMI + * watchdog as that APIC treats NMIs as level-triggered. + * The AEOI mode will finish them in the 8259A + * automatically. + */ + apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); + init_8259A(1); +#ifdef CONFIG_X86_32 + timer_ack = (nmi_watchdog == NMI_IO_APIC && !APIC_INTEGRATED(ver)); +#endif + + pin1 = find_isa_irq_pin(0, mp_INT); + apic1 = find_isa_irq_apic(0, mp_INT); + pin2 = ioapic_i8259.pin; + apic2 = ioapic_i8259.apic; + + apic_printk(APIC_QUIET, KERN_INFO "..TIMER: vector=0x%02X " + "apic1=%d pin1=%d apic2=%d pin2=%d\n", + cfg->vector, apic1, pin1, apic2, pin2); + + /* + * Some BIOS writers are clueless and report the ExtINTA + * I/O APIC input from the cascaded 8259A as the timer + * interrupt input. So just in case, if only one pin + * was found above, try it both directly and through the + * 8259A. + */ + if (pin1 == -1) { +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + panic("BIOS bug: timer not connected to IO-APIC"); +#endif + pin1 = pin2; + apic1 = apic2; + no_pin1 = 1; + } else if (pin2 == -1) { + pin2 = pin1; + apic2 = apic1; + } + + if (pin1 != -1) { + /* + * Ok, does IRQ0 through the IOAPIC work? + */ + if (no_pin1) { + add_pin_to_irq(0, apic1, pin1); + setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); + } + unmask_IO_APIC_irq(0); + if (timer_irq_works()) { + if (nmi_watchdog == NMI_IO_APIC) { + setup_nmi(); + enable_8259A_irq(0); + } + if (disable_timer_pin_1 > 0) + clear_IO_APIC_pin(0, pin1); + goto out; + } +#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) + panic("timer doesn't work through Interrupt-remapped IO-APIC"); +#endif + clear_IO_APIC_pin(apic1, pin1); + if (!no_pin1) + apic_printk(APIC_QUIET, KERN_ERR "..MP-BIOS bug: " + "8254 timer not connected to IO-APIC\n"); + + apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer " + "(IRQ0) through the 8259A ...\n"); + apic_printk(APIC_QUIET, KERN_INFO + "..... (found apic %d pin %d) ...\n", apic2, pin2); + /* + * legacy devices should be connected to IO APIC #0 + */ + replace_pin_at_irq(0, apic1, pin1, apic2, pin2); + setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); + unmask_IO_APIC_irq(0); + enable_8259A_irq(0); + if (timer_irq_works()) { + apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); + timer_through_8259 = 1; + if (nmi_watchdog == NMI_IO_APIC) { + disable_8259A_irq(0); + setup_nmi(); + enable_8259A_irq(0); + } + goto out; + } + /* + * Cleanup, just in case ... + */ + disable_8259A_irq(0); + clear_IO_APIC_pin(apic2, pin2); + apic_printk(APIC_QUIET, KERN_INFO "....... failed.\n"); + } + + if (nmi_watchdog == NMI_IO_APIC) { + apic_printk(APIC_QUIET, KERN_WARNING "timer doesn't work " + "through the IO-APIC - disabling NMI Watchdog!\n"); + nmi_watchdog = NMI_NONE; + } +#ifdef CONFIG_X86_32 + timer_ack = 0; +#endif + + apic_printk(APIC_QUIET, KERN_INFO + "...trying to set up timer as Virtual Wire IRQ...\n"); + + lapic_register_intr(0); + apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ + enable_8259A_irq(0); + + if (timer_irq_works()) { + apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); + goto out; + } + disable_8259A_irq(0); + apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector); + apic_printk(APIC_QUIET, KERN_INFO "..... failed.\n"); + + apic_printk(APIC_QUIET, KERN_INFO + "...trying to set up timer as ExtINT IRQ...\n"); + + init_8259A(0); + make_8259A_irq(0); + apic_write(APIC_LVT0, APIC_DM_EXTINT); + + unlock_ExtINT_logic(); + + if (timer_irq_works()) { + apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); + goto out; + } + apic_printk(APIC_QUIET, KERN_INFO "..... failed :(.\n"); + panic("IO-APIC + timer doesn't work! Boot with apic=debug and send a " + "report. Then try booting with the 'noapic' option.\n"); +out: + local_irq_restore(flags); +} + +/* + * Traditionally ISA IRQ2 is the cascade IRQ, and is not available + * to devices. However there may be an I/O APIC pin available for + * this interrupt regardless. The pin may be left unconnected, but + * typically it will be reused as an ExtINT cascade interrupt for + * the master 8259A. In the MPS case such a pin will normally be + * reported as an ExtINT interrupt in the MP table. With ACPI + * there is no provision for ExtINT interrupts, and in the absence + * of an override it would be treated as an ordinary ISA I/O APIC + * interrupt, that is edge-triggered and unmasked by default. We + * used to do this, but it caused problems on some systems because + * of the NMI watchdog and sometimes IRQ0 of the 8254 timer using + * the same ExtINT cascade interrupt to drive the local APIC of the + * bootstrap processor. Therefore we refrain from routing IRQ2 to + * the I/O APIC in all cases now. No actual device should request + * it anyway. --macro + */ +#define PIC_IRQS (1 << PIC_CASCADE_IR) + +void __init setup_IO_APIC(void) +{ + +#ifdef CONFIG_X86_32 + enable_IO_APIC(); +#else + /* + * calling enable_IO_APIC() is moved to setup_local_APIC for BP + */ +#endif + + io_apic_irqs = ~PIC_IRQS; + + apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); + /* + * Set up IO-APIC IRQ routing. + */ +#ifdef CONFIG_X86_32 + if (!acpi_ioapic) + setup_ioapic_ids_from_mpc(); +#endif + sync_Arb_IDs(); + setup_IO_APIC_irqs(); + init_IO_APIC_traps(); + check_timer(); +} + +/* + * Called after all the initialization is done. If we didnt find any + * APIC bugs then we can allow the modify fast path + */ + +static int __init io_apic_bug_finalize(void) +{ + if (sis_apic_bug == -1) + sis_apic_bug = 0; + return 0; +} + +late_initcall(io_apic_bug_finalize); + +struct sysfs_ioapic_data { + struct sys_device dev; + struct IO_APIC_route_entry entry[0]; +}; +static struct sysfs_ioapic_data * mp_ioapic_data[MAX_IO_APICS]; + +static int ioapic_suspend(struct sys_device *dev, pm_message_t state) +{ + struct IO_APIC_route_entry *entry; + struct sysfs_ioapic_data *data; + int i; + + data = container_of(dev, struct sysfs_ioapic_data, dev); + entry = data->entry; + for (i = 0; i < nr_ioapic_registers[dev->id]; i ++, entry ++ ) + *entry = ioapic_read_entry(dev->id, i); + + return 0; +} + +static int ioapic_resume(struct sys_device *dev) +{ + struct IO_APIC_route_entry *entry; + struct sysfs_ioapic_data *data; + unsigned long flags; + union IO_APIC_reg_00 reg_00; + int i; + + data = container_of(dev, struct sysfs_ioapic_data, dev); + entry = data->entry; + + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(dev->id, 0); + if (reg_00.bits.ID != mp_ioapics[dev->id].mp_apicid) { + reg_00.bits.ID = mp_ioapics[dev->id].mp_apicid; + io_apic_write(dev->id, 0, reg_00.raw); + } + spin_unlock_irqrestore(&ioapic_lock, flags); + for (i = 0; i < nr_ioapic_registers[dev->id]; i++) + ioapic_write_entry(dev->id, i, entry[i]); + + return 0; +} + +static struct sysdev_class ioapic_sysdev_class = { + .name = "ioapic", + .suspend = ioapic_suspend, + .resume = ioapic_resume, +}; + +static int __init ioapic_init_sysfs(void) +{ + struct sys_device * dev; + int i, size, error; + + error = sysdev_class_register(&ioapic_sysdev_class); + if (error) + return error; + + for (i = 0; i < nr_ioapics; i++ ) { + size = sizeof(struct sys_device) + nr_ioapic_registers[i] + * sizeof(struct IO_APIC_route_entry); + mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL); + if (!mp_ioapic_data[i]) { + printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); + continue; + } + dev = &mp_ioapic_data[i]->dev; + dev->id = i; + dev->cls = &ioapic_sysdev_class; + error = sysdev_register(dev); + if (error) { + kfree(mp_ioapic_data[i]); + mp_ioapic_data[i] = NULL; + printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); + continue; + } + } + + return 0; +} + +device_initcall(ioapic_init_sysfs); + +/* + * Dynamic irq allocate and deallocation + */ +unsigned int create_irq_nr(unsigned int irq_want) +{ + /* Allocate an unused irq */ + unsigned int irq; + unsigned int new; + unsigned long flags; + struct irq_cfg *cfg_new; + +#ifndef CONFIG_HAVE_SPARSE_IRQ + irq_want = nr_irqs - 1; +#endif + + irq = 0; + spin_lock_irqsave(&vector_lock, flags); + for (new = irq_want; new > 0; new--) { + if (platform_legacy_irq(new)) + continue; + cfg_new = irq_cfg(new); + if (cfg_new && cfg_new->vector != 0) + continue; + /* check if need to create one */ + if (!cfg_new) + cfg_new = irq_cfg_alloc(new); + if (__assign_irq_vector(new, TARGET_CPUS) == 0) + irq = new; + break; + } + spin_unlock_irqrestore(&vector_lock, flags); + + if (irq > 0) { + dynamic_irq_init(irq); + } + return irq; +} + +int create_irq(void) +{ + int irq; + + irq = create_irq_nr(nr_irqs - 1); + + if (irq == 0) + irq = -1; + + return irq; +} + +void destroy_irq(unsigned int irq) +{ + unsigned long flags; + + dynamic_irq_cleanup(irq); + +#ifdef CONFIG_INTR_REMAP + free_irte(irq); +#endif + spin_lock_irqsave(&vector_lock, flags); + __clear_irq_vector(irq); + spin_unlock_irqrestore(&vector_lock, flags); +} + +/* + * MSI message composition + */ +#ifdef CONFIG_PCI_MSI +static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg) +{ + struct irq_cfg *cfg; + int err; + unsigned dest; + cpumask_t tmp; + + tmp = TARGET_CPUS; + err = assign_irq_vector(irq, tmp); + if (err) + return err; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, tmp); + dest = cpu_mask_to_apicid(tmp); + +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + struct irte irte; + int ir_index; + u16 sub_handle; + + ir_index = map_irq_to_irte_handle(irq, &sub_handle); + BUG_ON(ir_index == -1); + + memset (&irte, 0, sizeof(irte)); + + irte.present = 1; + irte.dst_mode = INT_DEST_MODE; + irte.trigger_mode = 0; /* edge */ + irte.dlvry_mode = INT_DELIVERY_MODE; + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + modify_irte(irq, &irte); + + msg->address_hi = MSI_ADDR_BASE_HI; + msg->data = sub_handle; + msg->address_lo = MSI_ADDR_BASE_LO | MSI_ADDR_IR_EXT_INT | + MSI_ADDR_IR_SHV | + MSI_ADDR_IR_INDEX1(ir_index) | + MSI_ADDR_IR_INDEX2(ir_index); + } else +#endif + { + msg->address_hi = MSI_ADDR_BASE_HI; + msg->address_lo = + MSI_ADDR_BASE_LO | + ((INT_DEST_MODE == 0) ? + MSI_ADDR_DEST_MODE_PHYSICAL: + MSI_ADDR_DEST_MODE_LOGICAL) | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_ADDR_REDIRECTION_CPU: + MSI_ADDR_REDIRECTION_LOWPRI) | + MSI_ADDR_DEST_ID(dest); + + msg->data = + MSI_DATA_TRIGGER_EDGE | + MSI_DATA_LEVEL_ASSERT | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + MSI_DATA_DELIVERY_FIXED: + MSI_DATA_DELIVERY_LOWPRI) | + MSI_DATA_VECTOR(cfg->vector); + } + return err; +} + +#ifdef CONFIG_SMP +static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + struct msi_msg msg; + unsigned int dest; + cpumask_t tmp; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + read_msi_msg(irq, &msg); + + msg.data &= ~MSI_DATA_VECTOR_MASK; + msg.data |= MSI_DATA_VECTOR(cfg->vector); + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; + msg.address_lo |= MSI_ADDR_DEST_ID(dest); + + write_msi_msg(irq, &msg); + desc = irq_to_desc(irq); + desc->affinity = mask; +} + +#ifdef CONFIG_INTR_REMAP +/* + * Migrate the MSI irq to another cpumask. This migration is + * done in the process context using interrupt-remapping hardware. + */ +static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + unsigned int dest; + cpumask_t tmp, cleanup_mask; + struct irte irte; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (get_irte(irq, &irte)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + irte.vector = cfg->vector; + irte.dest_id = IRTE_DEST(dest); + + /* + * atomically update the IRTE with the new destination and vector. + */ + modify_irte(irq, &irte); + + /* + * After this point, all the interrupts will start arriving + * at the new destination. So, time to cleanup the previous + * vector allocation. + */ + if (cfg->move_in_progress) { + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cfg->move_in_progress = 0; + } + + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif +#endif /* CONFIG_SMP */ + +/* + * IRQ Chip for MSI PCI/PCI-X/PCI-Express Devices, + * which implement the MSI or MSI-X Capability Structure. + */ +static struct irq_chip msi_chip = { + .name = "PCI-MSI", + .unmask = unmask_msi_irq, + .mask = mask_msi_irq, + .ack = ack_apic_edge, +#ifdef CONFIG_SMP + .set_affinity = set_msi_irq_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +#ifdef CONFIG_INTR_REMAP +static struct irq_chip msi_ir_chip = { + .name = "IR-PCI-MSI", + .unmask = unmask_msi_irq, + .mask = mask_msi_irq, + .ack = ack_x2apic_edge, +#ifdef CONFIG_SMP + .set_affinity = ir_set_msi_irq_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +/* + * Map the PCI dev to the corresponding remapping hardware unit + * and allocate 'nvec' consecutive interrupt-remapping table entries + * in it. + */ +static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) +{ + struct intel_iommu *iommu; + int index; + + iommu = map_dev_to_ir(dev); + if (!iommu) { + printk(KERN_ERR + "Unable to map PCI %s to iommu\n", pci_name(dev)); + return -ENOENT; + } + + index = alloc_irte(iommu, irq, nvec); + if (index < 0) { + printk(KERN_ERR + "Unable to allocate %d IRTE for PCI %s\n", nvec, + pci_name(dev)); + return -ENOSPC; + } + return index; +} +#endif + +static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) +{ + int ret; + struct msi_msg msg; + + ret = msi_compose_msg(dev, irq, &msg); + if (ret < 0) + return ret; + + set_irq_msi(irq, desc); + write_msi_msg(irq, &msg); + +#ifdef CONFIG_INTR_REMAP + if (irq_remapped(irq)) { + struct irq_desc *desc = irq_to_desc(irq); + /* + * irq migration in process context + */ + desc->status |= IRQ_MOVE_PCNTXT; + set_irq_chip_and_handler_name(irq, &msi_ir_chip, handle_edge_irq, "edge"); + } else +#endif + set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); + + return 0; +} + +static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) +{ + unsigned int irq; + + irq = dev->bus->number; + irq <<= 8; + irq |= dev->devfn; + irq <<= 12; + + return irq; +} + +int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) +{ + unsigned int irq; + int ret; + unsigned int irq_want; + + irq_want = build_irq_for_pci_dev(dev) + 0x100; + + irq = create_irq_nr(irq_want); + if (irq == 0) + return -1; + +#ifdef CONFIG_INTR_REMAP + if (!intr_remapping_enabled) + goto no_ir; + + ret = msi_alloc_irte(dev, irq, 1); + if (ret < 0) + goto error; +no_ir: +#endif + ret = setup_msi_irq(dev, desc, irq); + if (ret < 0) { + destroy_irq(irq); + return ret; + } + return 0; + +#ifdef CONFIG_INTR_REMAP +error: + destroy_irq(irq); + return ret; +#endif +} + +int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) +{ + unsigned int irq; + int ret, sub_handle; + struct msi_desc *desc; + unsigned int irq_want; + +#ifdef CONFIG_INTR_REMAP + struct intel_iommu *iommu = 0; + int index = 0; +#endif + + irq_want = build_irq_for_pci_dev(dev) + 0x100; + sub_handle = 0; + list_for_each_entry(desc, &dev->msi_list, list) { + irq = create_irq_nr(irq_want--); + if (irq == 0) + return -1; +#ifdef CONFIG_INTR_REMAP + if (!intr_remapping_enabled) + goto no_ir; + + if (!sub_handle) { + /* + * allocate the consecutive block of IRTE's + * for 'nvec' + */ + index = msi_alloc_irte(dev, irq, nvec); + if (index < 0) { + ret = index; + goto error; + } + } else { + iommu = map_dev_to_ir(dev); + if (!iommu) { + ret = -ENOENT; + goto error; + } + /* + * setup the mapping between the irq and the IRTE + * base index, the sub_handle pointing to the + * appropriate interrupt remap table entry. + */ + set_irte_irq(irq, iommu, index, sub_handle); + } +no_ir: +#endif + ret = setup_msi_irq(dev, desc, irq); + if (ret < 0) + goto error; + sub_handle++; + } + return 0; + +error: + destroy_irq(irq); + return ret; +} + +void arch_teardown_msi_irq(unsigned int irq) +{ + destroy_irq(irq); +} + +#ifdef CONFIG_DMAR +#ifdef CONFIG_SMP +static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + struct msi_msg msg; + unsigned int dest; + cpumask_t tmp; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + dmar_msi_read(irq, &msg); + + msg.data &= ~MSI_DATA_VECTOR_MASK; + msg.data |= MSI_DATA_VECTOR(cfg->vector); + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; + msg.address_lo |= MSI_ADDR_DEST_ID(dest); + + dmar_msi_write(irq, &msg); + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif /* CONFIG_SMP */ + +struct irq_chip dmar_msi_type = { + .name = "DMAR_MSI", + .unmask = dmar_msi_unmask, + .mask = dmar_msi_mask, + .ack = ack_apic_edge, +#ifdef CONFIG_SMP + .set_affinity = dmar_msi_set_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +int arch_setup_dmar_msi(unsigned int irq) +{ + int ret; + struct msi_msg msg; + + ret = msi_compose_msg(NULL, irq, &msg); + if (ret < 0) + return ret; + dmar_msi_write(irq, &msg); + set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq, + "edge"); + return 0; +} +#endif + +#endif /* CONFIG_PCI_MSI */ +/* + * Hypertransport interrupt support + */ +#ifdef CONFIG_HT_IRQ + +#ifdef CONFIG_SMP + +static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) +{ + struct ht_irq_msg msg; + fetch_ht_irq_msg(irq, &msg); + + msg.address_lo &= ~(HT_IRQ_LOW_VECTOR_MASK | HT_IRQ_LOW_DEST_ID_MASK); + msg.address_hi &= ~(HT_IRQ_HIGH_DEST_ID_MASK); + + msg.address_lo |= HT_IRQ_LOW_VECTOR(vector) | HT_IRQ_LOW_DEST_ID(dest); + msg.address_hi |= HT_IRQ_HIGH_DEST_ID(dest); + + write_ht_irq_msg(irq, &msg); +} + +static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + unsigned int dest; + cpumask_t tmp; + struct irq_desc *desc; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + target_ht_irq(irq, dest, cfg->vector); + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif + +static struct irq_chip ht_irq_chip = { + .name = "PCI-HT", + .mask = mask_ht_irq, + .unmask = unmask_ht_irq, + .ack = ack_apic_edge, +#ifdef CONFIG_SMP + .set_affinity = set_ht_irq_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) +{ + struct irq_cfg *cfg; + int err; + cpumask_t tmp; + + tmp = TARGET_CPUS; + err = assign_irq_vector(irq, tmp); + if (!err) { + struct ht_irq_msg msg; + unsigned dest; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, tmp); + dest = cpu_mask_to_apicid(tmp); + + msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); + + msg.address_lo = + HT_IRQ_LOW_BASE | + HT_IRQ_LOW_DEST_ID(dest) | + HT_IRQ_LOW_VECTOR(cfg->vector) | + ((INT_DEST_MODE == 0) ? + HT_IRQ_LOW_DM_PHYSICAL : + HT_IRQ_LOW_DM_LOGICAL) | + HT_IRQ_LOW_RQEOI_EDGE | + ((INT_DELIVERY_MODE != dest_LowestPrio) ? + HT_IRQ_LOW_MT_FIXED : + HT_IRQ_LOW_MT_ARBITRATED) | + HT_IRQ_LOW_IRQ_MASKED; + + write_ht_irq_msg(irq, &msg); + + set_irq_chip_and_handler_name(irq, &ht_irq_chip, + handle_edge_irq, "edge"); + } + return err; +} +#endif /* CONFIG_HT_IRQ */ + +/* -------------------------------------------------------------------------- + ACPI-based IOAPIC Configuration + -------------------------------------------------------------------------- */ + +#ifdef CONFIG_ACPI + +#ifdef CONFIG_X86_32 +int __init io_apic_get_unique_id(int ioapic, int apic_id) +{ + union IO_APIC_reg_00 reg_00; + static physid_mask_t apic_id_map = PHYSID_MASK_NONE; + physid_mask_t tmp; + unsigned long flags; + int i = 0; + + /* + * The P4 platform supports up to 256 APIC IDs on two separate APIC + * buses (one for LAPICs, one for IOAPICs), where predecessors only + * supports up to 16 on one shared APIC bus. + * + * TBD: Expand LAPIC/IOAPIC support on P4-class systems to take full + * advantage of new APIC bus architecture. + */ + + if (physids_empty(apic_id_map)) + apic_id_map = ioapic_phys_id_map(phys_cpu_present_map); + + spin_lock_irqsave(&ioapic_lock, flags); + reg_00.raw = io_apic_read(ioapic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + if (apic_id >= get_physical_broadcast()) { + printk(KERN_WARNING "IOAPIC[%d]: Invalid apic_id %d, trying " + "%d\n", ioapic, apic_id, reg_00.bits.ID); + apic_id = reg_00.bits.ID; + } + + /* + * Every APIC in a system must have a unique ID or we get lots of nice + * 'stuck on smp_invalidate_needed IPI wait' messages. + */ + if (check_apicid_used(apic_id_map, apic_id)) { + + for (i = 0; i < get_physical_broadcast(); i++) { + if (!check_apicid_used(apic_id_map, i)) + break; + } + + if (i == get_physical_broadcast()) + panic("Max apic_id exceeded!\n"); + + printk(KERN_WARNING "IOAPIC[%d]: apic_id %d already used, " + "trying %d\n", ioapic, apic_id, i); + + apic_id = i; + } + + tmp = apicid_to_cpu_present(apic_id); + physids_or(apic_id_map, apic_id_map, tmp); + + if (reg_00.bits.ID != apic_id) { + reg_00.bits.ID = apic_id; + + spin_lock_irqsave(&ioapic_lock, flags); + io_apic_write(ioapic, 0, reg_00.raw); + reg_00.raw = io_apic_read(ioapic, 0); + spin_unlock_irqrestore(&ioapic_lock, flags); + + /* Sanity check */ + if (reg_00.bits.ID != apic_id) { + printk("IOAPIC[%d]: Unable to change apic_id!\n", ioapic); + return -1; + } + } + + apic_printk(APIC_VERBOSE, KERN_INFO + "IOAPIC[%d]: Assigned apic_id %d\n", ioapic, apic_id); + + return apic_id; +} + +int __init io_apic_get_version(int ioapic) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(ioapic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + + return reg_01.bits.version; +} +#endif + +int __init io_apic_get_redir_entries (int ioapic) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(ioapic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + + return reg_01.bits.entries; +} + + +int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) +{ + if (!IO_APIC_IRQ(irq)) { + apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", + ioapic); + return -EINVAL; + } + + /* + * IRQs < 16 are already in the irq_2_pin[] map + */ + if (irq >= 16) + add_pin_to_irq(irq, ioapic, pin); + + setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); + + return 0; +} + + +int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) +{ + int i; + + if (skip_ioapic_setup) + return -1; + + for (i = 0; i < mp_irq_entries; i++) + if (mp_irqs[i].mp_irqtype == mp_INT && + mp_irqs[i].mp_srcbusirq == bus_irq) + break; + if (i >= mp_irq_entries) + return -1; + + *trigger = irq_trigger(i); + *polarity = irq_polarity(i); + return 0; +} + +#endif /* CONFIG_ACPI */ + +/* + * This function currently is only a helper for the i386 smp boot process where + * we need to reprogram the ioredtbls to cater for the cpus which have come online + * so mask in all cases should simply be TARGET_CPUS + */ +#ifdef CONFIG_SMP +void __init setup_ioapic_dest(void) +{ + int pin, ioapic, irq, irq_entry; + struct irq_cfg *cfg; + + if (skip_ioapic_setup == 1) + return; + + for (ioapic = 0; ioapic < nr_ioapics; ioapic++) { + for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { + irq_entry = find_irq_entry(ioapic, pin, mp_INT); + if (irq_entry == -1) + continue; + irq = pin_2_irq(irq_entry, ioapic, pin); + + /* setup_IO_APIC_irqs could fail to get vector for some device + * when you have too many devices, because at that time only boot + * cpu is online. + */ + cfg = irq_cfg(irq); + if (!cfg->vector) + setup_IO_APIC_irq(ioapic, pin, irq, + irq_trigger(irq_entry), + irq_polarity(irq_entry)); +#ifdef CONFIG_INTR_REMAP + else if (intr_remapping_enabled) + set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); +#endif + else + set_ioapic_affinity_irq(irq, TARGET_CPUS); + } + + } +} +#endif + +#ifdef CONFIG_X86_64 +#define IOAPIC_RESOURCE_NAME_SIZE 11 + +static struct resource *ioapic_resources; + +static struct resource * __init ioapic_setup_resources(void) +{ + unsigned long n; + struct resource *res; + char *mem; + int i; + + if (nr_ioapics <= 0) + return NULL; + + n = IOAPIC_RESOURCE_NAME_SIZE + sizeof(struct resource); + n *= nr_ioapics; + + mem = alloc_bootmem(n); + res = (void *)mem; + + if (mem != NULL) { + mem += sizeof(struct resource) * nr_ioapics; + + for (i = 0; i < nr_ioapics; i++) { + res[i].name = mem; + res[i].flags = IORESOURCE_MEM | IORESOURCE_BUSY; + sprintf(mem, "IOAPIC %u", i); + mem += IOAPIC_RESOURCE_NAME_SIZE; + } + } + + ioapic_resources = res; + + return res; +} +#endif + +void __init ioapic_init_mappings(void) +{ + unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; + int i; +#ifdef CONFIG_X86_64 + struct resource *ioapic_res; + + ioapic_res = ioapic_setup_resources(); +#endif + for (i = 0; i < nr_ioapics; i++) { + if (smp_found_config) { + ioapic_phys = mp_ioapics[i].mp_apicaddr; +#ifdef CONFIG_X86_32 + if (!ioapic_phys) { + printk(KERN_ERR + "WARNING: bogus zero IO-APIC " + "address found in MPTABLE, " + "disabling IO/APIC support!\n"); + smp_found_config = 0; + skip_ioapic_setup = 1; + goto fake_ioapic_page; + } +#endif + } else { +#ifdef CONFIG_X86_32 +fake_ioapic_page: +#endif + ioapic_phys = (unsigned long) + alloc_bootmem_pages(PAGE_SIZE); + ioapic_phys = __pa(ioapic_phys); + } + set_fixmap_nocache(idx, ioapic_phys); + apic_printk(APIC_VERBOSE, + "mapped IOAPIC to %08lx (%08lx)\n", + __fix_to_virt(idx), ioapic_phys); + idx++; + +#ifdef CONFIG_X86_64 + if (ioapic_res != NULL) { + ioapic_res->start = ioapic_phys; + ioapic_res->end = ioapic_phys + (4 * 1024) - 1; + ioapic_res++; + } +#endif + } +} + +#ifdef CONFIG_X86_64 +static int __init ioapic_insert_resources(void) +{ + int i; + struct resource *r = ioapic_resources; + + if (!r) { + printk(KERN_ERR + "IO APIC resources could be not be allocated.\n"); + return -1; + } + + for (i = 0; i < nr_ioapics; i++) { + insert_resource(&iomem_resource, r); + r++; + } + + return 0; +} + +/* Insert the IO APIC resources after PCI initialization has occured to handle + * IO APICS that are mapped in on a BAR in PCI space. */ +late_initcall(ioapic_insert_resources); +#endif diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c deleted file mode 100644 index fba6d6ee3480..000000000000 --- a/arch/x86/kernel/io_apic_32.c +++ /dev/null @@ -1,3913 +0,0 @@ -/* - * Intel IO-APIC support for multi-Pentium hosts. - * - * Copyright (C) 1997, 1998, 1999, 2000 Ingo Molnar, Hajnalka Szabo - * - * Many thanks to Stig Venaas for trying out countless experimental - * patches and reporting/debugging problems patiently! - * - * (c) 1999, Multiple IO-APIC support, developed by - * Ken-ichi Yaku and - * Hidemi Kishimoto , - * further tested and cleaned up by Zach Brown - * and Ingo Molnar - * - * Fixes - * Maciej W. Rozycki : Bits for genuine 82489DX APICs; - * thanks to Eric Gilmore - * and Rolf G. Tews - * for testing these extensively - * Paul Diefenbaugh : Added full ACPI support - */ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include /* time_after() */ -#ifdef CONFIG_ACPI -#include -#endif -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include - -#define __apicdebuginit(type) static type __init - -/* - * Is the SiS APIC rmw bug present ? - * -1 = don't know, 0 = no, 1 = yes - */ -int sis_apic_bug = -1; - -static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); - -int first_free_entry; -/* - * Rough estimation of how many shared IRQs there are, can - * be changed anytime. - */ -int pin_map_size; - -/* - * # of IRQ routing registers - */ -int nr_ioapic_registers[MAX_IO_APICS]; - -/* I/O APIC entries */ -struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; -int nr_ioapics; - -/* MP IRQ source entries */ -struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; - -/* # of MP IRQ source entries */ -int mp_irq_entries; - -#if defined (CONFIG_MCA) || defined (CONFIG_EISA) -int mp_bus_id_to_type[MAX_MP_BUSSES]; -#endif - -DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); - -int skip_ioapic_setup; - -static int __init parse_noapic(char *str) -{ - /* disable IO-APIC */ - disable_ioapic_setup(); - return 0; -} -early_param("noapic", parse_noapic); - -struct irq_cfg; -struct irq_pin_list; -struct irq_cfg { - unsigned int irq; - struct irq_cfg *next; - struct irq_pin_list *irq_2_pin; - cpumask_t domain; - cpumask_t old_domain; - unsigned move_cleanup_count; - u8 vector; - u8 move_in_progress : 1; -}; - -/* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ -static struct irq_cfg irq_cfg_legacy[] __initdata = { - [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, -}; - -static struct irq_cfg irq_cfg_init = { .irq = -1U, }; -/* need to be biger than size of irq_cfg_legacy */ -static int nr_irq_cfg = 32; - -static int __init parse_nr_irq_cfg(char *arg) -{ - if (arg) { - nr_irq_cfg = simple_strtoul(arg, NULL, 0); - if (nr_irq_cfg < 32) - nr_irq_cfg = 32; - } - return 0; -} - -early_param("nr_irq_cfg", parse_nr_irq_cfg); - -static void init_one_irq_cfg(struct irq_cfg *cfg) -{ - memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); -} - -static struct irq_cfg *irq_cfgx; -static struct irq_cfg *irq_cfgx_free; -static void __init init_work(void *data) -{ - struct dyn_array *da = data; - struct irq_cfg *cfg; - int legacy_count; - int i; - - cfg = *da->name; - - memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - - legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); - for (i = legacy_count; i < *da->nr; i++) - init_one_irq_cfg(&cfg[i]); - - for (i = 1; i < *da->nr; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = &irq_cfgx[legacy_count]; - irq_cfgx[legacy_count - 1].next = NULL; -} - -#define for_each_irq_cfg(cfg) \ - for (cfg = irq_cfgx; cfg; cfg = cfg->next) - -DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); - -static struct irq_cfg *irq_cfg(unsigned int irq) -{ - struct irq_cfg *cfg; - - cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg = cfg->next; - } - - return NULL; -} - -static struct irq_cfg *irq_cfg_alloc(unsigned int irq) -{ - struct irq_cfg *cfg, *cfg_pri; - int i; - int count = 0; - - cfg_pri = cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg_pri = cfg; - cfg = cfg->next; - count++; - } - - if (!irq_cfgx_free) { - unsigned long phys; - unsigned long total_bytes; - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); - - total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; - if (after_bootmem) - cfg = kzalloc(total_bytes, GFP_ATOMIC); - else - cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); - - if (!cfg) - panic("please boot with nr_irq_cfg= %d\n", count * 2); - - phys = __pa(cfg); - printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); - - for (i = 0; i < nr_irq_cfg; i++) - init_one_irq_cfg(&cfg[i]); - - for (i = 1; i < nr_irq_cfg; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = cfg; - } - - cfg = irq_cfgx_free; - irq_cfgx_free = irq_cfgx_free->next; - cfg->next = NULL; - if (cfg_pri) - cfg_pri->next = cfg; - else - irq_cfgx = cfg; - cfg->irq = irq; - printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); -#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG - { - /* dump the results */ - struct irq_cfg *cfg; - unsigned long phys; - unsigned long bytes = sizeof(struct irq_cfg); - - printk(KERN_DEBUG "=========================== %d\n", irq); - printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); - for_each_irq_cfg(cfg) { - phys = __pa(cfg); - printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); - } - printk(KERN_DEBUG "===========================\n"); - } -#endif - return cfg; -} - -/* - * This is performance-critical, we want to do it O(1) - * - * the indexing order of this array favors 1:1 mappings - * between pins and IRQs. - */ - -struct irq_pin_list { - int apic, pin; - struct irq_pin_list *next; -}; - -static struct irq_pin_list *irq_2_pin_head; -/* fill one page ? */ -static int nr_irq_2_pin = 0x100; -static struct irq_pin_list *irq_2_pin_ptr; -static void __init irq_2_pin_init_work(void *data) -{ - struct dyn_array *da = data; - struct irq_pin_list *pin; - int i; - - pin = *da->name; - - for (i = 1; i < *da->nr; i++) - pin[i-1].next = &pin[i]; - - irq_2_pin_ptr = &pin[0]; -} -DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); - -static struct irq_pin_list *get_one_free_irq_2_pin(void) -{ - struct irq_pin_list *pin; - int i; - - pin = irq_2_pin_ptr; - - if (pin) { - irq_2_pin_ptr = pin->next; - pin->next = NULL; - return pin; - } - - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); - - if (after_bootmem) - pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, - GFP_ATOMIC); - else - pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * - nr_irq_2_pin, PAGE_SIZE, 0); - - if (!pin) - panic("can not get more irq_2_pin\n"); - - for (i = 1; i < nr_irq_2_pin; i++) - pin[i-1].next = &pin[i]; - - irq_2_pin_ptr = pin->next; - pin->next = NULL; - - return pin; -} - -struct io_apic { - unsigned int index; - unsigned int unused[3]; - unsigned int data; -}; - -static __attribute_const__ struct io_apic __iomem *io_apic_base(int idx) -{ - return (void __iomem *) __fix_to_virt(FIX_IO_APIC_BASE_0 + idx) - + (mp_ioapics[idx].mp_apicaddr & ~PAGE_MASK); -} - -static inline unsigned int io_apic_read(unsigned int apic, unsigned int reg) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - writel(reg, &io_apic->index); - return readl(&io_apic->data); -} - -static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned int value) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - writel(reg, &io_apic->index); - writel(value, &io_apic->data); -} - -/* - * Re-write a value: to be used for read-modify-write - * cycles where the read already set up the index register. - * - * Older SiS APIC requires we rewrite the index register - */ -static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - if (sis_apic_bug) - writel(reg, &io_apic->index); - writel(value, &io_apic->data); -} - -#ifdef CONFIG_X86_64 -static bool io_apic_level_ack_pending(unsigned int irq) -{ - struct irq_pin_list *entry; - unsigned long flags; - struct irq_cfg *cfg = irq_cfg(irq); - - spin_lock_irqsave(&ioapic_lock, flags); - entry = cfg->irq_2_pin; - for (;;) { - unsigned int reg; - int pin; - - if (!entry) - break; - pin = entry->pin; - reg = io_apic_read(entry->apic, 0x10 + pin*2); - /* Is the remote IRR bit set? */ - if (reg & IO_APIC_REDIR_REMOTE_IRR) { - spin_unlock_irqrestore(&ioapic_lock, flags); - return true; - } - if (!entry->next) - break; - entry = entry->next; - } - spin_unlock_irqrestore(&ioapic_lock, flags); - - return false; -} -#endif - -union entry_union { - struct { u32 w1, w2; }; - struct IO_APIC_route_entry entry; -}; - -static struct IO_APIC_route_entry ioapic_read_entry(int apic, int pin) -{ - union entry_union eu; - unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); - eu.w1 = io_apic_read(apic, 0x10 + 2 * pin); - eu.w2 = io_apic_read(apic, 0x11 + 2 * pin); - spin_unlock_irqrestore(&ioapic_lock, flags); - return eu.entry; -} - -/* - * When we write a new IO APIC routing entry, we need to write the high - * word first! If the mask bit in the low word is clear, we will enable - * the interrupt, and we need to make sure the entry is fully populated - * before that happens. - */ -static void -__ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) -{ - union entry_union eu; - eu.entry = e; - io_apic_write(apic, 0x11 + 2*pin, eu.w2); - io_apic_write(apic, 0x10 + 2*pin, eu.w1); -} - -static void ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) -{ - unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); - __ioapic_write_entry(apic, pin, e); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -/* - * When we mask an IO APIC routing entry, we need to write the low - * word first, in order to set the mask bit before we change the - * high bits! - */ -static void ioapic_mask_entry(int apic, int pin) -{ - unsigned long flags; - union entry_union eu = { .entry.mask = 1 }; - - spin_lock_irqsave(&ioapic_lock, flags); - io_apic_write(apic, 0x10 + 2*pin, eu.w1); - io_apic_write(apic, 0x11 + 2*pin, eu.w2); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -#ifdef CONFIG_SMP -static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) -{ - int apic, pin; - struct irq_cfg *cfg; - struct irq_pin_list *entry; - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - for (;;) { - unsigned int reg; - - if (!entry) - break; - - apic = entry->apic; - pin = entry->pin; -#ifdef CONFIG_INTR_REMAP - /* - * With interrupt-remapping, destination information comes - * from interrupt-remapping table entry. - */ - if (!irq_remapped(irq)) - io_apic_write(apic, 0x11 + pin*2, dest); -#else - io_apic_write(apic, 0x11 + pin*2, dest); -#endif - reg = io_apic_read(apic, 0x10 + pin*2); - reg &= ~IO_APIC_REDIR_VECTOR_MASK; - reg |= vector; - io_apic_modify(apic, 0x10 + pin*2, reg); - if (!entry->next) - break; - entry = entry->next; - } -} - -static int assign_irq_vector(int irq, cpumask_t mask); - -static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned long flags; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - cfg = irq_cfg(irq); - if (assign_irq_vector(irq, mask)) - return; - - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - /* - * Only the high 8 bits are valid. - */ - dest = SET_APIC_LOGICAL_ID(dest); - - desc = irq_to_desc(irq); - spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); - desc->affinity = mask; - spin_unlock_irqrestore(&ioapic_lock, flags); -} -#endif /* CONFIG_SMP */ - -/* - * The common case is 1:1 IRQ<->pin mappings. Sometimes there are - * shared ISA-space IRQs, so we have to support them. We are super - * fast in the common case, and fast for shared ISA-space IRQs. - */ -static void add_pin_to_irq(unsigned int irq, int apic, int pin) -{ - struct irq_cfg *cfg; - struct irq_pin_list *entry; - - /* first time to refer irq_cfg, so with new */ - cfg = irq_cfg_alloc(irq); - entry = cfg->irq_2_pin; - if (!entry) { - entry = get_one_free_irq_2_pin(); - cfg->irq_2_pin = entry; - entry->apic = apic; - entry->pin = pin; - printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); - return; - } - - while (entry->next) { - /* not again, please */ - if (entry->apic == apic && entry->pin == pin) - return; - - entry = entry->next; - } - - entry->next = get_one_free_irq_2_pin(); - entry = entry->next; - entry->apic = apic; - entry->pin = pin; - printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); -} - -/* - * Reroute an IRQ to a different pin. - */ -static void __init replace_pin_at_irq(unsigned int irq, - int oldapic, int oldpin, - int newapic, int newpin) -{ - struct irq_cfg *cfg = irq_cfg(irq); - struct irq_pin_list *entry = cfg->irq_2_pin; - int replaced = 0; - - while (entry) { - if (entry->apic == oldapic && entry->pin == oldpin) { - entry->apic = newapic; - entry->pin = newpin; - replaced = 1; - /* every one is different, right? */ - break; - } - entry = entry->next; - } - - /* why? call replace before add? */ - if (!replaced) - add_pin_to_irq(irq, newapic, newpin); -} - -#ifdef CONFIG_X86_64 -/* - * Synchronize the IO-APIC and the CPU by doing - * a dummy read from the IO-APIC - */ -static inline void io_apic_sync(unsigned int apic) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - readl(&io_apic->data); -} - -#define __DO_ACTION(R, ACTION, FINAL) \ - \ -{ \ - int pin; \ - struct irq_cfg *cfg; \ - struct irq_pin_list *entry; \ - \ - cfg = irq_cfg(irq); \ - entry = cfg->irq_2_pin; \ - for (;;) { \ - unsigned int reg; \ - if (!entry) \ - break; \ - pin = entry->pin; \ - reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ - reg ACTION; \ - io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ - FINAL; \ - if (!entry->next) \ - break; \ - entry = entry->next; \ - } \ -} - -#define DO_ACTION(name,R,ACTION, FINAL) \ - \ - static void name##_IO_APIC_irq (unsigned int irq) \ - __DO_ACTION(R, ACTION, FINAL) - -/* mask = 1 */ -DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) - -/* mask = 0 */ -DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) - -#else - -static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) -{ - struct irq_cfg *cfg; - struct irq_pin_list *entry; - unsigned int pin, reg; - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - for (;;) { - if (!entry) - break; - pin = entry->pin; - reg = io_apic_read(entry->apic, 0x10 + pin*2); - reg &= ~disable; - reg |= enable; - io_apic_modify(entry->apic, 0x10 + pin*2, reg); - if (!entry->next) - break; - entry = entry->next; - } -} - -/* mask = 1 */ -static void __mask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, 0); -} - -/* mask = 0 */ -static void __unmask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, 0, IO_APIC_REDIR_MASKED); -} - -/* mask = 1, trigger = 0 */ -static void __mask_and_edge_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, - IO_APIC_REDIR_LEVEL_TRIGGER); -} - -/* mask = 0, trigger = 1 */ -static void __unmask_and_level_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_LEVEL_TRIGGER, - IO_APIC_REDIR_MASKED); -} - -#endif - -static void mask_IO_APIC_irq (unsigned int irq) -{ - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - __mask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -static void unmask_IO_APIC_irq (unsigned int irq) -{ - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - __unmask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) -{ - struct IO_APIC_route_entry entry; - - /* Check delivery_mode to be sure we're not clearing an SMI pin */ - entry = ioapic_read_entry(apic, pin); - if (entry.delivery_mode == dest_SMI) - return; - /* - * Disable it in the IO-APIC irq-routing table: - */ - ioapic_mask_entry(apic, pin); -} - -static void clear_IO_APIC (void) -{ - int apic, pin; - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) - clear_IO_APIC_pin(apic, pin); -} - -#if !defined(CONFIG_SMP) && defined(CONFIG_X86_32) -void send_IPI_self(int vector) -{ - unsigned int cfg; - - /* - * Wait for idle. - */ - apic_wait_icr_idle(); - cfg = APIC_DM_FIXED | APIC_DEST_SELF | vector | APIC_DEST_LOGICAL; - /* - * Send the IPI. The write to APIC_ICR fires this off. - */ - apic_write(APIC_ICR, cfg); -} -#endif /* !CONFIG_SMP && CONFIG_X86_32*/ - -#ifdef CONFIG_X86_32 -/* - * support for broken MP BIOSs, enables hand-redirection of PIRQ0-7 to - * specific CPU-side IRQs. - */ - -#define MAX_PIRQS 8 -static int pirq_entries [MAX_PIRQS]; -static int pirqs_enabled; - -static int __init ioapic_pirq_setup(char *str) -{ - int i, max; - int ints[MAX_PIRQS+1]; - - get_options(str, ARRAY_SIZE(ints), ints); - - for (i = 0; i < MAX_PIRQS; i++) - pirq_entries[i] = -1; - - pirqs_enabled = 1; - apic_printk(APIC_VERBOSE, KERN_INFO - "PIRQ redirection, working around broken MP-BIOS.\n"); - max = MAX_PIRQS; - if (ints[0] < MAX_PIRQS) - max = ints[0]; - - for (i = 0; i < max; i++) { - apic_printk(APIC_VERBOSE, KERN_DEBUG - "... PIRQ%d -> IRQ %d\n", i, ints[i+1]); - /* - * PIRQs are mapped upside down, usually. - */ - pirq_entries[MAX_PIRQS-i-1] = ints[i+1]; - } - return 1; -} - -__setup("pirq=", ioapic_pirq_setup); -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_INTR_REMAP -/* I/O APIC RTE contents at the OS boot up */ -static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; - -/* - * Saves and masks all the unmasked IO-APIC RTE's - */ -int save_mask_IO_APIC_setup(void) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - int apic, pin; - - /* - * The number of IO-APIC IRQ registers (== #pins): - */ - for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(apic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - nr_ioapic_registers[apic] = reg_01.bits.entries+1; - } - - for (apic = 0; apic < nr_ioapics; apic++) { - early_ioapic_entries[apic] = - kzalloc(sizeof(struct IO_APIC_route_entry) * - nr_ioapic_registers[apic], GFP_KERNEL); - if (!early_ioapic_entries[apic]) - return -ENOMEM; - } - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - struct IO_APIC_route_entry entry; - - entry = early_ioapic_entries[apic][pin] = - ioapic_read_entry(apic, pin); - if (!entry.mask) { - entry.mask = 1; - ioapic_write_entry(apic, pin, entry); - } - } - return 0; -} - -void restore_IO_APIC_setup(void) -{ - int apic, pin; - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) - ioapic_write_entry(apic, pin, - early_ioapic_entries[apic][pin]); -} - -void reinit_intr_remapped_IO_APIC(int intr_remapping) -{ - /* - * for now plain restore of previous settings. - * TBD: In the case of OS enabling interrupt-remapping, - * IO-APIC RTE's need to be setup to point to interrupt-remapping - * table entries. for now, do a plain restore, and wait for - * the setup_IO_APIC_irqs() to do proper initialization. - */ - restore_IO_APIC_setup(); -} -#endif - -/* - * Find the IRQ entry number of a certain pin. - */ -static int find_irq_entry(int apic, int pin, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == type && - (mp_irqs[i].mp_dstapic == mp_ioapics[apic].mp_apicid || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) && - mp_irqs[i].mp_dstirq == pin) - return i; - - return -1; -} - -/* - * Find the pin to which IRQ[irq] (ISA) is connected - */ -static int __init find_isa_irq_pin(int irq, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) - - return mp_irqs[i].mp_dstirq; - } - return -1; -} - -static int __init find_isa_irq_apic(int irq, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) - break; - } - if (i < mp_irq_entries) { - int apic; - for(apic = 0; apic < nr_ioapics; apic++) { - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic) - return apic; - } - } - - return -1; -} - -/* - * Find a specific PCI IRQ entry. - * Not an __init, possibly needed by modules - */ -static int pin_2_irq(int idx, int apic, int pin); - -int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) -{ - int apic, i, best_guess = -1; - - apic_printk(APIC_DEBUG, "querying PCI -> IRQ mapping bus:%d, slot:%d, pin:%d.\n", - bus, slot, pin); - if (test_bit(bus, mp_bus_not_pci)) { - apic_printk(APIC_VERBOSE, "PCI BIOS passed nonexistent PCI bus %d!\n", bus); - return -1; - } - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - for (apic = 0; apic < nr_ioapics; apic++) - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) - break; - - if (!test_bit(lbus, mp_bus_not_pci) && - !mp_irqs[i].mp_irqtype && - (bus == lbus) && - (slot == ((mp_irqs[i].mp_srcbusirq >> 2) & 0x1f))) { - int irq = pin_2_irq(i,apic,mp_irqs[i].mp_dstirq); - - if (!(apic || IO_APIC_IRQ(irq))) - continue; - - if (pin == (mp_irqs[i].mp_srcbusirq & 3)) - return irq; - /* - * Use the first all-but-pin matching entry as a - * best-guess fuzzy result for broken mptables. - */ - if (best_guess < 0) - best_guess = irq; - } - } - return best_guess; -} - -EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); - -#if defined(CONFIG_EISA) || defined(CONFIG_MCA) -/* - * EISA Edge/Level control register, ELCR - */ -static int EISA_ELCR(unsigned int irq) -{ - if (irq < 16) { - unsigned int port = 0x4d0 + (irq >> 3); - return (inb(port) >> (irq & 7)) & 1; - } - apic_printk(APIC_VERBOSE, KERN_INFO - "Broken MPtable reports ISA irq %d\n", irq); - return 0; -} - -#endif - -/* ISA interrupts are always polarity zero edge triggered, - * when listed as conforming in the MP table. */ - -#define default_ISA_trigger(idx) (0) -#define default_ISA_polarity(idx) (0) - -/* EISA interrupts are always polarity zero and can be edge or level - * trigger depending on the ELCR value. If an interrupt is listed as - * EISA conforming in the MP table, that means its trigger type must - * be read in from the ELCR */ - -#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].mp_srcbusirq)) -#define default_EISA_polarity(idx) default_ISA_polarity(idx) - -/* PCI interrupts are always polarity one level triggered, - * when listed as conforming in the MP table. */ - -#define default_PCI_trigger(idx) (1) -#define default_PCI_polarity(idx) (1) - -/* MCA interrupts are always polarity zero level triggered, - * when listed as conforming in the MP table. */ - -#define default_MCA_trigger(idx) (1) -#define default_MCA_polarity(idx) default_ISA_polarity(idx) - -static int MPBIOS_polarity(int idx) -{ - int bus = mp_irqs[idx].mp_srcbus; - int polarity; - - /* - * Determine IRQ line polarity (high active or low active): - */ - switch (mp_irqs[idx].mp_irqflag & 3) - { - case 0: /* conforms, ie. bus-type dependent polarity */ - if (test_bit(bus, mp_bus_not_pci)) - polarity = default_ISA_polarity(idx); - else - polarity = default_PCI_polarity(idx); - break; - case 1: /* high active */ - { - polarity = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } - case 3: /* low active */ - { - polarity = 1; - break; - } - default: /* invalid */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } - } - return polarity; -} - -static int MPBIOS_trigger(int idx) -{ - int bus = mp_irqs[idx].mp_srcbus; - int trigger; - - /* - * Determine IRQ trigger mode (edge or level sensitive): - */ - switch ((mp_irqs[idx].mp_irqflag>>2) & 3) - { - case 0: /* conforms, ie. bus-type dependent */ - if (test_bit(bus, mp_bus_not_pci)) - trigger = default_ISA_trigger(idx); - else - trigger = default_PCI_trigger(idx); -#if defined(CONFIG_EISA) || defined(CONFIG_MCA) - switch (mp_bus_id_to_type[bus]) { - case MP_BUS_ISA: /* ISA pin */ - { - /* set before the switch */ - break; - } - case MP_BUS_EISA: /* EISA pin */ - { - trigger = default_EISA_trigger(idx); - break; - } - case MP_BUS_PCI: /* PCI pin */ - { - /* set before the switch */ - break; - } - case MP_BUS_MCA: /* MCA pin */ - { - trigger = default_MCA_trigger(idx); - break; - } - default: - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; - break; - } - } -#endif - break; - case 1: /* edge */ - { - trigger = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; - break; - } - case 3: /* level */ - { - trigger = 1; - break; - } - default: /* invalid */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 0; - break; - } - } - return trigger; -} - -static inline int irq_polarity(int idx) -{ - return MPBIOS_polarity(idx); -} - -static inline int irq_trigger(int idx) -{ - return MPBIOS_trigger(idx); -} - -int (*ioapic_renumber_irq)(int ioapic, int irq); -static int pin_2_irq(int idx, int apic, int pin) -{ - int irq, i; - int bus = mp_irqs[idx].mp_srcbus; - - /* - * Debugging check, we are in big trouble if this message pops up! - */ - if (mp_irqs[idx].mp_dstirq != pin) - printk(KERN_ERR "broken BIOS or MPTABLE parser, ayiee!!\n"); - - if (test_bit(bus, mp_bus_not_pci)) { - irq = mp_irqs[idx].mp_srcbusirq; - } else { - /* - * PCI IRQs are mapped in order - */ - i = irq = 0; - while (i < apic) - irq += nr_ioapic_registers[i++]; - irq += pin; - /* - * For MPS mode, so far only needed by ES7000 platform - */ - if (ioapic_renumber_irq) - irq = ioapic_renumber_irq(apic, irq); - } - -#ifdef CONFIG_X86_32 - /* - * PCI IRQ command line redirection. Yes, limits are hardcoded. - */ - if ((pin >= 16) && (pin <= 23)) { - if (pirq_entries[pin-16] != -1) { - if (!pirq_entries[pin-16]) { - apic_printk(APIC_VERBOSE, KERN_DEBUG - "disabling PIRQ%d\n", pin-16); - } else { - irq = pirq_entries[pin-16]; - apic_printk(APIC_VERBOSE, KERN_DEBUG - "using PIRQ%d -> IRQ %d\n", - pin-16, irq); - } - } - } -#endif - - return irq; -} - -void lock_vector_lock(void) -{ - /* Used to the online set of cpus does not change - * during assign_irq_vector. - */ - spin_lock(&vector_lock); -} - -void unlock_vector_lock(void) -{ - spin_unlock(&vector_lock); -} - -static int __assign_irq_vector(int irq, cpumask_t mask) -{ - /* - * NOTE! The local APIC isn't very good at handling - * multiple interrupts at the same interrupt level. - * As the interrupt level is determined by taking the - * vector number and shifting that right by 4, we - * want to spread these out a bit so that they don't - * all fall in the same interrupt level. - * - * Also, we've got to be careful not to trash gate - * 0x80, because int 0x80 is hm, kind of importantish. ;) - */ - static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; - unsigned int old_vector; - int cpu; - struct irq_cfg *cfg; - - cfg = irq_cfg(irq); - - /* Only try and allocate irqs on cpus that are present */ - cpus_and(mask, mask, cpu_online_map); - - if ((cfg->move_in_progress) || cfg->move_cleanup_count) - return -EBUSY; - - old_vector = cfg->vector; - if (old_vector) { - cpumask_t tmp; - cpus_and(tmp, cfg->domain, mask); - if (!cpus_empty(tmp)) - return 0; - } - - for_each_cpu_mask_nr(cpu, mask) { - cpumask_t domain, new_mask; - int new_cpu; - int vector, offset; - - domain = vector_allocation_domain(cpu); - cpus_and(new_mask, domain, cpu_online_map); - - vector = current_vector; - offset = current_offset; -next: - vector += 8; - if (vector >= first_system_vector) { - /* If we run out of vectors on large boxen, must share them. */ - offset = (offset + 1) % 8; - vector = FIRST_DEVICE_VECTOR + offset; - } - if (unlikely(current_vector == vector)) - continue; -#ifdef CONFIG_X86_64 - if (vector == IA32_SYSCALL_VECTOR) - goto next; -#else - if (vector == SYSCALL_VECTOR) - goto next; -#endif - for_each_cpu_mask_nr(new_cpu, new_mask) - if (per_cpu(vector_irq, new_cpu)[vector] != -1) - goto next; - /* Found one! */ - current_vector = vector; - current_offset = offset; - if (old_vector) { - cfg->move_in_progress = 1; - cfg->old_domain = cfg->domain; - } - for_each_cpu_mask_nr(new_cpu, new_mask) - per_cpu(vector_irq, new_cpu)[vector] = irq; - cfg->vector = vector; - cfg->domain = domain; - return 0; - } - return -ENOSPC; -} - -static int assign_irq_vector(int irq, cpumask_t mask) -{ - int err; - unsigned long flags; - - spin_lock_irqsave(&vector_lock, flags); - err = __assign_irq_vector(irq, mask); - spin_unlock_irqrestore(&vector_lock, flags); - return err; -} - -static void __clear_irq_vector(int irq) -{ - struct irq_cfg *cfg; - cpumask_t mask; - int cpu, vector; - - cfg = irq_cfg(irq); - BUG_ON(!cfg->vector); - - vector = cfg->vector; - cpus_and(mask, cfg->domain, cpu_online_map); - for_each_cpu_mask_nr(cpu, mask) - per_cpu(vector_irq, cpu)[vector] = -1; - - cfg->vector = 0; - cpus_clear(cfg->domain); -} - -void __setup_vector_irq(int cpu) -{ - /* Initialize vector_irq on a new cpu */ - /* This function must be called with vector_lock held */ - int irq, vector; - struct irq_cfg *cfg; - - /* Mark the inuse vectors */ - for_each_irq_cfg(cfg) { - if (!cpu_isset(cpu, cfg->domain)) - continue; - vector = cfg->vector; - irq = cfg->irq; - per_cpu(vector_irq, cpu)[vector] = irq; - } - /* Mark the free vectors */ - for (vector = 0; vector < NR_VECTORS; ++vector) { - irq = per_cpu(vector_irq, cpu)[vector]; - if (irq < 0) - continue; - - cfg = irq_cfg(irq); - if (!cpu_isset(cpu, cfg->domain)) - per_cpu(vector_irq, cpu)[vector] = -1; - } -} - -static struct irq_chip ioapic_chip; -#ifdef CONFIG_INTR_REMAP -static struct irq_chip ir_ioapic_chip; -#endif - -#define IOAPIC_AUTO -1 -#define IOAPIC_EDGE 0 -#define IOAPIC_LEVEL 1 - -#ifdef CONFIG_X86_32 -static inline int IO_APIC_irq_trigger(int irq) -{ - int apic, idx, pin; - - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - idx = find_irq_entry(apic, pin, mp_INT); - if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) - return irq_trigger(idx); - } - } - /* - * nonexistent IRQs are edge default - */ - return 0; -} -#else -static inline int IO_APIC_irq_trigger(int irq) -{ - return 1; -} -#endif - -static void ioapic_register_intr(int irq, unsigned long trigger) -{ - struct irq_desc *desc; - - /* first time to use this irq_desc */ - if (irq < 16) - desc = irq_to_desc(irq); - else - desc = irq_to_desc_alloc(irq); - - if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || - trigger == IOAPIC_LEVEL) - desc->status |= IRQ_LEVEL; - else - desc->status &= ~IRQ_LEVEL; - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - desc->status |= IRQ_MOVE_PCNTXT; - if (trigger) - set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, - handle_fasteoi_irq, - "fasteoi"); - else - set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, - handle_edge_irq, "edge"); - return; - } -#endif - if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || - trigger == IOAPIC_LEVEL) - set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_fasteoi_irq, - "fasteoi"); - else - set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_edge_irq, "edge"); -} - -static int setup_ioapic_entry(int apic, int irq, - struct IO_APIC_route_entry *entry, - unsigned int destination, int trigger, - int polarity, int vector) -{ - /* - * add it to the IO-APIC irq-routing table: - */ - memset(entry,0,sizeof(*entry)); - -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) { - struct intel_iommu *iommu = map_ioapic_to_ir(apic); - struct irte irte; - struct IR_IO_APIC_route_entry *ir_entry = - (struct IR_IO_APIC_route_entry *) entry; - int index; - - if (!iommu) - panic("No mapping iommu for ioapic %d\n", apic); - - index = alloc_irte(iommu, irq, 1); - if (index < 0) - panic("Failed to allocate IRTE for ioapic %d\n", apic); - - memset(&irte, 0, sizeof(irte)); - - irte.present = 1; - irte.dst_mode = INT_DEST_MODE; - irte.trigger_mode = trigger; - irte.dlvry_mode = INT_DELIVERY_MODE; - irte.vector = vector; - irte.dest_id = IRTE_DEST(destination); - - modify_irte(irq, &irte); - - ir_entry->index2 = (index >> 15) & 0x1; - ir_entry->zero = 0; - ir_entry->format = 1; - ir_entry->index = (index & 0x7fff); - } else -#endif - { - entry->delivery_mode = INT_DELIVERY_MODE; - entry->dest_mode = INT_DEST_MODE; - entry->dest = destination; - } - - entry->mask = 0; /* enable IRQ */ - entry->trigger = trigger; - entry->polarity = polarity; - entry->vector = vector; - - /* Mask level triggered irqs. - * Use IRQ_DELAYED_DISABLE for edge triggered irqs. - */ - if (trigger) - entry->mask = 1; - return 0; -} - -static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, - int trigger, int polarity) -{ - struct irq_cfg *cfg; - struct IO_APIC_route_entry entry; - cpumask_t mask; - - if (!IO_APIC_IRQ(irq)) - return; - - cfg = irq_cfg(irq); - - mask = TARGET_CPUS; - if (assign_irq_vector(irq, mask)) - return; - - cpus_and(mask, cfg->domain, mask); - - apic_printk(APIC_VERBOSE,KERN_DEBUG - "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " - "IRQ %d Mode:%i Active:%i)\n", - apic, mp_ioapics[apic].mp_apicid, pin, cfg->vector, - irq, trigger, polarity); - - - if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, - cpu_mask_to_apicid(mask), trigger, polarity, - cfg->vector)) { - printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", - mp_ioapics[apic].mp_apicid, pin); - __clear_irq_vector(irq); - return; - } - - ioapic_register_intr(irq, trigger); - if (irq < 16) - disable_8259A_irq(irq); - - ioapic_write_entry(apic, pin, entry); -} - -static void __init setup_IO_APIC_irqs(void) -{ - int apic, pin, idx, irq, first_notcon = 1; - - apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); - - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - - idx = find_irq_entry(apic,pin,mp_INT); - if (idx == -1) { - if (first_notcon) { - apic_printk(APIC_VERBOSE, KERN_DEBUG " IO-APIC (apicid-pin) %d-%d", mp_ioapics[apic].mp_apicid, pin); - first_notcon = 0; - } else - apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mp_apicid, pin); - continue; - } - if (!first_notcon) { - apic_printk(APIC_VERBOSE, " not connected.\n"); - first_notcon = 1; - } - - irq = pin_2_irq(idx, apic, pin); -#ifdef CONFIG_X86_32 - if (multi_timer_check(apic, irq)) - continue; -#endif - add_pin_to_irq(irq, apic, pin); - - setup_IO_APIC_irq(apic, pin, irq, - irq_trigger(idx), irq_polarity(idx)); - } - } - - if (!first_notcon) - apic_printk(APIC_VERBOSE, " not connected.\n"); -} - -/* - * Set up the timer pin, possibly with the 8259A-master behind. - */ -static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, - int vector) -{ - struct IO_APIC_route_entry entry; - -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - return; -#endif - - memset(&entry, 0, sizeof(entry)); - - /* - * We use logical delivery to get the timer IRQ - * to the first CPU. - */ - entry.dest_mode = INT_DEST_MODE; - entry.mask = 1; /* mask IRQ now */ - entry.dest = cpu_mask_to_apicid(TARGET_CPUS); - entry.delivery_mode = INT_DELIVERY_MODE; - entry.polarity = 0; - entry.trigger = 0; - entry.vector = vector; - - /* - * The timer IRQ doesn't have to know that behind the - * scene we may have a 8259A-master in AEOI mode ... - */ - set_irq_chip_and_handler_name(0, &ioapic_chip, handle_edge_irq, "edge"); - - /* - * Add it to the IO-APIC irq-routing table: - */ - ioapic_write_entry(apic, pin, entry); -} - - -__apicdebuginit(void) print_IO_APIC(void) -{ - int apic, i; - union IO_APIC_reg_00 reg_00; - union IO_APIC_reg_01 reg_01; - union IO_APIC_reg_02 reg_02; - union IO_APIC_reg_03 reg_03; - unsigned long flags; - struct irq_cfg *cfg; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "number of MP IRQ sources: %d.\n", mp_irq_entries); - for (i = 0; i < nr_ioapics; i++) - printk(KERN_DEBUG "number of IO-APIC #%d registers: %d.\n", - mp_ioapics[i].mp_apicid, nr_ioapic_registers[i]); - - /* - * We are a bit conservative about what we expect. We have to - * know about every hardware change ASAP. - */ - printk(KERN_INFO "testing the IO APIC.......................\n"); - - for (apic = 0; apic < nr_ioapics; apic++) { - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - reg_01.raw = io_apic_read(apic, 1); - if (reg_01.bits.version >= 0x10) - reg_02.raw = io_apic_read(apic, 2); - if (reg_01.bits.version >= 0x20) - reg_03.raw = io_apic_read(apic, 3); - spin_unlock_irqrestore(&ioapic_lock, flags); - - printk("\n"); - printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); - printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); - printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); - printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); - printk(KERN_DEBUG "....... : LTS : %X\n", reg_00.bits.LTS); - - printk(KERN_DEBUG ".... register #01: %08X\n", *(int *)®_01); - printk(KERN_DEBUG "....... : max redirection entries: %04X\n", reg_01.bits.entries); - - printk(KERN_DEBUG "....... : PRQ implemented: %X\n", reg_01.bits.PRQ); - printk(KERN_DEBUG "....... : IO APIC version: %04X\n", reg_01.bits.version); - - /* - * Some Intel chipsets with IO APIC VERSION of 0x1? don't have reg_02, - * but the value of reg_02 is read as the previous read register - * value, so ignore it if reg_02 == reg_01. - */ - if (reg_01.bits.version >= 0x10 && reg_02.raw != reg_01.raw) { - printk(KERN_DEBUG ".... register #02: %08X\n", reg_02.raw); - printk(KERN_DEBUG "....... : arbitration: %02X\n", reg_02.bits.arbitration); - } - - /* - * Some Intel chipsets with IO APIC VERSION of 0x2? don't have reg_02 - * or reg_03, but the value of reg_0[23] is read as the previous read - * register value, so ignore it if reg_03 == reg_0[12]. - */ - if (reg_01.bits.version >= 0x20 && reg_03.raw != reg_02.raw && - reg_03.raw != reg_01.raw) { - printk(KERN_DEBUG ".... register #03: %08X\n", reg_03.raw); - printk(KERN_DEBUG "....... : Boot DT : %X\n", reg_03.bits.boot_DT); - } - - printk(KERN_DEBUG ".... IRQ redirection table:\n"); - - printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" - " Stat Dmod Deli Vect: \n"); - - for (i = 0; i <= reg_01.bits.entries; i++) { - struct IO_APIC_route_entry entry; - - entry = ioapic_read_entry(apic, i); - - printk(KERN_DEBUG " %02x %03X ", - i, - entry.dest - ); - - printk("%1d %1d %1d %1d %1d %1d %1d %02X\n", - entry.mask, - entry.trigger, - entry.irr, - entry.polarity, - entry.delivery_status, - entry.dest_mode, - entry.delivery_mode, - entry.vector - ); - } - } - printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for_each_irq_cfg(cfg) { - struct irq_pin_list *entry = cfg->irq_2_pin; - if (!entry) - continue; - printk(KERN_DEBUG "IRQ%d ", cfg->irq); - for (;;) { - printk("-> %d:%d", entry->apic, entry->pin); - if (!entry->next) - break; - entry = entry->next; - } - printk("\n"); - } - - printk(KERN_INFO ".................................... done.\n"); - - return; -} - -__apicdebuginit(void) print_APIC_bitfield(int base) -{ - unsigned int v; - int i, j; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "0123456789abcdef0123456789abcdef\n" KERN_DEBUG); - for (i = 0; i < 8; i++) { - v = apic_read(base + i*0x10); - for (j = 0; j < 32; j++) { - if (v & (1< 3) /* Due to the Pentium erratum 3AP. */ - apic_write(APIC_ESR, 0); - - v = apic_read(APIC_ESR); - printk(KERN_DEBUG "... APIC ESR: %08x\n", v); - } - - icr = apic_icr_read(); - printk(KERN_DEBUG "... APIC ICR: %08x\n", icr); - printk(KERN_DEBUG "... APIC ICR2: %08x\n", icr >> 32); - - v = apic_read(APIC_LVTT); - printk(KERN_DEBUG "... APIC LVTT: %08x\n", v); - - if (maxlvt > 3) { /* PC is LVT#4. */ - v = apic_read(APIC_LVTPC); - printk(KERN_DEBUG "... APIC LVTPC: %08x\n", v); - } - v = apic_read(APIC_LVT0); - printk(KERN_DEBUG "... APIC LVT0: %08x\n", v); - v = apic_read(APIC_LVT1); - printk(KERN_DEBUG "... APIC LVT1: %08x\n", v); - - if (maxlvt > 2) { /* ERR is LVT#3. */ - v = apic_read(APIC_LVTERR); - printk(KERN_DEBUG "... APIC LVTERR: %08x\n", v); - } - - v = apic_read(APIC_TMICT); - printk(KERN_DEBUG "... APIC TMICT: %08x\n", v); - v = apic_read(APIC_TMCCT); - printk(KERN_DEBUG "... APIC TMCCT: %08x\n", v); - v = apic_read(APIC_TDCR); - printk(KERN_DEBUG "... APIC TDCR: %08x\n", v); - printk("\n"); -} - -__apicdebuginit(void) print_all_local_APICs(void) -{ - on_each_cpu(print_local_APIC, NULL, 1); -} - -__apicdebuginit(void) print_PIC(void) -{ - unsigned int v; - unsigned long flags; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "\nprinting PIC contents\n"); - - spin_lock_irqsave(&i8259A_lock, flags); - - v = inb(0xa1) << 8 | inb(0x21); - printk(KERN_DEBUG "... PIC IMR: %04x\n", v); - - v = inb(0xa0) << 8 | inb(0x20); - printk(KERN_DEBUG "... PIC IRR: %04x\n", v); - - outb(0x0b,0xa0); - outb(0x0b,0x20); - v = inb(0xa0) << 8 | inb(0x20); - outb(0x0a,0xa0); - outb(0x0a,0x20); - - spin_unlock_irqrestore(&i8259A_lock, flags); - - printk(KERN_DEBUG "... PIC ISR: %04x\n", v); - - v = inb(0x4d1) << 8 | inb(0x4d0); - printk(KERN_DEBUG "... PIC ELCR: %04x\n", v); -} - -__apicdebuginit(int) print_all_ICs(void) -{ - print_PIC(); - print_all_local_APICs(); - print_IO_APIC(); - - return 0; -} - -fs_initcall(print_all_ICs); - - -/* Where if anywhere is the i8259 connect in external int mode */ -static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; - -void __init enable_IO_APIC(void) -{ - union IO_APIC_reg_01 reg_01; - int i8259_apic, i8259_pin; - int apic; - unsigned long flags; - -#ifdef CONFIG_X86_32 - int i; - if (!pirqs_enabled) - for (i = 0; i < MAX_PIRQS; i++) - pirq_entries[i] = -1; -#endif - - /* - * The number of IO-APIC IRQ registers (== #pins): - */ - for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(apic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - nr_ioapic_registers[apic] = reg_01.bits.entries+1; - } - for(apic = 0; apic < nr_ioapics; apic++) { - int pin; - /* See if any of the pins is in ExtINT mode */ - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - struct IO_APIC_route_entry entry; - entry = ioapic_read_entry(apic, pin); - - /* If the interrupt line is enabled and in ExtInt mode - * I have found the pin where the i8259 is connected. - */ - if ((entry.mask == 0) && (entry.delivery_mode == dest_ExtINT)) { - ioapic_i8259.apic = apic; - ioapic_i8259.pin = pin; - goto found_i8259; - } - } - } - found_i8259: - /* Look to see what if the MP table has reported the ExtINT */ - /* If we could not find the appropriate pin by looking at the ioapic - * the i8259 probably is not connected the ioapic but give the - * mptable a chance anyway. - */ - i8259_pin = find_isa_irq_pin(0, mp_ExtINT); - i8259_apic = find_isa_irq_apic(0, mp_ExtINT); - /* Trust the MP table if nothing is setup in the hardware */ - if ((ioapic_i8259.pin == -1) && (i8259_pin >= 0)) { - printk(KERN_WARNING "ExtINT not setup in hardware but reported by MP table\n"); - ioapic_i8259.pin = i8259_pin; - ioapic_i8259.apic = i8259_apic; - } - /* Complain if the MP table and the hardware disagree */ - if (((ioapic_i8259.apic != i8259_apic) || (ioapic_i8259.pin != i8259_pin)) && - (i8259_pin >= 0) && (ioapic_i8259.pin >= 0)) - { - printk(KERN_WARNING "ExtINT in hardware and MP table differ\n"); - } - - /* - * Do not trust the IO-APIC being empty at bootup - */ - clear_IO_APIC(); -} - -/* - * Not an __init, needed by the reboot code - */ -void disable_IO_APIC(void) -{ - /* - * Clear the IO-APIC before rebooting: - */ - clear_IO_APIC(); - - /* - * If the i8259 is routed through an IOAPIC - * Put that IOAPIC in virtual wire mode - * so legacy interrupts can be delivered. - */ - if (ioapic_i8259.pin != -1) { - struct IO_APIC_route_entry entry; - - memset(&entry, 0, sizeof(entry)); - entry.mask = 0; /* Enabled */ - entry.trigger = 0; /* Edge */ - entry.irr = 0; - entry.polarity = 0; /* High */ - entry.delivery_status = 0; - entry.dest_mode = 0; /* Physical */ - entry.delivery_mode = dest_ExtINT; /* ExtInt */ - entry.vector = 0; - entry.dest = read_apic_id(); - - /* - * Add it to the IO-APIC irq-routing table: - */ - ioapic_write_entry(ioapic_i8259.apic, ioapic_i8259.pin, entry); - } - - disconnect_bsp_APIC(ioapic_i8259.pin != -1); -} - -#ifdef CONFIG_X86_32 -/* - * function to set the IO-APIC physical IDs based on the - * values stored in the MPC table. - * - * by Matt Domsch Tue Dec 21 12:25:05 CST 1999 - */ - -static void __init setup_ioapic_ids_from_mpc(void) -{ - union IO_APIC_reg_00 reg_00; - physid_mask_t phys_id_present_map; - int apic; - int i; - unsigned char old_id; - unsigned long flags; - - if (x86_quirks->setup_ioapic_ids && x86_quirks->setup_ioapic_ids()) - return; - - /* - * Don't check I/O APIC IDs for xAPIC systems. They have - * no meaning without the serial APIC bus. - */ - if (!(boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) - || APIC_XAPIC(apic_version[boot_cpu_physical_apicid])) - return; - /* - * This is broken; anything with a real cpu count has to - * circumvent this idiocy regardless. - */ - phys_id_present_map = ioapic_phys_id_map(phys_cpu_present_map); - - /* - * Set the IOAPIC ID to the value stored in the MPC table. - */ - for (apic = 0; apic < nr_ioapics; apic++) { - - /* Read the register 0 value */ - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - old_id = mp_ioapics[apic].mp_apicid; - - if (mp_ioapics[apic].mp_apicid >= get_physical_broadcast()) { - printk(KERN_ERR "BIOS bug, IO-APIC#%d ID is %d in the MPC table!...\n", - apic, mp_ioapics[apic].mp_apicid); - printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", - reg_00.bits.ID); - mp_ioapics[apic].mp_apicid = reg_00.bits.ID; - } - - /* - * Sanity check, is the ID really free? Every APIC in a - * system must have a unique ID or we get lots of nice - * 'stuck on smp_invalidate_needed IPI wait' messages. - */ - if (check_apicid_used(phys_id_present_map, - mp_ioapics[apic].mp_apicid)) { - printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n", - apic, mp_ioapics[apic].mp_apicid); - for (i = 0; i < get_physical_broadcast(); i++) - if (!physid_isset(i, phys_id_present_map)) - break; - if (i >= get_physical_broadcast()) - panic("Max APIC ID exceeded!\n"); - printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", - i); - physid_set(i, phys_id_present_map); - mp_ioapics[apic].mp_apicid = i; - } else { - physid_mask_t tmp; - tmp = apicid_to_cpu_present(mp_ioapics[apic].mp_apicid); - apic_printk(APIC_VERBOSE, "Setting %d in the " - "phys_id_present_map\n", - mp_ioapics[apic].mp_apicid); - physids_or(phys_id_present_map, phys_id_present_map, tmp); - } - - - /* - * We need to adjust the IRQ routing table - * if the ID changed. - */ - if (old_id != mp_ioapics[apic].mp_apicid) - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_dstapic == old_id) - mp_irqs[i].mp_dstapic - = mp_ioapics[apic].mp_apicid; - - /* - * Read the right value from the MPC table and - * write it into the ID register. - */ - apic_printk(APIC_VERBOSE, KERN_INFO - "...changing IO-APIC physical APIC ID to %d ...", - mp_ioapics[apic].mp_apicid); - - reg_00.bits.ID = mp_ioapics[apic].mp_apicid; - spin_lock_irqsave(&ioapic_lock, flags); - - /* - * Sanity check - */ - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - if (reg_00.bits.ID != mp_ioapics[apic].mp_apicid) - printk("could not set ID!\n"); - else - apic_printk(APIC_VERBOSE, " ok.\n"); - } -} -#endif - -int no_timer_check __initdata; - -static int __init notimercheck(char *s) -{ - no_timer_check = 1; - return 1; -} -__setup("no_timer_check", notimercheck); - -/* - * There is a nasty bug in some older SMP boards, their mptable lies - * about the timer IRQ. We do the following to work around the situation: - * - * - timer IRQ defaults to IO-APIC IRQ - * - if this function detects that timer IRQs are defunct, then we fall - * back to ISA timer IRQs - */ -static int __init timer_irq_works(void) -{ - unsigned long t1 = jiffies; - unsigned long flags; - - if (no_timer_check) - return 1; - - local_save_flags(flags); - local_irq_enable(); - /* Let ten ticks pass... */ - mdelay((10 * 1000) / HZ); - local_irq_restore(flags); - - /* - * Expect a few ticks at least, to be sure some possible - * glue logic does not lock up after one or two first - * ticks in a non-ExtINT mode. Also the local APIC - * might have cached one ExtINT interrupt. Finally, at - * least one tick may be lost due to delays. - */ - - /* jiffies wrap? */ - if (time_after(jiffies, t1 + 4)) - return 1; - return 0; -} - -/* - * In the SMP+IOAPIC case it might happen that there are an unspecified - * number of pending IRQ events unhandled. These cases are very rare, - * so we 'resend' these IRQs via IPIs, to the same CPU. It's much - * better to do it this way as thus we do not have to be aware of - * 'pending' interrupts in the IRQ path, except at this point. - */ -/* - * Edge triggered needs to resend any interrupt - * that was delayed but this is now handled in the device - * independent code. - */ - -/* - * Starting up a edge-triggered IO-APIC interrupt is - * nasty - we need to make sure that we get the edge. - * If it is already asserted for some reason, we need - * return 1 to indicate that is was pending. - * - * This is not complete - we should be able to fake - * an edge even if it isn't on the 8259A... - */ - -static unsigned int startup_ioapic_irq(unsigned int irq) -{ - int was_pending = 0; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - if (irq < 16) { - disable_8259A_irq(irq); - if (i8259A_irq_pending(irq)) - was_pending = 1; - } - __unmask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return was_pending; -} - -#ifdef CONFIG_X86_64 -static int ioapic_retrigger_irq(unsigned int irq) -{ - - struct irq_cfg *cfg = irq_cfg(irq); - unsigned long flags; - - spin_lock_irqsave(&vector_lock, flags); - send_IPI_mask(cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); - spin_unlock_irqrestore(&vector_lock, flags); - - return 1; -} -#else -static int ioapic_retrigger_irq(unsigned int irq) -{ - send_IPI_self(irq_cfg(irq)->vector); - - return 1; -} -#endif - -/* - * Level and edge triggered IO-APIC interrupts need different handling, - * so we use two separate IRQ descriptors. Edge triggered IRQs can be - * handled with the level-triggered descriptor, but that one has slightly - * more overhead. Level-triggered interrupts cannot be handled with the - * edge-triggered handler, without risking IRQ storms and other ugly - * races. - */ - -#ifdef CONFIG_SMP - -#ifdef CONFIG_INTR_REMAP -static void ir_irq_migration(struct work_struct *work); - -static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); - -/* - * Migrate the IO-APIC irq in the presence of intr-remapping. - * - * For edge triggered, irq migration is a simple atomic update(of vector - * and cpu destination) of IRTE and flush the hardware cache. - * - * For level triggered, we need to modify the io-apic RTE aswell with the update - * vector information, along with modifying IRTE with vector and destination. - * So irq migration for level triggered is little bit more complex compared to - * edge triggered migration. But the good news is, we use the same algorithm - * for level triggered migration as we have today, only difference being, - * we now initiate the irq migration from process context instead of the - * interrupt context. - * - * In future, when we do a directed EOI (combined with cpu EOI broadcast - * suppression) to the IO-APIC, level triggered irq migration will also be - * as simple as edge triggered migration and we can do the irq migration - * with a simple atomic update to IO-APIC RTE. - */ -static void migrate_ioapic_irq(int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct irq_desc *desc; - cpumask_t tmp, cleanup_mask; - struct irte irte; - int modify_ioapic_rte; - unsigned int dest; - unsigned long flags; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (get_irte(irq, &irte)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - desc = irq_to_desc(irq); - modify_ioapic_rte = desc->status & IRQ_LEVEL; - if (modify_ioapic_rte) { - spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); - spin_unlock_irqrestore(&ioapic_lock, flags); - } - - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - /* - * Modified the IRTE and flushes the Interrupt entry cache. - */ - modify_irte(irq, &irte); - - if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } - - desc->affinity = mask; -} - -static int migrate_irq_remapped_level(int irq) -{ - int ret = -1; - struct irq_desc *desc = irq_to_desc(irq); - - mask_IO_APIC_irq(irq); - - if (io_apic_level_ack_pending(irq)) { - /* - * Interrupt in progress. Migrating irq now will change the - * vector information in the IO-APIC RTE and that will confuse - * the EOI broadcast performed by cpu. - * So, delay the irq migration to the next instance. - */ - schedule_delayed_work(&ir_migration_work, 1); - goto unmask; - } - - /* everthing is clear. we have right of way */ - migrate_ioapic_irq(irq, desc->pending_mask); - - ret = 0; - desc->status &= ~IRQ_MOVE_PENDING; - cpus_clear(desc->pending_mask); - -unmask: - unmask_IO_APIC_irq(irq); - return ret; -} - -static void ir_irq_migration(struct work_struct *work) -{ - unsigned int irq; - struct irq_desc *desc; - - for_each_irq_desc(irq, desc) { - if (desc->status & IRQ_MOVE_PENDING) { - unsigned long flags; - - spin_lock_irqsave(&desc->lock, flags); - if (!desc->chip->set_affinity || - !(desc->status & IRQ_MOVE_PENDING)) { - desc->status &= ~IRQ_MOVE_PENDING; - spin_unlock_irqrestore(&desc->lock, flags); - continue; - } - - desc->chip->set_affinity(irq, desc->pending_mask); - spin_unlock_irqrestore(&desc->lock, flags); - } - } -} - -/* - * Migrates the IRQ destination in the process context. - */ -static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) -{ - struct irq_desc *desc = irq_to_desc(irq); - - if (desc->status & IRQ_LEVEL) { - desc->status |= IRQ_MOVE_PENDING; - desc->pending_mask = mask; - migrate_irq_remapped_level(irq); - return; - } - - migrate_ioapic_irq(irq, mask); -} -#endif - -asmlinkage void smp_irq_move_cleanup_interrupt(void) -{ - unsigned vector, me; - ack_APIC_irq(); -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - - me = smp_processor_id(); - for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { - unsigned int irq; - struct irq_desc *desc; - struct irq_cfg *cfg; - irq = __get_cpu_var(vector_irq)[vector]; - - desc = irq_to_desc(irq); - if (!desc) - continue; - - cfg = irq_cfg(irq); - spin_lock(&desc->lock); - if (!cfg->move_cleanup_count) - goto unlock; - - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) - goto unlock; - - __get_cpu_var(vector_irq)[vector] = -1; - cfg->move_cleanup_count--; -unlock: - spin_unlock(&desc->lock); - } - - irq_exit(); -} - -static void irq_complete_move(unsigned int irq) -{ - struct irq_cfg *cfg = irq_cfg(irq); - unsigned vector, me; - - if (likely(!cfg->move_in_progress)) - return; - - vector = ~get_irq_regs()->orig_ax; - me = smp_processor_id(); - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { - cpumask_t cleanup_mask; - - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } -} -#else -static inline void irq_complete_move(unsigned int irq) {} -#endif -#ifdef CONFIG_INTR_REMAP -static void ack_x2apic_level(unsigned int irq) -{ - ack_x2APIC_irq(); -} - -static void ack_x2apic_edge(unsigned int irq) -{ - ack_x2APIC_irq(); -} -#endif - -static void ack_apic_edge(unsigned int irq) -{ - irq_complete_move(irq); - move_native_irq(irq); - ack_APIC_irq(); -} - -#ifdef CONFIG_X86_64 -static void ack_apic_level(unsigned int irq) -{ - int do_unmask_irq = 0; - - irq_complete_move(irq); -#ifdef CONFIG_GENERIC_PENDING_IRQ - /* If we are moving the irq we need to mask it */ - if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { - do_unmask_irq = 1; - mask_IO_APIC_irq(irq); - } -#endif - - /* - * We must acknowledge the irq before we move it or the acknowledge will - * not propagate properly. - */ - ack_APIC_irq(); - - /* Now we can move and renable the irq */ - if (unlikely(do_unmask_irq)) { - /* Only migrate the irq if the ack has been received. - * - * On rare occasions the broadcast level triggered ack gets - * delayed going to ioapics, and if we reprogram the - * vector while Remote IRR is still set the irq will never - * fire again. - * - * To prevent this scenario we read the Remote IRR bit - * of the ioapic. This has two effects. - * - On any sane system the read of the ioapic will - * flush writes (and acks) going to the ioapic from - * this cpu. - * - We get to see if the ACK has actually been delivered. - * - * Based on failed experiments of reprogramming the - * ioapic entry from outside of irq context starting - * with masking the ioapic entry and then polling until - * Remote IRR was clear before reprogramming the - * ioapic I don't trust the Remote IRR bit to be - * completey accurate. - * - * However there appears to be no other way to plug - * this race, so if the Remote IRR bit is not - * accurate and is causing problems then it is a hardware bug - * and you can go talk to the chipset vendor about it. - */ - if (!io_apic_level_ack_pending(irq)) - move_masked_irq(irq); - unmask_IO_APIC_irq(irq); - } -} -#else -atomic_t irq_mis_count; -static void ack_apic_level(unsigned int irq) -{ - unsigned long v; - int i; - - irq_complete_move(irq); - move_native_irq(irq); - /* - * It appears there is an erratum which affects at least version 0x11 - * of I/O APIC (that's the 82093AA and cores integrated into various - * chipsets). Under certain conditions a level-triggered interrupt is - * erroneously delivered as edge-triggered one but the respective IRR - * bit gets set nevertheless. As a result the I/O unit expects an EOI - * message but it will never arrive and further interrupts are blocked - * from the source. The exact reason is so far unknown, but the - * phenomenon was observed when two consecutive interrupt requests - * from a given source get delivered to the same CPU and the source is - * temporarily disabled in between. - * - * A workaround is to simulate an EOI message manually. We achieve it - * by setting the trigger mode to edge and then to level when the edge - * trigger mode gets detected in the TMR of a local APIC for a - * level-triggered interrupt. We mask the source for the time of the - * operation to prevent an edge-triggered interrupt escaping meanwhile. - * The idea is from Manfred Spraul. --macro - */ - i = irq_cfg(irq)->vector; - - v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); - - ack_APIC_irq(); - - if (!(v & (1 << (i & 0x1f)))) { - atomic_inc(&irq_mis_count); - spin_lock(&ioapic_lock); - __mask_and_edge_IO_APIC_irq(irq); - __unmask_and_level_IO_APIC_irq(irq); - spin_unlock(&ioapic_lock); - } -} -#endif - -static struct irq_chip ioapic_chip __read_mostly = { - .name = "IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_apic_edge, - .eoi = ack_apic_level, -#ifdef CONFIG_SMP - .set_affinity = set_ioapic_affinity_irq, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -#ifdef CONFIG_INTR_REMAP -static struct irq_chip ir_ioapic_chip __read_mostly = { - .name = "IR-IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_x2apic_edge, - .eoi = ack_x2apic_level, -#ifdef CONFIG_SMP - .set_affinity = set_ir_ioapic_affinity_irq, -#endif - .retrigger = ioapic_retrigger_irq, -}; -#endif - -static inline void init_IO_APIC_traps(void) -{ - int irq; - struct irq_desc *desc; - struct irq_cfg *cfg; - - /* - * NOTE! The local APIC isn't very good at handling - * multiple interrupts at the same interrupt level. - * As the interrupt level is determined by taking the - * vector number and shifting that right by 4, we - * want to spread these out a bit so that they don't - * all fall in the same interrupt level. - * - * Also, we've got to be careful not to trash gate - * 0x80, because int 0x80 is hm, kind of importantish. ;) - */ - for_each_irq_cfg(cfg) { - irq = cfg->irq; - if (IO_APIC_IRQ(irq) && !cfg->vector) { - /* - * Hmm.. We don't have an entry for this, - * so default to an old-fashioned 8259 - * interrupt if we can.. - */ - if (irq < 16) - make_8259A_irq(irq); - else { - desc = irq_to_desc(irq); - /* Strange. Oh, well.. */ - desc->chip = &no_irq_chip; - } - } - } -} - -/* - * The local APIC irq-chip implementation: - */ - -static void mask_lapic_irq(unsigned int irq) -{ - unsigned long v; - - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v | APIC_LVT_MASKED); -} - -static void unmask_lapic_irq(unsigned int irq) -{ - unsigned long v; - - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); -} - -static void ack_lapic_irq (unsigned int irq) -{ - ack_APIC_irq(); -} - -static struct irq_chip lapic_chip __read_mostly = { - .name = "local-APIC", - .mask = mask_lapic_irq, - .unmask = unmask_lapic_irq, - .ack = ack_lapic_irq, -}; - -static void lapic_register_intr(int irq) -{ - struct irq_desc *desc; - - desc = irq_to_desc(irq); - desc->status &= ~IRQ_LEVEL; - set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, - "edge"); -} - -static void __init setup_nmi(void) -{ - /* - * Dirty trick to enable the NMI watchdog ... - * We put the 8259A master into AEOI mode and - * unmask on all local APICs LVT0 as NMI. - * - * The idea to use the 8259A in AEOI mode ('8259A Virtual Wire') - * is from Maciej W. Rozycki - so we do not have to EOI from - * the NMI handler or the timer interrupt. - */ - apic_printk(APIC_VERBOSE, KERN_INFO "activating NMI Watchdog ..."); - - enable_NMI_through_LVT0(); - - apic_printk(APIC_VERBOSE, " done.\n"); -} - -/* - * This looks a bit hackish but it's about the only one way of sending - * a few INTA cycles to 8259As and any associated glue logic. ICR does - * not support the ExtINT mode, unfortunately. We need to send these - * cycles as some i82489DX-based boards have glue logic that keeps the - * 8259A interrupt line asserted until INTA. --macro - */ -static inline void __init unlock_ExtINT_logic(void) -{ - int apic, pin, i; - struct IO_APIC_route_entry entry0, entry1; - unsigned char save_control, save_freq_select; - - pin = find_isa_irq_pin(8, mp_INT); - if (pin == -1) { - WARN_ON_ONCE(1); - return; - } - apic = find_isa_irq_apic(8, mp_INT); - if (apic == -1) { - WARN_ON_ONCE(1); - return; - } - - entry0 = ioapic_read_entry(apic, pin); - clear_IO_APIC_pin(apic, pin); - - memset(&entry1, 0, sizeof(entry1)); - - entry1.dest_mode = 0; /* physical delivery */ - entry1.mask = 0; /* unmask IRQ now */ - entry1.dest = hard_smp_processor_id(); - entry1.delivery_mode = dest_ExtINT; - entry1.polarity = entry0.polarity; - entry1.trigger = 0; - entry1.vector = 0; - - ioapic_write_entry(apic, pin, entry1); - - save_control = CMOS_READ(RTC_CONTROL); - save_freq_select = CMOS_READ(RTC_FREQ_SELECT); - CMOS_WRITE((save_freq_select & ~RTC_RATE_SELECT) | 0x6, - RTC_FREQ_SELECT); - CMOS_WRITE(save_control | RTC_PIE, RTC_CONTROL); - - i = 100; - while (i-- > 0) { - mdelay(10); - if ((CMOS_READ(RTC_INTR_FLAGS) & RTC_PF) == RTC_PF) - i -= 10; - } - - CMOS_WRITE(save_control, RTC_CONTROL); - CMOS_WRITE(save_freq_select, RTC_FREQ_SELECT); - clear_IO_APIC_pin(apic, pin); - - ioapic_write_entry(apic, pin, entry0); -} - -static int disable_timer_pin_1 __initdata; -/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ -static int __init disable_timer_pin_setup(char *arg) -{ - disable_timer_pin_1 = 1; - return 0; -} -early_param("disable_timer_pin_1", disable_timer_pin_setup); - -int timer_through_8259 __initdata; - -/* - * This code may look a bit paranoid, but it's supposed to cooperate with - * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ - * is so screwy. Thanks to Brian Perkins for testing/hacking this beast - * fanatically on his truly buggy board. - * - * FIXME: really need to revamp this for all platforms. - */ -static inline void __init check_timer(void) -{ - struct irq_cfg *cfg = irq_cfg(0); - int apic1, pin1, apic2, pin2; - unsigned long flags; - unsigned int ver; - int no_pin1 = 0; - - local_irq_save(flags); - - ver = apic_read(APIC_LVR); - ver = GET_APIC_VERSION(ver); - - /* - * get/set the timer IRQ vector: - */ - disable_8259A_irq(0); - assign_irq_vector(0, TARGET_CPUS); - - /* - * As IRQ0 is to be enabled in the 8259A, the virtual - * wire has to be disabled in the local APIC. Also - * timer interrupts need to be acknowledged manually in - * the 8259A for the i82489DX when using the NMI - * watchdog as that APIC treats NMIs as level-triggered. - * The AEOI mode will finish them in the 8259A - * automatically. - */ - apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); - init_8259A(1); -#ifdef CONFIG_X86_32 - timer_ack = (nmi_watchdog == NMI_IO_APIC && !APIC_INTEGRATED(ver)); -#endif - - pin1 = find_isa_irq_pin(0, mp_INT); - apic1 = find_isa_irq_apic(0, mp_INT); - pin2 = ioapic_i8259.pin; - apic2 = ioapic_i8259.apic; - - apic_printk(APIC_QUIET, KERN_INFO "..TIMER: vector=0x%02X " - "apic1=%d pin1=%d apic2=%d pin2=%d\n", - cfg->vector, apic1, pin1, apic2, pin2); - - /* - * Some BIOS writers are clueless and report the ExtINTA - * I/O APIC input from the cascaded 8259A as the timer - * interrupt input. So just in case, if only one pin - * was found above, try it both directly and through the - * 8259A. - */ - if (pin1 == -1) { -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - panic("BIOS bug: timer not connected to IO-APIC"); -#endif - pin1 = pin2; - apic1 = apic2; - no_pin1 = 1; - } else if (pin2 == -1) { - pin2 = pin1; - apic2 = apic1; - } - - if (pin1 != -1) { - /* - * Ok, does IRQ0 through the IOAPIC work? - */ - if (no_pin1) { - add_pin_to_irq(0, apic1, pin1); - setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); - } - unmask_IO_APIC_irq(0); - if (timer_irq_works()) { - if (nmi_watchdog == NMI_IO_APIC) { - setup_nmi(); - enable_8259A_irq(0); - } - if (disable_timer_pin_1 > 0) - clear_IO_APIC_pin(0, pin1); - goto out; - } -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - panic("timer doesn't work through Interrupt-remapped IO-APIC"); -#endif - clear_IO_APIC_pin(apic1, pin1); - if (!no_pin1) - apic_printk(APIC_QUIET, KERN_ERR "..MP-BIOS bug: " - "8254 timer not connected to IO-APIC\n"); - - apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer " - "(IRQ0) through the 8259A ...\n"); - apic_printk(APIC_QUIET, KERN_INFO - "..... (found apic %d pin %d) ...\n", apic2, pin2); - /* - * legacy devices should be connected to IO APIC #0 - */ - replace_pin_at_irq(0, apic1, pin1, apic2, pin2); - setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); - unmask_IO_APIC_irq(0); - enable_8259A_irq(0); - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); - timer_through_8259 = 1; - if (nmi_watchdog == NMI_IO_APIC) { - disable_8259A_irq(0); - setup_nmi(); - enable_8259A_irq(0); - } - goto out; - } - /* - * Cleanup, just in case ... - */ - disable_8259A_irq(0); - clear_IO_APIC_pin(apic2, pin2); - apic_printk(APIC_QUIET, KERN_INFO "....... failed.\n"); - } - - if (nmi_watchdog == NMI_IO_APIC) { - apic_printk(APIC_QUIET, KERN_WARNING "timer doesn't work " - "through the IO-APIC - disabling NMI Watchdog!\n"); - nmi_watchdog = NMI_NONE; - } -#ifdef CONFIG_X86_32 - timer_ack = 0; -#endif - - apic_printk(APIC_QUIET, KERN_INFO - "...trying to set up timer as Virtual Wire IRQ...\n"); - - lapic_register_intr(0); - apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ - enable_8259A_irq(0); - - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); - goto out; - } - disable_8259A_irq(0); - apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector); - apic_printk(APIC_QUIET, KERN_INFO "..... failed.\n"); - - apic_printk(APIC_QUIET, KERN_INFO - "...trying to set up timer as ExtINT IRQ...\n"); - - init_8259A(0); - make_8259A_irq(0); - apic_write(APIC_LVT0, APIC_DM_EXTINT); - - unlock_ExtINT_logic(); - - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); - goto out; - } - apic_printk(APIC_QUIET, KERN_INFO "..... failed :(.\n"); - panic("IO-APIC + timer doesn't work! Boot with apic=debug and send a " - "report. Then try booting with the 'noapic' option.\n"); -out: - local_irq_restore(flags); -} - -/* - * Traditionally ISA IRQ2 is the cascade IRQ, and is not available - * to devices. However there may be an I/O APIC pin available for - * this interrupt regardless. The pin may be left unconnected, but - * typically it will be reused as an ExtINT cascade interrupt for - * the master 8259A. In the MPS case such a pin will normally be - * reported as an ExtINT interrupt in the MP table. With ACPI - * there is no provision for ExtINT interrupts, and in the absence - * of an override it would be treated as an ordinary ISA I/O APIC - * interrupt, that is edge-triggered and unmasked by default. We - * used to do this, but it caused problems on some systems because - * of the NMI watchdog and sometimes IRQ0 of the 8254 timer using - * the same ExtINT cascade interrupt to drive the local APIC of the - * bootstrap processor. Therefore we refrain from routing IRQ2 to - * the I/O APIC in all cases now. No actual device should request - * it anyway. --macro - */ -#define PIC_IRQS (1 << PIC_CASCADE_IR) - -void __init setup_IO_APIC(void) -{ - -#ifdef CONFIG_X86_32 - enable_IO_APIC(); -#else - /* - * calling enable_IO_APIC() is moved to setup_local_APIC for BP - */ -#endif - - io_apic_irqs = ~PIC_IRQS; - - apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); - /* - * Set up IO-APIC IRQ routing. - */ -#ifdef CONFIG_X86_32 - if (!acpi_ioapic) - setup_ioapic_ids_from_mpc(); -#endif - sync_Arb_IDs(); - setup_IO_APIC_irqs(); - init_IO_APIC_traps(); - check_timer(); -} - -/* - * Called after all the initialization is done. If we didnt find any - * APIC bugs then we can allow the modify fast path - */ - -static int __init io_apic_bug_finalize(void) -{ - if (sis_apic_bug == -1) - sis_apic_bug = 0; - return 0; -} - -late_initcall(io_apic_bug_finalize); - -struct sysfs_ioapic_data { - struct sys_device dev; - struct IO_APIC_route_entry entry[0]; -}; -static struct sysfs_ioapic_data * mp_ioapic_data[MAX_IO_APICS]; - -static int ioapic_suspend(struct sys_device *dev, pm_message_t state) -{ - struct IO_APIC_route_entry *entry; - struct sysfs_ioapic_data *data; - int i; - - data = container_of(dev, struct sysfs_ioapic_data, dev); - entry = data->entry; - for (i = 0; i < nr_ioapic_registers[dev->id]; i ++, entry ++ ) - *entry = ioapic_read_entry(dev->id, i); - - return 0; -} - -static int ioapic_resume(struct sys_device *dev) -{ - struct IO_APIC_route_entry *entry; - struct sysfs_ioapic_data *data; - unsigned long flags; - union IO_APIC_reg_00 reg_00; - int i; - - data = container_of(dev, struct sysfs_ioapic_data, dev); - entry = data->entry; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(dev->id, 0); - if (reg_00.bits.ID != mp_ioapics[dev->id].mp_apicid) { - reg_00.bits.ID = mp_ioapics[dev->id].mp_apicid; - io_apic_write(dev->id, 0, reg_00.raw); - } - spin_unlock_irqrestore(&ioapic_lock, flags); - for (i = 0; i < nr_ioapic_registers[dev->id]; i++) - ioapic_write_entry(dev->id, i, entry[i]); - - return 0; -} - -static struct sysdev_class ioapic_sysdev_class = { - .name = "ioapic", - .suspend = ioapic_suspend, - .resume = ioapic_resume, -}; - -static int __init ioapic_init_sysfs(void) -{ - struct sys_device * dev; - int i, size, error; - - error = sysdev_class_register(&ioapic_sysdev_class); - if (error) - return error; - - for (i = 0; i < nr_ioapics; i++ ) { - size = sizeof(struct sys_device) + nr_ioapic_registers[i] - * sizeof(struct IO_APIC_route_entry); - mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL); - if (!mp_ioapic_data[i]) { - printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); - continue; - } - dev = &mp_ioapic_data[i]->dev; - dev->id = i; - dev->cls = &ioapic_sysdev_class; - error = sysdev_register(dev); - if (error) { - kfree(mp_ioapic_data[i]); - mp_ioapic_data[i] = NULL; - printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); - continue; - } - } - - return 0; -} - -device_initcall(ioapic_init_sysfs); - -/* - * Dynamic irq allocate and deallocation - */ -unsigned int create_irq_nr(unsigned int irq_want) -{ - /* Allocate an unused irq */ - unsigned int irq; - unsigned int new; - unsigned long flags; - struct irq_cfg *cfg_new; - -#ifndef CONFIG_HAVE_SPARSE_IRQ - irq_want = nr_irqs - 1; -#endif - - irq = 0; - spin_lock_irqsave(&vector_lock, flags); - for (new = irq_want; new > 0; new--) { - if (platform_legacy_irq(new)) - continue; - cfg_new = irq_cfg(new); - if (cfg_new && cfg_new->vector != 0) - continue; - /* check if need to create one */ - if (!cfg_new) - cfg_new = irq_cfg_alloc(new); - if (__assign_irq_vector(new, TARGET_CPUS) == 0) - irq = new; - break; - } - spin_unlock_irqrestore(&vector_lock, flags); - - if (irq > 0) { - dynamic_irq_init(irq); - } - return irq; -} - -int create_irq(void) -{ - int irq; - - irq = create_irq_nr(nr_irqs - 1); - - if (irq == 0) - irq = -1; - - return irq; -} - -void destroy_irq(unsigned int irq) -{ - unsigned long flags; - - dynamic_irq_cleanup(irq); - -#ifdef CONFIG_INTR_REMAP - free_irte(irq); -#endif - spin_lock_irqsave(&vector_lock, flags); - __clear_irq_vector(irq); - spin_unlock_irqrestore(&vector_lock, flags); -} - -/* - * MSI message composition - */ -#ifdef CONFIG_PCI_MSI -static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg) -{ - struct irq_cfg *cfg; - int err; - unsigned dest; - cpumask_t tmp; - - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); - if (err) - return err; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - struct irte irte; - int ir_index; - u16 sub_handle; - - ir_index = map_irq_to_irte_handle(irq, &sub_handle); - BUG_ON(ir_index == -1); - - memset (&irte, 0, sizeof(irte)); - - irte.present = 1; - irte.dst_mode = INT_DEST_MODE; - irte.trigger_mode = 0; /* edge */ - irte.dlvry_mode = INT_DELIVERY_MODE; - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - modify_irte(irq, &irte); - - msg->address_hi = MSI_ADDR_BASE_HI; - msg->data = sub_handle; - msg->address_lo = MSI_ADDR_BASE_LO | MSI_ADDR_IR_EXT_INT | - MSI_ADDR_IR_SHV | - MSI_ADDR_IR_INDEX1(ir_index) | - MSI_ADDR_IR_INDEX2(ir_index); - } else -#endif - { - msg->address_hi = MSI_ADDR_BASE_HI; - msg->address_lo = - MSI_ADDR_BASE_LO | - ((INT_DEST_MODE == 0) ? - MSI_ADDR_DEST_MODE_PHYSICAL: - MSI_ADDR_DEST_MODE_LOGICAL) | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_ADDR_REDIRECTION_CPU: - MSI_ADDR_REDIRECTION_LOWPRI) | - MSI_ADDR_DEST_ID(dest); - - msg->data = - MSI_DATA_TRIGGER_EDGE | - MSI_DATA_LEVEL_ASSERT | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_DATA_DELIVERY_FIXED: - MSI_DATA_DELIVERY_LOWPRI) | - MSI_DATA_VECTOR(cfg->vector); - } - return err; -} - -#ifdef CONFIG_SMP -static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct msi_msg msg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - read_msi_msg(irq, &msg); - - msg.data &= ~MSI_DATA_VECTOR_MASK; - msg.data |= MSI_DATA_VECTOR(cfg->vector); - msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; - msg.address_lo |= MSI_ADDR_DEST_ID(dest); - - write_msi_msg(irq, &msg); - desc = irq_to_desc(irq); - desc->affinity = mask; -} - -#ifdef CONFIG_INTR_REMAP -/* - * Migrate the MSI irq to another cpumask. This migration is - * done in the process context using interrupt-remapping hardware. - */ -static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned int dest; - cpumask_t tmp, cleanup_mask; - struct irte irte; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (get_irte(irq, &irte)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - /* - * atomically update the IRTE with the new destination and vector. - */ - modify_irte(irq, &irte); - - /* - * After this point, all the interrupts will start arriving - * at the new destination. So, time to cleanup the previous - * vector allocation. - */ - if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } - - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif -#endif /* CONFIG_SMP */ - -/* - * IRQ Chip for MSI PCI/PCI-X/PCI-Express Devices, - * which implement the MSI or MSI-X Capability Structure. - */ -static struct irq_chip msi_chip = { - .name = "PCI-MSI", - .unmask = unmask_msi_irq, - .mask = mask_msi_irq, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = set_msi_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -#ifdef CONFIG_INTR_REMAP -static struct irq_chip msi_ir_chip = { - .name = "IR-PCI-MSI", - .unmask = unmask_msi_irq, - .mask = mask_msi_irq, - .ack = ack_x2apic_edge, -#ifdef CONFIG_SMP - .set_affinity = ir_set_msi_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -/* - * Map the PCI dev to the corresponding remapping hardware unit - * and allocate 'nvec' consecutive interrupt-remapping table entries - * in it. - */ -static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) -{ - struct intel_iommu *iommu; - int index; - - iommu = map_dev_to_ir(dev); - if (!iommu) { - printk(KERN_ERR - "Unable to map PCI %s to iommu\n", pci_name(dev)); - return -ENOENT; - } - - index = alloc_irte(iommu, irq, nvec); - if (index < 0) { - printk(KERN_ERR - "Unable to allocate %d IRTE for PCI %s\n", nvec, - pci_name(dev)); - return -ENOSPC; - } - return index; -} -#endif - -static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) -{ - int ret; - struct msi_msg msg; - - ret = msi_compose_msg(dev, irq, &msg); - if (ret < 0) - return ret; - - set_irq_msi(irq, desc); - write_msi_msg(irq, &msg); - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - struct irq_desc *desc = irq_to_desc(irq); - /* - * irq migration in process context - */ - desc->status |= IRQ_MOVE_PCNTXT; - set_irq_chip_and_handler_name(irq, &msi_ir_chip, handle_edge_irq, "edge"); - } else -#endif - set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); - - return 0; -} - -static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) -{ - unsigned int irq; - - irq = dev->bus->number; - irq <<= 8; - irq |= dev->devfn; - irq <<= 12; - - return irq; -} - -int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) -{ - unsigned int irq; - int ret; - unsigned int irq_want; - - irq_want = build_irq_for_pci_dev(dev) + 0x100; - - irq = create_irq_nr(irq_want); - if (irq == 0) - return -1; - -#ifdef CONFIG_INTR_REMAP - if (!intr_remapping_enabled) - goto no_ir; - - ret = msi_alloc_irte(dev, irq, 1); - if (ret < 0) - goto error; -no_ir: -#endif - ret = setup_msi_irq(dev, desc, irq); - if (ret < 0) { - destroy_irq(irq); - return ret; - } - return 0; - -#ifdef CONFIG_INTR_REMAP -error: - destroy_irq(irq); - return ret; -#endif -} - -int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) -{ - unsigned int irq; - int ret, sub_handle; - struct msi_desc *desc; - unsigned int irq_want; - -#ifdef CONFIG_INTR_REMAP - struct intel_iommu *iommu = 0; - int index = 0; -#endif - - irq_want = build_irq_for_pci_dev(dev) + 0x100; - sub_handle = 0; - list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq_nr(irq_want--); - if (irq == 0) - return -1; -#ifdef CONFIG_INTR_REMAP - if (!intr_remapping_enabled) - goto no_ir; - - if (!sub_handle) { - /* - * allocate the consecutive block of IRTE's - * for 'nvec' - */ - index = msi_alloc_irte(dev, irq, nvec); - if (index < 0) { - ret = index; - goto error; - } - } else { - iommu = map_dev_to_ir(dev); - if (!iommu) { - ret = -ENOENT; - goto error; - } - /* - * setup the mapping between the irq and the IRTE - * base index, the sub_handle pointing to the - * appropriate interrupt remap table entry. - */ - set_irte_irq(irq, iommu, index, sub_handle); - } -no_ir: -#endif - ret = setup_msi_irq(dev, desc, irq); - if (ret < 0) - goto error; - sub_handle++; - } - return 0; - -error: - destroy_irq(irq); - return ret; -} - -void arch_teardown_msi_irq(unsigned int irq) -{ - destroy_irq(irq); -} - -#ifdef CONFIG_DMAR -#ifdef CONFIG_SMP -static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct msi_msg msg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - dmar_msi_read(irq, &msg); - - msg.data &= ~MSI_DATA_VECTOR_MASK; - msg.data |= MSI_DATA_VECTOR(cfg->vector); - msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; - msg.address_lo |= MSI_ADDR_DEST_ID(dest); - - dmar_msi_write(irq, &msg); - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif /* CONFIG_SMP */ - -struct irq_chip dmar_msi_type = { - .name = "DMAR_MSI", - .unmask = dmar_msi_unmask, - .mask = dmar_msi_mask, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = dmar_msi_set_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -int arch_setup_dmar_msi(unsigned int irq) -{ - int ret; - struct msi_msg msg; - - ret = msi_compose_msg(NULL, irq, &msg); - if (ret < 0) - return ret; - dmar_msi_write(irq, &msg); - set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq, - "edge"); - return 0; -} -#endif - -#endif /* CONFIG_PCI_MSI */ -/* - * Hypertransport interrupt support - */ -#ifdef CONFIG_HT_IRQ - -#ifdef CONFIG_SMP - -static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) -{ - struct ht_irq_msg msg; - fetch_ht_irq_msg(irq, &msg); - - msg.address_lo &= ~(HT_IRQ_LOW_VECTOR_MASK | HT_IRQ_LOW_DEST_ID_MASK); - msg.address_hi &= ~(HT_IRQ_HIGH_DEST_ID_MASK); - - msg.address_lo |= HT_IRQ_LOW_VECTOR(vector) | HT_IRQ_LOW_DEST_ID(dest); - msg.address_hi |= HT_IRQ_HIGH_DEST_ID(dest); - - write_ht_irq_msg(irq, &msg); -} - -static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - target_ht_irq(irq, dest, cfg->vector); - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif - -static struct irq_chip ht_irq_chip = { - .name = "PCI-HT", - .mask = mask_ht_irq, - .unmask = unmask_ht_irq, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = set_ht_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) -{ - struct irq_cfg *cfg; - int err; - cpumask_t tmp; - - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); - if (!err) { - struct ht_irq_msg msg; - unsigned dest; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); - - msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); - - msg.address_lo = - HT_IRQ_LOW_BASE | - HT_IRQ_LOW_DEST_ID(dest) | - HT_IRQ_LOW_VECTOR(cfg->vector) | - ((INT_DEST_MODE == 0) ? - HT_IRQ_LOW_DM_PHYSICAL : - HT_IRQ_LOW_DM_LOGICAL) | - HT_IRQ_LOW_RQEOI_EDGE | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - HT_IRQ_LOW_MT_FIXED : - HT_IRQ_LOW_MT_ARBITRATED) | - HT_IRQ_LOW_IRQ_MASKED; - - write_ht_irq_msg(irq, &msg); - - set_irq_chip_and_handler_name(irq, &ht_irq_chip, - handle_edge_irq, "edge"); - } - return err; -} -#endif /* CONFIG_HT_IRQ */ - -/* -------------------------------------------------------------------------- - ACPI-based IOAPIC Configuration - -------------------------------------------------------------------------- */ - -#ifdef CONFIG_ACPI - -#ifdef CONFIG_X86_32 -int __init io_apic_get_unique_id(int ioapic, int apic_id) -{ - union IO_APIC_reg_00 reg_00; - static physid_mask_t apic_id_map = PHYSID_MASK_NONE; - physid_mask_t tmp; - unsigned long flags; - int i = 0; - - /* - * The P4 platform supports up to 256 APIC IDs on two separate APIC - * buses (one for LAPICs, one for IOAPICs), where predecessors only - * supports up to 16 on one shared APIC bus. - * - * TBD: Expand LAPIC/IOAPIC support on P4-class systems to take full - * advantage of new APIC bus architecture. - */ - - if (physids_empty(apic_id_map)) - apic_id_map = ioapic_phys_id_map(phys_cpu_present_map); - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - if (apic_id >= get_physical_broadcast()) { - printk(KERN_WARNING "IOAPIC[%d]: Invalid apic_id %d, trying " - "%d\n", ioapic, apic_id, reg_00.bits.ID); - apic_id = reg_00.bits.ID; - } - - /* - * Every APIC in a system must have a unique ID or we get lots of nice - * 'stuck on smp_invalidate_needed IPI wait' messages. - */ - if (check_apicid_used(apic_id_map, apic_id)) { - - for (i = 0; i < get_physical_broadcast(); i++) { - if (!check_apicid_used(apic_id_map, i)) - break; - } - - if (i == get_physical_broadcast()) - panic("Max apic_id exceeded!\n"); - - printk(KERN_WARNING "IOAPIC[%d]: apic_id %d already used, " - "trying %d\n", ioapic, apic_id, i); - - apic_id = i; - } - - tmp = apicid_to_cpu_present(apic_id); - physids_or(apic_id_map, apic_id_map, tmp); - - if (reg_00.bits.ID != apic_id) { - reg_00.bits.ID = apic_id; - - spin_lock_irqsave(&ioapic_lock, flags); - io_apic_write(ioapic, 0, reg_00.raw); - reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - /* Sanity check */ - if (reg_00.bits.ID != apic_id) { - printk("IOAPIC[%d]: Unable to change apic_id!\n", ioapic); - return -1; - } - } - - apic_printk(APIC_VERBOSE, KERN_INFO - "IOAPIC[%d]: Assigned apic_id %d\n", ioapic, apic_id); - - return apic_id; -} - -int __init io_apic_get_version(int ioapic) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return reg_01.bits.version; -} -#endif - -int __init io_apic_get_redir_entries (int ioapic) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return reg_01.bits.entries; -} - - -int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) -{ - if (!IO_APIC_IRQ(irq)) { - apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", - ioapic); - return -EINVAL; - } - - /* - * IRQs < 16 are already in the irq_2_pin[] map - */ - if (irq >= 16) - add_pin_to_irq(irq, ioapic, pin); - - setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); - - return 0; -} - - -int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) -{ - int i; - - if (skip_ioapic_setup) - return -1; - - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == mp_INT && - mp_irqs[i].mp_srcbusirq == bus_irq) - break; - if (i >= mp_irq_entries) - return -1; - - *trigger = irq_trigger(i); - *polarity = irq_polarity(i); - return 0; -} - -#endif /* CONFIG_ACPI */ - -/* - * This function currently is only a helper for the i386 smp boot process where - * we need to reprogram the ioredtbls to cater for the cpus which have come online - * so mask in all cases should simply be TARGET_CPUS - */ -#ifdef CONFIG_SMP -void __init setup_ioapic_dest(void) -{ - int pin, ioapic, irq, irq_entry; - struct irq_cfg *cfg; - - if (skip_ioapic_setup == 1) - return; - - for (ioapic = 0; ioapic < nr_ioapics; ioapic++) { - for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { - irq_entry = find_irq_entry(ioapic, pin, mp_INT); - if (irq_entry == -1) - continue; - irq = pin_2_irq(irq_entry, ioapic, pin); - - /* setup_IO_APIC_irqs could fail to get vector for some device - * when you have too many devices, because at that time only boot - * cpu is online. - */ - cfg = irq_cfg(irq); - if (!cfg->vector) - setup_IO_APIC_irq(ioapic, pin, irq, - irq_trigger(irq_entry), - irq_polarity(irq_entry)); -#ifdef CONFIG_INTR_REMAP - else if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); -#endif - else - set_ioapic_affinity_irq(irq, TARGET_CPUS); - } - - } -} -#endif - -#ifdef CONFIG_X86_64 -#define IOAPIC_RESOURCE_NAME_SIZE 11 - -static struct resource *ioapic_resources; - -static struct resource * __init ioapic_setup_resources(void) -{ - unsigned long n; - struct resource *res; - char *mem; - int i; - - if (nr_ioapics <= 0) - return NULL; - - n = IOAPIC_RESOURCE_NAME_SIZE + sizeof(struct resource); - n *= nr_ioapics; - - mem = alloc_bootmem(n); - res = (void *)mem; - - if (mem != NULL) { - mem += sizeof(struct resource) * nr_ioapics; - - for (i = 0; i < nr_ioapics; i++) { - res[i].name = mem; - res[i].flags = IORESOURCE_MEM | IORESOURCE_BUSY; - sprintf(mem, "IOAPIC %u", i); - mem += IOAPIC_RESOURCE_NAME_SIZE; - } - } - - ioapic_resources = res; - - return res; -} -#endif - -void __init ioapic_init_mappings(void) -{ - unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; - int i; -#ifdef CONFIG_X86_64 - struct resource *ioapic_res; - - ioapic_res = ioapic_setup_resources(); -#endif - for (i = 0; i < nr_ioapics; i++) { - if (smp_found_config) { - ioapic_phys = mp_ioapics[i].mp_apicaddr; -#ifdef CONFIG_X86_32 - if (!ioapic_phys) { - printk(KERN_ERR - "WARNING: bogus zero IO-APIC " - "address found in MPTABLE, " - "disabling IO/APIC support!\n"); - smp_found_config = 0; - skip_ioapic_setup = 1; - goto fake_ioapic_page; - } -#endif - } else { -#ifdef CONFIG_X86_32 -fake_ioapic_page: -#endif - ioapic_phys = (unsigned long) - alloc_bootmem_pages(PAGE_SIZE); - ioapic_phys = __pa(ioapic_phys); - } - set_fixmap_nocache(idx, ioapic_phys); - apic_printk(APIC_VERBOSE, - "mapped IOAPIC to %08lx (%08lx)\n", - __fix_to_virt(idx), ioapic_phys); - idx++; - -#ifdef CONFIG_X86_64 - if (ioapic_res != NULL) { - ioapic_res->start = ioapic_phys; - ioapic_res->end = ioapic_phys + (4 * 1024) - 1; - ioapic_res++; - } -#endif - } -} - -#ifdef CONFIG_X86_64 -static int __init ioapic_insert_resources(void) -{ - int i; - struct resource *r = ioapic_resources; - - if (!r) { - printk(KERN_ERR - "IO APIC resources could be not be allocated.\n"); - return -1; - } - - for (i = 0; i < nr_ioapics; i++) { - insert_resource(&iomem_resource, r); - r++; - } - - return 0; -} - -/* Insert the IO APIC resources after PCI initialization has occured to handle - * IO APICS that are mapped in on a BAR in PCI space. */ -late_initcall(ioapic_insert_resources); -#endif diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c deleted file mode 100644 index 940c4167b325..000000000000 --- a/arch/x86/kernel/io_apic_64.c +++ /dev/null @@ -1,3913 +0,0 @@ -/* - * Intel IO-APIC support for multi-Pentium hosts. - * - * Copyright (C) 1997, 1998, 1999, 2000 Ingo Molnar, Hajnalka Szabo - * - * Many thanks to Stig Venaas for trying out countless experimental - * patches and reporting/debugging problems patiently! - * - * (c) 1999, Multiple IO-APIC support, developed by - * Ken-ichi Yaku and - * Hidemi Kishimoto , - * further tested and cleaned up by Zach Brown - * and Ingo Molnar - * - * Fixes - * Maciej W. Rozycki : Bits for genuine 82489DX APICs; - * thanks to Eric Gilmore - * and Rolf G. Tews - * for testing these extensively - * Paul Diefenbaugh : Added full ACPI support - */ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include /* time_after() */ -#ifdef CONFIG_ACPI -#include -#endif -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include - -#define __apicdebuginit(type) static type __init - -/* - * Is the SiS APIC rmw bug present ? - * -1 = don't know, 0 = no, 1 = yes - */ -int sis_apic_bug = -1; - -static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); - -int first_free_entry; -/* - * Rough estimation of how many shared IRQs there are, can - * be changed anytime. - */ -int pin_map_size; - -/* - * # of IRQ routing registers - */ -int nr_ioapic_registers[MAX_IO_APICS]; - -/* I/O APIC entries */ -struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; -int nr_ioapics; - -/* MP IRQ source entries */ -struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; - -/* # of MP IRQ source entries */ -int mp_irq_entries; - -#if defined (CONFIG_MCA) || defined (CONFIG_EISA) -int mp_bus_id_to_type[MAX_MP_BUSSES]; -#endif - -DECLARE_BITMAP(mp_bus_not_pci, MAX_MP_BUSSES); - -int skip_ioapic_setup; - -static int __init parse_noapic(char *str) -{ - /* disable IO-APIC */ - disable_ioapic_setup(); - return 0; -} -early_param("noapic", parse_noapic); - -struct irq_cfg; -struct irq_pin_list; -struct irq_cfg { - unsigned int irq; - struct irq_cfg *next; - struct irq_pin_list *irq_2_pin; - cpumask_t domain; - cpumask_t old_domain; - unsigned move_cleanup_count; - u8 vector; - u8 move_in_progress : 1; -}; - -/* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ -static struct irq_cfg irq_cfg_legacy[] __initdata = { - [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, -}; - -static struct irq_cfg irq_cfg_init = { .irq = -1U, }; -/* need to be biger than size of irq_cfg_legacy */ -static int nr_irq_cfg = 32; - -static int __init parse_nr_irq_cfg(char *arg) -{ - if (arg) { - nr_irq_cfg = simple_strtoul(arg, NULL, 0); - if (nr_irq_cfg < 32) - nr_irq_cfg = 32; - } - return 0; -} - -early_param("nr_irq_cfg", parse_nr_irq_cfg); - -static void init_one_irq_cfg(struct irq_cfg *cfg) -{ - memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); -} - -static struct irq_cfg *irq_cfgx; -static struct irq_cfg *irq_cfgx_free; -static void __init init_work(void *data) -{ - struct dyn_array *da = data; - struct irq_cfg *cfg; - int legacy_count; - int i; - - cfg = *da->name; - - memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - - legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); - for (i = legacy_count; i < *da->nr; i++) - init_one_irq_cfg(&cfg[i]); - - for (i = 1; i < *da->nr; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = &irq_cfgx[legacy_count]; - irq_cfgx[legacy_count - 1].next = NULL; -} - -#define for_each_irq_cfg(cfg) \ - for (cfg = irq_cfgx; cfg; cfg = cfg->next) - -DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); - -static struct irq_cfg *irq_cfg(unsigned int irq) -{ - struct irq_cfg *cfg; - - cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg = cfg->next; - } - - return NULL; -} - -static struct irq_cfg *irq_cfg_alloc(unsigned int irq) -{ - struct irq_cfg *cfg, *cfg_pri; - int i; - int count = 0; - - cfg_pri = cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg_pri = cfg; - cfg = cfg->next; - count++; - } - - if (!irq_cfgx_free) { - unsigned long phys; - unsigned long total_bytes; - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); - - total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; - if (after_bootmem) - cfg = kzalloc(total_bytes, GFP_ATOMIC); - else - cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); - - if (!cfg) - panic("please boot with nr_irq_cfg= %d\n", count * 2); - - phys = __pa(cfg); - printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); - - for (i = 0; i < nr_irq_cfg; i++) - init_one_irq_cfg(&cfg[i]); - - for (i = 1; i < nr_irq_cfg; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = cfg; - } - - cfg = irq_cfgx_free; - irq_cfgx_free = irq_cfgx_free->next; - cfg->next = NULL; - if (cfg_pri) - cfg_pri->next = cfg; - else - irq_cfgx = cfg; - cfg->irq = irq; - printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); -#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG - { - /* dump the results */ - struct irq_cfg *cfg; - unsigned long phys; - unsigned long bytes = sizeof(struct irq_cfg); - - printk(KERN_DEBUG "=========================== %d\n", irq); - printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); - for_each_irq_cfg(cfg) { - phys = __pa(cfg); - printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); - } - printk(KERN_DEBUG "===========================\n"); - } -#endif - return cfg; -} - -/* - * This is performance-critical, we want to do it O(1) - * - * the indexing order of this array favors 1:1 mappings - * between pins and IRQs. - */ - -struct irq_pin_list { - int apic, pin; - struct irq_pin_list *next; -}; - -static struct irq_pin_list *irq_2_pin_head; -/* fill one page ? */ -static int nr_irq_2_pin = 0x100; -static struct irq_pin_list *irq_2_pin_ptr; -static void __init irq_2_pin_init_work(void *data) -{ - struct dyn_array *da = data; - struct irq_pin_list *pin; - int i; - - pin = *da->name; - - for (i = 1; i < *da->nr; i++) - pin[i-1].next = &pin[i]; - - irq_2_pin_ptr = &pin[0]; -} -DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); - -static struct irq_pin_list *get_one_free_irq_2_pin(void) -{ - struct irq_pin_list *pin; - int i; - - pin = irq_2_pin_ptr; - - if (pin) { - irq_2_pin_ptr = pin->next; - pin->next = NULL; - return pin; - } - - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); - - if (after_bootmem) - pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, - GFP_ATOMIC); - else - pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * - nr_irq_2_pin, PAGE_SIZE, 0); - - if (!pin) - panic("can not get more irq_2_pin\n"); - - for (i = 1; i < nr_irq_2_pin; i++) - pin[i-1].next = &pin[i]; - - irq_2_pin_ptr = pin->next; - pin->next = NULL; - - return pin; -} - -struct io_apic { - unsigned int index; - unsigned int unused[3]; - unsigned int data; -}; - -static __attribute_const__ struct io_apic __iomem *io_apic_base(int idx) -{ - return (void __iomem *) __fix_to_virt(FIX_IO_APIC_BASE_0 + idx) - + (mp_ioapics[idx].mp_apicaddr & ~PAGE_MASK); -} - -static inline unsigned int io_apic_read(unsigned int apic, unsigned int reg) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - writel(reg, &io_apic->index); - return readl(&io_apic->data); -} - -static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned int value) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - writel(reg, &io_apic->index); - writel(value, &io_apic->data); -} - -/* - * Re-write a value: to be used for read-modify-write - * cycles where the read already set up the index register. - * - * Older SiS APIC requires we rewrite the index register - */ -static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - if (sis_apic_bug) - writel(reg, &io_apic->index); - writel(value, &io_apic->data); -} - -#ifdef CONFIG_X86_64 -static bool io_apic_level_ack_pending(unsigned int irq) -{ - struct irq_pin_list *entry; - unsigned long flags; - struct irq_cfg *cfg = irq_cfg(irq); - - spin_lock_irqsave(&ioapic_lock, flags); - entry = cfg->irq_2_pin; - for (;;) { - unsigned int reg; - int pin; - - if (!entry) - break; - pin = entry->pin; - reg = io_apic_read(entry->apic, 0x10 + pin*2); - /* Is the remote IRR bit set? */ - if (reg & IO_APIC_REDIR_REMOTE_IRR) { - spin_unlock_irqrestore(&ioapic_lock, flags); - return true; - } - if (!entry->next) - break; - entry = entry->next; - } - spin_unlock_irqrestore(&ioapic_lock, flags); - - return false; -} -#endif - -union entry_union { - struct { u32 w1, w2; }; - struct IO_APIC_route_entry entry; -}; - -static struct IO_APIC_route_entry ioapic_read_entry(int apic, int pin) -{ - union entry_union eu; - unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); - eu.w1 = io_apic_read(apic, 0x10 + 2 * pin); - eu.w2 = io_apic_read(apic, 0x11 + 2 * pin); - spin_unlock_irqrestore(&ioapic_lock, flags); - return eu.entry; -} - -/* - * When we write a new IO APIC routing entry, we need to write the high - * word first! If the mask bit in the low word is clear, we will enable - * the interrupt, and we need to make sure the entry is fully populated - * before that happens. - */ -static void -__ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) -{ - union entry_union eu; - eu.entry = e; - io_apic_write(apic, 0x11 + 2*pin, eu.w2); - io_apic_write(apic, 0x10 + 2*pin, eu.w1); -} - -static void ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) -{ - unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); - __ioapic_write_entry(apic, pin, e); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -/* - * When we mask an IO APIC routing entry, we need to write the low - * word first, in order to set the mask bit before we change the - * high bits! - */ -static void ioapic_mask_entry(int apic, int pin) -{ - unsigned long flags; - union entry_union eu = { .entry.mask = 1 }; - - spin_lock_irqsave(&ioapic_lock, flags); - io_apic_write(apic, 0x10 + 2*pin, eu.w1); - io_apic_write(apic, 0x11 + 2*pin, eu.w2); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -#ifdef CONFIG_SMP -static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) -{ - int apic, pin; - struct irq_cfg *cfg; - struct irq_pin_list *entry; - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - for (;;) { - unsigned int reg; - - if (!entry) - break; - - apic = entry->apic; - pin = entry->pin; -#ifdef CONFIG_INTR_REMAP - /* - * With interrupt-remapping, destination information comes - * from interrupt-remapping table entry. - */ - if (!irq_remapped(irq)) - io_apic_write(apic, 0x11 + pin*2, dest); -#else - io_apic_write(apic, 0x11 + pin*2, dest); -#endif - reg = io_apic_read(apic, 0x10 + pin*2); - reg &= ~IO_APIC_REDIR_VECTOR_MASK; - reg |= vector; - io_apic_modify(apic, 0x10 + pin*2, reg); - if (!entry->next) - break; - entry = entry->next; - } -} - -static int assign_irq_vector(int irq, cpumask_t mask); - -static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned long flags; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - cfg = irq_cfg(irq); - if (assign_irq_vector(irq, mask)) - return; - - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - /* - * Only the high 8 bits are valid. - */ - dest = SET_APIC_LOGICAL_ID(dest); - - desc = irq_to_desc(irq); - spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); - desc->affinity = mask; - spin_unlock_irqrestore(&ioapic_lock, flags); -} -#endif /* CONFIG_SMP */ - -/* - * The common case is 1:1 IRQ<->pin mappings. Sometimes there are - * shared ISA-space IRQs, so we have to support them. We are super - * fast in the common case, and fast for shared ISA-space IRQs. - */ -static void add_pin_to_irq(unsigned int irq, int apic, int pin) -{ - struct irq_cfg *cfg; - struct irq_pin_list *entry; - - /* first time to refer irq_cfg, so with new */ - cfg = irq_cfg_alloc(irq); - entry = cfg->irq_2_pin; - if (!entry) { - entry = get_one_free_irq_2_pin(); - cfg->irq_2_pin = entry; - entry->apic = apic; - entry->pin = pin; - printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); - return; - } - - while (entry->next) { - /* not again, please */ - if (entry->apic == apic && entry->pin == pin) - return; - - entry = entry->next; - } - - entry->next = get_one_free_irq_2_pin(); - entry = entry->next; - entry->apic = apic; - entry->pin = pin; - printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); -} - -/* - * Reroute an IRQ to a different pin. - */ -static void __init replace_pin_at_irq(unsigned int irq, - int oldapic, int oldpin, - int newapic, int newpin) -{ - struct irq_cfg *cfg = irq_cfg(irq); - struct irq_pin_list *entry = cfg->irq_2_pin; - int replaced = 0; - - while (entry) { - if (entry->apic == oldapic && entry->pin == oldpin) { - entry->apic = newapic; - entry->pin = newpin; - replaced = 1; - /* every one is different, right? */ - break; - } - entry = entry->next; - } - - /* why? call replace before add? */ - if (!replaced) - add_pin_to_irq(irq, newapic, newpin); -} - -#ifdef CONFIG_X86_64 -/* - * Synchronize the IO-APIC and the CPU by doing - * a dummy read from the IO-APIC - */ -static inline void io_apic_sync(unsigned int apic) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - readl(&io_apic->data); -} - -#define __DO_ACTION(R, ACTION, FINAL) \ - \ -{ \ - int pin; \ - struct irq_cfg *cfg; \ - struct irq_pin_list *entry; \ - \ - cfg = irq_cfg(irq); \ - entry = cfg->irq_2_pin; \ - for (;;) { \ - unsigned int reg; \ - if (!entry) \ - break; \ - pin = entry->pin; \ - reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ - reg ACTION; \ - io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ - FINAL; \ - if (!entry->next) \ - break; \ - entry = entry->next; \ - } \ -} - -#define DO_ACTION(name,R,ACTION, FINAL) \ - \ - static void name##_IO_APIC_irq (unsigned int irq) \ - __DO_ACTION(R, ACTION, FINAL) - -/* mask = 1 */ -DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) - -/* mask = 0 */ -DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) - -#else - -static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) -{ - struct irq_cfg *cfg; - struct irq_pin_list *entry; - unsigned int pin, reg; - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - for (;;) { - if (!entry) - break; - pin = entry->pin; - reg = io_apic_read(entry->apic, 0x10 + pin*2); - reg &= ~disable; - reg |= enable; - io_apic_modify(entry->apic, 0x10 + pin*2, reg); - if (!entry->next) - break; - entry = entry->next; - } -} - -/* mask = 1 */ -static void __mask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, 0); -} - -/* mask = 0 */ -static void __unmask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, 0, IO_APIC_REDIR_MASKED); -} - -/* mask = 1, trigger = 0 */ -static void __mask_and_edge_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, - IO_APIC_REDIR_LEVEL_TRIGGER); -} - -/* mask = 0, trigger = 1 */ -static void __unmask_and_level_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_LEVEL_TRIGGER, - IO_APIC_REDIR_MASKED); -} - -#endif - -static void mask_IO_APIC_irq (unsigned int irq) -{ - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - __mask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -static void unmask_IO_APIC_irq (unsigned int irq) -{ - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - __unmask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); -} - -static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) -{ - struct IO_APIC_route_entry entry; - - /* Check delivery_mode to be sure we're not clearing an SMI pin */ - entry = ioapic_read_entry(apic, pin); - if (entry.delivery_mode == dest_SMI) - return; - /* - * Disable it in the IO-APIC irq-routing table: - */ - ioapic_mask_entry(apic, pin); -} - -static void clear_IO_APIC (void) -{ - int apic, pin; - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) - clear_IO_APIC_pin(apic, pin); -} - -#if !defined(CONFIG_SMP) && defined(CONFIG_X86_32) -void send_IPI_self(int vector) -{ - unsigned int cfg; - - /* - * Wait for idle. - */ - apic_wait_icr_idle(); - cfg = APIC_DM_FIXED | APIC_DEST_SELF | vector | APIC_DEST_LOGICAL; - /* - * Send the IPI. The write to APIC_ICR fires this off. - */ - apic_write(APIC_ICR, cfg); -} -#endif /* !CONFIG_SMP && CONFIG_X86_32*/ - -#ifdef CONFIG_X86_32 -/* - * support for broken MP BIOSs, enables hand-redirection of PIRQ0-7 to - * specific CPU-side IRQs. - */ - -#define MAX_PIRQS 8 -static int pirq_entries [MAX_PIRQS]; -static int pirqs_enabled; - -static int __init ioapic_pirq_setup(char *str) -{ - int i, max; - int ints[MAX_PIRQS+1]; - - get_options(str, ARRAY_SIZE(ints), ints); - - for (i = 0; i < MAX_PIRQS; i++) - pirq_entries[i] = -1; - - pirqs_enabled = 1; - apic_printk(APIC_VERBOSE, KERN_INFO - "PIRQ redirection, working around broken MP-BIOS.\n"); - max = MAX_PIRQS; - if (ints[0] < MAX_PIRQS) - max = ints[0]; - - for (i = 0; i < max; i++) { - apic_printk(APIC_VERBOSE, KERN_DEBUG - "... PIRQ%d -> IRQ %d\n", i, ints[i+1]); - /* - * PIRQs are mapped upside down, usually. - */ - pirq_entries[MAX_PIRQS-i-1] = ints[i+1]; - } - return 1; -} - -__setup("pirq=", ioapic_pirq_setup); -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_INTR_REMAP -/* I/O APIC RTE contents at the OS boot up */ -static struct IO_APIC_route_entry *early_ioapic_entries[MAX_IO_APICS]; - -/* - * Saves and masks all the unmasked IO-APIC RTE's - */ -int save_mask_IO_APIC_setup(void) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - int apic, pin; - - /* - * The number of IO-APIC IRQ registers (== #pins): - */ - for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(apic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - nr_ioapic_registers[apic] = reg_01.bits.entries+1; - } - - for (apic = 0; apic < nr_ioapics; apic++) { - early_ioapic_entries[apic] = - kzalloc(sizeof(struct IO_APIC_route_entry) * - nr_ioapic_registers[apic], GFP_KERNEL); - if (!early_ioapic_entries[apic]) - return -ENOMEM; - } - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - struct IO_APIC_route_entry entry; - - entry = early_ioapic_entries[apic][pin] = - ioapic_read_entry(apic, pin); - if (!entry.mask) { - entry.mask = 1; - ioapic_write_entry(apic, pin, entry); - } - } - return 0; -} - -void restore_IO_APIC_setup(void) -{ - int apic, pin; - - for (apic = 0; apic < nr_ioapics; apic++) - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) - ioapic_write_entry(apic, pin, - early_ioapic_entries[apic][pin]); -} - -void reinit_intr_remapped_IO_APIC(int intr_remapping) -{ - /* - * for now plain restore of previous settings. - * TBD: In the case of OS enabling interrupt-remapping, - * IO-APIC RTE's need to be setup to point to interrupt-remapping - * table entries. for now, do a plain restore, and wait for - * the setup_IO_APIC_irqs() to do proper initialization. - */ - restore_IO_APIC_setup(); -} -#endif - -/* - * Find the IRQ entry number of a certain pin. - */ -static int find_irq_entry(int apic, int pin, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == type && - (mp_irqs[i].mp_dstapic == mp_ioapics[apic].mp_apicid || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) && - mp_irqs[i].mp_dstirq == pin) - return i; - - return -1; -} - -/* - * Find the pin to which IRQ[irq] (ISA) is connected - */ -static int __init find_isa_irq_pin(int irq, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) - - return mp_irqs[i].mp_dstirq; - } - return -1; -} - -static int __init find_isa_irq_apic(int irq, int type) -{ - int i; - - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) - break; - } - if (i < mp_irq_entries) { - int apic; - for(apic = 0; apic < nr_ioapics; apic++) { - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic) - return apic; - } - } - - return -1; -} - -/* - * Find a specific PCI IRQ entry. - * Not an __init, possibly needed by modules - */ -static int pin_2_irq(int idx, int apic, int pin); - -int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) -{ - int apic, i, best_guess = -1; - - apic_printk(APIC_DEBUG, "querying PCI -> IRQ mapping bus:%d, slot:%d, pin:%d.\n", - bus, slot, pin); - if (test_bit(bus, mp_bus_not_pci)) { - apic_printk(APIC_VERBOSE, "PCI BIOS passed nonexistent PCI bus %d!\n", bus); - return -1; - } - for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; - - for (apic = 0; apic < nr_ioapics; apic++) - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) - break; - - if (!test_bit(lbus, mp_bus_not_pci) && - !mp_irqs[i].mp_irqtype && - (bus == lbus) && - (slot == ((mp_irqs[i].mp_srcbusirq >> 2) & 0x1f))) { - int irq = pin_2_irq(i,apic,mp_irqs[i].mp_dstirq); - - if (!(apic || IO_APIC_IRQ(irq))) - continue; - - if (pin == (mp_irqs[i].mp_srcbusirq & 3)) - return irq; - /* - * Use the first all-but-pin matching entry as a - * best-guess fuzzy result for broken mptables. - */ - if (best_guess < 0) - best_guess = irq; - } - } - return best_guess; -} - -EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); - -#if defined(CONFIG_EISA) || defined(CONFIG_MCA) -/* - * EISA Edge/Level control register, ELCR - */ -static int EISA_ELCR(unsigned int irq) -{ - if (irq < 16) { - unsigned int port = 0x4d0 + (irq >> 3); - return (inb(port) >> (irq & 7)) & 1; - } - apic_printk(APIC_VERBOSE, KERN_INFO - "Broken MPtable reports ISA irq %d\n", irq); - return 0; -} - -#endif - -/* ISA interrupts are always polarity zero edge triggered, - * when listed as conforming in the MP table. */ - -#define default_ISA_trigger(idx) (0) -#define default_ISA_polarity(idx) (0) - -/* EISA interrupts are always polarity zero and can be edge or level - * trigger depending on the ELCR value. If an interrupt is listed as - * EISA conforming in the MP table, that means its trigger type must - * be read in from the ELCR */ - -#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].mp_srcbusirq)) -#define default_EISA_polarity(idx) default_ISA_polarity(idx) - -/* PCI interrupts are always polarity one level triggered, - * when listed as conforming in the MP table. */ - -#define default_PCI_trigger(idx) (1) -#define default_PCI_polarity(idx) (1) - -/* MCA interrupts are always polarity zero level triggered, - * when listed as conforming in the MP table. */ - -#define default_MCA_trigger(idx) (1) -#define default_MCA_polarity(idx) default_ISA_polarity(idx) - -static int MPBIOS_polarity(int idx) -{ - int bus = mp_irqs[idx].mp_srcbus; - int polarity; - - /* - * Determine IRQ line polarity (high active or low active): - */ - switch (mp_irqs[idx].mp_irqflag & 3) - { - case 0: /* conforms, ie. bus-type dependent polarity */ - if (test_bit(bus, mp_bus_not_pci)) - polarity = default_ISA_polarity(idx); - else - polarity = default_PCI_polarity(idx); - break; - case 1: /* high active */ - { - polarity = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } - case 3: /* low active */ - { - polarity = 1; - break; - } - default: /* invalid */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - polarity = 1; - break; - } - } - return polarity; -} - -static int MPBIOS_trigger(int idx) -{ - int bus = mp_irqs[idx].mp_srcbus; - int trigger; - - /* - * Determine IRQ trigger mode (edge or level sensitive): - */ - switch ((mp_irqs[idx].mp_irqflag>>2) & 3) - { - case 0: /* conforms, ie. bus-type dependent */ - if (test_bit(bus, mp_bus_not_pci)) - trigger = default_ISA_trigger(idx); - else - trigger = default_PCI_trigger(idx); -#if defined(CONFIG_EISA) || defined(CONFIG_MCA) - switch (mp_bus_id_to_type[bus]) { - case MP_BUS_ISA: /* ISA pin */ - { - /* set before the switch */ - break; - } - case MP_BUS_EISA: /* EISA pin */ - { - trigger = default_EISA_trigger(idx); - break; - } - case MP_BUS_PCI: /* PCI pin */ - { - /* set before the switch */ - break; - } - case MP_BUS_MCA: /* MCA pin */ - { - trigger = default_MCA_trigger(idx); - break; - } - default: - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; - break; - } - } -#endif - break; - case 1: /* edge */ - { - trigger = 0; - break; - } - case 2: /* reserved */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 1; - break; - } - case 3: /* level */ - { - trigger = 1; - break; - } - default: /* invalid */ - { - printk(KERN_WARNING "broken BIOS!!\n"); - trigger = 0; - break; - } - } - return trigger; -} - -static inline int irq_polarity(int idx) -{ - return MPBIOS_polarity(idx); -} - -static inline int irq_trigger(int idx) -{ - return MPBIOS_trigger(idx); -} - -int (*ioapic_renumber_irq)(int ioapic, int irq); -static int pin_2_irq(int idx, int apic, int pin) -{ - int irq, i; - int bus = mp_irqs[idx].mp_srcbus; - - /* - * Debugging check, we are in big trouble if this message pops up! - */ - if (mp_irqs[idx].mp_dstirq != pin) - printk(KERN_ERR "broken BIOS or MPTABLE parser, ayiee!!\n"); - - if (test_bit(bus, mp_bus_not_pci)) { - irq = mp_irqs[idx].mp_srcbusirq; - } else { - /* - * PCI IRQs are mapped in order - */ - i = irq = 0; - while (i < apic) - irq += nr_ioapic_registers[i++]; - irq += pin; - /* - * For MPS mode, so far only needed by ES7000 platform - */ - if (ioapic_renumber_irq) - irq = ioapic_renumber_irq(apic, irq); - } - -#ifdef CONFIG_X86_32 - /* - * PCI IRQ command line redirection. Yes, limits are hardcoded. - */ - if ((pin >= 16) && (pin <= 23)) { - if (pirq_entries[pin-16] != -1) { - if (!pirq_entries[pin-16]) { - apic_printk(APIC_VERBOSE, KERN_DEBUG - "disabling PIRQ%d\n", pin-16); - } else { - irq = pirq_entries[pin-16]; - apic_printk(APIC_VERBOSE, KERN_DEBUG - "using PIRQ%d -> IRQ %d\n", - pin-16, irq); - } - } - } -#endif - - return irq; -} - -void lock_vector_lock(void) -{ - /* Used to the online set of cpus does not change - * during assign_irq_vector. - */ - spin_lock(&vector_lock); -} - -void unlock_vector_lock(void) -{ - spin_unlock(&vector_lock); -} - -static int __assign_irq_vector(int irq, cpumask_t mask) -{ - /* - * NOTE! The local APIC isn't very good at handling - * multiple interrupts at the same interrupt level. - * As the interrupt level is determined by taking the - * vector number and shifting that right by 4, we - * want to spread these out a bit so that they don't - * all fall in the same interrupt level. - * - * Also, we've got to be careful not to trash gate - * 0x80, because int 0x80 is hm, kind of importantish. ;) - */ - static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; - unsigned int old_vector; - int cpu; - struct irq_cfg *cfg; - - cfg = irq_cfg(irq); - - /* Only try and allocate irqs on cpus that are present */ - cpus_and(mask, mask, cpu_online_map); - - if ((cfg->move_in_progress) || cfg->move_cleanup_count) - return -EBUSY; - - old_vector = cfg->vector; - if (old_vector) { - cpumask_t tmp; - cpus_and(tmp, cfg->domain, mask); - if (!cpus_empty(tmp)) - return 0; - } - - for_each_cpu_mask_nr(cpu, mask) { - cpumask_t domain, new_mask; - int new_cpu; - int vector, offset; - - domain = vector_allocation_domain(cpu); - cpus_and(new_mask, domain, cpu_online_map); - - vector = current_vector; - offset = current_offset; -next: - vector += 8; - if (vector >= first_system_vector) { - /* If we run out of vectors on large boxen, must share them. */ - offset = (offset + 1) % 8; - vector = FIRST_DEVICE_VECTOR + offset; - } - if (unlikely(current_vector == vector)) - continue; -#ifdef CONFIG_X86_64 - if (vector == IA32_SYSCALL_VECTOR) - goto next; -#else - if (vector == SYSCALL_VECTOR) - goto next; -#endif - for_each_cpu_mask_nr(new_cpu, new_mask) - if (per_cpu(vector_irq, new_cpu)[vector] != -1) - goto next; - /* Found one! */ - current_vector = vector; - current_offset = offset; - if (old_vector) { - cfg->move_in_progress = 1; - cfg->old_domain = cfg->domain; - } - for_each_cpu_mask_nr(new_cpu, new_mask) - per_cpu(vector_irq, new_cpu)[vector] = irq; - cfg->vector = vector; - cfg->domain = domain; - return 0; - } - return -ENOSPC; -} - -static int assign_irq_vector(int irq, cpumask_t mask) -{ - int err; - unsigned long flags; - - spin_lock_irqsave(&vector_lock, flags); - err = __assign_irq_vector(irq, mask); - spin_unlock_irqrestore(&vector_lock, flags); - return err; -} - -static void __clear_irq_vector(int irq) -{ - struct irq_cfg *cfg; - cpumask_t mask; - int cpu, vector; - - cfg = irq_cfg(irq); - BUG_ON(!cfg->vector); - - vector = cfg->vector; - cpus_and(mask, cfg->domain, cpu_online_map); - for_each_cpu_mask_nr(cpu, mask) - per_cpu(vector_irq, cpu)[vector] = -1; - - cfg->vector = 0; - cpus_clear(cfg->domain); -} - -void __setup_vector_irq(int cpu) -{ - /* Initialize vector_irq on a new cpu */ - /* This function must be called with vector_lock held */ - int irq, vector; - struct irq_cfg *cfg; - - /* Mark the inuse vectors */ - for_each_irq_cfg(cfg) { - if (!cpu_isset(cpu, cfg->domain)) - continue; - vector = cfg->vector; - irq = cfg->irq; - per_cpu(vector_irq, cpu)[vector] = irq; - } - /* Mark the free vectors */ - for (vector = 0; vector < NR_VECTORS; ++vector) { - irq = per_cpu(vector_irq, cpu)[vector]; - if (irq < 0) - continue; - - cfg = irq_cfg(irq); - if (!cpu_isset(cpu, cfg->domain)) - per_cpu(vector_irq, cpu)[vector] = -1; - } -} - -static struct irq_chip ioapic_chip; -#ifdef CONFIG_INTR_REMAP -static struct irq_chip ir_ioapic_chip; -#endif - -#define IOAPIC_AUTO -1 -#define IOAPIC_EDGE 0 -#define IOAPIC_LEVEL 1 - -#ifdef CONFIG_X86_32 -static inline int IO_APIC_irq_trigger(int irq) -{ - int apic, idx, pin; - - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - idx = find_irq_entry(apic, pin, mp_INT); - if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) - return irq_trigger(idx); - } - } - /* - * nonexistent IRQs are edge default - */ - return 0; -} -#else -static inline int IO_APIC_irq_trigger(int irq) -{ - return 1; -} -#endif - -static void ioapic_register_intr(int irq, unsigned long trigger) -{ - struct irq_desc *desc; - - /* first time to use this irq_desc */ - if (irq < 16) - desc = irq_to_desc(irq); - else - desc = irq_to_desc_alloc(irq); - - if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || - trigger == IOAPIC_LEVEL) - desc->status |= IRQ_LEVEL; - else - desc->status &= ~IRQ_LEVEL; - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - desc->status |= IRQ_MOVE_PCNTXT; - if (trigger) - set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, - handle_fasteoi_irq, - "fasteoi"); - else - set_irq_chip_and_handler_name(irq, &ir_ioapic_chip, - handle_edge_irq, "edge"); - return; - } -#endif - if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || - trigger == IOAPIC_LEVEL) - set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_fasteoi_irq, - "fasteoi"); - else - set_irq_chip_and_handler_name(irq, &ioapic_chip, - handle_edge_irq, "edge"); -} - -static int setup_ioapic_entry(int apic, int irq, - struct IO_APIC_route_entry *entry, - unsigned int destination, int trigger, - int polarity, int vector) -{ - /* - * add it to the IO-APIC irq-routing table: - */ - memset(entry,0,sizeof(*entry)); - -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) { - struct intel_iommu *iommu = map_ioapic_to_ir(apic); - struct irte irte; - struct IR_IO_APIC_route_entry *ir_entry = - (struct IR_IO_APIC_route_entry *) entry; - int index; - - if (!iommu) - panic("No mapping iommu for ioapic %d\n", apic); - - index = alloc_irte(iommu, irq, 1); - if (index < 0) - panic("Failed to allocate IRTE for ioapic %d\n", apic); - - memset(&irte, 0, sizeof(irte)); - - irte.present = 1; - irte.dst_mode = INT_DEST_MODE; - irte.trigger_mode = trigger; - irte.dlvry_mode = INT_DELIVERY_MODE; - irte.vector = vector; - irte.dest_id = IRTE_DEST(destination); - - modify_irte(irq, &irte); - - ir_entry->index2 = (index >> 15) & 0x1; - ir_entry->zero = 0; - ir_entry->format = 1; - ir_entry->index = (index & 0x7fff); - } else -#endif - { - entry->delivery_mode = INT_DELIVERY_MODE; - entry->dest_mode = INT_DEST_MODE; - entry->dest = destination; - } - - entry->mask = 0; /* enable IRQ */ - entry->trigger = trigger; - entry->polarity = polarity; - entry->vector = vector; - - /* Mask level triggered irqs. - * Use IRQ_DELAYED_DISABLE for edge triggered irqs. - */ - if (trigger) - entry->mask = 1; - return 0; -} - -static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, - int trigger, int polarity) -{ - struct irq_cfg *cfg; - struct IO_APIC_route_entry entry; - cpumask_t mask; - - if (!IO_APIC_IRQ(irq)) - return; - - cfg = irq_cfg(irq); - - mask = TARGET_CPUS; - if (assign_irq_vector(irq, mask)) - return; - - cpus_and(mask, cfg->domain, mask); - - apic_printk(APIC_VERBOSE,KERN_DEBUG - "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " - "IRQ %d Mode:%i Active:%i)\n", - apic, mp_ioapics[apic].mp_apicid, pin, cfg->vector, - irq, trigger, polarity); - - - if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, - cpu_mask_to_apicid(mask), trigger, polarity, - cfg->vector)) { - printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", - mp_ioapics[apic].mp_apicid, pin); - __clear_irq_vector(irq); - return; - } - - ioapic_register_intr(irq, trigger); - if (irq < 16) - disable_8259A_irq(irq); - - ioapic_write_entry(apic, pin, entry); -} - -static void __init setup_IO_APIC_irqs(void) -{ - int apic, pin, idx, irq, first_notcon = 1; - - apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); - - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - - idx = find_irq_entry(apic,pin,mp_INT); - if (idx == -1) { - if (first_notcon) { - apic_printk(APIC_VERBOSE, KERN_DEBUG " IO-APIC (apicid-pin) %d-%d", mp_ioapics[apic].mp_apicid, pin); - first_notcon = 0; - } else - apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mp_apicid, pin); - continue; - } - if (!first_notcon) { - apic_printk(APIC_VERBOSE, " not connected.\n"); - first_notcon = 1; - } - - irq = pin_2_irq(idx, apic, pin); -#ifdef CONFIG_X86_32 - if (multi_timer_check(apic, irq)) - continue; -#endif - add_pin_to_irq(irq, apic, pin); - - setup_IO_APIC_irq(apic, pin, irq, - irq_trigger(idx), irq_polarity(idx)); - } - } - - if (!first_notcon) - apic_printk(APIC_VERBOSE, " not connected.\n"); -} - -/* - * Set up the timer pin, possibly with the 8259A-master behind. - */ -static void __init setup_timer_IRQ0_pin(unsigned int apic, unsigned int pin, - int vector) -{ - struct IO_APIC_route_entry entry; - -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - return; -#endif - - memset(&entry, 0, sizeof(entry)); - - /* - * We use logical delivery to get the timer IRQ - * to the first CPU. - */ - entry.dest_mode = INT_DEST_MODE; - entry.mask = 1; /* mask IRQ now */ - entry.dest = cpu_mask_to_apicid(TARGET_CPUS); - entry.delivery_mode = INT_DELIVERY_MODE; - entry.polarity = 0; - entry.trigger = 0; - entry.vector = vector; - - /* - * The timer IRQ doesn't have to know that behind the - * scene we may have a 8259A-master in AEOI mode ... - */ - set_irq_chip_and_handler_name(0, &ioapic_chip, handle_edge_irq, "edge"); - - /* - * Add it to the IO-APIC irq-routing table: - */ - ioapic_write_entry(apic, pin, entry); -} - - -__apicdebuginit(void) print_IO_APIC(void) -{ - int apic, i; - union IO_APIC_reg_00 reg_00; - union IO_APIC_reg_01 reg_01; - union IO_APIC_reg_02 reg_02; - union IO_APIC_reg_03 reg_03; - unsigned long flags; - struct irq_cfg *cfg; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "number of MP IRQ sources: %d.\n", mp_irq_entries); - for (i = 0; i < nr_ioapics; i++) - printk(KERN_DEBUG "number of IO-APIC #%d registers: %d.\n", - mp_ioapics[i].mp_apicid, nr_ioapic_registers[i]); - - /* - * We are a bit conservative about what we expect. We have to - * know about every hardware change ASAP. - */ - printk(KERN_INFO "testing the IO APIC.......................\n"); - - for (apic = 0; apic < nr_ioapics; apic++) { - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - reg_01.raw = io_apic_read(apic, 1); - if (reg_01.bits.version >= 0x10) - reg_02.raw = io_apic_read(apic, 2); - if (reg_01.bits.version >= 0x20) - reg_03.raw = io_apic_read(apic, 3); - spin_unlock_irqrestore(&ioapic_lock, flags); - - printk("\n"); - printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); - printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); - printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); - printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); - printk(KERN_DEBUG "....... : LTS : %X\n", reg_00.bits.LTS); - - printk(KERN_DEBUG ".... register #01: %08X\n", *(int *)®_01); - printk(KERN_DEBUG "....... : max redirection entries: %04X\n", reg_01.bits.entries); - - printk(KERN_DEBUG "....... : PRQ implemented: %X\n", reg_01.bits.PRQ); - printk(KERN_DEBUG "....... : IO APIC version: %04X\n", reg_01.bits.version); - - /* - * Some Intel chipsets with IO APIC VERSION of 0x1? don't have reg_02, - * but the value of reg_02 is read as the previous read register - * value, so ignore it if reg_02 == reg_01. - */ - if (reg_01.bits.version >= 0x10 && reg_02.raw != reg_01.raw) { - printk(KERN_DEBUG ".... register #02: %08X\n", reg_02.raw); - printk(KERN_DEBUG "....... : arbitration: %02X\n", reg_02.bits.arbitration); - } - - /* - * Some Intel chipsets with IO APIC VERSION of 0x2? don't have reg_02 - * or reg_03, but the value of reg_0[23] is read as the previous read - * register value, so ignore it if reg_03 == reg_0[12]. - */ - if (reg_01.bits.version >= 0x20 && reg_03.raw != reg_02.raw && - reg_03.raw != reg_01.raw) { - printk(KERN_DEBUG ".... register #03: %08X\n", reg_03.raw); - printk(KERN_DEBUG "....... : Boot DT : %X\n", reg_03.bits.boot_DT); - } - - printk(KERN_DEBUG ".... IRQ redirection table:\n"); - - printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" - " Stat Dmod Deli Vect: \n"); - - for (i = 0; i <= reg_01.bits.entries; i++) { - struct IO_APIC_route_entry entry; - - entry = ioapic_read_entry(apic, i); - - printk(KERN_DEBUG " %02x %03X ", - i, - entry.dest - ); - - printk("%1d %1d %1d %1d %1d %1d %1d %02X\n", - entry.mask, - entry.trigger, - entry.irr, - entry.polarity, - entry.delivery_status, - entry.dest_mode, - entry.delivery_mode, - entry.vector - ); - } - } - printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for_each_irq_cfg(cfg) { - struct irq_pin_list *entry = cfg->irq_2_pin; - if (!entry) - continue; - printk(KERN_DEBUG "IRQ%d ", cfg->irq); - for (;;) { - printk("-> %d:%d", entry->apic, entry->pin); - if (!entry->next) - break; - entry = entry->next; - } - printk("\n"); - } - - printk(KERN_INFO ".................................... done.\n"); - - return; -} - -__apicdebuginit(void) print_APIC_bitfield(int base) -{ - unsigned int v; - int i, j; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "0123456789abcdef0123456789abcdef\n" KERN_DEBUG); - for (i = 0; i < 8; i++) { - v = apic_read(base + i*0x10); - for (j = 0; j < 32; j++) { - if (v & (1< 3) /* Due to the Pentium erratum 3AP. */ - apic_write(APIC_ESR, 0); - - v = apic_read(APIC_ESR); - printk(KERN_DEBUG "... APIC ESR: %08x\n", v); - } - - icr = apic_icr_read(); - printk(KERN_DEBUG "... APIC ICR: %08x\n", (u32)icr); - printk(KERN_DEBUG "... APIC ICR2: %08x\n", (u32)(icr >> 32)); - - v = apic_read(APIC_LVTT); - printk(KERN_DEBUG "... APIC LVTT: %08x\n", v); - - if (maxlvt > 3) { /* PC is LVT#4. */ - v = apic_read(APIC_LVTPC); - printk(KERN_DEBUG "... APIC LVTPC: %08x\n", v); - } - v = apic_read(APIC_LVT0); - printk(KERN_DEBUG "... APIC LVT0: %08x\n", v); - v = apic_read(APIC_LVT1); - printk(KERN_DEBUG "... APIC LVT1: %08x\n", v); - - if (maxlvt > 2) { /* ERR is LVT#3. */ - v = apic_read(APIC_LVTERR); - printk(KERN_DEBUG "... APIC LVTERR: %08x\n", v); - } - - v = apic_read(APIC_TMICT); - printk(KERN_DEBUG "... APIC TMICT: %08x\n", v); - v = apic_read(APIC_TMCCT); - printk(KERN_DEBUG "... APIC TMCCT: %08x\n", v); - v = apic_read(APIC_TDCR); - printk(KERN_DEBUG "... APIC TDCR: %08x\n", v); - printk("\n"); -} - -__apicdebuginit(void) print_all_local_APICs(void) -{ - on_each_cpu(print_local_APIC, NULL, 1); -} - -__apicdebuginit(void) print_PIC(void) -{ - unsigned int v; - unsigned long flags; - - if (apic_verbosity == APIC_QUIET) - return; - - printk(KERN_DEBUG "\nprinting PIC contents\n"); - - spin_lock_irqsave(&i8259A_lock, flags); - - v = inb(0xa1) << 8 | inb(0x21); - printk(KERN_DEBUG "... PIC IMR: %04x\n", v); - - v = inb(0xa0) << 8 | inb(0x20); - printk(KERN_DEBUG "... PIC IRR: %04x\n", v); - - outb(0x0b,0xa0); - outb(0x0b,0x20); - v = inb(0xa0) << 8 | inb(0x20); - outb(0x0a,0xa0); - outb(0x0a,0x20); - - spin_unlock_irqrestore(&i8259A_lock, flags); - - printk(KERN_DEBUG "... PIC ISR: %04x\n", v); - - v = inb(0x4d1) << 8 | inb(0x4d0); - printk(KERN_DEBUG "... PIC ELCR: %04x\n", v); -} - -__apicdebuginit(int) print_all_ICs(void) -{ - print_PIC(); - print_all_local_APICs(); - print_IO_APIC(); - - return 0; -} - -fs_initcall(print_all_ICs); - - -/* Where if anywhere is the i8259 connect in external int mode */ -static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; - -void __init enable_IO_APIC(void) -{ - union IO_APIC_reg_01 reg_01; - int i8259_apic, i8259_pin; - int apic; - unsigned long flags; - -#ifdef CONFIG_X86_32 - int i; - if (!pirqs_enabled) - for (i = 0; i < MAX_PIRQS; i++) - pirq_entries[i] = -1; -#endif - - /* - * The number of IO-APIC IRQ registers (== #pins): - */ - for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(apic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - nr_ioapic_registers[apic] = reg_01.bits.entries+1; - } - for(apic = 0; apic < nr_ioapics; apic++) { - int pin; - /* See if any of the pins is in ExtINT mode */ - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - struct IO_APIC_route_entry entry; - entry = ioapic_read_entry(apic, pin); - - /* If the interrupt line is enabled and in ExtInt mode - * I have found the pin where the i8259 is connected. - */ - if ((entry.mask == 0) && (entry.delivery_mode == dest_ExtINT)) { - ioapic_i8259.apic = apic; - ioapic_i8259.pin = pin; - goto found_i8259; - } - } - } - found_i8259: - /* Look to see what if the MP table has reported the ExtINT */ - /* If we could not find the appropriate pin by looking at the ioapic - * the i8259 probably is not connected the ioapic but give the - * mptable a chance anyway. - */ - i8259_pin = find_isa_irq_pin(0, mp_ExtINT); - i8259_apic = find_isa_irq_apic(0, mp_ExtINT); - /* Trust the MP table if nothing is setup in the hardware */ - if ((ioapic_i8259.pin == -1) && (i8259_pin >= 0)) { - printk(KERN_WARNING "ExtINT not setup in hardware but reported by MP table\n"); - ioapic_i8259.pin = i8259_pin; - ioapic_i8259.apic = i8259_apic; - } - /* Complain if the MP table and the hardware disagree */ - if (((ioapic_i8259.apic != i8259_apic) || (ioapic_i8259.pin != i8259_pin)) && - (i8259_pin >= 0) && (ioapic_i8259.pin >= 0)) - { - printk(KERN_WARNING "ExtINT in hardware and MP table differ\n"); - } - - /* - * Do not trust the IO-APIC being empty at bootup - */ - clear_IO_APIC(); -} - -/* - * Not an __init, needed by the reboot code - */ -void disable_IO_APIC(void) -{ - /* - * Clear the IO-APIC before rebooting: - */ - clear_IO_APIC(); - - /* - * If the i8259 is routed through an IOAPIC - * Put that IOAPIC in virtual wire mode - * so legacy interrupts can be delivered. - */ - if (ioapic_i8259.pin != -1) { - struct IO_APIC_route_entry entry; - - memset(&entry, 0, sizeof(entry)); - entry.mask = 0; /* Enabled */ - entry.trigger = 0; /* Edge */ - entry.irr = 0; - entry.polarity = 0; /* High */ - entry.delivery_status = 0; - entry.dest_mode = 0; /* Physical */ - entry.delivery_mode = dest_ExtINT; /* ExtInt */ - entry.vector = 0; - entry.dest = read_apic_id(); - - /* - * Add it to the IO-APIC irq-routing table: - */ - ioapic_write_entry(ioapic_i8259.apic, ioapic_i8259.pin, entry); - } - - disconnect_bsp_APIC(ioapic_i8259.pin != -1); -} - -#ifdef CONFIG_X86_32 -/* - * function to set the IO-APIC physical IDs based on the - * values stored in the MPC table. - * - * by Matt Domsch Tue Dec 21 12:25:05 CST 1999 - */ - -static void __init setup_ioapic_ids_from_mpc(void) -{ - union IO_APIC_reg_00 reg_00; - physid_mask_t phys_id_present_map; - int apic; - int i; - unsigned char old_id; - unsigned long flags; - - if (x86_quirks->setup_ioapic_ids && x86_quirks->setup_ioapic_ids()) - return; - - /* - * Don't check I/O APIC IDs for xAPIC systems. They have - * no meaning without the serial APIC bus. - */ - if (!(boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) - || APIC_XAPIC(apic_version[boot_cpu_physical_apicid])) - return; - /* - * This is broken; anything with a real cpu count has to - * circumvent this idiocy regardless. - */ - phys_id_present_map = ioapic_phys_id_map(phys_cpu_present_map); - - /* - * Set the IOAPIC ID to the value stored in the MPC table. - */ - for (apic = 0; apic < nr_ioapics; apic++) { - - /* Read the register 0 value */ - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - old_id = mp_ioapics[apic].mp_apicid; - - if (mp_ioapics[apic].mp_apicid >= get_physical_broadcast()) { - printk(KERN_ERR "BIOS bug, IO-APIC#%d ID is %d in the MPC table!...\n", - apic, mp_ioapics[apic].mp_apicid); - printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", - reg_00.bits.ID); - mp_ioapics[apic].mp_apicid = reg_00.bits.ID; - } - - /* - * Sanity check, is the ID really free? Every APIC in a - * system must have a unique ID or we get lots of nice - * 'stuck on smp_invalidate_needed IPI wait' messages. - */ - if (check_apicid_used(phys_id_present_map, - mp_ioapics[apic].mp_apicid)) { - printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n", - apic, mp_ioapics[apic].mp_apicid); - for (i = 0; i < get_physical_broadcast(); i++) - if (!physid_isset(i, phys_id_present_map)) - break; - if (i >= get_physical_broadcast()) - panic("Max APIC ID exceeded!\n"); - printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", - i); - physid_set(i, phys_id_present_map); - mp_ioapics[apic].mp_apicid = i; - } else { - physid_mask_t tmp; - tmp = apicid_to_cpu_present(mp_ioapics[apic].mp_apicid); - apic_printk(APIC_VERBOSE, "Setting %d in the " - "phys_id_present_map\n", - mp_ioapics[apic].mp_apicid); - physids_or(phys_id_present_map, phys_id_present_map, tmp); - } - - - /* - * We need to adjust the IRQ routing table - * if the ID changed. - */ - if (old_id != mp_ioapics[apic].mp_apicid) - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_dstapic == old_id) - mp_irqs[i].mp_dstapic - = mp_ioapics[apic].mp_apicid; - - /* - * Read the right value from the MPC table and - * write it into the ID register. - */ - apic_printk(APIC_VERBOSE, KERN_INFO - "...changing IO-APIC physical APIC ID to %d ...", - mp_ioapics[apic].mp_apicid); - - reg_00.bits.ID = mp_ioapics[apic].mp_apicid; - spin_lock_irqsave(&ioapic_lock, flags); - - /* - * Sanity check - */ - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(apic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - if (reg_00.bits.ID != mp_ioapics[apic].mp_apicid) - printk("could not set ID!\n"); - else - apic_printk(APIC_VERBOSE, " ok.\n"); - } -} -#endif - -int no_timer_check __initdata; - -static int __init notimercheck(char *s) -{ - no_timer_check = 1; - return 1; -} -__setup("no_timer_check", notimercheck); - -/* - * There is a nasty bug in some older SMP boards, their mptable lies - * about the timer IRQ. We do the following to work around the situation: - * - * - timer IRQ defaults to IO-APIC IRQ - * - if this function detects that timer IRQs are defunct, then we fall - * back to ISA timer IRQs - */ -static int __init timer_irq_works(void) -{ - unsigned long t1 = jiffies; - unsigned long flags; - - if (no_timer_check) - return 1; - - local_save_flags(flags); - local_irq_enable(); - /* Let ten ticks pass... */ - mdelay((10 * 1000) / HZ); - local_irq_restore(flags); - - /* - * Expect a few ticks at least, to be sure some possible - * glue logic does not lock up after one or two first - * ticks in a non-ExtINT mode. Also the local APIC - * might have cached one ExtINT interrupt. Finally, at - * least one tick may be lost due to delays. - */ - - /* jiffies wrap? */ - if (time_after(jiffies, t1 + 4)) - return 1; - return 0; -} - -/* - * In the SMP+IOAPIC case it might happen that there are an unspecified - * number of pending IRQ events unhandled. These cases are very rare, - * so we 'resend' these IRQs via IPIs, to the same CPU. It's much - * better to do it this way as thus we do not have to be aware of - * 'pending' interrupts in the IRQ path, except at this point. - */ -/* - * Edge triggered needs to resend any interrupt - * that was delayed but this is now handled in the device - * independent code. - */ - -/* - * Starting up a edge-triggered IO-APIC interrupt is - * nasty - we need to make sure that we get the edge. - * If it is already asserted for some reason, we need - * return 1 to indicate that is was pending. - * - * This is not complete - we should be able to fake - * an edge even if it isn't on the 8259A... - */ - -static unsigned int startup_ioapic_irq(unsigned int irq) -{ - int was_pending = 0; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - if (irq < 16) { - disable_8259A_irq(irq); - if (i8259A_irq_pending(irq)) - was_pending = 1; - } - __unmask_IO_APIC_irq(irq); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return was_pending; -} - -#ifdef CONFIG_X86_64 -static int ioapic_retrigger_irq(unsigned int irq) -{ - - struct irq_cfg *cfg = irq_cfg(irq); - unsigned long flags; - - spin_lock_irqsave(&vector_lock, flags); - send_IPI_mask(cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); - spin_unlock_irqrestore(&vector_lock, flags); - - return 1; -} -#else -static int ioapic_retrigger_irq(unsigned int irq) -{ - send_IPI_self(irq_cfg(irq)->vector); - - return 1; -} -#endif - -/* - * Level and edge triggered IO-APIC interrupts need different handling, - * so we use two separate IRQ descriptors. Edge triggered IRQs can be - * handled with the level-triggered descriptor, but that one has slightly - * more overhead. Level-triggered interrupts cannot be handled with the - * edge-triggered handler, without risking IRQ storms and other ugly - * races. - */ - -#ifdef CONFIG_SMP - -#ifdef CONFIG_INTR_REMAP -static void ir_irq_migration(struct work_struct *work); - -static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); - -/* - * Migrate the IO-APIC irq in the presence of intr-remapping. - * - * For edge triggered, irq migration is a simple atomic update(of vector - * and cpu destination) of IRTE and flush the hardware cache. - * - * For level triggered, we need to modify the io-apic RTE aswell with the update - * vector information, along with modifying IRTE with vector and destination. - * So irq migration for level triggered is little bit more complex compared to - * edge triggered migration. But the good news is, we use the same algorithm - * for level triggered migration as we have today, only difference being, - * we now initiate the irq migration from process context instead of the - * interrupt context. - * - * In future, when we do a directed EOI (combined with cpu EOI broadcast - * suppression) to the IO-APIC, level triggered irq migration will also be - * as simple as edge triggered migration and we can do the irq migration - * with a simple atomic update to IO-APIC RTE. - */ -static void migrate_ioapic_irq(int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct irq_desc *desc; - cpumask_t tmp, cleanup_mask; - struct irte irte; - int modify_ioapic_rte; - unsigned int dest; - unsigned long flags; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (get_irte(irq, &irte)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - desc = irq_to_desc(irq); - modify_ioapic_rte = desc->status & IRQ_LEVEL; - if (modify_ioapic_rte) { - spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); - spin_unlock_irqrestore(&ioapic_lock, flags); - } - - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - /* - * Modified the IRTE and flushes the Interrupt entry cache. - */ - modify_irte(irq, &irte); - - if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } - - desc->affinity = mask; -} - -static int migrate_irq_remapped_level(int irq) -{ - int ret = -1; - struct irq_desc *desc = irq_to_desc(irq); - - mask_IO_APIC_irq(irq); - - if (io_apic_level_ack_pending(irq)) { - /* - * Interrupt in progress. Migrating irq now will change the - * vector information in the IO-APIC RTE and that will confuse - * the EOI broadcast performed by cpu. - * So, delay the irq migration to the next instance. - */ - schedule_delayed_work(&ir_migration_work, 1); - goto unmask; - } - - /* everthing is clear. we have right of way */ - migrate_ioapic_irq(irq, desc->pending_mask); - - ret = 0; - desc->status &= ~IRQ_MOVE_PENDING; - cpus_clear(desc->pending_mask); - -unmask: - unmask_IO_APIC_irq(irq); - return ret; -} - -static void ir_irq_migration(struct work_struct *work) -{ - unsigned int irq; - struct irq_desc *desc; - - for_each_irq_desc(irq, desc) { - if (desc->status & IRQ_MOVE_PENDING) { - unsigned long flags; - - spin_lock_irqsave(&desc->lock, flags); - if (!desc->chip->set_affinity || - !(desc->status & IRQ_MOVE_PENDING)) { - desc->status &= ~IRQ_MOVE_PENDING; - spin_unlock_irqrestore(&desc->lock, flags); - continue; - } - - desc->chip->set_affinity(irq, desc->pending_mask); - spin_unlock_irqrestore(&desc->lock, flags); - } - } -} - -/* - * Migrates the IRQ destination in the process context. - */ -static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) -{ - struct irq_desc *desc = irq_to_desc(irq); - - if (desc->status & IRQ_LEVEL) { - desc->status |= IRQ_MOVE_PENDING; - desc->pending_mask = mask; - migrate_irq_remapped_level(irq); - return; - } - - migrate_ioapic_irq(irq, mask); -} -#endif - -asmlinkage void smp_irq_move_cleanup_interrupt(void) -{ - unsigned vector, me; - ack_APIC_irq(); -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - - me = smp_processor_id(); - for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { - unsigned int irq; - struct irq_desc *desc; - struct irq_cfg *cfg; - irq = __get_cpu_var(vector_irq)[vector]; - - desc = irq_to_desc(irq); - if (!desc) - continue; - - cfg = irq_cfg(irq); - spin_lock(&desc->lock); - if (!cfg->move_cleanup_count) - goto unlock; - - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) - goto unlock; - - __get_cpu_var(vector_irq)[vector] = -1; - cfg->move_cleanup_count--; -unlock: - spin_unlock(&desc->lock); - } - - irq_exit(); -} - -static void irq_complete_move(unsigned int irq) -{ - struct irq_cfg *cfg = irq_cfg(irq); - unsigned vector, me; - - if (likely(!cfg->move_in_progress)) - return; - - vector = ~get_irq_regs()->orig_ax; - me = smp_processor_id(); - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { - cpumask_t cleanup_mask; - - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } -} -#else -static inline void irq_complete_move(unsigned int irq) {} -#endif -#ifdef CONFIG_INTR_REMAP -static void ack_x2apic_level(unsigned int irq) -{ - ack_x2APIC_irq(); -} - -static void ack_x2apic_edge(unsigned int irq) -{ - ack_x2APIC_irq(); -} -#endif - -static void ack_apic_edge(unsigned int irq) -{ - irq_complete_move(irq); - move_native_irq(irq); - ack_APIC_irq(); -} - -#ifdef CONFIG_X86_64 -static void ack_apic_level(unsigned int irq) -{ - int do_unmask_irq = 0; - - irq_complete_move(irq); -#ifdef CONFIG_GENERIC_PENDING_IRQ - /* If we are moving the irq we need to mask it */ - if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { - do_unmask_irq = 1; - mask_IO_APIC_irq(irq); - } -#endif - - /* - * We must acknowledge the irq before we move it or the acknowledge will - * not propagate properly. - */ - ack_APIC_irq(); - - /* Now we can move and renable the irq */ - if (unlikely(do_unmask_irq)) { - /* Only migrate the irq if the ack has been received. - * - * On rare occasions the broadcast level triggered ack gets - * delayed going to ioapics, and if we reprogram the - * vector while Remote IRR is still set the irq will never - * fire again. - * - * To prevent this scenario we read the Remote IRR bit - * of the ioapic. This has two effects. - * - On any sane system the read of the ioapic will - * flush writes (and acks) going to the ioapic from - * this cpu. - * - We get to see if the ACK has actually been delivered. - * - * Based on failed experiments of reprogramming the - * ioapic entry from outside of irq context starting - * with masking the ioapic entry and then polling until - * Remote IRR was clear before reprogramming the - * ioapic I don't trust the Remote IRR bit to be - * completey accurate. - * - * However there appears to be no other way to plug - * this race, so if the Remote IRR bit is not - * accurate and is causing problems then it is a hardware bug - * and you can go talk to the chipset vendor about it. - */ - if (!io_apic_level_ack_pending(irq)) - move_masked_irq(irq); - unmask_IO_APIC_irq(irq); - } -} -#else -atomic_t irq_mis_count; -static void ack_apic_level(unsigned int irq) -{ - unsigned long v; - int i; - - irq_complete_move(irq); - move_native_irq(irq); - /* - * It appears there is an erratum which affects at least version 0x11 - * of I/O APIC (that's the 82093AA and cores integrated into various - * chipsets). Under certain conditions a level-triggered interrupt is - * erroneously delivered as edge-triggered one but the respective IRR - * bit gets set nevertheless. As a result the I/O unit expects an EOI - * message but it will never arrive and further interrupts are blocked - * from the source. The exact reason is so far unknown, but the - * phenomenon was observed when two consecutive interrupt requests - * from a given source get delivered to the same CPU and the source is - * temporarily disabled in between. - * - * A workaround is to simulate an EOI message manually. We achieve it - * by setting the trigger mode to edge and then to level when the edge - * trigger mode gets detected in the TMR of a local APIC for a - * level-triggered interrupt. We mask the source for the time of the - * operation to prevent an edge-triggered interrupt escaping meanwhile. - * The idea is from Manfred Spraul. --macro - */ - i = irq_cfg(irq)->vector; - - v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); - - ack_APIC_irq(); - - if (!(v & (1 << (i & 0x1f)))) { - atomic_inc(&irq_mis_count); - spin_lock(&ioapic_lock); - __mask_and_edge_IO_APIC_irq(irq); - __unmask_and_level_IO_APIC_irq(irq); - spin_unlock(&ioapic_lock); - } -} -#endif - -static struct irq_chip ioapic_chip __read_mostly = { - .name = "IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_apic_edge, - .eoi = ack_apic_level, -#ifdef CONFIG_SMP - .set_affinity = set_ioapic_affinity_irq, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -#ifdef CONFIG_INTR_REMAP -static struct irq_chip ir_ioapic_chip __read_mostly = { - .name = "IR-IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_x2apic_edge, - .eoi = ack_x2apic_level, -#ifdef CONFIG_SMP - .set_affinity = set_ir_ioapic_affinity_irq, -#endif - .retrigger = ioapic_retrigger_irq, -}; -#endif - -static inline void init_IO_APIC_traps(void) -{ - int irq; - struct irq_desc *desc; - struct irq_cfg *cfg; - - /* - * NOTE! The local APIC isn't very good at handling - * multiple interrupts at the same interrupt level. - * As the interrupt level is determined by taking the - * vector number and shifting that right by 4, we - * want to spread these out a bit so that they don't - * all fall in the same interrupt level. - * - * Also, we've got to be careful not to trash gate - * 0x80, because int 0x80 is hm, kind of importantish. ;) - */ - for_each_irq_cfg(cfg) { - irq = cfg->irq; - if (IO_APIC_IRQ(irq) && !cfg->vector) { - /* - * Hmm.. We don't have an entry for this, - * so default to an old-fashioned 8259 - * interrupt if we can.. - */ - if (irq < 16) - make_8259A_irq(irq); - else { - desc = irq_to_desc(irq); - /* Strange. Oh, well.. */ - desc->chip = &no_irq_chip; - } - } - } -} - -/* - * The local APIC irq-chip implementation: - */ - -static void mask_lapic_irq(unsigned int irq) -{ - unsigned long v; - - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v | APIC_LVT_MASKED); -} - -static void unmask_lapic_irq(unsigned int irq) -{ - unsigned long v; - - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); -} - -static void ack_lapic_irq (unsigned int irq) -{ - ack_APIC_irq(); -} - -static struct irq_chip lapic_chip __read_mostly = { - .name = "local-APIC", - .mask = mask_lapic_irq, - .unmask = unmask_lapic_irq, - .ack = ack_lapic_irq, -}; - -static void lapic_register_intr(int irq) -{ - struct irq_desc *desc; - - desc = irq_to_desc(irq); - desc->status &= ~IRQ_LEVEL; - set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, - "edge"); -} - -static void __init setup_nmi(void) -{ - /* - * Dirty trick to enable the NMI watchdog ... - * We put the 8259A master into AEOI mode and - * unmask on all local APICs LVT0 as NMI. - * - * The idea to use the 8259A in AEOI mode ('8259A Virtual Wire') - * is from Maciej W. Rozycki - so we do not have to EOI from - * the NMI handler or the timer interrupt. - */ - apic_printk(APIC_VERBOSE, KERN_INFO "activating NMI Watchdog ..."); - - enable_NMI_through_LVT0(); - - apic_printk(APIC_VERBOSE, " done.\n"); -} - -/* - * This looks a bit hackish but it's about the only one way of sending - * a few INTA cycles to 8259As and any associated glue logic. ICR does - * not support the ExtINT mode, unfortunately. We need to send these - * cycles as some i82489DX-based boards have glue logic that keeps the - * 8259A interrupt line asserted until INTA. --macro - */ -static inline void __init unlock_ExtINT_logic(void) -{ - int apic, pin, i; - struct IO_APIC_route_entry entry0, entry1; - unsigned char save_control, save_freq_select; - - pin = find_isa_irq_pin(8, mp_INT); - if (pin == -1) { - WARN_ON_ONCE(1); - return; - } - apic = find_isa_irq_apic(8, mp_INT); - if (apic == -1) { - WARN_ON_ONCE(1); - return; - } - - entry0 = ioapic_read_entry(apic, pin); - clear_IO_APIC_pin(apic, pin); - - memset(&entry1, 0, sizeof(entry1)); - - entry1.dest_mode = 0; /* physical delivery */ - entry1.mask = 0; /* unmask IRQ now */ - entry1.dest = hard_smp_processor_id(); - entry1.delivery_mode = dest_ExtINT; - entry1.polarity = entry0.polarity; - entry1.trigger = 0; - entry1.vector = 0; - - ioapic_write_entry(apic, pin, entry1); - - save_control = CMOS_READ(RTC_CONTROL); - save_freq_select = CMOS_READ(RTC_FREQ_SELECT); - CMOS_WRITE((save_freq_select & ~RTC_RATE_SELECT) | 0x6, - RTC_FREQ_SELECT); - CMOS_WRITE(save_control | RTC_PIE, RTC_CONTROL); - - i = 100; - while (i-- > 0) { - mdelay(10); - if ((CMOS_READ(RTC_INTR_FLAGS) & RTC_PF) == RTC_PF) - i -= 10; - } - - CMOS_WRITE(save_control, RTC_CONTROL); - CMOS_WRITE(save_freq_select, RTC_FREQ_SELECT); - clear_IO_APIC_pin(apic, pin); - - ioapic_write_entry(apic, pin, entry0); -} - -static int disable_timer_pin_1 __initdata; -/* Actually the next is obsolete, but keep it for paranoid reasons -AK */ -static int __init disable_timer_pin_setup(char *arg) -{ - disable_timer_pin_1 = 1; - return 0; -} -early_param("disable_timer_pin_1", disable_timer_pin_setup); - -int timer_through_8259 __initdata; - -/* - * This code may look a bit paranoid, but it's supposed to cooperate with - * a wide range of boards and BIOS bugs. Fortunately only the timer IRQ - * is so screwy. Thanks to Brian Perkins for testing/hacking this beast - * fanatically on his truly buggy board. - * - * FIXME: really need to revamp this for all platforms. - */ -static inline void __init check_timer(void) -{ - struct irq_cfg *cfg = irq_cfg(0); - int apic1, pin1, apic2, pin2; - unsigned long flags; - unsigned int ver; - int no_pin1 = 0; - - local_irq_save(flags); - - ver = apic_read(APIC_LVR); - ver = GET_APIC_VERSION(ver); - - /* - * get/set the timer IRQ vector: - */ - disable_8259A_irq(0); - assign_irq_vector(0, TARGET_CPUS); - - /* - * As IRQ0 is to be enabled in the 8259A, the virtual - * wire has to be disabled in the local APIC. Also - * timer interrupts need to be acknowledged manually in - * the 8259A for the i82489DX when using the NMI - * watchdog as that APIC treats NMIs as level-triggered. - * The AEOI mode will finish them in the 8259A - * automatically. - */ - apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); - init_8259A(1); -#ifdef CONFIG_X86_32 - timer_ack = (nmi_watchdog == NMI_IO_APIC && !APIC_INTEGRATED(ver)); -#endif - - pin1 = find_isa_irq_pin(0, mp_INT); - apic1 = find_isa_irq_apic(0, mp_INT); - pin2 = ioapic_i8259.pin; - apic2 = ioapic_i8259.apic; - - apic_printk(APIC_QUIET, KERN_INFO "..TIMER: vector=0x%02X " - "apic1=%d pin1=%d apic2=%d pin2=%d\n", - cfg->vector, apic1, pin1, apic2, pin2); - - /* - * Some BIOS writers are clueless and report the ExtINTA - * I/O APIC input from the cascaded 8259A as the timer - * interrupt input. So just in case, if only one pin - * was found above, try it both directly and through the - * 8259A. - */ - if (pin1 == -1) { -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - panic("BIOS bug: timer not connected to IO-APIC"); -#endif - pin1 = pin2; - apic1 = apic2; - no_pin1 = 1; - } else if (pin2 == -1) { - pin2 = pin1; - apic2 = apic1; - } - - if (pin1 != -1) { - /* - * Ok, does IRQ0 through the IOAPIC work? - */ - if (no_pin1) { - add_pin_to_irq(0, apic1, pin1); - setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); - } - unmask_IO_APIC_irq(0); - if (timer_irq_works()) { - if (nmi_watchdog == NMI_IO_APIC) { - setup_nmi(); - enable_8259A_irq(0); - } - if (disable_timer_pin_1 > 0) - clear_IO_APIC_pin(0, pin1); - goto out; - } -#ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - panic("timer doesn't work through Interrupt-remapped IO-APIC"); -#endif - clear_IO_APIC_pin(apic1, pin1); - if (!no_pin1) - apic_printk(APIC_QUIET, KERN_ERR "..MP-BIOS bug: " - "8254 timer not connected to IO-APIC\n"); - - apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer " - "(IRQ0) through the 8259A ...\n"); - apic_printk(APIC_QUIET, KERN_INFO - "..... (found apic %d pin %d) ...\n", apic2, pin2); - /* - * legacy devices should be connected to IO APIC #0 - */ - replace_pin_at_irq(0, apic1, pin1, apic2, pin2); - setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); - unmask_IO_APIC_irq(0); - enable_8259A_irq(0); - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); - timer_through_8259 = 1; - if (nmi_watchdog == NMI_IO_APIC) { - disable_8259A_irq(0); - setup_nmi(); - enable_8259A_irq(0); - } - goto out; - } - /* - * Cleanup, just in case ... - */ - disable_8259A_irq(0); - clear_IO_APIC_pin(apic2, pin2); - apic_printk(APIC_QUIET, KERN_INFO "....... failed.\n"); - } - - if (nmi_watchdog == NMI_IO_APIC) { - apic_printk(APIC_QUIET, KERN_WARNING "timer doesn't work " - "through the IO-APIC - disabling NMI Watchdog!\n"); - nmi_watchdog = NMI_NONE; - } -#ifdef CONFIG_X86_32 - timer_ack = 0; -#endif - - apic_printk(APIC_QUIET, KERN_INFO - "...trying to set up timer as Virtual Wire IRQ...\n"); - - lapic_register_intr(0); - apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ - enable_8259A_irq(0); - - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); - goto out; - } - disable_8259A_irq(0); - apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector); - apic_printk(APIC_QUIET, KERN_INFO "..... failed.\n"); - - apic_printk(APIC_QUIET, KERN_INFO - "...trying to set up timer as ExtINT IRQ...\n"); - - init_8259A(0); - make_8259A_irq(0); - apic_write(APIC_LVT0, APIC_DM_EXTINT); - - unlock_ExtINT_logic(); - - if (timer_irq_works()) { - apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); - goto out; - } - apic_printk(APIC_QUIET, KERN_INFO "..... failed :(.\n"); - panic("IO-APIC + timer doesn't work! Boot with apic=debug and send a " - "report. Then try booting with the 'noapic' option.\n"); -out: - local_irq_restore(flags); -} - -/* - * Traditionally ISA IRQ2 is the cascade IRQ, and is not available - * to devices. However there may be an I/O APIC pin available for - * this interrupt regardless. The pin may be left unconnected, but - * typically it will be reused as an ExtINT cascade interrupt for - * the master 8259A. In the MPS case such a pin will normally be - * reported as an ExtINT interrupt in the MP table. With ACPI - * there is no provision for ExtINT interrupts, and in the absence - * of an override it would be treated as an ordinary ISA I/O APIC - * interrupt, that is edge-triggered and unmasked by default. We - * used to do this, but it caused problems on some systems because - * of the NMI watchdog and sometimes IRQ0 of the 8254 timer using - * the same ExtINT cascade interrupt to drive the local APIC of the - * bootstrap processor. Therefore we refrain from routing IRQ2 to - * the I/O APIC in all cases now. No actual device should request - * it anyway. --macro - */ -#define PIC_IRQS (1 << PIC_CASCADE_IR) - -void __init setup_IO_APIC(void) -{ - -#ifdef CONFIG_X86_32 - enable_IO_APIC(); -#else - /* - * calling enable_IO_APIC() is moved to setup_local_APIC for BP - */ -#endif - - io_apic_irqs = ~PIC_IRQS; - - apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); - /* - * Set up IO-APIC IRQ routing. - */ -#ifdef CONFIG_X86_32 - if (!acpi_ioapic) - setup_ioapic_ids_from_mpc(); -#endif - sync_Arb_IDs(); - setup_IO_APIC_irqs(); - init_IO_APIC_traps(); - check_timer(); -} - -/* - * Called after all the initialization is done. If we didnt find any - * APIC bugs then we can allow the modify fast path - */ - -static int __init io_apic_bug_finalize(void) -{ - if (sis_apic_bug == -1) - sis_apic_bug = 0; - return 0; -} - -late_initcall(io_apic_bug_finalize); - -struct sysfs_ioapic_data { - struct sys_device dev; - struct IO_APIC_route_entry entry[0]; -}; -static struct sysfs_ioapic_data * mp_ioapic_data[MAX_IO_APICS]; - -static int ioapic_suspend(struct sys_device *dev, pm_message_t state) -{ - struct IO_APIC_route_entry *entry; - struct sysfs_ioapic_data *data; - int i; - - data = container_of(dev, struct sysfs_ioapic_data, dev); - entry = data->entry; - for (i = 0; i < nr_ioapic_registers[dev->id]; i ++, entry ++ ) - *entry = ioapic_read_entry(dev->id, i); - - return 0; -} - -static int ioapic_resume(struct sys_device *dev) -{ - struct IO_APIC_route_entry *entry; - struct sysfs_ioapic_data *data; - unsigned long flags; - union IO_APIC_reg_00 reg_00; - int i; - - data = container_of(dev, struct sysfs_ioapic_data, dev); - entry = data->entry; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(dev->id, 0); - if (reg_00.bits.ID != mp_ioapics[dev->id].mp_apicid) { - reg_00.bits.ID = mp_ioapics[dev->id].mp_apicid; - io_apic_write(dev->id, 0, reg_00.raw); - } - spin_unlock_irqrestore(&ioapic_lock, flags); - for (i = 0; i < nr_ioapic_registers[dev->id]; i++) - ioapic_write_entry(dev->id, i, entry[i]); - - return 0; -} - -static struct sysdev_class ioapic_sysdev_class = { - .name = "ioapic", - .suspend = ioapic_suspend, - .resume = ioapic_resume, -}; - -static int __init ioapic_init_sysfs(void) -{ - struct sys_device * dev; - int i, size, error; - - error = sysdev_class_register(&ioapic_sysdev_class); - if (error) - return error; - - for (i = 0; i < nr_ioapics; i++ ) { - size = sizeof(struct sys_device) + nr_ioapic_registers[i] - * sizeof(struct IO_APIC_route_entry); - mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL); - if (!mp_ioapic_data[i]) { - printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); - continue; - } - dev = &mp_ioapic_data[i]->dev; - dev->id = i; - dev->cls = &ioapic_sysdev_class; - error = sysdev_register(dev); - if (error) { - kfree(mp_ioapic_data[i]); - mp_ioapic_data[i] = NULL; - printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i); - continue; - } - } - - return 0; -} - -device_initcall(ioapic_init_sysfs); - -/* - * Dynamic irq allocate and deallocation - */ -unsigned int create_irq_nr(unsigned int irq_want) -{ - /* Allocate an unused irq */ - unsigned int irq; - unsigned int new; - unsigned long flags; - struct irq_cfg *cfg_new; - -#ifndef CONFIG_HAVE_SPARSE_IRQ - irq_want = nr_irqs - 1; -#endif - - irq = 0; - spin_lock_irqsave(&vector_lock, flags); - for (new = irq_want; new > 0; new--) { - if (platform_legacy_irq(new)) - continue; - cfg_new = irq_cfg(new); - if (cfg_new && cfg_new->vector != 0) - continue; - /* check if need to create one */ - if (!cfg_new) - cfg_new = irq_cfg_alloc(new); - if (__assign_irq_vector(new, TARGET_CPUS) == 0) - irq = new; - break; - } - spin_unlock_irqrestore(&vector_lock, flags); - - if (irq > 0) { - dynamic_irq_init(irq); - } - return irq; -} - -int create_irq(void) -{ - int irq; - - irq = create_irq_nr(nr_irqs - 1); - - if (irq == 0) - irq = -1; - - return irq; -} - -void destroy_irq(unsigned int irq) -{ - unsigned long flags; - - dynamic_irq_cleanup(irq); - -#ifdef CONFIG_INTR_REMAP - free_irte(irq); -#endif - spin_lock_irqsave(&vector_lock, flags); - __clear_irq_vector(irq); - spin_unlock_irqrestore(&vector_lock, flags); -} - -/* - * MSI message composition - */ -#ifdef CONFIG_PCI_MSI -static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg) -{ - struct irq_cfg *cfg; - int err; - unsigned dest; - cpumask_t tmp; - - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); - if (err) - return err; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - struct irte irte; - int ir_index; - u16 sub_handle; - - ir_index = map_irq_to_irte_handle(irq, &sub_handle); - BUG_ON(ir_index == -1); - - memset (&irte, 0, sizeof(irte)); - - irte.present = 1; - irte.dst_mode = INT_DEST_MODE; - irte.trigger_mode = 0; /* edge */ - irte.dlvry_mode = INT_DELIVERY_MODE; - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - modify_irte(irq, &irte); - - msg->address_hi = MSI_ADDR_BASE_HI; - msg->data = sub_handle; - msg->address_lo = MSI_ADDR_BASE_LO | MSI_ADDR_IR_EXT_INT | - MSI_ADDR_IR_SHV | - MSI_ADDR_IR_INDEX1(ir_index) | - MSI_ADDR_IR_INDEX2(ir_index); - } else -#endif - { - msg->address_hi = MSI_ADDR_BASE_HI; - msg->address_lo = - MSI_ADDR_BASE_LO | - ((INT_DEST_MODE == 0) ? - MSI_ADDR_DEST_MODE_PHYSICAL: - MSI_ADDR_DEST_MODE_LOGICAL) | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_ADDR_REDIRECTION_CPU: - MSI_ADDR_REDIRECTION_LOWPRI) | - MSI_ADDR_DEST_ID(dest); - - msg->data = - MSI_DATA_TRIGGER_EDGE | - MSI_DATA_LEVEL_ASSERT | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - MSI_DATA_DELIVERY_FIXED: - MSI_DATA_DELIVERY_LOWPRI) | - MSI_DATA_VECTOR(cfg->vector); - } - return err; -} - -#ifdef CONFIG_SMP -static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct msi_msg msg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - read_msi_msg(irq, &msg); - - msg.data &= ~MSI_DATA_VECTOR_MASK; - msg.data |= MSI_DATA_VECTOR(cfg->vector); - msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; - msg.address_lo |= MSI_ADDR_DEST_ID(dest); - - write_msi_msg(irq, &msg); - desc = irq_to_desc(irq); - desc->affinity = mask; -} - -#ifdef CONFIG_INTR_REMAP -/* - * Migrate the MSI irq to another cpumask. This migration is - * done in the process context using interrupt-remapping hardware. - */ -static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned int dest; - cpumask_t tmp, cleanup_mask; - struct irte irte; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (get_irte(irq, &irte)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - irte.vector = cfg->vector; - irte.dest_id = IRTE_DEST(dest); - - /* - * atomically update the IRTE with the new destination and vector. - */ - modify_irte(irq, &irte); - - /* - * After this point, all the interrupts will start arriving - * at the new destination. So, time to cleanup the previous - * vector allocation. - */ - if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } - - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif -#endif /* CONFIG_SMP */ - -/* - * IRQ Chip for MSI PCI/PCI-X/PCI-Express Devices, - * which implement the MSI or MSI-X Capability Structure. - */ -static struct irq_chip msi_chip = { - .name = "PCI-MSI", - .unmask = unmask_msi_irq, - .mask = mask_msi_irq, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = set_msi_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -#ifdef CONFIG_INTR_REMAP -static struct irq_chip msi_ir_chip = { - .name = "IR-PCI-MSI", - .unmask = unmask_msi_irq, - .mask = mask_msi_irq, - .ack = ack_x2apic_edge, -#ifdef CONFIG_SMP - .set_affinity = ir_set_msi_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -/* - * Map the PCI dev to the corresponding remapping hardware unit - * and allocate 'nvec' consecutive interrupt-remapping table entries - * in it. - */ -static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) -{ - struct intel_iommu *iommu; - int index; - - iommu = map_dev_to_ir(dev); - if (!iommu) { - printk(KERN_ERR - "Unable to map PCI %s to iommu\n", pci_name(dev)); - return -ENOENT; - } - - index = alloc_irte(iommu, irq, nvec); - if (index < 0) { - printk(KERN_ERR - "Unable to allocate %d IRTE for PCI %s\n", nvec, - pci_name(dev)); - return -ENOSPC; - } - return index; -} -#endif - -static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) -{ - int ret; - struct msi_msg msg; - - ret = msi_compose_msg(dev, irq, &msg); - if (ret < 0) - return ret; - - set_irq_msi(irq, desc); - write_msi_msg(irq, &msg); - -#ifdef CONFIG_INTR_REMAP - if (irq_remapped(irq)) { - struct irq_desc *desc = irq_to_desc(irq); - /* - * irq migration in process context - */ - desc->status |= IRQ_MOVE_PCNTXT; - set_irq_chip_and_handler_name(irq, &msi_ir_chip, handle_edge_irq, "edge"); - } else -#endif - set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); - - return 0; -} - -static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) -{ - unsigned int irq; - - irq = dev->bus->number; - irq <<= 8; - irq |= dev->devfn; - irq <<= 12; - - return irq; -} - -int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) -{ - unsigned int irq; - int ret; - unsigned int irq_want; - - irq_want = build_irq_for_pci_dev(dev) + 0x100; - - irq = create_irq_nr(irq_want); - if (irq == 0) - return -1; - -#ifdef CONFIG_INTR_REMAP - if (!intr_remapping_enabled) - goto no_ir; - - ret = msi_alloc_irte(dev, irq, 1); - if (ret < 0) - goto error; -no_ir: -#endif - ret = setup_msi_irq(dev, desc, irq); - if (ret < 0) { - destroy_irq(irq); - return ret; - } - return 0; - -#ifdef CONFIG_INTR_REMAP -error: - destroy_irq(irq); - return ret; -#endif -} - -int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) -{ - unsigned int irq; - int ret, sub_handle; - struct msi_desc *desc; - unsigned int irq_want; - -#ifdef CONFIG_INTR_REMAP - struct intel_iommu *iommu = 0; - int index = 0; -#endif - - irq_want = build_irq_for_pci_dev(dev) + 0x100; - sub_handle = 0; - list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq_nr(irq_want--); - if (irq == 0) - return -1; -#ifdef CONFIG_INTR_REMAP - if (!intr_remapping_enabled) - goto no_ir; - - if (!sub_handle) { - /* - * allocate the consecutive block of IRTE's - * for 'nvec' - */ - index = msi_alloc_irte(dev, irq, nvec); - if (index < 0) { - ret = index; - goto error; - } - } else { - iommu = map_dev_to_ir(dev); - if (!iommu) { - ret = -ENOENT; - goto error; - } - /* - * setup the mapping between the irq and the IRTE - * base index, the sub_handle pointing to the - * appropriate interrupt remap table entry. - */ - set_irte_irq(irq, iommu, index, sub_handle); - } -no_ir: -#endif - ret = setup_msi_irq(dev, desc, irq); - if (ret < 0) - goto error; - sub_handle++; - } - return 0; - -error: - destroy_irq(irq); - return ret; -} - -void arch_teardown_msi_irq(unsigned int irq) -{ - destroy_irq(irq); -} - -#ifdef CONFIG_DMAR -#ifdef CONFIG_SMP -static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - struct msi_msg msg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - dmar_msi_read(irq, &msg); - - msg.data &= ~MSI_DATA_VECTOR_MASK; - msg.data |= MSI_DATA_VECTOR(cfg->vector); - msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; - msg.address_lo |= MSI_ADDR_DEST_ID(dest); - - dmar_msi_write(irq, &msg); - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif /* CONFIG_SMP */ - -struct irq_chip dmar_msi_type = { - .name = "DMAR_MSI", - .unmask = dmar_msi_unmask, - .mask = dmar_msi_mask, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = dmar_msi_set_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -int arch_setup_dmar_msi(unsigned int irq) -{ - int ret; - struct msi_msg msg; - - ret = msi_compose_msg(NULL, irq, &msg); - if (ret < 0) - return ret; - dmar_msi_write(irq, &msg); - set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq, - "edge"); - return 0; -} -#endif - -#endif /* CONFIG_PCI_MSI */ -/* - * Hypertransport interrupt support - */ -#ifdef CONFIG_HT_IRQ - -#ifdef CONFIG_SMP - -static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) -{ - struct ht_irq_msg msg; - fetch_ht_irq_msg(irq, &msg); - - msg.address_lo &= ~(HT_IRQ_LOW_VECTOR_MASK | HT_IRQ_LOW_DEST_ID_MASK); - msg.address_hi &= ~(HT_IRQ_HIGH_DEST_ID_MASK); - - msg.address_lo |= HT_IRQ_LOW_VECTOR(vector) | HT_IRQ_LOW_DEST_ID(dest); - msg.address_hi |= HT_IRQ_HIGH_DEST_ID(dest); - - write_ht_irq_msg(irq, &msg); -} - -static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) -{ - struct irq_cfg *cfg; - unsigned int dest; - cpumask_t tmp; - struct irq_desc *desc; - - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) - return; - - if (assign_irq_vector(irq, mask)) - return; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); - - target_ht_irq(irq, dest, cfg->vector); - desc = irq_to_desc(irq); - desc->affinity = mask; -} -#endif - -static struct irq_chip ht_irq_chip = { - .name = "PCI-HT", - .mask = mask_ht_irq, - .unmask = unmask_ht_irq, - .ack = ack_apic_edge, -#ifdef CONFIG_SMP - .set_affinity = set_ht_irq_affinity, -#endif - .retrigger = ioapic_retrigger_irq, -}; - -int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) -{ - struct irq_cfg *cfg; - int err; - cpumask_t tmp; - - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); - if (!err) { - struct ht_irq_msg msg; - unsigned dest; - - cfg = irq_cfg(irq); - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); - - msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); - - msg.address_lo = - HT_IRQ_LOW_BASE | - HT_IRQ_LOW_DEST_ID(dest) | - HT_IRQ_LOW_VECTOR(cfg->vector) | - ((INT_DEST_MODE == 0) ? - HT_IRQ_LOW_DM_PHYSICAL : - HT_IRQ_LOW_DM_LOGICAL) | - HT_IRQ_LOW_RQEOI_EDGE | - ((INT_DELIVERY_MODE != dest_LowestPrio) ? - HT_IRQ_LOW_MT_FIXED : - HT_IRQ_LOW_MT_ARBITRATED) | - HT_IRQ_LOW_IRQ_MASKED; - - write_ht_irq_msg(irq, &msg); - - set_irq_chip_and_handler_name(irq, &ht_irq_chip, - handle_edge_irq, "edge"); - } - return err; -} -#endif /* CONFIG_HT_IRQ */ - -/* -------------------------------------------------------------------------- - ACPI-based IOAPIC Configuration - -------------------------------------------------------------------------- */ - -#ifdef CONFIG_ACPI - -#ifdef CONFIG_X86_32 -int __init io_apic_get_unique_id(int ioapic, int apic_id) -{ - union IO_APIC_reg_00 reg_00; - static physid_mask_t apic_id_map = PHYSID_MASK_NONE; - physid_mask_t tmp; - unsigned long flags; - int i = 0; - - /* - * The P4 platform supports up to 256 APIC IDs on two separate APIC - * buses (one for LAPICs, one for IOAPICs), where predecessors only - * supports up to 16 on one shared APIC bus. - * - * TBD: Expand LAPIC/IOAPIC support on P4-class systems to take full - * advantage of new APIC bus architecture. - */ - - if (physids_empty(apic_id_map)) - apic_id_map = ioapic_phys_id_map(phys_cpu_present_map); - - spin_lock_irqsave(&ioapic_lock, flags); - reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - if (apic_id >= get_physical_broadcast()) { - printk(KERN_WARNING "IOAPIC[%d]: Invalid apic_id %d, trying " - "%d\n", ioapic, apic_id, reg_00.bits.ID); - apic_id = reg_00.bits.ID; - } - - /* - * Every APIC in a system must have a unique ID or we get lots of nice - * 'stuck on smp_invalidate_needed IPI wait' messages. - */ - if (check_apicid_used(apic_id_map, apic_id)) { - - for (i = 0; i < get_physical_broadcast(); i++) { - if (!check_apicid_used(apic_id_map, i)) - break; - } - - if (i == get_physical_broadcast()) - panic("Max apic_id exceeded!\n"); - - printk(KERN_WARNING "IOAPIC[%d]: apic_id %d already used, " - "trying %d\n", ioapic, apic_id, i); - - apic_id = i; - } - - tmp = apicid_to_cpu_present(apic_id); - physids_or(apic_id_map, apic_id_map, tmp); - - if (reg_00.bits.ID != apic_id) { - reg_00.bits.ID = apic_id; - - spin_lock_irqsave(&ioapic_lock, flags); - io_apic_write(ioapic, 0, reg_00.raw); - reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); - - /* Sanity check */ - if (reg_00.bits.ID != apic_id) { - printk("IOAPIC[%d]: Unable to change apic_id!\n", ioapic); - return -1; - } - } - - apic_printk(APIC_VERBOSE, KERN_INFO - "IOAPIC[%d]: Assigned apic_id %d\n", ioapic, apic_id); - - return apic_id; -} - -int __init io_apic_get_version(int ioapic) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return reg_01.bits.version; -} -#endif - -int __init io_apic_get_redir_entries (int ioapic) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return reg_01.bits.entries; -} - - -int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) -{ - if (!IO_APIC_IRQ(irq)) { - apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", - ioapic); - return -EINVAL; - } - - /* - * IRQs < 16 are already in the irq_2_pin[] map - */ - if (irq >= 16) - add_pin_to_irq(irq, ioapic, pin); - - setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); - - return 0; -} - - -int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) -{ - int i; - - if (skip_ioapic_setup) - return -1; - - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == mp_INT && - mp_irqs[i].mp_srcbusirq == bus_irq) - break; - if (i >= mp_irq_entries) - return -1; - - *trigger = irq_trigger(i); - *polarity = irq_polarity(i); - return 0; -} - -#endif /* CONFIG_ACPI */ - -/* - * This function currently is only a helper for the i386 smp boot process where - * we need to reprogram the ioredtbls to cater for the cpus which have come online - * so mask in all cases should simply be TARGET_CPUS - */ -#ifdef CONFIG_SMP -void __init setup_ioapic_dest(void) -{ - int pin, ioapic, irq, irq_entry; - struct irq_cfg *cfg; - - if (skip_ioapic_setup == 1) - return; - - for (ioapic = 0; ioapic < nr_ioapics; ioapic++) { - for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { - irq_entry = find_irq_entry(ioapic, pin, mp_INT); - if (irq_entry == -1) - continue; - irq = pin_2_irq(irq_entry, ioapic, pin); - - /* setup_IO_APIC_irqs could fail to get vector for some device - * when you have too many devices, because at that time only boot - * cpu is online. - */ - cfg = irq_cfg(irq); - if (!cfg->vector) - setup_IO_APIC_irq(ioapic, pin, irq, - irq_trigger(irq_entry), - irq_polarity(irq_entry)); -#ifdef CONFIG_INTR_REMAP - else if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); -#endif - else - set_ioapic_affinity_irq(irq, TARGET_CPUS); - } - - } -} -#endif - -#ifdef CONFIG_X86_64 -#define IOAPIC_RESOURCE_NAME_SIZE 11 - -static struct resource *ioapic_resources; - -static struct resource * __init ioapic_setup_resources(void) -{ - unsigned long n; - struct resource *res; - char *mem; - int i; - - if (nr_ioapics <= 0) - return NULL; - - n = IOAPIC_RESOURCE_NAME_SIZE + sizeof(struct resource); - n *= nr_ioapics; - - mem = alloc_bootmem(n); - res = (void *)mem; - - if (mem != NULL) { - mem += sizeof(struct resource) * nr_ioapics; - - for (i = 0; i < nr_ioapics; i++) { - res[i].name = mem; - res[i].flags = IORESOURCE_MEM | IORESOURCE_BUSY; - sprintf(mem, "IOAPIC %u", i); - mem += IOAPIC_RESOURCE_NAME_SIZE; - } - } - - ioapic_resources = res; - - return res; -} -#endif - -void __init ioapic_init_mappings(void) -{ - unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; - int i; -#ifdef CONFIG_X86_64 - struct resource *ioapic_res; - - ioapic_res = ioapic_setup_resources(); -#endif - for (i = 0; i < nr_ioapics; i++) { - if (smp_found_config) { - ioapic_phys = mp_ioapics[i].mp_apicaddr; -#ifdef CONFIG_X86_32 - if (!ioapic_phys) { - printk(KERN_ERR - "WARNING: bogus zero IO-APIC " - "address found in MPTABLE, " - "disabling IO/APIC support!\n"); - smp_found_config = 0; - skip_ioapic_setup = 1; - goto fake_ioapic_page; - } -#endif - } else { -#ifdef CONFIG_X86_32 -fake_ioapic_page: -#endif - ioapic_phys = (unsigned long) - alloc_bootmem_pages(PAGE_SIZE); - ioapic_phys = __pa(ioapic_phys); - } - set_fixmap_nocache(idx, ioapic_phys); - apic_printk(APIC_VERBOSE, - "mapped IOAPIC to %08lx (%08lx)\n", - __fix_to_virt(idx), ioapic_phys); - idx++; - -#ifdef CONFIG_X86_64 - if (ioapic_res != NULL) { - ioapic_res->start = ioapic_phys; - ioapic_res->end = ioapic_phys + (4 * 1024) - 1; - ioapic_res++; - } -#endif - } -} - -#ifdef CONFIG_X86_64 -static int __init ioapic_insert_resources(void) -{ - int i; - struct resource *r = ioapic_resources; - - if (!r) { - printk(KERN_ERR - "IO APIC resources could be not be allocated.\n"); - return -1; - } - - for (i = 0; i < nr_ioapics; i++) { - insert_resource(&iomem_resource, r); - r++; - } - - return 0; -} - -/* Insert the IO APIC resources after PCI initialization has occured to handle - * IO APICS that are mapped in on a BAR in PCI space. */ -late_initcall(ioapic_insert_resources); -#endif -- cgit v1.2.3 From c691cc84529ec88ccb32b174535bb61875888c90 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:43 -0700 Subject: io_apic: make 32 bit have io_apic resource in /proc/iomem Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 8 -------- 1 file changed, 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index fba6d6ee3480..7e303e0967a4 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3802,7 +3802,6 @@ void __init setup_ioapic_dest(void) } #endif -#ifdef CONFIG_X86_64 #define IOAPIC_RESOURCE_NAME_SIZE 11 static struct resource *ioapic_resources; @@ -3838,17 +3837,14 @@ static struct resource * __init ioapic_setup_resources(void) return res; } -#endif void __init ioapic_init_mappings(void) { unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; int i; -#ifdef CONFIG_X86_64 struct resource *ioapic_res; ioapic_res = ioapic_setup_resources(); -#endif for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { ioapic_phys = mp_ioapics[i].mp_apicaddr; @@ -3877,17 +3873,14 @@ fake_ioapic_page: __fix_to_virt(idx), ioapic_phys); idx++; -#ifdef CONFIG_X86_64 if (ioapic_res != NULL) { ioapic_res->start = ioapic_phys; ioapic_res->end = ioapic_phys + (4 * 1024) - 1; ioapic_res++; } -#endif } } -#ifdef CONFIG_X86_64 static int __init ioapic_insert_resources(void) { int i; @@ -3910,4 +3903,3 @@ static int __init ioapic_insert_resources(void) /* Insert the IO APIC resources after PCI initialization has occured to handle * IO APICS that are mapped in on a BAR in PCI space. */ late_initcall(ioapic_insert_resources); -#endif -- cgit v1.2.3 From 4e738e2f307113feaedebae147c3e0d072e39648 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:47 -0700 Subject: x86: unify mask_IO_APIC_irq use MACRO for 32 bit too Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 80 +++++++++++++---------------------------------- 1 file changed, 21 insertions(+), 59 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 7e303e0967a4..099696109ca8 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -610,18 +610,7 @@ static void __init replace_pin_at_irq(unsigned int irq, add_pin_to_irq(irq, newapic, newpin); } -#ifdef CONFIG_X86_64 -/* - * Synchronize the IO-APIC and the CPU by doing - * a dummy read from the IO-APIC - */ -static inline void io_apic_sync(unsigned int apic) -{ - struct io_apic __iomem *io_apic = io_apic_base(apic); - readl(&io_apic->data); -} - -#define __DO_ACTION(R, ACTION, FINAL) \ +#define __DO_ACTION(R, ACTION_ENABLE, ACTION_DISABLE, FINAL) \ \ { \ int pin; \ @@ -636,7 +625,8 @@ static inline void io_apic_sync(unsigned int apic) break; \ pin = entry->pin; \ reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ - reg ACTION; \ + reg ACTION_DISABLE; \ + reg ACTION_ENABLE; \ io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ FINAL; \ if (!entry->next) \ @@ -645,66 +635,38 @@ static inline void io_apic_sync(unsigned int apic) } \ } -#define DO_ACTION(name,R,ACTION, FINAL) \ +#define DO_ACTION(name,R, ACTION_ENABLE, ACTION_DISABLE, FINAL) \ \ static void name##_IO_APIC_irq (unsigned int irq) \ - __DO_ACTION(R, ACTION, FINAL) - -/* mask = 1 */ -DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, io_apic_sync(entry->apic)) + __DO_ACTION(R, ACTION_ENABLE, ACTION_DISABLE, FINAL) /* mask = 0 */ -DO_ACTION(__unmask, 0, &= ~IO_APIC_REDIR_MASKED, ) +DO_ACTION(__unmask, 0, |= 0, &= ~IO_APIC_REDIR_MASKED, ) -#else - -static void __modify_IO_APIC_irq(unsigned int irq, unsigned long enable, unsigned long disable) +#ifdef CONFIG_X86_64 +/* + * Synchronize the IO-APIC and the CPU by doing + * a dummy read from the IO-APIC + */ +static inline void io_apic_sync(unsigned int apic) { - struct irq_cfg *cfg; - struct irq_pin_list *entry; - unsigned int pin, reg; - - cfg = irq_cfg(irq); - entry = cfg->irq_2_pin; - for (;;) { - if (!entry) - break; - pin = entry->pin; - reg = io_apic_read(entry->apic, 0x10 + pin*2); - reg &= ~disable; - reg |= enable; - io_apic_modify(entry->apic, 0x10 + pin*2, reg); - if (!entry->next) - break; - entry = entry->next; - } + struct io_apic __iomem *io_apic = io_apic_base(apic); + readl(&io_apic->data); } /* mask = 1 */ -static void __mask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, 0); -} +DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, &= ~0, io_apic_sync(entry->apic)) -/* mask = 0 */ -static void __unmask_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, 0, IO_APIC_REDIR_MASKED); -} +#else + +/* mask = 1 */ +DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, &= ~0, ) /* mask = 1, trigger = 0 */ -static void __mask_and_edge_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_MASKED, - IO_APIC_REDIR_LEVEL_TRIGGER); -} +DO_ACTION(__mask_and_edge, 0, |= IO_APIC_REDIR_MASKED, &= ~IO_APIC_REDIR_LEVEL_TRIGGER, ) /* mask = 0, trigger = 1 */ -static void __unmask_and_level_IO_APIC_irq(unsigned int irq) -{ - __modify_IO_APIC_irq(irq, IO_APIC_REDIR_LEVEL_TRIGGER, - IO_APIC_REDIR_MASKED); -} +DO_ACTION(__unmask_and_level, 0, |= IO_APIC_REDIR_LEVEL_TRIGGER, &= ~IO_APIC_REDIR_MASKED, ) #endif -- cgit v1.2.3 From 3eb2cce84beae8fd41de950569cafd5bca7edd5d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:48 -0700 Subject: x86: unify ack_apic_edge use code in 64 to replace move_native_irq(irq, desc); in 32 bit Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 73 +++++++++++++++++++++++------------------------ 1 file changed, 35 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 099696109ca8..a466b04ad5a4 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -389,7 +389,6 @@ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned writel(value, &io_apic->data); } -#ifdef CONFIG_X86_64 static bool io_apic_level_ack_pending(unsigned int irq) { struct irq_pin_list *entry; @@ -419,7 +418,6 @@ static bool io_apic_level_ack_pending(unsigned int irq) return false; } -#endif union entry_union { struct { u32 w1, w2; }; @@ -2398,9 +2396,16 @@ static void ack_apic_edge(unsigned int irq) ack_APIC_irq(); } -#ifdef CONFIG_X86_64 +#ifdef CONFIG_X86_32 +atomic_t irq_mis_count; +#endif + static void ack_apic_level(unsigned int irq) { +#ifdef CONFIG_X86_32 + unsigned long v; + int i; +#endif int do_unmask_irq = 0; irq_complete_move(irq); @@ -2412,6 +2417,31 @@ static void ack_apic_level(unsigned int irq) } #endif +#ifdef CONFIG_X86_32 + /* + * It appears there is an erratum which affects at least version 0x11 + * of I/O APIC (that's the 82093AA and cores integrated into various + * chipsets). Under certain conditions a level-triggered interrupt is + * erroneously delivered as edge-triggered one but the respective IRR + * bit gets set nevertheless. As a result the I/O unit expects an EOI + * message but it will never arrive and further interrupts are blocked + * from the source. The exact reason is so far unknown, but the + * phenomenon was observed when two consecutive interrupt requests + * from a given source get delivered to the same CPU and the source is + * temporarily disabled in between. + * + * A workaround is to simulate an EOI message manually. We achieve it + * by setting the trigger mode to edge and then to level when the edge + * trigger mode gets detected in the TMR of a local APIC for a + * level-triggered interrupt. We mask the source for the time of the + * operation to prevent an edge-triggered interrupt escaping meanwhile. + * The idea is from Manfred Spraul. --macro + */ + i = irq_cfg(irq)->vector; + + v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); +#endif + /* * We must acknowledge the irq before we move it or the acknowledge will * not propagate properly. @@ -2450,41 +2480,8 @@ static void ack_apic_level(unsigned int irq) move_masked_irq(irq); unmask_IO_APIC_irq(irq); } -} -#else -atomic_t irq_mis_count; -static void ack_apic_level(unsigned int irq) -{ - unsigned long v; - int i; - - irq_complete_move(irq); - move_native_irq(irq); - /* - * It appears there is an erratum which affects at least version 0x11 - * of I/O APIC (that's the 82093AA and cores integrated into various - * chipsets). Under certain conditions a level-triggered interrupt is - * erroneously delivered as edge-triggered one but the respective IRR - * bit gets set nevertheless. As a result the I/O unit expects an EOI - * message but it will never arrive and further interrupts are blocked - * from the source. The exact reason is so far unknown, but the - * phenomenon was observed when two consecutive interrupt requests - * from a given source get delivered to the same CPU and the source is - * temporarily disabled in between. - * - * A workaround is to simulate an EOI message manually. We achieve it - * by setting the trigger mode to edge and then to level when the edge - * trigger mode gets detected in the TMR of a local APIC for a - * level-triggered interrupt. We mask the source for the time of the - * operation to prevent an edge-triggered interrupt escaping meanwhile. - * The idea is from Manfred Spraul. --macro - */ - i = irq_cfg(irq)->vector; - - v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); - - ack_APIC_irq(); +#ifdef CONFIG_X86_32 if (!(v & (1 << (i & 0x1f)))) { atomic_inc(&irq_mis_count); spin_lock(&ioapic_lock); @@ -2492,8 +2489,8 @@ static void ack_apic_level(unsigned int irq) __unmask_and_level_IO_APIC_irq(irq); spin_unlock(&ioapic_lock); } -} #endif +} static struct irq_chip ioapic_chip __read_mostly = { .name = "IO-APIC", -- cgit v1.2.3 From 29ccbbf232c035b8c7ff0c5060fbe30a66ed9b99 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:49 -0700 Subject: x86: remove first_free_entry/pin_map_size no user now Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 7 ------- arch/x86/kernel/setup.c | 4 ---- 2 files changed, 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index a466b04ad5a4..5de2d38812aa 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -72,13 +72,6 @@ int sis_apic_bug = -1; static DEFINE_SPINLOCK(ioapic_lock); static DEFINE_SPINLOCK(vector_lock); -int first_free_entry; -/* - * Rough estimation of how many shared IRQs there are, can - * be changed anytime. - */ -int pin_map_size; - /* * # of IRQ routing registers */ diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 02e3a6697977..d90c659b70e9 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1074,10 +1074,6 @@ void __init setup_arch(char **cmdline_p) nr_irqs = 32 * nr_cpu_ids + 224; init_cpu_to_node(); #endif -#ifdef CONFIG_X86_IO_APIC - pin_map_size = nr_irqs * 2; - first_free_entry = nr_irqs; -#endif init_apic_mappings(); ioapic_init_mappings(); -- cgit v1.2.3 From ffd5aae7817fba22c5c3e304a31c44fa0a4e9a97 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:50 -0700 Subject: x86: print local APIC of APs one by one instead of print that of all APs at the time Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 5de2d38812aa..bf0e66d73030 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1777,7 +1777,12 @@ __apicdebuginit(void) print_local_APIC(void *dummy) __apicdebuginit(void) print_all_local_APICs(void) { - on_each_cpu(print_local_APIC, NULL, 1); + int cpu; + + preempt_disable(); + for_each_online_cpu(cpu) + smp_call_function_single(cpu, print_local_APIC, NULL, 1); + preempt_enable(); } __apicdebuginit(void) print_PIC(void) -- cgit v1.2.3 From 8f09cd20a24c5f13c571bc73ddcd47be0af3b70f Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:51 -0700 Subject: x86: make HAVE_SPARSE_IRQ support selectable Ingo said sparse_irq is some intrusive. need to make it selectable to make it simple, remove irq_desc as parameter in some functions. (ack, eoi, set_affinity). may need to make member if irq_chip to take irq_desc, or struct irq later. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 71 ++++++++++++++++++++++++++++++++--------------- arch/x86/kernel/irq_32.c | 2 +- arch/x86/kernel/irq_64.c | 2 +- 3 files changed, 51 insertions(+), 24 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index bf0e66d73030..f853b667fa5c 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -107,7 +107,9 @@ struct irq_cfg; struct irq_pin_list; struct irq_cfg { unsigned int irq; +#ifdef CONFIG_HAVE_SPARSE_IRQ struct irq_cfg *next; +#endif struct irq_pin_list *irq_2_pin; cpumask_t domain; cpumask_t old_domain; @@ -137,20 +139,6 @@ static struct irq_cfg irq_cfg_legacy[] __initdata = { }; static struct irq_cfg irq_cfg_init = { .irq = -1U, }; -/* need to be biger than size of irq_cfg_legacy */ -static int nr_irq_cfg = 32; - -static int __init parse_nr_irq_cfg(char *arg) -{ - if (arg) { - nr_irq_cfg = simple_strtoul(arg, NULL, 0); - if (nr_irq_cfg < 32) - nr_irq_cfg = 32; - } - return 0; -} - -early_param("nr_irq_cfg", parse_nr_irq_cfg); static void init_one_irq_cfg(struct irq_cfg *cfg) { @@ -158,7 +146,9 @@ static void init_one_irq_cfg(struct irq_cfg *cfg) } static struct irq_cfg *irq_cfgx; +#ifdef CONFIG_HAVE_SPARSE_IRQ static struct irq_cfg *irq_cfgx_free; +#endif static void __init init_work(void *data) { struct dyn_array *da = data; @@ -174,15 +164,34 @@ static void __init init_work(void *data) for (i = legacy_count; i < *da->nr; i++) init_one_irq_cfg(&cfg[i]); +#ifdef CONFIG_HAVE_SPARSE_IRQ for (i = 1; i < *da->nr; i++) cfg[i-1].next = &cfg[i]; irq_cfgx_free = &irq_cfgx[legacy_count]; irq_cfgx[legacy_count - 1].next = NULL; +#endif +} + +#ifdef CONFIG_HAVE_SPARSE_IRQ +/* need to be biger than size of irq_cfg_legacy */ +static int nr_irq_cfg = 32; + +static int __init parse_nr_irq_cfg(char *arg) +{ + if (arg) { + nr_irq_cfg = simple_strtoul(arg, NULL, 0); + if (nr_irq_cfg < 32) + nr_irq_cfg = 32; + } + return 0; } -#define for_each_irq_cfg(cfg) \ - for (cfg = irq_cfgx; cfg; cfg = cfg->next) +early_param("nr_irq_cfg", parse_nr_irq_cfg); + +#define for_each_irq_cfg(irqX, cfg) \ + for (cfg = irq_cfgx, irqX = cfg->irq; cfg; cfg = cfg->next, irqX = cfg ? cfg->irq : -1U) + DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); @@ -273,7 +282,26 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) #endif return cfg; } +#else + +#define for_each_irq_cfg(irq, cfg) \ + for (irq = 0, cfg = &irq_cfgx[irq]; irq < nr_irqs; irq++, cfg = &irq_cfgx[irq]) + +DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); +struct irq_cfg *irq_cfg(unsigned int irq) +{ + if (irq < nr_irqs) + return &irq_cfgx[irq]; + + return NULL; +} +struct irq_cfg *irq_cfg_alloc(unsigned int irq) +{ + return irq_cfg(irq); +} + +#endif /* * This is performance-critical, we want to do it O(1) * @@ -1282,11 +1310,10 @@ void __setup_vector_irq(int cpu) struct irq_cfg *cfg; /* Mark the inuse vectors */ - for_each_irq_cfg(cfg) { + for_each_irq_cfg(irq, cfg) { if (!cpu_isset(cpu, cfg->domain)) continue; vector = cfg->vector; - irq = cfg->irq; per_cpu(vector_irq, cpu)[vector] = irq; } /* Mark the free vectors */ @@ -1563,6 +1590,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_03 reg_03; unsigned long flags; struct irq_cfg *cfg; + unsigned int irq; if (apic_verbosity == APIC_QUIET) return; @@ -1651,11 +1679,11 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for_each_irq_cfg(cfg) { + for_each_irq_cfg(irq, cfg) { struct irq_pin_list *entry = cfg->irq_2_pin; if (!entry) continue; - printk(KERN_DEBUG "IRQ%d ", cfg->irq); + printk(KERN_DEBUG "IRQ%d ", irq); for (;;) { printk("-> %d:%d", entry->apic, entry->pin); if (!entry->next) @@ -2535,8 +2563,7 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for_each_irq_cfg(cfg) { - irq = cfg->irq; + for_each_irq_cfg(irq, cfg) { if (IO_APIC_IRQ(irq) && !cfg->vector) { /* * Hmm.. We don't have an entry for this, diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index cc929f2f84f1..b2e1082cf5ad 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -269,7 +269,7 @@ int show_interrupts(struct seq_file *p, void *v) struct irqaction * action; unsigned long flags; unsigned int entries; - struct irq_desc *desc; + struct irq_desc *desc = NULL; int tail = 0; #ifdef CONFIG_HAVE_SPARSE_IRQ diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 348a11168c2b..c2fca89a3f7e 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -74,7 +74,7 @@ int show_interrupts(struct seq_file *p, void *v) struct irqaction * action; unsigned long flags; unsigned int entries; - struct irq_desc *desc; + struct irq_desc *desc = NULL; int tail = 0; #ifdef CONFIG_HAVE_SPARSE_IRQ -- cgit v1.2.3 From 9d6a4d0823b3b8e29156f5e698b5a68687afad32 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 19 Aug 2008 20:50:52 -0700 Subject: x86: probe nr_irqs even only mptable is used for !CONFIG_HAVE_SPARSE_IRQ fix: In file included from arch/x86/kernel/early-quirks.c:18: include/asm/io_apic.h: In function 'probe_nr_irqs': include/asm/io_apic.h:209: error: 'NR_IRQS' undeclared (first use in this function) include/asm/io_apic.h:209: error: (Each undeclared identifier is reported only once include/asm/io_apic.h:209: error: for each function it appears in.) v2: fix by Ingo Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 25 ------------------------- arch/x86/kernel/io_apic.c | 43 ++++++++++++++++++++++++++++++------------- arch/x86/kernel/setup.c | 6 +++--- 3 files changed, 33 insertions(+), 41 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 3e9d163fd92f..5fef4fece4a5 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -957,29 +957,6 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) nr_ioapics++; } -int get_nr_irqs_via_madt(void) -{ - int idx; - int nr = 0; - - for (idx = 0; idx < nr_ioapics; idx++) { - if (mp_ioapic_routing[idx].gsi_end > nr) - nr = mp_ioapic_routing[idx].gsi_end; - } - - nr++; - - /* double it for hotplug and msi and nmi */ - nr <<= 1; - - /* something wrong ? */ - if (nr < 32) - nr = 32; - - return nr; - -} - static void assign_to_mp_irq(struct mp_config_intsrc *m, struct mp_config_intsrc *mp_irq) { @@ -1278,8 +1255,6 @@ static int __init acpi_parse_madt_ioapic_entries(void) } - nr_irqs = get_nr_irqs_via_madt(); - count = acpi_table_parse_madt(ACPI_MADT_TYPE_INTERRUPT_OVERRIDE, acpi_parse_int_src_ovr, nr_irqs); diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index f853b667fa5c..f7e80262cbbb 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3596,6 +3596,36 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) } #endif /* CONFIG_HT_IRQ */ +int __init io_apic_get_redir_entries (int ioapic) +{ + union IO_APIC_reg_01 reg_01; + unsigned long flags; + + spin_lock_irqsave(&ioapic_lock, flags); + reg_01.raw = io_apic_read(ioapic, 1); + spin_unlock_irqrestore(&ioapic_lock, flags); + + return reg_01.bits.entries; +} + +int __init probe_nr_irqs(void) +{ + int idx; + int nr = 0; + + for (idx = 0; idx < nr_ioapics; idx++) + nr += io_apic_get_redir_entries(idx); + + /* double it for hotplug and msi and nmi */ + nr <<= 1; + + /* something wrong ? */ + if (nr < 32) + nr = 32; + + return nr; +} + /* -------------------------------------------------------------------------- ACPI-based IOAPIC Configuration -------------------------------------------------------------------------- */ @@ -3690,19 +3720,6 @@ int __init io_apic_get_version(int ioapic) } #endif -int __init io_apic_get_redir_entries (int ioapic) -{ - union IO_APIC_reg_01 reg_01; - unsigned long flags; - - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - - return reg_01.bits.entries; -} - - int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) { if (!IO_APIC_IRQ(irq)) { diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index d90c659b70e9..61335306696a 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1069,15 +1069,15 @@ void __init setup_arch(char **cmdline_p) prefill_possible_map(); #ifdef CONFIG_X86_64 - /* need to wait for nr_cpu_ids settle down */ - if (nr_irqs == NR_IRQS) - nr_irqs = 32 * nr_cpu_ids + 224; init_cpu_to_node(); #endif init_apic_mappings(); ioapic_init_mappings(); + /* need to wait for io_apic is mapped */ + nr_irqs = probe_nr_irqs(); + kvm_guest_init(); e820_reserve_resources(); -- cgit v1.2.3 From 0c425cec64eb0c0d0dd7037c21a25585cbe3636c Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 18 Aug 2008 13:04:26 +0200 Subject: warning: fix arch x86 kernel io_apic c MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit fix warning: arch/x86/kernel/io_apic.c: In function ‘print_local_APIC’: arch/x86/kernel/io_apic.c:1786: warning: format ‘%08x’ expects type ‘unsigned int’, but argument 2 has type ‘u64’ arch/x86/kernel/io_apic.c:1787: warning: format ‘%08x’ expects type ‘unsigned int’, but argument 2 has type ‘u64’ By creating uniform behavior on 32-bit and 64-bit and printing out the ICR value in two 32-bit words. Code has changed: text data bss dec hex filename 22901 19650 17040 59591 e8c7 io_apic.o.before 22899 19650 17040 59589 e8c5 io_apic.o.after Due to the 32-bit cast narrowing the printed out value on 64-bit. Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index f7e80262cbbb..34c74cf5c244 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1774,8 +1774,8 @@ __apicdebuginit(void) print_local_APIC(void *dummy) } icr = apic_icr_read(); - printk(KERN_DEBUG "... APIC ICR: %08x\n", icr); - printk(KERN_DEBUG "... APIC ICR2: %08x\n", icr >> 32); + printk(KERN_DEBUG "... APIC ICR: %08x\n", (u32)icr); + printk(KERN_DEBUG "... APIC ICR2: %08x\n", (u32)(icr >> 32)); v = apic_read(APIC_LVTT); printk(KERN_DEBUG "... APIC LVTT: %08x\n", v); -- cgit v1.2.3 From e89eb43863c2d9f11a3bbe766766fe646e6c50d9 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 20 Aug 2008 20:46:25 -0700 Subject: x86: sparse_irq needs spin_lock in allocations Suresh Siddha noticed that we should have a spinlock around it. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 34c74cf5c244..7ca556690474 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -146,6 +146,12 @@ static void init_one_irq_cfg(struct irq_cfg *cfg) } static struct irq_cfg *irq_cfgx; + +/* + * Protect the irq_cfgx_free freelist: + */ +static DEFINE_SPINLOCK(irq_cfg_lock); + #ifdef CONFIG_HAVE_SPARSE_IRQ static struct irq_cfg *irq_cfgx_free; #endif @@ -213,8 +219,9 @@ static struct irq_cfg *irq_cfg(unsigned int irq) static struct irq_cfg *irq_cfg_alloc(unsigned int irq) { struct irq_cfg *cfg, *cfg_pri; - int i; + unsigned long flags; int count = 0; + int i; cfg_pri = cfg = irq_cfgx; while (cfg) { @@ -226,6 +233,7 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) count++; } + spin_lock_irqsave(&irq_cfg_lock, flags); if (!irq_cfgx_free) { unsigned long phys; unsigned long total_bytes; @@ -263,6 +271,9 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) else irq_cfgx = cfg; cfg->irq = irq; + + spin_unlock_irqrestore(&irq_cfg_lock, flags); + printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); #ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG { -- cgit v1.2.3 From a2d332fa3445160519de03c350a59602ac1c3df9 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 21 Aug 2008 12:56:32 -0700 Subject: x86: fix 32-bit ioapic lockup with sparseirqs Missed two lines when copying. Fix panic on one of Ingo's machines that need to adjust ioapic id when acpi off/ 32bit. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 7ca556690474..4e44fd1f466e 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2077,6 +2077,8 @@ static void __init setup_ioapic_ids_from_mpc(void) reg_00.bits.ID = mp_ioapics[apic].mp_apicid; spin_lock_irqsave(&ioapic_lock, flags); + io_apic_write(apic, 0, reg_00.raw); + spin_unlock_irqrestore(&ioapic_lock, flags); /* * Sanity check -- cgit v1.2.3 From 052c0bff9b83a578654dfa513d6e3d0b3795f1e8 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 21 Aug 2008 13:10:09 -0700 Subject: x86: fix probe_nr_irqs for xen otherwise Xen is _completely_ unusable with 5 or more VCPUs. (when !CONFIG_HAVE_SPARSE_IRQ). based on Alex Nixon's patch. also add +1 offset after redir_entries Signed-off-by: Yinghai Lu Acked-by: Jeremy Fitzhardinge Acked-by: Alex Nixon Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 4e44fd1f466e..d28128e0392c 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3625,16 +3625,21 @@ int __init probe_nr_irqs(void) { int idx; int nr = 0; +#ifndef CONFIG_XEN + int nr_min = 32; +#else + int nr_min = NR_IRQS; +#endif for (idx = 0; idx < nr_ioapics; idx++) - nr += io_apic_get_redir_entries(idx); + nr += io_apic_get_redir_entries(idx) + 1; /* double it for hotplug and msi and nmi */ nr <<= 1; /* something wrong ? */ - if (nr < 32) - nr = 32; + if (nr < nr_min) + nr = nr_min; return nr; } -- cgit v1.2.3 From 2699574b3c04351f5a23c8a842e17c303d7ebb6f Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Thu, 21 Aug 2008 11:26:43 -0700 Subject: x86: VMI, initialize IRQ vector Initialize vector_irq for the vmi used vector, to point to correct irq. Signed-off-by: Alok N Kataria Signed-off-by: Ingo Molnar --- arch/x86/kernel/vmiclock_32.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vmiclock_32.c b/arch/x86/kernel/vmiclock_32.c index 6953859fe289..254ee07f8635 100644 --- a/arch/x86/kernel/vmiclock_32.c +++ b/arch/x86/kernel/vmiclock_32.c @@ -235,11 +235,14 @@ static void __devinit vmi_time_init_clockevent(void) void __init vmi_time_init(void) { + unsigned int cpu; /* Disable PIT: BIOSes start PIT CH0 with 18.2hz peridic. */ outb_pit(0x3a, PIT_MODE); /* binary, mode 5, LSB/MSB, ch 0 */ vmi_time_init_clockevent(); setup_irq(0, &vmi_clock_action); + for_each_possible_cpu(cpu) + per_cpu(vector_irq, cpu)[vmi_get_timer_vector()] = 0; } #ifdef CONFIG_X86_LOCAL_APIC -- cgit v1.2.3 From db4b5525caafd846ec20f95afbc6403c792e22cf Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:39 -0700 Subject: x86: apic_64.c - setup_APIC_timer has to be __cpuinit function There is no need to hold this code if CPU_HOTPLUG is not defined. Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index cd860856a0b7..8f551276681e 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -400,7 +400,7 @@ static void lapic_timer_broadcast(cpumask_t mask) * Setup the local APIC timer for this CPU. Copy the initilized values * of the boot CPU and register the clock event in the framework. */ -static void setup_APIC_timer(void) +static void __cpuinit setup_APIC_timer(void) { struct clock_event_device *levt = &__get_cpu_var(lapic_events); -- cgit v1.2.3 From 7c37e48b5125fdb0acac846a0a3af42806175d44 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:40 -0700 Subject: x86: apic - introduce get_physical_broadcast for 64bit We don't really use it now on 64bit mode but could reserve it for future. Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 8f551276681e..bb46711a0b1e 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -244,6 +244,16 @@ void __cpuinit enable_NMI_through_LVT0(void) apic_write(APIC_LVT0, v); } +#ifdef CONFIG_X86_32 +/** + * get_physical_broadcast - Get number of physical broadcast IDs + */ +int get_physical_broadcast(void) +{ + return modern_apic() ? 0xff : 0xf; +} +#endif + /** * lapic_get_maxlvt - get the maximum number of local vector table entries */ -- cgit v1.2.3 From 920fa7a507c3b1004a9ebe07a2c9d38605b3406a Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:41 -0700 Subject: x86: apic - unify setup_apicpmtimer Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 10 ++++++++++ arch/x86/kernel/apic_64.c | 2 ++ 2 files changed, 12 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 21c831d96af3..df6ed603e547 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1774,6 +1774,16 @@ static int __init parse_nolapic_timer(char *arg) } early_param("nolapic_timer", parse_nolapic_timer); +#ifdef CONFIG_X86_64 +static __init int setup_apicpmtimer(char *s) +{ + apic_calibrate_pmtmr = 1; + notsc_setup(NULL); + return 0; +} +__setup("apicpmtimer", setup_apicpmtimer); +#endif + static int __init apic_set_verbosity(char *arg) { if (!arg) { diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index bb46711a0b1e..ddc5b245faf2 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -1810,6 +1810,7 @@ static int __init parse_nolapic_timer(char *arg) } early_param("nolapic_timer", parse_nolapic_timer); +#ifdef CONFIG_X86_64 static __init int setup_apicpmtimer(char *s) { apic_calibrate_pmtmr = 1; @@ -1817,6 +1818,7 @@ static __init int setup_apicpmtimer(char *s) return 0; } __setup("apicpmtimer", setup_apicpmtimer); +#endif static int __init apic_set_verbosity(char *arg) { -- cgit v1.2.3 From 80e5609cabd7e4321769701a70297f819a15b08d Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:42 -0700 Subject: x86: apic_64.c - add sanity check for spurious vector definition Do not check for SPUTIOUS_APIC_VECTOR definition twice. Check it once - is what we need. Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index ddc5b245faf2..4587e16f73ec 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -46,6 +46,13 @@ #include #include +/* + * Sanity check + */ +#if ((SPURIOUS_APIC_VECTOR & 0x0F) != 0x0F) +# error SPURIOUS_APIC_VECTOR definition error +#endif + /* Disable local APIC timer from the kernel commandline or via dmi quirk */ static int disable_apic_timer __cpuinitdata; static int apic_calibrate_pmtmr __initdata; @@ -939,8 +946,6 @@ void __cpuinit setup_local_APIC(void) preempt_disable(); value = apic_read(APIC_LVR); - BUILD_BUG_ON((SPURIOUS_APIC_VECTOR & 0x0f) != 0x0f); - /* * Double-check whether this APIC is really registered. * This is meaningless in clustered apic mode, so we skip it. -- cgit v1.2.3 From 89c38c2867ebe37c4c5aee23e7fa1bffb025b171 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:43 -0700 Subject: x86: apic - unify setup_local_APIC - remove useless read of APIC_LVR - wrap with preempt_disable/enable - check for integrated APIC just in place v2: fix by Yinghai Lu. fix lapic_is_integrated using let 64-bit too have pic_mode Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 21 ++++++++++++++----- arch/x86/kernel/apic_64.c | 51 ++++++++++++++++++++++++++++++++++++++++++----- 2 files changed, 62 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index df6ed603e547..f8c1251984e2 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1033,9 +1033,10 @@ static void __cpuinit lapic_setup_esr(void) */ void __cpuinit setup_local_APIC(void) { - unsigned long value, integrated; + unsigned int value; int i, j; +#ifdef CONFIG_X86_32 /* Pound the ESR really hard over the head with a big hammer - mbligh */ if (esr_disable) { apic_write(APIC_ESR, 0); @@ -1043,14 +1044,16 @@ void __cpuinit setup_local_APIC(void) apic_write(APIC_ESR, 0); apic_write(APIC_ESR, 0); } +#endif - integrated = lapic_is_integrated(); + preempt_disable(); /* * Double-check whether this APIC is really registered. + * This is meaningless in clustered apic mode, so we skip it. */ if (!apic_id_registered()) - WARN_ON_ONCE(1); + BUG(); /* * Intel recommends to set DFR, LDR and TPR before enabling @@ -1096,6 +1099,7 @@ void __cpuinit setup_local_APIC(void) */ value |= APIC_SPIV_APIC_ENABLED; +#ifdef CONFIG_X86_32 /* * Some unknown Intel IO/APIC (or APIC) errata is biting us with * certain networking cards. If high frequency interrupts are @@ -1116,8 +1120,13 @@ void __cpuinit setup_local_APIC(void) * See also the comment in end_level_ioapic_irq(). --macro */ - /* Enable focus processor (bit==0) */ + /* + * - enable focus processor (bit==0) + * - 64bit mode always use processor focus + * so no need to set it + */ value &= ~APIC_SPIV_FOCUS_DISABLED; +#endif /* * Set spurious IRQ vector @@ -1154,9 +1163,11 @@ void __cpuinit setup_local_APIC(void) value = APIC_DM_NMI; else value = APIC_DM_NMI | APIC_LVT_MASKED; - if (!integrated) /* 82489DX */ + if (!lapic_is_integrated()) /* 82489DX */ value |= APIC_LVT_LEVEL_TRIGGER; apic_write(APIC_LVT1, value); + + preempt_enable(); } void __cpuinit end_local_APIC_setup(void) diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 4587e16f73ec..283968d4e024 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -76,6 +76,8 @@ char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; */ unsigned int apic_verbosity; +int pic_mode; + /* Have we found an MP table */ int smp_found_config; @@ -943,8 +945,17 @@ void __cpuinit setup_local_APIC(void) unsigned int value; int i, j; +#ifdef CONFIG_X86_32 + /* Pound the ESR really hard over the head with a big hammer - mbligh */ + if (esr_disable) { + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + } +#endif + preempt_disable(); - value = apic_read(APIC_LVR); /* * Double-check whether this APIC is really registered. @@ -997,7 +1008,34 @@ void __cpuinit setup_local_APIC(void) */ value |= APIC_SPIV_APIC_ENABLED; - /* We always use processor focus */ +#ifdef CONFIG_X86_32 + /* + * Some unknown Intel IO/APIC (or APIC) errata is biting us with + * certain networking cards. If high frequency interrupts are + * happening on a particular IOAPIC pin, plus the IOAPIC routing + * entry is masked/unmasked at a high rate as well then sooner or + * later IOAPIC line gets 'stuck', no more interrupts are received + * from the device. If focus CPU is disabled then the hang goes + * away, oh well :-( + * + * [ This bug can be reproduced easily with a level-triggered + * PCI Ne2000 networking cards and PII/PIII processors, dual + * BX chipset. ] + */ + /* + * Actually disabling the focus CPU check just makes the hang less + * frequent as it makes the interrupt distributon model be more + * like LRU than MRU (the short-term load is more even across CPUs). + * See also the comment in end_level_ioapic_irq(). --macro + */ + + /* + * - enable focus processor (bit==0) + * - 64bit mode always use processor focus + * so no need to set it + */ + value &= ~APIC_SPIV_FOCUS_DISABLED; +#endif /* * Set spurious IRQ vector @@ -1016,14 +1054,14 @@ void __cpuinit setup_local_APIC(void) * TODO: set up through-local-APIC from through-I/O-APIC? --macro */ value = apic_read(APIC_LVT0) & APIC_LVT_MASKED; - if (!smp_processor_id() && !value) { + if (!smp_processor_id() && (pic_mode || !value)) { value = APIC_DM_EXTINT; apic_printk(APIC_VERBOSE, "enabled ExtINT on CPU#%d\n", - smp_processor_id()); + smp_processor_id()); } else { value = APIC_DM_EXTINT | APIC_LVT_MASKED; apic_printk(APIC_VERBOSE, "masked ExtINT on CPU#%d\n", - smp_processor_id()); + smp_processor_id()); } apic_write(APIC_LVT0, value); @@ -1034,7 +1072,10 @@ void __cpuinit setup_local_APIC(void) value = APIC_DM_NMI; else value = APIC_DM_NMI | APIC_LVT_MASKED; + if (!lapic_is_integrated()) /* 82489DX */ + value |= APIC_LVT_LEVEL_TRIGGER; apic_write(APIC_LVT1, value); + preempt_enable(); } -- cgit v1.2.3 From 457cc52d4670bcf1470606a108bbf35aac28eb7f Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:44 -0700 Subject: x86: apic_32.c should use __cpuinit section All callers are __init or __cpuinit so there is no need to hold this code without CPU_HOTPLUG being set. Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index f8c1251984e2..4ca3717b3026 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -383,7 +383,7 @@ static void lapic_timer_broadcast(cpumask_t mask) * Setup the local APIC timer for this CPU. Copy the initilized values * of the boot CPU and register the clock event in the framework. */ -static void __devinit setup_APIC_timer(void) +static void __cpuinit setup_APIC_timer(void) { struct clock_event_device *levt = &__get_cpu_var(lapic_events); @@ -652,7 +652,7 @@ void __init setup_boot_APIC_clock(void) setup_APIC_timer(); } -void __devinit setup_secondary_APIC_clock(void) +void __cpuinit setup_secondary_APIC_clock(void) { setup_APIC_timer(); } @@ -1713,7 +1713,7 @@ static struct sys_device device_lapic = { .cls = &lapic_sysclass, }; -static void __devinit apic_pm_activate(void) +static void __cpuinit apic_pm_activate(void) { apic_pm_state.active = 1; } -- cgit v1.2.3 From 6460bc73aac970135104a0bc407c2c8b85394d59 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 24 Aug 2008 02:01:45 -0700 Subject: x86: apic - unify smp_apic_timer_interrupt Signed-off-by: Cyrill Gorcunov Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 3 +++ arch/x86/kernel/apic_64.c | 2 ++ 2 files changed, 5 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 4ca3717b3026..913d9924b101 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -718,6 +718,9 @@ void smp_apic_timer_interrupt(struct pt_regs *regs) * Besides, if we don't timer interrupts ignore the global * interrupt lock, which is the WrongThing (tm) to do. */ +#ifdef CONFIG_X86_64 + exit_idle(); +#endif irq_enter(); local_apic_timer_interrupt(); irq_exit(); diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 283968d4e024..2e109119f647 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -625,7 +625,9 @@ void smp_apic_timer_interrupt(struct pt_regs *regs) * Besides, if we don't timer interrupts ignore the global * interrupt lock, which is the WrongThing (tm) to do. */ +#ifdef CONFIG_X86_64 exit_idle(); +#endif irq_enter(); local_apic_timer_interrupt(); irq_exit(); -- cgit v1.2.3 From b3c5117050e8028d48b2fa0ea09c7a50dd7f3414 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:46 -0700 Subject: x86: apic_xx.c order variables Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 46 +++++++++++++++++++++++---------------------- arch/x86/kernel/apic_64.c | 48 +++++++++++++++++++++++++++++++---------------- 2 files changed, 56 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 913d9924b101..a54512bea629 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -50,16 +50,37 @@ # error SPURIOUS_APIC_VECTOR definition error #endif -unsigned long mp_lapic_addr; - +#ifdef CONFIG_X86_32 /* * Knob to control our willingness to enable the local APIC. * * +1=force-enable */ static int force_enable_local_apic; -int disable_apic; +/* + * APIC command line parameters + */ +static int __init parse_lapic(char *arg) +{ + force_enable_local_apic = 1; + return 0; +} +early_param("lapic", parse_lapic); +#endif +#ifdef CONFIG_X86_64 +static int apic_calibrate_pmtmr __initdata; +static __init int setup_apicpmtimer(char *s) +{ + apic_calibrate_pmtmr = 1; + notsc_setup(NULL); + return 0; +} +__setup("apicpmtimer", setup_apicpmtimer); +#endif + +unsigned long mp_lapic_addr; +int disable_apic; /* Disable local APIC timer from the kernel commandline or via dmi quirk */ static int disable_apic_timer __cpuinitdata; /* Local APIC timer works in C2 */ @@ -1742,15 +1763,6 @@ static void apic_pm_activate(void) { } #endif /* CONFIG_PM */ -/* - * APIC command line parameters - */ -static int __init parse_lapic(char *arg) -{ - force_enable_local_apic = 1; - return 0; -} -early_param("lapic", parse_lapic); static int __init setup_disableapic(char *arg) { @@ -1788,16 +1800,6 @@ static int __init parse_nolapic_timer(char *arg) } early_param("nolapic_timer", parse_nolapic_timer); -#ifdef CONFIG_X86_64 -static __init int setup_apicpmtimer(char *s) -{ - apic_calibrate_pmtmr = 1; - notsc_setup(NULL); - return 0; -} -__setup("apicpmtimer", setup_apicpmtimer); -#endif - static int __init apic_set_verbosity(char *arg) { if (!arg) { diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 2e109119f647..c40a900e155b 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -53,16 +53,44 @@ # error SPURIOUS_APIC_VECTOR definition error #endif -/* Disable local APIC timer from the kernel commandline or via dmi quirk */ -static int disable_apic_timer __cpuinitdata; +#ifdef CONFIG_X86_32 +/* + * Knob to control our willingness to enable the local APIC. + * + * +1=force-enable + */ +static int force_enable_local_apic; +/* + * APIC command line parameters + */ +static int __init parse_lapic(char *arg) +{ + force_enable_local_apic = 1; + return 0; +} +early_param("lapic", parse_lapic); +#endif + +#ifdef CONFIG_X86_64 static int apic_calibrate_pmtmr __initdata; -int disable_apic; +static __init int setup_apicpmtimer(char *s) +{ + apic_calibrate_pmtmr = 1; + notsc_setup(NULL); + return 0; +} +__setup("apicpmtimer", setup_apicpmtimer); +#endif + int disable_x2apic; int x2apic; - /* x2apic enabled before OS handover */ int x2apic_preenabled; +unsigned long mp_lapic_addr; +int disable_apic; +/* Disable local APIC timer from the kernel commandline or via dmi quirk */ +static int disable_apic_timer __cpuinitdata; /* Local APIC timer works in C2 */ int local_apic_timer_c2_ok; EXPORT_SYMBOL_GPL(local_apic_timer_c2_ok); @@ -113,8 +141,6 @@ static DEFINE_PER_CPU(struct clock_event_device, lapic_events); static unsigned long apic_phys; -unsigned long mp_lapic_addr; - /* * Get the LAPIC version */ @@ -1858,16 +1884,6 @@ static int __init parse_nolapic_timer(char *arg) } early_param("nolapic_timer", parse_nolapic_timer); -#ifdef CONFIG_X86_64 -static __init int setup_apicpmtimer(char *s) -{ - apic_calibrate_pmtmr = 1; - notsc_setup(NULL); - return 0; -} -__setup("apicpmtimer", setup_apicpmtimer); -#endif - static int __init apic_set_verbosity(char *arg) { if (!arg) { -- cgit v1.2.3 From 49899eacce79ce39faf531dad3e00f771eba2eb1 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:47 -0700 Subject: x86: use HAVE_X2APIC in apic_64.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_64.c | 35 +++++++++++++++++++++++------------ 1 file changed, 23 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index c40a900e155b..d3ec746aede4 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -82,10 +82,23 @@ static __init int setup_apicpmtimer(char *s) __setup("apicpmtimer", setup_apicpmtimer); #endif -int disable_x2apic; +#ifdef CONFIG_X86_64 +#define HAVE_X2APIC +#endif + +#ifdef HAVE_X2APIC int x2apic; /* x2apic enabled before OS handover */ int x2apic_preenabled; +int disable_x2apic; +static __init int setup_nox2apic(char *str) +{ + disable_x2apic = 1; + setup_clear_cpu_cap(X86_FEATURE_X2APIC); + return 0; +} +early_param("nox2apic", setup_nox2apic); +#endif unsigned long mp_lapic_addr; int disable_apic; @@ -228,6 +241,7 @@ static struct apic_ops xapic_ops = { struct apic_ops __read_mostly *apic_ops = &xapic_ops; EXPORT_SYMBOL_GPL(apic_ops); +#ifdef HAVE_X2APIC static void x2apic_wait_icr_idle(void) { /* no need to wait for icr idle in x2apic */ @@ -261,6 +275,7 @@ static struct apic_ops x2apic_ops = { .wait_icr_idle = x2apic_wait_icr_idle, .safe_wait_icr_idle = safe_x2apic_wait_icr_idle, }; +#endif /** * enable_NMI_through_LVT0 - enable NMI through local vector table 0 @@ -1125,6 +1140,7 @@ void __cpuinit end_local_APIC_setup(void) apic_pm_activate(); } +#ifdef HAVE_X2APIC void check_x2apic(void) { int msr, msr2; @@ -1243,6 +1259,7 @@ end: return; } +#endif /* HAVE_X2APIC */ /* * Detect and enable local APICs on non-SMP boards. @@ -1291,10 +1308,12 @@ void __init early_init_lapic_mapping(void) */ void __init init_apic_mappings(void) { +#ifdef HAVE_X2APIC if (x2apic) { boot_cpu_physical_apicid = read_apic_id(); return; } +#endif /* * If no local APIC can be found then set up a fake all @@ -1335,8 +1354,9 @@ int __init APIC_init_uniprocessor(void) printk(KERN_INFO "Apic disabled by BIOS\n"); return -1; } - +#ifdef HAVE_X2APIC enable_IR_x2apic(); +#endif setup_apic_routing(); verify_local_APIC(); @@ -1672,7 +1692,7 @@ static int lapic_resume(struct sys_device *dev) local_irq_save(flags); -#ifdef CONFIG_X86_64 +#ifdef HAVE_X2APIC if (x2apic) enable_x2apic(); else @@ -1836,15 +1856,6 @@ __cpuinit int apic_is_clustered_box(void) return (clusters > 2); } -static __init int setup_nox2apic(char *str) -{ - disable_x2apic = 1; - clear_cpu_cap(&boot_cpu_data, X86_FEATURE_X2APIC); - return 0; -} -early_param("nox2apic", setup_nox2apic); - - /* * APIC command line parameters */ -- cgit v1.2.3 From 3491998dd54f6d4ef7344518fe5463b299fdf537 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:48 -0700 Subject: x86: add hard_smp_prossor_id with MACRO in io_apic_xx.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 7 +++++++ arch/x86/kernel/apic_64.c | 2 ++ 2 files changed, 9 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index a54512bea629..93718ffb9b9c 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1600,6 +1600,13 @@ void __cpuinit generic_processor_info(int apicid, int version) cpu_set(cpu, cpu_present_map); } +#ifdef CONFIG_X86_64 +int hard_smp_processor_id(void) +{ + return read_apic_id(); +} +#endif + /* * Power management */ diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index d3ec746aede4..eeb69838c2f8 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -1612,10 +1612,12 @@ void __cpuinit generic_processor_info(int apicid, int version) cpu_set(cpu, cpu_present_map); } +#ifdef CONFIG_X86_64 int hard_smp_processor_id(void) { return read_apic_id(); } +#endif /* * Power management -- cgit v1.2.3 From f28c0ae21d80ffd6eb0987901c5273843387e341 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:49 -0700 Subject: x86: make apic_32/64.c more like except x2apic, detec_init_APIC, and calibrating_APIC_clock Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 125 +++++++++++++++++++++++++++++++++++++++++----- arch/x86/kernel/apic_64.c | 10 +++- 2 files changed, 122 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 93718ffb9b9c..bfac26a16099 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -66,6 +66,9 @@ static int __init parse_lapic(char *arg) return 0; } early_param("lapic", parse_lapic); +/* Local APIC was disabled by the BIOS and enabled by the kernel */ +static int enabled_via_apicbase; + #endif #ifdef CONFIG_X86_64 @@ -131,9 +134,6 @@ static struct clock_event_device lapic_clockevent = { }; static DEFINE_PER_CPU(struct clock_event_device, lapic_events); -/* Local APIC was disabled by the BIOS and enabled by the kernel */ -static int enabled_via_apicbase; - static unsigned long apic_phys; /* @@ -240,6 +240,7 @@ void __cpuinit enable_NMI_through_LVT0(void) apic_write(APIC_LVT0, v); } +#ifdef CONFIG_X86_32 /** * get_physical_broadcast - Get number of physical broadcast IDs */ @@ -247,6 +248,7 @@ int get_physical_broadcast(void) { return modern_apic() ? 0xff : 0xf; } +#endif /** * lapic_get_maxlvt - get the maximum number of local vector table entries @@ -1291,6 +1293,32 @@ no_apic: return -1; } +#ifdef CONFIG_X86_64 +void __init early_init_lapic_mapping(void) +{ + unsigned long phys_addr; + + /* + * If no local APIC can be found then go out + * : it means there is no mpatable and MADT + */ + if (!smp_found_config) + return; + + phys_addr = mp_lapic_addr; + + set_fixmap_nocache(FIX_APIC_BASE, phys_addr); + apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", + APIC_BASE, phys_addr); + + /* + * Fetch the APIC ID of the BSP in case we have a + * default configuration (or the MP table is broken). + */ + boot_cpu_physical_apicid = read_apic_id(); +} +#endif + /** * init_apic_mappings - initialize APIC mappings */ @@ -1308,8 +1336,8 @@ void __init init_apic_mappings(void) apic_phys = mp_lapic_addr; set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - printk(KERN_DEBUG "mapped APIC to %08lx (%08lx)\n", APIC_BASE, - apic_phys); + apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", + APIC_BASE, apic_phys); /* * Fetch the APIC ID of the BSP in case we have a @@ -1317,14 +1345,12 @@ void __init init_apic_mappings(void) */ if (boot_cpu_physical_apicid == -1U) boot_cpu_physical_apicid = read_apic_id(); - } /* * This initializes the IO-APIC and APIC hardware if this is * a UP kernel. */ - int apic_version[MAX_APICS]; int __init APIC_init_uniprocessor(void) @@ -1682,11 +1708,6 @@ static int lapic_resume(struct sys_device *dev) local_irq_save(flags); -#ifdef CONFIG_X86_64 - if (x2apic) - enable_x2apic(); - else -#endif { /* * Make sure the APICBASE points to the right address @@ -1770,7 +1791,87 @@ static void apic_pm_activate(void) { } #endif /* CONFIG_PM */ +#ifdef CONFIG_X86_64 +/* + * apic_is_clustered_box() -- Check if we can expect good TSC + * + * Thus far, the major user of this is IBM's Summit2 series: + * + * Clustered boxes may have unsynced TSC problems if they are + * multi-chassis. Use available data to take a good guess. + * If in doubt, go HPET. + */ +__cpuinit int apic_is_clustered_box(void) +{ + int i, clusters, zeros; + unsigned id; + u16 *bios_cpu_apicid; + DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS); + + /* + * there is not this kind of box with AMD CPU yet. + * Some AMD box with quadcore cpu and 8 sockets apicid + * will be [4, 0x23] or [8, 0x27] could be thought to + * vsmp box still need checking... + */ + if ((boot_cpu_data.x86_vendor == X86_VENDOR_AMD) && !is_vsmp_box()) + return 0; + + bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); + bitmap_zero(clustermap, NUM_APIC_CLUSTERS); + + for (i = 0; i < NR_CPUS; i++) { + /* are we being called early in kernel startup? */ + if (bios_cpu_apicid) { + id = bios_cpu_apicid[i]; + } + else if (i < nr_cpu_ids) { + if (cpu_present(i)) + id = per_cpu(x86_bios_cpu_apicid, i); + else + continue; + } + else + break; + if (id != BAD_APICID) + __set_bit(APIC_CLUSTERID(id), clustermap); + } + + /* Problem: Partially populated chassis may not have CPUs in some of + * the APIC clusters they have been allocated. Only present CPUs have + * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. + * Since clusters are allocated sequentially, count zeros only if + * they are bounded by ones. + */ + clusters = 0; + zeros = 0; + for (i = 0; i < NUM_APIC_CLUSTERS; i++) { + if (test_bit(i, clustermap)) { + clusters += 1 + zeros; + zeros = 0; + } else + ++zeros; + } + + /* ScaleMP vSMPowered boxes have one cluster per board and TSCs are + * not guaranteed to be synced between boards + */ + if (is_vsmp_box() && clusters > 1) + return 1; + + /* + * If clusters > 2, then should be multi-chassis. + * May have to revisit this when multi-core + hyperthreaded CPUs come + * out, but AFAIK this will work even for them. + */ + return (clusters > 2); +} +#endif + +/* + * APIC command line parameters + */ static int __init setup_disableapic(char *arg) { disable_apic = 1; diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index eeb69838c2f8..dca6c246e731 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -69,6 +69,9 @@ static int __init parse_lapic(char *arg) return 0; } early_param("lapic", parse_lapic); +/* Local APIC was disabled by the BIOS and enabled by the kernel */ +static int enabled_via_apicbase; + #endif #ifdef CONFIG_X86_64 @@ -1279,6 +1282,7 @@ static int __init detect_init_APIC(void) return 0; } +#ifdef CONFIG_X86_64 void __init early_init_lapic_mapping(void) { unsigned long phys_addr; @@ -1302,6 +1306,7 @@ void __init early_init_lapic_mapping(void) */ boot_cpu_physical_apicid = read_apic_id(); } +#endif /** * init_apic_mappings - initialize APIC mappings @@ -1334,7 +1339,8 @@ void __init init_apic_mappings(void) * Fetch the APIC ID of the BSP in case we have a * default configuration (or the MP table is broken). */ - boot_cpu_physical_apicid = read_apic_id(); + if (boot_cpu_physical_apicid == -1U) + boot_cpu_physical_apicid = read_apic_id(); } /* @@ -1782,6 +1788,7 @@ static void apic_pm_activate(void) { } #endif /* CONFIG_PM */ +#ifdef CONFIG_X86_64 /* * apic_is_clustered_box() -- Check if we can expect good TSC * @@ -1857,6 +1864,7 @@ __cpuinit int apic_is_clustered_box(void) */ return (clusters > 2); } +#endif /* * APIC command line parameters -- cgit v1.2.3 From fa2bd35a8d5c88c03b638c72daf7f38a132d0e8c Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:50 -0700 Subject: x86: merge APIC_init_uniprocessor Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 51 +++++++++++++++++++++++++++++++++++++++++------ arch/x86/kernel/apic_64.c | 48 +++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 90 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index bfac26a16099..8f8b0e1f3eb3 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1355,6 +1355,17 @@ int apic_version[MAX_APICS]; int __init APIC_init_uniprocessor(void) { +#ifdef CONFIG_X86_64 + if (disable_apic) { + printk(KERN_INFO "Apic disabled\n"); + return -1; + } + if (!cpu_has_apic) { + disable_apic = 1; + printk(KERN_INFO "Apic disabled by BIOS\n"); + return -1; + } +#else if (!smp_found_config && !cpu_has_apic) return -1; @@ -1368,34 +1379,62 @@ int __init APIC_init_uniprocessor(void) clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); return -1; } +#endif +#ifdef HAVE_X2APIC + enable_IR_x2apic(); +#endif +#ifdef CONFIG_X86_64 + setup_apic_routing(); +#endif verify_local_APIC(); - connect_bsp_APIC(); +#ifdef CONFIG_X86_64 + apic_write(APIC_ID, SET_APIC_ID(boot_cpu_physical_apicid)); +#else /* * Hack: In case of kdump, after a crash, kernel might be booting * on a cpu with non-zero lapic id. But boot_cpu_physical_apicid * might be zero if read from MP tables. Get it from LAPIC. */ -#ifdef CONFIG_CRASH_DUMP +# ifdef CONFIG_CRASH_DUMP boot_cpu_physical_apicid = read_apic_id(); +# endif #endif physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); - setup_local_APIC(); +#ifdef CONFIG_X86_64 + /* + * Now enable IO-APICs, actually call clear_IO_APIC + * We need clear_IO_APIC before enabling vector on BP + */ + if (!skip_ioapic_setup && nr_ioapics) + enable_IO_APIC(); +#endif + #ifdef CONFIG_X86_IO_APIC if (!smp_found_config || skip_ioapic_setup || !nr_ioapics) #endif localise_nmi_watchdog(); end_local_APIC_setup(); + #ifdef CONFIG_X86_IO_APIC - if (smp_found_config) - if (!skip_ioapic_setup && nr_ioapics) - setup_IO_APIC(); + if (smp_found_config && !skip_ioapic_setup && nr_ioapics) + setup_IO_APIC(); +# ifdef CONFIG_X86_64 + else + nr_ioapics = 0; +# endif #endif + +#ifdef CONFIG_X86_64 + setup_boot_APIC_clock(); + check_nmi_watchdog(); +#else setup_boot_clock(); +#endif return 0; } diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index dca6c246e731..16a30c22b07c 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -1351,6 +1351,7 @@ int apic_version[MAX_APICS]; int __init APIC_init_uniprocessor(void) { +#ifdef CONFIG_X86_64 if (disable_apic) { printk(KERN_INFO "Apic disabled\n"); return -1; @@ -1360,37 +1361,78 @@ int __init APIC_init_uniprocessor(void) printk(KERN_INFO "Apic disabled by BIOS\n"); return -1; } +#else + if (!smp_found_config && !cpu_has_apic) + return -1; + + /* + * Complain if the BIOS pretends there is one. + */ + if (!cpu_has_apic && + APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { + printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", + boot_cpu_physical_apicid); + clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); + return -1; + } +#endif + #ifdef HAVE_X2APIC enable_IR_x2apic(); #endif +#ifdef CONFIG_X86_64 setup_apic_routing(); +#endif verify_local_APIC(); - connect_bsp_APIC(); - physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); +#ifdef CONFIG_X86_64 apic_write(APIC_ID, SET_APIC_ID(boot_cpu_physical_apicid)); - +#else + /* + * Hack: In case of kdump, after a crash, kernel might be booting + * on a cpu with non-zero lapic id. But boot_cpu_physical_apicid + * might be zero if read from MP tables. Get it from LAPIC. + */ +# ifdef CONFIG_CRASH_DUMP + boot_cpu_physical_apicid = read_apic_id(); +# endif +#endif + physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); setup_local_APIC(); +#ifdef CONFIG_X86_64 /* * Now enable IO-APICs, actually call clear_IO_APIC * We need clear_IO_APIC before enabling vector on BP */ if (!skip_ioapic_setup && nr_ioapics) enable_IO_APIC(); +#endif +#ifdef CONFIG_X86_IO_APIC if (!smp_found_config || skip_ioapic_setup || !nr_ioapics) +#endif localise_nmi_watchdog(); end_local_APIC_setup(); +#ifdef CONFIG_X86_IO_APIC if (smp_found_config && !skip_ioapic_setup && nr_ioapics) setup_IO_APIC(); +# ifdef CONFIG_X86_64 else nr_ioapics = 0; +# endif +#endif + +#ifdef CONFIG_X86_64 setup_boot_APIC_clock(); check_nmi_watchdog(); +#else + setup_boot_clock(); +#endif + return 0; } -- cgit v1.2.3 From be7a656fe131cba088912bcafb079b029320504d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:51 -0700 Subject: x86: copy detect_init_APIC to the other Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 20 ++++++++++++ arch/x86/kernel/apic_64.c | 82 ++++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 101 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 8f8b0e1f3eb3..1103e05aa1ba 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1214,6 +1214,25 @@ void __cpuinit end_local_APIC_setup(void) apic_pm_activate(); } +#ifdef CONFIG_X86_64 +/* + * Detect and enable local APICs on non-SMP boards. + * Original code written by Keir Fraser. + * On AMD64 we trust the BIOS - if it says no APIC it is likely + * not correctly set up (usually the APIC timer won't work etc.) + */ +static int __init detect_init_APIC(void) +{ + if (!cpu_has_apic) { + printk(KERN_INFO "No local APIC present\n"); + return -1; + } + + mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; + boot_cpu_physical_apicid = 0; + return 0; +} +#else /* * Detect and initialize APIC */ @@ -1292,6 +1311,7 @@ no_apic: printk(KERN_INFO "No local APIC present or hardware disabled\n"); return -1; } +#endif #ifdef CONFIG_X86_64 void __init early_init_lapic_mapping(void) diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 16a30c22b07c..b7268f5c8b18 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -684,7 +684,6 @@ int setup_profiling_timer(unsigned int multiplier) return -EINVAL; } - /* * Local APIC start and shutdown */ @@ -1264,6 +1263,7 @@ end: } #endif /* HAVE_X2APIC */ +#ifdef CONFIG_X86_64 /* * Detect and enable local APICs on non-SMP boards. * Original code written by Keir Fraser. @@ -1281,6 +1281,86 @@ static int __init detect_init_APIC(void) boot_cpu_physical_apicid = 0; return 0; } +#else +/* + * Detect and initialize APIC + */ +static int __init detect_init_APIC(void) +{ + u32 h, l, features; + + /* Disabled by kernel option? */ + if (disable_apic) + return -1; + + switch (boot_cpu_data.x86_vendor) { + case X86_VENDOR_AMD: + if ((boot_cpu_data.x86 == 6 && boot_cpu_data.x86_model > 1) || + (boot_cpu_data.x86 == 15)) + break; + goto no_apic; + case X86_VENDOR_INTEL: + if (boot_cpu_data.x86 == 6 || boot_cpu_data.x86 == 15 || + (boot_cpu_data.x86 == 5 && cpu_has_apic)) + break; + goto no_apic; + default: + goto no_apic; + } + + if (!cpu_has_apic) { + /* + * Over-ride BIOS and try to enable the local APIC only if + * "lapic" specified. + */ + if (!force_enable_local_apic) { + printk(KERN_INFO "Local APIC disabled by BIOS -- " + "you can enable it with \"lapic\"\n"); + return -1; + } + /* + * Some BIOSes disable the local APIC in the APIC_BASE + * MSR. This can only be done in software for Intel P6 or later + * and AMD K7 (Model > 1) or later. + */ + rdmsr(MSR_IA32_APICBASE, l, h); + if (!(l & MSR_IA32_APICBASE_ENABLE)) { + printk(KERN_INFO + "Local APIC disabled by BIOS -- reenabling.\n"); + l &= ~MSR_IA32_APICBASE_BASE; + l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; + wrmsr(MSR_IA32_APICBASE, l, h); + enabled_via_apicbase = 1; + } + } + /* + * The APIC feature bit should now be enabled + * in `cpuid' + */ + features = cpuid_edx(1); + if (!(features & (1 << X86_FEATURE_APIC))) { + printk(KERN_WARNING "Could not enable APIC!\n"); + return -1; + } + set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); + mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; + + /* The BIOS may have set up the APIC at some other address */ + rdmsr(MSR_IA32_APICBASE, l, h); + if (l & MSR_IA32_APICBASE_ENABLE) + mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; + + printk(KERN_INFO "Found and enabled local APIC!\n"); + + apic_pm_activate(); + + return 0; + +no_apic: + printk(KERN_INFO "No local APIC present or hardware disabled\n"); + return -1; +} +#endif #ifdef CONFIG_X86_64 void __init early_init_lapic_mapping(void) -- cgit v1.2.3 From 773763df7de881e65ff2600c024c9ce2dde64750 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:52 -0700 Subject: x86: merge header files in apic_xx.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 8 ++++++++ arch/x86/kernel/apic_64.c | 7 ++++++- 2 files changed, 14 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 1103e05aa1ba..0ec8321e687c 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -23,11 +23,13 @@ #include #include #include +#include #include #include #include #include #include +#include #include #include @@ -36,8 +38,14 @@ #include #include #include +#include #include #include +#include +#include +#include +#include +#include #include #include diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index b7268f5c8b18..ebe417b4d7fc 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -24,9 +24,11 @@ #include #include #include +#include #include #include #include +#include #include #include @@ -34,8 +36,10 @@ #include #include #include +#include #include #include +#include #include #include #include @@ -43,8 +47,9 @@ #include #include -#include #include +#include +#include /* * Sanity check -- cgit v1.2.3 From dc1528dd864a0b79fa67b60b3ca5674fe94fdce5 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:53 -0700 Subject: x86: apic unify smp_spurious/error_interrupt Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 24 +++++++++++++++++++++--- arch/x86/kernel/apic_64.c | 24 ++++++++++++++++++++++-- 2 files changed, 43 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 0ec8321e687c..60a901b2d4fe 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1474,10 +1474,17 @@ int __init APIC_init_uniprocessor(void) /* * This interrupt should _never_ happen with our APIC/SMP architecture */ +#ifdef CONFIG_X86_64 +asmlinkage void smp_spurious_interrupt(void) +#else void smp_spurious_interrupt(struct pt_regs *regs) +#endif { - unsigned long v; + u32 v; +#ifdef CONFIG_X86_64 + exit_idle(); +#endif irq_enter(); /* * Check if this really is a spurious interrupt and ACK it @@ -1488,20 +1495,31 @@ void smp_spurious_interrupt(struct pt_regs *regs) if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) ack_APIC_irq(); +#ifdef CONFIG_X86_64 + add_pda(irq_spurious_count, 1); +#else /* see sw-dev-man vol 3, chapter 7.4.13.5 */ printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " "should never happen.\n", smp_processor_id()); __get_cpu_var(irq_stat).irq_spurious_count++; +#endif irq_exit(); } /* * This interrupt should never happen with our APIC/SMP architecture */ +#ifdef CONFIG_X86_64 +asmlinkage void smp_error_interrupt(void) +#else void smp_error_interrupt(struct pt_regs *regs) +#endif { - unsigned long v, v1; + u32 v, v1; +#ifdef CONFIG_X86_64 + exit_idle(); +#endif irq_enter(); /* First tickle the hardware, only then report what went on. -- REW */ v = apic_read(APIC_ESR); @@ -1520,7 +1538,7 @@ void smp_error_interrupt(struct pt_regs *regs) 6: Received illegal vector 7: Illegal register address */ - printk(KERN_DEBUG "APIC error on CPU%d: %02lx(%02lx)\n", + printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", smp_processor_id(), v , v1); irq_exit(); } diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index ebe417b4d7fc..c728885e4f4a 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -1528,10 +1528,17 @@ int __init APIC_init_uniprocessor(void) /* * This interrupt should _never_ happen with our APIC/SMP architecture */ +#ifdef CONFIG_X86_64 asmlinkage void smp_spurious_interrupt(void) +#else +void smp_spurious_interrupt(struct pt_regs *regs) +#endif { - unsigned int v; + u32 v; + +#ifdef CONFIG_X86_64 exit_idle(); +#endif irq_enter(); /* * Check if this really is a spurious interrupt and ACK it @@ -1542,18 +1549,31 @@ asmlinkage void smp_spurious_interrupt(void) if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) ack_APIC_irq(); +#ifdef CONFIG_X86_64 add_pda(irq_spurious_count, 1); +#else + /* see sw-dev-man vol 3, chapter 7.4.13.5 */ + printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " + "should never happen.\n", smp_processor_id()); + __get_cpu_var(irq_stat).irq_spurious_count++; +#endif irq_exit(); } /* * This interrupt should never happen with our APIC/SMP architecture */ +#ifdef CONFIG_X86_64 asmlinkage void smp_error_interrupt(void) +#else +void smp_error_interrupt(struct pt_regs *regs) +#endif { - unsigned int v, v1; + u32 v, v1; +#ifdef CONFIG_X86_64 exit_idle(); +#endif irq_enter(); /* First tickle the hardware, only then report what went on. -- REW */ v = apic_read(APIC_ESR); -- cgit v1.2.3 From 2f04fa888d270951b9e0fe9e641ddd560d77ad1b Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:54 -0700 Subject: x86: apic copy calibrate_APIC_clock to each other in apic_32/64.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 86 +++++++++++++++++++ arch/x86/kernel/apic_64.c | 215 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 301 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 60a901b2d4fe..05498c37f8d9 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -424,6 +424,90 @@ static void __cpuinit setup_APIC_timer(void) clockevents_register_device(levt); } +#ifdef CONFIG_X86_64 +/* + * In this function we calibrate APIC bus clocks to the external + * timer. Unfortunately we cannot use jiffies and the timer irq + * to calibrate, since some later bootup code depends on getting + * the first irq? Ugh. + * + * We want to do the calibration only once since we + * want to have local timer irqs syncron. CPUs connected + * by the same APIC bus have the very same bus frequency. + * And we want to have irqs off anyways, no accidental + * APIC irq that way. + */ + +#define TICK_COUNT 100000000 + +static int __init calibrate_APIC_clock(void) +{ + unsigned apic, apic_start; + unsigned long tsc, tsc_start; + int result; + + local_irq_disable(); + + /* + * Put whatever arbitrary (but long enough) timeout + * value into the APIC clock, we just want to get the + * counter running for calibration. + * + * No interrupt enable ! + */ + __setup_APIC_LVTT(250000000, 0, 0); + + apic_start = apic_read(APIC_TMCCT); +#ifdef CONFIG_X86_PM_TIMER + if (apic_calibrate_pmtmr && pmtmr_ioport) { + pmtimer_wait(5000); /* 5ms wait */ + apic = apic_read(APIC_TMCCT); + result = (apic_start - apic) * 1000L / 5; + } else +#endif + { + rdtscll(tsc_start); + + do { + apic = apic_read(APIC_TMCCT); + rdtscll(tsc); + } while ((tsc - tsc_start) < TICK_COUNT && + (apic_start - apic) < TICK_COUNT); + + result = (apic_start - apic) * 1000L * tsc_khz / + (tsc - tsc_start); + } + + local_irq_enable(); + + printk(KERN_DEBUG "APIC timer calibration result %d\n", result); + + printk(KERN_INFO "Detected %d.%03d MHz APIC timer.\n", + result / 1000 / 1000, result / 1000 % 1000); + + /* Calculate the scaled math multiplication factor */ + lapic_clockevent.mult = div_sc(result, NSEC_PER_SEC, + lapic_clockevent.shift); + lapic_clockevent.max_delta_ns = + clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); + lapic_clockevent.min_delta_ns = + clockevent_delta2ns(0xF, &lapic_clockevent); + + calibration_result = (result * APIC_DIVISOR) / HZ; + + /* + * Do a sanity check on the APIC calibration result + */ + if (calibration_result < (1000000 / HZ)) { + printk(KERN_WARNING + "APIC frequency too slow, disabling apic timer\n"); + return -1; + } + + return 0; +} + +#else /* * In this functions we calibrate APIC bus clocks to the external timer. * @@ -635,6 +719,8 @@ static int __init calibrate_APIC_clock(void) return 0; } +#endif + /* * Setup the boot APIC * diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index c728885e4f4a..6e99af5ce678 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -478,6 +478,7 @@ static void __cpuinit setup_APIC_timer(void) clockevents_register_device(levt); } +#ifdef CONFIG_X86_64 /* * In this function we calibrate APIC bus clocks to the external * timer. Unfortunately we cannot use jiffies and the timer irq @@ -560,6 +561,220 @@ static int __init calibrate_APIC_clock(void) return 0; } +#else +/* + * In this functions we calibrate APIC bus clocks to the external timer. + * + * We want to do the calibration only once since we want to have local timer + * irqs syncron. CPUs connected by the same APIC bus have the very same bus + * frequency. + * + * This was previously done by reading the PIT/HPET and waiting for a wrap + * around to find out, that a tick has elapsed. I have a box, where the PIT + * readout is broken, so it never gets out of the wait loop again. This was + * also reported by others. + * + * Monitoring the jiffies value is inaccurate and the clockevents + * infrastructure allows us to do a simple substitution of the interrupt + * handler. + * + * The calibration routine also uses the pm_timer when possible, as the PIT + * happens to run way too slow (factor 2.3 on my VAIO CoreDuo, which goes + * back to normal later in the boot process). + */ + +#define LAPIC_CAL_LOOPS (HZ/10) + +static __initdata int lapic_cal_loops = -1; +static __initdata long lapic_cal_t1, lapic_cal_t2; +static __initdata unsigned long long lapic_cal_tsc1, lapic_cal_tsc2; +static __initdata unsigned long lapic_cal_pm1, lapic_cal_pm2; +static __initdata unsigned long lapic_cal_j1, lapic_cal_j2; + +/* + * Temporary interrupt handler. + */ +static void __init lapic_cal_handler(struct clock_event_device *dev) +{ + unsigned long long tsc = 0; + long tapic = apic_read(APIC_TMCCT); + unsigned long pm = acpi_pm_read_early(); + + if (cpu_has_tsc) + rdtscll(tsc); + + switch (lapic_cal_loops++) { + case 0: + lapic_cal_t1 = tapic; + lapic_cal_tsc1 = tsc; + lapic_cal_pm1 = pm; + lapic_cal_j1 = jiffies; + break; + + case LAPIC_CAL_LOOPS: + lapic_cal_t2 = tapic; + lapic_cal_tsc2 = tsc; + if (pm < lapic_cal_pm1) + pm += ACPI_PM_OVRRUN; + lapic_cal_pm2 = pm; + lapic_cal_j2 = jiffies; + break; + } +} + +static int __init calibrate_APIC_clock(void) +{ + struct clock_event_device *levt = &__get_cpu_var(lapic_events); + const long pm_100ms = PMTMR_TICKS_PER_SEC/10; + const long pm_thresh = pm_100ms/100; + void (*real_handler)(struct clock_event_device *dev); + unsigned long deltaj; + long delta, deltapm; + int pm_referenced = 0; + + local_irq_disable(); + + /* Replace the global interrupt handler */ + real_handler = global_clock_event->event_handler; + global_clock_event->event_handler = lapic_cal_handler; + + /* + * Setup the APIC counter to 1e9. There is no way the lapic + * can underflow in the 100ms detection time frame + */ + __setup_APIC_LVTT(1000000000, 0, 0); + + /* Let the interrupts run */ + local_irq_enable(); + + while (lapic_cal_loops <= LAPIC_CAL_LOOPS) + cpu_relax(); + + local_irq_disable(); + + /* Restore the real event handler */ + global_clock_event->event_handler = real_handler; + + /* Build delta t1-t2 as apic timer counts down */ + delta = lapic_cal_t1 - lapic_cal_t2; + apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); + + /* Check, if the PM timer is available */ + deltapm = lapic_cal_pm2 - lapic_cal_pm1; + apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); + + if (deltapm) { + unsigned long mult; + u64 res; + + mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); + + if (deltapm > (pm_100ms - pm_thresh) && + deltapm < (pm_100ms + pm_thresh)) { + apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); + } else { + res = (((u64) deltapm) * mult) >> 22; + do_div(res, 1000000); + printk(KERN_WARNING "APIC calibration not consistent " + "with PM Timer: %ldms instead of 100ms\n", + (long)res); + /* Correct the lapic counter value */ + res = (((u64) delta) * pm_100ms); + do_div(res, deltapm); + printk(KERN_INFO "APIC delta adjusted to PM-Timer: " + "%lu (%ld)\n", (unsigned long) res, delta); + delta = (long) res; + } + pm_referenced = 1; + } + + /* Calculate the scaled math multiplication factor */ + lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, + lapic_clockevent.shift); + lapic_clockevent.max_delta_ns = + clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); + lapic_clockevent.min_delta_ns = + clockevent_delta2ns(0xF, &lapic_clockevent); + + calibration_result = (delta * APIC_DIVISOR) / LAPIC_CAL_LOOPS; + + apic_printk(APIC_VERBOSE, "..... delta %ld\n", delta); + apic_printk(APIC_VERBOSE, "..... mult: %ld\n", lapic_clockevent.mult); + apic_printk(APIC_VERBOSE, "..... calibration result: %u\n", + calibration_result); + + if (cpu_has_tsc) { + delta = (long)(lapic_cal_tsc2 - lapic_cal_tsc1); + apic_printk(APIC_VERBOSE, "..... CPU clock speed is " + "%ld.%04ld MHz.\n", + (delta / LAPIC_CAL_LOOPS) / (1000000 / HZ), + (delta / LAPIC_CAL_LOOPS) % (1000000 / HZ)); + } + + apic_printk(APIC_VERBOSE, "..... host bus clock speed is " + "%u.%04u MHz.\n", + calibration_result / (1000000 / HZ), + calibration_result % (1000000 / HZ)); + + /* + * Do a sanity check on the APIC calibration result + */ + if (calibration_result < (1000000 / HZ)) { + local_irq_enable(); + printk(KERN_WARNING + "APIC frequency too slow, disabling apic timer\n"); + return -1; + } + + levt->features &= ~CLOCK_EVT_FEAT_DUMMY; + + /* We trust the pm timer based calibration */ + if (!pm_referenced) { + apic_printk(APIC_VERBOSE, "... verify APIC timer\n"); + + /* + * Setup the apic timer manually + */ + levt->event_handler = lapic_cal_handler; + lapic_timer_setup(CLOCK_EVT_MODE_PERIODIC, levt); + lapic_cal_loops = -1; + + /* Let the interrupts run */ + local_irq_enable(); + + while (lapic_cal_loops <= LAPIC_CAL_LOOPS) + cpu_relax(); + + local_irq_disable(); + + /* Stop the lapic timer */ + lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); + + local_irq_enable(); + + /* Jiffies delta */ + deltaj = lapic_cal_j2 - lapic_cal_j1; + apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); + + /* Check, if the jiffies result is consistent */ + if (deltaj >= LAPIC_CAL_LOOPS-2 && deltaj <= LAPIC_CAL_LOOPS+2) + apic_printk(APIC_VERBOSE, "... jiffies result ok\n"); + else + levt->features |= CLOCK_EVT_FEAT_DUMMY; + } else + local_irq_enable(); + + if (levt->features & CLOCK_EVT_FEAT_DUMMY) { + printk(KERN_WARNING + "APIC timer disabled due to verification failure.\n"); + return -1; + } + + return 0; +} + +#endif + /* * Setup the boot APIC * -- cgit v1.2.3 From 6c15822752a13748241e1a34068f2b15b660d28e Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:55 -0700 Subject: x86: apic copy apic_64.c to apic_32.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 188 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 188 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 05498c37f8d9..0b11d6e3f091 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -90,6 +90,24 @@ static __init int setup_apicpmtimer(char *s) __setup("apicpmtimer", setup_apicpmtimer); #endif +#ifdef CONFIG_X86_64 +#define HAVE_X2APIC +#endif + +#ifdef HAVE_X2APIC +int x2apic; +/* x2apic enabled before OS handover */ +int x2apic_preenabled; +int disable_x2apic; +static __init int setup_nox2apic(char *str) +{ + disable_x2apic = 1; + setup_clear_cpu_cap(X86_FEATURE_X2APIC); + return 0; +} +early_param("nox2apic", setup_nox2apic); +#endif + unsigned long mp_lapic_addr; int disable_apic; /* Disable local APIC timer from the kernel commandline or via dmi quirk */ @@ -231,6 +249,42 @@ static struct apic_ops xapic_ops = { struct apic_ops __read_mostly *apic_ops = &xapic_ops; EXPORT_SYMBOL_GPL(apic_ops); +#ifdef HAVE_X2APIC +static void x2apic_wait_icr_idle(void) +{ + /* no need to wait for icr idle in x2apic */ + return; +} + +static u32 safe_x2apic_wait_icr_idle(void) +{ + /* no need to wait for icr idle in x2apic */ + return 0; +} + +void x2apic_icr_write(u32 low, u32 id) +{ + wrmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), ((__u64) id) << 32 | low); +} + +u64 x2apic_icr_read(void) +{ + unsigned long val; + + rdmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), val); + return val; +} + +static struct apic_ops x2apic_ops = { + .read = native_apic_msr_read, + .write = native_apic_msr_write, + .icr_read = x2apic_icr_read, + .icr_write = x2apic_icr_write, + .wait_icr_idle = x2apic_wait_icr_idle, + .safe_wait_icr_idle = safe_x2apic_wait_icr_idle, +}; +#endif + /** * enable_NMI_through_LVT0 - enable NMI through local vector table 0 */ @@ -1308,6 +1362,127 @@ void __cpuinit end_local_APIC_setup(void) apic_pm_activate(); } +#ifdef HAVE_X2APIC +void check_x2apic(void) +{ + int msr, msr2; + + rdmsr(MSR_IA32_APICBASE, msr, msr2); + + if (msr & X2APIC_ENABLE) { + printk("x2apic enabled by BIOS, switching to x2apic ops\n"); + x2apic_preenabled = x2apic = 1; + apic_ops = &x2apic_ops; + } +} + +void enable_x2apic(void) +{ + int msr, msr2; + + rdmsr(MSR_IA32_APICBASE, msr, msr2); + if (!(msr & X2APIC_ENABLE)) { + printk("Enabling x2apic\n"); + wrmsr(MSR_IA32_APICBASE, msr | X2APIC_ENABLE, 0); + } +} + +void enable_IR_x2apic(void) +{ +#ifdef CONFIG_INTR_REMAP + int ret; + unsigned long flags; + + if (!cpu_has_x2apic) + return; + + if (!x2apic_preenabled && disable_x2apic) { + printk(KERN_INFO + "Skipped enabling x2apic and Interrupt-remapping " + "because of nox2apic\n"); + return; + } + + if (x2apic_preenabled && disable_x2apic) + panic("Bios already enabled x2apic, can't enforce nox2apic"); + + if (!x2apic_preenabled && skip_ioapic_setup) { + printk(KERN_INFO + "Skipped enabling x2apic and Interrupt-remapping " + "because of skipping io-apic setup\n"); + return; + } + + ret = dmar_table_init(); + if (ret) { + printk(KERN_INFO + "dmar_table_init() failed with %d:\n", ret); + + if (x2apic_preenabled) + panic("x2apic enabled by bios. But IR enabling failed"); + else + printk(KERN_INFO + "Not enabling x2apic,Intr-remapping\n"); + return; + } + + local_irq_save(flags); + mask_8259A(); + save_mask_IO_APIC_setup(); + + ret = enable_intr_remapping(1); + + if (ret && x2apic_preenabled) { + local_irq_restore(flags); + panic("x2apic enabled by bios. But IR enabling failed"); + } + + if (ret) + goto end; + + if (!x2apic) { + x2apic = 1; + apic_ops = &x2apic_ops; + enable_x2apic(); + } +end: + if (ret) + /* + * IR enabling failed + */ + restore_IO_APIC_setup(); + else + reinit_intr_remapped_IO_APIC(x2apic_preenabled); + + unmask_8259A(); + local_irq_restore(flags); + + if (!ret) { + if (!x2apic_preenabled) + printk(KERN_INFO + "Enabled x2apic and interrupt-remapping\n"); + else + printk(KERN_INFO + "Enabled Interrupt-remapping\n"); + } else + printk(KERN_ERR + "Failed to enable Interrupt-remapping and x2apic\n"); +#else + if (!cpu_has_x2apic) + return; + + if (x2apic_preenabled) + panic("x2apic enabled prior OS handover," + " enable CONFIG_INTR_REMAP"); + + printk(KERN_INFO "Enable CONFIG_INTR_REMAP for enabling intr-remapping " + " and x2apic\n"); +#endif + + return; +} +#endif /* HAVE_X2APIC */ + #ifdef CONFIG_X86_64 /* * Detect and enable local APICs on non-SMP boards. @@ -1438,6 +1613,13 @@ void __init early_init_lapic_mapping(void) */ void __init init_apic_mappings(void) { +#ifdef HAVE_X2APIC + if (x2apic) { + boot_cpu_physical_apicid = read_apic_id(); + return; + } +#endif + /* * If no local APIC can be found then set up a fake all * zeroes page to simulate the local APIC and another @@ -1501,6 +1683,7 @@ int __init APIC_init_uniprocessor(void) #ifdef CONFIG_X86_64 setup_apic_routing(); #endif + verify_local_APIC(); connect_bsp_APIC(); @@ -1879,6 +2062,11 @@ static int lapic_resume(struct sys_device *dev) local_irq_save(flags); +#ifdef HAVE_X2APIC + if (x2apic) + enable_x2apic(); + else +#endif { /* * Make sure the APICBASE points to the right address -- cgit v1.2.3 From 0f611ffaea81b8e1c69682188ba1ccaf7683a2ba Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 02:01:56 -0700 Subject: x86: rename apic_32.c and apic_64.c to apic.c Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/apic.c | 2311 ++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/apic_32.c | 2312 --------------------------------------------- arch/x86/kernel/apic_64.c | 2311 -------------------------------------------- 4 files changed, 2312 insertions(+), 4624 deletions(-) create mode 100644 arch/x86/kernel/apic.c delete mode 100644 arch/x86/kernel/apic_32.c delete mode 100644 arch/x86/kernel/apic_64.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 7264f9c3af8f..45cecc59d894 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -60,7 +60,7 @@ obj-$(CONFIG_X86_32_SMP) += smpcommon.o obj-$(CONFIG_X86_64_SMP) += tsc_sync.o smpcommon.o obj-$(CONFIG_X86_TRAMPOLINE) += trampoline_$(BITS).o obj-$(CONFIG_X86_MPPARSE) += mpparse.o -obj-$(CONFIG_X86_LOCAL_APIC) += apic_$(BITS).o nmi.o +obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c new file mode 100644 index 000000000000..6e99af5ce678 --- /dev/null +++ b/arch/x86/kernel/apic.c @@ -0,0 +1,2311 @@ +/* + * Local APIC handling, local APIC timers + * + * (c) 1999, 2000 Ingo Molnar + * + * Fixes + * Maciej W. Rozycki : Bits for genuine 82489DX APICs; + * thanks to Eric Gilmore + * and Rolf G. Tews + * for testing these extensively. + * Maciej W. Rozycki : Various updates and fixes. + * Mikael Pettersson : Power Management for UP-APIC. + * Pavel Machek and + * Mikael Pettersson : PM converted to driver model. + */ + +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +/* + * Sanity check + */ +#if ((SPURIOUS_APIC_VECTOR & 0x0F) != 0x0F) +# error SPURIOUS_APIC_VECTOR definition error +#endif + +#ifdef CONFIG_X86_32 +/* + * Knob to control our willingness to enable the local APIC. + * + * +1=force-enable + */ +static int force_enable_local_apic; +/* + * APIC command line parameters + */ +static int __init parse_lapic(char *arg) +{ + force_enable_local_apic = 1; + return 0; +} +early_param("lapic", parse_lapic); +/* Local APIC was disabled by the BIOS and enabled by the kernel */ +static int enabled_via_apicbase; + +#endif + +#ifdef CONFIG_X86_64 +static int apic_calibrate_pmtmr __initdata; +static __init int setup_apicpmtimer(char *s) +{ + apic_calibrate_pmtmr = 1; + notsc_setup(NULL); + return 0; +} +__setup("apicpmtimer", setup_apicpmtimer); +#endif + +#ifdef CONFIG_X86_64 +#define HAVE_X2APIC +#endif + +#ifdef HAVE_X2APIC +int x2apic; +/* x2apic enabled before OS handover */ +int x2apic_preenabled; +int disable_x2apic; +static __init int setup_nox2apic(char *str) +{ + disable_x2apic = 1; + setup_clear_cpu_cap(X86_FEATURE_X2APIC); + return 0; +} +early_param("nox2apic", setup_nox2apic); +#endif + +unsigned long mp_lapic_addr; +int disable_apic; +/* Disable local APIC timer from the kernel commandline or via dmi quirk */ +static int disable_apic_timer __cpuinitdata; +/* Local APIC timer works in C2 */ +int local_apic_timer_c2_ok; +EXPORT_SYMBOL_GPL(local_apic_timer_c2_ok); + +int first_system_vector = 0xfe; + +char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; + +/* + * Debug level, exported for io_apic.c + */ +unsigned int apic_verbosity; + +int pic_mode; + +/* Have we found an MP table */ +int smp_found_config; + +static struct resource lapic_resource = { + .name = "Local APIC", + .flags = IORESOURCE_MEM | IORESOURCE_BUSY, +}; + +static unsigned int calibration_result; + +static int lapic_next_event(unsigned long delta, + struct clock_event_device *evt); +static void lapic_timer_setup(enum clock_event_mode mode, + struct clock_event_device *evt); +static void lapic_timer_broadcast(cpumask_t mask); +static void apic_pm_activate(void); + +/* + * The local apic timer can be used for any function which is CPU local. + */ +static struct clock_event_device lapic_clockevent = { + .name = "lapic", + .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT + | CLOCK_EVT_FEAT_C3STOP | CLOCK_EVT_FEAT_DUMMY, + .shift = 32, + .set_mode = lapic_timer_setup, + .set_next_event = lapic_next_event, + .broadcast = lapic_timer_broadcast, + .rating = 100, + .irq = -1, +}; +static DEFINE_PER_CPU(struct clock_event_device, lapic_events); + +static unsigned long apic_phys; + +/* + * Get the LAPIC version + */ +static inline int lapic_get_version(void) +{ + return GET_APIC_VERSION(apic_read(APIC_LVR)); +} + +/* + * Check, if the APIC is integrated or a separate chip + */ +static inline int lapic_is_integrated(void) +{ +#ifdef CONFIG_X86_64 + return 1; +#else + return APIC_INTEGRATED(lapic_get_version()); +#endif +} + +/* + * Check, whether this is a modern or a first generation APIC + */ +static int modern_apic(void) +{ + /* AMD systems use old APIC versions, so check the CPU */ + if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD && + boot_cpu_data.x86 >= 0xf) + return 1; + return lapic_get_version() >= 0x14; +} + +/* + * Paravirt kernels also might be using these below ops. So we still + * use generic apic_read()/apic_write(), which might be pointing to different + * ops in PARAVIRT case. + */ +void xapic_wait_icr_idle(void) +{ + while (apic_read(APIC_ICR) & APIC_ICR_BUSY) + cpu_relax(); +} + +u32 safe_xapic_wait_icr_idle(void) +{ + u32 send_status; + int timeout; + + timeout = 0; + do { + send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY; + if (!send_status) + break; + udelay(100); + } while (timeout++ < 1000); + + return send_status; +} + +void xapic_icr_write(u32 low, u32 id) +{ + apic_write(APIC_ICR2, SET_APIC_DEST_FIELD(id)); + apic_write(APIC_ICR, low); +} + +u64 xapic_icr_read(void) +{ + u32 icr1, icr2; + + icr2 = apic_read(APIC_ICR2); + icr1 = apic_read(APIC_ICR); + + return icr1 | ((u64)icr2 << 32); +} + +static struct apic_ops xapic_ops = { + .read = native_apic_mem_read, + .write = native_apic_mem_write, + .icr_read = xapic_icr_read, + .icr_write = xapic_icr_write, + .wait_icr_idle = xapic_wait_icr_idle, + .safe_wait_icr_idle = safe_xapic_wait_icr_idle, +}; + +struct apic_ops __read_mostly *apic_ops = &xapic_ops; +EXPORT_SYMBOL_GPL(apic_ops); + +#ifdef HAVE_X2APIC +static void x2apic_wait_icr_idle(void) +{ + /* no need to wait for icr idle in x2apic */ + return; +} + +static u32 safe_x2apic_wait_icr_idle(void) +{ + /* no need to wait for icr idle in x2apic */ + return 0; +} + +void x2apic_icr_write(u32 low, u32 id) +{ + wrmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), ((__u64) id) << 32 | low); +} + +u64 x2apic_icr_read(void) +{ + unsigned long val; + + rdmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), val); + return val; +} + +static struct apic_ops x2apic_ops = { + .read = native_apic_msr_read, + .write = native_apic_msr_write, + .icr_read = x2apic_icr_read, + .icr_write = x2apic_icr_write, + .wait_icr_idle = x2apic_wait_icr_idle, + .safe_wait_icr_idle = safe_x2apic_wait_icr_idle, +}; +#endif + +/** + * enable_NMI_through_LVT0 - enable NMI through local vector table 0 + */ +void __cpuinit enable_NMI_through_LVT0(void) +{ + unsigned int v; + + /* unmask and set to NMI */ + v = APIC_DM_NMI; + + /* Level triggered for 82489DX (32bit mode) */ + if (!lapic_is_integrated()) + v |= APIC_LVT_LEVEL_TRIGGER; + + apic_write(APIC_LVT0, v); +} + +#ifdef CONFIG_X86_32 +/** + * get_physical_broadcast - Get number of physical broadcast IDs + */ +int get_physical_broadcast(void) +{ + return modern_apic() ? 0xff : 0xf; +} +#endif + +/** + * lapic_get_maxlvt - get the maximum number of local vector table entries + */ +int lapic_get_maxlvt(void) +{ + unsigned int v; + + v = apic_read(APIC_LVR); + /* + * - we always have APIC integrated on 64bit mode + * - 82489DXs do not report # of LVT entries + */ + return APIC_INTEGRATED(GET_APIC_VERSION(v)) ? GET_APIC_MAXLVT(v) : 2; +} + +/* + * Local APIC timer + */ + +/* Clock divisor */ +#ifdef CONFG_X86_64 +#define APIC_DIVISOR 1 +#else +#define APIC_DIVISOR 16 +#endif + +/* + * This function sets up the local APIC timer, with a timeout of + * 'clocks' APIC bus clock. During calibration we actually call + * this function twice on the boot CPU, once with a bogus timeout + * value, second time for real. The other (noncalibrating) CPUs + * call this function only once, with the real, calibrated value. + * + * We do reads before writes even if unnecessary, to get around the + * P5 APIC double write bug. + */ +static void __setup_APIC_LVTT(unsigned int clocks, int oneshot, int irqen) +{ + unsigned int lvtt_value, tmp_value; + + lvtt_value = LOCAL_TIMER_VECTOR; + if (!oneshot) + lvtt_value |= APIC_LVT_TIMER_PERIODIC; + if (!lapic_is_integrated()) + lvtt_value |= SET_APIC_TIMER_BASE(APIC_TIMER_BASE_DIV); + + if (!irqen) + lvtt_value |= APIC_LVT_MASKED; + + apic_write(APIC_LVTT, lvtt_value); + + /* + * Divide PICLK by 16 + */ + tmp_value = apic_read(APIC_TDCR); + apic_write(APIC_TDCR, + (tmp_value & ~(APIC_TDR_DIV_1 | APIC_TDR_DIV_TMBASE)) | + APIC_TDR_DIV_16); + + if (!oneshot) + apic_write(APIC_TMICT, clocks / APIC_DIVISOR); +} + +/* + * Setup extended LVT, AMD specific (K8, family 10h) + * + * Vector mappings are hard coded. On K8 only offset 0 (APIC500) and + * MCE interrupts are supported. Thus MCE offset must be set to 0. + * + * If mask=1, the LVT entry does not generate interrupts while mask=0 + * enables the vector. See also the BKDGs. + */ + +#define APIC_EILVT_LVTOFF_MCE 0 +#define APIC_EILVT_LVTOFF_IBS 1 + +static void setup_APIC_eilvt(u8 lvt_off, u8 vector, u8 msg_type, u8 mask) +{ + unsigned long reg = (lvt_off << 4) + APIC_EILVT0; + unsigned int v = (mask << 16) | (msg_type << 8) | vector; + + apic_write(reg, v); +} + +u8 setup_APIC_eilvt_mce(u8 vector, u8 msg_type, u8 mask) +{ + setup_APIC_eilvt(APIC_EILVT_LVTOFF_MCE, vector, msg_type, mask); + return APIC_EILVT_LVTOFF_MCE; +} + +u8 setup_APIC_eilvt_ibs(u8 vector, u8 msg_type, u8 mask) +{ + setup_APIC_eilvt(APIC_EILVT_LVTOFF_IBS, vector, msg_type, mask); + return APIC_EILVT_LVTOFF_IBS; +} +EXPORT_SYMBOL_GPL(setup_APIC_eilvt_ibs); + +/* + * Program the next event, relative to now + */ +static int lapic_next_event(unsigned long delta, + struct clock_event_device *evt) +{ + apic_write(APIC_TMICT, delta); + return 0; +} + +/* + * Setup the lapic timer in periodic or oneshot mode + */ +static void lapic_timer_setup(enum clock_event_mode mode, + struct clock_event_device *evt) +{ + unsigned long flags; + unsigned int v; + + /* Lapic used as dummy for broadcast ? */ + if (evt->features & CLOCK_EVT_FEAT_DUMMY) + return; + + local_irq_save(flags); + + switch (mode) { + case CLOCK_EVT_MODE_PERIODIC: + case CLOCK_EVT_MODE_ONESHOT: + __setup_APIC_LVTT(calibration_result, + mode != CLOCK_EVT_MODE_PERIODIC, 1); + break; + case CLOCK_EVT_MODE_UNUSED: + case CLOCK_EVT_MODE_SHUTDOWN: + v = apic_read(APIC_LVTT); + v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR); + apic_write(APIC_LVTT, v); + break; + case CLOCK_EVT_MODE_RESUME: + /* Nothing to do here */ + break; + } + + local_irq_restore(flags); +} + +/* + * Local APIC timer broadcast function + */ +static void lapic_timer_broadcast(cpumask_t mask) +{ +#ifdef CONFIG_SMP + send_IPI_mask(mask, LOCAL_TIMER_VECTOR); +#endif +} + +/* + * Setup the local APIC timer for this CPU. Copy the initilized values + * of the boot CPU and register the clock event in the framework. + */ +static void __cpuinit setup_APIC_timer(void) +{ + struct clock_event_device *levt = &__get_cpu_var(lapic_events); + + memcpy(levt, &lapic_clockevent, sizeof(*levt)); + levt->cpumask = cpumask_of_cpu(smp_processor_id()); + + clockevents_register_device(levt); +} + +#ifdef CONFIG_X86_64 +/* + * In this function we calibrate APIC bus clocks to the external + * timer. Unfortunately we cannot use jiffies and the timer irq + * to calibrate, since some later bootup code depends on getting + * the first irq? Ugh. + * + * We want to do the calibration only once since we + * want to have local timer irqs syncron. CPUs connected + * by the same APIC bus have the very same bus frequency. + * And we want to have irqs off anyways, no accidental + * APIC irq that way. + */ + +#define TICK_COUNT 100000000 + +static int __init calibrate_APIC_clock(void) +{ + unsigned apic, apic_start; + unsigned long tsc, tsc_start; + int result; + + local_irq_disable(); + + /* + * Put whatever arbitrary (but long enough) timeout + * value into the APIC clock, we just want to get the + * counter running for calibration. + * + * No interrupt enable ! + */ + __setup_APIC_LVTT(250000000, 0, 0); + + apic_start = apic_read(APIC_TMCCT); +#ifdef CONFIG_X86_PM_TIMER + if (apic_calibrate_pmtmr && pmtmr_ioport) { + pmtimer_wait(5000); /* 5ms wait */ + apic = apic_read(APIC_TMCCT); + result = (apic_start - apic) * 1000L / 5; + } else +#endif + { + rdtscll(tsc_start); + + do { + apic = apic_read(APIC_TMCCT); + rdtscll(tsc); + } while ((tsc - tsc_start) < TICK_COUNT && + (apic_start - apic) < TICK_COUNT); + + result = (apic_start - apic) * 1000L * tsc_khz / + (tsc - tsc_start); + } + + local_irq_enable(); + + printk(KERN_DEBUG "APIC timer calibration result %d\n", result); + + printk(KERN_INFO "Detected %d.%03d MHz APIC timer.\n", + result / 1000 / 1000, result / 1000 % 1000); + + /* Calculate the scaled math multiplication factor */ + lapic_clockevent.mult = div_sc(result, NSEC_PER_SEC, + lapic_clockevent.shift); + lapic_clockevent.max_delta_ns = + clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); + lapic_clockevent.min_delta_ns = + clockevent_delta2ns(0xF, &lapic_clockevent); + + calibration_result = (result * APIC_DIVISOR) / HZ; + + /* + * Do a sanity check on the APIC calibration result + */ + if (calibration_result < (1000000 / HZ)) { + printk(KERN_WARNING + "APIC frequency too slow, disabling apic timer\n"); + return -1; + } + + return 0; +} + +#else +/* + * In this functions we calibrate APIC bus clocks to the external timer. + * + * We want to do the calibration only once since we want to have local timer + * irqs syncron. CPUs connected by the same APIC bus have the very same bus + * frequency. + * + * This was previously done by reading the PIT/HPET and waiting for a wrap + * around to find out, that a tick has elapsed. I have a box, where the PIT + * readout is broken, so it never gets out of the wait loop again. This was + * also reported by others. + * + * Monitoring the jiffies value is inaccurate and the clockevents + * infrastructure allows us to do a simple substitution of the interrupt + * handler. + * + * The calibration routine also uses the pm_timer when possible, as the PIT + * happens to run way too slow (factor 2.3 on my VAIO CoreDuo, which goes + * back to normal later in the boot process). + */ + +#define LAPIC_CAL_LOOPS (HZ/10) + +static __initdata int lapic_cal_loops = -1; +static __initdata long lapic_cal_t1, lapic_cal_t2; +static __initdata unsigned long long lapic_cal_tsc1, lapic_cal_tsc2; +static __initdata unsigned long lapic_cal_pm1, lapic_cal_pm2; +static __initdata unsigned long lapic_cal_j1, lapic_cal_j2; + +/* + * Temporary interrupt handler. + */ +static void __init lapic_cal_handler(struct clock_event_device *dev) +{ + unsigned long long tsc = 0; + long tapic = apic_read(APIC_TMCCT); + unsigned long pm = acpi_pm_read_early(); + + if (cpu_has_tsc) + rdtscll(tsc); + + switch (lapic_cal_loops++) { + case 0: + lapic_cal_t1 = tapic; + lapic_cal_tsc1 = tsc; + lapic_cal_pm1 = pm; + lapic_cal_j1 = jiffies; + break; + + case LAPIC_CAL_LOOPS: + lapic_cal_t2 = tapic; + lapic_cal_tsc2 = tsc; + if (pm < lapic_cal_pm1) + pm += ACPI_PM_OVRRUN; + lapic_cal_pm2 = pm; + lapic_cal_j2 = jiffies; + break; + } +} + +static int __init calibrate_APIC_clock(void) +{ + struct clock_event_device *levt = &__get_cpu_var(lapic_events); + const long pm_100ms = PMTMR_TICKS_PER_SEC/10; + const long pm_thresh = pm_100ms/100; + void (*real_handler)(struct clock_event_device *dev); + unsigned long deltaj; + long delta, deltapm; + int pm_referenced = 0; + + local_irq_disable(); + + /* Replace the global interrupt handler */ + real_handler = global_clock_event->event_handler; + global_clock_event->event_handler = lapic_cal_handler; + + /* + * Setup the APIC counter to 1e9. There is no way the lapic + * can underflow in the 100ms detection time frame + */ + __setup_APIC_LVTT(1000000000, 0, 0); + + /* Let the interrupts run */ + local_irq_enable(); + + while (lapic_cal_loops <= LAPIC_CAL_LOOPS) + cpu_relax(); + + local_irq_disable(); + + /* Restore the real event handler */ + global_clock_event->event_handler = real_handler; + + /* Build delta t1-t2 as apic timer counts down */ + delta = lapic_cal_t1 - lapic_cal_t2; + apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); + + /* Check, if the PM timer is available */ + deltapm = lapic_cal_pm2 - lapic_cal_pm1; + apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); + + if (deltapm) { + unsigned long mult; + u64 res; + + mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); + + if (deltapm > (pm_100ms - pm_thresh) && + deltapm < (pm_100ms + pm_thresh)) { + apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); + } else { + res = (((u64) deltapm) * mult) >> 22; + do_div(res, 1000000); + printk(KERN_WARNING "APIC calibration not consistent " + "with PM Timer: %ldms instead of 100ms\n", + (long)res); + /* Correct the lapic counter value */ + res = (((u64) delta) * pm_100ms); + do_div(res, deltapm); + printk(KERN_INFO "APIC delta adjusted to PM-Timer: " + "%lu (%ld)\n", (unsigned long) res, delta); + delta = (long) res; + } + pm_referenced = 1; + } + + /* Calculate the scaled math multiplication factor */ + lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, + lapic_clockevent.shift); + lapic_clockevent.max_delta_ns = + clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); + lapic_clockevent.min_delta_ns = + clockevent_delta2ns(0xF, &lapic_clockevent); + + calibration_result = (delta * APIC_DIVISOR) / LAPIC_CAL_LOOPS; + + apic_printk(APIC_VERBOSE, "..... delta %ld\n", delta); + apic_printk(APIC_VERBOSE, "..... mult: %ld\n", lapic_clockevent.mult); + apic_printk(APIC_VERBOSE, "..... calibration result: %u\n", + calibration_result); + + if (cpu_has_tsc) { + delta = (long)(lapic_cal_tsc2 - lapic_cal_tsc1); + apic_printk(APIC_VERBOSE, "..... CPU clock speed is " + "%ld.%04ld MHz.\n", + (delta / LAPIC_CAL_LOOPS) / (1000000 / HZ), + (delta / LAPIC_CAL_LOOPS) % (1000000 / HZ)); + } + + apic_printk(APIC_VERBOSE, "..... host bus clock speed is " + "%u.%04u MHz.\n", + calibration_result / (1000000 / HZ), + calibration_result % (1000000 / HZ)); + + /* + * Do a sanity check on the APIC calibration result + */ + if (calibration_result < (1000000 / HZ)) { + local_irq_enable(); + printk(KERN_WARNING + "APIC frequency too slow, disabling apic timer\n"); + return -1; + } + + levt->features &= ~CLOCK_EVT_FEAT_DUMMY; + + /* We trust the pm timer based calibration */ + if (!pm_referenced) { + apic_printk(APIC_VERBOSE, "... verify APIC timer\n"); + + /* + * Setup the apic timer manually + */ + levt->event_handler = lapic_cal_handler; + lapic_timer_setup(CLOCK_EVT_MODE_PERIODIC, levt); + lapic_cal_loops = -1; + + /* Let the interrupts run */ + local_irq_enable(); + + while (lapic_cal_loops <= LAPIC_CAL_LOOPS) + cpu_relax(); + + local_irq_disable(); + + /* Stop the lapic timer */ + lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); + + local_irq_enable(); + + /* Jiffies delta */ + deltaj = lapic_cal_j2 - lapic_cal_j1; + apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); + + /* Check, if the jiffies result is consistent */ + if (deltaj >= LAPIC_CAL_LOOPS-2 && deltaj <= LAPIC_CAL_LOOPS+2) + apic_printk(APIC_VERBOSE, "... jiffies result ok\n"); + else + levt->features |= CLOCK_EVT_FEAT_DUMMY; + } else + local_irq_enable(); + + if (levt->features & CLOCK_EVT_FEAT_DUMMY) { + printk(KERN_WARNING + "APIC timer disabled due to verification failure.\n"); + return -1; + } + + return 0; +} + +#endif + +/* + * Setup the boot APIC + * + * Calibrate and verify the result. + */ +void __init setup_boot_APIC_clock(void) +{ + /* + * The local apic timer can be disabled via the kernel + * commandline or from the CPU detection code. Register the lapic + * timer as a dummy clock event source on SMP systems, so the + * broadcast mechanism is used. On UP systems simply ignore it. + */ + if (disable_apic_timer) { + printk(KERN_INFO "Disabling APIC timer\n"); + /* No broadcast on UP ! */ + if (num_possible_cpus() > 1) { + lapic_clockevent.mult = 1; + setup_APIC_timer(); + } + return; + } + + apic_printk(APIC_VERBOSE, "Using local APIC timer interrupts.\n" + "calibrating APIC timer ...\n"); + + if (calibrate_APIC_clock()) { + /* No broadcast on UP ! */ + if (num_possible_cpus() > 1) + setup_APIC_timer(); + return; + } + + /* + * If nmi_watchdog is set to IO_APIC, we need the + * PIT/HPET going. Otherwise register lapic as a dummy + * device. + */ + if (nmi_watchdog != NMI_IO_APIC) + lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY; + else + printk(KERN_WARNING "APIC timer registered as dummy," + " due to nmi_watchdog=%d!\n", nmi_watchdog); + + /* Setup the lapic or request the broadcast */ + setup_APIC_timer(); +} + +void __cpuinit setup_secondary_APIC_clock(void) +{ + setup_APIC_timer(); +} + +/* + * The guts of the apic timer interrupt + */ +static void local_apic_timer_interrupt(void) +{ + int cpu = smp_processor_id(); + struct clock_event_device *evt = &per_cpu(lapic_events, cpu); + + /* + * Normally we should not be here till LAPIC has been initialized but + * in some cases like kdump, its possible that there is a pending LAPIC + * timer interrupt from previous kernel's context and is delivered in + * new kernel the moment interrupts are enabled. + * + * Interrupts are enabled early and LAPIC is setup much later, hence + * its possible that when we get here evt->event_handler is NULL. + * Check for event_handler being NULL and discard the interrupt as + * spurious. + */ + if (!evt->event_handler) { + printk(KERN_WARNING + "Spurious LAPIC timer interrupt on cpu %d\n", cpu); + /* Switch it off */ + lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt); + return; + } + + /* + * the NMI deadlock-detector uses this. + */ +#ifdef CONFIG_X86_64 + add_pda(apic_timer_irqs, 1); +#else + per_cpu(irq_stat, cpu).apic_timer_irqs++; +#endif + + evt->event_handler(evt); +} + +/* + * Local APIC timer interrupt. This is the most natural way for doing + * local interrupts, but local timer interrupts can be emulated by + * broadcast interrupts too. [in case the hw doesn't support APIC timers] + * + * [ if a single-CPU system runs an SMP kernel then we call the local + * interrupt as well. Thus we cannot inline the local irq ... ] + */ +void smp_apic_timer_interrupt(struct pt_regs *regs) +{ + struct pt_regs *old_regs = set_irq_regs(regs); + + /* + * NOTE! We'd better ACK the irq immediately, + * because timer handling can be slow. + */ + ack_APIC_irq(); + /* + * update_process_times() expects us to have done irq_enter(). + * Besides, if we don't timer interrupts ignore the global + * interrupt lock, which is the WrongThing (tm) to do. + */ +#ifdef CONFIG_X86_64 + exit_idle(); +#endif + irq_enter(); + local_apic_timer_interrupt(); + irq_exit(); + + set_irq_regs(old_regs); +} + +int setup_profiling_timer(unsigned int multiplier) +{ + return -EINVAL; +} + +/* + * Local APIC start and shutdown + */ + +/** + * clear_local_APIC - shutdown the local APIC + * + * This is called, when a CPU is disabled and before rebooting, so the state of + * the local APIC has no dangling leftovers. Also used to cleanout any BIOS + * leftovers during boot. + */ +void clear_local_APIC(void) +{ + int maxlvt; + u32 v; + + /* APIC hasn't been mapped yet */ + if (!apic_phys) + return; + + maxlvt = lapic_get_maxlvt(); + /* + * Masking an LVT entry can trigger a local APIC error + * if the vector is zero. Mask LVTERR first to prevent this. + */ + if (maxlvt >= 3) { + v = ERROR_APIC_VECTOR; /* any non-zero vector will do */ + apic_write(APIC_LVTERR, v | APIC_LVT_MASKED); + } + /* + * Careful: we have to set masks only first to deassert + * any level-triggered sources. + */ + v = apic_read(APIC_LVTT); + apic_write(APIC_LVTT, v | APIC_LVT_MASKED); + v = apic_read(APIC_LVT0); + apic_write(APIC_LVT0, v | APIC_LVT_MASKED); + v = apic_read(APIC_LVT1); + apic_write(APIC_LVT1, v | APIC_LVT_MASKED); + if (maxlvt >= 4) { + v = apic_read(APIC_LVTPC); + apic_write(APIC_LVTPC, v | APIC_LVT_MASKED); + } + + /* lets not touch this if we didn't frob it */ +#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(X86_MCE_INTEL) + if (maxlvt >= 5) { + v = apic_read(APIC_LVTTHMR); + apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED); + } +#endif + /* + * Clean APIC state for other OSs: + */ + apic_write(APIC_LVTT, APIC_LVT_MASKED); + apic_write(APIC_LVT0, APIC_LVT_MASKED); + apic_write(APIC_LVT1, APIC_LVT_MASKED); + if (maxlvt >= 3) + apic_write(APIC_LVTERR, APIC_LVT_MASKED); + if (maxlvt >= 4) + apic_write(APIC_LVTPC, APIC_LVT_MASKED); + + /* Integrated APIC (!82489DX) ? */ + if (lapic_is_integrated()) { + if (maxlvt > 3) + /* Clear ESR due to Pentium errata 3AP and 11AP */ + apic_write(APIC_ESR, 0); + apic_read(APIC_ESR); + } +} + +/** + * disable_local_APIC - clear and disable the local APIC + */ +void disable_local_APIC(void) +{ + unsigned int value; + + clear_local_APIC(); + + /* + * Disable APIC (implies clearing of registers + * for 82489DX!). + */ + value = apic_read(APIC_SPIV); + value &= ~APIC_SPIV_APIC_ENABLED; + apic_write(APIC_SPIV, value); + +#ifdef CONFIG_X86_32 + /* + * When LAPIC was disabled by the BIOS and enabled by the kernel, + * restore the disabled state. + */ + if (enabled_via_apicbase) { + unsigned int l, h; + + rdmsr(MSR_IA32_APICBASE, l, h); + l &= ~MSR_IA32_APICBASE_ENABLE; + wrmsr(MSR_IA32_APICBASE, l, h); + } +#endif +} + +/* + * If Linux enabled the LAPIC against the BIOS default disable it down before + * re-entering the BIOS on shutdown. Otherwise the BIOS may get confused and + * not power-off. Additionally clear all LVT entries before disable_local_APIC + * for the case where Linux didn't enable the LAPIC. + */ +void lapic_shutdown(void) +{ + unsigned long flags; + + if (!cpu_has_apic) + return; + + local_irq_save(flags); + +#ifdef CONFIG_X86_32 + if (!enabled_via_apicbase) + clear_local_APIC(); + else +#endif + disable_local_APIC(); + + + local_irq_restore(flags); +} + +/* + * This is to verify that we're looking at a real local APIC. + * Check these against your board if the CPUs aren't getting + * started for no apparent reason. + */ +int __init verify_local_APIC(void) +{ + unsigned int reg0, reg1; + + /* + * The version register is read-only in a real APIC. + */ + reg0 = apic_read(APIC_LVR); + apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg0); + apic_write(APIC_LVR, reg0 ^ APIC_LVR_MASK); + reg1 = apic_read(APIC_LVR); + apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg1); + + /* + * The two version reads above should print the same + * numbers. If the second one is different, then we + * poke at a non-APIC. + */ + if (reg1 != reg0) + return 0; + + /* + * Check if the version looks reasonably. + */ + reg1 = GET_APIC_VERSION(reg0); + if (reg1 == 0x00 || reg1 == 0xff) + return 0; + reg1 = lapic_get_maxlvt(); + if (reg1 < 0x02 || reg1 == 0xff) + return 0; + + /* + * The ID register is read/write in a real APIC. + */ + reg0 = apic_read(APIC_ID); + apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg0); + apic_write(APIC_ID, reg0 ^ APIC_ID_MASK); + reg1 = apic_read(APIC_ID); + apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg1); + apic_write(APIC_ID, reg0); + if (reg1 != (reg0 ^ APIC_ID_MASK)) + return 0; + + /* + * The next two are just to see if we have sane values. + * They're only really relevant if we're in Virtual Wire + * compatibility mode, but most boxes are anymore. + */ + reg0 = apic_read(APIC_LVT0); + apic_printk(APIC_DEBUG, "Getting LVT0: %x\n", reg0); + reg1 = apic_read(APIC_LVT1); + apic_printk(APIC_DEBUG, "Getting LVT1: %x\n", reg1); + + return 1; +} + +/** + * sync_Arb_IDs - synchronize APIC bus arbitration IDs + */ +void __init sync_Arb_IDs(void) +{ + /* + * Unsupported on P4 - see Intel Dev. Manual Vol. 3, Ch. 8.6.1 And not + * needed on AMD. + */ + if (modern_apic() || boot_cpu_data.x86_vendor == X86_VENDOR_AMD) + return; + + /* + * Wait for idle. + */ + apic_wait_icr_idle(); + + apic_printk(APIC_DEBUG, "Synchronizing Arb IDs.\n"); + apic_write(APIC_ICR, APIC_DEST_ALLINC | + APIC_INT_LEVELTRIG | APIC_DM_INIT); +} + +/* + * An initial setup of the virtual wire mode. + */ +void __init init_bsp_APIC(void) +{ + unsigned int value; + + /* + * Don't do the setup now if we have a SMP BIOS as the + * through-I/O-APIC virtual wire mode might be active. + */ + if (smp_found_config || !cpu_has_apic) + return; + + /* + * Do not trust the local APIC being empty at bootup. + */ + clear_local_APIC(); + + /* + * Enable APIC. + */ + value = apic_read(APIC_SPIV); + value &= ~APIC_VECTOR_MASK; + value |= APIC_SPIV_APIC_ENABLED; + +#ifdef CONFIG_X86_32 + /* This bit is reserved on P4/Xeon and should be cleared */ + if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) && + (boot_cpu_data.x86 == 15)) + value &= ~APIC_SPIV_FOCUS_DISABLED; + else +#endif + value |= APIC_SPIV_FOCUS_DISABLED; + value |= SPURIOUS_APIC_VECTOR; + apic_write(APIC_SPIV, value); + + /* + * Set up the virtual wire mode. + */ + apic_write(APIC_LVT0, APIC_DM_EXTINT); + value = APIC_DM_NMI; + if (!lapic_is_integrated()) /* 82489DX */ + value |= APIC_LVT_LEVEL_TRIGGER; + apic_write(APIC_LVT1, value); +} + +static void __cpuinit lapic_setup_esr(void) +{ + unsigned long oldvalue, value, maxlvt; + if (lapic_is_integrated() && !esr_disable) { + if (esr_disable) { + /* + * Something untraceable is creating bad interrupts on + * secondary quads ... for the moment, just leave the + * ESR disabled - we can't do anything useful with the + * errors anyway - mbligh + */ + printk(KERN_INFO "Leaving ESR disabled.\n"); + return; + } + /* !82489DX */ + maxlvt = lapic_get_maxlvt(); + if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ + apic_write(APIC_ESR, 0); + oldvalue = apic_read(APIC_ESR); + + /* enables sending errors */ + value = ERROR_APIC_VECTOR; + apic_write(APIC_LVTERR, value); + /* + * spec says clear errors after enabling vector. + */ + if (maxlvt > 3) + apic_write(APIC_ESR, 0); + value = apic_read(APIC_ESR); + if (value != oldvalue) + apic_printk(APIC_VERBOSE, "ESR value before enabling " + "vector: 0x%08lx after: 0x%08lx\n", + oldvalue, value); + } else { + printk(KERN_INFO "No ESR for 82489DX.\n"); + } +} + + +/** + * setup_local_APIC - setup the local APIC + */ +void __cpuinit setup_local_APIC(void) +{ + unsigned int value; + int i, j; + +#ifdef CONFIG_X86_32 + /* Pound the ESR really hard over the head with a big hammer - mbligh */ + if (esr_disable) { + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + apic_write(APIC_ESR, 0); + } +#endif + + preempt_disable(); + + /* + * Double-check whether this APIC is really registered. + * This is meaningless in clustered apic mode, so we skip it. + */ + if (!apic_id_registered()) + BUG(); + + /* + * Intel recommends to set DFR, LDR and TPR before enabling + * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel + * document number 292116). So here it goes... + */ + init_apic_ldr(); + + /* + * Set Task Priority to 'accept all'. We never change this + * later on. + */ + value = apic_read(APIC_TASKPRI); + value &= ~APIC_TPRI_MASK; + apic_write(APIC_TASKPRI, value); + + /* + * After a crash, we no longer service the interrupts and a pending + * interrupt from previous kernel might still have ISR bit set. + * + * Most probably by now CPU has serviced that pending interrupt and + * it might not have done the ack_APIC_irq() because it thought, + * interrupt came from i8259 as ExtInt. LAPIC did not get EOI so it + * does not clear the ISR bit and cpu thinks it has already serivced + * the interrupt. Hence a vector might get locked. It was noticed + * for timer irq (vector 0x31). Issue an extra EOI to clear ISR. + */ + for (i = APIC_ISR_NR - 1; i >= 0; i--) { + value = apic_read(APIC_ISR + i*0x10); + for (j = 31; j >= 0; j--) { + if (value & (1< 1) || + (boot_cpu_data.x86 == 15)) + break; + goto no_apic; + case X86_VENDOR_INTEL: + if (boot_cpu_data.x86 == 6 || boot_cpu_data.x86 == 15 || + (boot_cpu_data.x86 == 5 && cpu_has_apic)) + break; + goto no_apic; + default: + goto no_apic; + } + + if (!cpu_has_apic) { + /* + * Over-ride BIOS and try to enable the local APIC only if + * "lapic" specified. + */ + if (!force_enable_local_apic) { + printk(KERN_INFO "Local APIC disabled by BIOS -- " + "you can enable it with \"lapic\"\n"); + return -1; + } + /* + * Some BIOSes disable the local APIC in the APIC_BASE + * MSR. This can only be done in software for Intel P6 or later + * and AMD K7 (Model > 1) or later. + */ + rdmsr(MSR_IA32_APICBASE, l, h); + if (!(l & MSR_IA32_APICBASE_ENABLE)) { + printk(KERN_INFO + "Local APIC disabled by BIOS -- reenabling.\n"); + l &= ~MSR_IA32_APICBASE_BASE; + l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; + wrmsr(MSR_IA32_APICBASE, l, h); + enabled_via_apicbase = 1; + } + } + /* + * The APIC feature bit should now be enabled + * in `cpuid' + */ + features = cpuid_edx(1); + if (!(features & (1 << X86_FEATURE_APIC))) { + printk(KERN_WARNING "Could not enable APIC!\n"); + return -1; + } + set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); + mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; + + /* The BIOS may have set up the APIC at some other address */ + rdmsr(MSR_IA32_APICBASE, l, h); + if (l & MSR_IA32_APICBASE_ENABLE) + mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; + + printk(KERN_INFO "Found and enabled local APIC!\n"); + + apic_pm_activate(); + + return 0; + +no_apic: + printk(KERN_INFO "No local APIC present or hardware disabled\n"); + return -1; +} +#endif + +#ifdef CONFIG_X86_64 +void __init early_init_lapic_mapping(void) +{ + unsigned long phys_addr; + + /* + * If no local APIC can be found then go out + * : it means there is no mpatable and MADT + */ + if (!smp_found_config) + return; + + phys_addr = mp_lapic_addr; + + set_fixmap_nocache(FIX_APIC_BASE, phys_addr); + apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", + APIC_BASE, phys_addr); + + /* + * Fetch the APIC ID of the BSP in case we have a + * default configuration (or the MP table is broken). + */ + boot_cpu_physical_apicid = read_apic_id(); +} +#endif + +/** + * init_apic_mappings - initialize APIC mappings + */ +void __init init_apic_mappings(void) +{ +#ifdef HAVE_X2APIC + if (x2apic) { + boot_cpu_physical_apicid = read_apic_id(); + return; + } +#endif + + /* + * If no local APIC can be found then set up a fake all + * zeroes page to simulate the local APIC and another + * one for the IO-APIC. + */ + if (!smp_found_config && detect_init_APIC()) { + apic_phys = (unsigned long) alloc_bootmem_pages(PAGE_SIZE); + apic_phys = __pa(apic_phys); + } else + apic_phys = mp_lapic_addr; + + set_fixmap_nocache(FIX_APIC_BASE, apic_phys); + apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", + APIC_BASE, apic_phys); + + /* + * Fetch the APIC ID of the BSP in case we have a + * default configuration (or the MP table is broken). + */ + if (boot_cpu_physical_apicid == -1U) + boot_cpu_physical_apicid = read_apic_id(); +} + +/* + * This initializes the IO-APIC and APIC hardware if this is + * a UP kernel. + */ +int apic_version[MAX_APICS]; + +int __init APIC_init_uniprocessor(void) +{ +#ifdef CONFIG_X86_64 + if (disable_apic) { + printk(KERN_INFO "Apic disabled\n"); + return -1; + } + if (!cpu_has_apic) { + disable_apic = 1; + printk(KERN_INFO "Apic disabled by BIOS\n"); + return -1; + } +#else + if (!smp_found_config && !cpu_has_apic) + return -1; + + /* + * Complain if the BIOS pretends there is one. + */ + if (!cpu_has_apic && + APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { + printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", + boot_cpu_physical_apicid); + clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); + return -1; + } +#endif + +#ifdef HAVE_X2APIC + enable_IR_x2apic(); +#endif +#ifdef CONFIG_X86_64 + setup_apic_routing(); +#endif + + verify_local_APIC(); + connect_bsp_APIC(); + +#ifdef CONFIG_X86_64 + apic_write(APIC_ID, SET_APIC_ID(boot_cpu_physical_apicid)); +#else + /* + * Hack: In case of kdump, after a crash, kernel might be booting + * on a cpu with non-zero lapic id. But boot_cpu_physical_apicid + * might be zero if read from MP tables. Get it from LAPIC. + */ +# ifdef CONFIG_CRASH_DUMP + boot_cpu_physical_apicid = read_apic_id(); +# endif +#endif + physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); + setup_local_APIC(); + +#ifdef CONFIG_X86_64 + /* + * Now enable IO-APICs, actually call clear_IO_APIC + * We need clear_IO_APIC before enabling vector on BP + */ + if (!skip_ioapic_setup && nr_ioapics) + enable_IO_APIC(); +#endif + +#ifdef CONFIG_X86_IO_APIC + if (!smp_found_config || skip_ioapic_setup || !nr_ioapics) +#endif + localise_nmi_watchdog(); + end_local_APIC_setup(); + +#ifdef CONFIG_X86_IO_APIC + if (smp_found_config && !skip_ioapic_setup && nr_ioapics) + setup_IO_APIC(); +# ifdef CONFIG_X86_64 + else + nr_ioapics = 0; +# endif +#endif + +#ifdef CONFIG_X86_64 + setup_boot_APIC_clock(); + check_nmi_watchdog(); +#else + setup_boot_clock(); +#endif + + return 0; +} + +/* + * Local APIC interrupts + */ + +/* + * This interrupt should _never_ happen with our APIC/SMP architecture + */ +#ifdef CONFIG_X86_64 +asmlinkage void smp_spurious_interrupt(void) +#else +void smp_spurious_interrupt(struct pt_regs *regs) +#endif +{ + u32 v; + +#ifdef CONFIG_X86_64 + exit_idle(); +#endif + irq_enter(); + /* + * Check if this really is a spurious interrupt and ACK it + * if it is a vectored one. Just in case... + * Spurious interrupts should not be ACKed. + */ + v = apic_read(APIC_ISR + ((SPURIOUS_APIC_VECTOR & ~0x1f) >> 1)); + if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) + ack_APIC_irq(); + +#ifdef CONFIG_X86_64 + add_pda(irq_spurious_count, 1); +#else + /* see sw-dev-man vol 3, chapter 7.4.13.5 */ + printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " + "should never happen.\n", smp_processor_id()); + __get_cpu_var(irq_stat).irq_spurious_count++; +#endif + irq_exit(); +} + +/* + * This interrupt should never happen with our APIC/SMP architecture + */ +#ifdef CONFIG_X86_64 +asmlinkage void smp_error_interrupt(void) +#else +void smp_error_interrupt(struct pt_regs *regs) +#endif +{ + u32 v, v1; + +#ifdef CONFIG_X86_64 + exit_idle(); +#endif + irq_enter(); + /* First tickle the hardware, only then report what went on. -- REW */ + v = apic_read(APIC_ESR); + apic_write(APIC_ESR, 0); + v1 = apic_read(APIC_ESR); + ack_APIC_irq(); + atomic_inc(&irq_err_count); + + /* Here is what the APIC error bits mean: + 0: Send CS error + 1: Receive CS error + 2: Send accept error + 3: Receive accept error + 4: Reserved + 5: Send illegal vector + 6: Received illegal vector + 7: Illegal register address + */ + printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", + smp_processor_id(), v , v1); + irq_exit(); +} + +/** + * connect_bsp_APIC - attach the APIC to the interrupt system + */ +void __init connect_bsp_APIC(void) +{ +#ifdef CONFIG_X86_32 + if (pic_mode) { + /* + * Do not trust the local APIC being empty at bootup. + */ + clear_local_APIC(); + /* + * PIC mode, enable APIC mode in the IMCR, i.e. connect BSP's + * local APIC to INT and NMI lines. + */ + apic_printk(APIC_VERBOSE, "leaving PIC mode, " + "enabling APIC mode.\n"); + outb(0x70, 0x22); + outb(0x01, 0x23); + } +#endif + enable_apic_mode(); +} + +/** + * disconnect_bsp_APIC - detach the APIC from the interrupt system + * @virt_wire_setup: indicates, whether virtual wire mode is selected + * + * Virtual wire mode is necessary to deliver legacy interrupts even when the + * APIC is disabled. + */ +void disconnect_bsp_APIC(int virt_wire_setup) +{ + unsigned int value; + +#ifdef CONFIG_X86_32 + if (pic_mode) { + /* + * Put the board back into PIC mode (has an effect only on + * certain older boards). Note that APIC interrupts, including + * IPIs, won't work beyond this point! The only exception are + * INIT IPIs. + */ + apic_printk(APIC_VERBOSE, "disabling APIC mode, " + "entering PIC mode.\n"); + outb(0x70, 0x22); + outb(0x00, 0x23); + return; + } +#endif + + /* Go back to Virtual Wire compatibility mode */ + + /* For the spurious interrupt use vector F, and enable it */ + value = apic_read(APIC_SPIV); + value &= ~APIC_VECTOR_MASK; + value |= APIC_SPIV_APIC_ENABLED; + value |= 0xf; + apic_write(APIC_SPIV, value); + + if (!virt_wire_setup) { + /* + * For LVT0 make it edge triggered, active high, + * external and enabled + */ + value = apic_read(APIC_LVT0); + value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | + APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | + APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); + value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; + value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_EXTINT); + apic_write(APIC_LVT0, value); + } else { + /* Disable LVT0 */ + apic_write(APIC_LVT0, APIC_LVT_MASKED); + } + + /* + * For LVT1 make it edge triggered, active high, + * nmi and enabled + */ + value = apic_read(APIC_LVT1); + value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | + APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | + APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); + value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; + value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_NMI); + apic_write(APIC_LVT1, value); +} + +void __cpuinit generic_processor_info(int apicid, int version) +{ + int cpu; + cpumask_t tmp_map; + + /* + * Validate version + */ + if (version == 0x0) { + printk(KERN_WARNING "BIOS bug, APIC version is 0 for CPU#%d! " + "fixing up to 0x10. (tell your hw vendor)\n", + version); + version = 0x10; + } + apic_version[apicid] = version; + + if (num_processors >= NR_CPUS) { + printk(KERN_WARNING "WARNING: NR_CPUS limit of %i reached." + " Processor ignored.\n", NR_CPUS); + return; + } + + num_processors++; + cpus_complement(tmp_map, cpu_present_map); + cpu = first_cpu(tmp_map); + + physid_set(apicid, phys_cpu_present_map); + if (apicid == boot_cpu_physical_apicid) { + /* + * x86_bios_cpu_apicid is required to have processors listed + * in same order as logical cpu numbers. Hence the first + * entry is BSP, and so on. + */ + cpu = 0; + } + if (apicid > max_physical_apicid) + max_physical_apicid = apicid; + +#ifdef CONFIG_X86_32 + /* + * Would be preferable to switch to bigsmp when CONFIG_HOTPLUG_CPU=y + * but we need to work other dependencies like SMP_SUSPEND etc + * before this can be done without some confusion. + * if (CPU_HOTPLUG_ENABLED || num_processors > 8) + * - Ashok Raj + */ + if (max_physical_apicid >= 8) { + switch (boot_cpu_data.x86_vendor) { + case X86_VENDOR_INTEL: + if (!APIC_XAPIC(version)) { + def_to_bigsmp = 0; + break; + } + /* If P4 and above fall through */ + case X86_VENDOR_AMD: + def_to_bigsmp = 1; + } + } +#endif + +#if defined(CONFIG_X86_SMP) || defined(CONFIG_X86_64) + /* are we being called early in kernel startup? */ + if (early_per_cpu_ptr(x86_cpu_to_apicid)) { + u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid); + u16 *bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); + + cpu_to_apicid[cpu] = apicid; + bios_cpu_apicid[cpu] = apicid; + } else { + per_cpu(x86_cpu_to_apicid, cpu) = apicid; + per_cpu(x86_bios_cpu_apicid, cpu) = apicid; + } +#endif + + cpu_set(cpu, cpu_possible_map); + cpu_set(cpu, cpu_present_map); +} + +#ifdef CONFIG_X86_64 +int hard_smp_processor_id(void) +{ + return read_apic_id(); +} +#endif + +/* + * Power management + */ +#ifdef CONFIG_PM + +static struct { + /* + * 'active' is true if the local APIC was enabled by us and + * not the BIOS; this signifies that we are also responsible + * for disabling it before entering apm/acpi suspend + */ + int active; + /* r/w apic fields */ + unsigned int apic_id; + unsigned int apic_taskpri; + unsigned int apic_ldr; + unsigned int apic_dfr; + unsigned int apic_spiv; + unsigned int apic_lvtt; + unsigned int apic_lvtpc; + unsigned int apic_lvt0; + unsigned int apic_lvt1; + unsigned int apic_lvterr; + unsigned int apic_tmict; + unsigned int apic_tdcr; + unsigned int apic_thmr; +} apic_pm_state; + +static int lapic_suspend(struct sys_device *dev, pm_message_t state) +{ + unsigned long flags; + int maxlvt; + + if (!apic_pm_state.active) + return 0; + + maxlvt = lapic_get_maxlvt(); + + apic_pm_state.apic_id = apic_read(APIC_ID); + apic_pm_state.apic_taskpri = apic_read(APIC_TASKPRI); + apic_pm_state.apic_ldr = apic_read(APIC_LDR); + apic_pm_state.apic_dfr = apic_read(APIC_DFR); + apic_pm_state.apic_spiv = apic_read(APIC_SPIV); + apic_pm_state.apic_lvtt = apic_read(APIC_LVTT); + if (maxlvt >= 4) + apic_pm_state.apic_lvtpc = apic_read(APIC_LVTPC); + apic_pm_state.apic_lvt0 = apic_read(APIC_LVT0); + apic_pm_state.apic_lvt1 = apic_read(APIC_LVT1); + apic_pm_state.apic_lvterr = apic_read(APIC_LVTERR); + apic_pm_state.apic_tmict = apic_read(APIC_TMICT); + apic_pm_state.apic_tdcr = apic_read(APIC_TDCR); +#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) + if (maxlvt >= 5) + apic_pm_state.apic_thmr = apic_read(APIC_LVTTHMR); +#endif + + local_irq_save(flags); + disable_local_APIC(); + local_irq_restore(flags); + return 0; +} + +static int lapic_resume(struct sys_device *dev) +{ + unsigned int l, h; + unsigned long flags; + int maxlvt; + + if (!apic_pm_state.active) + return 0; + + maxlvt = lapic_get_maxlvt(); + + local_irq_save(flags); + +#ifdef HAVE_X2APIC + if (x2apic) + enable_x2apic(); + else +#endif + { + /* + * Make sure the APICBASE points to the right address + * + * FIXME! This will be wrong if we ever support suspend on + * SMP! We'll need to do this as part of the CPU restore! + */ + rdmsr(MSR_IA32_APICBASE, l, h); + l &= ~MSR_IA32_APICBASE_BASE; + l |= MSR_IA32_APICBASE_ENABLE | mp_lapic_addr; + wrmsr(MSR_IA32_APICBASE, l, h); + } + + apic_write(APIC_LVTERR, ERROR_APIC_VECTOR | APIC_LVT_MASKED); + apic_write(APIC_ID, apic_pm_state.apic_id); + apic_write(APIC_DFR, apic_pm_state.apic_dfr); + apic_write(APIC_LDR, apic_pm_state.apic_ldr); + apic_write(APIC_TASKPRI, apic_pm_state.apic_taskpri); + apic_write(APIC_SPIV, apic_pm_state.apic_spiv); + apic_write(APIC_LVT0, apic_pm_state.apic_lvt0); + apic_write(APIC_LVT1, apic_pm_state.apic_lvt1); +#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) + if (maxlvt >= 5) + apic_write(APIC_LVTTHMR, apic_pm_state.apic_thmr); +#endif + if (maxlvt >= 4) + apic_write(APIC_LVTPC, apic_pm_state.apic_lvtpc); + apic_write(APIC_LVTT, apic_pm_state.apic_lvtt); + apic_write(APIC_TDCR, apic_pm_state.apic_tdcr); + apic_write(APIC_TMICT, apic_pm_state.apic_tmict); + apic_write(APIC_ESR, 0); + apic_read(APIC_ESR); + apic_write(APIC_LVTERR, apic_pm_state.apic_lvterr); + apic_write(APIC_ESR, 0); + apic_read(APIC_ESR); + + local_irq_restore(flags); + + return 0; +} + +/* + * This device has no shutdown method - fully functioning local APICs + * are needed on every CPU up until machine_halt/restart/poweroff. + */ + +static struct sysdev_class lapic_sysclass = { + .name = "lapic", + .resume = lapic_resume, + .suspend = lapic_suspend, +}; + +static struct sys_device device_lapic = { + .id = 0, + .cls = &lapic_sysclass, +}; + +static void __cpuinit apic_pm_activate(void) +{ + apic_pm_state.active = 1; +} + +static int __init init_lapic_sysfs(void) +{ + int error; + + if (!cpu_has_apic) + return 0; + /* XXX: remove suspend/resume procs if !apic_pm_state.active? */ + + error = sysdev_class_register(&lapic_sysclass); + if (!error) + error = sysdev_register(&device_lapic); + return error; +} +device_initcall(init_lapic_sysfs); + +#else /* CONFIG_PM */ + +static void apic_pm_activate(void) { } + +#endif /* CONFIG_PM */ + +#ifdef CONFIG_X86_64 +/* + * apic_is_clustered_box() -- Check if we can expect good TSC + * + * Thus far, the major user of this is IBM's Summit2 series: + * + * Clustered boxes may have unsynced TSC problems if they are + * multi-chassis. Use available data to take a good guess. + * If in doubt, go HPET. + */ +__cpuinit int apic_is_clustered_box(void) +{ + int i, clusters, zeros; + unsigned id; + u16 *bios_cpu_apicid; + DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS); + + /* + * there is not this kind of box with AMD CPU yet. + * Some AMD box with quadcore cpu and 8 sockets apicid + * will be [4, 0x23] or [8, 0x27] could be thought to + * vsmp box still need checking... + */ + if ((boot_cpu_data.x86_vendor == X86_VENDOR_AMD) && !is_vsmp_box()) + return 0; + + bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); + bitmap_zero(clustermap, NUM_APIC_CLUSTERS); + + for (i = 0; i < NR_CPUS; i++) { + /* are we being called early in kernel startup? */ + if (bios_cpu_apicid) { + id = bios_cpu_apicid[i]; + } + else if (i < nr_cpu_ids) { + if (cpu_present(i)) + id = per_cpu(x86_bios_cpu_apicid, i); + else + continue; + } + else + break; + + if (id != BAD_APICID) + __set_bit(APIC_CLUSTERID(id), clustermap); + } + + /* Problem: Partially populated chassis may not have CPUs in some of + * the APIC clusters they have been allocated. Only present CPUs have + * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. + * Since clusters are allocated sequentially, count zeros only if + * they are bounded by ones. + */ + clusters = 0; + zeros = 0; + for (i = 0; i < NUM_APIC_CLUSTERS; i++) { + if (test_bit(i, clustermap)) { + clusters += 1 + zeros; + zeros = 0; + } else + ++zeros; + } + + /* ScaleMP vSMPowered boxes have one cluster per board and TSCs are + * not guaranteed to be synced between boards + */ + if (is_vsmp_box() && clusters > 1) + return 1; + + /* + * If clusters > 2, then should be multi-chassis. + * May have to revisit this when multi-core + hyperthreaded CPUs come + * out, but AFAIK this will work even for them. + */ + return (clusters > 2); +} +#endif + +/* + * APIC command line parameters + */ +static int __init setup_disableapic(char *arg) +{ + disable_apic = 1; + setup_clear_cpu_cap(X86_FEATURE_APIC); + return 0; +} +early_param("disableapic", setup_disableapic); + +/* same as disableapic, for compatibility */ +static int __init setup_nolapic(char *arg) +{ + return setup_disableapic(arg); +} +early_param("nolapic", setup_nolapic); + +static int __init parse_lapic_timer_c2_ok(char *arg) +{ + local_apic_timer_c2_ok = 1; + return 0; +} +early_param("lapic_timer_c2_ok", parse_lapic_timer_c2_ok); + +static int __init parse_disable_apic_timer(char *arg) +{ + disable_apic_timer = 1; + return 0; +} +early_param("noapictimer", parse_disable_apic_timer); + +static int __init parse_nolapic_timer(char *arg) +{ + disable_apic_timer = 1; + return 0; +} +early_param("nolapic_timer", parse_nolapic_timer); + +static int __init apic_set_verbosity(char *arg) +{ + if (!arg) { +#ifdef CONFIG_X86_64 + skip_ioapic_setup = 0; + return 0; +#endif + return -EINVAL; + } + + if (strcmp("debug", arg) == 0) + apic_verbosity = APIC_DEBUG; + else if (strcmp("verbose", arg) == 0) + apic_verbosity = APIC_VERBOSE; + else { + printk(KERN_WARNING "APIC Verbosity level %s not recognised" + " use apic=verbose or apic=debug\n", arg); + return -EINVAL; + } + + return 0; +} +early_param("apic", apic_set_verbosity); + +static int __init lapic_insert_resource(void) +{ + if (!apic_phys) + return -1; + + /* Put local APIC into the resource map. */ + lapic_resource.start = apic_phys; + lapic_resource.end = lapic_resource.start + PAGE_SIZE - 1; + insert_resource(&iomem_resource, &lapic_resource); + + return 0; +} + +/* + * need call insert after e820_reserve_resources() + * that is using request_resource + */ +late_initcall(lapic_insert_resource); diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c deleted file mode 100644 index 0b11d6e3f091..000000000000 --- a/arch/x86/kernel/apic_32.c +++ /dev/null @@ -1,2312 +0,0 @@ -/* - * Local APIC handling, local APIC timers - * - * (c) 1999, 2000 Ingo Molnar - * - * Fixes - * Maciej W. Rozycki : Bits for genuine 82489DX APICs; - * thanks to Eric Gilmore - * and Rolf G. Tews - * for testing these extensively. - * Maciej W. Rozycki : Various updates and fixes. - * Mikael Pettersson : Power Management for UP-APIC. - * Pavel Machek and - * Mikael Pettersson : PM converted to driver model. - */ - -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include - -/* - * Sanity check - */ -#if ((SPURIOUS_APIC_VECTOR & 0x0F) != 0x0F) -# error SPURIOUS_APIC_VECTOR definition error -#endif - -#ifdef CONFIG_X86_32 -/* - * Knob to control our willingness to enable the local APIC. - * - * +1=force-enable - */ -static int force_enable_local_apic; -/* - * APIC command line parameters - */ -static int __init parse_lapic(char *arg) -{ - force_enable_local_apic = 1; - return 0; -} -early_param("lapic", parse_lapic); -/* Local APIC was disabled by the BIOS and enabled by the kernel */ -static int enabled_via_apicbase; - -#endif - -#ifdef CONFIG_X86_64 -static int apic_calibrate_pmtmr __initdata; -static __init int setup_apicpmtimer(char *s) -{ - apic_calibrate_pmtmr = 1; - notsc_setup(NULL); - return 0; -} -__setup("apicpmtimer", setup_apicpmtimer); -#endif - -#ifdef CONFIG_X86_64 -#define HAVE_X2APIC -#endif - -#ifdef HAVE_X2APIC -int x2apic; -/* x2apic enabled before OS handover */ -int x2apic_preenabled; -int disable_x2apic; -static __init int setup_nox2apic(char *str) -{ - disable_x2apic = 1; - setup_clear_cpu_cap(X86_FEATURE_X2APIC); - return 0; -} -early_param("nox2apic", setup_nox2apic); -#endif - -unsigned long mp_lapic_addr; -int disable_apic; -/* Disable local APIC timer from the kernel commandline or via dmi quirk */ -static int disable_apic_timer __cpuinitdata; -/* Local APIC timer works in C2 */ -int local_apic_timer_c2_ok; -EXPORT_SYMBOL_GPL(local_apic_timer_c2_ok); - -int first_system_vector = 0xfe; - -char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; - -/* - * Debug level, exported for io_apic.c - */ -unsigned int apic_verbosity; - -int pic_mode; - -/* Have we found an MP table */ -int smp_found_config; - -static struct resource lapic_resource = { - .name = "Local APIC", - .flags = IORESOURCE_MEM | IORESOURCE_BUSY, -}; - -static unsigned int calibration_result; - -static int lapic_next_event(unsigned long delta, - struct clock_event_device *evt); -static void lapic_timer_setup(enum clock_event_mode mode, - struct clock_event_device *evt); -static void lapic_timer_broadcast(cpumask_t mask); -static void apic_pm_activate(void); - -/* - * The local apic timer can be used for any function which is CPU local. - */ -static struct clock_event_device lapic_clockevent = { - .name = "lapic", - .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT - | CLOCK_EVT_FEAT_C3STOP | CLOCK_EVT_FEAT_DUMMY, - .shift = 32, - .set_mode = lapic_timer_setup, - .set_next_event = lapic_next_event, - .broadcast = lapic_timer_broadcast, - .rating = 100, - .irq = -1, -}; -static DEFINE_PER_CPU(struct clock_event_device, lapic_events); - -static unsigned long apic_phys; - -/* - * Get the LAPIC version - */ -static inline int lapic_get_version(void) -{ - return GET_APIC_VERSION(apic_read(APIC_LVR)); -} - -/* - * Check, if the APIC is integrated or a separate chip - */ -static inline int lapic_is_integrated(void) -{ -#ifdef CONFIG_X86_64 - return 1; -#else - return APIC_INTEGRATED(lapic_get_version()); -#endif -} - -/* - * Check, whether this is a modern or a first generation APIC - */ -static int modern_apic(void) -{ - /* AMD systems use old APIC versions, so check the CPU */ - if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD && - boot_cpu_data.x86 >= 0xf) - return 1; - return lapic_get_version() >= 0x14; -} - -/* - * Paravirt kernels also might be using these below ops. So we still - * use generic apic_read()/apic_write(), which might be pointing to different - * ops in PARAVIRT case. - */ -void xapic_wait_icr_idle(void) -{ - while (apic_read(APIC_ICR) & APIC_ICR_BUSY) - cpu_relax(); -} - -u32 safe_xapic_wait_icr_idle(void) -{ - u32 send_status; - int timeout; - - timeout = 0; - do { - send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY; - if (!send_status) - break; - udelay(100); - } while (timeout++ < 1000); - - return send_status; -} - -void xapic_icr_write(u32 low, u32 id) -{ - apic_write(APIC_ICR2, SET_APIC_DEST_FIELD(id)); - apic_write(APIC_ICR, low); -} - -u64 xapic_icr_read(void) -{ - u32 icr1, icr2; - - icr2 = apic_read(APIC_ICR2); - icr1 = apic_read(APIC_ICR); - - return icr1 | ((u64)icr2 << 32); -} - -static struct apic_ops xapic_ops = { - .read = native_apic_mem_read, - .write = native_apic_mem_write, - .icr_read = xapic_icr_read, - .icr_write = xapic_icr_write, - .wait_icr_idle = xapic_wait_icr_idle, - .safe_wait_icr_idle = safe_xapic_wait_icr_idle, -}; - -struct apic_ops __read_mostly *apic_ops = &xapic_ops; -EXPORT_SYMBOL_GPL(apic_ops); - -#ifdef HAVE_X2APIC -static void x2apic_wait_icr_idle(void) -{ - /* no need to wait for icr idle in x2apic */ - return; -} - -static u32 safe_x2apic_wait_icr_idle(void) -{ - /* no need to wait for icr idle in x2apic */ - return 0; -} - -void x2apic_icr_write(u32 low, u32 id) -{ - wrmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), ((__u64) id) << 32 | low); -} - -u64 x2apic_icr_read(void) -{ - unsigned long val; - - rdmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), val); - return val; -} - -static struct apic_ops x2apic_ops = { - .read = native_apic_msr_read, - .write = native_apic_msr_write, - .icr_read = x2apic_icr_read, - .icr_write = x2apic_icr_write, - .wait_icr_idle = x2apic_wait_icr_idle, - .safe_wait_icr_idle = safe_x2apic_wait_icr_idle, -}; -#endif - -/** - * enable_NMI_through_LVT0 - enable NMI through local vector table 0 - */ -void __cpuinit enable_NMI_through_LVT0(void) -{ - unsigned int v; - - /* unmask and set to NMI */ - v = APIC_DM_NMI; - - /* Level triggered for 82489DX (32bit mode) */ - if (!lapic_is_integrated()) - v |= APIC_LVT_LEVEL_TRIGGER; - - apic_write(APIC_LVT0, v); -} - -#ifdef CONFIG_X86_32 -/** - * get_physical_broadcast - Get number of physical broadcast IDs - */ -int get_physical_broadcast(void) -{ - return modern_apic() ? 0xff : 0xf; -} -#endif - -/** - * lapic_get_maxlvt - get the maximum number of local vector table entries - */ -int lapic_get_maxlvt(void) -{ - unsigned int v; - - v = apic_read(APIC_LVR); - /* - * - we always have APIC integrated on 64bit mode - * - 82489DXs do not report # of LVT entries - */ - return APIC_INTEGRATED(GET_APIC_VERSION(v)) ? GET_APIC_MAXLVT(v) : 2; -} - -/* - * Local APIC timer - */ - -/* Clock divisor */ -#ifdef CONFG_X86_64 -#define APIC_DIVISOR 1 -#else -#define APIC_DIVISOR 16 -#endif - -/* - * This function sets up the local APIC timer, with a timeout of - * 'clocks' APIC bus clock. During calibration we actually call - * this function twice on the boot CPU, once with a bogus timeout - * value, second time for real. The other (noncalibrating) CPUs - * call this function only once, with the real, calibrated value. - * - * We do reads before writes even if unnecessary, to get around the - * P5 APIC double write bug. - */ -static void __setup_APIC_LVTT(unsigned int clocks, int oneshot, int irqen) -{ - unsigned int lvtt_value, tmp_value; - - lvtt_value = LOCAL_TIMER_VECTOR; - if (!oneshot) - lvtt_value |= APIC_LVT_TIMER_PERIODIC; - if (!lapic_is_integrated()) - lvtt_value |= SET_APIC_TIMER_BASE(APIC_TIMER_BASE_DIV); - - if (!irqen) - lvtt_value |= APIC_LVT_MASKED; - - apic_write(APIC_LVTT, lvtt_value); - - /* - * Divide PICLK by 16 - */ - tmp_value = apic_read(APIC_TDCR); - apic_write(APIC_TDCR, - (tmp_value & ~(APIC_TDR_DIV_1 | APIC_TDR_DIV_TMBASE)) | - APIC_TDR_DIV_16); - - if (!oneshot) - apic_write(APIC_TMICT, clocks / APIC_DIVISOR); -} - -/* - * Setup extended LVT, AMD specific (K8, family 10h) - * - * Vector mappings are hard coded. On K8 only offset 0 (APIC500) and - * MCE interrupts are supported. Thus MCE offset must be set to 0. - * - * If mask=1, the LVT entry does not generate interrupts while mask=0 - * enables the vector. See also the BKDGs. - */ - -#define APIC_EILVT_LVTOFF_MCE 0 -#define APIC_EILVT_LVTOFF_IBS 1 - -static void setup_APIC_eilvt(u8 lvt_off, u8 vector, u8 msg_type, u8 mask) -{ - unsigned long reg = (lvt_off << 4) + APIC_EILVT0; - unsigned int v = (mask << 16) | (msg_type << 8) | vector; - - apic_write(reg, v); -} - -u8 setup_APIC_eilvt_mce(u8 vector, u8 msg_type, u8 mask) -{ - setup_APIC_eilvt(APIC_EILVT_LVTOFF_MCE, vector, msg_type, mask); - return APIC_EILVT_LVTOFF_MCE; -} - -u8 setup_APIC_eilvt_ibs(u8 vector, u8 msg_type, u8 mask) -{ - setup_APIC_eilvt(APIC_EILVT_LVTOFF_IBS, vector, msg_type, mask); - return APIC_EILVT_LVTOFF_IBS; -} -EXPORT_SYMBOL_GPL(setup_APIC_eilvt_ibs); - -/* - * Program the next event, relative to now - */ -static int lapic_next_event(unsigned long delta, - struct clock_event_device *evt) -{ - apic_write(APIC_TMICT, delta); - return 0; -} - -/* - * Setup the lapic timer in periodic or oneshot mode - */ -static void lapic_timer_setup(enum clock_event_mode mode, - struct clock_event_device *evt) -{ - unsigned long flags; - unsigned int v; - - /* Lapic used as dummy for broadcast ? */ - if (evt->features & CLOCK_EVT_FEAT_DUMMY) - return; - - local_irq_save(flags); - - switch (mode) { - case CLOCK_EVT_MODE_PERIODIC: - case CLOCK_EVT_MODE_ONESHOT: - __setup_APIC_LVTT(calibration_result, - mode != CLOCK_EVT_MODE_PERIODIC, 1); - break; - case CLOCK_EVT_MODE_UNUSED: - case CLOCK_EVT_MODE_SHUTDOWN: - v = apic_read(APIC_LVTT); - v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR); - apic_write(APIC_LVTT, v); - break; - case CLOCK_EVT_MODE_RESUME: - /* Nothing to do here */ - break; - } - - local_irq_restore(flags); -} - -/* - * Local APIC timer broadcast function - */ -static void lapic_timer_broadcast(cpumask_t mask) -{ -#ifdef CONFIG_SMP - send_IPI_mask(mask, LOCAL_TIMER_VECTOR); -#endif -} - -/* - * Setup the local APIC timer for this CPU. Copy the initilized values - * of the boot CPU and register the clock event in the framework. - */ -static void __cpuinit setup_APIC_timer(void) -{ - struct clock_event_device *levt = &__get_cpu_var(lapic_events); - - memcpy(levt, &lapic_clockevent, sizeof(*levt)); - levt->cpumask = cpumask_of_cpu(smp_processor_id()); - - clockevents_register_device(levt); -} - -#ifdef CONFIG_X86_64 -/* - * In this function we calibrate APIC bus clocks to the external - * timer. Unfortunately we cannot use jiffies and the timer irq - * to calibrate, since some later bootup code depends on getting - * the first irq? Ugh. - * - * We want to do the calibration only once since we - * want to have local timer irqs syncron. CPUs connected - * by the same APIC bus have the very same bus frequency. - * And we want to have irqs off anyways, no accidental - * APIC irq that way. - */ - -#define TICK_COUNT 100000000 - -static int __init calibrate_APIC_clock(void) -{ - unsigned apic, apic_start; - unsigned long tsc, tsc_start; - int result; - - local_irq_disable(); - - /* - * Put whatever arbitrary (but long enough) timeout - * value into the APIC clock, we just want to get the - * counter running for calibration. - * - * No interrupt enable ! - */ - __setup_APIC_LVTT(250000000, 0, 0); - - apic_start = apic_read(APIC_TMCCT); -#ifdef CONFIG_X86_PM_TIMER - if (apic_calibrate_pmtmr && pmtmr_ioport) { - pmtimer_wait(5000); /* 5ms wait */ - apic = apic_read(APIC_TMCCT); - result = (apic_start - apic) * 1000L / 5; - } else -#endif - { - rdtscll(tsc_start); - - do { - apic = apic_read(APIC_TMCCT); - rdtscll(tsc); - } while ((tsc - tsc_start) < TICK_COUNT && - (apic_start - apic) < TICK_COUNT); - - result = (apic_start - apic) * 1000L * tsc_khz / - (tsc - tsc_start); - } - - local_irq_enable(); - - printk(KERN_DEBUG "APIC timer calibration result %d\n", result); - - printk(KERN_INFO "Detected %d.%03d MHz APIC timer.\n", - result / 1000 / 1000, result / 1000 % 1000); - - /* Calculate the scaled math multiplication factor */ - lapic_clockevent.mult = div_sc(result, NSEC_PER_SEC, - lapic_clockevent.shift); - lapic_clockevent.max_delta_ns = - clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); - lapic_clockevent.min_delta_ns = - clockevent_delta2ns(0xF, &lapic_clockevent); - - calibration_result = (result * APIC_DIVISOR) / HZ; - - /* - * Do a sanity check on the APIC calibration result - */ - if (calibration_result < (1000000 / HZ)) { - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); - return -1; - } - - return 0; -} - -#else -/* - * In this functions we calibrate APIC bus clocks to the external timer. - * - * We want to do the calibration only once since we want to have local timer - * irqs syncron. CPUs connected by the same APIC bus have the very same bus - * frequency. - * - * This was previously done by reading the PIT/HPET and waiting for a wrap - * around to find out, that a tick has elapsed. I have a box, where the PIT - * readout is broken, so it never gets out of the wait loop again. This was - * also reported by others. - * - * Monitoring the jiffies value is inaccurate and the clockevents - * infrastructure allows us to do a simple substitution of the interrupt - * handler. - * - * The calibration routine also uses the pm_timer when possible, as the PIT - * happens to run way too slow (factor 2.3 on my VAIO CoreDuo, which goes - * back to normal later in the boot process). - */ - -#define LAPIC_CAL_LOOPS (HZ/10) - -static __initdata int lapic_cal_loops = -1; -static __initdata long lapic_cal_t1, lapic_cal_t2; -static __initdata unsigned long long lapic_cal_tsc1, lapic_cal_tsc2; -static __initdata unsigned long lapic_cal_pm1, lapic_cal_pm2; -static __initdata unsigned long lapic_cal_j1, lapic_cal_j2; - -/* - * Temporary interrupt handler. - */ -static void __init lapic_cal_handler(struct clock_event_device *dev) -{ - unsigned long long tsc = 0; - long tapic = apic_read(APIC_TMCCT); - unsigned long pm = acpi_pm_read_early(); - - if (cpu_has_tsc) - rdtscll(tsc); - - switch (lapic_cal_loops++) { - case 0: - lapic_cal_t1 = tapic; - lapic_cal_tsc1 = tsc; - lapic_cal_pm1 = pm; - lapic_cal_j1 = jiffies; - break; - - case LAPIC_CAL_LOOPS: - lapic_cal_t2 = tapic; - lapic_cal_tsc2 = tsc; - if (pm < lapic_cal_pm1) - pm += ACPI_PM_OVRRUN; - lapic_cal_pm2 = pm; - lapic_cal_j2 = jiffies; - break; - } -} - -static int __init calibrate_APIC_clock(void) -{ - struct clock_event_device *levt = &__get_cpu_var(lapic_events); - const long pm_100ms = PMTMR_TICKS_PER_SEC/10; - const long pm_thresh = pm_100ms/100; - void (*real_handler)(struct clock_event_device *dev); - unsigned long deltaj; - long delta, deltapm; - int pm_referenced = 0; - - local_irq_disable(); - - /* Replace the global interrupt handler */ - real_handler = global_clock_event->event_handler; - global_clock_event->event_handler = lapic_cal_handler; - - /* - * Setup the APIC counter to 1e9. There is no way the lapic - * can underflow in the 100ms detection time frame - */ - __setup_APIC_LVTT(1000000000, 0, 0); - - /* Let the interrupts run */ - local_irq_enable(); - - while (lapic_cal_loops <= LAPIC_CAL_LOOPS) - cpu_relax(); - - local_irq_disable(); - - /* Restore the real event handler */ - global_clock_event->event_handler = real_handler; - - /* Build delta t1-t2 as apic timer counts down */ - delta = lapic_cal_t1 - lapic_cal_t2; - apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); - - /* Check, if the PM timer is available */ - deltapm = lapic_cal_pm2 - lapic_cal_pm1; - apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); - - if (deltapm) { - unsigned long mult; - u64 res; - - mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); - - if (deltapm > (pm_100ms - pm_thresh) && - deltapm < (pm_100ms + pm_thresh)) { - apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); - } else { - res = (((u64) deltapm) * mult) >> 22; - do_div(res, 1000000); - printk(KERN_WARNING "APIC calibration not consistent " - "with PM Timer: %ldms instead of 100ms\n", - (long)res); - /* Correct the lapic counter value */ - res = (((u64) delta) * pm_100ms); - do_div(res, deltapm); - printk(KERN_INFO "APIC delta adjusted to PM-Timer: " - "%lu (%ld)\n", (unsigned long) res, delta); - delta = (long) res; - } - pm_referenced = 1; - } - - /* Calculate the scaled math multiplication factor */ - lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, - lapic_clockevent.shift); - lapic_clockevent.max_delta_ns = - clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); - lapic_clockevent.min_delta_ns = - clockevent_delta2ns(0xF, &lapic_clockevent); - - calibration_result = (delta * APIC_DIVISOR) / LAPIC_CAL_LOOPS; - - apic_printk(APIC_VERBOSE, "..... delta %ld\n", delta); - apic_printk(APIC_VERBOSE, "..... mult: %ld\n", lapic_clockevent.mult); - apic_printk(APIC_VERBOSE, "..... calibration result: %u\n", - calibration_result); - - if (cpu_has_tsc) { - delta = (long)(lapic_cal_tsc2 - lapic_cal_tsc1); - apic_printk(APIC_VERBOSE, "..... CPU clock speed is " - "%ld.%04ld MHz.\n", - (delta / LAPIC_CAL_LOOPS) / (1000000 / HZ), - (delta / LAPIC_CAL_LOOPS) % (1000000 / HZ)); - } - - apic_printk(APIC_VERBOSE, "..... host bus clock speed is " - "%u.%04u MHz.\n", - calibration_result / (1000000 / HZ), - calibration_result % (1000000 / HZ)); - - /* - * Do a sanity check on the APIC calibration result - */ - if (calibration_result < (1000000 / HZ)) { - local_irq_enable(); - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); - return -1; - } - - levt->features &= ~CLOCK_EVT_FEAT_DUMMY; - - /* We trust the pm timer based calibration */ - if (!pm_referenced) { - apic_printk(APIC_VERBOSE, "... verify APIC timer\n"); - - /* - * Setup the apic timer manually - */ - levt->event_handler = lapic_cal_handler; - lapic_timer_setup(CLOCK_EVT_MODE_PERIODIC, levt); - lapic_cal_loops = -1; - - /* Let the interrupts run */ - local_irq_enable(); - - while (lapic_cal_loops <= LAPIC_CAL_LOOPS) - cpu_relax(); - - local_irq_disable(); - - /* Stop the lapic timer */ - lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); - - local_irq_enable(); - - /* Jiffies delta */ - deltaj = lapic_cal_j2 - lapic_cal_j1; - apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); - - /* Check, if the jiffies result is consistent */ - if (deltaj >= LAPIC_CAL_LOOPS-2 && deltaj <= LAPIC_CAL_LOOPS+2) - apic_printk(APIC_VERBOSE, "... jiffies result ok\n"); - else - levt->features |= CLOCK_EVT_FEAT_DUMMY; - } else - local_irq_enable(); - - if (levt->features & CLOCK_EVT_FEAT_DUMMY) { - printk(KERN_WARNING - "APIC timer disabled due to verification failure.\n"); - return -1; - } - - return 0; -} - -#endif - -/* - * Setup the boot APIC - * - * Calibrate and verify the result. - */ -void __init setup_boot_APIC_clock(void) -{ - /* - * The local apic timer can be disabled via the kernel - * commandline or from the CPU detection code. Register the lapic - * timer as a dummy clock event source on SMP systems, so the - * broadcast mechanism is used. On UP systems simply ignore it. - */ - if (disable_apic_timer) { - printk(KERN_INFO "Disabling APIC timer\n"); - /* No broadcast on UP ! */ - if (num_possible_cpus() > 1) { - lapic_clockevent.mult = 1; - setup_APIC_timer(); - } - return; - } - - apic_printk(APIC_VERBOSE, "Using local APIC timer interrupts.\n" - "calibrating APIC timer ...\n"); - - if (calibrate_APIC_clock()) { - /* No broadcast on UP ! */ - if (num_possible_cpus() > 1) - setup_APIC_timer(); - return; - } - - /* - * If nmi_watchdog is set to IO_APIC, we need the - * PIT/HPET going. Otherwise register lapic as a dummy - * device. - */ - if (nmi_watchdog != NMI_IO_APIC) - lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY; - else - printk(KERN_WARNING "APIC timer registered as dummy," - " due to nmi_watchdog=%d!\n", nmi_watchdog); - - /* Setup the lapic or request the broadcast */ - setup_APIC_timer(); -} - -void __cpuinit setup_secondary_APIC_clock(void) -{ - setup_APIC_timer(); -} - -/* - * The guts of the apic timer interrupt - */ -static void local_apic_timer_interrupt(void) -{ - int cpu = smp_processor_id(); - struct clock_event_device *evt = &per_cpu(lapic_events, cpu); - - /* - * Normally we should not be here till LAPIC has been initialized but - * in some cases like kdump, its possible that there is a pending LAPIC - * timer interrupt from previous kernel's context and is delivered in - * new kernel the moment interrupts are enabled. - * - * Interrupts are enabled early and LAPIC is setup much later, hence - * its possible that when we get here evt->event_handler is NULL. - * Check for event_handler being NULL and discard the interrupt as - * spurious. - */ - if (!evt->event_handler) { - printk(KERN_WARNING - "Spurious LAPIC timer interrupt on cpu %d\n", cpu); - /* Switch it off */ - lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt); - return; - } - - /* - * the NMI deadlock-detector uses this. - */ -#ifdef CONFIG_X86_64 - add_pda(apic_timer_irqs, 1); -#else - per_cpu(irq_stat, cpu).apic_timer_irqs++; -#endif - - evt->event_handler(evt); -} - -/* - * Local APIC timer interrupt. This is the most natural way for doing - * local interrupts, but local timer interrupts can be emulated by - * broadcast interrupts too. [in case the hw doesn't support APIC timers] - * - * [ if a single-CPU system runs an SMP kernel then we call the local - * interrupt as well. Thus we cannot inline the local irq ... ] - */ -void smp_apic_timer_interrupt(struct pt_regs *regs) -{ - struct pt_regs *old_regs = set_irq_regs(regs); - - /* - * NOTE! We'd better ACK the irq immediately, - * because timer handling can be slow. - */ - ack_APIC_irq(); - /* - * update_process_times() expects us to have done irq_enter(). - * Besides, if we don't timer interrupts ignore the global - * interrupt lock, which is the WrongThing (tm) to do. - */ -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - local_apic_timer_interrupt(); - irq_exit(); - - set_irq_regs(old_regs); -} - -int setup_profiling_timer(unsigned int multiplier) -{ - return -EINVAL; -} - -/* - * Local APIC start and shutdown - */ - -/** - * clear_local_APIC - shutdown the local APIC - * - * This is called, when a CPU is disabled and before rebooting, so the state of - * the local APIC has no dangling leftovers. Also used to cleanout any BIOS - * leftovers during boot. - */ -void clear_local_APIC(void) -{ - int maxlvt; - u32 v; - - /* APIC hasn't been mapped yet */ - if (!apic_phys) - return; - - maxlvt = lapic_get_maxlvt(); - /* - * Masking an LVT entry can trigger a local APIC error - * if the vector is zero. Mask LVTERR first to prevent this. - */ - if (maxlvt >= 3) { - v = ERROR_APIC_VECTOR; /* any non-zero vector will do */ - apic_write(APIC_LVTERR, v | APIC_LVT_MASKED); - } - /* - * Careful: we have to set masks only first to deassert - * any level-triggered sources. - */ - v = apic_read(APIC_LVTT); - apic_write(APIC_LVTT, v | APIC_LVT_MASKED); - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v | APIC_LVT_MASKED); - v = apic_read(APIC_LVT1); - apic_write(APIC_LVT1, v | APIC_LVT_MASKED); - if (maxlvt >= 4) { - v = apic_read(APIC_LVTPC); - apic_write(APIC_LVTPC, v | APIC_LVT_MASKED); - } - - /* lets not touch this if we didn't frob it */ -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(X86_MCE_INTEL) - if (maxlvt >= 5) { - v = apic_read(APIC_LVTTHMR); - apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED); - } -#endif - /* - * Clean APIC state for other OSs: - */ - apic_write(APIC_LVTT, APIC_LVT_MASKED); - apic_write(APIC_LVT0, APIC_LVT_MASKED); - apic_write(APIC_LVT1, APIC_LVT_MASKED); - if (maxlvt >= 3) - apic_write(APIC_LVTERR, APIC_LVT_MASKED); - if (maxlvt >= 4) - apic_write(APIC_LVTPC, APIC_LVT_MASKED); - - /* Integrated APIC (!82489DX) ? */ - if (lapic_is_integrated()) { - if (maxlvt > 3) - /* Clear ESR due to Pentium errata 3AP and 11AP */ - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - } -} - -/** - * disable_local_APIC - clear and disable the local APIC - */ -void disable_local_APIC(void) -{ - unsigned int value; - - clear_local_APIC(); - - /* - * Disable APIC (implies clearing of registers - * for 82489DX!). - */ - value = apic_read(APIC_SPIV); - value &= ~APIC_SPIV_APIC_ENABLED; - apic_write(APIC_SPIV, value); - -#ifdef CONFIG_X86_32 - /* - * When LAPIC was disabled by the BIOS and enabled by the kernel, - * restore the disabled state. - */ - if (enabled_via_apicbase) { - unsigned int l, h; - - rdmsr(MSR_IA32_APICBASE, l, h); - l &= ~MSR_IA32_APICBASE_ENABLE; - wrmsr(MSR_IA32_APICBASE, l, h); - } -#endif -} - -/* - * If Linux enabled the LAPIC against the BIOS default disable it down before - * re-entering the BIOS on shutdown. Otherwise the BIOS may get confused and - * not power-off. Additionally clear all LVT entries before disable_local_APIC - * for the case where Linux didn't enable the LAPIC. - */ -void lapic_shutdown(void) -{ - unsigned long flags; - - if (!cpu_has_apic) - return; - - local_irq_save(flags); - -#ifdef CONFIG_X86_32 - if (!enabled_via_apicbase) - clear_local_APIC(); - else -#endif - disable_local_APIC(); - - - local_irq_restore(flags); -} - -/* - * This is to verify that we're looking at a real local APIC. - * Check these against your board if the CPUs aren't getting - * started for no apparent reason. - */ -int __init verify_local_APIC(void) -{ - unsigned int reg0, reg1; - - /* - * The version register is read-only in a real APIC. - */ - reg0 = apic_read(APIC_LVR); - apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg0); - apic_write(APIC_LVR, reg0 ^ APIC_LVR_MASK); - reg1 = apic_read(APIC_LVR); - apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg1); - - /* - * The two version reads above should print the same - * numbers. If the second one is different, then we - * poke at a non-APIC. - */ - if (reg1 != reg0) - return 0; - - /* - * Check if the version looks reasonably. - */ - reg1 = GET_APIC_VERSION(reg0); - if (reg1 == 0x00 || reg1 == 0xff) - return 0; - reg1 = lapic_get_maxlvt(); - if (reg1 < 0x02 || reg1 == 0xff) - return 0; - - /* - * The ID register is read/write in a real APIC. - */ - reg0 = apic_read(APIC_ID); - apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg0); - apic_write(APIC_ID, reg0 ^ APIC_ID_MASK); - reg1 = apic_read(APIC_ID); - apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg1); - apic_write(APIC_ID, reg0); - if (reg1 != (reg0 ^ APIC_ID_MASK)) - return 0; - - /* - * The next two are just to see if we have sane values. - * They're only really relevant if we're in Virtual Wire - * compatibility mode, but most boxes are anymore. - */ - reg0 = apic_read(APIC_LVT0); - apic_printk(APIC_DEBUG, "Getting LVT0: %x\n", reg0); - reg1 = apic_read(APIC_LVT1); - apic_printk(APIC_DEBUG, "Getting LVT1: %x\n", reg1); - - return 1; -} - -/** - * sync_Arb_IDs - synchronize APIC bus arbitration IDs - */ -void __init sync_Arb_IDs(void) -{ - /* - * Unsupported on P4 - see Intel Dev. Manual Vol. 3, Ch. 8.6.1 And not - * needed on AMD. - */ - if (modern_apic() || boot_cpu_data.x86_vendor == X86_VENDOR_AMD) - return; - - /* - * Wait for idle. - */ - apic_wait_icr_idle(); - - apic_printk(APIC_DEBUG, "Synchronizing Arb IDs.\n"); - apic_write(APIC_ICR, APIC_DEST_ALLINC | - APIC_INT_LEVELTRIG | APIC_DM_INIT); -} - -/* - * An initial setup of the virtual wire mode. - */ -void __init init_bsp_APIC(void) -{ - unsigned int value; - - /* - * Don't do the setup now if we have a SMP BIOS as the - * through-I/O-APIC virtual wire mode might be active. - */ - if (smp_found_config || !cpu_has_apic) - return; - - /* - * Do not trust the local APIC being empty at bootup. - */ - clear_local_APIC(); - - /* - * Enable APIC. - */ - value = apic_read(APIC_SPIV); - value &= ~APIC_VECTOR_MASK; - value |= APIC_SPIV_APIC_ENABLED; - -#ifdef CONFIG_X86_32 - /* This bit is reserved on P4/Xeon and should be cleared */ - if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) && - (boot_cpu_data.x86 == 15)) - value &= ~APIC_SPIV_FOCUS_DISABLED; - else -#endif - value |= APIC_SPIV_FOCUS_DISABLED; - value |= SPURIOUS_APIC_VECTOR; - apic_write(APIC_SPIV, value); - - /* - * Set up the virtual wire mode. - */ - apic_write(APIC_LVT0, APIC_DM_EXTINT); - value = APIC_DM_NMI; - if (!lapic_is_integrated()) /* 82489DX */ - value |= APIC_LVT_LEVEL_TRIGGER; - apic_write(APIC_LVT1, value); -} - -static void __cpuinit lapic_setup_esr(void) -{ - unsigned long oldvalue, value, maxlvt; - if (lapic_is_integrated() && !esr_disable) { - if (esr_disable) { - /* - * Something untraceable is creating bad interrupts on - * secondary quads ... for the moment, just leave the - * ESR disabled - we can't do anything useful with the - * errors anyway - mbligh - */ - printk(KERN_INFO "Leaving ESR disabled.\n"); - return; - } - /* !82489DX */ - maxlvt = lapic_get_maxlvt(); - if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ - apic_write(APIC_ESR, 0); - oldvalue = apic_read(APIC_ESR); - - /* enables sending errors */ - value = ERROR_APIC_VECTOR; - apic_write(APIC_LVTERR, value); - /* - * spec says clear errors after enabling vector. - */ - if (maxlvt > 3) - apic_write(APIC_ESR, 0); - value = apic_read(APIC_ESR); - if (value != oldvalue) - apic_printk(APIC_VERBOSE, "ESR value before enabling " - "vector: 0x%08lx after: 0x%08lx\n", - oldvalue, value); - } else { - printk(KERN_INFO "No ESR for 82489DX.\n"); - } -} - - -/** - * setup_local_APIC - setup the local APIC - */ -void __cpuinit setup_local_APIC(void) -{ - unsigned int value; - int i, j; - -#ifdef CONFIG_X86_32 - /* Pound the ESR really hard over the head with a big hammer - mbligh */ - if (esr_disable) { - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - } -#endif - - preempt_disable(); - - /* - * Double-check whether this APIC is really registered. - * This is meaningless in clustered apic mode, so we skip it. - */ - if (!apic_id_registered()) - BUG(); - - /* - * Intel recommends to set DFR, LDR and TPR before enabling - * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel - * document number 292116). So here it goes... - */ - init_apic_ldr(); - - /* - * Set Task Priority to 'accept all'. We never change this - * later on. - */ - value = apic_read(APIC_TASKPRI); - value &= ~APIC_TPRI_MASK; - apic_write(APIC_TASKPRI, value); - - /* - * After a crash, we no longer service the interrupts and a pending - * interrupt from previous kernel might still have ISR bit set. - * - * Most probably by now CPU has serviced that pending interrupt and - * it might not have done the ack_APIC_irq() because it thought, - * interrupt came from i8259 as ExtInt. LAPIC did not get EOI so it - * does not clear the ISR bit and cpu thinks it has already serivced - * the interrupt. Hence a vector might get locked. It was noticed - * for timer irq (vector 0x31). Issue an extra EOI to clear ISR. - */ - for (i = APIC_ISR_NR - 1; i >= 0; i--) { - value = apic_read(APIC_ISR + i*0x10); - for (j = 31; j >= 0; j--) { - if (value & (1< 1) || - (boot_cpu_data.x86 == 15)) - break; - goto no_apic; - case X86_VENDOR_INTEL: - if (boot_cpu_data.x86 == 6 || boot_cpu_data.x86 == 15 || - (boot_cpu_data.x86 == 5 && cpu_has_apic)) - break; - goto no_apic; - default: - goto no_apic; - } - - if (!cpu_has_apic) { - /* - * Over-ride BIOS and try to enable the local APIC only if - * "lapic" specified. - */ - if (!force_enable_local_apic) { - printk(KERN_INFO "Local APIC disabled by BIOS -- " - "you can enable it with \"lapic\"\n"); - return -1; - } - /* - * Some BIOSes disable the local APIC in the APIC_BASE - * MSR. This can only be done in software for Intel P6 or later - * and AMD K7 (Model > 1) or later. - */ - rdmsr(MSR_IA32_APICBASE, l, h); - if (!(l & MSR_IA32_APICBASE_ENABLE)) { - printk(KERN_INFO - "Local APIC disabled by BIOS -- reenabling.\n"); - l &= ~MSR_IA32_APICBASE_BASE; - l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; - wrmsr(MSR_IA32_APICBASE, l, h); - enabled_via_apicbase = 1; - } - } - /* - * The APIC feature bit should now be enabled - * in `cpuid' - */ - features = cpuid_edx(1); - if (!(features & (1 << X86_FEATURE_APIC))) { - printk(KERN_WARNING "Could not enable APIC!\n"); - return -1; - } - set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); - mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; - - /* The BIOS may have set up the APIC at some other address */ - rdmsr(MSR_IA32_APICBASE, l, h); - if (l & MSR_IA32_APICBASE_ENABLE) - mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; - - printk(KERN_INFO "Found and enabled local APIC!\n"); - - apic_pm_activate(); - - return 0; - -no_apic: - printk(KERN_INFO "No local APIC present or hardware disabled\n"); - return -1; -} -#endif - -#ifdef CONFIG_X86_64 -void __init early_init_lapic_mapping(void) -{ - unsigned long phys_addr; - - /* - * If no local APIC can be found then go out - * : it means there is no mpatable and MADT - */ - if (!smp_found_config) - return; - - phys_addr = mp_lapic_addr; - - set_fixmap_nocache(FIX_APIC_BASE, phys_addr); - apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", - APIC_BASE, phys_addr); - - /* - * Fetch the APIC ID of the BSP in case we have a - * default configuration (or the MP table is broken). - */ - boot_cpu_physical_apicid = read_apic_id(); -} -#endif - -/** - * init_apic_mappings - initialize APIC mappings - */ -void __init init_apic_mappings(void) -{ -#ifdef HAVE_X2APIC - if (x2apic) { - boot_cpu_physical_apicid = read_apic_id(); - return; - } -#endif - - /* - * If no local APIC can be found then set up a fake all - * zeroes page to simulate the local APIC and another - * one for the IO-APIC. - */ - if (!smp_found_config && detect_init_APIC()) { - apic_phys = (unsigned long) alloc_bootmem_pages(PAGE_SIZE); - apic_phys = __pa(apic_phys); - } else - apic_phys = mp_lapic_addr; - - set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", - APIC_BASE, apic_phys); - - /* - * Fetch the APIC ID of the BSP in case we have a - * default configuration (or the MP table is broken). - */ - if (boot_cpu_physical_apicid == -1U) - boot_cpu_physical_apicid = read_apic_id(); -} - -/* - * This initializes the IO-APIC and APIC hardware if this is - * a UP kernel. - */ -int apic_version[MAX_APICS]; - -int __init APIC_init_uniprocessor(void) -{ -#ifdef CONFIG_X86_64 - if (disable_apic) { - printk(KERN_INFO "Apic disabled\n"); - return -1; - } - if (!cpu_has_apic) { - disable_apic = 1; - printk(KERN_INFO "Apic disabled by BIOS\n"); - return -1; - } -#else - if (!smp_found_config && !cpu_has_apic) - return -1; - - /* - * Complain if the BIOS pretends there is one. - */ - if (!cpu_has_apic && - APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { - printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", - boot_cpu_physical_apicid); - clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); - return -1; - } -#endif - -#ifdef HAVE_X2APIC - enable_IR_x2apic(); -#endif -#ifdef CONFIG_X86_64 - setup_apic_routing(); -#endif - - verify_local_APIC(); - connect_bsp_APIC(); - -#ifdef CONFIG_X86_64 - apic_write(APIC_ID, SET_APIC_ID(boot_cpu_physical_apicid)); -#else - /* - * Hack: In case of kdump, after a crash, kernel might be booting - * on a cpu with non-zero lapic id. But boot_cpu_physical_apicid - * might be zero if read from MP tables. Get it from LAPIC. - */ -# ifdef CONFIG_CRASH_DUMP - boot_cpu_physical_apicid = read_apic_id(); -# endif -#endif - physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); - setup_local_APIC(); - -#ifdef CONFIG_X86_64 - /* - * Now enable IO-APICs, actually call clear_IO_APIC - * We need clear_IO_APIC before enabling vector on BP - */ - if (!skip_ioapic_setup && nr_ioapics) - enable_IO_APIC(); -#endif - -#ifdef CONFIG_X86_IO_APIC - if (!smp_found_config || skip_ioapic_setup || !nr_ioapics) -#endif - localise_nmi_watchdog(); - end_local_APIC_setup(); - -#ifdef CONFIG_X86_IO_APIC - if (smp_found_config && !skip_ioapic_setup && nr_ioapics) - setup_IO_APIC(); -# ifdef CONFIG_X86_64 - else - nr_ioapics = 0; -# endif -#endif - -#ifdef CONFIG_X86_64 - setup_boot_APIC_clock(); - check_nmi_watchdog(); -#else - setup_boot_clock(); -#endif - - return 0; -} - -/* - * Local APIC interrupts - */ - -/* - * This interrupt should _never_ happen with our APIC/SMP architecture - */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_spurious_interrupt(void) -#else -void smp_spurious_interrupt(struct pt_regs *regs) -#endif -{ - u32 v; - -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - /* - * Check if this really is a spurious interrupt and ACK it - * if it is a vectored one. Just in case... - * Spurious interrupts should not be ACKed. - */ - v = apic_read(APIC_ISR + ((SPURIOUS_APIC_VECTOR & ~0x1f) >> 1)); - if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) - ack_APIC_irq(); - -#ifdef CONFIG_X86_64 - add_pda(irq_spurious_count, 1); -#else - /* see sw-dev-man vol 3, chapter 7.4.13.5 */ - printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " - "should never happen.\n", smp_processor_id()); - __get_cpu_var(irq_stat).irq_spurious_count++; -#endif - irq_exit(); -} - -/* - * This interrupt should never happen with our APIC/SMP architecture - */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_error_interrupt(void) -#else -void smp_error_interrupt(struct pt_regs *regs) -#endif -{ - u32 v, v1; - -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - /* First tickle the hardware, only then report what went on. -- REW */ - v = apic_read(APIC_ESR); - apic_write(APIC_ESR, 0); - v1 = apic_read(APIC_ESR); - ack_APIC_irq(); - atomic_inc(&irq_err_count); - - /* Here is what the APIC error bits mean: - 0: Send CS error - 1: Receive CS error - 2: Send accept error - 3: Receive accept error - 4: Reserved - 5: Send illegal vector - 6: Received illegal vector - 7: Illegal register address - */ - printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", - smp_processor_id(), v , v1); - irq_exit(); -} - -/** - * connect_bsp_APIC - attach the APIC to the interrupt system - */ -void __init connect_bsp_APIC(void) -{ -#ifdef CONFIG_X86_32 - if (pic_mode) { - /* - * Do not trust the local APIC being empty at bootup. - */ - clear_local_APIC(); - /* - * PIC mode, enable APIC mode in the IMCR, i.e. connect BSP's - * local APIC to INT and NMI lines. - */ - apic_printk(APIC_VERBOSE, "leaving PIC mode, " - "enabling APIC mode.\n"); - outb(0x70, 0x22); - outb(0x01, 0x23); - } -#endif - enable_apic_mode(); -} - -/** - * disconnect_bsp_APIC - detach the APIC from the interrupt system - * @virt_wire_setup: indicates, whether virtual wire mode is selected - * - * Virtual wire mode is necessary to deliver legacy interrupts even when the - * APIC is disabled. - */ -void disconnect_bsp_APIC(int virt_wire_setup) -{ - unsigned int value; - -#ifdef CONFIG_X86_32 - if (pic_mode) { - /* - * Put the board back into PIC mode (has an effect only on - * certain older boards). Note that APIC interrupts, including - * IPIs, won't work beyond this point! The only exception are - * INIT IPIs. - */ - apic_printk(APIC_VERBOSE, "disabling APIC mode, " - "entering PIC mode.\n"); - outb(0x70, 0x22); - outb(0x00, 0x23); - return; - } -#endif - - /* Go back to Virtual Wire compatibility mode */ - - /* For the spurious interrupt use vector F, and enable it */ - value = apic_read(APIC_SPIV); - value &= ~APIC_VECTOR_MASK; - value |= APIC_SPIV_APIC_ENABLED; - value |= 0xf; - apic_write(APIC_SPIV, value); - - if (!virt_wire_setup) { - /* - * For LVT0 make it edge triggered, active high, - * external and enabled - */ - value = apic_read(APIC_LVT0); - value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | - APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | - APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); - value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; - value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_EXTINT); - apic_write(APIC_LVT0, value); - } else { - /* Disable LVT0 */ - apic_write(APIC_LVT0, APIC_LVT_MASKED); - } - - /* - * For LVT1 make it edge triggered, active high, - * nmi and enabled - */ - value = apic_read(APIC_LVT1); - value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | - APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | - APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); - value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; - value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_NMI); - apic_write(APIC_LVT1, value); -} - -void __cpuinit generic_processor_info(int apicid, int version) -{ - int cpu; - cpumask_t tmp_map; - - /* - * Validate version - */ - if (version == 0x0) { - printk(KERN_WARNING "BIOS bug, APIC version is 0 for CPU#%d! " - "fixing up to 0x10. (tell your hw vendor)\n", - version); - version = 0x10; - } - apic_version[apicid] = version; - - if (num_processors >= NR_CPUS) { - printk(KERN_WARNING "WARNING: NR_CPUS limit of %i reached." - " Processor ignored.\n", NR_CPUS); - return; - } - - num_processors++; - cpus_complement(tmp_map, cpu_present_map); - cpu = first_cpu(tmp_map); - - physid_set(apicid, phys_cpu_present_map); - if (apicid == boot_cpu_physical_apicid) { - /* - * x86_bios_cpu_apicid is required to have processors listed - * in same order as logical cpu numbers. Hence the first - * entry is BSP, and so on. - */ - cpu = 0; - } - if (apicid > max_physical_apicid) - max_physical_apicid = apicid; - -#ifdef CONFIG_X86_32 - /* - * Would be preferable to switch to bigsmp when CONFIG_HOTPLUG_CPU=y - * but we need to work other dependencies like SMP_SUSPEND etc - * before this can be done without some confusion. - * if (CPU_HOTPLUG_ENABLED || num_processors > 8) - * - Ashok Raj - */ - if (max_physical_apicid >= 8) { - switch (boot_cpu_data.x86_vendor) { - case X86_VENDOR_INTEL: - if (!APIC_XAPIC(version)) { - def_to_bigsmp = 0; - break; - } - /* If P4 and above fall through */ - case X86_VENDOR_AMD: - def_to_bigsmp = 1; - } - } -#endif - -#if defined(CONFIG_X86_SMP) || defined(CONFIG_X86_64) - /* are we being called early in kernel startup? */ - if (early_per_cpu_ptr(x86_cpu_to_apicid)) { - u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid); - u16 *bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); - - cpu_to_apicid[cpu] = apicid; - bios_cpu_apicid[cpu] = apicid; - } else { - per_cpu(x86_cpu_to_apicid, cpu) = apicid; - per_cpu(x86_bios_cpu_apicid, cpu) = apicid; - } -#endif - - cpu_set(cpu, cpu_possible_map); - cpu_set(cpu, cpu_present_map); -} - -#ifdef CONFIG_X86_64 -int hard_smp_processor_id(void) -{ - return read_apic_id(); -} -#endif - -/* - * Power management - */ -#ifdef CONFIG_PM - -static struct { - /* - * 'active' is true if the local APIC was enabled by us and - * not the BIOS; this signifies that we are also responsible - * for disabling it before entering apm/acpi suspend - */ - int active; - /* r/w apic fields */ - unsigned int apic_id; - unsigned int apic_taskpri; - unsigned int apic_ldr; - unsigned int apic_dfr; - unsigned int apic_spiv; - unsigned int apic_lvtt; - unsigned int apic_lvtpc; - unsigned int apic_lvt0; - unsigned int apic_lvt1; - unsigned int apic_lvterr; - unsigned int apic_tmict; - unsigned int apic_tdcr; - unsigned int apic_thmr; -} apic_pm_state; - -static int lapic_suspend(struct sys_device *dev, pm_message_t state) -{ - unsigned long flags; - int maxlvt; - - if (!apic_pm_state.active) - return 0; - - maxlvt = lapic_get_maxlvt(); - - apic_pm_state.apic_id = apic_read(APIC_ID); - apic_pm_state.apic_taskpri = apic_read(APIC_TASKPRI); - apic_pm_state.apic_ldr = apic_read(APIC_LDR); - apic_pm_state.apic_dfr = apic_read(APIC_DFR); - apic_pm_state.apic_spiv = apic_read(APIC_SPIV); - apic_pm_state.apic_lvtt = apic_read(APIC_LVTT); - if (maxlvt >= 4) - apic_pm_state.apic_lvtpc = apic_read(APIC_LVTPC); - apic_pm_state.apic_lvt0 = apic_read(APIC_LVT0); - apic_pm_state.apic_lvt1 = apic_read(APIC_LVT1); - apic_pm_state.apic_lvterr = apic_read(APIC_LVTERR); - apic_pm_state.apic_tmict = apic_read(APIC_TMICT); - apic_pm_state.apic_tdcr = apic_read(APIC_TDCR); -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) - if (maxlvt >= 5) - apic_pm_state.apic_thmr = apic_read(APIC_LVTTHMR); -#endif - - local_irq_save(flags); - disable_local_APIC(); - local_irq_restore(flags); - return 0; -} - -static int lapic_resume(struct sys_device *dev) -{ - unsigned int l, h; - unsigned long flags; - int maxlvt; - - if (!apic_pm_state.active) - return 0; - - maxlvt = lapic_get_maxlvt(); - - local_irq_save(flags); - -#ifdef HAVE_X2APIC - if (x2apic) - enable_x2apic(); - else -#endif - { - /* - * Make sure the APICBASE points to the right address - * - * FIXME! This will be wrong if we ever support suspend on - * SMP! We'll need to do this as part of the CPU restore! - */ - rdmsr(MSR_IA32_APICBASE, l, h); - l &= ~MSR_IA32_APICBASE_BASE; - l |= MSR_IA32_APICBASE_ENABLE | mp_lapic_addr; - wrmsr(MSR_IA32_APICBASE, l, h); - } - - apic_write(APIC_LVTERR, ERROR_APIC_VECTOR | APIC_LVT_MASKED); - apic_write(APIC_ID, apic_pm_state.apic_id); - apic_write(APIC_DFR, apic_pm_state.apic_dfr); - apic_write(APIC_LDR, apic_pm_state.apic_ldr); - apic_write(APIC_TASKPRI, apic_pm_state.apic_taskpri); - apic_write(APIC_SPIV, apic_pm_state.apic_spiv); - apic_write(APIC_LVT0, apic_pm_state.apic_lvt0); - apic_write(APIC_LVT1, apic_pm_state.apic_lvt1); -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) - if (maxlvt >= 5) - apic_write(APIC_LVTTHMR, apic_pm_state.apic_thmr); -#endif - if (maxlvt >= 4) - apic_write(APIC_LVTPC, apic_pm_state.apic_lvtpc); - apic_write(APIC_LVTT, apic_pm_state.apic_lvtt); - apic_write(APIC_TDCR, apic_pm_state.apic_tdcr); - apic_write(APIC_TMICT, apic_pm_state.apic_tmict); - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - apic_write(APIC_LVTERR, apic_pm_state.apic_lvterr); - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - - local_irq_restore(flags); - - return 0; -} - -/* - * This device has no shutdown method - fully functioning local APICs - * are needed on every CPU up until machine_halt/restart/poweroff. - */ - -static struct sysdev_class lapic_sysclass = { - .name = "lapic", - .resume = lapic_resume, - .suspend = lapic_suspend, -}; - -static struct sys_device device_lapic = { - .id = 0, - .cls = &lapic_sysclass, -}; - -static void __cpuinit apic_pm_activate(void) -{ - apic_pm_state.active = 1; -} - -static int __init init_lapic_sysfs(void) -{ - int error; - - if (!cpu_has_apic) - return 0; - /* XXX: remove suspend/resume procs if !apic_pm_state.active? */ - - error = sysdev_class_register(&lapic_sysclass); - if (!error) - error = sysdev_register(&device_lapic); - return error; -} -device_initcall(init_lapic_sysfs); - -#else /* CONFIG_PM */ - -static void apic_pm_activate(void) { } - -#endif /* CONFIG_PM */ - -#ifdef CONFIG_X86_64 -/* - * apic_is_clustered_box() -- Check if we can expect good TSC - * - * Thus far, the major user of this is IBM's Summit2 series: - * - * Clustered boxes may have unsynced TSC problems if they are - * multi-chassis. Use available data to take a good guess. - * If in doubt, go HPET. - */ -__cpuinit int apic_is_clustered_box(void) -{ - int i, clusters, zeros; - unsigned id; - u16 *bios_cpu_apicid; - DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS); - - /* - * there is not this kind of box with AMD CPU yet. - * Some AMD box with quadcore cpu and 8 sockets apicid - * will be [4, 0x23] or [8, 0x27] could be thought to - * vsmp box still need checking... - */ - if ((boot_cpu_data.x86_vendor == X86_VENDOR_AMD) && !is_vsmp_box()) - return 0; - - bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); - bitmap_zero(clustermap, NUM_APIC_CLUSTERS); - - for (i = 0; i < NR_CPUS; i++) { - /* are we being called early in kernel startup? */ - if (bios_cpu_apicid) { - id = bios_cpu_apicid[i]; - } - else if (i < nr_cpu_ids) { - if (cpu_present(i)) - id = per_cpu(x86_bios_cpu_apicid, i); - else - continue; - } - else - break; - - if (id != BAD_APICID) - __set_bit(APIC_CLUSTERID(id), clustermap); - } - - /* Problem: Partially populated chassis may not have CPUs in some of - * the APIC clusters they have been allocated. Only present CPUs have - * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. - * Since clusters are allocated sequentially, count zeros only if - * they are bounded by ones. - */ - clusters = 0; - zeros = 0; - for (i = 0; i < NUM_APIC_CLUSTERS; i++) { - if (test_bit(i, clustermap)) { - clusters += 1 + zeros; - zeros = 0; - } else - ++zeros; - } - - /* ScaleMP vSMPowered boxes have one cluster per board and TSCs are - * not guaranteed to be synced between boards - */ - if (is_vsmp_box() && clusters > 1) - return 1; - - /* - * If clusters > 2, then should be multi-chassis. - * May have to revisit this when multi-core + hyperthreaded CPUs come - * out, but AFAIK this will work even for them. - */ - return (clusters > 2); -} -#endif - -/* - * APIC command line parameters - */ -static int __init setup_disableapic(char *arg) -{ - disable_apic = 1; - setup_clear_cpu_cap(X86_FEATURE_APIC); - return 0; -} -early_param("disableapic", setup_disableapic); - -/* same as disableapic, for compatibility */ -static int __init setup_nolapic(char *arg) -{ - return setup_disableapic(arg); -} -early_param("nolapic", setup_nolapic); - -static int __init parse_lapic_timer_c2_ok(char *arg) -{ - local_apic_timer_c2_ok = 1; - return 0; -} -early_param("lapic_timer_c2_ok", parse_lapic_timer_c2_ok); - -static int __init parse_disable_apic_timer(char *arg) -{ - disable_apic_timer = 1; - return 0; -} -early_param("noapictimer", parse_disable_apic_timer); - -static int __init parse_nolapic_timer(char *arg) -{ - disable_apic_timer = 1; - return 0; -} -early_param("nolapic_timer", parse_nolapic_timer); - -static int __init apic_set_verbosity(char *arg) -{ - if (!arg) { -#ifdef CONFIG_X86_64 - skip_ioapic_setup = 0; - ioapic_force = 1; - return 0; -#endif - return -EINVAL; - } - - if (strcmp("debug", arg) == 0) - apic_verbosity = APIC_DEBUG; - else if (strcmp("verbose", arg) == 0) - apic_verbosity = APIC_VERBOSE; - else { - printk(KERN_WARNING "APIC Verbosity level %s not recognised" - " use apic=verbose or apic=debug\n", arg); - return -EINVAL; - } - - return 0; -} -early_param("apic", apic_set_verbosity); - -static int __init lapic_insert_resource(void) -{ - if (!apic_phys) - return -1; - - /* Put local APIC into the resource map. */ - lapic_resource.start = apic_phys; - lapic_resource.end = lapic_resource.start + PAGE_SIZE - 1; - insert_resource(&iomem_resource, &lapic_resource); - - return 0; -} - -/* - * need call insert after e820_reserve_resources() - * that is using request_resource - */ -late_initcall(lapic_insert_resource); diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c deleted file mode 100644 index 6e99af5ce678..000000000000 --- a/arch/x86/kernel/apic_64.c +++ /dev/null @@ -1,2311 +0,0 @@ -/* - * Local APIC handling, local APIC timers - * - * (c) 1999, 2000 Ingo Molnar - * - * Fixes - * Maciej W. Rozycki : Bits for genuine 82489DX APICs; - * thanks to Eric Gilmore - * and Rolf G. Tews - * for testing these extensively. - * Maciej W. Rozycki : Various updates and fixes. - * Mikael Pettersson : Power Management for UP-APIC. - * Pavel Machek and - * Mikael Pettersson : PM converted to driver model. - */ - -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include - -/* - * Sanity check - */ -#if ((SPURIOUS_APIC_VECTOR & 0x0F) != 0x0F) -# error SPURIOUS_APIC_VECTOR definition error -#endif - -#ifdef CONFIG_X86_32 -/* - * Knob to control our willingness to enable the local APIC. - * - * +1=force-enable - */ -static int force_enable_local_apic; -/* - * APIC command line parameters - */ -static int __init parse_lapic(char *arg) -{ - force_enable_local_apic = 1; - return 0; -} -early_param("lapic", parse_lapic); -/* Local APIC was disabled by the BIOS and enabled by the kernel */ -static int enabled_via_apicbase; - -#endif - -#ifdef CONFIG_X86_64 -static int apic_calibrate_pmtmr __initdata; -static __init int setup_apicpmtimer(char *s) -{ - apic_calibrate_pmtmr = 1; - notsc_setup(NULL); - return 0; -} -__setup("apicpmtimer", setup_apicpmtimer); -#endif - -#ifdef CONFIG_X86_64 -#define HAVE_X2APIC -#endif - -#ifdef HAVE_X2APIC -int x2apic; -/* x2apic enabled before OS handover */ -int x2apic_preenabled; -int disable_x2apic; -static __init int setup_nox2apic(char *str) -{ - disable_x2apic = 1; - setup_clear_cpu_cap(X86_FEATURE_X2APIC); - return 0; -} -early_param("nox2apic", setup_nox2apic); -#endif - -unsigned long mp_lapic_addr; -int disable_apic; -/* Disable local APIC timer from the kernel commandline or via dmi quirk */ -static int disable_apic_timer __cpuinitdata; -/* Local APIC timer works in C2 */ -int local_apic_timer_c2_ok; -EXPORT_SYMBOL_GPL(local_apic_timer_c2_ok); - -int first_system_vector = 0xfe; - -char system_vectors[NR_VECTORS] = { [0 ... NR_VECTORS-1] = SYS_VECTOR_FREE}; - -/* - * Debug level, exported for io_apic.c - */ -unsigned int apic_verbosity; - -int pic_mode; - -/* Have we found an MP table */ -int smp_found_config; - -static struct resource lapic_resource = { - .name = "Local APIC", - .flags = IORESOURCE_MEM | IORESOURCE_BUSY, -}; - -static unsigned int calibration_result; - -static int lapic_next_event(unsigned long delta, - struct clock_event_device *evt); -static void lapic_timer_setup(enum clock_event_mode mode, - struct clock_event_device *evt); -static void lapic_timer_broadcast(cpumask_t mask); -static void apic_pm_activate(void); - -/* - * The local apic timer can be used for any function which is CPU local. - */ -static struct clock_event_device lapic_clockevent = { - .name = "lapic", - .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT - | CLOCK_EVT_FEAT_C3STOP | CLOCK_EVT_FEAT_DUMMY, - .shift = 32, - .set_mode = lapic_timer_setup, - .set_next_event = lapic_next_event, - .broadcast = lapic_timer_broadcast, - .rating = 100, - .irq = -1, -}; -static DEFINE_PER_CPU(struct clock_event_device, lapic_events); - -static unsigned long apic_phys; - -/* - * Get the LAPIC version - */ -static inline int lapic_get_version(void) -{ - return GET_APIC_VERSION(apic_read(APIC_LVR)); -} - -/* - * Check, if the APIC is integrated or a separate chip - */ -static inline int lapic_is_integrated(void) -{ -#ifdef CONFIG_X86_64 - return 1; -#else - return APIC_INTEGRATED(lapic_get_version()); -#endif -} - -/* - * Check, whether this is a modern or a first generation APIC - */ -static int modern_apic(void) -{ - /* AMD systems use old APIC versions, so check the CPU */ - if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD && - boot_cpu_data.x86 >= 0xf) - return 1; - return lapic_get_version() >= 0x14; -} - -/* - * Paravirt kernels also might be using these below ops. So we still - * use generic apic_read()/apic_write(), which might be pointing to different - * ops in PARAVIRT case. - */ -void xapic_wait_icr_idle(void) -{ - while (apic_read(APIC_ICR) & APIC_ICR_BUSY) - cpu_relax(); -} - -u32 safe_xapic_wait_icr_idle(void) -{ - u32 send_status; - int timeout; - - timeout = 0; - do { - send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY; - if (!send_status) - break; - udelay(100); - } while (timeout++ < 1000); - - return send_status; -} - -void xapic_icr_write(u32 low, u32 id) -{ - apic_write(APIC_ICR2, SET_APIC_DEST_FIELD(id)); - apic_write(APIC_ICR, low); -} - -u64 xapic_icr_read(void) -{ - u32 icr1, icr2; - - icr2 = apic_read(APIC_ICR2); - icr1 = apic_read(APIC_ICR); - - return icr1 | ((u64)icr2 << 32); -} - -static struct apic_ops xapic_ops = { - .read = native_apic_mem_read, - .write = native_apic_mem_write, - .icr_read = xapic_icr_read, - .icr_write = xapic_icr_write, - .wait_icr_idle = xapic_wait_icr_idle, - .safe_wait_icr_idle = safe_xapic_wait_icr_idle, -}; - -struct apic_ops __read_mostly *apic_ops = &xapic_ops; -EXPORT_SYMBOL_GPL(apic_ops); - -#ifdef HAVE_X2APIC -static void x2apic_wait_icr_idle(void) -{ - /* no need to wait for icr idle in x2apic */ - return; -} - -static u32 safe_x2apic_wait_icr_idle(void) -{ - /* no need to wait for icr idle in x2apic */ - return 0; -} - -void x2apic_icr_write(u32 low, u32 id) -{ - wrmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), ((__u64) id) << 32 | low); -} - -u64 x2apic_icr_read(void) -{ - unsigned long val; - - rdmsrl(APIC_BASE_MSR + (APIC_ICR >> 4), val); - return val; -} - -static struct apic_ops x2apic_ops = { - .read = native_apic_msr_read, - .write = native_apic_msr_write, - .icr_read = x2apic_icr_read, - .icr_write = x2apic_icr_write, - .wait_icr_idle = x2apic_wait_icr_idle, - .safe_wait_icr_idle = safe_x2apic_wait_icr_idle, -}; -#endif - -/** - * enable_NMI_through_LVT0 - enable NMI through local vector table 0 - */ -void __cpuinit enable_NMI_through_LVT0(void) -{ - unsigned int v; - - /* unmask and set to NMI */ - v = APIC_DM_NMI; - - /* Level triggered for 82489DX (32bit mode) */ - if (!lapic_is_integrated()) - v |= APIC_LVT_LEVEL_TRIGGER; - - apic_write(APIC_LVT0, v); -} - -#ifdef CONFIG_X86_32 -/** - * get_physical_broadcast - Get number of physical broadcast IDs - */ -int get_physical_broadcast(void) -{ - return modern_apic() ? 0xff : 0xf; -} -#endif - -/** - * lapic_get_maxlvt - get the maximum number of local vector table entries - */ -int lapic_get_maxlvt(void) -{ - unsigned int v; - - v = apic_read(APIC_LVR); - /* - * - we always have APIC integrated on 64bit mode - * - 82489DXs do not report # of LVT entries - */ - return APIC_INTEGRATED(GET_APIC_VERSION(v)) ? GET_APIC_MAXLVT(v) : 2; -} - -/* - * Local APIC timer - */ - -/* Clock divisor */ -#ifdef CONFG_X86_64 -#define APIC_DIVISOR 1 -#else -#define APIC_DIVISOR 16 -#endif - -/* - * This function sets up the local APIC timer, with a timeout of - * 'clocks' APIC bus clock. During calibration we actually call - * this function twice on the boot CPU, once with a bogus timeout - * value, second time for real. The other (noncalibrating) CPUs - * call this function only once, with the real, calibrated value. - * - * We do reads before writes even if unnecessary, to get around the - * P5 APIC double write bug. - */ -static void __setup_APIC_LVTT(unsigned int clocks, int oneshot, int irqen) -{ - unsigned int lvtt_value, tmp_value; - - lvtt_value = LOCAL_TIMER_VECTOR; - if (!oneshot) - lvtt_value |= APIC_LVT_TIMER_PERIODIC; - if (!lapic_is_integrated()) - lvtt_value |= SET_APIC_TIMER_BASE(APIC_TIMER_BASE_DIV); - - if (!irqen) - lvtt_value |= APIC_LVT_MASKED; - - apic_write(APIC_LVTT, lvtt_value); - - /* - * Divide PICLK by 16 - */ - tmp_value = apic_read(APIC_TDCR); - apic_write(APIC_TDCR, - (tmp_value & ~(APIC_TDR_DIV_1 | APIC_TDR_DIV_TMBASE)) | - APIC_TDR_DIV_16); - - if (!oneshot) - apic_write(APIC_TMICT, clocks / APIC_DIVISOR); -} - -/* - * Setup extended LVT, AMD specific (K8, family 10h) - * - * Vector mappings are hard coded. On K8 only offset 0 (APIC500) and - * MCE interrupts are supported. Thus MCE offset must be set to 0. - * - * If mask=1, the LVT entry does not generate interrupts while mask=0 - * enables the vector. See also the BKDGs. - */ - -#define APIC_EILVT_LVTOFF_MCE 0 -#define APIC_EILVT_LVTOFF_IBS 1 - -static void setup_APIC_eilvt(u8 lvt_off, u8 vector, u8 msg_type, u8 mask) -{ - unsigned long reg = (lvt_off << 4) + APIC_EILVT0; - unsigned int v = (mask << 16) | (msg_type << 8) | vector; - - apic_write(reg, v); -} - -u8 setup_APIC_eilvt_mce(u8 vector, u8 msg_type, u8 mask) -{ - setup_APIC_eilvt(APIC_EILVT_LVTOFF_MCE, vector, msg_type, mask); - return APIC_EILVT_LVTOFF_MCE; -} - -u8 setup_APIC_eilvt_ibs(u8 vector, u8 msg_type, u8 mask) -{ - setup_APIC_eilvt(APIC_EILVT_LVTOFF_IBS, vector, msg_type, mask); - return APIC_EILVT_LVTOFF_IBS; -} -EXPORT_SYMBOL_GPL(setup_APIC_eilvt_ibs); - -/* - * Program the next event, relative to now - */ -static int lapic_next_event(unsigned long delta, - struct clock_event_device *evt) -{ - apic_write(APIC_TMICT, delta); - return 0; -} - -/* - * Setup the lapic timer in periodic or oneshot mode - */ -static void lapic_timer_setup(enum clock_event_mode mode, - struct clock_event_device *evt) -{ - unsigned long flags; - unsigned int v; - - /* Lapic used as dummy for broadcast ? */ - if (evt->features & CLOCK_EVT_FEAT_DUMMY) - return; - - local_irq_save(flags); - - switch (mode) { - case CLOCK_EVT_MODE_PERIODIC: - case CLOCK_EVT_MODE_ONESHOT: - __setup_APIC_LVTT(calibration_result, - mode != CLOCK_EVT_MODE_PERIODIC, 1); - break; - case CLOCK_EVT_MODE_UNUSED: - case CLOCK_EVT_MODE_SHUTDOWN: - v = apic_read(APIC_LVTT); - v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR); - apic_write(APIC_LVTT, v); - break; - case CLOCK_EVT_MODE_RESUME: - /* Nothing to do here */ - break; - } - - local_irq_restore(flags); -} - -/* - * Local APIC timer broadcast function - */ -static void lapic_timer_broadcast(cpumask_t mask) -{ -#ifdef CONFIG_SMP - send_IPI_mask(mask, LOCAL_TIMER_VECTOR); -#endif -} - -/* - * Setup the local APIC timer for this CPU. Copy the initilized values - * of the boot CPU and register the clock event in the framework. - */ -static void __cpuinit setup_APIC_timer(void) -{ - struct clock_event_device *levt = &__get_cpu_var(lapic_events); - - memcpy(levt, &lapic_clockevent, sizeof(*levt)); - levt->cpumask = cpumask_of_cpu(smp_processor_id()); - - clockevents_register_device(levt); -} - -#ifdef CONFIG_X86_64 -/* - * In this function we calibrate APIC bus clocks to the external - * timer. Unfortunately we cannot use jiffies and the timer irq - * to calibrate, since some later bootup code depends on getting - * the first irq? Ugh. - * - * We want to do the calibration only once since we - * want to have local timer irqs syncron. CPUs connected - * by the same APIC bus have the very same bus frequency. - * And we want to have irqs off anyways, no accidental - * APIC irq that way. - */ - -#define TICK_COUNT 100000000 - -static int __init calibrate_APIC_clock(void) -{ - unsigned apic, apic_start; - unsigned long tsc, tsc_start; - int result; - - local_irq_disable(); - - /* - * Put whatever arbitrary (but long enough) timeout - * value into the APIC clock, we just want to get the - * counter running for calibration. - * - * No interrupt enable ! - */ - __setup_APIC_LVTT(250000000, 0, 0); - - apic_start = apic_read(APIC_TMCCT); -#ifdef CONFIG_X86_PM_TIMER - if (apic_calibrate_pmtmr && pmtmr_ioport) { - pmtimer_wait(5000); /* 5ms wait */ - apic = apic_read(APIC_TMCCT); - result = (apic_start - apic) * 1000L / 5; - } else -#endif - { - rdtscll(tsc_start); - - do { - apic = apic_read(APIC_TMCCT); - rdtscll(tsc); - } while ((tsc - tsc_start) < TICK_COUNT && - (apic_start - apic) < TICK_COUNT); - - result = (apic_start - apic) * 1000L * tsc_khz / - (tsc - tsc_start); - } - - local_irq_enable(); - - printk(KERN_DEBUG "APIC timer calibration result %d\n", result); - - printk(KERN_INFO "Detected %d.%03d MHz APIC timer.\n", - result / 1000 / 1000, result / 1000 % 1000); - - /* Calculate the scaled math multiplication factor */ - lapic_clockevent.mult = div_sc(result, NSEC_PER_SEC, - lapic_clockevent.shift); - lapic_clockevent.max_delta_ns = - clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); - lapic_clockevent.min_delta_ns = - clockevent_delta2ns(0xF, &lapic_clockevent); - - calibration_result = (result * APIC_DIVISOR) / HZ; - - /* - * Do a sanity check on the APIC calibration result - */ - if (calibration_result < (1000000 / HZ)) { - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); - return -1; - } - - return 0; -} - -#else -/* - * In this functions we calibrate APIC bus clocks to the external timer. - * - * We want to do the calibration only once since we want to have local timer - * irqs syncron. CPUs connected by the same APIC bus have the very same bus - * frequency. - * - * This was previously done by reading the PIT/HPET and waiting for a wrap - * around to find out, that a tick has elapsed. I have a box, where the PIT - * readout is broken, so it never gets out of the wait loop again. This was - * also reported by others. - * - * Monitoring the jiffies value is inaccurate and the clockevents - * infrastructure allows us to do a simple substitution of the interrupt - * handler. - * - * The calibration routine also uses the pm_timer when possible, as the PIT - * happens to run way too slow (factor 2.3 on my VAIO CoreDuo, which goes - * back to normal later in the boot process). - */ - -#define LAPIC_CAL_LOOPS (HZ/10) - -static __initdata int lapic_cal_loops = -1; -static __initdata long lapic_cal_t1, lapic_cal_t2; -static __initdata unsigned long long lapic_cal_tsc1, lapic_cal_tsc2; -static __initdata unsigned long lapic_cal_pm1, lapic_cal_pm2; -static __initdata unsigned long lapic_cal_j1, lapic_cal_j2; - -/* - * Temporary interrupt handler. - */ -static void __init lapic_cal_handler(struct clock_event_device *dev) -{ - unsigned long long tsc = 0; - long tapic = apic_read(APIC_TMCCT); - unsigned long pm = acpi_pm_read_early(); - - if (cpu_has_tsc) - rdtscll(tsc); - - switch (lapic_cal_loops++) { - case 0: - lapic_cal_t1 = tapic; - lapic_cal_tsc1 = tsc; - lapic_cal_pm1 = pm; - lapic_cal_j1 = jiffies; - break; - - case LAPIC_CAL_LOOPS: - lapic_cal_t2 = tapic; - lapic_cal_tsc2 = tsc; - if (pm < lapic_cal_pm1) - pm += ACPI_PM_OVRRUN; - lapic_cal_pm2 = pm; - lapic_cal_j2 = jiffies; - break; - } -} - -static int __init calibrate_APIC_clock(void) -{ - struct clock_event_device *levt = &__get_cpu_var(lapic_events); - const long pm_100ms = PMTMR_TICKS_PER_SEC/10; - const long pm_thresh = pm_100ms/100; - void (*real_handler)(struct clock_event_device *dev); - unsigned long deltaj; - long delta, deltapm; - int pm_referenced = 0; - - local_irq_disable(); - - /* Replace the global interrupt handler */ - real_handler = global_clock_event->event_handler; - global_clock_event->event_handler = lapic_cal_handler; - - /* - * Setup the APIC counter to 1e9. There is no way the lapic - * can underflow in the 100ms detection time frame - */ - __setup_APIC_LVTT(1000000000, 0, 0); - - /* Let the interrupts run */ - local_irq_enable(); - - while (lapic_cal_loops <= LAPIC_CAL_LOOPS) - cpu_relax(); - - local_irq_disable(); - - /* Restore the real event handler */ - global_clock_event->event_handler = real_handler; - - /* Build delta t1-t2 as apic timer counts down */ - delta = lapic_cal_t1 - lapic_cal_t2; - apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); - - /* Check, if the PM timer is available */ - deltapm = lapic_cal_pm2 - lapic_cal_pm1; - apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); - - if (deltapm) { - unsigned long mult; - u64 res; - - mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); - - if (deltapm > (pm_100ms - pm_thresh) && - deltapm < (pm_100ms + pm_thresh)) { - apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); - } else { - res = (((u64) deltapm) * mult) >> 22; - do_div(res, 1000000); - printk(KERN_WARNING "APIC calibration not consistent " - "with PM Timer: %ldms instead of 100ms\n", - (long)res); - /* Correct the lapic counter value */ - res = (((u64) delta) * pm_100ms); - do_div(res, deltapm); - printk(KERN_INFO "APIC delta adjusted to PM-Timer: " - "%lu (%ld)\n", (unsigned long) res, delta); - delta = (long) res; - } - pm_referenced = 1; - } - - /* Calculate the scaled math multiplication factor */ - lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, - lapic_clockevent.shift); - lapic_clockevent.max_delta_ns = - clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); - lapic_clockevent.min_delta_ns = - clockevent_delta2ns(0xF, &lapic_clockevent); - - calibration_result = (delta * APIC_DIVISOR) / LAPIC_CAL_LOOPS; - - apic_printk(APIC_VERBOSE, "..... delta %ld\n", delta); - apic_printk(APIC_VERBOSE, "..... mult: %ld\n", lapic_clockevent.mult); - apic_printk(APIC_VERBOSE, "..... calibration result: %u\n", - calibration_result); - - if (cpu_has_tsc) { - delta = (long)(lapic_cal_tsc2 - lapic_cal_tsc1); - apic_printk(APIC_VERBOSE, "..... CPU clock speed is " - "%ld.%04ld MHz.\n", - (delta / LAPIC_CAL_LOOPS) / (1000000 / HZ), - (delta / LAPIC_CAL_LOOPS) % (1000000 / HZ)); - } - - apic_printk(APIC_VERBOSE, "..... host bus clock speed is " - "%u.%04u MHz.\n", - calibration_result / (1000000 / HZ), - calibration_result % (1000000 / HZ)); - - /* - * Do a sanity check on the APIC calibration result - */ - if (calibration_result < (1000000 / HZ)) { - local_irq_enable(); - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); - return -1; - } - - levt->features &= ~CLOCK_EVT_FEAT_DUMMY; - - /* We trust the pm timer based calibration */ - if (!pm_referenced) { - apic_printk(APIC_VERBOSE, "... verify APIC timer\n"); - - /* - * Setup the apic timer manually - */ - levt->event_handler = lapic_cal_handler; - lapic_timer_setup(CLOCK_EVT_MODE_PERIODIC, levt); - lapic_cal_loops = -1; - - /* Let the interrupts run */ - local_irq_enable(); - - while (lapic_cal_loops <= LAPIC_CAL_LOOPS) - cpu_relax(); - - local_irq_disable(); - - /* Stop the lapic timer */ - lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); - - local_irq_enable(); - - /* Jiffies delta */ - deltaj = lapic_cal_j2 - lapic_cal_j1; - apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); - - /* Check, if the jiffies result is consistent */ - if (deltaj >= LAPIC_CAL_LOOPS-2 && deltaj <= LAPIC_CAL_LOOPS+2) - apic_printk(APIC_VERBOSE, "... jiffies result ok\n"); - else - levt->features |= CLOCK_EVT_FEAT_DUMMY; - } else - local_irq_enable(); - - if (levt->features & CLOCK_EVT_FEAT_DUMMY) { - printk(KERN_WARNING - "APIC timer disabled due to verification failure.\n"); - return -1; - } - - return 0; -} - -#endif - -/* - * Setup the boot APIC - * - * Calibrate and verify the result. - */ -void __init setup_boot_APIC_clock(void) -{ - /* - * The local apic timer can be disabled via the kernel - * commandline or from the CPU detection code. Register the lapic - * timer as a dummy clock event source on SMP systems, so the - * broadcast mechanism is used. On UP systems simply ignore it. - */ - if (disable_apic_timer) { - printk(KERN_INFO "Disabling APIC timer\n"); - /* No broadcast on UP ! */ - if (num_possible_cpus() > 1) { - lapic_clockevent.mult = 1; - setup_APIC_timer(); - } - return; - } - - apic_printk(APIC_VERBOSE, "Using local APIC timer interrupts.\n" - "calibrating APIC timer ...\n"); - - if (calibrate_APIC_clock()) { - /* No broadcast on UP ! */ - if (num_possible_cpus() > 1) - setup_APIC_timer(); - return; - } - - /* - * If nmi_watchdog is set to IO_APIC, we need the - * PIT/HPET going. Otherwise register lapic as a dummy - * device. - */ - if (nmi_watchdog != NMI_IO_APIC) - lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY; - else - printk(KERN_WARNING "APIC timer registered as dummy," - " due to nmi_watchdog=%d!\n", nmi_watchdog); - - /* Setup the lapic or request the broadcast */ - setup_APIC_timer(); -} - -void __cpuinit setup_secondary_APIC_clock(void) -{ - setup_APIC_timer(); -} - -/* - * The guts of the apic timer interrupt - */ -static void local_apic_timer_interrupt(void) -{ - int cpu = smp_processor_id(); - struct clock_event_device *evt = &per_cpu(lapic_events, cpu); - - /* - * Normally we should not be here till LAPIC has been initialized but - * in some cases like kdump, its possible that there is a pending LAPIC - * timer interrupt from previous kernel's context and is delivered in - * new kernel the moment interrupts are enabled. - * - * Interrupts are enabled early and LAPIC is setup much later, hence - * its possible that when we get here evt->event_handler is NULL. - * Check for event_handler being NULL and discard the interrupt as - * spurious. - */ - if (!evt->event_handler) { - printk(KERN_WARNING - "Spurious LAPIC timer interrupt on cpu %d\n", cpu); - /* Switch it off */ - lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt); - return; - } - - /* - * the NMI deadlock-detector uses this. - */ -#ifdef CONFIG_X86_64 - add_pda(apic_timer_irqs, 1); -#else - per_cpu(irq_stat, cpu).apic_timer_irqs++; -#endif - - evt->event_handler(evt); -} - -/* - * Local APIC timer interrupt. This is the most natural way for doing - * local interrupts, but local timer interrupts can be emulated by - * broadcast interrupts too. [in case the hw doesn't support APIC timers] - * - * [ if a single-CPU system runs an SMP kernel then we call the local - * interrupt as well. Thus we cannot inline the local irq ... ] - */ -void smp_apic_timer_interrupt(struct pt_regs *regs) -{ - struct pt_regs *old_regs = set_irq_regs(regs); - - /* - * NOTE! We'd better ACK the irq immediately, - * because timer handling can be slow. - */ - ack_APIC_irq(); - /* - * update_process_times() expects us to have done irq_enter(). - * Besides, if we don't timer interrupts ignore the global - * interrupt lock, which is the WrongThing (tm) to do. - */ -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - local_apic_timer_interrupt(); - irq_exit(); - - set_irq_regs(old_regs); -} - -int setup_profiling_timer(unsigned int multiplier) -{ - return -EINVAL; -} - -/* - * Local APIC start and shutdown - */ - -/** - * clear_local_APIC - shutdown the local APIC - * - * This is called, when a CPU is disabled and before rebooting, so the state of - * the local APIC has no dangling leftovers. Also used to cleanout any BIOS - * leftovers during boot. - */ -void clear_local_APIC(void) -{ - int maxlvt; - u32 v; - - /* APIC hasn't been mapped yet */ - if (!apic_phys) - return; - - maxlvt = lapic_get_maxlvt(); - /* - * Masking an LVT entry can trigger a local APIC error - * if the vector is zero. Mask LVTERR first to prevent this. - */ - if (maxlvt >= 3) { - v = ERROR_APIC_VECTOR; /* any non-zero vector will do */ - apic_write(APIC_LVTERR, v | APIC_LVT_MASKED); - } - /* - * Careful: we have to set masks only first to deassert - * any level-triggered sources. - */ - v = apic_read(APIC_LVTT); - apic_write(APIC_LVTT, v | APIC_LVT_MASKED); - v = apic_read(APIC_LVT0); - apic_write(APIC_LVT0, v | APIC_LVT_MASKED); - v = apic_read(APIC_LVT1); - apic_write(APIC_LVT1, v | APIC_LVT_MASKED); - if (maxlvt >= 4) { - v = apic_read(APIC_LVTPC); - apic_write(APIC_LVTPC, v | APIC_LVT_MASKED); - } - - /* lets not touch this if we didn't frob it */ -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(X86_MCE_INTEL) - if (maxlvt >= 5) { - v = apic_read(APIC_LVTTHMR); - apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED); - } -#endif - /* - * Clean APIC state for other OSs: - */ - apic_write(APIC_LVTT, APIC_LVT_MASKED); - apic_write(APIC_LVT0, APIC_LVT_MASKED); - apic_write(APIC_LVT1, APIC_LVT_MASKED); - if (maxlvt >= 3) - apic_write(APIC_LVTERR, APIC_LVT_MASKED); - if (maxlvt >= 4) - apic_write(APIC_LVTPC, APIC_LVT_MASKED); - - /* Integrated APIC (!82489DX) ? */ - if (lapic_is_integrated()) { - if (maxlvt > 3) - /* Clear ESR due to Pentium errata 3AP and 11AP */ - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - } -} - -/** - * disable_local_APIC - clear and disable the local APIC - */ -void disable_local_APIC(void) -{ - unsigned int value; - - clear_local_APIC(); - - /* - * Disable APIC (implies clearing of registers - * for 82489DX!). - */ - value = apic_read(APIC_SPIV); - value &= ~APIC_SPIV_APIC_ENABLED; - apic_write(APIC_SPIV, value); - -#ifdef CONFIG_X86_32 - /* - * When LAPIC was disabled by the BIOS and enabled by the kernel, - * restore the disabled state. - */ - if (enabled_via_apicbase) { - unsigned int l, h; - - rdmsr(MSR_IA32_APICBASE, l, h); - l &= ~MSR_IA32_APICBASE_ENABLE; - wrmsr(MSR_IA32_APICBASE, l, h); - } -#endif -} - -/* - * If Linux enabled the LAPIC against the BIOS default disable it down before - * re-entering the BIOS on shutdown. Otherwise the BIOS may get confused and - * not power-off. Additionally clear all LVT entries before disable_local_APIC - * for the case where Linux didn't enable the LAPIC. - */ -void lapic_shutdown(void) -{ - unsigned long flags; - - if (!cpu_has_apic) - return; - - local_irq_save(flags); - -#ifdef CONFIG_X86_32 - if (!enabled_via_apicbase) - clear_local_APIC(); - else -#endif - disable_local_APIC(); - - - local_irq_restore(flags); -} - -/* - * This is to verify that we're looking at a real local APIC. - * Check these against your board if the CPUs aren't getting - * started for no apparent reason. - */ -int __init verify_local_APIC(void) -{ - unsigned int reg0, reg1; - - /* - * The version register is read-only in a real APIC. - */ - reg0 = apic_read(APIC_LVR); - apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg0); - apic_write(APIC_LVR, reg0 ^ APIC_LVR_MASK); - reg1 = apic_read(APIC_LVR); - apic_printk(APIC_DEBUG, "Getting VERSION: %x\n", reg1); - - /* - * The two version reads above should print the same - * numbers. If the second one is different, then we - * poke at a non-APIC. - */ - if (reg1 != reg0) - return 0; - - /* - * Check if the version looks reasonably. - */ - reg1 = GET_APIC_VERSION(reg0); - if (reg1 == 0x00 || reg1 == 0xff) - return 0; - reg1 = lapic_get_maxlvt(); - if (reg1 < 0x02 || reg1 == 0xff) - return 0; - - /* - * The ID register is read/write in a real APIC. - */ - reg0 = apic_read(APIC_ID); - apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg0); - apic_write(APIC_ID, reg0 ^ APIC_ID_MASK); - reg1 = apic_read(APIC_ID); - apic_printk(APIC_DEBUG, "Getting ID: %x\n", reg1); - apic_write(APIC_ID, reg0); - if (reg1 != (reg0 ^ APIC_ID_MASK)) - return 0; - - /* - * The next two are just to see if we have sane values. - * They're only really relevant if we're in Virtual Wire - * compatibility mode, but most boxes are anymore. - */ - reg0 = apic_read(APIC_LVT0); - apic_printk(APIC_DEBUG, "Getting LVT0: %x\n", reg0); - reg1 = apic_read(APIC_LVT1); - apic_printk(APIC_DEBUG, "Getting LVT1: %x\n", reg1); - - return 1; -} - -/** - * sync_Arb_IDs - synchronize APIC bus arbitration IDs - */ -void __init sync_Arb_IDs(void) -{ - /* - * Unsupported on P4 - see Intel Dev. Manual Vol. 3, Ch. 8.6.1 And not - * needed on AMD. - */ - if (modern_apic() || boot_cpu_data.x86_vendor == X86_VENDOR_AMD) - return; - - /* - * Wait for idle. - */ - apic_wait_icr_idle(); - - apic_printk(APIC_DEBUG, "Synchronizing Arb IDs.\n"); - apic_write(APIC_ICR, APIC_DEST_ALLINC | - APIC_INT_LEVELTRIG | APIC_DM_INIT); -} - -/* - * An initial setup of the virtual wire mode. - */ -void __init init_bsp_APIC(void) -{ - unsigned int value; - - /* - * Don't do the setup now if we have a SMP BIOS as the - * through-I/O-APIC virtual wire mode might be active. - */ - if (smp_found_config || !cpu_has_apic) - return; - - /* - * Do not trust the local APIC being empty at bootup. - */ - clear_local_APIC(); - - /* - * Enable APIC. - */ - value = apic_read(APIC_SPIV); - value &= ~APIC_VECTOR_MASK; - value |= APIC_SPIV_APIC_ENABLED; - -#ifdef CONFIG_X86_32 - /* This bit is reserved on P4/Xeon and should be cleared */ - if ((boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) && - (boot_cpu_data.x86 == 15)) - value &= ~APIC_SPIV_FOCUS_DISABLED; - else -#endif - value |= APIC_SPIV_FOCUS_DISABLED; - value |= SPURIOUS_APIC_VECTOR; - apic_write(APIC_SPIV, value); - - /* - * Set up the virtual wire mode. - */ - apic_write(APIC_LVT0, APIC_DM_EXTINT); - value = APIC_DM_NMI; - if (!lapic_is_integrated()) /* 82489DX */ - value |= APIC_LVT_LEVEL_TRIGGER; - apic_write(APIC_LVT1, value); -} - -static void __cpuinit lapic_setup_esr(void) -{ - unsigned long oldvalue, value, maxlvt; - if (lapic_is_integrated() && !esr_disable) { - if (esr_disable) { - /* - * Something untraceable is creating bad interrupts on - * secondary quads ... for the moment, just leave the - * ESR disabled - we can't do anything useful with the - * errors anyway - mbligh - */ - printk(KERN_INFO "Leaving ESR disabled.\n"); - return; - } - /* !82489DX */ - maxlvt = lapic_get_maxlvt(); - if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ - apic_write(APIC_ESR, 0); - oldvalue = apic_read(APIC_ESR); - - /* enables sending errors */ - value = ERROR_APIC_VECTOR; - apic_write(APIC_LVTERR, value); - /* - * spec says clear errors after enabling vector. - */ - if (maxlvt > 3) - apic_write(APIC_ESR, 0); - value = apic_read(APIC_ESR); - if (value != oldvalue) - apic_printk(APIC_VERBOSE, "ESR value before enabling " - "vector: 0x%08lx after: 0x%08lx\n", - oldvalue, value); - } else { - printk(KERN_INFO "No ESR for 82489DX.\n"); - } -} - - -/** - * setup_local_APIC - setup the local APIC - */ -void __cpuinit setup_local_APIC(void) -{ - unsigned int value; - int i, j; - -#ifdef CONFIG_X86_32 - /* Pound the ESR really hard over the head with a big hammer - mbligh */ - if (esr_disable) { - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - apic_write(APIC_ESR, 0); - } -#endif - - preempt_disable(); - - /* - * Double-check whether this APIC is really registered. - * This is meaningless in clustered apic mode, so we skip it. - */ - if (!apic_id_registered()) - BUG(); - - /* - * Intel recommends to set DFR, LDR and TPR before enabling - * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel - * document number 292116). So here it goes... - */ - init_apic_ldr(); - - /* - * Set Task Priority to 'accept all'. We never change this - * later on. - */ - value = apic_read(APIC_TASKPRI); - value &= ~APIC_TPRI_MASK; - apic_write(APIC_TASKPRI, value); - - /* - * After a crash, we no longer service the interrupts and a pending - * interrupt from previous kernel might still have ISR bit set. - * - * Most probably by now CPU has serviced that pending interrupt and - * it might not have done the ack_APIC_irq() because it thought, - * interrupt came from i8259 as ExtInt. LAPIC did not get EOI so it - * does not clear the ISR bit and cpu thinks it has already serivced - * the interrupt. Hence a vector might get locked. It was noticed - * for timer irq (vector 0x31). Issue an extra EOI to clear ISR. - */ - for (i = APIC_ISR_NR - 1; i >= 0; i--) { - value = apic_read(APIC_ISR + i*0x10); - for (j = 31; j >= 0; j--) { - if (value & (1< 1) || - (boot_cpu_data.x86 == 15)) - break; - goto no_apic; - case X86_VENDOR_INTEL: - if (boot_cpu_data.x86 == 6 || boot_cpu_data.x86 == 15 || - (boot_cpu_data.x86 == 5 && cpu_has_apic)) - break; - goto no_apic; - default: - goto no_apic; - } - - if (!cpu_has_apic) { - /* - * Over-ride BIOS and try to enable the local APIC only if - * "lapic" specified. - */ - if (!force_enable_local_apic) { - printk(KERN_INFO "Local APIC disabled by BIOS -- " - "you can enable it with \"lapic\"\n"); - return -1; - } - /* - * Some BIOSes disable the local APIC in the APIC_BASE - * MSR. This can only be done in software for Intel P6 or later - * and AMD K7 (Model > 1) or later. - */ - rdmsr(MSR_IA32_APICBASE, l, h); - if (!(l & MSR_IA32_APICBASE_ENABLE)) { - printk(KERN_INFO - "Local APIC disabled by BIOS -- reenabling.\n"); - l &= ~MSR_IA32_APICBASE_BASE; - l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; - wrmsr(MSR_IA32_APICBASE, l, h); - enabled_via_apicbase = 1; - } - } - /* - * The APIC feature bit should now be enabled - * in `cpuid' - */ - features = cpuid_edx(1); - if (!(features & (1 << X86_FEATURE_APIC))) { - printk(KERN_WARNING "Could not enable APIC!\n"); - return -1; - } - set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); - mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; - - /* The BIOS may have set up the APIC at some other address */ - rdmsr(MSR_IA32_APICBASE, l, h); - if (l & MSR_IA32_APICBASE_ENABLE) - mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; - - printk(KERN_INFO "Found and enabled local APIC!\n"); - - apic_pm_activate(); - - return 0; - -no_apic: - printk(KERN_INFO "No local APIC present or hardware disabled\n"); - return -1; -} -#endif - -#ifdef CONFIG_X86_64 -void __init early_init_lapic_mapping(void) -{ - unsigned long phys_addr; - - /* - * If no local APIC can be found then go out - * : it means there is no mpatable and MADT - */ - if (!smp_found_config) - return; - - phys_addr = mp_lapic_addr; - - set_fixmap_nocache(FIX_APIC_BASE, phys_addr); - apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", - APIC_BASE, phys_addr); - - /* - * Fetch the APIC ID of the BSP in case we have a - * default configuration (or the MP table is broken). - */ - boot_cpu_physical_apicid = read_apic_id(); -} -#endif - -/** - * init_apic_mappings - initialize APIC mappings - */ -void __init init_apic_mappings(void) -{ -#ifdef HAVE_X2APIC - if (x2apic) { - boot_cpu_physical_apicid = read_apic_id(); - return; - } -#endif - - /* - * If no local APIC can be found then set up a fake all - * zeroes page to simulate the local APIC and another - * one for the IO-APIC. - */ - if (!smp_found_config && detect_init_APIC()) { - apic_phys = (unsigned long) alloc_bootmem_pages(PAGE_SIZE); - apic_phys = __pa(apic_phys); - } else - apic_phys = mp_lapic_addr; - - set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", - APIC_BASE, apic_phys); - - /* - * Fetch the APIC ID of the BSP in case we have a - * default configuration (or the MP table is broken). - */ - if (boot_cpu_physical_apicid == -1U) - boot_cpu_physical_apicid = read_apic_id(); -} - -/* - * This initializes the IO-APIC and APIC hardware if this is - * a UP kernel. - */ -int apic_version[MAX_APICS]; - -int __init APIC_init_uniprocessor(void) -{ -#ifdef CONFIG_X86_64 - if (disable_apic) { - printk(KERN_INFO "Apic disabled\n"); - return -1; - } - if (!cpu_has_apic) { - disable_apic = 1; - printk(KERN_INFO "Apic disabled by BIOS\n"); - return -1; - } -#else - if (!smp_found_config && !cpu_has_apic) - return -1; - - /* - * Complain if the BIOS pretends there is one. - */ - if (!cpu_has_apic && - APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { - printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", - boot_cpu_physical_apicid); - clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); - return -1; - } -#endif - -#ifdef HAVE_X2APIC - enable_IR_x2apic(); -#endif -#ifdef CONFIG_X86_64 - setup_apic_routing(); -#endif - - verify_local_APIC(); - connect_bsp_APIC(); - -#ifdef CONFIG_X86_64 - apic_write(APIC_ID, SET_APIC_ID(boot_cpu_physical_apicid)); -#else - /* - * Hack: In case of kdump, after a crash, kernel might be booting - * on a cpu with non-zero lapic id. But boot_cpu_physical_apicid - * might be zero if read from MP tables. Get it from LAPIC. - */ -# ifdef CONFIG_CRASH_DUMP - boot_cpu_physical_apicid = read_apic_id(); -# endif -#endif - physid_set_mask_of_physid(boot_cpu_physical_apicid, &phys_cpu_present_map); - setup_local_APIC(); - -#ifdef CONFIG_X86_64 - /* - * Now enable IO-APICs, actually call clear_IO_APIC - * We need clear_IO_APIC before enabling vector on BP - */ - if (!skip_ioapic_setup && nr_ioapics) - enable_IO_APIC(); -#endif - -#ifdef CONFIG_X86_IO_APIC - if (!smp_found_config || skip_ioapic_setup || !nr_ioapics) -#endif - localise_nmi_watchdog(); - end_local_APIC_setup(); - -#ifdef CONFIG_X86_IO_APIC - if (smp_found_config && !skip_ioapic_setup && nr_ioapics) - setup_IO_APIC(); -# ifdef CONFIG_X86_64 - else - nr_ioapics = 0; -# endif -#endif - -#ifdef CONFIG_X86_64 - setup_boot_APIC_clock(); - check_nmi_watchdog(); -#else - setup_boot_clock(); -#endif - - return 0; -} - -/* - * Local APIC interrupts - */ - -/* - * This interrupt should _never_ happen with our APIC/SMP architecture - */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_spurious_interrupt(void) -#else -void smp_spurious_interrupt(struct pt_regs *regs) -#endif -{ - u32 v; - -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - /* - * Check if this really is a spurious interrupt and ACK it - * if it is a vectored one. Just in case... - * Spurious interrupts should not be ACKed. - */ - v = apic_read(APIC_ISR + ((SPURIOUS_APIC_VECTOR & ~0x1f) >> 1)); - if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) - ack_APIC_irq(); - -#ifdef CONFIG_X86_64 - add_pda(irq_spurious_count, 1); -#else - /* see sw-dev-man vol 3, chapter 7.4.13.5 */ - printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " - "should never happen.\n", smp_processor_id()); - __get_cpu_var(irq_stat).irq_spurious_count++; -#endif - irq_exit(); -} - -/* - * This interrupt should never happen with our APIC/SMP architecture - */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_error_interrupt(void) -#else -void smp_error_interrupt(struct pt_regs *regs) -#endif -{ - u32 v, v1; - -#ifdef CONFIG_X86_64 - exit_idle(); -#endif - irq_enter(); - /* First tickle the hardware, only then report what went on. -- REW */ - v = apic_read(APIC_ESR); - apic_write(APIC_ESR, 0); - v1 = apic_read(APIC_ESR); - ack_APIC_irq(); - atomic_inc(&irq_err_count); - - /* Here is what the APIC error bits mean: - 0: Send CS error - 1: Receive CS error - 2: Send accept error - 3: Receive accept error - 4: Reserved - 5: Send illegal vector - 6: Received illegal vector - 7: Illegal register address - */ - printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", - smp_processor_id(), v , v1); - irq_exit(); -} - -/** - * connect_bsp_APIC - attach the APIC to the interrupt system - */ -void __init connect_bsp_APIC(void) -{ -#ifdef CONFIG_X86_32 - if (pic_mode) { - /* - * Do not trust the local APIC being empty at bootup. - */ - clear_local_APIC(); - /* - * PIC mode, enable APIC mode in the IMCR, i.e. connect BSP's - * local APIC to INT and NMI lines. - */ - apic_printk(APIC_VERBOSE, "leaving PIC mode, " - "enabling APIC mode.\n"); - outb(0x70, 0x22); - outb(0x01, 0x23); - } -#endif - enable_apic_mode(); -} - -/** - * disconnect_bsp_APIC - detach the APIC from the interrupt system - * @virt_wire_setup: indicates, whether virtual wire mode is selected - * - * Virtual wire mode is necessary to deliver legacy interrupts even when the - * APIC is disabled. - */ -void disconnect_bsp_APIC(int virt_wire_setup) -{ - unsigned int value; - -#ifdef CONFIG_X86_32 - if (pic_mode) { - /* - * Put the board back into PIC mode (has an effect only on - * certain older boards). Note that APIC interrupts, including - * IPIs, won't work beyond this point! The only exception are - * INIT IPIs. - */ - apic_printk(APIC_VERBOSE, "disabling APIC mode, " - "entering PIC mode.\n"); - outb(0x70, 0x22); - outb(0x00, 0x23); - return; - } -#endif - - /* Go back to Virtual Wire compatibility mode */ - - /* For the spurious interrupt use vector F, and enable it */ - value = apic_read(APIC_SPIV); - value &= ~APIC_VECTOR_MASK; - value |= APIC_SPIV_APIC_ENABLED; - value |= 0xf; - apic_write(APIC_SPIV, value); - - if (!virt_wire_setup) { - /* - * For LVT0 make it edge triggered, active high, - * external and enabled - */ - value = apic_read(APIC_LVT0); - value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | - APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | - APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); - value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; - value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_EXTINT); - apic_write(APIC_LVT0, value); - } else { - /* Disable LVT0 */ - apic_write(APIC_LVT0, APIC_LVT_MASKED); - } - - /* - * For LVT1 make it edge triggered, active high, - * nmi and enabled - */ - value = apic_read(APIC_LVT1); - value &= ~(APIC_MODE_MASK | APIC_SEND_PENDING | - APIC_INPUT_POLARITY | APIC_LVT_REMOTE_IRR | - APIC_LVT_LEVEL_TRIGGER | APIC_LVT_MASKED); - value |= APIC_LVT_REMOTE_IRR | APIC_SEND_PENDING; - value = SET_APIC_DELIVERY_MODE(value, APIC_MODE_NMI); - apic_write(APIC_LVT1, value); -} - -void __cpuinit generic_processor_info(int apicid, int version) -{ - int cpu; - cpumask_t tmp_map; - - /* - * Validate version - */ - if (version == 0x0) { - printk(KERN_WARNING "BIOS bug, APIC version is 0 for CPU#%d! " - "fixing up to 0x10. (tell your hw vendor)\n", - version); - version = 0x10; - } - apic_version[apicid] = version; - - if (num_processors >= NR_CPUS) { - printk(KERN_WARNING "WARNING: NR_CPUS limit of %i reached." - " Processor ignored.\n", NR_CPUS); - return; - } - - num_processors++; - cpus_complement(tmp_map, cpu_present_map); - cpu = first_cpu(tmp_map); - - physid_set(apicid, phys_cpu_present_map); - if (apicid == boot_cpu_physical_apicid) { - /* - * x86_bios_cpu_apicid is required to have processors listed - * in same order as logical cpu numbers. Hence the first - * entry is BSP, and so on. - */ - cpu = 0; - } - if (apicid > max_physical_apicid) - max_physical_apicid = apicid; - -#ifdef CONFIG_X86_32 - /* - * Would be preferable to switch to bigsmp when CONFIG_HOTPLUG_CPU=y - * but we need to work other dependencies like SMP_SUSPEND etc - * before this can be done without some confusion. - * if (CPU_HOTPLUG_ENABLED || num_processors > 8) - * - Ashok Raj - */ - if (max_physical_apicid >= 8) { - switch (boot_cpu_data.x86_vendor) { - case X86_VENDOR_INTEL: - if (!APIC_XAPIC(version)) { - def_to_bigsmp = 0; - break; - } - /* If P4 and above fall through */ - case X86_VENDOR_AMD: - def_to_bigsmp = 1; - } - } -#endif - -#if defined(CONFIG_X86_SMP) || defined(CONFIG_X86_64) - /* are we being called early in kernel startup? */ - if (early_per_cpu_ptr(x86_cpu_to_apicid)) { - u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid); - u16 *bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); - - cpu_to_apicid[cpu] = apicid; - bios_cpu_apicid[cpu] = apicid; - } else { - per_cpu(x86_cpu_to_apicid, cpu) = apicid; - per_cpu(x86_bios_cpu_apicid, cpu) = apicid; - } -#endif - - cpu_set(cpu, cpu_possible_map); - cpu_set(cpu, cpu_present_map); -} - -#ifdef CONFIG_X86_64 -int hard_smp_processor_id(void) -{ - return read_apic_id(); -} -#endif - -/* - * Power management - */ -#ifdef CONFIG_PM - -static struct { - /* - * 'active' is true if the local APIC was enabled by us and - * not the BIOS; this signifies that we are also responsible - * for disabling it before entering apm/acpi suspend - */ - int active; - /* r/w apic fields */ - unsigned int apic_id; - unsigned int apic_taskpri; - unsigned int apic_ldr; - unsigned int apic_dfr; - unsigned int apic_spiv; - unsigned int apic_lvtt; - unsigned int apic_lvtpc; - unsigned int apic_lvt0; - unsigned int apic_lvt1; - unsigned int apic_lvterr; - unsigned int apic_tmict; - unsigned int apic_tdcr; - unsigned int apic_thmr; -} apic_pm_state; - -static int lapic_suspend(struct sys_device *dev, pm_message_t state) -{ - unsigned long flags; - int maxlvt; - - if (!apic_pm_state.active) - return 0; - - maxlvt = lapic_get_maxlvt(); - - apic_pm_state.apic_id = apic_read(APIC_ID); - apic_pm_state.apic_taskpri = apic_read(APIC_TASKPRI); - apic_pm_state.apic_ldr = apic_read(APIC_LDR); - apic_pm_state.apic_dfr = apic_read(APIC_DFR); - apic_pm_state.apic_spiv = apic_read(APIC_SPIV); - apic_pm_state.apic_lvtt = apic_read(APIC_LVTT); - if (maxlvt >= 4) - apic_pm_state.apic_lvtpc = apic_read(APIC_LVTPC); - apic_pm_state.apic_lvt0 = apic_read(APIC_LVT0); - apic_pm_state.apic_lvt1 = apic_read(APIC_LVT1); - apic_pm_state.apic_lvterr = apic_read(APIC_LVTERR); - apic_pm_state.apic_tmict = apic_read(APIC_TMICT); - apic_pm_state.apic_tdcr = apic_read(APIC_TDCR); -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) - if (maxlvt >= 5) - apic_pm_state.apic_thmr = apic_read(APIC_LVTTHMR); -#endif - - local_irq_save(flags); - disable_local_APIC(); - local_irq_restore(flags); - return 0; -} - -static int lapic_resume(struct sys_device *dev) -{ - unsigned int l, h; - unsigned long flags; - int maxlvt; - - if (!apic_pm_state.active) - return 0; - - maxlvt = lapic_get_maxlvt(); - - local_irq_save(flags); - -#ifdef HAVE_X2APIC - if (x2apic) - enable_x2apic(); - else -#endif - { - /* - * Make sure the APICBASE points to the right address - * - * FIXME! This will be wrong if we ever support suspend on - * SMP! We'll need to do this as part of the CPU restore! - */ - rdmsr(MSR_IA32_APICBASE, l, h); - l &= ~MSR_IA32_APICBASE_BASE; - l |= MSR_IA32_APICBASE_ENABLE | mp_lapic_addr; - wrmsr(MSR_IA32_APICBASE, l, h); - } - - apic_write(APIC_LVTERR, ERROR_APIC_VECTOR | APIC_LVT_MASKED); - apic_write(APIC_ID, apic_pm_state.apic_id); - apic_write(APIC_DFR, apic_pm_state.apic_dfr); - apic_write(APIC_LDR, apic_pm_state.apic_ldr); - apic_write(APIC_TASKPRI, apic_pm_state.apic_taskpri); - apic_write(APIC_SPIV, apic_pm_state.apic_spiv); - apic_write(APIC_LVT0, apic_pm_state.apic_lvt0); - apic_write(APIC_LVT1, apic_pm_state.apic_lvt1); -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) - if (maxlvt >= 5) - apic_write(APIC_LVTTHMR, apic_pm_state.apic_thmr); -#endif - if (maxlvt >= 4) - apic_write(APIC_LVTPC, apic_pm_state.apic_lvtpc); - apic_write(APIC_LVTT, apic_pm_state.apic_lvtt); - apic_write(APIC_TDCR, apic_pm_state.apic_tdcr); - apic_write(APIC_TMICT, apic_pm_state.apic_tmict); - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - apic_write(APIC_LVTERR, apic_pm_state.apic_lvterr); - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); - - local_irq_restore(flags); - - return 0; -} - -/* - * This device has no shutdown method - fully functioning local APICs - * are needed on every CPU up until machine_halt/restart/poweroff. - */ - -static struct sysdev_class lapic_sysclass = { - .name = "lapic", - .resume = lapic_resume, - .suspend = lapic_suspend, -}; - -static struct sys_device device_lapic = { - .id = 0, - .cls = &lapic_sysclass, -}; - -static void __cpuinit apic_pm_activate(void) -{ - apic_pm_state.active = 1; -} - -static int __init init_lapic_sysfs(void) -{ - int error; - - if (!cpu_has_apic) - return 0; - /* XXX: remove suspend/resume procs if !apic_pm_state.active? */ - - error = sysdev_class_register(&lapic_sysclass); - if (!error) - error = sysdev_register(&device_lapic); - return error; -} -device_initcall(init_lapic_sysfs); - -#else /* CONFIG_PM */ - -static void apic_pm_activate(void) { } - -#endif /* CONFIG_PM */ - -#ifdef CONFIG_X86_64 -/* - * apic_is_clustered_box() -- Check if we can expect good TSC - * - * Thus far, the major user of this is IBM's Summit2 series: - * - * Clustered boxes may have unsynced TSC problems if they are - * multi-chassis. Use available data to take a good guess. - * If in doubt, go HPET. - */ -__cpuinit int apic_is_clustered_box(void) -{ - int i, clusters, zeros; - unsigned id; - u16 *bios_cpu_apicid; - DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS); - - /* - * there is not this kind of box with AMD CPU yet. - * Some AMD box with quadcore cpu and 8 sockets apicid - * will be [4, 0x23] or [8, 0x27] could be thought to - * vsmp box still need checking... - */ - if ((boot_cpu_data.x86_vendor == X86_VENDOR_AMD) && !is_vsmp_box()) - return 0; - - bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); - bitmap_zero(clustermap, NUM_APIC_CLUSTERS); - - for (i = 0; i < NR_CPUS; i++) { - /* are we being called early in kernel startup? */ - if (bios_cpu_apicid) { - id = bios_cpu_apicid[i]; - } - else if (i < nr_cpu_ids) { - if (cpu_present(i)) - id = per_cpu(x86_bios_cpu_apicid, i); - else - continue; - } - else - break; - - if (id != BAD_APICID) - __set_bit(APIC_CLUSTERID(id), clustermap); - } - - /* Problem: Partially populated chassis may not have CPUs in some of - * the APIC clusters they have been allocated. Only present CPUs have - * x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. - * Since clusters are allocated sequentially, count zeros only if - * they are bounded by ones. - */ - clusters = 0; - zeros = 0; - for (i = 0; i < NUM_APIC_CLUSTERS; i++) { - if (test_bit(i, clustermap)) { - clusters += 1 + zeros; - zeros = 0; - } else - ++zeros; - } - - /* ScaleMP vSMPowered boxes have one cluster per board and TSCs are - * not guaranteed to be synced between boards - */ - if (is_vsmp_box() && clusters > 1) - return 1; - - /* - * If clusters > 2, then should be multi-chassis. - * May have to revisit this when multi-core + hyperthreaded CPUs come - * out, but AFAIK this will work even for them. - */ - return (clusters > 2); -} -#endif - -/* - * APIC command line parameters - */ -static int __init setup_disableapic(char *arg) -{ - disable_apic = 1; - setup_clear_cpu_cap(X86_FEATURE_APIC); - return 0; -} -early_param("disableapic", setup_disableapic); - -/* same as disableapic, for compatibility */ -static int __init setup_nolapic(char *arg) -{ - return setup_disableapic(arg); -} -early_param("nolapic", setup_nolapic); - -static int __init parse_lapic_timer_c2_ok(char *arg) -{ - local_apic_timer_c2_ok = 1; - return 0; -} -early_param("lapic_timer_c2_ok", parse_lapic_timer_c2_ok); - -static int __init parse_disable_apic_timer(char *arg) -{ - disable_apic_timer = 1; - return 0; -} -early_param("noapictimer", parse_disable_apic_timer); - -static int __init parse_nolapic_timer(char *arg) -{ - disable_apic_timer = 1; - return 0; -} -early_param("nolapic_timer", parse_nolapic_timer); - -static int __init apic_set_verbosity(char *arg) -{ - if (!arg) { -#ifdef CONFIG_X86_64 - skip_ioapic_setup = 0; - return 0; -#endif - return -EINVAL; - } - - if (strcmp("debug", arg) == 0) - apic_verbosity = APIC_DEBUG; - else if (strcmp("verbose", arg) == 0) - apic_verbosity = APIC_VERBOSE; - else { - printk(KERN_WARNING "APIC Verbosity level %s not recognised" - " use apic=verbose or apic=debug\n", arg); - return -EINVAL; - } - - return 0; -} -early_param("apic", apic_set_verbosity); - -static int __init lapic_insert_resource(void) -{ - if (!apic_phys) - return -1; - - /* Put local APIC into the resource map. */ - lapic_resource.start = apic_phys; - lapic_resource.end = lapic_resource.start + PAGE_SIZE - 1; - insert_resource(&iomem_resource, &lapic_resource); - - return 0; -} - -/* - * need call insert after e820_reserve_resources() - * that is using request_resource - */ -late_initcall(lapic_insert_resource); -- cgit v1.2.3 From e492c5ae85428d4a3815d06bf308c590120b928b Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 24 Aug 2008 22:41:26 -0700 Subject: x86: let 64 bit to use 32 bit calibrate_apic_clock Use the 32-bit APIC calibration code - it's more mature. Signed-of-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 88 ++------------------------------------------------ 1 file changed, 2 insertions(+), 86 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 6e99af5ce678..ca5ef71f4208 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -478,90 +478,6 @@ static void __cpuinit setup_APIC_timer(void) clockevents_register_device(levt); } -#ifdef CONFIG_X86_64 -/* - * In this function we calibrate APIC bus clocks to the external - * timer. Unfortunately we cannot use jiffies and the timer irq - * to calibrate, since some later bootup code depends on getting - * the first irq? Ugh. - * - * We want to do the calibration only once since we - * want to have local timer irqs syncron. CPUs connected - * by the same APIC bus have the very same bus frequency. - * And we want to have irqs off anyways, no accidental - * APIC irq that way. - */ - -#define TICK_COUNT 100000000 - -static int __init calibrate_APIC_clock(void) -{ - unsigned apic, apic_start; - unsigned long tsc, tsc_start; - int result; - - local_irq_disable(); - - /* - * Put whatever arbitrary (but long enough) timeout - * value into the APIC clock, we just want to get the - * counter running for calibration. - * - * No interrupt enable ! - */ - __setup_APIC_LVTT(250000000, 0, 0); - - apic_start = apic_read(APIC_TMCCT); -#ifdef CONFIG_X86_PM_TIMER - if (apic_calibrate_pmtmr && pmtmr_ioport) { - pmtimer_wait(5000); /* 5ms wait */ - apic = apic_read(APIC_TMCCT); - result = (apic_start - apic) * 1000L / 5; - } else -#endif - { - rdtscll(tsc_start); - - do { - apic = apic_read(APIC_TMCCT); - rdtscll(tsc); - } while ((tsc - tsc_start) < TICK_COUNT && - (apic_start - apic) < TICK_COUNT); - - result = (apic_start - apic) * 1000L * tsc_khz / - (tsc - tsc_start); - } - - local_irq_enable(); - - printk(KERN_DEBUG "APIC timer calibration result %d\n", result); - - printk(KERN_INFO "Detected %d.%03d MHz APIC timer.\n", - result / 1000 / 1000, result / 1000 % 1000); - - /* Calculate the scaled math multiplication factor */ - lapic_clockevent.mult = div_sc(result, NSEC_PER_SEC, - lapic_clockevent.shift); - lapic_clockevent.max_delta_ns = - clockevent_delta2ns(0x7FFFFF, &lapic_clockevent); - lapic_clockevent.min_delta_ns = - clockevent_delta2ns(0xF, &lapic_clockevent); - - calibration_result = (result * APIC_DIVISOR) / HZ; - - /* - * Do a sanity check on the APIC calibration result - */ - if (calibration_result < (1000000 / HZ)) { - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); - return -1; - } - - return 0; -} - -#else /* * In this functions we calibrate APIC bus clocks to the external timer. * @@ -659,6 +575,7 @@ static int __init calibrate_APIC_clock(void) delta = lapic_cal_t1 - lapic_cal_t2; apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); +#ifdef CONFIG_X86_PM_TIMER /* Check, if the PM timer is available */ deltapm = lapic_cal_pm2 - lapic_cal_pm1; apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); @@ -687,6 +604,7 @@ static int __init calibrate_APIC_clock(void) } pm_referenced = 1; } +#endif /* Calculate the scaled math multiplication factor */ lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, @@ -773,8 +691,6 @@ static int __init calibrate_APIC_clock(void) return 0; } -#endif - /* * Setup the boot APIC * -- cgit v1.2.3 From 1dd6ba2e179773597e20f17f66049a64e6c4b2ec Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Mon, 25 Aug 2008 21:27:26 +0400 Subject: x86: apic - unify smp_spurious/error_interrupt declaration According to entry_64.S we do pass pt_regs pointer into interrupt handlers but don't use them. So we safely may merge the declarations. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 8 -------- 1 file changed, 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index ca5ef71f4208..0556f375e40b 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1659,11 +1659,7 @@ int __init APIC_init_uniprocessor(void) /* * This interrupt should _never_ happen with our APIC/SMP architecture */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_spurious_interrupt(void) -#else void smp_spurious_interrupt(struct pt_regs *regs) -#endif { u32 v; @@ -1694,11 +1690,7 @@ void smp_spurious_interrupt(struct pt_regs *regs) /* * This interrupt should never happen with our APIC/SMP architecture */ -#ifdef CONFIG_X86_64 -asmlinkage void smp_error_interrupt(void) -#else void smp_error_interrupt(struct pt_regs *regs) -#endif { u32 v, v1; -- cgit v1.2.3 From a11b5abef50722e42a7d13f6b799c4f606fcb797 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 3 Sep 2008 16:58:31 -0700 Subject: x2apic: fix reserved APIC register accesses in print_local_APIC() APIC_ARBPRI is a reserved register for XAPIC and beyond. APIC_RRR is a reserved register except for 82489DX, APIC for Pentium processors. APIC_EOI is a write only register. APIC_DFR is reserved in x2apic mode. Access to these registers in x2apic will result in #GP fault. Fix these apic register accesses. Signed-off-by: Yinghai Lu Signed-off-by: Suresh Siddha Cc: Maciej W. Rozycki Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 27 ++++++++++++++++++--------- 1 file changed, 18 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index d28128e0392c..93ceb19d1e90 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1751,21 +1751,30 @@ __apicdebuginit(void) print_local_APIC(void *dummy) printk(KERN_DEBUG "... APIC TASKPRI: %08x (%02x)\n", v, v & APIC_TPRI_MASK); if (APIC_INTEGRATED(ver)) { /* !82489DX */ - v = apic_read(APIC_ARBPRI); - printk(KERN_DEBUG "... APIC ARBPRI: %08x (%02x)\n", v, - v & APIC_ARBPRI_MASK); + if (!APIC_XAPIC(ver)) { + v = apic_read(APIC_ARBPRI); + printk(KERN_DEBUG "... APIC ARBPRI: %08x (%02x)\n", v, + v & APIC_ARBPRI_MASK); + } v = apic_read(APIC_PROCPRI); printk(KERN_DEBUG "... APIC PROCPRI: %08x\n", v); } - v = apic_read(APIC_EOI); - printk(KERN_DEBUG "... APIC EOI: %08x\n", v); - v = apic_read(APIC_RRR); - printk(KERN_DEBUG "... APIC RRR: %08x\n", v); + /* + * Remote read supported only in the 82489DX and local APIC for + * Pentium processors. + */ + if (!APIC_INTEGRATED(ver) || maxlvt == 3) { + v = apic_read(APIC_RRR); + printk(KERN_DEBUG "... APIC RRR: %08x\n", v); + } + v = apic_read(APIC_LDR); printk(KERN_DEBUG "... APIC LDR: %08x\n", v); - v = apic_read(APIC_DFR); - printk(KERN_DEBUG "... APIC DFR: %08x\n", v); + if (!x2apic_enabled()) { + v = apic_read(APIC_DFR); + printk(KERN_DEBUG "... APIC DFR: %08x\n", v); + } v = apic_read(APIC_SPIV); printk(KERN_DEBUG "... APIC SPIV: %08x\n", v); -- cgit v1.2.3 From 02c1df199c990cd21c09a4dffaa06d4e0b7bf2bf Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 4 Sep 2008 20:57:11 +0200 Subject: x86: print out if acpi want physical flat of all Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/genapic_flat_64.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index 9eca5ba7a6b1..2ec2de8d8c46 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -179,8 +179,10 @@ static int __init physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) * is an example). */ if (acpi_gbl_FADT.header.revision > FADT2_REVISION_ID && - (acpi_gbl_FADT.flags & ACPI_FADT_APIC_PHYSICAL)) + (acpi_gbl_FADT.flags & ACPI_FADT_APIC_PHYSICAL)) { + printk(KERN_DEBUG "system APIC only can use physical flat"); return 1; + } #endif return 0; -- cgit v1.2.3 From 676f4a920be27160747439fe71026aa15ec78e5a Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 4 Sep 2008 22:37:49 +0400 Subject: x86: io-apic - use ARRAY_SIZE macro Make the code width a bit shorter with ARRAY_SIZE macro. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 93ceb19d1e90..9b01fdadcb9b 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -166,7 +166,7 @@ static void __init init_work(void *data) memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - legacy_count = sizeof(irq_cfg_legacy)/sizeof(irq_cfg_legacy[0]); + legacy_count = ARRAY_SIZE(irq_cfg_legacy); for (i = legacy_count; i < *da->nr; i++) init_one_irq_cfg(&cfg[i]); -- cgit v1.2.3 From ac54a6c9371bacb86bee1db23f7d82e8685c7e17 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 4 Sep 2008 22:37:50 +0400 Subject: x86: io-apic - declare irq_cfg_lock for SPARSE_IRQ only We use irq_cfg_lock lock in SPARSE_IRQ only context so move it under #ifdef and compiler will be happy. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9b01fdadcb9b..d22fecf828b8 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -147,14 +147,15 @@ static void init_one_irq_cfg(struct irq_cfg *cfg) static struct irq_cfg *irq_cfgx; +#ifdef CONFIG_HAVE_SPARSE_IRQ /* * Protect the irq_cfgx_free freelist: */ static DEFINE_SPINLOCK(irq_cfg_lock); -#ifdef CONFIG_HAVE_SPARSE_IRQ static struct irq_cfg *irq_cfgx_free; #endif + static void __init init_work(void *data) { struct dyn_array *da = data; -- cgit v1.2.3 From b40d575bf0679c45aaf9e1161fc51a6b041b7210 Mon Sep 17 00:00:00 2001 From: "venkatesh.pallipadi@intel.com" Date: Fri, 5 Sep 2008 18:02:16 -0700 Subject: x86: HPET_MSI Refactor code in preparation for HPET_MSI Preparatory patch before the actual HPET MSI changes. Sets up hpet_set_mode and hpet_next_event for the MSI related changes. Just the code refactoring and should be zero functional change. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Shaohua Li Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 42 +++++++++++++++++++++++++++--------------- 1 file changed, 27 insertions(+), 15 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index acf62fc233da..f7cb5e9e261e 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -227,44 +227,44 @@ static void hpet_legacy_clockevent_register(void) printk(KERN_DEBUG "hpet clockevent registered\n"); } -static void hpet_legacy_set_mode(enum clock_event_mode mode, - struct clock_event_device *evt) +static void hpet_set_mode(enum clock_event_mode mode, + struct clock_event_device *evt, int timer) { unsigned long cfg, cmp, now; uint64_t delta; switch(mode) { case CLOCK_EVT_MODE_PERIODIC: - delta = ((uint64_t)(NSEC_PER_SEC/HZ)) * hpet_clockevent.mult; - delta >>= hpet_clockevent.shift; + delta = ((uint64_t)(NSEC_PER_SEC/HZ)) * evt->mult; + delta >>= evt->shift; now = hpet_readl(HPET_COUNTER); cmp = now + (unsigned long) delta; - cfg = hpet_readl(HPET_T0_CFG); + cfg = hpet_readl(HPET_Tn_CFG(timer)); cfg |= HPET_TN_ENABLE | HPET_TN_PERIODIC | HPET_TN_SETVAL | HPET_TN_32BIT; - hpet_writel(cfg, HPET_T0_CFG); + hpet_writel(cfg, HPET_Tn_CFG(timer)); /* * The first write after writing TN_SETVAL to the * config register sets the counter value, the second * write sets the period. */ - hpet_writel(cmp, HPET_T0_CMP); + hpet_writel(cmp, HPET_Tn_CMP(timer)); udelay(1); - hpet_writel((unsigned long) delta, HPET_T0_CMP); + hpet_writel((unsigned long) delta, HPET_Tn_CMP(timer)); break; case CLOCK_EVT_MODE_ONESHOT: - cfg = hpet_readl(HPET_T0_CFG); + cfg = hpet_readl(HPET_Tn_CFG(timer)); cfg &= ~HPET_TN_PERIODIC; cfg |= HPET_TN_ENABLE | HPET_TN_32BIT; - hpet_writel(cfg, HPET_T0_CFG); + hpet_writel(cfg, HPET_Tn_CFG(timer)); break; case CLOCK_EVT_MODE_UNUSED: case CLOCK_EVT_MODE_SHUTDOWN: - cfg = hpet_readl(HPET_T0_CFG); + cfg = hpet_readl(HPET_Tn_CFG(timer)); cfg &= ~HPET_TN_ENABLE; - hpet_writel(cfg, HPET_T0_CFG); + hpet_writel(cfg, HPET_Tn_CFG(timer)); break; case CLOCK_EVT_MODE_RESUME: @@ -273,14 +273,14 @@ static void hpet_legacy_set_mode(enum clock_event_mode mode, } } -static int hpet_legacy_next_event(unsigned long delta, - struct clock_event_device *evt) +static int hpet_next_event(unsigned long delta, + struct clock_event_device *evt, int timer) { u32 cnt; cnt = hpet_readl(HPET_COUNTER); cnt += (u32) delta; - hpet_writel(cnt, HPET_T0_CMP); + hpet_writel(cnt, HPET_Tn_CMP(timer)); /* * We need to read back the CMP register to make sure that @@ -292,6 +292,18 @@ static int hpet_legacy_next_event(unsigned long delta, return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } +static void hpet_legacy_set_mode(enum clock_event_mode mode, + struct clock_event_device *evt) +{ + hpet_set_mode(mode, evt, 0); +} + +static int hpet_legacy_next_event(unsigned long delta, + struct clock_event_device *evt) +{ + return hpet_next_event(delta, evt, 0); +} + /* * Clock source related code */ -- cgit v1.2.3 From 58ac1e76ce77d515bd5cb65dbc465a040da341c6 Mon Sep 17 00:00:00 2001 From: "venkatesh.pallipadi@intel.com" Date: Fri, 5 Sep 2008 18:02:17 -0700 Subject: x86: HPET_MSI Basic HPET_MSI setup code Basic HPET MSI setup code. Routines to perform basic MSI read write in HPET memory map and setting up irq_chip for HPET MSI. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Shaohua Li Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 54 +++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/io_apic.c | 64 +++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 118 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index f7cb5e9e261e..3f10d16a8348 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -6,6 +6,8 @@ #include #include #include +#include +#include #include #include @@ -25,6 +27,15 @@ unsigned long hpet_address; static void __iomem *hpet_virt_address; +struct hpet_dev { + struct clock_event_device evt; + unsigned int num; + int cpu; + unsigned int irq; + unsigned int flags; + char name[10]; +}; + unsigned long hpet_readl(unsigned long a) { return readl(hpet_virt_address + a); @@ -304,6 +315,49 @@ static int hpet_legacy_next_event(unsigned long delta, return hpet_next_event(delta, evt, 0); } +/* + * HPET MSI Support + */ + +void hpet_msi_unmask(unsigned int irq) +{ + struct hpet_dev *hdev = get_irq_data(irq); + unsigned long cfg; + + /* unmask it */ + cfg = hpet_readl(HPET_Tn_CFG(hdev->num)); + cfg |= HPET_TN_FSB; + hpet_writel(cfg, HPET_Tn_CFG(hdev->num)); +} + +void hpet_msi_mask(unsigned int irq) +{ + unsigned long cfg; + struct hpet_dev *hdev = get_irq_data(irq); + + /* mask it */ + cfg = hpet_readl(HPET_Tn_CFG(hdev->num)); + cfg &= ~HPET_TN_FSB; + hpet_writel(cfg, HPET_Tn_CFG(hdev->num)); +} + +void hpet_msi_write(unsigned int irq, struct msi_msg *msg) +{ + struct hpet_dev *hdev = get_irq_data(irq); + + hpet_writel(msg->data, HPET_Tn_ROUTE(hdev->num)); + hpet_writel(msg->address_lo, HPET_Tn_ROUTE(hdev->num) + 4); +} + +void hpet_msi_read(unsigned int irq, struct msi_msg *msg) +{ + struct hpet_dev *hdev = get_irq_data(irq); + + msg->data = hpet_readl(HPET_Tn_ROUTE(hdev->num)); + msg->address_lo = hpet_readl(HPET_Tn_ROUTE(hdev->num) + 4); + msg->address_hi = 0; +} + /* * Clock source related code */ diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index d22fecf828b8..77fa155becf6 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -41,6 +41,7 @@ #endif #include #include +#include #include #include @@ -56,6 +57,7 @@ #include #include #include +#include #include #include @@ -3522,6 +3524,68 @@ int arch_setup_dmar_msi(unsigned int irq) } #endif +#ifdef CONFIG_HPET_TIMER + +#ifdef CONFIG_SMP +static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) +{ + struct irq_cfg *cfg; + struct irq_desc *desc; + struct msi_msg msg; + unsigned int dest; + cpumask_t tmp; + + cpus_and(tmp, mask, cpu_online_map); + if (cpus_empty(tmp)) + return; + + if (assign_irq_vector(irq, mask)) + return; + + cfg = irq_cfg(irq); + cpus_and(tmp, cfg->domain, mask); + dest = cpu_mask_to_apicid(tmp); + + hpet_msi_read(irq, &msg); + + msg.data &= ~MSI_DATA_VECTOR_MASK; + msg.data |= MSI_DATA_VECTOR(cfg->vector); + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; + msg.address_lo |= MSI_ADDR_DEST_ID(dest); + + hpet_msi_write(irq, &msg); + desc = irq_to_desc(irq); + desc->affinity = mask; +} +#endif /* CONFIG_SMP */ + +struct irq_chip hpet_msi_type = { + .name = "HPET_MSI", + .unmask = hpet_msi_unmask, + .mask = hpet_msi_mask, + .ack = ack_apic_edge, +#ifdef CONFIG_SMP + .set_affinity = hpet_msi_set_affinity, +#endif + .retrigger = ioapic_retrigger_irq, +}; + +int arch_setup_hpet_msi(unsigned int irq) +{ + int ret; + struct msi_msg msg; + + ret = msi_compose_msg(NULL, irq, &msg); + if (ret < 0) + return ret; + + hpet_msi_write(irq, &msg); + set_irq_chip_and_handler_name(irq, &hpet_msi_type, handle_edge_irq, + "edge"); + return 0; +} +#endif + #endif /* CONFIG_PCI_MSI */ /* * Hypertransport interrupt support -- cgit v1.2.3 From 4588c1f0354ac96a358b3f9e8e4331c51cf3336f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sat, 6 Sep 2008 14:19:17 +0200 Subject: x86: HPET_MSI Basic HPET_MSI setup code, cleanups small style cleanups. Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 50 +++++++++++++++++++++++++------------------------- 1 file changed, 25 insertions(+), 25 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 3f10d16a8348..03d3655734b4 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -1,39 +1,39 @@ #include #include +#include +#include #include #include #include #include -#include -#include -#include #include +#include +#include #include -#include #include -#include +#include -#define HPET_MASK CLOCKSOURCE_MASK(32) -#define HPET_SHIFT 22 +#define HPET_MASK CLOCKSOURCE_MASK(32) +#define HPET_SHIFT 22 /* FSEC = 10^-15 NSEC = 10^-9 */ -#define FSEC_PER_NSEC 1000000L +#define FSEC_PER_NSEC 1000000L /* * HPET address is set in acpi/boot.c, when an ACPI entry exists */ -unsigned long hpet_address; -static void __iomem *hpet_virt_address; +unsigned long hpet_address; +static void __iomem *hpet_virt_address; struct hpet_dev { - struct clock_event_device evt; - unsigned int num; - int cpu; - unsigned int irq; - unsigned int flags; - char name[10]; + struct clock_event_device evt; + unsigned int num; + int cpu; + unsigned int irq; + unsigned int flags; + char name[10]; }; unsigned long hpet_readl(unsigned long a) @@ -70,7 +70,7 @@ static inline void hpet_clear_mapping(void) static int boot_hpet_disable; int hpet_force_user; -static int __init hpet_setup(char* str) +static int __init hpet_setup(char *str) { if (str) { if (!strncmp("disable", str, 7)) @@ -91,7 +91,7 @@ __setup("nohpet", disable_hpet); static inline int is_hpet_capable(void) { - return (!boot_hpet_disable && hpet_address); + return !boot_hpet_disable && hpet_address; } /* @@ -122,10 +122,10 @@ static void hpet_reserve_platform_timers(unsigned long id) nrtimers = ((id & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT) + 1; - memset(&hd, 0, sizeof (hd)); - hd.hd_phys_address = hpet_address; - hd.hd_address = hpet; - hd.hd_nirqs = nrtimers; + memset(&hd, 0, sizeof(hd)); + hd.hd_phys_address = hpet_address; + hd.hd_address = hpet; + hd.hd_nirqs = nrtimers; hpet_reserve_timer(&hd, 0); #ifdef CONFIG_HPET_EMULATE_RTC @@ -141,8 +141,8 @@ static void hpet_reserve_platform_timers(unsigned long id) hd.hd_irq[1] = HPET_LEGACY_RTC; for (i = 2; i < nrtimers; timer++, i++) { - hd.hd_irq[i] = (readl(&timer->hpet_config) & Tn_INT_ROUTE_CNF_MASK) >> - Tn_INT_ROUTE_CNF_SHIFT; + hd.hd_irq[i] = (readl(&timer->hpet_config) & + Tn_INT_ROUTE_CNF_MASK) >> Tn_INT_ROUTE_CNF_SHIFT; } hpet_alloc(&hd); @@ -244,7 +244,7 @@ static void hpet_set_mode(enum clock_event_mode mode, unsigned long cfg, cmp, now; uint64_t delta; - switch(mode) { + switch (mode) { case CLOCK_EVT_MODE_PERIODIC: delta = ((uint64_t)(NSEC_PER_SEC/HZ)) * evt->mult; delta >>= evt->shift; -- cgit v1.2.3 From 26afe5f2fbf06ea0765aaa316640c4dd472310c0 Mon Sep 17 00:00:00 2001 From: "venkatesh.pallipadi@intel.com" Date: Fri, 5 Sep 2008 18:02:18 -0700 Subject: x86: HPET_MSI Initialise per-cpu HPET timers Initialize a per CPU HPET MSI timer when possible. We retain the HPET timer 0 (IRQ 0) and timer 1 (IRQ 8) as is when legacy mode is being used. We setup the remaining HPET timers as per CPU MSI based timers. This per CPU timer will eliminate the need for timer broadcasting with IRQ 0 when there is non-functional LAPIC timer across CPU deep C-states. If there are more CPUs than number of available timers, CPUs that do not find any timer to use will continue using LAPIC and IRQ 0 broadcast. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Shaohua Li Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 295 ++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 293 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 03d3655734b4..31e9191b7e19 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -21,10 +21,19 @@ NSEC = 10^-9 */ #define FSEC_PER_NSEC 1000000L +#define HPET_DEV_USED_BIT 2 +#define HPET_DEV_USED (1 << HPET_DEV_USED_BIT) +#define HPET_DEV_VALID 0x8 +#define HPET_DEV_FSB_CAP 0x1000 +#define HPET_DEV_PERI_CAP 0x2000 + +#define EVT_TO_HPET_DEV(evt) container_of(evt, struct hpet_dev, evt) + /* * HPET address is set in acpi/boot.c, when an ACPI entry exists */ unsigned long hpet_address; +unsigned long hpet_num_timers; static void __iomem *hpet_virt_address; struct hpet_dev { @@ -36,6 +45,10 @@ struct hpet_dev { char name[10]; }; +static struct hpet_dev *hpet_devs; + +static DEFINE_PER_CPU(struct hpet_dev *, cpu_hpet_dev); + unsigned long hpet_readl(unsigned long a) { return readl(hpet_virt_address + a); @@ -145,6 +158,16 @@ static void hpet_reserve_platform_timers(unsigned long id) Tn_INT_ROUTE_CNF_MASK) >> Tn_INT_ROUTE_CNF_SHIFT; } + for (i = 0; i < nrtimers; i++) { + struct hpet_dev *hdev = &hpet_devs[i]; + + if (!(hdev->flags & HPET_DEV_VALID)) + continue; + + hd.hd_irq[hdev->num] = hdev->irq; + hpet_reserve_timer(&hd, hdev->num); + } + hpet_alloc(&hd); } @@ -238,6 +261,8 @@ static void hpet_legacy_clockevent_register(void) printk(KERN_DEBUG "hpet clockevent registered\n"); } +static int hpet_setup_msi_irq(unsigned int irq); + static void hpet_set_mode(enum clock_event_mode mode, struct clock_event_device *evt, int timer) { @@ -279,7 +304,15 @@ static void hpet_set_mode(enum clock_event_mode mode, break; case CLOCK_EVT_MODE_RESUME: - hpet_enable_legacy_int(); + if (timer == 0) { + hpet_enable_legacy_int(); + } else { + struct hpet_dev *hdev = EVT_TO_HPET_DEV(evt); + hpet_setup_msi_irq(hdev->irq); + disable_irq(hdev->irq); + irq_set_affinity(hdev->irq, cpumask_of_cpu(hdev->cpu)); + enable_irq(hdev->irq); + } break; } } @@ -318,7 +351,7 @@ static int hpet_legacy_next_event(unsigned long delta, /* * HPET MSI Support */ - +#ifdef CONFIG_PCI_MSI void hpet_msi_unmask(unsigned int irq) { struct hpet_dev *hdev = get_irq_data(irq); @@ -358,6 +391,253 @@ void hpet_msi_read(unsigned int irq, struct msi_msg *msg) msg->address_hi = 0; } +static void hpet_msi_set_mode(enum clock_event_mode mode, + struct clock_event_device *evt) +{ + struct hpet_dev *hdev = EVT_TO_HPET_DEV(evt); + hpet_set_mode(mode, evt, hdev->num); +} + +static int hpet_msi_next_event(unsigned long delta, + struct clock_event_device *evt) +{ + struct hpet_dev *hdev = EVT_TO_HPET_DEV(evt); + return hpet_next_event(delta, evt, hdev->num); +} + +static int hpet_setup_msi_irq(unsigned int irq) +{ + if (arch_setup_hpet_msi(irq)) { + destroy_irq(irq); + return -EINVAL; + } + return 0; +} + +static int hpet_assign_irq(struct hpet_dev *dev) +{ + unsigned int irq; + + irq = create_irq(); + if (!irq) + return -EINVAL; + + set_irq_data(irq, dev); + + if (hpet_setup_msi_irq(irq)) + return -EINVAL; + + dev->irq = irq; + return 0; +} + +static irqreturn_t hpet_interrupt_handler(int irq, void *data) +{ + struct hpet_dev *dev = (struct hpet_dev *)data; + struct clock_event_device *hevt = &dev->evt; + + if (!hevt->event_handler) { + printk(KERN_INFO "Spurious HPET timer interrupt on HPET timer %d\n", + dev->num); + return IRQ_HANDLED; + } + + hevt->event_handler(hevt); + return IRQ_HANDLED; +} + +static int hpet_setup_irq(struct hpet_dev *dev) +{ + + if (request_irq(dev->irq, hpet_interrupt_handler, + IRQF_SHARED|IRQF_NOBALANCING, dev->name, dev)) + return -1; + + disable_irq(dev->irq); + irq_set_affinity(dev->irq, cpumask_of_cpu(dev->cpu)); + enable_irq(dev->irq); + + return 0; +} + +/* This should be called in specific @cpu */ +static void init_one_hpet_msi_clockevent(struct hpet_dev *hdev, int cpu) +{ + struct clock_event_device *evt = &hdev->evt; + uint64_t hpet_freq; + + WARN_ON(cpu != smp_processor_id()); + if (!(hdev->flags & HPET_DEV_VALID)) + return; + + if (hpet_setup_msi_irq(hdev->irq)) + return; + + hdev->cpu = cpu; + per_cpu(cpu_hpet_dev, cpu) = hdev; + evt->name = hdev->name; + hpet_setup_irq(hdev); + evt->irq = hdev->irq; + + evt->rating = 110; + evt->features = CLOCK_EVT_FEAT_ONESHOT; + if (hdev->flags & HPET_DEV_PERI_CAP) + evt->features |= CLOCK_EVT_FEAT_PERIODIC; + + evt->set_mode = hpet_msi_set_mode; + evt->set_next_event = hpet_msi_next_event; + evt->shift = 32; + + /* + * The period is a femto seconds value. We need to calculate the + * scaled math multiplication factor for nanosecond to hpet tick + * conversion. + */ + hpet_freq = 1000000000000000ULL; + do_div(hpet_freq, hpet_period); + evt->mult = div_sc((unsigned long) hpet_freq, + NSEC_PER_SEC, evt->shift); + /* Calculate the max delta */ + evt->max_delta_ns = clockevent_delta2ns(0x7FFFFFFF, evt); + /* 5 usec minimum reprogramming delta. */ + evt->min_delta_ns = 5000; + + evt->cpumask = cpumask_of_cpu(hdev->cpu); + clockevents_register_device(evt); +} + +#ifdef CONFIG_HPET +/* Reserve at least one timer for userspace (/dev/hpet) */ +#define RESERVE_TIMERS 1 +#else +#define RESERVE_TIMERS 0 +#endif +void hpet_msi_capability_lookup(unsigned int start_timer) +{ + unsigned int id; + unsigned int num_timers; + unsigned int num_timers_used = 0; + int i; + + id = hpet_readl(HPET_ID); + + num_timers = ((id & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT); + num_timers++; /* Value read out starts from 0 */ + + hpet_devs = kzalloc(sizeof(struct hpet_dev) * num_timers, GFP_KERNEL); + if (!hpet_devs) + return; + + hpet_num_timers = num_timers; + + for (i = start_timer; i < num_timers - RESERVE_TIMERS; i++) { + struct hpet_dev *hdev = &hpet_devs[num_timers_used]; + unsigned long cfg = hpet_readl(HPET_Tn_CFG(i)); + + /* Only consider HPET timer with MSI support */ + if (!(cfg & HPET_TN_FSB_CAP)) + continue; + + hdev->flags = 0; + if (cfg & HPET_TN_PERIODIC_CAP) + hdev->flags |= HPET_DEV_PERI_CAP; + hdev->num = i; + + sprintf(hdev->name, "hpet%d", i); + if (hpet_assign_irq(hdev)) + continue; + + hdev->flags |= HPET_DEV_FSB_CAP; + hdev->flags |= HPET_DEV_VALID; + num_timers_used++; + if (num_timers_used == num_possible_cpus()) + break; + } + + printk(KERN_INFO "HPET: %d timers in total, %d timers will be used for per-cpu timer\n", + num_timers, num_timers_used); +} + +static struct hpet_dev *hpet_get_unused_timer(void) +{ + int i; + + if (!hpet_devs) + return NULL; + + for (i = 0; i < hpet_num_timers; i++) { + struct hpet_dev *hdev = &hpet_devs[i]; + + if (!(hdev->flags & HPET_DEV_VALID)) + continue; + if (test_and_set_bit(HPET_DEV_USED_BIT, + (unsigned long *)&hdev->flags)) + continue; + return hdev; + } + return NULL; +} + +struct hpet_work_struct { + struct delayed_work work; + struct completion complete; +}; + +static void hpet_work(struct work_struct *w) +{ + struct hpet_dev *hdev; + int cpu = smp_processor_id(); + struct hpet_work_struct *hpet_work; + + hpet_work = container_of(w, struct hpet_work_struct, work.work); + + hdev = hpet_get_unused_timer(); + if (hdev) + init_one_hpet_msi_clockevent(hdev, cpu); + + complete(&hpet_work->complete); +} + +static int hpet_cpuhp_notify(struct notifier_block *n, + unsigned long action, void *hcpu) +{ + unsigned long cpu = (unsigned long)hcpu; + struct hpet_work_struct work; + struct hpet_dev *hdev = per_cpu(cpu_hpet_dev, cpu); + + switch (action & 0xf) { + case CPU_ONLINE: + INIT_DELAYED_WORK(&work.work, hpet_work); + init_completion(&work.complete); + /* FIXME: add schedule_work_on() */ + schedule_delayed_work_on(cpu, &work.work, 0); + wait_for_completion(&work.complete); + break; + case CPU_DEAD: + if (hdev) { + free_irq(hdev->irq, hdev); + hdev->flags &= ~HPET_DEV_USED; + per_cpu(cpu_hpet_dev, cpu) = NULL; + } + break; + } + return NOTIFY_OK; +} +#else + +void hpet_msi_capability_lookup(unsigned int start_timer) +{ + return; +} + +static int hpet_cpuhp_notify(struct notifier_block *n, + unsigned long action, void *hcpu) +{ + return NOTIFY_OK; +} + +#endif + /* * Clock source related code */ @@ -493,8 +773,10 @@ int __init hpet_enable(void) if (id & HPET_ID_LEGSUP) { hpet_legacy_clockevent_register(); + hpet_msi_capability_lookup(2); return 1; } + hpet_msi_capability_lookup(0); return 0; out_nohpet: @@ -511,6 +793,8 @@ out_nohpet: */ static __init int hpet_late_init(void) { + int cpu; + if (boot_hpet_disable) return -ENODEV; @@ -526,6 +810,13 @@ static __init int hpet_late_init(void) hpet_reserve_platform_timers(hpet_readl(HPET_ID)); + for_each_online_cpu(cpu) { + hpet_cpuhp_notify(NULL, CPU_ONLINE, (void *)(long)cpu); + } + + /* This notifier should be called after workqueue is ready */ + hotcpu_notifier(hpet_cpuhp_notify, -20); + return 0; } fs_initcall(hpet_late_init); -- cgit v1.2.3 From 3c2cbd2490656fb4b6ede586c557a2b09811a432 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sat, 6 Sep 2008 14:15:33 +0400 Subject: x86: io-apic - code style cleaning for setup_IO_APIC_irqs By changing printout form we are able to shrink (and clean up) code a bit. Former printout example: init IO_APIC IRQs IO-APIC (apicid-pin) 1-1, 1-2, 1-3 not connected. IO-APIC (apicid-pin) 2-1, 2-2, 2-3 not connected. New printout example: init IO_APIC IRQs 1-1 1-2 1-3 (apicid-pin) not connected 2-1 2-2 2-3 (apicid-pin) not connected Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 53 +++++++++++++++++++++++++---------------------- 1 file changed, 28 insertions(+), 25 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 77fa155becf6..4040d575a21e 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1518,41 +1518,44 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, static void __init setup_IO_APIC_irqs(void) { - int apic, pin, idx, irq, first_notcon = 1; + int apic, pin, idx, irq; + int notcon = 0; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - - idx = find_irq_entry(apic,pin,mp_INT); - if (idx == -1) { - if (first_notcon) { - apic_printk(APIC_VERBOSE, KERN_DEBUG " IO-APIC (apicid-pin) %d-%d", mp_ioapics[apic].mp_apicid, pin); - first_notcon = 0; - } else - apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mp_apicid, pin); - continue; - } - if (!first_notcon) { - apic_printk(APIC_VERBOSE, " not connected.\n"); - first_notcon = 1; - } + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + + idx = find_irq_entry(apic, pin, mp_INT); + if (idx == -1) { + apic_printk(APIC_VERBOSE, + KERN_DEBUG " %d-%d", + mp_ioapics[apic].mp_apicid, pin); + if (!notcon) + notcon = 1; + continue; + } - irq = pin_2_irq(idx, apic, pin); + irq = pin_2_irq(idx, apic, pin); #ifdef CONFIG_X86_32 - if (multi_timer_check(apic, irq)) - continue; + if (multi_timer_check(apic, irq)) + continue; #endif - add_pin_to_irq(irq, apic, pin); + add_pin_to_irq(irq, apic, pin); - setup_IO_APIC_irq(apic, pin, irq, - irq_trigger(idx), irq_polarity(idx)); - } + setup_IO_APIC_irq(apic, pin, irq, + irq_trigger(idx), irq_polarity(idx)); + } + if (notcon) { + apic_printk(APIC_VERBOSE, + KERN_DEBUG " (apicid-pin) not connected\n"); + notcon = 0; + } } - if (!first_notcon) - apic_printk(APIC_VERBOSE, " not connected.\n"); + if (notcon) + apic_printk(APIC_VERBOSE, + KERN_DEBUG " (apicid-pin) not connected\n"); } /* -- cgit v1.2.3 From 79c09698cac8df16c5539447d5e1047fde9742e5 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 7 Sep 2008 17:58:57 -0700 Subject: x86: lapic address print out like io apic addr Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 0556f375e40b..f3f50858048e 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1548,7 +1548,7 @@ void __init init_apic_mappings(void) apic_phys = mp_lapic_addr; set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - apic_printk(APIC_VERBOSE, "mapped APIC to %16lx (%16lx)\n", + apic_printk(APIC_VERBOSE, "mapped APIC to %08lx (%08lx)\n", APIC_BASE, apic_phys); /* -- cgit v1.2.3 From 2a554fb132cf804477087057b9b0ff2162984507 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Mon, 8 Sep 2008 19:38:06 +0400 Subject: x86: io-apic - do not use KERN_DEBUG marker too much Do not use KERN_DEBUG several times on the same line being printed. Introduced by mine previous patch, sorry. Reported-by: Yinghai Lu Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 4040d575a21e..12e59df026db 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1528,11 +1528,16 @@ static void __init setup_IO_APIC_irqs(void) idx = find_irq_entry(apic, pin, mp_INT); if (idx == -1) { - apic_printk(APIC_VERBOSE, - KERN_DEBUG " %d-%d", - mp_ioapics[apic].mp_apicid, pin); - if (!notcon) + if (!notcon) { notcon = 1; + apic_printk(APIC_VERBOSE, + KERN_DEBUG " %d-%d", + mp_ioapics[apic].mp_apicid, + pin); + } else + apic_printk(APIC_VERBOSE, " %d-%d", + mp_ioapics[apic].mp_apicid, + pin); continue; } @@ -1548,14 +1553,14 @@ static void __init setup_IO_APIC_irqs(void) } if (notcon) { apic_printk(APIC_VERBOSE, - KERN_DEBUG " (apicid-pin) not connected\n"); + " (apicid-pin) not connected\n"); notcon = 0; } } if (notcon) apic_printk(APIC_VERBOSE, - KERN_DEBUG " (apicid-pin) not connected\n"); + " (apicid-pin) not connected\n"); } /* -- cgit v1.2.3 From f0ed4e695faf6766927c8cfbda2bc7530c7210c2 Mon Sep 17 00:00:00 2001 From: Venki Pallipadi Date: Mon, 8 Sep 2008 10:18:40 -0700 Subject: x86: using HPET in MSI mode and setting up per CPU HPET timers, fix On Sat, Sep 06, 2008 at 06:03:53AM -0700, Ingo Molnar wrote: > > it crashes two testsystems, the fault on a NULL pointer in hpet init, > with: > > initcall print_all_ICs+0x0/0x520 returned 0 after 26 msecs > calling hpet_late_init+0x0/0x1c0 > BUG: unable to handle kernel NULL pointer dereference at 000000000000008c > IP: [] hpet_late_init+0xfe/0x1c0 > PGD 0 > Oops: 0000 [1] SMP > CPU 0 > Modules linked in: > Pid: 1, comm: swapper Not tainted 2.6.27-rc5 #29725 > RIP: 0010:[] [] hpet_late_init+0xfe/0x1c0 > RSP: 0018:ffff88003fa07dd0 EFLAGS: 00010246 > RAX: 0000000000000000 RBX: 0000000000000003 RCX: 0000000000000000 > RDX: ffffc20000000160 RSI: 0000000000000000 RDI: 0000000000000003 > RBP: ffff88003fa07e90 R08: 0000000000000000 R09: ffff88003fa07dd0 > R10: 0000000000000001 R11: 0000000000000000 R12: ffff88003fa07dd0 > R13: 0000000000000002 R14: ffffc20000000000 R15: 000000006f57e511 > FS: 0000000000000000(0000) GS:ffffffff80cf6a80(0000) knlGS:0000000000000000 > CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b > CR2: 000000000000008c CR3: 0000000000201000 CR4: 00000000000006e0 > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 > DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 > Process swapper (pid: 1, threadinfo ffff88003fa06000, task ffff88003fa08000) > Stack: 00000000fed00000 ffffc20000000000 0000000100000003 0000000800000002 > 0000000000000000 0000000000000000 0000000000000000 0000000000000000 > 0000000000000000 0000000000000000 0000000000000000 0000000000000000 > Call Trace: > [] ? hpet_late_init+0x0/0x1c0 > [] do_one_initcall+0x45/0x190 > [] ? register_irq_proc+0x19/0xe0 > [] ? early_idt_handler+0x0/0x73 > [] kernel_init+0x14c/0x1b0 > [] ? trace_hardirqs_on_thunk+0x3a/0x3f > [] child_rip+0xa/0x11 > [] ? restore_args+0x0/0x30 > [] ? kernel_init+0x0/0x1b0 > [] ? child_rip+0x0/0x11 > Code: 20 48 83 c1 01 48 39 f1 75 e3 44 89 e8 4c 8b 05 29 29 22 00 31 f6 48 8d 78 01 66 66 90 89 f0 48 8d 04 80 48 c1 e0 05 4a 8d 0c 00 81 8c 00 00 00 08 74 26 8b 81 80 00 00 00 8b 91 88 00 00 00 > RIP [] hpet_late_init+0xfe/0x1c0 > RSP > CR2: 000000000000008c > Kernel panic - not syncing: Fatal exception There was one code path, with CONFIG_PCI_MSI disabled, where we were accessing hpet_devs without initialization. That resulted in the above crash. The change below adds a check for hpet_devs. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Shaohua Li Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 28 +++++++++++++++++++--------- 1 file changed, 19 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 31e9191b7e19..01005aeda7d9 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -126,6 +126,24 @@ EXPORT_SYMBOL_GPL(is_hpet_enabled); * timer 0 and timer 1 in case of RTC emulation. */ #ifdef CONFIG_HPET +static void hpet_reserve_msi_timers(struct hpet_data *hd) +{ + int i; + + if (!hpet_devs) + return; + + for (i = 0; i < hpet_num_timers; i++) { + struct hpet_dev *hdev = &hpet_devs[i]; + + if (!(hdev->flags & HPET_DEV_VALID)) + continue; + + hd->hd_irq[hdev->num] = hdev->irq; + hpet_reserve_timer(hd, hdev->num); + } +} + static void hpet_reserve_platform_timers(unsigned long id) { struct hpet __iomem *hpet = hpet_virt_address; @@ -158,15 +176,7 @@ static void hpet_reserve_platform_timers(unsigned long id) Tn_INT_ROUTE_CNF_MASK) >> Tn_INT_ROUTE_CNF_SHIFT; } - for (i = 0; i < nrtimers; i++) { - struct hpet_dev *hdev = &hpet_devs[i]; - - if (!(hdev->flags & HPET_DEV_VALID)) - continue; - - hd.hd_irq[hdev->num] = hdev->irq; - hpet_reserve_timer(&hd, hdev->num); - } + hpet_reserve_msi_timers(&hd); hpet_alloc(&hd); -- cgit v1.2.3 From ba374c9baef910fbc5373541d98c50f15e82c3f8 Mon Sep 17 00:00:00 2001 From: Steven Noonan Date: Mon, 8 Sep 2008 16:19:09 -0700 Subject: x86: fix HPET compiler error when not using CONFIG_PCI_MSI Added dummy function for hpet_setup_msi_irq(). Signed-off-by: Steven Noonan Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 01005aeda7d9..422c577ef691 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -635,6 +635,10 @@ static int hpet_cpuhp_notify(struct notifier_block *n, } #else +static int hpet_setup_msi_irq(unsigned int irq) +{ + return 0; +} void hpet_msi_capability_lookup(unsigned int start_timer) { return; -- cgit v1.2.3 From 87783be4c26d79f5cde245e6e8a9fd2b295e92d7 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Wed, 10 Sep 2008 22:19:50 +0400 Subject: x86: io-apic - get rid of __DO_ACTION macro Replace __DO_ACTION macro with io_apic_modify_irq function. This allow us to 'grep' definitions being hided by __DO_ACTION macro: __unmask_IO_APIC_irq __mask_IO_APIC_irq __mask_and_edge_IO_APIC_irq __unmask_and_level_IO_APIC_irq Signed-off-by: Cyrill Gorcunov Acked-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 103 +++++++++++++++++++++++----------------------- 1 file changed, 52 insertions(+), 51 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 12e59df026db..ac47384724e3 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -643,65 +643,66 @@ static void __init replace_pin_at_irq(unsigned int irq, add_pin_to_irq(irq, newapic, newpin); } -#define __DO_ACTION(R, ACTION_ENABLE, ACTION_DISABLE, FINAL) \ - \ -{ \ - int pin; \ - struct irq_cfg *cfg; \ - struct irq_pin_list *entry; \ - \ - cfg = irq_cfg(irq); \ - entry = cfg->irq_2_pin; \ - for (;;) { \ - unsigned int reg; \ - if (!entry) \ - break; \ - pin = entry->pin; \ - reg = io_apic_read(entry->apic, 0x10 + R + pin*2); \ - reg ACTION_DISABLE; \ - reg ACTION_ENABLE; \ - io_apic_modify(entry->apic, 0x10 + R + pin*2, reg); \ - FINAL; \ - if (!entry->next) \ - break; \ - entry = entry->next; \ - } \ -} - -#define DO_ACTION(name,R, ACTION_ENABLE, ACTION_DISABLE, FINAL) \ - \ - static void name##_IO_APIC_irq (unsigned int irq) \ - __DO_ACTION(R, ACTION_ENABLE, ACTION_DISABLE, FINAL) - -/* mask = 0 */ -DO_ACTION(__unmask, 0, |= 0, &= ~IO_APIC_REDIR_MASKED, ) +static inline void io_apic_modify_irq(unsigned int irq, + int mask_and, int mask_or, + void (*final)(struct irq_pin_list *entry)) +{ + int pin; + struct irq_cfg *cfg; + struct irq_pin_list *entry; + + cfg = irq_cfg(irq); + for (entry = cfg->irq_2_pin; entry != NULL; entry = entry->next) { + unsigned int reg; + pin = entry->pin; + reg = io_apic_read(entry->apic, 0x10 + pin * 2); + reg &= mask_and; + reg |= mask_or; + io_apic_modify(entry->apic, 0x10 + pin * 2, reg); + if (final) + final(entry); + } +} + +static void __unmask_IO_APIC_irq(unsigned int irq) +{ + io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, 0, NULL); +} #ifdef CONFIG_X86_64 -/* - * Synchronize the IO-APIC and the CPU by doing - * a dummy read from the IO-APIC - */ -static inline void io_apic_sync(unsigned int apic) +void io_apic_sync(struct irq_pin_list *entry) { - struct io_apic __iomem *io_apic = io_apic_base(apic); + /* + * Synchronize the IO-APIC and the CPU by doing + * a dummy read from the IO-APIC + */ + struct io_apic __iomem *io_apic; + io_apic = io_apic_base(entry->apic); readl(&io_apic->data); } -/* mask = 1 */ -DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, &= ~0, io_apic_sync(entry->apic)) - -#else - -/* mask = 1 */ -DO_ACTION(__mask, 0, |= IO_APIC_REDIR_MASKED, &= ~0, ) - -/* mask = 1, trigger = 0 */ -DO_ACTION(__mask_and_edge, 0, |= IO_APIC_REDIR_MASKED, &= ~IO_APIC_REDIR_LEVEL_TRIGGER, ) +static void __mask_IO_APIC_irq(unsigned int irq) +{ + io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, &io_apic_sync); +} +#else /* CONFIG_X86_32 */ +static void __mask_IO_APIC_irq(unsigned int irq) +{ + io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, NULL); +} -/* mask = 0, trigger = 1 */ -DO_ACTION(__unmask_and_level, 0, |= IO_APIC_REDIR_LEVEL_TRIGGER, &= ~IO_APIC_REDIR_MASKED, ) +static void __mask_and_edge_IO_APIC_irq(unsigned int irq) +{ + io_apic_modify_irq(irq, ~IO_APIC_REDIR_LEVEL_TRIGGER, + IO_APIC_REDIR_MASKED, NULL); +} -#endif +static void __unmask_and_level_IO_APIC_irq(unsigned int irq) +{ + io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, + IO_APIC_REDIR_LEVEL_TRIGGER, NULL); +} +#endif /* CONFIG_X86_32 */ static void mask_IO_APIC_irq (unsigned int irq) { -- cgit v1.2.3 From 823b259b80158a5fb694f6784e18b5bae669c599 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 10 Sep 2008 21:56:46 -0700 Subject: x86: print out apic id in hex format Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 2 +- arch/x86/kernel/cpu/amd.c | 2 +- arch/x86/kernel/cpu/intel.c | 2 +- arch/x86/kernel/smpboot.c | 6 +++--- 4 files changed, 6 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index f3f50858048e..11cb18639581 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1586,7 +1586,7 @@ int __init APIC_init_uniprocessor(void) */ if (!cpu_has_apic && APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { - printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", + printk(KERN_ERR "BIOS bug, local APIC 0x%x not detected!...\n", boot_cpu_physical_apicid); clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); return -1; diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 32e73520adf7..8f1e31db2ad5 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -249,7 +249,7 @@ static void __cpuinit srat_detect_node(struct cpuinfo_x86 *c) } numa_set_node(cpu, node); - printk(KERN_INFO "CPU %d/%x -> Node %d\n", cpu, apicid, node); + printk(KERN_INFO "CPU %d/0x%x -> Node %d\n", cpu, apicid, node); #endif } diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 99468dbd08da..cce0b6118d55 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -174,7 +174,7 @@ static void __cpuinit srat_detect_node(void) node = first_node(node_online_map); numa_set_node(cpu, node); - printk(KERN_INFO "CPU %d/%x -> Node %d\n", cpu, apicid, node); + printk(KERN_INFO "CPU %d/0x%x -> Node %d\n", cpu, apicid, node); #endif } diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 8c3aca7cb343..f84de2ff933c 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -543,10 +543,10 @@ static inline void __inquire_remote_apic(int apicid) int timeout; u32 status; - printk(KERN_INFO "Inquiring remote APIC #%d...\n", apicid); + printk(KERN_INFO "Inquiring remote APIC 0x%x...\n", apicid); for (i = 0; i < ARRAY_SIZE(regs); i++) { - printk(KERN_INFO "... APIC #%d %s: ", apicid, names[i]); + printk(KERN_INFO "... APIC 0x%x %s: ", apicid, names[i]); /* * Wait for idle. @@ -874,7 +874,7 @@ do_rest: start_ip = setup_trampoline(); /* So we see what's up */ - printk(KERN_INFO "Booting processor %d/%d ip %lx\n", + printk(KERN_INFO "Booting processor %d APIC 0x%x ip 0x%lx\n", cpu, apicid, start_ip); /* -- cgit v1.2.3 From 9df08f109572e88fbdab9a61d5cb0f0eae4ac9fa Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 14 Sep 2008 11:55:37 +0400 Subject: x86: apic - lapic_setup_esr does not handle esr_disable - fix it lapic_setup_esr doesn't handle esr_disable inquire. The error brought in during unification process. Fix it. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 63 ++++++++++++++++++++++++++------------------------ 1 file changed, 33 insertions(+), 30 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 11cb18639581..59550cc017dc 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1081,40 +1081,43 @@ void __init init_bsp_APIC(void) static void __cpuinit lapic_setup_esr(void) { - unsigned long oldvalue, value, maxlvt; - if (lapic_is_integrated() && !esr_disable) { - if (esr_disable) { - /* - * Something untraceable is creating bad interrupts on - * secondary quads ... for the moment, just leave the - * ESR disabled - we can't do anything useful with the - * errors anyway - mbligh - */ - printk(KERN_INFO "Leaving ESR disabled.\n"); - return; - } - /* !82489DX */ - maxlvt = lapic_get_maxlvt(); - if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ - apic_write(APIC_ESR, 0); - oldvalue = apic_read(APIC_ESR); + unsigned int oldvalue, value, maxlvt; + + if (!lapic_is_integrated()) { + printk(KERN_INFO "No ESR for 82489DX.\n"); + return; + } - /* enables sending errors */ - value = ERROR_APIC_VECTOR; - apic_write(APIC_LVTERR, value); + if (esr_disable) { /* - * spec says clear errors after enabling vector. + * Something untraceable is creating bad interrupts on + * secondary quads ... for the moment, just leave the + * ESR disabled - we can't do anything useful with the + * errors anyway - mbligh */ - if (maxlvt > 3) - apic_write(APIC_ESR, 0); - value = apic_read(APIC_ESR); - if (value != oldvalue) - apic_printk(APIC_VERBOSE, "ESR value before enabling " - "vector: 0x%08lx after: 0x%08lx\n", - oldvalue, value); - } else { - printk(KERN_INFO "No ESR for 82489DX.\n"); + printk(KERN_INFO "Leaving ESR disabled.\n"); + return; } + + maxlvt = lapic_get_maxlvt(); + if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ + apic_write(APIC_ESR, 0); + oldvalue = apic_read(APIC_ESR); + + /* enables sending errors */ + value = ERROR_APIC_VECTOR; + apic_write(APIC_LVTERR, value); + + /* + * spec says clear errors after enabling vector. + */ + if (maxlvt > 3) + apic_write(APIC_ESR, 0); + value = apic_read(APIC_ESR); + if (value != oldvalue) + apic_printk(APIC_VERBOSE, "ESR value before enabling " + "vector: 0x%08x after: 0x%08x\n", + oldvalue, value); } -- cgit v1.2.3 From 08ad776e3c54e6fe8b50939f176538063b69f89e Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 14 Sep 2008 11:55:38 +0400 Subject: x86: apic - skip writting ESR register if we dont have on On 82489DX we don't have ESR register so we should not write it. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 59550cc017dc..d730e8d31727 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1131,7 +1131,7 @@ void __cpuinit setup_local_APIC(void) #ifdef CONFIG_X86_32 /* Pound the ESR really hard over the head with a big hammer - mbligh */ - if (esr_disable) { + if (lapic_is_integrated() && esr_disable) { apic_write(APIC_ESR, 0); apic_write(APIC_ESR, 0); apic_write(APIC_ESR, 0); -- cgit v1.2.3 From b189892de4da4634560657aedd774752094dbfa0 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Fri, 12 Sep 2008 23:58:24 +0400 Subject: x86: apic - fix unused vars warning in calibrate_APIC_clock If we don't have CONFIG_X86_PM_TIMER=y compiler warns about unused variables. Move PM timer based calibration into a separate function and make the code cleaner and the compiler happy as well. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 81 +++++++++++++++++++++++++++++--------------------- 1 file changed, 47 insertions(+), 34 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index d730e8d31727..784116933c70 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -538,14 +538,51 @@ static void __init lapic_cal_handler(struct clock_event_device *dev) } } +static int __init calibrate_by_pmtimer(long deltapm, long *delta) +{ + const long pm_100ms = PMTMR_TICKS_PER_SEC / 10; + const long pm_thresh = pm_100ms / 100; + unsigned long mult; + u64 res; + +#ifndef CONFIG_X86_PM_TIMER + return -1; +#endif + + apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); + + /* Check, if the PM timer is available */ + if (!deltapm) + return -1; + + mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); + + if (deltapm > (pm_100ms - pm_thresh) && + deltapm < (pm_100ms + pm_thresh)) { + apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); + } else { + res = (((u64)deltapm) * mult) >> 22; + do_div(res, 1000000); + printk(KERN_WARNING "APIC calibration not consistent " + "with PM Timer: %ldms instead of 100ms\n", + (long)res); + /* Correct the lapic counter value */ + res = (((u64)(*delta)) * pm_100ms); + do_div(res, deltapm); + printk(KERN_INFO "APIC delta adjusted to PM-Timer: " + "%lu (%ld)\n", (unsigned long)res, *delta); + *delta = (long)res; + } + + return 0; +} + static int __init calibrate_APIC_clock(void) { struct clock_event_device *levt = &__get_cpu_var(lapic_events); - const long pm_100ms = PMTMR_TICKS_PER_SEC/10; - const long pm_thresh = pm_100ms/100; void (*real_handler)(struct clock_event_device *dev); unsigned long deltaj; - long delta, deltapm; + long delta; int pm_referenced = 0; local_irq_disable(); @@ -575,36 +612,9 @@ static int __init calibrate_APIC_clock(void) delta = lapic_cal_t1 - lapic_cal_t2; apic_printk(APIC_VERBOSE, "... lapic delta = %ld\n", delta); -#ifdef CONFIG_X86_PM_TIMER - /* Check, if the PM timer is available */ - deltapm = lapic_cal_pm2 - lapic_cal_pm1; - apic_printk(APIC_VERBOSE, "... PM timer delta = %ld\n", deltapm); - - if (deltapm) { - unsigned long mult; - u64 res; - - mult = clocksource_hz2mult(PMTMR_TICKS_PER_SEC, 22); - - if (deltapm > (pm_100ms - pm_thresh) && - deltapm < (pm_100ms + pm_thresh)) { - apic_printk(APIC_VERBOSE, "... PM timer result ok\n"); - } else { - res = (((u64) deltapm) * mult) >> 22; - do_div(res, 1000000); - printk(KERN_WARNING "APIC calibration not consistent " - "with PM Timer: %ldms instead of 100ms\n", - (long)res); - /* Correct the lapic counter value */ - res = (((u64) delta) * pm_100ms); - do_div(res, deltapm); - printk(KERN_INFO "APIC delta adjusted to PM-Timer: " - "%lu (%ld)\n", (unsigned long) res, delta); - delta = (long) res; - } - pm_referenced = 1; - } -#endif + /* we trust the PM based calibration if possible */ + pm_referenced = !calibrate_by_pmtimer(lapic_cal_pm2 - lapic_cal_pm1, + &delta); /* Calculate the scaled math multiplication factor */ lapic_clockevent.mult = div_sc(delta, TICK_NSEC * LAPIC_CAL_LOOPS, @@ -646,7 +656,10 @@ static int __init calibrate_APIC_clock(void) levt->features &= ~CLOCK_EVT_FEAT_DUMMY; - /* We trust the pm timer based calibration */ + /* + * PM timer calibration failed or not turned on + * so lets try APIC timer based calibration + */ if (!pm_referenced) { apic_printk(APIC_VERBOSE, "... verify APIC timer\n"); -- cgit v1.2.3 From 56ffa1a028b9fce3860a247c6fe79fce7cbf425b Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sat, 13 Sep 2008 13:11:16 +0400 Subject: x86: io-apic - do not use KERN_DEBUG marker too much, fix Yinghai Lu reported: | > 0 add_pin_to_irq: irq 15 --> apic 0 pin 15 | > IOAPIC[0]: Set routing entry (8-15 -> 0x3f -> IRQ 15 Mode:0 Active:0) | > 8-16 8-17 8-18 8-19 8-20 8-21 8-22 8-23 (apicid-pin) not connected | > 9-0 9-1 9-2 9-3 9-4 9-5 9-6 9-7 9-8 9-9 9-10 9-11 9-12 9-13 9-14 9-15 | > 9-16 9-17 9-18 9-19 9-20 9-21 9-22 9-23 (apicid-pin) not connected | > | | only first one not connected at first, and ... here is a quick fix for this. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index ac47384724e3..6ce5873a406c 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1541,6 +1541,11 @@ static void __init setup_IO_APIC_irqs(void) pin); continue; } + if (notcon) { + apic_printk(APIC_VERBOSE, + " (apicid-pin) not connected\n"); + notcon = 0; + } irq = pin_2_irq(idx, apic, pin); #ifdef CONFIG_X86_32 @@ -1552,11 +1557,6 @@ static void __init setup_IO_APIC_irqs(void) setup_IO_APIC_irq(apic, pin, irq, irq_trigger(idx), irq_polarity(idx)); } - if (notcon) { - apic_printk(APIC_VERBOSE, - " (apicid-pin) not connected\n"); - notcon = 0; - } } if (notcon) -- cgit v1.2.3 From 9d98598d2fc286c8dbcd0b681168639528428db0 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 19 Sep 2008 00:12:26 -0700 Subject: sparseirq: remove some debug print out Signed-off-by: Yinghai Lu Cc: Yinghai Lu Cc: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 19 ------------------- 1 file changed, 19 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 6ce5873a406c..01419fdc1d5d 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -277,23 +277,6 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) spin_unlock_irqrestore(&irq_cfg_lock, flags); - printk(KERN_DEBUG "found new irq_cfg for irq %d\n", cfg->irq); -#ifdef CONFIG_HAVE_SPARSE_IRQ_DEBUG - { - /* dump the results */ - struct irq_cfg *cfg; - unsigned long phys; - unsigned long bytes = sizeof(struct irq_cfg); - - printk(KERN_DEBUG "=========================== %d\n", irq); - printk(KERN_DEBUG "irq_cfg dump after get that for %d\n", irq); - for_each_irq_cfg(cfg) { - phys = __pa(cfg); - printk(KERN_DEBUG "irq_cfg %d ==> [%#lx - %#lx]\n", cfg->irq, phys, phys + bytes); - } - printk(KERN_DEBUG "===========================\n"); - } -#endif return cfg; } #else @@ -597,7 +580,6 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) cfg->irq_2_pin = entry; entry->apic = apic; entry->pin = pin; - printk(KERN_DEBUG " 0 add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); return; } @@ -613,7 +595,6 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) entry = entry->next; entry->apic = apic; entry->pin = pin; - printk(KERN_DEBUG " x add_pin_to_irq: irq %d --> apic %d pin %d\n", irq, apic, pin); } /* -- cgit v1.2.3 From 5ffa4eb2224343ec3fbd492ffe71e28e797435b7 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 18 Sep 2008 23:37:57 +0400 Subject: x86: io-apic - interrupt remapping fix Interrupt remapping could lead to NULL dereference in case of kzalloc failed and memory leak in other way. So fix the both cases. Signed-off-by: Cyrill Gorcunov Cc: "Maciej W. Rozycki" Cc: Suresh Siddha Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 13 ++++++++++--- arch/x86/kernel/io_apic.c | 19 +++++++++++++++++-- 2 files changed, 27 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 784116933c70..1f48bd1c9f2d 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1360,7 +1360,12 @@ void enable_IR_x2apic(void) local_irq_save(flags); mask_8259A(); - save_mask_IO_APIC_setup(); + + ret = save_mask_IO_APIC_setup(); + if (ret) { + printk(KERN_INFO "Saving IO-APIC state failed: %d\n", ret); + goto end; + } ret = enable_intr_remapping(1); @@ -1370,14 +1375,15 @@ void enable_IR_x2apic(void) } if (ret) - goto end; + goto end_restore; if (!x2apic) { x2apic = 1; apic_ops = &x2apic_ops; enable_x2apic(); } -end: + +end_restore: if (ret) /* * IR enabling failed @@ -1386,6 +1392,7 @@ end: else reinit_intr_remapped_IO_APIC(x2apic_preenabled); +end: unmask_8259A(); local_irq_restore(flags); diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 01419fdc1d5d..4cc9cb64c811 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -812,7 +812,7 @@ int save_mask_IO_APIC_setup(void) kzalloc(sizeof(struct IO_APIC_route_entry) * nr_ioapic_registers[apic], GFP_KERNEL); if (!early_ioapic_entries[apic]) - return -ENOMEM; + goto nomem; } for (apic = 0; apic < nr_ioapics; apic++) @@ -826,17 +826,32 @@ int save_mask_IO_APIC_setup(void) ioapic_write_entry(apic, pin, entry); } } + return 0; + +nomem: + for (; apic > 0; apic--) + kfree(early_ioapic_entries[apic]); + kfree(early_ioapic_entries[apic]); + memset(early_ioapic_entries, 0, + ARRAY_SIZE(early_ioapic_entries)); + + return -ENOMEM; } void restore_IO_APIC_setup(void) { int apic, pin; - for (apic = 0; apic < nr_ioapics; apic++) + for (apic = 0; apic < nr_ioapics; apic++) { + if (!early_ioapic_entries[apic]) + break; for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) ioapic_write_entry(apic, pin, early_ioapic_entries[apic][pin]); + kfree(early_ioapic_entries[apic]); + early_ioapic_entries[apic] = NULL; + } } void reinit_intr_remapped_IO_APIC(int intr_remapping) -- cgit v1.2.3 From 8da077d6f31da291ee3a7dd559671cb8ca48cbe2 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Tue, 23 Sep 2008 08:42:05 -0500 Subject: x86, uv: fix ordering of calls to uv_system_init & uv_cpu_init Fix problem caused by reordering of the calls to uv_cpu_init() & uv_system_init. Originally, uv_cpu_init() was called AFTER uv_system_init. This order was recently broken as a side-effect of other patches. With this patch, initialization of cpu 0 is now done by the system_init call. Signed-off-by: Jack Steiner Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 34 +++++++++++++++++----------------- 1 file changed, 17 insertions(+), 17 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 33581d94a90e..b689075bbe2f 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -356,7 +356,22 @@ static __init void uv_rtc_init(void) sn_rtc_cycles_per_second = ticks_per_sec; } -static bool uv_system_inited; +/* + * Called on each cpu to initialize the per_cpu UV data area. + * ZZZ hotplug not supported yet + */ +void __cpuinit uv_cpu_init(void) +{ + /* CPU 0 initilization will be done via uv_system_init. */ + if (!uv_blade_info) + return; + + uv_blade_info[uv_numa_blade_id()].nr_online_cpus++; + + if (get_uv_system_type() == UV_NON_UNIQUE_APIC) + set_x2apic_extra_bits(uv_hub_info->pnode); +} + void __init uv_system_init(void) { @@ -448,21 +463,6 @@ void __init uv_system_init(void) map_mmr_high(max_pnode); map_config_high(max_pnode); map_mmioh_high(max_pnode); - uv_system_inited = true; -} -/* - * Called on each cpu to initialize the per_cpu UV data area. - * ZZZ hotplug not supported yet - */ -void __cpuinit uv_cpu_init(void) -{ - BUG_ON(!uv_system_inited); - - uv_blade_info[uv_numa_blade_id()].nr_online_cpus++; - - if (get_uv_system_type() == UV_NON_UNIQUE_APIC) - set_x2apic_extra_bits(uv_hub_info->pnode); + uv_cpu_init(); } - - -- cgit v1.2.3 From 7564676813780e0ba4dacf95338202cb72d2172d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 24 Sep 2008 19:04:36 -0700 Subject: x86: irq no should not use hex in /proc/interrupts Arjan van de Ven noticed that we changed IRQ numbers from decimal to hex in /proc/interrupts - that can break user-space utilities like irqbalanced. Reported-by: Arjan van de Ven Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_32.c | 2 +- arch/x86/kernel/irq_64.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index b2e1082cf5ad..001772ffc918 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -307,7 +307,7 @@ int show_interrupts(struct seq_file *p, void *v) action = desc->action; if (!action && !any_count) goto skip; - seq_printf(p, "%#x: ",i); + seq_printf(p, "%3d: ", i); #ifndef CONFIG_SMP seq_printf(p, "%10u ", kstat_irqs(i)); #else diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index c2fca89a3f7e..ec2661091283 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -112,7 +112,7 @@ int show_interrupts(struct seq_file *p, void *v) action = desc->action; if (!action && !any_count) goto skip; - seq_printf(p, "%#x: ",i); + seq_printf(p, "%3d: ", i); #ifndef CONFIG_SMP seq_printf(p, "%10u ", kstat_irqs(i)); #else -- cgit v1.2.3 From c1370b49cc4f09de5b447ddf688a3988a297dbfc Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Tue, 23 Sep 2008 23:00:02 +0400 Subject: x86: io-apic - interrupt remapping fix Clean up obscure for() cycle with straight while() form Signed-off-by: Cyrill Gorcunov Cc: "Maciej W. Rozycki" Acked-by: Suresh Siddha Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 4cc9cb64c811..ed68a6f99dc2 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -830,9 +830,8 @@ int save_mask_IO_APIC_setup(void) return 0; nomem: - for (; apic > 0; apic--) - kfree(early_ioapic_entries[apic]); - kfree(early_ioapic_entries[apic]); + while (apic >= 0) + kfree(early_ioapic_entries[apic--]); memset(early_ioapic_entries, 0, ARRAY_SIZE(early_ioapic_entries)); -- cgit v1.2.3 From c81bba49a13cb3376654d0cc93dc069fd619ed76 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 25 Sep 2008 11:53:11 -0700 Subject: x86: print out irq nr for msi/ht, v3 v2: fix hpet compiling error v3: Bjorn want to use dev_printk instead Signed-off-by: Yinghai Lu Cc: Bjorn Helgaas Cc: Jesse Barnes Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 3 +++ arch/x86/kernel/io_apic.c | 5 +++++ 2 files changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 422c577ef691..2913913f4a46 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -467,6 +467,9 @@ static int hpet_setup_irq(struct hpet_dev *dev) irq_set_affinity(dev->irq, cpumask_of_cpu(dev->cpu)); enable_irq(dev->irq); + printk(KERN_DEBUG "hpet: %s irq %d for MSI\n", + dev->name, dev->irq); + return 0; } diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index ed68a6f99dc2..4ee270d30358 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3354,6 +3354,8 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) #endif set_irq_chip_and_handler_name(irq, &msi_chip, handle_edge_irq, "edge"); + dev_printk(KERN_DEBUG, &dev->dev, "irq %d for MSI/MSI-X\n", irq); + return 0; } @@ -3586,6 +3588,7 @@ int arch_setup_hpet_msi(unsigned int irq) hpet_msi_write(irq, &msg); set_irq_chip_and_handler_name(irq, &hpet_msi_type, handle_edge_irq, "edge"); + return 0; } #endif @@ -3682,6 +3685,8 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) set_irq_chip_and_handler_name(irq, &ht_irq_chip, handle_edge_irq, "edge"); + + dev_printk(KERN_DEBUG, &dev->dev, "irq %d for HT\n", irq); } return err; } -- cgit v1.2.3 From 5f79f2f2ad39b5177c52ed08ffd066ea0c1da924 Mon Sep 17 00:00:00 2001 From: Venki Pallipadi Date: Wed, 24 Sep 2008 10:03:17 -0700 Subject: hpet: clean up warning Fix the below compile warnings due to recent HPET MSI changes arch/x86/kernel/hpet.c:48: warning: 'hpet_devs' defined but not used arch/x86/kernel/hpet.c:50: warning: 'per_cpu__cpu_hpet_dev' defined but not used Reported-by: Andrew Morton Signed-off-by: Venkatesh Pallipadi Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 57 +++++++++++++++++++++++++++++++------------------- 1 file changed, 35 insertions(+), 22 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 2913913f4a46..77017e834cf7 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -45,10 +45,6 @@ struct hpet_dev { char name[10]; }; -static struct hpet_dev *hpet_devs; - -static DEFINE_PER_CPU(struct hpet_dev *, cpu_hpet_dev); - unsigned long hpet_readl(unsigned long a) { return readl(hpet_virt_address + a); @@ -126,23 +122,8 @@ EXPORT_SYMBOL_GPL(is_hpet_enabled); * timer 0 and timer 1 in case of RTC emulation. */ #ifdef CONFIG_HPET -static void hpet_reserve_msi_timers(struct hpet_data *hd) -{ - int i; - if (!hpet_devs) - return; - - for (i = 0; i < hpet_num_timers; i++) { - struct hpet_dev *hdev = &hpet_devs[i]; - - if (!(hdev->flags & HPET_DEV_VALID)) - continue; - - hd->hd_irq[hdev->num] = hdev->irq; - hpet_reserve_timer(hd, hdev->num); - } -} +static void hpet_reserve_msi_timers(struct hpet_data *hd); static void hpet_reserve_platform_timers(unsigned long id) { @@ -362,6 +343,10 @@ static int hpet_legacy_next_event(unsigned long delta, * HPET MSI Support */ #ifdef CONFIG_PCI_MSI + +static DEFINE_PER_CPU(struct hpet_dev *, cpu_hpet_dev); +static struct hpet_dev *hpet_devs; + void hpet_msi_unmask(unsigned int irq) { struct hpet_dev *hdev = get_irq_data(irq); @@ -525,7 +510,8 @@ static void init_one_hpet_msi_clockevent(struct hpet_dev *hdev, int cpu) #else #define RESERVE_TIMERS 0 #endif -void hpet_msi_capability_lookup(unsigned int start_timer) + +static void hpet_msi_capability_lookup(unsigned int start_timer) { unsigned int id; unsigned int num_timers; @@ -571,6 +557,26 @@ void hpet_msi_capability_lookup(unsigned int start_timer) num_timers, num_timers_used); } +#ifdef CONFIG_HPET +static void hpet_reserve_msi_timers(struct hpet_data *hd) +{ + int i; + + if (!hpet_devs) + return; + + for (i = 0; i < hpet_num_timers; i++) { + struct hpet_dev *hdev = &hpet_devs[i]; + + if (!(hdev->flags & HPET_DEV_VALID)) + continue; + + hd->hd_irq[hdev->num] = hdev->irq; + hpet_reserve_timer(hd, hdev->num); + } +} +#endif + static struct hpet_dev *hpet_get_unused_timer(void) { int i; @@ -642,10 +648,17 @@ static int hpet_setup_msi_irq(unsigned int irq) { return 0; } -void hpet_msi_capability_lookup(unsigned int start_timer) +static void hpet_msi_capability_lookup(unsigned int start_timer) +{ + return; +} + +#ifdef CONFIG_HPET +static void hpet_reserve_msi_timers(struct hpet_data *hd) { return; } +#endif static int hpet_cpuhp_notify(struct notifier_block *n, unsigned long action, void *hcpu) -- cgit v1.2.3 From 4173a0e7371ece227559b44943c6fd456ee470d1 Mon Sep 17 00:00:00 2001 From: Dean Nelson Date: Thu, 2 Oct 2008 12:18:21 -0500 Subject: x86, UV: add uv_setup_irq() and uv_teardown_irq() functions, v3 Provide a means for UV interrupt MMRs to be setup with the message to be sent when an MSI is raised. Signed-off-by: Dean Nelson Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/io_apic.c | 68 +++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/uv_irq.c | 77 +++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 146 insertions(+), 1 deletion(-) create mode 100644 arch/x86/kernel/uv_irq.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 45cecc59d894..acc0e8bf043e 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -108,7 +108,7 @@ obj-$(CONFIG_MICROCODE) += microcode.o # 64 bit specific files ifeq ($(CONFIG_X86_64),y) obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o tlb_uv.o - obj-y += bios_uv.o + obj-y += bios_uv.o uv_irq.o obj-y += genx2apic_cluster.o obj-y += genx2apic_phys.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 4ee270d30358..260c95a5e6dc 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -58,6 +58,8 @@ #include #include #include +#include +#include #include #include @@ -3692,6 +3694,72 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) } #endif /* CONFIG_HT_IRQ */ +#ifdef CONFIG_X86_64 +/* + * Re-target the irq to the specified CPU and enable the specified MMR located + * on the specified blade to allow the sending of MSIs to the specified CPU. + */ +int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, + unsigned long mmr_offset) +{ + const cpumask_t *eligible_cpu = get_cpu_mask(cpu); + struct irq_cfg *cfg; + int mmr_pnode; + unsigned long mmr_value; + struct uv_IO_APIC_route_entry *entry; + unsigned long flags; + int err; + + err = assign_irq_vector(irq, *eligible_cpu); + if (err != 0) + return err; + + spin_lock_irqsave(&vector_lock, flags); + set_irq_chip_and_handler_name(irq, &uv_irq_chip, handle_percpu_irq, + irq_name); + spin_unlock_irqrestore(&vector_lock, flags); + + cfg = irq_cfg(irq); + + mmr_value = 0; + entry = (struct uv_IO_APIC_route_entry *)&mmr_value; + BUG_ON(sizeof(struct uv_IO_APIC_route_entry) != sizeof(unsigned long)); + + entry->vector = cfg->vector; + entry->delivery_mode = INT_DELIVERY_MODE; + entry->dest_mode = INT_DEST_MODE; + entry->polarity = 0; + entry->trigger = 0; + entry->mask = 0; + entry->dest = cpu_mask_to_apicid(*eligible_cpu); + + mmr_pnode = uv_blade_to_pnode(mmr_blade); + uv_write_global_mmr64(mmr_pnode, mmr_offset, mmr_value); + + return irq; +} + +/* + * Disable the specified MMR located on the specified blade so that MSIs are + * longer allowed to be sent. + */ +void arch_disable_uv_irq(int mmr_blade, unsigned long mmr_offset) +{ + unsigned long mmr_value; + struct uv_IO_APIC_route_entry *entry; + int mmr_pnode; + + mmr_value = 0; + entry = (struct uv_IO_APIC_route_entry *)&mmr_value; + BUG_ON(sizeof(struct uv_IO_APIC_route_entry) != sizeof(unsigned long)); + + entry->mask = 1; + + mmr_pnode = uv_blade_to_pnode(mmr_blade); + uv_write_global_mmr64(mmr_pnode, mmr_offset, mmr_value); +} +#endif /* CONFIG_X86_64 */ + int __init io_apic_get_redir_entries (int ioapic) { union IO_APIC_reg_01 reg_01; diff --git a/arch/x86/kernel/uv_irq.c b/arch/x86/kernel/uv_irq.c new file mode 100644 index 000000000000..6bd26c91c30b --- /dev/null +++ b/arch/x86/kernel/uv_irq.c @@ -0,0 +1,77 @@ +/* + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * SGI UV IRQ functions + * + * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved. + */ + +#include +#include +#include + +static void uv_noop(unsigned int irq) +{ +} + +static unsigned int uv_noop_ret(unsigned int irq) +{ + return 0; +} + +static void uv_ack_apic(unsigned int irq) +{ + ack_APIC_irq(); +} + +struct irq_chip uv_irq_chip = { + .name = "UV-CORE", + .startup = uv_noop_ret, + .shutdown = uv_noop, + .enable = uv_noop, + .disable = uv_noop, + .ack = uv_noop, + .mask = uv_noop, + .unmask = uv_noop, + .eoi = uv_ack_apic, + .end = uv_noop, +}; + +/* + * Set up a mapping of an available irq and vector, and enable the specified + * MMR that defines the MSI that is to be sent to the specified CPU when an + * interrupt is raised. + */ +int uv_setup_irq(char *irq_name, int cpu, int mmr_blade, + unsigned long mmr_offset) +{ + int irq; + int ret; + + irq = create_irq(); + if (irq <= 0) + return -EBUSY; + + ret = arch_enable_uv_irq(irq_name, irq, cpu, mmr_blade, mmr_offset); + if (ret != irq) + destroy_irq(irq); + + return ret; +} +EXPORT_SYMBOL_GPL(uv_setup_irq); + +/* + * Tear down a mapping of an irq and vector, and disable the specified MMR that + * defined the MSI that was to be sent to the specified CPU when an interrupt + * was raised. + * + * Set mmr_blade and mmr_offset to what was passed in on uv_setup_irq(). + */ +void uv_teardown_irq(unsigned int irq, int mmr_blade, unsigned long mmr_offset) +{ + arch_disable_uv_irq(mmr_blade, mmr_offset); + destroy_irq(irq); +} +EXPORT_SYMBOL_GPL(uv_teardown_irq); -- cgit v1.2.3 From 37762b6ffb37f9e21cbc6f80902aa06b7a053fd7 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 3 Oct 2008 11:38:37 +0200 Subject: x86, UV: add uv_setup_irq() and uv_teardown_irq() functions, v3, fix fix: arch/x86/kernel/uv_irq.c: In function 'uv_ack_apic': arch/x86/kernel/uv_irq.c:26: error: implicit declaration of function 'ack_APIC_irq' Signed-off-by: Ingo Molnar --- arch/x86/kernel/uv_irq.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/uv_irq.c b/arch/x86/kernel/uv_irq.c index 6bd26c91c30b..aeef529917e4 100644 --- a/arch/x86/kernel/uv_irq.c +++ b/arch/x86/kernel/uv_irq.c @@ -10,6 +10,8 @@ #include #include + +#include #include static void uv_noop(unsigned int irq) -- cgit v1.2.3 From a50f70b17541c0060967c6df61133e968bad3652 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Fri, 3 Oct 2008 11:58:54 -0500 Subject: x86: Add UV EFI table entry v4 Look for a UV entry in the EFI tables. Signed-off-by: Russ Anderson Signed-off-by: Paul Jackson Acked-by: Huang Ying Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/efi.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/efi.c b/arch/x86/kernel/efi.c index 945a31cdd81f..1119d247fe11 100644 --- a/arch/x86/kernel/efi.c +++ b/arch/x86/kernel/efi.c @@ -366,6 +366,10 @@ void __init efi_init(void) SMBIOS_TABLE_GUID)) { efi.smbios = config_tables[i].table; printk(" SMBIOS=0x%lx ", config_tables[i].table); + } else if (!efi_guidcmp(config_tables[i].guid, + UV_SYSTEM_TABLE_GUID)) { + efi.uv_systab = config_tables[i].table; + printk(" UVsystab=0x%lx ", config_tables[i].table); } else if (!efi_guidcmp(config_tables[i].guid, HCDP_TABLE_GUID)) { efi.hcdp = config_tables[i].table; -- cgit v1.2.3 From 7f5942329e0787087a5e4dced838cee711ac2b58 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Fri, 3 Oct 2008 11:59:15 -0500 Subject: x86: Add UV bios call infrastructure v4 Add the EFI callback function and associated wrapper code. Initialize SAL system table entry info at boot time. Signed-off-by: Russ Anderson Signed-off-by: Paul Jackson Acked-by: Huang Ying Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/bios_uv.c | 101 +++++++++++++++++++++++++++++++-------- arch/x86/kernel/genx2apic_uv_x.c | 1 + 2 files changed, 81 insertions(+), 21 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index fdd585f9c53d..5481eb59f783 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -1,8 +1,6 @@ /* * BIOS run time interface routines. * - * Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. - * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or @@ -16,33 +14,94 @@ * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) Russ Anderson */ +#include +#include +#include #include -const char * -x86_bios_strerror(long status) +struct uv_systab uv_systab; + +s64 uv_bios_call(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, u64 a4, u64 a5) { - const char *str; - switch (status) { - case 0: str = "Call completed without error"; break; - case -1: str = "Not implemented"; break; - case -2: str = "Invalid argument"; break; - case -3: str = "Call completed with error"; break; - default: str = "Unknown BIOS status code"; break; - } - return str; + struct uv_systab *tab = &uv_systab; + + if (!tab->function) + /* + * BIOS does not support UV systab + */ + return BIOS_STATUS_UNIMPLEMENTED; + + return efi_call6((void *)__va(tab->function), + (u64)which, a1, a2, a3, a4, a5); } -long -x86_bios_freq_base(unsigned long which, unsigned long *ticks_per_second, - unsigned long *drift_info) +s64 uv_bios_call_irqsave(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, + u64 a4, u64 a5) { - struct uv_bios_retval isrv; + unsigned long bios_flags; + s64 ret; + + local_irq_save(bios_flags); + ret = uv_bios_call(which, a1, a2, a3, a4, a5); + local_irq_restore(bios_flags); + + return ret; +} + +s64 uv_bios_call_reentrant(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, + u64 a4, u64 a5) +{ + s64 ret; + + preempt_disable(); + ret = uv_bios_call(which, a1, a2, a3, a4, a5); + preempt_enable(); - BIOS_CALL(isrv, BIOS_FREQ_BASE, which, 0, 0, 0, 0, 0, 0); - *ticks_per_second = isrv.v0; - *drift_info = isrv.v1; - return isrv.status; + return ret; +} + +long +x86_bios_freq_base(unsigned long clock_type, unsigned long *ticks_per_second, + unsigned long *drift_info) +{ + return uv_bios_call(UV_BIOS_FREQ_BASE, clock_type, + (u64)ticks_per_second, 0, 0, 0); } EXPORT_SYMBOL_GPL(x86_bios_freq_base); + + +#ifdef CONFIG_EFI +void uv_bios_init(void) +{ + struct uv_systab *tab; + + if ((efi.uv_systab == EFI_INVALID_TABLE_ADDR) || + (efi.uv_systab == (unsigned long)NULL)) { + printk(KERN_CRIT "No EFI UV System Table.\n"); + uv_systab.function = (unsigned long)NULL; + return; + } + + tab = (struct uv_systab *)ioremap(efi.uv_systab, + sizeof(struct uv_systab)); + if (strncmp(tab->signature, "UVST", 4) != 0) + printk(KERN_ERR "bad signature in UV system table!"); + + /* + * Copy table to permanent spot for later use. + */ + memcpy(&uv_systab, tab, sizeof(struct uv_systab)); + iounmap(tab); + + printk(KERN_INFO "EFI UV System Table Revision %d\n", tab->revision); +} +#else /* !CONFIG_EFI */ + +void uv_bios_init(void) { } +#endif + diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index b689075bbe2f..aa2e5e15bf69 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -427,6 +427,7 @@ void __init uv_system_init(void) gnode_upper = (((unsigned long)node_id.s.node_id) & ~((1 << n_val) - 1)) << m_val; + uv_bios_init(); uv_rtc_init(); for_each_present_cpu(cpu) { -- cgit v1.2.3 From 922402f15a85f7a064926eb1db68cc52bc4d4a91 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Fri, 3 Oct 2008 11:59:33 -0500 Subject: x86: Add UV partition call v4 Add a bios call to return partitioning related info. Signed-off-by: Russ Anderson Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/bios_uv.c | 44 +++++++++++++++++++++++++++++++++++----- arch/x86/kernel/genx2apic_uv_x.c | 14 +++++++------ 2 files changed, 47 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index 5481eb59f783..f0dfe6f17e7e 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -23,6 +23,7 @@ #include #include #include +#include struct uv_systab uv_systab; @@ -65,14 +66,47 @@ s64 uv_bios_call_reentrant(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, return ret; } -long -x86_bios_freq_base(unsigned long clock_type, unsigned long *ticks_per_second, - unsigned long *drift_info) + +long sn_partition_id; +EXPORT_SYMBOL_GPL(sn_partition_id); +long uv_coherency_id; +EXPORT_SYMBOL_GPL(uv_coherency_id); +long uv_region_size; +EXPORT_SYMBOL_GPL(uv_region_size); +int uv_type; + + +s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher, + long *region) +{ + s64 ret; + u64 v0, v1; + union partition_info_u part; + + ret = uv_bios_call_irqsave(UV_BIOS_GET_SN_INFO, fc, + (u64)(&v0), (u64)(&v1), 0, 0); + if (ret != BIOS_STATUS_SUCCESS) + return ret; + + part.val = v0; + if (uvtype) + *uvtype = part.hub_version; + if (partid) + *partid = part.partition_id; + if (coher) + *coher = part.coherence_id; + if (region) + *region = part.region_size; + return ret; +} + + +s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) { return uv_bios_call(UV_BIOS_FREQ_BASE, clock_type, - (u64)ticks_per_second, 0, 0, 0); + (u64)ticks_per_second, 0, 0, 0); } -EXPORT_SYMBOL_GPL(x86_bios_freq_base); +EXPORT_SYMBOL_GPL(uv_bios_freq_base); #ifdef CONFIG_EFI diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index aa2e5e15bf69..bfd532843df6 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -341,12 +341,12 @@ static __init void map_mmioh_high(int max_pnode) static __init void uv_rtc_init(void) { - long status, ticks_per_sec, drift; + long status; + u64 ticks_per_sec; - status = - x86_bios_freq_base(BIOS_FREQ_BASE_REALTIME_CLOCK, &ticks_per_sec, - &drift); - if (status != 0 || ticks_per_sec < 100000) { + status = uv_bios_freq_base(BIOS_FREQ_BASE_REALTIME_CLOCK, + &ticks_per_sec); + if (status != BIOS_STATUS_SUCCESS || ticks_per_sec < 100000) { printk(KERN_WARNING "unable to determine platform RTC clock frequency, " "guessing.\n"); @@ -428,6 +428,8 @@ void __init uv_system_init(void) ~((1 << n_val) - 1)) << m_val; uv_bios_init(); + uv_bios_get_sn_info(0, &uv_type, &sn_partition_id, + &uv_coherency_id, &uv_region_size); uv_rtc_init(); for_each_present_cpu(cpu) { @@ -449,7 +451,7 @@ void __init uv_system_init(void) uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1; uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; - uv_cpu_hub_info(cpu)->coherency_domain_number = 0;/* ZZZ */ + uv_cpu_hub_info(cpu)->coherency_domain_number = uv_coherency_id; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode = max(pnode, max_pnode); -- cgit v1.2.3 From fc8c2d763bacc02962048fa042e287debb1416aa Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Fri, 3 Oct 2008 11:59:50 -0500 Subject: x86: Add sysfs entries for UV v4 Create /sys/firmware/sgi_uv sysfs entries for partition_id and coherence_id. Signed-off-by: Russ Anderson Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/uv_sysfs.c | 72 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 73 insertions(+), 1 deletion(-) create mode 100644 arch/x86/kernel/uv_sysfs.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index acc0e8bf043e..cb6ade443f00 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -108,7 +108,7 @@ obj-$(CONFIG_MICROCODE) += microcode.o # 64 bit specific files ifeq ($(CONFIG_X86_64),y) obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o tlb_uv.o - obj-y += bios_uv.o uv_irq.o + obj-y += bios_uv.o uv_irq.o uv_sysfs.o obj-y += genx2apic_cluster.o obj-y += genx2apic_phys.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o diff --git a/arch/x86/kernel/uv_sysfs.c b/arch/x86/kernel/uv_sysfs.c new file mode 100644 index 000000000000..67f9b9dbf800 --- /dev/null +++ b/arch/x86/kernel/uv_sysfs.c @@ -0,0 +1,72 @@ +/* + * This file supports the /sys/firmware/sgi_uv interfaces for SGI UV. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) Russ Anderson + */ + +#include +#include + +struct kobject *sgi_uv_kobj; + +static ssize_t partition_id_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + return snprintf(buf, PAGE_SIZE, "%ld\n", sn_partition_id); +} + +static ssize_t coherence_id_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + return snprintf(buf, PAGE_SIZE, "%ld\n", partition_coherence_id()); +} + +static struct kobj_attribute partition_id_attr = + __ATTR(partition_id, S_IRUGO, partition_id_show, NULL); + +static struct kobj_attribute coherence_id_attr = + __ATTR(coherence_id, S_IRUGO, coherence_id_show, NULL); + + +static int __init sgi_uv_sysfs_init(void) +{ + unsigned long ret; + + if (!sgi_uv_kobj) + sgi_uv_kobj = kobject_create_and_add("sgi_uv", firmware_kobj); + if (!sgi_uv_kobj) { + printk(KERN_WARNING "kobject_create_and_add sgi_uv failed \n"); + return -EINVAL; + } + + ret = sysfs_create_file(sgi_uv_kobj, &partition_id_attr.attr); + if (ret) { + printk(KERN_WARNING "sysfs_create_file partition_id failed \n"); + return ret; + } + + ret = sysfs_create_file(sgi_uv_kobj, &coherence_id_attr.attr); + if (ret) { + printk(KERN_WARNING "sysfs_create_file coherence_id failed \n"); + return ret; + } + + return 0; +} + +device_initcall(sgi_uv_sysfs_init); -- cgit v1.2.3 From 4c66a73f0796dacc2ff0d4af75794ec843ceb3d1 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sat, 4 Oct 2008 15:52:15 -0700 Subject: x86: sparse_irq: fix typo in debug print out Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 260c95a5e6dc..f959acbc0db2 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -257,7 +257,7 @@ static struct irq_cfg *irq_cfg_alloc(unsigned int irq) panic("please boot with nr_irq_cfg= %d\n", count * 2); phys = __pa(cfg); - printk(KERN_DEBUG "irq_irq ==> [%#lx - %#lx]\n", phys, phys + total_bytes); + printk(KERN_DEBUG "irq_cfg ==> [%#lx - %#lx]\n", phys, phys + total_bytes); for (i = 0; i < nr_irq_cfg; i++) init_one_irq_cfg(&cfg[i]); -- cgit v1.2.3 From 81608f3c254512b906ab78082ec5966b376aacd5 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Fri, 10 Oct 2008 19:00:17 +0400 Subject: x86: apic - unify APIC_DIVISOR Use APIC_DIVISOR being set to 16 for both 32/64bit mode. To escape APIC timer underflow during calibration set it to the maximum possible value. Also typo error (CONFG instead of proper CONFIG) fixed. The error was caught by Venkatesh Pallipadi, thanks a lot Venkatesh! See details on http://lkml.org/lkml/2008/10/9/425 Reported-by: Venkatesh Pallipad Signed-off-by: Cyrill Gorcunov Acked-by: "Maciej W. Rozycki" Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 1f48bd1c9f2d..04a7f960bbc0 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -332,11 +332,7 @@ int lapic_get_maxlvt(void) */ /* Clock divisor */ -#ifdef CONFG_X86_64 -#define APIC_DIVISOR 1 -#else #define APIC_DIVISOR 16 -#endif /* * This function sets up the local APIC timer, with a timeout of @@ -592,10 +588,10 @@ static int __init calibrate_APIC_clock(void) global_clock_event->event_handler = lapic_cal_handler; /* - * Setup the APIC counter to 1e9. There is no way the lapic + * Setup the APIC counter to maximum. There is no way the lapic * can underflow in the 100ms detection time frame */ - __setup_APIC_LVTT(1000000000, 0, 0); + __setup_APIC_LVTT(0xffffffff, 0, 0); /* Let the interrupts run */ local_irq_enable(); -- cgit v1.2.3 From 2cc21ef843d4fb7da122239b644a1f6f0aca60a6 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 15 Oct 2008 14:16:55 +0200 Subject: genirq: remove sparse irq code This code is not ready, but we need to rip it out instead of rebasing as we would lose the APIC/IO_APIC unification otherwise. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/io_apic.c | 130 ++-------------------------------------------- arch/x86/kernel/irq_32.c | 8 --- arch/x86/kernel/irq_64.c | 8 --- 3 files changed, 4 insertions(+), 142 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index f959acbc0db2..683610517d2a 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -111,9 +111,6 @@ struct irq_cfg; struct irq_pin_list; struct irq_cfg { unsigned int irq; -#ifdef CONFIG_HAVE_SPARSE_IRQ - struct irq_cfg *next; -#endif struct irq_pin_list *irq_2_pin; cpumask_t domain; cpumask_t old_domain; @@ -151,15 +148,6 @@ static void init_one_irq_cfg(struct irq_cfg *cfg) static struct irq_cfg *irq_cfgx; -#ifdef CONFIG_HAVE_SPARSE_IRQ -/* - * Protect the irq_cfgx_free freelist: - */ -static DEFINE_SPINLOCK(irq_cfg_lock); - -static struct irq_cfg *irq_cfgx_free; -#endif - static void __init init_work(void *data) { struct dyn_array *da = data; @@ -174,114 +162,7 @@ static void __init init_work(void *data) legacy_count = ARRAY_SIZE(irq_cfg_legacy); for (i = legacy_count; i < *da->nr; i++) init_one_irq_cfg(&cfg[i]); - -#ifdef CONFIG_HAVE_SPARSE_IRQ - for (i = 1; i < *da->nr; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = &irq_cfgx[legacy_count]; - irq_cfgx[legacy_count - 1].next = NULL; -#endif -} - -#ifdef CONFIG_HAVE_SPARSE_IRQ -/* need to be biger than size of irq_cfg_legacy */ -static int nr_irq_cfg = 32; - -static int __init parse_nr_irq_cfg(char *arg) -{ - if (arg) { - nr_irq_cfg = simple_strtoul(arg, NULL, 0); - if (nr_irq_cfg < 32) - nr_irq_cfg = 32; - } - return 0; -} - -early_param("nr_irq_cfg", parse_nr_irq_cfg); - -#define for_each_irq_cfg(irqX, cfg) \ - for (cfg = irq_cfgx, irqX = cfg->irq; cfg; cfg = cfg->next, irqX = cfg ? cfg->irq : -1U) - - -DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irq_cfg, PAGE_SIZE, init_work); - -static struct irq_cfg *irq_cfg(unsigned int irq) -{ - struct irq_cfg *cfg; - - cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg = cfg->next; - } - - return NULL; -} - -static struct irq_cfg *irq_cfg_alloc(unsigned int irq) -{ - struct irq_cfg *cfg, *cfg_pri; - unsigned long flags; - int count = 0; - int i; - - cfg_pri = cfg = irq_cfgx; - while (cfg) { - if (cfg->irq == irq) - return cfg; - - cfg_pri = cfg; - cfg = cfg->next; - count++; - } - - spin_lock_irqsave(&irq_cfg_lock, flags); - if (!irq_cfgx_free) { - unsigned long phys; - unsigned long total_bytes; - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_cfg %d\n", nr_irq_cfg); - - total_bytes = sizeof(struct irq_cfg) * nr_irq_cfg; - if (after_bootmem) - cfg = kzalloc(total_bytes, GFP_ATOMIC); - else - cfg = __alloc_bootmem_nopanic(total_bytes, PAGE_SIZE, 0); - - if (!cfg) - panic("please boot with nr_irq_cfg= %d\n", count * 2); - - phys = __pa(cfg); - printk(KERN_DEBUG "irq_cfg ==> [%#lx - %#lx]\n", phys, phys + total_bytes); - - for (i = 0; i < nr_irq_cfg; i++) - init_one_irq_cfg(&cfg[i]); - - for (i = 1; i < nr_irq_cfg; i++) - cfg[i-1].next = &cfg[i]; - - irq_cfgx_free = cfg; - } - - cfg = irq_cfgx_free; - irq_cfgx_free = irq_cfgx_free->next; - cfg->next = NULL; - if (cfg_pri) - cfg_pri->next = cfg; - else - irq_cfgx = cfg; - cfg->irq = irq; - - spin_unlock_irqrestore(&irq_cfg_lock, flags); - - return cfg; } -#else #define for_each_irq_cfg(irq, cfg) \ for (irq = 0, cfg = &irq_cfgx[irq]; irq < nr_irqs; irq++, cfg = &irq_cfgx[irq]) @@ -290,17 +171,16 @@ DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work struct irq_cfg *irq_cfg(unsigned int irq) { - if (irq < nr_irqs) - return &irq_cfgx[irq]; + if (irq < nr_irqs) + return &irq_cfgx[irq]; - return NULL; + return NULL; } struct irq_cfg *irq_cfg_alloc(unsigned int irq) { - return irq_cfg(irq); + return irq_cfg(irq); } -#endif /* * This is performance-critical, we want to do it O(1) * @@ -3068,9 +2948,7 @@ unsigned int create_irq_nr(unsigned int irq_want) unsigned long flags; struct irq_cfg *cfg_new; -#ifndef CONFIG_HAVE_SPARSE_IRQ irq_want = nr_irqs - 1; -#endif irq = 0; spin_lock_irqsave(&vector_lock, flags); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 001772ffc918..ccf6c1bf7120 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -272,20 +272,12 @@ int show_interrupts(struct seq_file *p, void *v) struct irq_desc *desc = NULL; int tail = 0; -#ifdef CONFIG_HAVE_SPARSE_IRQ - desc = (struct irq_desc *)v; - entries = -1U; - i = desc->irq; - if (!desc->next) - tail = 1; -#else entries = nr_irqs - 1; i = *(loff_t *) v; if (i == nr_irqs) tail = 1; else desc = irq_to_desc(i); -#endif if (i == 0) { seq_printf(p, " "); diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index ec2661091283..21f53b911113 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -77,20 +77,12 @@ int show_interrupts(struct seq_file *p, void *v) struct irq_desc *desc = NULL; int tail = 0; -#ifdef CONFIG_HAVE_SPARSE_IRQ - desc = (struct irq_desc *)v; - entries = -1U; - i = desc->irq; - if (!desc->next) - tail = 1; -#else entries = nr_irqs - 1; i = *(loff_t *) v; if (i == nr_irqs) tail = 1; else desc = irq_to_desc(i); -#endif if (i == 0) { seq_printf(p, " "); -- cgit v1.2.3 From ee32c9732244bde4b9b59eeac2814c23e2b71f8d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 15 Oct 2008 14:34:09 +0200 Subject: genirq: remove irq_to_desc_alloc Remove the leftover of sparseirqs. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/io_apic.c | 6 +----- arch/x86/kernel/irqinit_32.c | 2 +- arch/x86/kernel/irqinit_64.c | 2 +- 3 files changed, 3 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 683610517d2a..e03bc0f87eef 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1257,11 +1257,7 @@ static void ioapic_register_intr(int irq, unsigned long trigger) { struct irq_desc *desc; - /* first time to use this irq_desc */ - if (irq < 16) - desc = irq_to_desc(irq); - else - desc = irq_to_desc_alloc(irq); + desc = irq_to_desc(irq); if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 9092103a18eb..a8d35998d308 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -70,7 +70,7 @@ void __init init_ISA_irqs (void) */ for (i = 0; i < 16; i++) { /* first time call this irq_desc */ - struct irq_desc *desc = irq_to_desc_alloc(i); + struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; desc->action = NULL; diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index d17fbc26d96f..ff0235391285 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -144,7 +144,7 @@ void __init init_ISA_irqs(void) for (i = 0; i < 16; i++) { /* first time call this irq_desc */ - struct irq_desc *desc = irq_to_desc_alloc(i); + struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; desc->action = NULL; -- cgit v1.2.3 From d6c88a507ef0b6afdb013cba4e7804ba7324d99a Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 15 Oct 2008 15:27:23 +0200 Subject: genirq: revert dynarray Revert the dynarray changes. They need more thought and polishing. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/io_apic.c | 199 +++++++++++++++------------------------ arch/x86/kernel/setup_percpu.c | 8 +- arch/x86/kernel/visws_quirks.c | 2 +- arch/x86/kernel/vmlinux_32.lds.S | 1 - arch/x86/kernel/vmlinux_64.lds.S | 2 - 5 files changed, 77 insertions(+), 135 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index e03bc0f87eef..6f80dc2f137e 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -107,7 +107,6 @@ static int __init parse_noapic(char *str) } early_param("noapic", parse_noapic); -struct irq_cfg; struct irq_pin_list; struct irq_cfg { unsigned int irq; @@ -120,7 +119,7 @@ struct irq_cfg { }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ -static struct irq_cfg irq_cfg_legacy[] __initdata = { +static struct irq_cfg irq_cfgx[NR_IRQS] = { [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, @@ -139,48 +138,26 @@ static struct irq_cfg irq_cfg_legacy[] __initdata = { [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, }; -static struct irq_cfg irq_cfg_init = { .irq = -1U, }; - -static void init_one_irq_cfg(struct irq_cfg *cfg) -{ - memcpy(cfg, &irq_cfg_init, sizeof(struct irq_cfg)); -} - -static struct irq_cfg *irq_cfgx; - -static void __init init_work(void *data) -{ - struct dyn_array *da = data; - struct irq_cfg *cfg; - int legacy_count; - int i; - - cfg = *da->name; - - memcpy(cfg, irq_cfg_legacy, sizeof(irq_cfg_legacy)); - - legacy_count = ARRAY_SIZE(irq_cfg_legacy); - for (i = legacy_count; i < *da->nr; i++) - init_one_irq_cfg(&cfg[i]); -} - #define for_each_irq_cfg(irq, cfg) \ - for (irq = 0, cfg = &irq_cfgx[irq]; irq < nr_irqs; irq++, cfg = &irq_cfgx[irq]) + for (irq = 0, cfg = irq_cfgx; irq < nr_irqs; irq++, cfg++) -DEFINE_DYN_ARRAY(irq_cfgx, sizeof(struct irq_cfg), nr_irqs, PAGE_SIZE, init_work); - -struct irq_cfg *irq_cfg(unsigned int irq) +static struct irq_cfg *irq_cfg(unsigned int irq) { - if (irq < nr_irqs) - return &irq_cfgx[irq]; - - return NULL; + return irq < nr_irqs ? irq_cfgx + irq : NULL; } -struct irq_cfg *irq_cfg_alloc(unsigned int irq) + +static struct irq_cfg *irq_cfg_alloc(unsigned int irq) { return irq_cfg(irq); } +/* + * Rough estimation of how many shared IRQs there are, can be changed + * anytime. + */ +#define MAX_PLUS_SHARED_IRQS NR_IRQS +#define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) + /* * This is performance-critical, we want to do it O(1) * @@ -193,59 +170,29 @@ struct irq_pin_list { struct irq_pin_list *next; }; -static struct irq_pin_list *irq_2_pin_head; -/* fill one page ? */ -static int nr_irq_2_pin = 0x100; +static struct irq_pin_list irq_2_pin_head[PIN_MAP_SIZE]; static struct irq_pin_list *irq_2_pin_ptr; -static void __init irq_2_pin_init_work(void *data) + +static void __init irq_2_pin_init(void) { - struct dyn_array *da = data; - struct irq_pin_list *pin; + struct irq_pin_list *pin = irq_2_pin_head; int i; - pin = *da->name; - - for (i = 1; i < *da->nr; i++) + for (i = 1; i < PIN_MAP_SIZE; i++) pin[i-1].next = &pin[i]; irq_2_pin_ptr = &pin[0]; } -DEFINE_DYN_ARRAY(irq_2_pin_head, sizeof(struct irq_pin_list), nr_irq_2_pin, PAGE_SIZE, irq_2_pin_init_work); static struct irq_pin_list *get_one_free_irq_2_pin(void) { - struct irq_pin_list *pin; - int i; - - pin = irq_2_pin_ptr; - - if (pin) { - irq_2_pin_ptr = pin->next; - pin->next = NULL; - return pin; - } - - /* - * we run out of pre-allocate ones, allocate more - */ - printk(KERN_DEBUG "try to get more irq_2_pin %d\n", nr_irq_2_pin); - - if (after_bootmem) - pin = kzalloc(sizeof(struct irq_pin_list)*nr_irq_2_pin, - GFP_ATOMIC); - else - pin = __alloc_bootmem_nopanic(sizeof(struct irq_pin_list) * - nr_irq_2_pin, PAGE_SIZE, 0); + struct irq_pin_list *pin = irq_2_pin_ptr; if (!pin) panic("can not get more irq_2_pin\n"); - for (i = 1; i < nr_irq_2_pin; i++) - pin[i-1].next = &pin[i]; - irq_2_pin_ptr = pin->next; pin->next = NULL; - return pin; } @@ -284,8 +231,9 @@ static inline void io_apic_write(unsigned int apic, unsigned int reg, unsigned i static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned int value) { struct io_apic __iomem *io_apic = io_apic_base(apic); - if (sis_apic_bug) - writel(reg, &io_apic->index); + + if (sis_apic_bug) + writel(reg, &io_apic->index); writel(value, &io_apic->data); } @@ -1044,11 +992,11 @@ static int pin_2_irq(int idx, int apic, int pin) while (i < apic) irq += nr_ioapic_registers[i++]; irq += pin; - /* + /* * For MPS mode, so far only needed by ES7000 platform */ - if (ioapic_renumber_irq) - irq = ioapic_renumber_irq(apic, irq); + if (ioapic_renumber_irq) + irq = ioapic_renumber_irq(apic, irq); } #ifdef CONFIG_X86_32 @@ -1232,19 +1180,19 @@ static struct irq_chip ir_ioapic_chip; #ifdef CONFIG_X86_32 static inline int IO_APIC_irq_trigger(int irq) { - int apic, idx, pin; + int apic, idx, pin; - for (apic = 0; apic < nr_ioapics; apic++) { - for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { - idx = find_irq_entry(apic, pin, mp_INT); - if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) - return irq_trigger(idx); - } - } - /* + for (apic = 0; apic < nr_ioapics; apic++) { + for (pin = 0; pin < nr_ioapic_registers[apic]; pin++) { + idx = find_irq_entry(apic, pin, mp_INT); + if ((idx != -1) && (irq == pin_2_irq(idx, apic, pin))) + return irq_trigger(idx); + } + } + /* * nonexistent IRQs are edge default */ - return 0; + return 0; } #else static inline int IO_APIC_irq_trigger(int irq) @@ -1509,8 +1457,8 @@ __apicdebuginit(void) print_IO_APIC(void) reg_01.raw = io_apic_read(apic, 1); if (reg_01.bits.version >= 0x10) reg_02.raw = io_apic_read(apic, 2); - if (reg_01.bits.version >= 0x20) - reg_03.raw = io_apic_read(apic, 3); + if (reg_01.bits.version >= 0x20) + reg_03.raw = io_apic_read(apic, 3); spin_unlock_irqrestore(&ioapic_lock, flags); printk("\n"); @@ -2089,9 +2037,9 @@ static int ioapic_retrigger_irq(unsigned int irq) #else static int ioapic_retrigger_irq(unsigned int irq) { - send_IPI_self(irq_cfg(irq)->vector); + send_IPI_self(irq_cfg(irq)->vector); - return 1; + return 1; } #endif @@ -2189,7 +2137,7 @@ static int migrate_irq_remapped_level(int irq) if (io_apic_level_ack_pending(irq)) { /* - * Interrupt in progress. Migrating irq now will change the + * Interrupt in progress. Migrating irq now will change the * vector information in the IO-APIC RTE and that will confuse * the EOI broadcast performed by cpu. * So, delay the irq migration to the next instance. @@ -2426,28 +2374,28 @@ static void ack_apic_level(unsigned int irq) } static struct irq_chip ioapic_chip __read_mostly = { - .name = "IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_apic_edge, - .eoi = ack_apic_level, + .name = "IO-APIC", + .startup = startup_ioapic_irq, + .mask = mask_IO_APIC_irq, + .unmask = unmask_IO_APIC_irq, + .ack = ack_apic_edge, + .eoi = ack_apic_level, #ifdef CONFIG_SMP - .set_affinity = set_ioapic_affinity_irq, + .set_affinity = set_ioapic_affinity_irq, #endif .retrigger = ioapic_retrigger_irq, }; #ifdef CONFIG_INTR_REMAP static struct irq_chip ir_ioapic_chip __read_mostly = { - .name = "IR-IO-APIC", - .startup = startup_ioapic_irq, - .mask = mask_IO_APIC_irq, - .unmask = unmask_IO_APIC_irq, - .ack = ack_x2apic_edge, - .eoi = ack_x2apic_level, + .name = "IR-IO-APIC", + .startup = startup_ioapic_irq, + .mask = mask_IO_APIC_irq, + .unmask = unmask_IO_APIC_irq, + .ack = ack_x2apic_edge, + .eoi = ack_x2apic_level, #ifdef CONFIG_SMP - .set_affinity = set_ir_ioapic_affinity_irq, + .set_affinity = set_ir_ioapic_affinity_irq, #endif .retrigger = ioapic_retrigger_irq, }; @@ -2636,8 +2584,8 @@ static inline void __init check_timer(void) local_irq_save(flags); - ver = apic_read(APIC_LVR); - ver = GET_APIC_VERSION(ver); + ver = apic_read(APIC_LVR); + ver = GET_APIC_VERSION(ver); /* * get/set the timer IRQ vector: @@ -2822,12 +2770,12 @@ void __init setup_IO_APIC(void) io_apic_irqs = ~PIC_IRQS; apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); - /* + /* * Set up IO-APIC IRQ routing. */ #ifdef CONFIG_X86_32 - if (!acpi_ioapic) - setup_ioapic_ids_from_mpc(); + if (!acpi_ioapic) + setup_ioapic_ids_from_mpc(); #endif sync_Arb_IDs(); setup_IO_APIC_irqs(); @@ -2842,9 +2790,9 @@ void __init setup_IO_APIC(void) static int __init io_apic_bug_finalize(void) { - if (sis_apic_bug == -1) - sis_apic_bug = 0; - return 0; + if (sis_apic_bug == -1) + sis_apic_bug = 0; + return 0; } late_initcall(io_apic_bug_finalize); @@ -3199,7 +3147,7 @@ static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) if (index < 0) { printk(KERN_ERR "Unable to allocate %d IRTE for PCI %s\n", nvec, - pci_name(dev)); + pci_name(dev)); return -ENOSPC; } return index; @@ -3885,23 +3833,24 @@ static struct resource * __init ioapic_setup_resources(void) void __init ioapic_init_mappings(void) { unsigned long ioapic_phys, idx = FIX_IO_APIC_BASE_0; - int i; struct resource *ioapic_res; + int i; + irq_2_pin_init(); ioapic_res = ioapic_setup_resources(); for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { ioapic_phys = mp_ioapics[i].mp_apicaddr; #ifdef CONFIG_X86_32 - if (!ioapic_phys) { - printk(KERN_ERR - "WARNING: bogus zero IO-APIC " - "address found in MPTABLE, " - "disabling IO/APIC support!\n"); - smp_found_config = 0; - skip_ioapic_setup = 1; - goto fake_ioapic_page; - } + if (!ioapic_phys) { + printk(KERN_ERR + "WARNING: bogus zero IO-APIC " + "address found in MPTABLE, " + "disabling IO/APIC support!\n"); + smp_found_config = 0; + skip_ioapic_setup = 1; + goto fake_ioapic_page; + } #endif } else { #ifdef CONFIG_X86_32 diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 2b7dab699e83..410c88f0bfeb 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -140,7 +140,7 @@ static void __init setup_cpu_pda_map(void) */ void __init setup_per_cpu_areas(void) { - ssize_t size, old_size, da_size; + ssize_t size, old_size; char *ptr; int cpu; unsigned long align = 1; @@ -150,9 +150,8 @@ void __init setup_per_cpu_areas(void) /* Copy section for each CPU (we discard the original) */ old_size = PERCPU_ENOUGH_ROOM; - da_size = per_cpu_dyn_array_size(&align); align = max_t(unsigned long, PAGE_SIZE, align); - size = roundup(old_size + da_size, align); + size = roundup(old_size, align); printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n", size); @@ -182,9 +181,6 @@ void __init setup_per_cpu_areas(void) #endif per_cpu_offset(cpu) = ptr - __per_cpu_start; memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); - - per_cpu_alloc_dyn_array(cpu, ptr + old_size); - } printk(KERN_DEBUG "NR_CPUS: %d, nr_cpu_ids: %d, nr_node_ids %d\n", diff --git a/arch/x86/kernel/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 817aa55a1209..0c9667f0752a 100644 --- a/arch/x86/kernel/visws_quirks.c +++ b/arch/x86/kernel/visws_quirks.c @@ -633,7 +633,7 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) /* * handle this 'virtual interrupt' as a Cobalt one now. */ - kstat_irqs_this_cpu(desc)++; + kstat_incr_irqs_this_cpu(realirq, desc); if (likely(desc->action != NULL)) handle_IRQ_event(realirq, desc->action); diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index c36007ab3940..a9b8560adbc2 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -145,7 +145,6 @@ SECTIONS *(.x86_cpu_dev.init) __x86_cpu_dev_end = .; } - DYN_ARRAY_INIT(8) SECURITY_INIT . = ALIGN(4); .altinstructions : AT(ADDR(.altinstructions) - LOAD_OFFSET) { diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 30973dbac8c2..3245ad72594a 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -174,8 +174,6 @@ SECTIONS } __x86_cpu_dev_end = .; - DYN_ARRAY_INIT(8) - SECURITY_INIT . = ALIGN(8); -- cgit v1.2.3 From a1aca5de08a0cb840a90fb3f729a5940f8d21185 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 15 Oct 2008 19:29:15 +0200 Subject: genirq: remove artifacts from sparseirq removal Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 1 - arch/x86/kernel/irqinit_32.c | 1 - arch/x86/kernel/vmlinux_64.lds.S | 1 - 3 files changed, 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 5fef4fece4a5..0d1c26a583c5 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -1254,7 +1254,6 @@ static int __init acpi_parse_madt_ioapic_entries(void) return count; } - count = acpi_table_parse_madt(ACPI_MADT_TYPE_INTERRUPT_OVERRIDE, acpi_parse_int_src_ovr, nr_irqs); diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index a8d35998d308..845aa9803e80 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -184,4 +184,3 @@ void __init native_init_IRQ(void) irq_ctx_init(smp_processor_id()); } - diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 3245ad72594a..46e05447405b 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -173,7 +173,6 @@ SECTIONS *(.x86_cpu_dev.init) } __x86_cpu_dev_end = .; - SECURITY_INIT . = ALIGN(8); -- cgit v1.2.3 From c0c168ca26b54a4a6ad34fc813fe00f275fbc94c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 16 Oct 2008 11:15:28 +0200 Subject: x86: cleanup show_interrupts The sparseirq patches introduced some more ugliness in show_interrupts(). Clean it up all together and make the code easier to read by splitting out the "tail" function which prints the special interrupts. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/irq_32.c | 160 +++++++++++++++++++++++------------------------ arch/x86/kernel/irq_64.c | 153 ++++++++++++++++++++++---------------------- 2 files changed, 154 insertions(+), 159 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index ccf6c1bf7120..8d525765a6c4 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -263,22 +263,67 @@ atomic_t irq_err_count; * /proc/interrupts printing: */ +static int show_other_interrupts(struct seq_file *p) +{ + int j; + + seq_printf(p, "NMI: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", nmi_count(j)); + seq_printf(p, " Non-maskable interrupts\n"); +#ifdef CONFIG_X86_LOCAL_APIC + seq_printf(p, "LOC: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).apic_timer_irqs); + seq_printf(p, " Local timer interrupts\n"); +#endif +#ifdef CONFIG_SMP + seq_printf(p, "RES: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_resched_count); + seq_printf(p, " Rescheduling interrupts\n"); + seq_printf(p, "CAL: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_call_count); + seq_printf(p, " Function call interrupts\n"); + seq_printf(p, "TLB: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_tlb_count); + seq_printf(p, " TLB shootdowns\n"); +#endif +#ifdef CONFIG_X86_MCE + seq_printf(p, "TRM: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_thermal_count); + seq_printf(p, " Thermal event interrupts\n"); +#endif +#ifdef CONFIG_X86_LOCAL_APIC + seq_printf(p, "SPU: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_spurious_count); + seq_printf(p, " Spurious interrupts\n"); +#endif + seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); +#if defined(CONFIG_X86_IO_APIC) + seq_printf(p, "MIS: %10u\n", atomic_read(&irq_mis_count)); +#endif + return 0; +} + int show_interrupts(struct seq_file *p, void *v) { + unsigned long flags, any_count = 0; int i = *(loff_t *) v, j; - struct irqaction * action; - unsigned long flags; - unsigned int entries; - struct irq_desc *desc = NULL; - int tail = 0; + struct irqaction *action; + struct irq_desc *desc; + + if (i > nr_irqs) + return 0; - entries = nr_irqs - 1; - i = *(loff_t *) v; if (i == nr_irqs) - tail = 1; - else - desc = irq_to_desc(i); + return show_other_interrupts(p); + /* print header */ if (i == 0) { seq_printf(p, " "); for_each_online_cpu(j) @@ -286,88 +331,37 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i <= entries) { - unsigned any_count = 0; - - spin_lock_irqsave(&desc->lock, flags); + desc = irq_to_desc(i); + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP - any_count = kstat_irqs(i); + any_count = kstat_irqs(i); #else - for_each_online_cpu(j) - any_count |= kstat_irqs_cpu(i, j); + for_each_online_cpu(j) + any_count |= kstat_irqs_cpu(i, j); #endif - action = desc->action; - if (!action && !any_count) - goto skip; - seq_printf(p, "%3d: ", i); + action = desc->action; + if (!action && !any_count) + goto out; + + seq_printf(p, "%3d: ", i); #ifndef CONFIG_SMP - seq_printf(p, "%10u ", kstat_irqs(i)); + seq_printf(p, "%10u ", kstat_irqs(i)); #else - for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); + for_each_online_cpu(j) + seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); #endif - seq_printf(p, " %8s", desc->chip->name); - seq_printf(p, "-%-8s", desc->name); - - if (action) { - seq_printf(p, " %s", action->name); - while ((action = action->next) != NULL) - seq_printf(p, ", %s", action->name); - } + seq_printf(p, " %8s", desc->chip->name); + seq_printf(p, "-%-8s", desc->name); - seq_putc(p, '\n'); -skip: - spin_unlock_irqrestore(&desc->lock, flags); + if (action) { + seq_printf(p, " %s", action->name); + while ((action = action->next) != NULL) + seq_printf(p, ", %s", action->name); } - if (tail) { - seq_printf(p, "NMI: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", nmi_count(j)); - seq_printf(p, " Non-maskable interrupts\n"); -#ifdef CONFIG_X86_LOCAL_APIC - seq_printf(p, "LOC: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).apic_timer_irqs); - seq_printf(p, " Local timer interrupts\n"); -#endif -#ifdef CONFIG_SMP - seq_printf(p, "RES: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).irq_resched_count); - seq_printf(p, " Rescheduling interrupts\n"); - seq_printf(p, "CAL: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).irq_call_count); - seq_printf(p, " Function call interrupts\n"); - seq_printf(p, "TLB: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).irq_tlb_count); - seq_printf(p, " TLB shootdowns\n"); -#endif -#ifdef CONFIG_X86_MCE - seq_printf(p, "TRM: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).irq_thermal_count); - seq_printf(p, " Thermal event interrupts\n"); -#endif -#ifdef CONFIG_X86_LOCAL_APIC - seq_printf(p, "SPU: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", - per_cpu(irq_stat,j).irq_spurious_count); - seq_printf(p, " Spurious interrupts\n"); -#endif - seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); -#if defined(CONFIG_X86_IO_APIC) - seq_printf(p, "MIS: %10u\n", atomic_read(&irq_mis_count)); -#endif - } + seq_putc(p, '\n'); +out: + spin_unlock_irqrestore(&desc->lock, flags); return 0; } diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 21f53b911113..4f374294f292 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -68,22 +68,65 @@ static inline void stack_overflow_check(struct pt_regs *regs) * Generic, controller-independent functions: */ +static int show_other_interrupts(struct seq_file *p) +{ + int j; + + seq_printf(p, "NMI: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count); + seq_printf(p, " Non-maskable interrupts\n"); + seq_printf(p, "LOC: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->apic_timer_irqs); + seq_printf(p, " Local timer interrupts\n"); +#ifdef CONFIG_SMP + seq_printf(p, "RES: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_resched_count); + seq_printf(p, " Rescheduling interrupts\n"); + seq_printf(p, "CAL: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_call_count); + seq_printf(p, " Function call interrupts\n"); + seq_printf(p, "TLB: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_tlb_count); + seq_printf(p, " TLB shootdowns\n"); +#endif +#ifdef CONFIG_X86_MCE + seq_printf(p, "TRM: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_thermal_count); + seq_printf(p, " Thermal event interrupts\n"); + seq_printf(p, "THR: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_threshold_count); + seq_printf(p, " Threshold APIC interrupts\n"); +#endif + seq_printf(p, "SPU: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", cpu_pda(j)->irq_spurious_count); + seq_printf(p, " Spurious interrupts\n"); + seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); + + return 0; +} + int show_interrupts(struct seq_file *p, void *v) { - int i, j; - struct irqaction * action; - unsigned long flags; - unsigned int entries; - struct irq_desc *desc = NULL; - int tail = 0; - - entries = nr_irqs - 1; - i = *(loff_t *) v; + unsigned long flags, any_count = 0; + int i = *(loff_t *) v, j; + struct irqaction *action; + struct irq_desc *desc; + + if (i > nr_irqs) + return 0; + if (i == nr_irqs) - tail = 1; - else - desc = irq_to_desc(i); + return show_other_interrupts(p); + /* print header */ if (i == 0) { seq_printf(p, " "); for_each_online_cpu(j) @@ -91,79 +134,37 @@ int show_interrupts(struct seq_file *p, void *v) seq_putc(p, '\n'); } - if (i <= entries) { - unsigned any_count = 0; - - spin_lock_irqsave(&desc->lock, flags); + desc = irq_to_desc(i); + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP - any_count = kstat_irqs(i); + any_count = kstat_irqs(i); #else - for_each_online_cpu(j) - any_count |= kstat_irqs_cpu(i, j); + for_each_online_cpu(j) + any_count |= kstat_irqs_cpu(i, j); #endif - action = desc->action; - if (!action && !any_count) - goto skip; - seq_printf(p, "%3d: ", i); + action = desc->action; + if (!action && !any_count) + goto out; + + seq_printf(p, "%3d: ", i); #ifndef CONFIG_SMP - seq_printf(p, "%10u ", kstat_irqs(i)); + seq_printf(p, "%10u ", kstat_irqs(i)); #else - for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); + for_each_online_cpu(j) + seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); #endif - seq_printf(p, " %8s", desc->chip->name); - seq_printf(p, "-%-8s", desc->name); + seq_printf(p, " %8s", desc->chip->name); + seq_printf(p, "-%-8s", desc->name); - if (action) { - seq_printf(p, " %s", action->name); - while ((action = action->next) != NULL) - seq_printf(p, ", %s", action->name); - } - seq_putc(p, '\n'); -skip: - spin_unlock_irqrestore(&desc->lock, flags); - } - - if (tail) { - seq_printf(p, "NMI: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count); - seq_printf(p, " Non-maskable interrupts\n"); - seq_printf(p, "LOC: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->apic_timer_irqs); - seq_printf(p, " Local timer interrupts\n"); -#ifdef CONFIG_SMP - seq_printf(p, "RES: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_resched_count); - seq_printf(p, " Rescheduling interrupts\n"); - seq_printf(p, "CAL: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_call_count); - seq_printf(p, " Function call interrupts\n"); - seq_printf(p, "TLB: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_tlb_count); - seq_printf(p, " TLB shootdowns\n"); -#endif -#ifdef CONFIG_X86_MCE - seq_printf(p, "TRM: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_thermal_count); - seq_printf(p, " Thermal event interrupts\n"); - seq_printf(p, "THR: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_threshold_count); - seq_printf(p, " Threshold APIC interrupts\n"); -#endif - seq_printf(p, "SPU: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_spurious_count); - seq_printf(p, " Spurious interrupts\n"); - seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); + if (action) { + seq_printf(p, " %s", action->name); + while ((action = action->next) != NULL) + seq_printf(p, ", %s", action->name); } + seq_putc(p, '\n'); +out: + spin_unlock_irqrestore(&desc->lock, flags); return 0; } -- cgit v1.2.3 From 6b39ba771e3c78d00e0abcebad270bd4212b28bc Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 16 Oct 2008 11:32:24 +0200 Subject: x86: unify show_interrupts() and proc helpers show_interrupts() and proc helpers are basically the same for 32 and 64 bit. Move them to a shared source file. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/irq.c | 166 +++++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/irq_32.c | 146 ----------------------------------------- arch/x86/kernel/irq_64.c | 132 ------------------------------------- 4 files changed, 167 insertions(+), 279 deletions(-) create mode 100644 arch/x86/kernel/irq.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index cb6ade443f00..d7e5a58ee22f 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -23,7 +23,7 @@ CFLAGS_hpet.o := $(nostackp) CFLAGS_tsc.o := $(nostackp) obj-y := process_$(BITS).o signal_$(BITS).o entry_$(BITS).o -obj-y += traps.o irq_$(BITS).o dumpstack_$(BITS).o +obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o obj-y += time_$(BITS).o ioport.o ldt.o obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o obj-$(CONFIG_X86_VISWS) += visws_quirks.o diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c new file mode 100644 index 000000000000..3b9128498976 --- /dev/null +++ b/arch/x86/kernel/irq.c @@ -0,0 +1,166 @@ +/* + * Common interrupt code for 32 and 64 bit + */ +#include +#include +#include +#include + +#include +#include +#include + +atomic_t irq_err_count; + +#ifdef CONFIG_X86_32 +# define irq_stats(x) (&per_cpu(irq_stat,x)) +#else +# define irq_stats(x) cpu_pda(x) +#endif +/* + * /proc/interrupts printing: + */ +static int show_other_interrupts(struct seq_file *p) +{ + int j; + + seq_printf(p, "NMI: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->__nmi_count); + seq_printf(p, " Non-maskable interrupts\n"); +#ifdef CONFIG_X86_LOCAL_APIC + seq_printf(p, "LOC: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->apic_timer_irqs); + seq_printf(p, " Local timer interrupts\n"); +#endif +#ifdef CONFIG_SMP + seq_printf(p, "RES: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_resched_count); + seq_printf(p, " Rescheduling interrupts\n"); + seq_printf(p, "CAL: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_call_count); + seq_printf(p, " Function call interrupts\n"); + seq_printf(p, "TLB: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_tlb_count); + seq_printf(p, " TLB shootdowns\n"); +#endif +#ifdef CONFIG_X86_MCE + seq_printf(p, "TRM: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_thermal_count); + seq_printf(p, " Thermal event interrupts\n"); +# ifdef CONFIG_X86_64 + seq_printf(p, "THR: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_threshold_count); + seq_printf(p, " Threshold APIC interrupts\n"); +# endif +#endif +#ifdef CONFIG_X86_LOCAL_APIC + seq_printf(p, "SPU: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->irq_spurious_count); + seq_printf(p, " Spurious interrupts\n"); +#endif + seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); +#if defined(CONFIG_X86_IO_APIC) + seq_printf(p, "MIS: %10u\n", atomic_read(&irq_mis_count)); +#endif + return 0; +} + +int show_interrupts(struct seq_file *p, void *v) +{ + unsigned long flags, any_count = 0; + int i = *(loff_t *) v, j; + struct irqaction *action; + struct irq_desc *desc; + + if (i > nr_irqs) + return 0; + + if (i == nr_irqs) + return show_other_interrupts(p); + + /* print header */ + if (i == 0) { + seq_printf(p, " "); + for_each_online_cpu(j) + seq_printf(p, "CPU%-8d",j); + seq_putc(p, '\n'); + } + + desc = irq_to_desc(i); + spin_lock_irqsave(&desc->lock, flags); +#ifndef CONFIG_SMP + any_count = kstat_irqs(i); +#else + for_each_online_cpu(j) + any_count |= kstat_irqs_cpu(i, j); +#endif + action = desc->action; + if (!action && !any_count) + goto out; + + seq_printf(p, "%3d: ", i); +#ifndef CONFIG_SMP + seq_printf(p, "%10u ", kstat_irqs(i)); +#else + for_each_online_cpu(j) + seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); +#endif + seq_printf(p, " %8s", desc->chip->name); + seq_printf(p, "-%-8s", desc->name); + + if (action) { + seq_printf(p, " %s", action->name); + while ((action = action->next) != NULL) + seq_printf(p, ", %s", action->name); + } + + seq_putc(p, '\n'); +out: + spin_unlock_irqrestore(&desc->lock, flags); + return 0; +} + +/* + * /proc/stat helpers + */ +u64 arch_irq_stat_cpu(unsigned int cpu) +{ + u64 sum = irq_stats(cpu)->__nmi_count; + +#ifdef CONFIG_X86_LOCAL_APIC + sum += irq_stats(cpu)->apic_timer_irqs; +#endif +#ifdef CONFIG_SMP + sum += irq_stats(cpu)->irq_resched_count; + sum += irq_stats(cpu)->irq_call_count; + sum += irq_stats(cpu)->irq_tlb_count; +#endif +#ifdef CONFIG_X86_MCE + sum += irq_stats(cpu)->irq_thermal_count; +# ifdef CONFIG_X86_64 + sum += irq_stats(cpu)->irq_threshold_count; +#endif +#endif +#ifdef CONFIG_X86_LOCAL_APIC + sum += irq_stats(cpu)->irq_spurious_count; +#endif + return sum; +} + +u64 arch_irq_stat(void) +{ + u64 sum = atomic_read(&irq_err_count); + +#ifdef CONFIG_X86_IO_APIC + sum += atomic_read(&irq_mis_count); +#endif + return sum; +} diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 8d525765a6c4..6d9bf3936c78 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -253,152 +253,6 @@ unsigned int do_IRQ(struct pt_regs *regs) return 1; } -/* - * Interrupt statistics: - */ - -atomic_t irq_err_count; - -/* - * /proc/interrupts printing: - */ - -static int show_other_interrupts(struct seq_file *p) -{ - int j; - - seq_printf(p, "NMI: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", nmi_count(j)); - seq_printf(p, " Non-maskable interrupts\n"); -#ifdef CONFIG_X86_LOCAL_APIC - seq_printf(p, "LOC: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).apic_timer_irqs); - seq_printf(p, " Local timer interrupts\n"); -#endif -#ifdef CONFIG_SMP - seq_printf(p, "RES: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_resched_count); - seq_printf(p, " Rescheduling interrupts\n"); - seq_printf(p, "CAL: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_call_count); - seq_printf(p, " Function call interrupts\n"); - seq_printf(p, "TLB: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_tlb_count); - seq_printf(p, " TLB shootdowns\n"); -#endif -#ifdef CONFIG_X86_MCE - seq_printf(p, "TRM: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_thermal_count); - seq_printf(p, " Thermal event interrupts\n"); -#endif -#ifdef CONFIG_X86_LOCAL_APIC - seq_printf(p, "SPU: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", per_cpu(irq_stat,j).irq_spurious_count); - seq_printf(p, " Spurious interrupts\n"); -#endif - seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); -#if defined(CONFIG_X86_IO_APIC) - seq_printf(p, "MIS: %10u\n", atomic_read(&irq_mis_count)); -#endif - return 0; -} - -int show_interrupts(struct seq_file *p, void *v) -{ - unsigned long flags, any_count = 0; - int i = *(loff_t *) v, j; - struct irqaction *action; - struct irq_desc *desc; - - if (i > nr_irqs) - return 0; - - if (i == nr_irqs) - return show_other_interrupts(p); - - /* print header */ - if (i == 0) { - seq_printf(p, " "); - for_each_online_cpu(j) - seq_printf(p, "CPU%-8d",j); - seq_putc(p, '\n'); - } - - desc = irq_to_desc(i); - spin_lock_irqsave(&desc->lock, flags); -#ifndef CONFIG_SMP - any_count = kstat_irqs(i); -#else - for_each_online_cpu(j) - any_count |= kstat_irqs_cpu(i, j); -#endif - action = desc->action; - if (!action && !any_count) - goto out; - - seq_printf(p, "%3d: ", i); -#ifndef CONFIG_SMP - seq_printf(p, "%10u ", kstat_irqs(i)); -#else - for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); -#endif - seq_printf(p, " %8s", desc->chip->name); - seq_printf(p, "-%-8s", desc->name); - - if (action) { - seq_printf(p, " %s", action->name); - while ((action = action->next) != NULL) - seq_printf(p, ", %s", action->name); - } - - seq_putc(p, '\n'); -out: - spin_unlock_irqrestore(&desc->lock, flags); - return 0; -} - -/* - * /proc/stat helpers - */ -u64 arch_irq_stat_cpu(unsigned int cpu) -{ - u64 sum = nmi_count(cpu); - -#ifdef CONFIG_X86_LOCAL_APIC - sum += per_cpu(irq_stat, cpu).apic_timer_irqs; -#endif -#ifdef CONFIG_SMP - sum += per_cpu(irq_stat, cpu).irq_resched_count; - sum += per_cpu(irq_stat, cpu).irq_call_count; - sum += per_cpu(irq_stat, cpu).irq_tlb_count; -#endif -#ifdef CONFIG_X86_MCE - sum += per_cpu(irq_stat, cpu).irq_thermal_count; -#endif -#ifdef CONFIG_X86_LOCAL_APIC - sum += per_cpu(irq_stat, cpu).irq_spurious_count; -#endif - return sum; -} - -u64 arch_irq_stat(void) -{ - u64 sum = atomic_read(&irq_err_count); - -#ifdef CONFIG_X86_IO_APIC - sum += atomic_read(&irq_mis_count); -#endif - return sum; -} - #ifdef CONFIG_HOTPLUG_CPU #include diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 4f374294f292..39ef7feb9ea4 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -18,8 +18,6 @@ #include #include -atomic_t irq_err_count; - /* * 'what should we do if we get a hw irq event on an illegal vector'. * each architecture has to answer this themselves. @@ -64,136 +62,6 @@ static inline void stack_overflow_check(struct pt_regs *regs) } #endif -/* - * Generic, controller-independent functions: - */ - -static int show_other_interrupts(struct seq_file *p) -{ - int j; - - seq_printf(p, "NMI: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count); - seq_printf(p, " Non-maskable interrupts\n"); - seq_printf(p, "LOC: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->apic_timer_irqs); - seq_printf(p, " Local timer interrupts\n"); -#ifdef CONFIG_SMP - seq_printf(p, "RES: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_resched_count); - seq_printf(p, " Rescheduling interrupts\n"); - seq_printf(p, "CAL: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_call_count); - seq_printf(p, " Function call interrupts\n"); - seq_printf(p, "TLB: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_tlb_count); - seq_printf(p, " TLB shootdowns\n"); -#endif -#ifdef CONFIG_X86_MCE - seq_printf(p, "TRM: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_thermal_count); - seq_printf(p, " Thermal event interrupts\n"); - seq_printf(p, "THR: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_threshold_count); - seq_printf(p, " Threshold APIC interrupts\n"); -#endif - seq_printf(p, "SPU: "); - for_each_online_cpu(j) - seq_printf(p, "%10u ", cpu_pda(j)->irq_spurious_count); - seq_printf(p, " Spurious interrupts\n"); - seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count)); - - return 0; -} - -int show_interrupts(struct seq_file *p, void *v) -{ - unsigned long flags, any_count = 0; - int i = *(loff_t *) v, j; - struct irqaction *action; - struct irq_desc *desc; - - if (i > nr_irqs) - return 0; - - if (i == nr_irqs) - return show_other_interrupts(p); - - /* print header */ - if (i == 0) { - seq_printf(p, " "); - for_each_online_cpu(j) - seq_printf(p, "CPU%-8d",j); - seq_putc(p, '\n'); - } - - desc = irq_to_desc(i); - spin_lock_irqsave(&desc->lock, flags); -#ifndef CONFIG_SMP - any_count = kstat_irqs(i); -#else - for_each_online_cpu(j) - any_count |= kstat_irqs_cpu(i, j); -#endif - action = desc->action; - if (!action && !any_count) - goto out; - - seq_printf(p, "%3d: ", i); -#ifndef CONFIG_SMP - seq_printf(p, "%10u ", kstat_irqs(i)); -#else - for_each_online_cpu(j) - seq_printf(p, "%10u ", kstat_irqs_cpu(i, j)); -#endif - seq_printf(p, " %8s", desc->chip->name); - seq_printf(p, "-%-8s", desc->name); - - if (action) { - seq_printf(p, " %s", action->name); - while ((action = action->next) != NULL) - seq_printf(p, ", %s", action->name); - } - - seq_putc(p, '\n'); -out: - spin_unlock_irqrestore(&desc->lock, flags); - return 0; -} - -/* - * /proc/stat helpers - */ -u64 arch_irq_stat_cpu(unsigned int cpu) -{ - u64 sum = cpu_pda(cpu)->__nmi_count; - - sum += cpu_pda(cpu)->apic_timer_irqs; -#ifdef CONFIG_SMP - sum += cpu_pda(cpu)->irq_resched_count; - sum += cpu_pda(cpu)->irq_call_count; - sum += cpu_pda(cpu)->irq_tlb_count; -#endif -#ifdef CONFIG_X86_MCE - sum += cpu_pda(cpu)->irq_thermal_count; - sum += cpu_pda(cpu)->irq_threshold_count; -#endif - sum += cpu_pda(cpu)->irq_spurious_count; - return sum; -} - -u64 arch_irq_stat(void) -{ - return atomic_read(&irq_err_count); -} - /* * do_IRQ handles all normal device IRQ's (the special * SMP cross-CPU interrupts have their own specific -- cgit v1.2.3 From 249f6d9eab372790579ada8991bba3384c204e06 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 16 Oct 2008 12:18:50 +0200 Subject: x86: move ack_bad_irq() to irq.c Share more duplicated code. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/irq.c | 23 +++++++++++++++++++++++ arch/x86/kernel/irq_32.c | 23 ----------------------- arch/x86/kernel/irq_64.c | 20 -------------------- 3 files changed, 23 insertions(+), 43 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index 3b9128498976..ccf6c503fc3b 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -12,6 +12,29 @@ atomic_t irq_err_count; +/* + * 'what should we do if we get a hw irq event on an illegal vector'. + * each architecture has to answer this themselves. + */ +void ack_bad_irq(unsigned int irq) +{ + printk(KERN_ERR "unexpected IRQ trap at vector %02x\n", irq); + +#ifdef CONFIG_X86_LOCAL_APIC + /* + * Currently unexpected vectors happen only on SMP and APIC. + * We _must_ ack these because every local APIC has only N + * irq slots per priority level, and a 'hanging, unacked' IRQ + * holds up an irq slot - in excessive cases (when multiple + * unexpected vectors occur) that might lock up the APIC + * completely. + * But only ack when the APIC is enabled -AK + */ + if (cpu_has_apic) + ack_APIC_irq(); +#endif +} + #ifdef CONFIG_X86_32 # define irq_stats(x) (&per_cpu(irq_stat,x)) #else diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 6d9bf3936c78..a51382672de0 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -25,29 +25,6 @@ EXPORT_PER_CPU_SYMBOL(irq_stat); DEFINE_PER_CPU(struct pt_regs *, irq_regs); EXPORT_PER_CPU_SYMBOL(irq_regs); -/* - * 'what should we do if we get a hw irq event on an illegal vector'. - * each architecture has to answer this themselves. - */ -void ack_bad_irq(unsigned int irq) -{ - printk(KERN_ERR "unexpected IRQ trap at vector %02x\n", irq); - -#ifdef CONFIG_X86_LOCAL_APIC - /* - * Currently unexpected vectors happen only on SMP and APIC. - * We _must_ ack these because every local APIC has only N - * irq slots per priority level, and a 'hanging, unacked' IRQ - * holds up an irq slot - in excessive cases (when multiple - * unexpected vectors occur) that might lock up the APIC - * completely. - * But only ack when the APIC is enabled -AK - */ - if (cpu_has_apic) - ack_APIC_irq(); -#endif -} - #ifdef CONFIG_DEBUG_STACKOVERFLOW /* Debugging check for stack overflow: is there less than 1KB free? */ static int check_stack_overflow(void) diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 39ef7feb9ea4..60eb84eb77a0 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -18,26 +18,6 @@ #include #include -/* - * 'what should we do if we get a hw irq event on an illegal vector'. - * each architecture has to answer this themselves. - */ -void ack_bad_irq(unsigned int irq) -{ - printk(KERN_WARNING "unexpected IRQ trap at vector %02x\n", irq); - /* - * Currently unexpected vectors happen only on SMP and APIC. - * We _must_ ack these because every local APIC has only N - * irq slots per priority level, and a 'hanging, unacked' IRQ - * holds up an irq slot - in excessive cases (when multiple - * unexpected vectors occur) that might lock up the APIC - * completely. - * But don't ack when the APIC is disabled. -AK - */ - if (!disable_apic) - ack_APIC_irq(); -} - #ifdef CONFIG_DEBUG_STACKOVERFLOW /* * Probabilistic stack overflow check: -- cgit v1.2.3 From 10e0298686be6249b0665465737a6b83446c4d35 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 16 Oct 2008 17:04:22 +0200 Subject: io_apic: make irq_mis_count available on 64-bit too Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 6f80dc2f137e..b764d7429c61 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2277,9 +2277,7 @@ static void ack_apic_edge(unsigned int irq) ack_APIC_irq(); } -#ifdef CONFIG_X86_32 atomic_t irq_mis_count; -#endif static void ack_apic_level(unsigned int irq) { -- cgit v1.2.3 From ae87221d3ce49d9de1e43756da834fd0bf05a2ad Mon Sep 17 00:00:00 2001 From: Andrew Morton Date: Fri, 24 Aug 2007 16:11:54 -0700 Subject: sysfs: crash debugging Print the name of the last-accessed sysfs file when we oops, to help track down oopses which occur in sysfs store/read handlers. Because these oopses tend to not leave any trace of the offending code in the stack traces. Cc: Kay Sievers Cc: Mathieu Desnoyers Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman --- arch/x86/kernel/dumpstack_32.c | 2 ++ arch/x86/kernel/dumpstack_64.c | 2 ++ 2 files changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 201ee359a1a9..1a78180f08d3 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -13,6 +13,7 @@ #include #include #include +#include #include @@ -343,6 +344,7 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) printk("DEBUG_PAGEALLOC"); #endif printk("\n"); + sysfs_printk_last_file(); if (notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) return 1; diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 086cc8118e39..96a5db7da8a7 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -13,6 +13,7 @@ #include #include #include +#include #include @@ -489,6 +490,7 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) printk("DEBUG_PAGEALLOC"); #endif printk("\n"); + sysfs_printk_last_file(); if (notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) return 1; -- cgit v1.2.3 From a9b12619f7b6f19c871437ec24a088787a04b1de Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Mon, 21 Jul 2008 20:03:34 -0700 Subject: device create: misc: convert device_create_drvdata to device_create Now that device_create() has been audited, rename things back to the original call to be sane. Signed-off-by: Greg Kroah-Hartman --- arch/x86/kernel/cpuid.c | 4 ++-- arch/x86/kernel/msr.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 6a44d6465991..72cefd1e649b 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -147,8 +147,8 @@ static __cpuinit int cpuid_device_create(int cpu) { struct device *dev; - dev = device_create_drvdata(cpuid_class, NULL, MKDEV(CPUID_MAJOR, cpu), - NULL, "cpu%d", cpu); + dev = device_create(cpuid_class, NULL, MKDEV(CPUID_MAJOR, cpu), NULL, + "cpu%d", cpu); return IS_ERR(dev) ? PTR_ERR(dev) : 0; } diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 2e2af5d18191..82a7c7ed6d45 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -163,8 +163,8 @@ static int __cpuinit msr_device_create(int cpu) { struct device *dev; - dev = device_create_drvdata(msr_class, NULL, MKDEV(MSR_MAJOR, cpu), - NULL, "msr%d", cpu); + dev = device_create(msr_class, NULL, MKDEV(MSR_MAJOR, cpu), NULL, + "msr%d", cpu); return IS_ERR(dev) ? PTR_ERR(dev) : 0; } -- cgit v1.2.3 From 25ddbb18aae33ad255eb9f35aacebe3af01e1e9c Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 15 Oct 2008 22:01:41 -0700 Subject: Make the taint flags reliable It's somewhat unlikely that it happens, but right now a race window between interrupts or machine checks or oopses could corrupt the tainted bitmap because it is modified in a non atomic fashion. Convert the taint variable to an unsigned long and use only atomic bit operations on it. Unfortunately this means the intvec sysctl functions cannot be used on it anymore. It turned out the taint sysctl handler could actually be simplified a bit (since it only increases capabilities) so this patch actually removes code. [akpm@linux-foundation.org: remove unneeded include] Signed-off-by: Andi Kleen Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/smpboot.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 8c3aca7cb343..7ed9e070a6e9 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -282,6 +282,8 @@ static void __cpuinit smp_callin(void) cpu_set(cpuid, cpu_callin_map); } +static int __cpuinitdata unsafe_smp; + /* * Activate a secondary processor. */ @@ -397,7 +399,7 @@ static void __cpuinit smp_apply_quirks(struct cpuinfo_x86 *c) goto valid_k7; /* If we get here, not a certified SMP capable AMD system. */ - add_taint(TAINT_UNSAFE_SMP); + unsafe_smp = 1; } valid_k7: @@ -414,12 +416,10 @@ static void __cpuinit smp_checks(void) * Don't taint if we are running SMP kernel on a single non-MP * approved Athlon */ - if (tainted & TAINT_UNSAFE_SMP) { - if (num_online_cpus()) - printk(KERN_INFO "WARNING: This combination of AMD" - "processors is not suitable for SMP.\n"); - else - tainted &= ~TAINT_UNSAFE_SMP; + if (unsafe_smp && num_online_cpus() > 1) { + printk(KERN_INFO "WARNING: This combination of AMD" + "processors is not suitable for SMP.\n"); + add_taint(TAINT_UNSAFE_SMP); } } -- cgit v1.2.3 From bdab0ba3d9ad8de257ee6236daf314723748fde6 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 15 Oct 2008 22:02:07 -0700 Subject: x86: rename iommu_num_pages function to iommu_nr_pages This series of patches re-introduces the iommu_num_pages function so that it can be used by each architecture specific IOMMU implementations. The series also changes IOMMU implementations for X86, Alpha, PowerPC and UltraSparc. The other implementations are not yet changed because the modifications required are not obvious and I can't test them on real hardware. This patch: This is a preparation patch for introducing a generic iommu_num_pages function. Signed-off-by: Joerg Roedel Cc: "David S. Miller" Cc: Richard Henderson Cc: Ivan Kokshaysky Cc: Benjamin Herrenschmidt Cc: Paul Mackerras Cc: Ingo Molnar Cc: Thomas Gleixner Cc: FUJITA Tomonori Cc: Muli Ben-Yehuda Cc: Dave Airlie Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/amd_iommu.c | 8 ++++---- arch/x86/kernel/pci-dma.c | 4 ++-- arch/x86/kernel/pci-gart_64.c | 8 ++++---- 3 files changed, 10 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 34e4d112b1ef..10646acba9be 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -295,7 +295,7 @@ static int iommu_flush_pages(struct amd_iommu *iommu, u16 domid, u64 address, size_t size) { int s = 0; - unsigned pages = iommu_num_pages(address, size); + unsigned pages = iommu_nr_pages(address, size); address &= PAGE_MASK; @@ -679,7 +679,7 @@ static struct dma_ops_domain *dma_ops_domain_alloc(struct amd_iommu *iommu, if (iommu->exclusion_start && iommu->exclusion_start < dma_dom->aperture_size) { unsigned long startpage = iommu->exclusion_start >> PAGE_SHIFT; - int pages = iommu_num_pages(iommu->exclusion_start, + int pages = iommu_nr_pages(iommu->exclusion_start, iommu->exclusion_length); dma_ops_reserve_addresses(dma_dom, startpage, pages); } @@ -935,7 +935,7 @@ static dma_addr_t __map_single(struct device *dev, unsigned long align_mask = 0; int i; - pages = iommu_num_pages(paddr, size); + pages = iommu_nr_pages(paddr, size); paddr &= PAGE_MASK; if (align) @@ -980,7 +980,7 @@ static void __unmap_single(struct amd_iommu *iommu, if ((dma_addr == 0) || (dma_addr + size > dma_dom->aperture_size)) return; - pages = iommu_num_pages(dma_addr, size); + pages = iommu_nr_pages(dma_addr, size); dma_addr &= PAGE_MASK; start = dma_addr; diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 0a3824e837b4..192624820217 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -125,13 +125,13 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } -unsigned long iommu_num_pages(unsigned long addr, unsigned long len) +unsigned long iommu_nr_pages(unsigned long addr, unsigned long len) { unsigned long size = roundup((addr & ~PAGE_MASK) + len, PAGE_SIZE); return size >> PAGE_SHIFT; } -EXPORT_SYMBOL(iommu_num_pages); +EXPORT_SYMBOL(iommu_nr_pages); #endif void *dma_generic_alloc_coherent(struct device *dev, size_t size, diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index 145f1c83369f..14f1b41348fd 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -231,7 +231,7 @@ nonforced_iommu(struct device *dev, unsigned long addr, size_t size) static dma_addr_t dma_map_area(struct device *dev, dma_addr_t phys_mem, size_t size, int dir, unsigned long align_mask) { - unsigned long npages = iommu_num_pages(phys_mem, size); + unsigned long npages = iommu_nr_pages(phys_mem, size); unsigned long iommu_page = alloc_iommu(dev, npages, align_mask); int i; @@ -285,7 +285,7 @@ static void gart_unmap_single(struct device *dev, dma_addr_t dma_addr, return; iommu_page = (dma_addr - iommu_bus_base)>>PAGE_SHIFT; - npages = iommu_num_pages(dma_addr, size); + npages = iommu_nr_pages(dma_addr, size); for (i = 0; i < npages; i++) { iommu_gatt_base[iommu_page + i] = gart_unmapped_entry; CLEAR_LEAK(iommu_page + i); @@ -368,7 +368,7 @@ static int __dma_map_cont(struct device *dev, struct scatterlist *start, } addr = phys_addr; - pages = iommu_num_pages(s->offset, s->length); + pages = iommu_nr_pages(s->offset, s->length); while (pages--) { iommu_gatt_base[iommu_page] = GPTE_ENCODE(addr); SET_LEAK(iommu_page); @@ -451,7 +451,7 @@ gart_map_sg(struct device *dev, struct scatterlist *sg, int nents, int dir) seg_size += s->length; need = nextneed; - pages += iommu_num_pages(s->offset, s->length); + pages += iommu_nr_pages(s->offset, s->length); ps = s; } if (dma_map_cont(dev, start_sg, i - start, sgmap, pages, need) < 0) -- cgit v1.2.3 From 1477b8e5f13266bbf0389baafd39a90ff29c5f61 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 15 Oct 2008 22:02:11 -0700 Subject: x86: convert GART driver to generic iommu_num_pages function Signed-off-by: Joerg Roedel Cc: Ingo Molnar Cc: Thomas Gleixner Cc: FUJITA Tomonori Cc: Muli Ben-Yehuda Cc: Dave Airlie Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/pci-gart_64.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index 14f1b41348fd..e3f75bbcedea 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -231,7 +231,7 @@ nonforced_iommu(struct device *dev, unsigned long addr, size_t size) static dma_addr_t dma_map_area(struct device *dev, dma_addr_t phys_mem, size_t size, int dir, unsigned long align_mask) { - unsigned long npages = iommu_nr_pages(phys_mem, size); + unsigned long npages = iommu_num_pages(phys_mem, size, PAGE_SIZE); unsigned long iommu_page = alloc_iommu(dev, npages, align_mask); int i; @@ -285,7 +285,7 @@ static void gart_unmap_single(struct device *dev, dma_addr_t dma_addr, return; iommu_page = (dma_addr - iommu_bus_base)>>PAGE_SHIFT; - npages = iommu_nr_pages(dma_addr, size); + npages = iommu_num_pages(dma_addr, size, PAGE_SIZE); for (i = 0; i < npages; i++) { iommu_gatt_base[iommu_page + i] = gart_unmapped_entry; CLEAR_LEAK(iommu_page + i); @@ -368,7 +368,7 @@ static int __dma_map_cont(struct device *dev, struct scatterlist *start, } addr = phys_addr; - pages = iommu_nr_pages(s->offset, s->length); + pages = iommu_num_pages(s->offset, s->length, PAGE_SIZE); while (pages--) { iommu_gatt_base[iommu_page] = GPTE_ENCODE(addr); SET_LEAK(iommu_page); @@ -451,7 +451,7 @@ gart_map_sg(struct device *dev, struct scatterlist *sg, int nents, int dir) seg_size += s->length; need = nextneed; - pages += iommu_nr_pages(s->offset, s->length); + pages += iommu_num_pages(s->offset, s->length, PAGE_SIZE); ps = s; } if (dma_map_cont(dev, start_sg, i - start, sgmap, pages, need) < 0) -- cgit v1.2.3 From e3c449f526cebb8d287241c7e82faafd9709668b Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 15 Oct 2008 22:02:11 -0700 Subject: x86, AMD IOMMU: convert driver to generic iommu_num_pages function Signed-off-by: Joerg Roedel Cc: Ingo Molnar Cc: Thomas Gleixner Cc: FUJITA Tomonori Cc: Muli Ben-Yehuda Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/amd_iommu.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 10646acba9be..a8fd9ebdc8e2 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -295,7 +295,7 @@ static int iommu_flush_pages(struct amd_iommu *iommu, u16 domid, u64 address, size_t size) { int s = 0; - unsigned pages = iommu_nr_pages(address, size); + unsigned pages = iommu_num_pages(address, size, PAGE_SIZE); address &= PAGE_MASK; @@ -679,8 +679,9 @@ static struct dma_ops_domain *dma_ops_domain_alloc(struct amd_iommu *iommu, if (iommu->exclusion_start && iommu->exclusion_start < dma_dom->aperture_size) { unsigned long startpage = iommu->exclusion_start >> PAGE_SHIFT; - int pages = iommu_nr_pages(iommu->exclusion_start, - iommu->exclusion_length); + int pages = iommu_num_pages(iommu->exclusion_start, + iommu->exclusion_length, + PAGE_SIZE); dma_ops_reserve_addresses(dma_dom, startpage, pages); } @@ -935,7 +936,7 @@ static dma_addr_t __map_single(struct device *dev, unsigned long align_mask = 0; int i; - pages = iommu_nr_pages(paddr, size); + pages = iommu_num_pages(paddr, size, PAGE_SIZE); paddr &= PAGE_MASK; if (align) @@ -980,7 +981,7 @@ static void __unmap_single(struct amd_iommu *iommu, if ((dma_addr == 0) || (dma_addr + size > dma_dom->aperture_size)) return; - pages = iommu_nr_pages(dma_addr, size); + pages = iommu_num_pages(dma_addr, size, PAGE_SIZE); dma_addr &= PAGE_MASK; start = dma_addr; -- cgit v1.2.3 From 036b4c50fe99a2f308f36561335b9904ab507972 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 15 Oct 2008 22:02:12 -0700 Subject: x86: convert Calgary IOMMU driver to generic iommu_num_pages function Signed-off-by: Joerg Roedel Cc: Ingo Molnar Cc: Thomas Gleixner Cc: FUJITA Tomonori Acked-by: Muli Ben-Yehuda Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/pci-calgary_64.c | 18 ++++-------------- 1 file changed, 4 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 080d1d27f37a..e1e731d78f38 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -217,16 +217,6 @@ static inline unsigned long verify_bit_range(unsigned long* bitmap, #endif /* CONFIG_IOMMU_DEBUG */ -static inline unsigned int num_dma_pages(unsigned long dma, unsigned int dmalen) -{ - unsigned int npages; - - npages = PAGE_ALIGN(dma + dmalen) - (dma & PAGE_MASK); - npages >>= PAGE_SHIFT; - - return npages; -} - static inline int translation_enabled(struct iommu_table *tbl) { /* only PHBs with translation enabled have an IOMMU table */ @@ -408,7 +398,7 @@ static void calgary_unmap_sg(struct device *dev, if (dmalen == 0) break; - npages = num_dma_pages(dma, dmalen); + npages = iommu_num_pages(dma, dmalen, PAGE_SIZE); iommu_free(tbl, dma, npages); } } @@ -427,7 +417,7 @@ static int calgary_map_sg(struct device *dev, struct scatterlist *sg, BUG_ON(!sg_page(s)); vaddr = (unsigned long) sg_virt(s); - npages = num_dma_pages(vaddr, s->length); + npages = iommu_num_pages(vaddr, s->length, PAGE_SIZE); entry = iommu_range_alloc(dev, tbl, npages); if (entry == bad_dma_address) { @@ -464,7 +454,7 @@ static dma_addr_t calgary_map_single(struct device *dev, phys_addr_t paddr, struct iommu_table *tbl = find_iommu_table(dev); uaddr = (unsigned long)vaddr; - npages = num_dma_pages(uaddr, size); + npages = iommu_num_pages(uaddr, size, PAGE_SIZE); return iommu_alloc(dev, tbl, vaddr, npages, direction); } @@ -475,7 +465,7 @@ static void calgary_unmap_single(struct device *dev, dma_addr_t dma_handle, struct iommu_table *tbl = find_iommu_table(dev); unsigned int npages; - npages = num_dma_pages(dma_handle, size); + npages = iommu_num_pages(dma_handle, size, PAGE_SIZE); iommu_free(tbl, dma_handle, npages); } -- cgit v1.2.3 From 26adcfbf00e0726b4469070aa2f530dcf963f484 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 14 Oct 2008 21:01:15 +0200 Subject: x86: SB600: skip ACPI IRQ0 override if it is not routed to INT2 of IOAPIC On some more HP laptops BIOS reports an IRQ0 override but the SB600 chipset is configured such that timer interrupts go to INT0 of IOAPIC. Check IRQ0 routing and if it is routed to INT0 of IOAPIC skip the timer override. http://bugzilla.kernel.org/show_bug.cgi?id=11715 http://bugzilla.kernel.org/show_bug.cgi?id=11516 Signed-off-by: Andreas Herrmann Signed-off-by: Len Brown --- arch/x86/kernel/early-quirks.c | 55 +++++++++++++++++++++++++++++++++++++++--- 1 file changed, 52 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 733c4f8d42ea..3ce029ffaa55 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -95,7 +95,8 @@ static void __init nvidia_bugs(int num, int slot, int func) } -static u32 ati_ixp4x0_rev(int num, int slot, int func) +#if defined(CONFIG_ACPI) && defined(CONFIG_X86_IO_APIC) +static u32 __init ati_ixp4x0_rev(int num, int slot, int func) { u32 d; u8 b; @@ -115,7 +116,6 @@ static u32 ati_ixp4x0_rev(int num, int slot, int func) static void __init ati_bugs(int num, int slot, int func) { -#if defined(CONFIG_ACPI) && defined (CONFIG_X86_IO_APIC) u32 d; u8 b; @@ -138,9 +138,56 @@ static void __init ati_bugs(int num, int slot, int func) printk(KERN_INFO "If you got timer trouble " "try acpi_use_timer_override\n"); } -#endif } +static u32 __init ati_sbx00_rev(int num, int slot, int func) +{ + u32 old, d; + + d = read_pci_config(num, slot, func, 0x70); + old = d; + d &= ~(1<<8); + write_pci_config(num, slot, func, 0x70, d); + d = read_pci_config(num, slot, func, 0x8); + d &= 0xff; + write_pci_config(num, slot, func, 0x70, old); + + return d; +} + +static void __init ati_bugs_contd(int num, int slot, int func) +{ + u32 d, rev; + + if (acpi_use_timer_override) + return; + + rev = ati_sbx00_rev(num, slot, func); + if (rev > 0x13) + return; + + /* check for IRQ0 interrupt swap */ + d = read_pci_config(num, slot, func, 0x64); + if (!(d & (1<<14))) + acpi_skip_timer_override = 1; + + if (acpi_skip_timer_override) { + printk(KERN_INFO "SB600 revision 0x%x\n", rev); + printk(KERN_INFO "Ignoring ACPI timer override.\n"); + printk(KERN_INFO "If you got timer trouble " + "try acpi_use_timer_override\n"); + } +} +#else +static void __init ati_bugs(int num, int slot, int func) +{ +} + +static void __init ati_bugs_contd(int num, int slot, int func) +{ +} +#endif + #ifdef CONFIG_DMAR static void __init intel_g33_dmar(int num, int slot, int func) { @@ -176,6 +223,8 @@ static struct chipset early_qrk[] __initdata = { PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, fix_hypertransport_config }, { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP400_SMBUS, PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, + { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, + PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, #ifdef CONFIG_DMAR { PCI_VENDOR_ID_INTEL, 0x29c0, PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, intel_g33_dmar }, -- cgit v1.2.3 From 3038edabf48f01421c621cb77a712b446d3a5d67 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Fri, 17 Oct 2008 01:26:27 +0200 Subject: x86 ACPI: fix breakage of resume on 64-bit UP systems with SMP kernel x86 ACPI: Fix breakage of resume on 64-bit UP systems with SMP kernel We are now using per CPU GDT tables in head_64.S and the original early_gdt_descr.address is invalidated after boot by setup_per_cpu_areas(). This breaks resume from suspend to RAM on x86_64 UP systems using SMP kernels, because this part of head_64.S is also executed during the resume and the invalid GDT address causes the system to crash. It doesn't break on 'true' SMP systems, because early_gdt_descr.address is modified every time native_cpu_up() runs. However, during resume it should point to the GDT of the boot CPU rather than to another CPU's GDT. For this reason, during suspend to RAM always make early_gdt_descr.address point to the boot CPU's GDT. This fixes http://bugzilla.kernel.org/show_bug.cgi?id=11568, which is a regression from 2.6.26. Signed-off-by: Rafael J. Wysocki Acked-by: Pavel Machek Cc: Signed-off-by: Ingo Molnar Reported-and-tested-by: Andy Wettstein --- arch/x86/kernel/acpi/sleep.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 426e5d91b63a..c44cd6dbfa14 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -10,6 +10,7 @@ #include #include #include +#include #include "realmode/wakeup.h" #include "sleep.h" @@ -98,6 +99,8 @@ int acpi_save_state_mem(void) header->trampoline_segment = setup_trampoline() >> 4; #ifdef CONFIG_SMP stack_start.sp = temp_stack + 4096; + early_gdt_descr.address = + (unsigned long)get_cpu_gdt_table(smp_processor_id()); #endif initial_code = (unsigned long)wakeup_long64; saved_magic = 0x123456789abcdef0; -- cgit v1.2.3 From 5b6985ce8ec7127b4d60ad450b64ca8b82748a3b Mon Sep 17 00:00:00 2001 From: Fenghua Yu Date: Thu, 16 Oct 2008 18:02:32 -0700 Subject: intel-iommu: IA64 support The current Intel IOMMU code assumes that both host page size and Intel IOMMU page size are 4KiB. The first patch supports variable page size. This provides support for IA64 which has multiple page sizes. This patch also adds some other code hooks for IA64 platform including DMAR_OPERATION_TIMEOUT definition. [dwmw2: some cleanup] Signed-off-by: Fenghua Yu Signed-off-by: Tony Luck Signed-off-by: David Woodhouse --- arch/x86/kernel/pci-dma.c | 16 ---------------- 1 file changed, 16 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 192624820217..1972266e8ba5 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -9,8 +9,6 @@ #include #include -static int forbid_dac __read_mostly; - struct dma_mapping_ops *dma_ops; EXPORT_SYMBOL(dma_ops); @@ -293,17 +291,3 @@ void pci_iommu_shutdown(void) } /* Must execute after PCI subsystem */ fs_initcall(pci_iommu_init); - -#ifdef CONFIG_PCI -/* Many VIA bridges seem to corrupt data for DAC. Disable it here */ - -static __devinit void via_no_dac(struct pci_dev *dev) -{ - if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI && forbid_dac == 0) { - printk(KERN_INFO "PCI: VIA PCI bridge detected." - "Disabling DAC.\n"); - forbid_dac = 1; - } -} -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_VIA, PCI_ANY_ID, via_no_dac); -#endif -- cgit v1.2.3 From f609891f428e1c20e270e7c350daf8c93cc459d7 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Thu, 16 Oct 2008 16:27:36 +0200 Subject: amd_iommu: fix nasty bug that caused ILLEGAL_DEVICE_TABLE_ENTRY errors We are on 64-bit so better use u64 instead of u32 to deal with addresses: static void __init iommu_set_device_table(struct amd_iommu *iommu) { u64 entry; ... entry = virt_to_phys(amd_iommu_dev_table); ... (I am wondering why gcc 4.2.x did not warn about the assignment between u32 and unsigned long.) Cc: iommu@lists.linux-foundation.org Cc: stable@kernel.org Signed-off-by: Andreas Herrmann Signed-off-by: Joerg Roedel Signed-off-by: David Woodhouse --- arch/x86/kernel/amd_iommu_init.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 4cd8083c58be..0cdcda35a05f 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -212,7 +212,7 @@ static void __init iommu_set_exclusion_range(struct amd_iommu *iommu) /* Programs the physical address of the device table into the IOMMU hardware */ static void __init iommu_set_device_table(struct amd_iommu *iommu) { - u32 entry; + u64 entry; BUG_ON(iommu->mmio_base == NULL); -- cgit v1.2.3 From 57cac4d1880527e0baf6c2fda529d2ad1d815aec Mon Sep 17 00:00:00 2001 From: Vivek Goyal Date: Sat, 18 Oct 2008 20:28:25 -0700 Subject: kdump: make elfcorehdr_addr independent of CONFIG_PROC_VMCORE o elfcorehdr_addr is used by not only the code under CONFIG_PROC_VMCORE but also by the code which is not inside CONFIG_PROC_VMCORE. For example, is_kdump_kernel() is used by powerpc code to determine if kernel is booting after a panic then use previous kernel's TCE table. So even if CONFIG_PROC_VMCORE is not set in second kernel, one should be able to correctly determine that we are booting after a panic and setup calgary iommu accordingly. o So remove the assumption that elfcorehdr_addr is under CONFIG_PROC_VMCORE. o Move definition of elfcorehdr_addr to arch dependent crash files. (Unfortunately crash dump does not have an arch independent file otherwise that would have been the best place). o kexec.c is not the right place as one can Have CRASH_DUMP enabled in second kernel without KEXEC being enabled. o I don't see sh setup code parsing the command line for elfcorehdr_addr. I am wondering how does vmcore interface work on sh. Anyway, I am atleast defining elfcoredhr_addr so that compilation is not broken on sh. Signed-off-by: Vivek Goyal Acked-by: "Eric W. Biederman" Acked-by: Simon Horman Acked-by: Paul Mundt Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/crash_dump_32.c | 3 +++ arch/x86/kernel/crash_dump_64.c | 3 +++ arch/x86/kernel/setup.c | 8 +++++++- 3 files changed, 13 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash_dump_32.c b/arch/x86/kernel/crash_dump_32.c index 72d0c56c1b48..f7cdb3b457aa 100644 --- a/arch/x86/kernel/crash_dump_32.c +++ b/arch/x86/kernel/crash_dump_32.c @@ -13,6 +13,9 @@ static void *kdump_buf_page; +/* Stores the physical address of elf header of crash image. */ +unsigned long long elfcorehdr_addr = ELFCORE_ADDR_MAX; + /** * copy_oldmem_page - copy one page from "oldmem" * @pfn: page frame number to be copied diff --git a/arch/x86/kernel/crash_dump_64.c b/arch/x86/kernel/crash_dump_64.c index e90a60ef10c2..045b36cada65 100644 --- a/arch/x86/kernel/crash_dump_64.c +++ b/arch/x86/kernel/crash_dump_64.c @@ -10,6 +10,9 @@ #include #include +/* Stores the physical address of elf header of crash image. */ +unsigned long long elfcorehdr_addr = ELFCORE_ADDR_MAX; + /** * copy_oldmem_page - copy one page from "oldmem" * @pfn: page frame number to be copied diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 2255782e8d4b..b2c97874ec0f 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -561,7 +561,13 @@ static void __init reserve_standard_io_resources(void) } -#ifdef CONFIG_PROC_VMCORE +/* + * Note: elfcorehdr_addr is not just limited to vmcore. It is also used by + * is_kdump_kernel() to determine if we are booting after a panic. Hence + * ifdef it under CONFIG_CRASH_DUMP and not CONFIG_PROC_VMCORE. + */ + +#ifdef CONFIG_CRASH_DUMP /* elfcorehdr= specifies the location of elf core header * stored by the crashed kernel. This option will be passed * by kexec loader to the capture kernel. -- cgit v1.2.3 From 357c6e63590895dc87cc9300f5a1c27544ea69e8 Mon Sep 17 00:00:00 2001 From: Adrian Bunk Date: Sat, 18 Oct 2008 20:28:42 -0700 Subject: rtc: use bcd2bin/bin2bcd Change various rtc related code to use the new bcd2bin/bin2bcd functions instead of the obsolete BCD_TO_BIN/BIN_TO_BCD/BCD2BIN/BIN2BCD macros. Signed-off-by: Adrian Bunk Acked-by: Alessandro Zummo Cc: David Brownell Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/rtc.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/rtc.c b/arch/x86/kernel/rtc.c index 0a23b5795b25..dd6f2b71561b 100644 --- a/arch/x86/kernel/rtc.c +++ b/arch/x86/kernel/rtc.c @@ -52,7 +52,7 @@ int mach_set_rtc_mmss(unsigned long nowtime) cmos_minutes = CMOS_READ(RTC_MINUTES); if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) - BCD_TO_BIN(cmos_minutes); + cmos_minutes = bcd2bin(cmos_minutes); /* * since we're only adjusting minutes and seconds, @@ -69,8 +69,8 @@ int mach_set_rtc_mmss(unsigned long nowtime) if (abs(real_minutes - cmos_minutes) < 30) { if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) { - BIN_TO_BCD(real_seconds); - BIN_TO_BCD(real_minutes); + real_seconds = bin2bcd(real_seconds); + real_minutes = bin2bcd(real_minutes); } CMOS_WRITE(real_seconds,RTC_SECONDS); CMOS_WRITE(real_minutes,RTC_MINUTES); @@ -124,16 +124,16 @@ unsigned long mach_get_cmos_time(void) WARN_ON_ONCE(RTC_ALWAYS_BCD && (status & RTC_DM_BINARY)); if (RTC_ALWAYS_BCD || !(status & RTC_DM_BINARY)) { - BCD_TO_BIN(sec); - BCD_TO_BIN(min); - BCD_TO_BIN(hour); - BCD_TO_BIN(day); - BCD_TO_BIN(mon); - BCD_TO_BIN(year); + sec = bcd2bin(sec); + min = bcd2bin(min); + hour = bcd2bin(hour); + day = bcd2bin(day); + mon = bcd2bin(mon); + year = bcd2bin(year); } if (century) { - BCD_TO_BIN(century); + century = bcd2bin(century); year += century * 100; printk(KERN_INFO "Extended CMOS year: %d\n", century * 100); } else -- cgit v1.2.3 From c513867561eeb07d24a0bdda1a18a8f91921a301 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 14 Oct 2008 18:08:48 -0400 Subject: ftrace: do not enclose logic in WARN_ON In ftrace, logic is defined in the WARN_ON_ONCE, which can become a nop with some configs. This patch fixes it. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index d073d981a730..8821ceabf51d 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -62,6 +62,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { unsigned char replaced[MCOUNT_INSN_SIZE]; + int ret; /* * Note: Due to modules and __init, code can @@ -77,8 +78,9 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) return 2; - WARN_ON_ONCE(__copy_to_user_inatomic((char __user *)ip, new_code, - MCOUNT_INSN_SIZE)); + ret = __copy_to_user_inatomic((char __user *)ip, new_code, + MCOUNT_INSN_SIZE); + WARN_ON_ONCE(ret); sync_core(); -- cgit v1.2.3 From 606576ce816603d9fe1fb453a88bc6eea16ca709 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 6 Oct 2008 19:06:12 -0400 Subject: ftrace: rename FTRACE to FUNCTION_TRACER Due to confusion between the ftrace infrastructure and the gcc profiling tracer "ftrace", this patch renames the config options from FTRACE to FUNCTION_TRACER. The other two names that are offspring from FTRACE DYNAMIC_FTRACE and FTRACE_MCOUNT_RECORD will stay the same. This patch was generated mostly by script, and partially by hand. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/entry_32.S | 4 ++-- arch/x86/kernel/entry_64.S | 4 ++-- arch/x86/kernel/i386_ksyms_32.c | 2 +- arch/x86/kernel/x8664_ksyms_64.c | 2 +- 5 files changed, 7 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 0d41f0343dc0..ec3d30136bf0 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -6,7 +6,7 @@ extra-y := head_$(BITS).o head$(BITS).o head.o init_task.o vmlinu CPPFLAGS_vmlinux.lds += -U$(UTS_MACHINE) -ifdef CONFIG_FTRACE +ifdef CONFIG_FUNCTION_TRACER # Do not profile debug and lowlevel utilities CFLAGS_REMOVE_tsc.o = -pg CFLAGS_REMOVE_rtc.o = -pg diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 4e4269c73bb7..9d49facc21f2 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1149,7 +1149,7 @@ ENDPROC(xen_failsafe_callback) #endif /* CONFIG_XEN */ -#ifdef CONFIG_FTRACE +#ifdef CONFIG_FUNCTION_TRACER #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) @@ -1204,7 +1204,7 @@ trace: jmp ftrace_stub END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ -#endif /* CONFIG_FTRACE */ +#endif /* CONFIG_FUNCTION_TRACER */ .section .rodata,"a" #include "syscall_table_32.S" diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 09e7145484c5..b86f332c96a6 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -61,7 +61,7 @@ .code64 -#ifdef CONFIG_FTRACE +#ifdef CONFIG_FUNCTION_TRACER #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) retq @@ -138,7 +138,7 @@ trace: jmp ftrace_stub END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ -#endif /* CONFIG_FTRACE */ +#endif /* CONFIG_FUNCTION_TRACER */ #ifndef CONFIG_PREEMPT #define retint_kernel retint_restore_args diff --git a/arch/x86/kernel/i386_ksyms_32.c b/arch/x86/kernel/i386_ksyms_32.c index dd7ebee446af..43cec6bdda63 100644 --- a/arch/x86/kernel/i386_ksyms_32.c +++ b/arch/x86/kernel/i386_ksyms_32.c @@ -5,7 +5,7 @@ #include #include -#ifdef CONFIG_FTRACE +#ifdef CONFIG_FUNCTION_TRACER /* mcount is defined in assembly */ EXPORT_SYMBOL(mcount); #endif diff --git a/arch/x86/kernel/x8664_ksyms_64.c b/arch/x86/kernel/x8664_ksyms_64.c index b545f371b5f5..695e426aa354 100644 --- a/arch/x86/kernel/x8664_ksyms_64.c +++ b/arch/x86/kernel/x8664_ksyms_64.c @@ -12,7 +12,7 @@ #include #include -#ifdef CONFIG_FTRACE +#ifdef CONFIG_FUNCTION_TRACER /* mcount is defined in assembly */ EXPORT_SYMBOL(mcount); #endif -- cgit v1.2.3 From db7a6d8d01b21829d28638258cbbc9553210bac1 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Mon, 20 Oct 2008 11:24:31 -0700 Subject: Update .gitignore files for generated targets The generated 'capflags.c' file wasn't properly ignored, and the list of files in scripts/basic/ wasn't up-to-date. Signed-off-by: Linus Torvalds --- arch/x86/kernel/cpu/.gitignore | 1 + 1 file changed, 1 insertion(+) create mode 100644 arch/x86/kernel/cpu/.gitignore (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/.gitignore b/arch/x86/kernel/cpu/.gitignore new file mode 100644 index 000000000000..667df55a4399 --- /dev/null +++ b/arch/x86/kernel/cpu/.gitignore @@ -0,0 +1 @@ +capflags.c -- cgit v1.2.3 From f4432c5caec5fa95ea7eefd00f8e6cee17e2e023 Mon Sep 17 00:00:00 2001 From: Dave Jones Date: Mon, 20 Oct 2008 13:31:45 -0400 Subject: Update email addresses. Update assorted email addresses and related info to point to a single current, valid address. additionally - trivial CREDITS entry updates. (Not that this file means much any more) - remove arjans dead redhat.com address from powernow driver Signed-off-by: Dave Jones Signed-off-by: Linus Torvalds --- arch/x86/kernel/cpu/cpufreq/longhaul.c | 4 ++-- arch/x86/kernel/cpu/cpufreq/powernow-k6.c | 2 +- arch/x86/kernel/cpu/cpufreq/powernow-k7.c | 4 ++-- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 2 +- arch/x86/kernel/cpu/cpufreq/speedstep-ich.c | 2 +- arch/x86/kernel/cpu/mcheck/k7.c | 4 ++-- arch/x86/kernel/cpu/mcheck/mce_32.c | 2 +- arch/x86/kernel/cpu/mcheck/non-fatal.c | 2 +- 8 files changed, 11 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/longhaul.c b/arch/x86/kernel/cpu/cpufreq/longhaul.c index 06fcce516d51..b0461856acfb 100644 --- a/arch/x86/kernel/cpu/cpufreq/longhaul.c +++ b/arch/x86/kernel/cpu/cpufreq/longhaul.c @@ -1,5 +1,5 @@ /* - * (C) 2001-2004 Dave Jones. + * (C) 2001-2004 Dave Jones. * (C) 2002 Padraig Brady. * * Licensed under the terms of the GNU GPL License version 2. @@ -1019,7 +1019,7 @@ MODULE_PARM_DESC(scale_voltage, "Scale voltage of processor"); module_param(revid_errata, int, 0644); MODULE_PARM_DESC(revid_errata, "Ignore CPU Revision ID"); -MODULE_AUTHOR ("Dave Jones "); +MODULE_AUTHOR ("Dave Jones "); MODULE_DESCRIPTION ("Longhaul driver for VIA Cyrix processors."); MODULE_LICENSE ("GPL"); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c index b5ced806a316..c1ac5790c63e 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c @@ -246,7 +246,7 @@ static void __exit powernow_k6_exit(void) } -MODULE_AUTHOR("Arjan van de Ven , Dave Jones , Dominik Brodowski "); +MODULE_AUTHOR("Arjan van de Ven, Dave Jones , Dominik Brodowski "); MODULE_DESCRIPTION("PowerNow! driver for AMD K6-2+ / K6-3+ processors."); MODULE_LICENSE("GPL"); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c index 0a61159d7b71..7c7d56b43136 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c @@ -1,6 +1,6 @@ /* * AMD K7 Powernow driver. - * (C) 2003 Dave Jones on behalf of SuSE Labs. + * (C) 2003 Dave Jones on behalf of SuSE Labs. * (C) 2003-2004 Dave Jones * * Licensed under the terms of the GNU GPL License version 2. @@ -692,7 +692,7 @@ static void __exit powernow_exit (void) module_param(acpi_force, int, 0444); MODULE_PARM_DESC(acpi_force, "Force ACPI to be used."); -MODULE_AUTHOR ("Dave Jones "); +MODULE_AUTHOR ("Dave Jones "); MODULE_DESCRIPTION ("Powernow driver for AMD K7 processors."); MODULE_LICENSE ("GPL"); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 84bb395038d8..008d23ba491b 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -7,7 +7,7 @@ * Support : mark.langsdorf@amd.com * * Based on the powernow-k7.c module written by Dave Jones. - * (C) 2003 Dave Jones on behalf of SuSE Labs + * (C) 2003 Dave Jones on behalf of SuSE Labs * (C) 2004 Dominik Brodowski * (C) 2004 Pavel Machek * Licensed under the terms of the GNU GPL License version 2. diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c index 191f7263c61d..04d0376b64b0 100644 --- a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c +++ b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c @@ -431,7 +431,7 @@ static void __exit speedstep_exit(void) } -MODULE_AUTHOR ("Dave Jones , Dominik Brodowski "); +MODULE_AUTHOR ("Dave Jones , Dominik Brodowski "); MODULE_DESCRIPTION ("Speedstep driver for Intel mobile processors on chipsets with ICH-M southbridges."); MODULE_LICENSE ("GPL"); diff --git a/arch/x86/kernel/cpu/mcheck/k7.c b/arch/x86/kernel/cpu/mcheck/k7.c index f390c9f66351..dd3af6e7b39a 100644 --- a/arch/x86/kernel/cpu/mcheck/k7.c +++ b/arch/x86/kernel/cpu/mcheck/k7.c @@ -1,6 +1,6 @@ /* - * Athlon/Hammer specific Machine Check Exception Reporting - * (C) Copyright 2002 Dave Jones + * Athlon specific Machine Check Exception Reporting + * (C) Copyright 2002 Dave Jones */ #include diff --git a/arch/x86/kernel/cpu/mcheck/mce_32.c b/arch/x86/kernel/cpu/mcheck/mce_32.c index 774d87cfd8cd..0ebf3fc6a610 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_32.c +++ b/arch/x86/kernel/cpu/mcheck/mce_32.c @@ -1,6 +1,6 @@ /* * mce.c - x86 Machine Check Exception Reporting - * (c) 2002 Alan Cox , Dave Jones + * (c) 2002 Alan Cox , Dave Jones */ #include diff --git a/arch/x86/kernel/cpu/mcheck/non-fatal.c b/arch/x86/kernel/cpu/mcheck/non-fatal.c index cc1fccdd31e0..a74af128efc9 100644 --- a/arch/x86/kernel/cpu/mcheck/non-fatal.c +++ b/arch/x86/kernel/cpu/mcheck/non-fatal.c @@ -1,7 +1,7 @@ /* * Non Fatal Machine Check Exception Reporting * - * (C) Copyright 2002 Dave Jones. + * (C) Copyright 2002 Dave Jones. * * This file contains routines to check for non-fatal MCEs every 15s * -- cgit v1.2.3 From e9f95e637320efe1936b647308ddf4ec5b8e0311 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 21 Oct 2008 15:49:59 +0200 Subject: genirq: fix off by one and coding style Fix off-by-one in for_each_irq_desc_reverse(). Impact is near zero in practice, because nothing substantial wants to iterate down to IRQ#0 - but fix it nevertheless. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- arch/x86/kernel/irq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index ccf6c503fc3b..d1d4dc52f649 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -36,7 +36,7 @@ void ack_bad_irq(unsigned int irq) } #ifdef CONFIG_X86_32 -# define irq_stats(x) (&per_cpu(irq_stat,x)) +# define irq_stats(x) (&per_cpu(irq_stat, x)) #else # define irq_stats(x) cpu_pda(x) #endif @@ -113,7 +113,7 @@ int show_interrupts(struct seq_file *p, void *v) if (i == 0) { seq_printf(p, " "); for_each_online_cpu(j) - seq_printf(p, "CPU%-8d",j); + seq_printf(p, "CPU%-8d", j); seq_putc(p, '\n'); } -- cgit v1.2.3 From c7d87d79d14cecab7a34dedf1b133377cf5a0203 Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Thu, 16 Oct 2008 16:34:43 -0400 Subject: x86 allow modules to register idle notifiers needed if the i7300_idle driver is to be modular. Signed-off-by: Venkatesh Pallipadi Acked-by: Ingo Molnar Signed-off-by: Len Brown --- arch/x86/kernel/process_64.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index e12e0e4dd256..3e3d503eadcf 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -62,6 +62,13 @@ void idle_notifier_register(struct notifier_block *n) { atomic_notifier_chain_register(&idle_notifier, n); } +EXPORT_SYMBOL_GPL(idle_notifier_register); + +void idle_notifier_unregister(struct notifier_block *n) +{ + atomic_notifier_chain_unregister(&idle_notifier, n); +} +EXPORT_SYMBOL_GPL(idle_notifier_unregister); void enter_idle(void) { -- cgit v1.2.3 From 8bcad30f2e6d4c20f7e71d2e2ac77acc0f0931e5 Mon Sep 17 00:00:00 2001 From: roel kluin Date: Tue, 21 Oct 2008 19:49:09 -0400 Subject: x86: make variables static These variables are only used in their source files, so make them static. Signed-off-by: Roel Kluin Signed-off-by: Ingo Molnar --- arch/x86/kernel/xsave.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c index 9abac8a9d823..b13acb75e822 100644 --- a/arch/x86/kernel/xsave.c +++ b/arch/x86/kernel/xsave.c @@ -248,7 +248,7 @@ clear: * This will be saved when ever the FP and extended state context is * saved on the user stack during the signal handler delivery to the user. */ -void prepare_fx_sw_frame(void) +static void prepare_fx_sw_frame(void) { int size_extended = (xstate_size - sizeof(struct i387_fxsave_struct)) + FP_XSTATE_MAGIC2_SIZE; -- cgit v1.2.3 From b0f209898f1a177bd503d49215b8c6628797a81c Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Tue, 21 Oct 2008 14:09:51 -0500 Subject: x86, uv: use consistent names for region size and conherence id on x86 and ia64 Use consistent names for region size and conherence id on x86 and ia64. The SGI xp drivers are used on both ia64 and x86. Using the same names (sn_coherency_id, sn_region_size) simplies the driver code. Signed-off-by: Russ Anderson Signed-off-by: Ingo Molnar --- arch/x86/kernel/bios_uv.c | 8 ++++---- arch/x86/kernel/genx2apic_uv_x.c | 4 ++-- 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index f0dfe6f17e7e..7cefb7170e75 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -69,10 +69,10 @@ s64 uv_bios_call_reentrant(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, long sn_partition_id; EXPORT_SYMBOL_GPL(sn_partition_id); -long uv_coherency_id; -EXPORT_SYMBOL_GPL(uv_coherency_id); -long uv_region_size; -EXPORT_SYMBOL_GPL(uv_region_size); +long sn_coherency_id; +EXPORT_SYMBOL_GPL(sn_coherency_id); +long sn_region_size; +EXPORT_SYMBOL_GPL(sn_region_size); int uv_type; diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index bfd532843df6..6cf35c8bd636 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -429,7 +429,7 @@ void __init uv_system_init(void) uv_bios_init(); uv_bios_get_sn_info(0, &uv_type, &sn_partition_id, - &uv_coherency_id, &uv_region_size); + &sn_coherency_id, &sn_region_size); uv_rtc_init(); for_each_present_cpu(cpu) { @@ -451,7 +451,7 @@ void __init uv_system_init(void) uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1; uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; - uv_cpu_hub_info(cpu)->coherency_domain_number = uv_coherency_id; + uv_cpu_hub_info(cpu)->coherency_domain_number = sn_coherency_id; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode = max(pnode, max_pnode); -- cgit v1.2.3 From 2bfef69d9e8cc056aa4dbc13f2136747340b4515 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Wed, 15 Oct 2008 11:51:53 +0200 Subject: x86: SB600: skip IRQ0 override if it is not routed to INT2 of IOAPIC Impact: fix hung bootup and other misbehavior on certain laptops On some more HP laptops BIOS reports an IRQ0 override but the SB600 chipset is configured such that timer interrupts go to INT0 of IOAPIC. Check IRQ0 routing and if it is routed to INT0 of IOAPIC skip the timer override. See following bug reports: http://bugzilla.kernel.org/show_bug.cgi?id=11715 http://bugzilla.kernel.org/show_bug.cgi?id=11516 Signed-off-by: Andreas Herrmann Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/early-quirks.c | 55 +++++++++++++++++++++++++++++++++++++++--- 1 file changed, 52 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 733c4f8d42ea..3ce029ffaa55 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -95,7 +95,8 @@ static void __init nvidia_bugs(int num, int slot, int func) } -static u32 ati_ixp4x0_rev(int num, int slot, int func) +#if defined(CONFIG_ACPI) && defined(CONFIG_X86_IO_APIC) +static u32 __init ati_ixp4x0_rev(int num, int slot, int func) { u32 d; u8 b; @@ -115,7 +116,6 @@ static u32 ati_ixp4x0_rev(int num, int slot, int func) static void __init ati_bugs(int num, int slot, int func) { -#if defined(CONFIG_ACPI) && defined (CONFIG_X86_IO_APIC) u32 d; u8 b; @@ -138,9 +138,56 @@ static void __init ati_bugs(int num, int slot, int func) printk(KERN_INFO "If you got timer trouble " "try acpi_use_timer_override\n"); } -#endif } +static u32 __init ati_sbx00_rev(int num, int slot, int func) +{ + u32 old, d; + + d = read_pci_config(num, slot, func, 0x70); + old = d; + d &= ~(1<<8); + write_pci_config(num, slot, func, 0x70, d); + d = read_pci_config(num, slot, func, 0x8); + d &= 0xff; + write_pci_config(num, slot, func, 0x70, old); + + return d; +} + +static void __init ati_bugs_contd(int num, int slot, int func) +{ + u32 d, rev; + + if (acpi_use_timer_override) + return; + + rev = ati_sbx00_rev(num, slot, func); + if (rev > 0x13) + return; + + /* check for IRQ0 interrupt swap */ + d = read_pci_config(num, slot, func, 0x64); + if (!(d & (1<<14))) + acpi_skip_timer_override = 1; + + if (acpi_skip_timer_override) { + printk(KERN_INFO "SB600 revision 0x%x\n", rev); + printk(KERN_INFO "Ignoring ACPI timer override.\n"); + printk(KERN_INFO "If you got timer trouble " + "try acpi_use_timer_override\n"); + } +} +#else +static void __init ati_bugs(int num, int slot, int func) +{ +} + +static void __init ati_bugs_contd(int num, int slot, int func) +{ +} +#endif + #ifdef CONFIG_DMAR static void __init intel_g33_dmar(int num, int slot, int func) { @@ -176,6 +223,8 @@ static struct chipset early_qrk[] __initdata = { PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, fix_hypertransport_config }, { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP400_SMBUS, PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, + { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, + PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, #ifdef CONFIG_DMAR { PCI_VENDOR_ID_INTEL, 0x29c0, PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, intel_g33_dmar }, -- cgit v1.2.3 From d2f6f7aeee890df445be29a60e34925ec15f620c Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Tue, 21 Oct 2008 22:45:22 +0200 Subject: MCE: Don't run 32bit machine checks with interrupts on Running machine checks with interrupt on is a extremly bad idea. The machine check handler only runs when the system is broken and needs to finish as quickly as possible. Remove the respective bogus post 2.6.27 regression and call the machine check vector directly again. This removes only code. Signed-off-by: Andi Kleen [Cherry-picked from x86/mce] Signed-off-by: Thomas Gleixner --- arch/x86/kernel/entry_32.S | 2 +- arch/x86/kernel/traps.c | 8 -------- 2 files changed, 1 insertion(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index c356423a6026..dd65143941a8 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1024,7 +1024,7 @@ ENTRY(machine_check) RING0_INT_FRAME pushl $0 CFI_ADJUST_CFA_OFFSET 4 - pushl $do_machine_check + pushl machine_check_vector CFI_ADJUST_CFA_OFFSET 4 jmp error_code CFI_ENDPROC diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index e062974cce34..04d242ab0161 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -931,14 +931,6 @@ do_device_not_available(struct pt_regs *regs, long error) } #ifdef CONFIG_X86_32 -#ifdef CONFIG_X86_MCE -dotraplinkage void __kprobes do_machine_check(struct pt_regs *regs, long error) -{ - conditional_sti(regs); - machine_check_vector(regs, error); -} -#endif - dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code) { siginfo_t info; -- cgit v1.2.3 From cf52ebedba77ee494b495dedd3a1f55944611275 Mon Sep 17 00:00:00 2001 From: Neil Horman Date: Fri, 17 Oct 2008 17:00:13 -0400 Subject: x86, kexec: fix hang on i386 when panic occurs while console_sem is held There's a corner case in 32 bit x86 kdump at the moment. When the box panics via nmi, we call bust_spinlocks(1) to disable sensitivity to the console_sem (allowing us to print to the console in all cases), but we don't call crash_kexec, until after we call bust_spinlocks(0), which re-enables console_sem sensitivity. The result is that, if we get an nmi while the console_sem is held and kdump is configured, and we try to print something to the console during kdump shutdown (which we often do) we deadlock the box. The fix is to simply do what 64 bit die_nmi does which is to not call bust_spinlocks(0) until after we call crash_kexec. Patch below tested successfully by me. Signed-off-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 1a78180f08d3..b3614752197b 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -405,7 +405,6 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic) panic("Non maskable interrupt"); console_silent(); spin_unlock(&nmi_print_lock); - bust_spinlocks(0); /* * If we are in kernel we are probably nested up pretty bad @@ -416,6 +415,7 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic) crash_kexec(regs); } + bust_spinlocks(0); do_exit(SIGSEGV); } -- cgit v1.2.3 From b4b8f87bf4958cbad620654efc0882ac46c19846 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:08 +0200 Subject: i386, dumpstack: move crash_kexec before bust_spinlocks(0) in oops_end crash_kexec should not be called with console_sem held. Move the call before bust_spinlocks(0) in oops_end to avoid the problem. Signed-off-by: Alexander van Heukelum Acked-by: "Neil Horman" Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index b3614752197b..5493d31be4e5 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -309,6 +309,9 @@ unsigned __kprobes long oops_begin(void) void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) { + if (regs && kexec_should_crash(current)) + crash_kexec(regs); + bust_spinlocks(0); die_owner = -1; add_taint(TAINT_DIE); @@ -318,8 +321,6 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) if (!regs) return; - if (kexec_should_crash(current)) - crash_kexec(regs); if (in_interrupt()) panic("Fatal exception in interrupt"); if (panic_on_oops) -- cgit v1.2.3 From 874d93d11823b2b861addac6a5dc31162e924ab2 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:09 +0200 Subject: x86, dumpstack: let signr=0 signal no do_exit Change oops_end such that signr=0 signals that do_exit is not to be called. Currently, each use of __die is soon followed by a call to oops_end and 'regs' is set to NULL if oops_end is expected not to call do_exit. Change all such pairs to set signr=0 instead. On x86_64 oops_end is used 'bare' in die_nmi; use signr=0 instead of regs=NULL there, too. Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 7 ++++--- arch/x86/kernel/dumpstack_64.c | 9 +++++---- 2 files changed, 9 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 5493d31be4e5..7c22f99f0efb 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -318,7 +318,7 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) __raw_spin_unlock(&die_lock); raw_local_irq_restore(flags); - if (!regs) + if (!signr) return; if (in_interrupt()) @@ -371,17 +371,18 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) void die(const char *str, struct pt_regs *regs, long err) { unsigned long flags = oops_begin(); + int sig = SIGSEGV; if (die_nest_count < 3) { report_bug(regs->ip, regs); if (__die(str, regs, err)) - regs = NULL; + sig = 0; } else { printk(KERN_EMERG "Recursive die() failure, output suppressed\n"); } - oops_end(flags, regs, SIGSEGV); + oops_end(flags, regs, sig); } static DEFINE_SPINLOCK(nmi_print_lock); diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 96a5db7da8a7..ffefea611ba3 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -465,7 +465,7 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) /* Nest count reaches zero, release the lock. */ __raw_spin_unlock(&die_lock); raw_local_irq_restore(flags); - if (!regs) { + if (!signr) { oops_exit(); return; } @@ -509,13 +509,14 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) void die(const char *str, struct pt_regs *regs, long err) { unsigned long flags = oops_begin(); + int sig = SIGSEGV; if (!user_mode(regs)) report_bug(regs->ip, regs); if (__die(str, regs, err)) - regs = NULL; - oops_end(flags, regs, SIGSEGV); + sig = 0; + oops_end(flags, regs, sig); } notrace __kprobes void @@ -539,7 +540,7 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic) crash_kexec(regs); if (do_panic || panic_on_oops) panic("Non maskable interrupt"); - oops_end(flags, NULL, SIGBUS); + oops_end(flags, regs, 0); nmi_exit(); local_irq_enable(); do_exit(SIGBUS); -- cgit v1.2.3 From 0ed7a498f416dcfa1cca478a559238a2a3396240 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:10 +0200 Subject: x86_64, dumpstack: move kexec_crash from __die to oops_end oops_end is preceded by either a call to __die, or a conditional call to crash_kexec. Move the conditional call to crash_kexec from the end of __die to the start of oops_end and remove the superfluous call to crash_kexec in die_nmi. Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_64.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index ffefea611ba3..57ce11b895ce 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -458,6 +458,9 @@ unsigned __kprobes long oops_begin(void) void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) { + if (regs && kexec_should_crash(current)) + crash_kexec(regs); + die_owner = -1; bust_spinlocks(0); die_nest_count--; @@ -501,8 +504,6 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) printk(KERN_ALERT "RIP "); printk_address(regs->ip, 1); printk(" RSP <%016lx>\n", regs->sp); - if (kexec_should_crash(current)) - crash_kexec(regs); return 0; } @@ -536,11 +537,9 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic) printk(" on CPU%d, ip %08lx, registers:\n", smp_processor_id(), regs->ip); show_registers(regs); - if (kexec_should_crash(current)) - crash_kexec(regs); + oops_end(flags, regs, 0); if (do_panic || panic_on_oops) panic("Non maskable interrupt"); - oops_end(flags, regs, 0); nmi_exit(); local_irq_enable(); do_exit(SIGBUS); -- cgit v1.2.3 From 10b14cb7eb7dd5bff8023f76a55c8ac20e586128 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:11 +0200 Subject: x86, dumpstack: always call oops_exit from oops_end Always call oops_exit from oops_end, even if signr==0. Also, move add_taint(TAINT_DIE) from __die to oops_end on x86_64 and interchange two lines to make oops_end more similar to the i386-version. Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 2 +- arch/x86/kernel/dumpstack_64.c | 11 +++++------ 2 files changed, 6 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 7c22f99f0efb..a29b88ffa346 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -318,6 +318,7 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) __raw_spin_unlock(&die_lock); raw_local_irq_restore(flags); + oops_exit(); if (!signr) return; @@ -325,7 +326,6 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) panic("Fatal exception in interrupt"); if (panic_on_oops) panic("Fatal exception"); - oops_exit(); do_exit(signr); } diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 57ce11b895ce..dc6162bf7454 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -461,22 +461,22 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) if (regs && kexec_should_crash(current)) crash_kexec(regs); - die_owner = -1; bust_spinlocks(0); + die_owner = -1; + add_taint(TAINT_DIE); die_nest_count--; if (!die_nest_count) /* Nest count reaches zero, release the lock. */ __raw_spin_unlock(&die_lock); raw_local_irq_restore(flags); - if (!signr) { - oops_exit(); + oops_exit(); + + if (!signr) return; - } if (in_interrupt()) panic("Fatal exception in interrupt"); if (panic_on_oops) panic("Fatal exception"); - oops_exit(); do_exit(signr); } @@ -499,7 +499,6 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) return 1; show_registers(regs); - add_taint(TAINT_DIE); /* Executive summary in case the oops scrolled away */ printk(KERN_ALERT "RIP "); printk_address(regs->ip, 1); -- cgit v1.2.3 From e4955cfd2f5c81eb708f55769aa60173f207fd63 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:12 +0200 Subject: i386, dumpstack: use x86_64's method to account die_nest_count oops_begin/oops_end should always be used in pairs. On x86_64 oops_begin increments die_nest_count, and oops_end decrements die_nest_count. Doing this makes oops_begin and oops_end equal to the x86_64 versions. Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 29 +++++++++++++++++------------ 1 file changed, 17 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index a29b88ffa346..7c7d691b32be 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -289,21 +289,24 @@ static unsigned int die_nest_count; unsigned __kprobes long oops_begin(void) { + int cpu; unsigned long flags; oops_enter(); - if (die_owner != raw_smp_processor_id()) { - console_verbose(); - raw_local_irq_save(flags); - __raw_spin_lock(&die_lock); - die_owner = smp_processor_id(); - die_nest_count = 0; - bust_spinlocks(1); - } else { - raw_local_irq_save(flags); + /* racy, but better than risking deadlock. */ + raw_local_irq_save(flags); + cpu = smp_processor_id(); + if (!__raw_spin_trylock(&die_lock)) { + if (cpu == die_owner) + /* nested oops. should stop eventually */; + else + __raw_spin_lock(&die_lock); } die_nest_count++; + die_owner = cpu; + console_verbose(); + bust_spinlocks(1); return flags; } @@ -315,13 +318,15 @@ void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) bust_spinlocks(0); die_owner = -1; add_taint(TAINT_DIE); - __raw_spin_unlock(&die_lock); + die_nest_count--; + if (!die_nest_count) + /* Nest count reaches zero, release the lock. */ + __raw_spin_unlock(&die_lock); raw_local_irq_restore(flags); - oops_exit(); + if (!signr) return; - if (in_interrupt()) panic("Fatal exception in interrupt"); if (panic_on_oops) -- cgit v1.2.3 From e06ca430c3d0fddbd1c901ab3fb3e1f0bc8a786b Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:13 +0200 Subject: i386, dumpstack: use oops_begin/oops_end in die_nmi Use oops_begin and oops_end in die_nmi. Whitespace-only changes on x86_64, to make it equal to i386's version. Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 33 +++++++++++---------------------- arch/x86/kernel/dumpstack_64.c | 4 ++-- 2 files changed, 13 insertions(+), 24 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 7c7d691b32be..e91ae34f9684 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -390,40 +390,29 @@ void die(const char *str, struct pt_regs *regs, long err) oops_end(flags, regs, sig); } -static DEFINE_SPINLOCK(nmi_print_lock); - void notrace __kprobes die_nmi(char *str, struct pt_regs *regs, int do_panic) { + unsigned long flags; + if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) return; - spin_lock(&nmi_print_lock); /* - * We are in trouble anyway, lets at least try - * to get a message out: - */ - bust_spinlocks(1); + * We are in trouble anyway, lets at least try + * to get a message out. + */ + flags = oops_begin(); printk(KERN_EMERG "%s", str); printk(" on CPU%d, ip %08lx, registers:\n", smp_processor_id(), regs->ip); show_registers(regs); - if (do_panic) + oops_end(flags, regs, 0); + if (do_panic || panic_on_oops) panic("Non maskable interrupt"); - console_silent(); - spin_unlock(&nmi_print_lock); - - /* - * If we are in kernel we are probably nested up pretty bad - * and might aswell get out now while we still can: - */ - if (!user_mode_vm(regs)) { - current->thread.trap_no = 2; - crash_kexec(regs); - } - - bust_spinlocks(0); - do_exit(SIGSEGV); + nmi_exit(); + local_irq_enable(); + do_exit(SIGBUS); } static int __init oops_setup(char *s) diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index dc6162bf7454..831e1e159cb4 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -519,7 +519,7 @@ void die(const char *str, struct pt_regs *regs, long err) oops_end(flags, regs, sig); } -notrace __kprobes void +void notrace __kprobes die_nmi(char *str, struct pt_regs *regs, int do_panic) { unsigned long flags; @@ -527,11 +527,11 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic) if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) return; - flags = oops_begin(); /* * We are in trouble anyway, lets at least try * to get a message out. */ + flags = oops_begin(); printk(KERN_EMERG "%s", str); printk(" on CPU%d, ip %08lx, registers:\n", smp_processor_id(), regs->ip); -- cgit v1.2.3 From 871d3779cba18b028e34d0d2f6cc6caae76a97b6 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 22 Oct 2008 12:00:14 +0200 Subject: i386, dumpstack: unify die() Make i386's die() equal to x86_64's version. Whitespace-only changes on x86_64, to make it equal to i386's version. (user_mode and user_mode_vm are equal on x86_64.) Signed-off-by: Alexander van Heukelum Acked-by: Neil Horman Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack_32.c | 10 +++------- arch/x86/kernel/dumpstack_64.c | 6 +++++- 2 files changed, 8 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index e91ae34f9684..f2046c5752d0 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -378,15 +378,11 @@ void die(const char *str, struct pt_regs *regs, long err) unsigned long flags = oops_begin(); int sig = SIGSEGV; - if (die_nest_count < 3) { + if (!user_mode_vm(regs)) report_bug(regs->ip, regs); - if (__die(str, regs, err)) - sig = 0; - } else { - printk(KERN_EMERG "Recursive die() failure, output suppressed\n"); - } - + if (__die(str, regs, err)) + sig = 0; oops_end(flags, regs, sig); } diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 831e1e159cb4..28c67aae5562 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -506,12 +506,16 @@ int __kprobes __die(const char *str, struct pt_regs *regs, long err) return 0; } +/* + * This is gone through when something in the kernel has done something bad + * and is about to be terminated: + */ void die(const char *str, struct pt_regs *regs, long err) { unsigned long flags = oops_begin(); int sig = SIGSEGV; - if (!user_mode(regs)) + if (!user_mode_vm(regs)) report_bug(regs->ip, regs); if (__die(str, regs, err)) -- cgit v1.2.3 From 35af28219e684a36cc8b1ff456c370ce22be157d Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Wed, 22 Oct 2008 13:08:31 +0200 Subject: x86: call dmi-quirks for HP Laptops after early-quirks are executed Impact: make warning message disappear - functionality unchanged Problems with bogus IRQ0 override of those laptops should be fixed with commits x86: SB600: skip IRQ0 override if it is not routed to INT2 of IOAPIC x86: SB450: skip IRQ0 override if it is not routed to INT2 of IOAPIC that introduce early-quirks based on chipset configuration. For further information, see http://bugzilla.kernel.org/show_bug.cgi?id=11516 Instead of removing the related dmi-quirks completely we'd like to keep them for (at least) one kernel version -- to double-check whether the early-quirks really took effect. But the dmi-quirks need to be called after early-quirks are executed. With this patch calling sequence for dmi-quriks is changed as follows: acpi_boot_table_init() (dmi-quirks) ... early_quirks() (detect bogus IRQ0 override) ... acpi_boot_init() (late dmi-quirks and setup IO APIC) Note: Plan is to remove the "late dmi-quirks" with next kernel version. Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 0d1c26a583c5..8072edb62478 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -1598,6 +1598,11 @@ static struct dmi_system_id __initdata acpi_dmi_table[] = { DMI_MATCH(DMI_PRODUCT_NAME, "TravelMate 360"), }, }, + {} +}; + +/* second table for DMI checks that should run after early-quirks */ +static struct dmi_system_id __initdata acpi_dmi_table_late[] = { /* * HP laptops which use a DSDT reporting as HP/SB400/10000, * which includes some code which overrides all temperature @@ -1726,6 +1731,9 @@ int __init early_acpi_boot_init(void) int __init acpi_boot_init(void) { + /* those are executed after early-quirks are executed */ + dmi_check_system(acpi_dmi_table_late); + /* * If acpi_disabled, bail out * One exception: acpi=ht continues far enough to enumerate LAPICs -- cgit v1.2.3 From bc8bcc79ea4203c7d04309f1307ab88c86f0b0cf Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 22 Oct 2008 12:42:30 +0800 Subject: x86/proc: fix /proc/cpuinfo cpu offline bug Impact: fix missing CPUs in /proc/cpuinfo after CPU hotunplug/hotreplug In my test, I found that if a cpu has been offline, the next cpus may not be shown in the /proc/cpuinfo. if one read() cannot consume the whole /proc/cpuinfo, c_start() will be called again in the next read() calls. And *pos has been increased by 1 by the caller(seq_read()). if this time the cpu#*pos is offline, c_start() will return NULL, and the next cpus can not be shown. this fix use next_cpu_nr(*pos - 1, cpu_online_map) to search the next unshown cpu. the most easy way to reproduce this bug: 1) offline cpu#1 (cpu#0 is online) 2) dd ibs=2 if=/proc/cpuinfo the result is that only cpu#0 is shown. cpu#2 and cpu#3 .... cannot be shown in /proc/cpuinfo it's bug. Signed-off-by: Lai Jiangshan Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/proc.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/proc.c b/arch/x86/kernel/cpu/proc.c index a26c480b9491..01b1244ef1c0 100644 --- a/arch/x86/kernel/cpu/proc.c +++ b/arch/x86/kernel/cpu/proc.c @@ -160,14 +160,16 @@ static void *c_start(struct seq_file *m, loff_t *pos) { if (*pos == 0) /* just in case, cpu 0 is not the first */ *pos = first_cpu(cpu_online_map); - if ((*pos) < nr_cpu_ids && cpu_online(*pos)) + else + *pos = next_cpu_nr(*pos - 1, cpu_online_map); + if ((*pos) < nr_cpu_ids) return &cpu_data(*pos); return NULL; } static void *c_next(struct seq_file *m, void *v, loff_t *pos) { - *pos = next_cpu(*pos, cpu_online_map); + (*pos)++; return c_start(m, pos); } -- cgit v1.2.3 From db96b0a0e4158806fcf03945a870f9320e6bac9b Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Wed, 22 Oct 2008 18:00:09 +0400 Subject: x86: do_boot_cpu - check if we have ESR register Impact: fix APIC IRQ irregularities on certain older boxes We should touch the APIC ESR register if only we have it. The patch fixes the problem mentioned by Max Kellermann: http://lkml.org/lkml/2008/10/17/147 Bisected-by: Max Kellermann Signed-off-by: Cyrill Gorcunov [ mingo@elte.hu: build fix ] Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 7ece815ea637..7b1093397319 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -893,9 +893,11 @@ do_rest: smpboot_setup_warm_reset_vector(start_ip); /* * Be paranoid about clearing APIC errors. - */ - apic_write(APIC_ESR, 0); - apic_read(APIC_ESR); + */ + if (APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { + apic_write(APIC_ESR, 0); + apic_read(APIC_ESR); + } } /* -- cgit v1.2.3 From 55410791c91f448737c90585d2e280edd3c4d5c7 Mon Sep 17 00:00:00 2001 From: "Gustavo F. Padovan" Date: Wed, 15 Oct 2008 10:37:04 -0300 Subject: x86: remove redundant KERN_DEBUG on pr_debug pr_debug don't need KERN_DEBUG. Signed-off-by: Gustavo F. Padovan Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 2 +- arch/x86/kernel/setup_percpu.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 8072edb62478..dc3eac353db2 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -1136,7 +1136,7 @@ int mp_register_gsi(u32 gsi, int triggering, int polarity) return gsi; } if (test_bit(ioapic_pin, mp_ioapic_routing[ioapic].pin_programmed)) { - pr_debug(KERN_DEBUG "Pin %d-%d already programmed\n", + pr_debug("Pin %d-%d already programmed\n", mp_ioapic_routing[ioapic].apic_id, ioapic_pin); #ifdef CONFIG_X86_32 return (gsi < IRQ_COMPRESSION_START ? gsi : gsi_to_irq[gsi]); diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 410c88f0bfeb..ae0c0d3bb770 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -218,7 +218,7 @@ static void __init setup_node_to_cpumask_map(void) /* allocate the map */ map = alloc_bootmem_low(nr_node_ids * sizeof(cpumask_t)); - pr_debug(KERN_DEBUG "Node to cpumask map at %p for %d nodes\n", + pr_debug("Node to cpumask map at %p for %d nodes\n", map, nr_node_ids); /* node_to_cpumask() will now work */ -- cgit v1.2.3 From aef8f5b8c2da706cb3efe701e2a35e11221ea5bd Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 14 Oct 2008 21:43:43 -0700 Subject: x86/tlb_uv: remove strange mc146818rtc include For some reason tlb_uv was including linux/mc146818rtc.h. It really just needs linux/seq_file.h Signed-off-by: Jeremy Fitzhardinge Cc: Cliff Wickman Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_uv.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index 8b8c0d6640fa..04431f34fd16 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -6,7 +6,7 @@ * This code is released under the GNU General Public License version 2 or * later. */ -#include +#include #include #include -- cgit v1.2.3 From 983f91ff949d9008eb7fb73ba4e35f9decc182b3 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 12 Oct 2008 11:44:08 +0200 Subject: x86: fix section mismatch warning - apic_flat Impact: cleanup only, no functionality changed WARNING: vmlinux.o(.data+0xbe08): Section mismatch in reference from the variable apic_flat to the function .init.text:flat_acpi_madt_oem_check() The variable apic_flat references the function __init flat_acpi_madt_oem_check() This is harmless, because the .acpi_madt_oem_check is only called during init time. But we keep the function pointer around in a .data function pointer template, so it's better we do not keep that stale - so mark this function non-__init. Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genapic_flat_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index 2ec2de8d8c46..a10ed42a35e0 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -25,7 +25,7 @@ #include #endif -static int __init flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) +static int flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) { return 1; } -- cgit v1.2.3 From fae1721601a008fc83a809fdd76085a56f2275f0 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 12 Oct 2008 11:44:09 +0200 Subject: x86: fix section mismatch warning - apic_physflat Impact: cleanup only, no functionality changed WARNING: vmlinux.o(.data+0xbe88): Section mismatch in reference from the variable apic_physflat to the function .init.text:physflat_acpi_madt_oem_check() The variable apic_physflat references the function __init physflat_acpi_madt_oem_check() Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genapic_flat_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index a10ed42a35e0..c0262791bda4 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -170,7 +170,7 @@ struct genapic apic_flat = { * We cannot use logical delivery in this case because the mask * overflows, so use physical mode. */ -static int __init physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) +static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) { #ifdef CONFIG_ACPI /* -- cgit v1.2.3 From f8827c017f19e1cd8834821f4303d5f163feab84 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 12 Oct 2008 11:44:10 +0200 Subject: x86: fix section mismatch warning - apic_x2apic_uv_x Impact: cleanup only, no functionality changed WARNING: vmlinux.o(.data+0xbf08): Section mismatch in reference from the variable apic_x2apic_uv_x to the function .init.text:uv_acpi_madt_oem_check() The variable apic_x2apic_uv_x references the function __init uv_acpi_madt_oem_check() Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index bfd532843df6..680a06557c5e 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -30,7 +30,7 @@ DEFINE_PER_CPU(int, x2apic_extra_bits); static enum uv_system_type uv_system_type; -static int __init uv_acpi_madt_oem_check(char *oem_id, char *oem_table_id) +static int uv_acpi_madt_oem_check(char *oem_id, char *oem_table_id) { if (!strcmp(oem_id, "SGI")) { if (!strcmp(oem_table_id, "UVL")) -- cgit v1.2.3 From 2caa37150d7aa110b6155cb77b07e581c103fef4 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 12 Oct 2008 11:44:11 +0200 Subject: x86: fix section mismatch warning - apic_x2apic_cluster Impact: cleanup only, no functionality changed WARNING: vmlinux.o(.data+0xbf88): Section mismatch in reference from the variable apic_x2apic_cluster to the function .init.text:x2apic_acpi_madt_oem_check() The variable apic_x2apic_cluster references the function __init x2apic_acpi_madt_oem_check() Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_cluster.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_cluster.c b/arch/x86/kernel/genx2apic_cluster.c index e4bf2cc0d743..f6a2c8eb48a6 100644 --- a/arch/x86/kernel/genx2apic_cluster.c +++ b/arch/x86/kernel/genx2apic_cluster.c @@ -12,7 +12,7 @@ DEFINE_PER_CPU(u32, x86_cpu_to_logical_apicid); -static int __init x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) +static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) { if (cpu_has_x2apic) return 1; -- cgit v1.2.3 From 3cfba0892585d4c8e7b4122b5dc0d206a76936de Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 12 Oct 2008 11:46:23 +0200 Subject: x86: fix section mismatch warning - apic_x2apic_phys Impact: cleanup only, no functionality changed WARNING: vmlinux.o(.data+0xc008): Section mismatch in reference from the variable apic_x2apic_phys to the function .init.text:x2apic_acpi_madt_oem_check() The variable apic_x2apic_phys references the function __init x2apic_acpi_madt_oem_check() Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_phys.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_phys.c b/arch/x86/kernel/genx2apic_phys.c index 8f1343df2627..d042211768b7 100644 --- a/arch/x86/kernel/genx2apic_phys.c +++ b/arch/x86/kernel/genx2apic_phys.c @@ -19,7 +19,7 @@ static int set_x2apic_phys_mode(char *arg) } early_param("x2apic_phys", set_x2apic_phys_mode); -static int __init x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) +static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) { if (cpu_has_x2apic && x2apic_phys) return 1; -- cgit v1.2.3 From bb8985586b7a906e116db835c64773b7a7d51663 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 17 Aug 2008 21:05:42 -0400 Subject: x86, um: ... and asm-x86 move Signed-off-by: Al Viro Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index 7f0b45a5d788..82ec6075c057 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -25,7 +25,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += perfctr-watchdog.o quiet_cmd_mkcapflags = MKCAP $@ cmd_mkcapflags = $(PERL) $(srctree)/$(src)/mkcapflags.pl $< $@ -cpufeature = $(src)/../../../../include/asm-x86/cpufeature.h +cpufeature = $(src)/../../include/asm/cpufeature.h targets += capflags.c $(obj)/capflags.c: $(cpufeature) $(src)/mkcapflags.pl FORCE -- cgit v1.2.3 From 1965aae3c98397aad957412413c07e97b1bd4e64 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Wed, 22 Oct 2008 22:26:29 -0700 Subject: x86: Fix ASM_X86__ header guards Change header guards named "ASM_X86__*" to "_ASM_X86_*" since: a. the double underscore is ugly and pointless. b. no leading underscore violates namespace constraints. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/asm-offsets_64.c | 2 +- arch/x86/kernel/syscall_64.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c index 505543a75a56..7fcf63d22f8b 100644 --- a/arch/x86/kernel/asm-offsets_64.c +++ b/arch/x86/kernel/asm-offsets_64.c @@ -22,7 +22,7 @@ #define __NO_STUBS 1 #undef __SYSCALL -#undef ASM_X86__UNISTD_64_H +#undef _ASM_X86_UNISTD_64_H #define __SYSCALL(nr, sym) [nr] = 1, static char syscalls[] = { #include diff --git a/arch/x86/kernel/syscall_64.c b/arch/x86/kernel/syscall_64.c index 3d1be4f0fac5..de87d6008295 100644 --- a/arch/x86/kernel/syscall_64.c +++ b/arch/x86/kernel/syscall_64.c @@ -8,12 +8,12 @@ #define __NO_STUBS #define __SYSCALL(nr, sym) extern asmlinkage void sym(void) ; -#undef ASM_X86__UNISTD_64_H +#undef _ASM_X86_UNISTD_64_H #include #undef __SYSCALL #define __SYSCALL(nr, sym) [nr] = sym, -#undef ASM_X86__UNISTD_64_H +#undef _ASM_X86_UNISTD_64_H typedef void (*sys_call_ptr_t)(void); -- cgit v1.2.3 From 593eb8a2d63e95772a5f22d746f18a997c5ee463 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 23 Oct 2008 09:32:59 -0400 Subject: ftrace: return error on failed modified text. Have the ftrace_modify_code return error values: -EFAULT on error of reading the address -EINVAL if what is read does not match what it expected -EPERM if the write fails to update after a successful match. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 8821ceabf51d..428291581cb2 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -62,7 +62,6 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { unsigned char replaced[MCOUNT_INSN_SIZE]; - int ret; /* * Note: Due to modules and __init, code can @@ -72,15 +71,16 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * No real locking needed, this code is run through * kstop_machine, or before SMP starts. */ - if (__copy_from_user_inatomic(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) - return 1; + if (__copy_from_user_inatomic(replaced, (char __user *)ip, + MCOUNT_INSN_SIZE)) + return -EFAULT; if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) - return 2; + return -EINVAL; - ret = __copy_to_user_inatomic((char __user *)ip, new_code, - MCOUNT_INSN_SIZE); - WARN_ON_ONCE(ret); + if (__copy_to_user_inatomic((char __user *)ip, new_code, + MCOUNT_INSN_SIZE)) + return -EPERM; sync_core(); -- cgit v1.2.3 From 76aefee57657428fb77cbd8624119c1a440bee44 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 23 Oct 2008 09:33:00 -0400 Subject: ftrace: comment arch ftrace code Add comments to explain what is happening in the x86 arch ftrace code. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 428291581cb2..783455454d78 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -66,18 +66,23 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, /* * Note: Due to modules and __init, code can * disappear and change, we need to protect against faulting - * as well as code changing. + * as well as code changing. We do this by using the + * __copy_*_user functions. * * No real locking needed, this code is run through * kstop_machine, or before SMP starts. */ + + /* read the text we want to modify */ if (__copy_from_user_inatomic(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) return -EFAULT; + /* Make sure it is what we expect it to be */ if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) return -EINVAL; + /* replace the text with the new text */ if (__copy_to_user_inatomic((char __user *)ip, new_code, MCOUNT_INSN_SIZE)) return -EPERM; -- cgit v1.2.3 From ab9a0918cbf0fa8883301838df8dbc8fc085ff50 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 23 Oct 2008 09:33:01 -0400 Subject: ftrace: use probe_kernel Andrew Morton suggested using the proper API for reading and writing kernel areas that might fault. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 783455454d78..da4fb0deecf7 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -67,15 +67,14 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * Note: Due to modules and __init, code can * disappear and change, we need to protect against faulting * as well as code changing. We do this by using the - * __copy_*_user functions. + * probe_kernel_* functions. * * No real locking needed, this code is run through * kstop_machine, or before SMP starts. */ /* read the text we want to modify */ - if (__copy_from_user_inatomic(replaced, (char __user *)ip, - MCOUNT_INSN_SIZE)) + if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE)) return -EFAULT; /* Make sure it is what we expect it to be */ @@ -83,8 +82,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return -EINVAL; /* replace the text with the new text */ - if (__copy_to_user_inatomic((char __user *)ip, new_code, - MCOUNT_INSN_SIZE)) + if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE)) return -EPERM; sync_core(); -- cgit v1.2.3 From 4d296c24326783bff1282ac72f310d8bac8df413 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 23 Oct 2008 09:33:06 -0400 Subject: ftrace: remove mcount set The arch dependent function ftrace_mcount_set was only used by the daemon start up code. This patch removes it. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 7 ------- 1 file changed, 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index da4fb0deecf7..b399eed23538 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -103,13 +103,6 @@ notrace int ftrace_update_ftrace_func(ftrace_func_t func) return ret; } -notrace int ftrace_mcount_set(unsigned long *data) -{ - /* mcount is initialized as a nop */ - *data = 0; - return 0; -} - int __init ftrace_dyn_arch_init(void *data) { extern const unsigned char ftrace_test_p6nop[]; -- cgit v1.2.3 From 15adc048986f6b54b6044f2b6fc4b48f49413e2f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 23 Oct 2008 09:33:08 -0400 Subject: ftrace, powerpc, sparc64, x86: remove notrace from arch ftrace file The entire file of ftrace.c in the arch code needs to be marked as notrace. It is much cleaner to do this from the Makefile with CFLAGS_REMOVE_ftrace.o. [ powerpc already had this in its Makefile. ] Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 1 + arch/x86/kernel/ftrace.c | 10 +++++----- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index f1283fe60723..e489ff9cb3e2 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -11,6 +11,7 @@ ifdef CONFIG_FUNCTION_TRACER CFLAGS_REMOVE_tsc.o = -pg CFLAGS_REMOVE_rtc.o = -pg CFLAGS_REMOVE_paravirt-spinlocks.o = -pg +CFLAGS_REMOVE_ftrace.o = -pg endif # diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index b399eed23538..b1e5e2244eca 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -33,17 +33,17 @@ union ftrace_code_union { }; -static int notrace ftrace_calc_offset(long ip, long addr) +static int ftrace_calc_offset(long ip, long addr) { return (int)(addr - ip); } -notrace unsigned char *ftrace_nop_replace(void) +unsigned char *ftrace_nop_replace(void) { return (char *)ftrace_nop; } -notrace unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) +unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -57,7 +57,7 @@ notrace unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) return calc.code; } -notrace int +int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { @@ -90,7 +90,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return 0; } -notrace int ftrace_update_ftrace_func(ftrace_func_t func) +int ftrace_update_ftrace_func(ftrace_func_t func) { unsigned long ip = (unsigned long)(&ftrace_call); unsigned char old[MCOUNT_INSN_SIZE], *new; -- cgit v1.2.3 From 03967c5267b0e7312d1d55dc814d94cf190ca573 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Thu, 23 Oct 2008 23:14:29 +0900 Subject: x86: restore the old swiotlb alloc_coherent behavior This restores the old swiotlb alloc_coherent behavior (before the alloc_coherent rewrite): http://lkml.org/lkml/2008/8/12/200 The old alloc_coherent avoids GFP_DMA allocation first and if the allocated address is not fit for the device's coherent_dma_mask, then dma_alloc_coherent does GFP_DMA allocation. If it fails, alloc_coherent calls swiotlb_alloc_coherent (in short, we rarely used swiotlb_alloc_coherent). After the alloc_coherent rewrite, dma_alloc_coherent (include/asm-x86/dma-mapping.h) directly calls swiotlb_alloc_coherent. It means that we possibly can't handle a device having dma_masks > 24bit < 32bits since swiotlb_alloc_coherent doesn't have the above GFP_DMA retry mechanism. This patch fixes x86's swiotlb alloc_coherent to use the GFP_DMA retry mechanism, which dma_generic_alloc_coherent() provides now (pci-nommu.c and GART IOMMU driver also use dma_generic_alloc_coherent). Signed-off-by: FUJITA Tomonori Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index c4ce0332759e..3c539d111abb 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -18,9 +18,21 @@ swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, return swiotlb_map_single(hwdev, phys_to_virt(paddr), size, direction); } +static void *x86_swiotlb_alloc_coherent(struct device *hwdev, size_t size, + dma_addr_t *dma_handle, gfp_t flags) +{ + void *vaddr; + + vaddr = dma_generic_alloc_coherent(hwdev, size, dma_handle, flags); + if (vaddr) + return vaddr; + + return swiotlb_alloc_coherent(hwdev, size, dma_handle, flags); +} + struct dma_mapping_ops swiotlb_dma_ops = { .mapping_error = swiotlb_dma_mapping_error, - .alloc_coherent = swiotlb_alloc_coherent, + .alloc_coherent = x86_swiotlb_alloc_coherent, .free_coherent = swiotlb_free_coherent, .map_single = swiotlb_map_single_phys, .unmap_single = swiotlb_unmap_single, -- cgit v1.2.3 From 3b15e581981b3ad35809f56d8131d5c19b6da1bd Mon Sep 17 00:00:00 2001 From: Fenghua Yu Date: Thu, 23 Oct 2008 16:51:00 -0700 Subject: x86/PCI: build failure at x86/kernel/pci-dma.c with !CONFIG_PCI On Thu, Oct 23, 2008 at 04:09:52PM -0700, Alexander Beregalov wrote: > arch/x86/kernel/built-in.o: In function `iommu_setup': > pci-dma.c:(.init.text+0x36ad): undefined reference to `forbid_dac' > pci-dma.c:(.init.text+0x36cc): undefined reference to `forbid_dac' > pci-dma.c:(.init.text+0x3711): undefined reference to `forbid_dac This patch partially reverts a patch to add IOMMU support to ia64. The forbid_dac variable was incorrectly moved to quirks.c, which isn't built when PCI is disabled. Tested-by: "Alexander Beregalov" Acked-by: FUJITA Tomonori Signed-off-by: Fenghua Yu Signed-off-by: Jesse Barnes --- arch/x86/kernel/pci-dma.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 1972266e8ba5..192624820217 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -9,6 +9,8 @@ #include #include +static int forbid_dac __read_mostly; + struct dma_mapping_ops *dma_ops; EXPORT_SYMBOL(dma_ops); @@ -291,3 +293,17 @@ void pci_iommu_shutdown(void) } /* Must execute after PCI subsystem */ fs_initcall(pci_iommu_init); + +#ifdef CONFIG_PCI +/* Many VIA bridges seem to corrupt data for DAC. Disable it here */ + +static __devinit void via_no_dac(struct pci_dev *dev) +{ + if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI && forbid_dac == 0) { + printk(KERN_INFO "PCI: VIA PCI bridge detected." + "Disabling DAC.\n"); + forbid_dac = 1; + } +} +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_VIA, PCI_ANY_ID, via_no_dac); +#endif -- cgit v1.2.3 From 7f1baa063e2582dd52d83bb31508e9e84468c666 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Fri, 24 Oct 2008 15:24:29 -0700 Subject: x86/uv: provide a System Activity Indicator driver Impact: start per CPU heartbeat LED timers on SGI UV systems The SGI UV system has no LEDS but uses one of the system controller regs to indicate the online internal state of the cpu. There is a heartbeat bit indicating that the cpu is responding to interrupts, and an idle bit indicating whether the cpu is idle when the heartbeat interrupt occurs. The current period is one second. When a cpu panics, an error code is written by BIOS to this same reg. This patchset provides the following: * x86_64: Add base functionality for writing to the specific SCIR's for each cpu. * heartbeat: Invert "heartbeat" bit to indicate the cpu is "interruptible". If the current thread is the idle thread, then indicate system is "idle". * if hotplug enabled, all bits are set (0xff) when the cpu is disabled. Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 102 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 102 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index dc6b46961523..84367d84bb10 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -10,6 +10,7 @@ #include #include +#include #include #include #include @@ -18,6 +19,8 @@ #include #include #include +#include +#include #include #include #include @@ -356,6 +359,103 @@ static __init void uv_rtc_init(void) sn_rtc_cycles_per_second = ticks_per_sec; } +/* + * percpu heartbeat timer + */ +static void uv_heartbeat(unsigned long ignored) +{ + struct timer_list *timer = &uv_hub_info->scir.timer; + unsigned char bits = uv_hub_info->scir.state; + + /* flip heartbeat bit */ + bits ^= SCIR_CPU_HEARTBEAT; + + /* are we the idle thread? */ + if (current->pid == 0) + bits &= ~SCIR_CPU_ACTIVITY; + else + bits |= SCIR_CPU_ACTIVITY; + + /* update system controller interface reg */ + uv_set_scir_bits(bits); + + /* enable next timer period */ + mod_timer(timer, jiffies + SCIR_CPU_HB_INTERVAL); +} + +static void __cpuinit uv_heartbeat_enable(int cpu) +{ + if (!uv_cpu_hub_info(cpu)->scir.enabled) { + struct timer_list *timer = &uv_cpu_hub_info(cpu)->scir.timer; + + uv_set_cpu_scir_bits(cpu, SCIR_CPU_HEARTBEAT|SCIR_CPU_ACTIVITY); + setup_timer(timer, uv_heartbeat, cpu); + timer->expires = jiffies + SCIR_CPU_HB_INTERVAL; + add_timer_on(timer, cpu); + uv_cpu_hub_info(cpu)->scir.enabled = 1; + } + + /* check boot cpu */ + if (!uv_cpu_hub_info(0)->scir.enabled) + uv_heartbeat_enable(0); +} + +static void __cpuinit uv_heartbeat_disable(int cpu) +{ + if (uv_cpu_hub_info(cpu)->scir.enabled) { + uv_cpu_hub_info(cpu)->scir.enabled = 0; + del_timer(&uv_cpu_hub_info(cpu)->scir.timer); + } + uv_set_cpu_scir_bits(cpu, 0xff); +} + +#ifdef CONFIG_HOTPLUG_CPU +/* + * cpu hotplug notifier + */ +static __cpuinit int uv_scir_cpu_notify(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + long cpu = (long)hcpu; + + switch (action) { + case CPU_ONLINE: + uv_heartbeat_enable(cpu); + break; + case CPU_DOWN_PREPARE: + uv_heartbeat_disable(cpu); + break; + default: + break; + } + return NOTIFY_OK; +} + +static __init void uv_scir_register_cpu_notifier(void) +{ + hotcpu_notifier(uv_scir_cpu_notify, 0); +} + +#else /* !CONFIG_HOTPLUG_CPU */ + +static __init void uv_scir_register_cpu_notifier(void) +{ +} + +static __init int uv_init_heartbeat(void) +{ + int cpu; + + if (is_uv_system()) + for_each_online_cpu(cpu) + uv_heartbeat_enable(cpu); + return 0; +} + +late_initcall(uv_init_heartbeat); + +#endif /* !CONFIG_HOTPLUG_CPU */ + /* * Called on each cpu to initialize the per_cpu UV data area. * ZZZ hotplug not supported yet @@ -452,6 +552,7 @@ void __init uv_system_init(void) uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper; uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base; uv_cpu_hub_info(cpu)->coherency_domain_number = sn_coherency_id; + uv_cpu_hub_info(cpu)->scir.offset = SCIR_LOCAL_MMR_BASE + lcpu; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode = max(pnode, max_pnode); @@ -468,4 +569,5 @@ void __init uv_system_init(void) map_mmioh_high(max_pnode); uv_cpu_init(); + uv_scir_register_cpu_notifier(); } -- cgit v1.2.3 From 709110bd5624094992579f5311541f2e2b7ce58a Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 23 Oct 2008 17:14:25 -0700 Subject: x86: signal: cosmetic unification of restore_sigcontext() Impact: cleanup Make restore_sigcontext() the same. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 22 ++++++++++++++++++++++ arch/x86/kernel/signal_64.c | 15 +++++++++++++++ 2 files changed, 37 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index d6dd057d0f22..85a0d37cdae9 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -149,14 +149,36 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, /* Always make any pending restarted system calls return -EINTR */ current_thread_info()->restart_block.fn = do_no_restart_syscall; +#ifdef CONFIG_X86_32 GET_SEG(gs); COPY_SEG(fs); COPY_SEG(es); COPY_SEG(ds); +#endif /* CONFIG_X86_32 */ + COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); COPY(dx); COPY(cx); COPY(ip); + +#ifdef CONFIG_X86_64 + COPY(r8); + COPY(r9); + COPY(r10); + COPY(r11); + COPY(r12); + COPY(r13); + COPY(r14); + COPY(r15); +#endif /* CONFIG_X86_64 */ + +#ifdef CONFIG_X86_32 COPY_SEG_STRICT(cs); COPY_SEG_STRICT(ss); +#else /* !CONFIG_X86_32 */ + /* Kernel saves and restores only the CS segment register on signals, + * which is the bare minimum needed to allow mixed 32/64-bit code. + * App's signal handler can save/restore other segments if needed. */ + COPY_SEG_STRICT(cs); +#endif /* CONFIG_X86_32 */ err |= __get_user(tmpflags, &sc->flags); regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index a5c9627f4db9..9c469da7f9e8 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -76,8 +76,17 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, /* Always make any pending restarted system calls return -EINTR */ current_thread_info()->restart_block.fn = do_no_restart_syscall; +#ifdef CONFIG_X86_32 + GET_SEG(gs); + COPY_SEG(fs); + COPY_SEG(es); + COPY_SEG(ds); +#endif /* CONFIG_X86_32 */ + COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); COPY(dx); COPY(cx); COPY(ip); + +#ifdef CONFIG_X86_64 COPY(r8); COPY(r9); COPY(r10); @@ -86,11 +95,17 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, COPY(r13); COPY(r14); COPY(r15); +#endif /* CONFIG_X86_64 */ +#ifdef CONFIG_X86_32 + COPY_SEG_STRICT(cs); + COPY_SEG_STRICT(ss); +#else /* !CONFIG_X86_32 */ /* Kernel saves and restores only the CS segment register on signals, * which is the bare minimum needed to allow mixed 32/64-bit code. * App's signal handler can save/restore other segments if needed. */ COPY_SEG_STRICT(cs); +#endif /* CONFIG_X86_32 */ err |= __get_user(tmpflags, &sc->flags); regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); -- cgit v1.2.3 From fd4a2030a358b4818646031049d9631bd45b9915 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 23 Oct 2008 17:15:28 -0700 Subject: x86: signal_64.c: get_stack() doesn't need entire regs Impact: cleanup get_stack() uses sp only, entire regs is not needed. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_64.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 9c469da7f9e8..3d0deb336745 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -199,12 +199,10 @@ setup_sigcontext(struct sigcontext __user *sc, struct pt_regs *regs, */ static void __user * -get_stack(struct k_sigaction *ka, struct pt_regs *regs, unsigned long size) +get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) { - unsigned long sp; - /* Default to using normal stack - redzone*/ - sp = regs->sp - 128; + sp -= 128; /* This is the X/Open sanctioned signal stack switching. */ if (ka->sa.sa_flags & SA_ONSTACK) { @@ -224,14 +222,14 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, struct task_struct *me = current; if (used_math()) { - fp = get_stack(ka, regs, sig_xstate_size); + fp = get_stack(ka, regs->sp, sig_xstate_size); frame = (void __user *)round_down( (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8; if (save_i387_xstate(fp) < 0) return -EFAULT; } else - frame = get_stack(ka, regs, sizeof(struct rt_sigframe)) - 8; + frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8; if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) return -EFAULT; -- cgit v1.2.3 From ef020ab0109aa5cd6eac2e93519b7641c9862828 Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Thu, 23 Oct 2008 17:54:05 -0500 Subject: x86/uv: memory allocation at initialization Impact: on SGI UV platforms, fix boot crash UV initialization is currently called too late to call alloc_bootmem_pages(). The current sequence is: start_kernel() mem_init() free_all_bootmem() <--- discard of bootmem rest_init() kernel_init() smp_prepare_cpus() native_smp_prepare_cpus() uv_system_init() <--- uses alloc_bootmem_pages() It should be calling kmalloc(). Signed-off-by: Cliff Wickman Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 680a06557c5e..2c7dbdb98278 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -15,7 +15,6 @@ #include #include #include -#include #include #include #include @@ -398,16 +397,16 @@ void __init uv_system_init(void) printk(KERN_DEBUG "UV: Found %d blades\n", uv_num_possible_blades()); bytes = sizeof(struct uv_blade_info) * uv_num_possible_blades(); - uv_blade_info = alloc_bootmem_pages(bytes); + uv_blade_info = kmalloc(bytes, GFP_KERNEL); get_lowmem_redirect(&lowmem_redir_base, &lowmem_redir_size); bytes = sizeof(uv_node_to_blade[0]) * num_possible_nodes(); - uv_node_to_blade = alloc_bootmem_pages(bytes); + uv_node_to_blade = kmalloc(bytes, GFP_KERNEL); memset(uv_node_to_blade, 255, bytes); bytes = sizeof(uv_cpu_to_blade[0]) * num_possible_cpus(); - uv_cpu_to_blade = alloc_bootmem_pages(bytes); + uv_cpu_to_blade = kmalloc(bytes, GFP_KERNEL); memset(uv_cpu_to_blade, 255, bytes); blade = 0; -- cgit v1.2.3 From 8115f3f0c939c5db0fe3c6c6c58911fd3a205b1e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 24 Oct 2008 09:12:17 -0400 Subject: ftrace: use a real variable for ftrace_nop in x86 Impact: avoid section mismatch warning, clean up The dynamic ftrace determines which nop is safe to use at start up. When it finds a safe nop for patching, it sets a pointer called ftrace_nop to point to the code. All call sites are then patched to this nop. Later, when tracing is turned on, this ftrace_nop variable is again used to compare the location to make sure it is a nop before we update it to an mcount call. If this fails just once, a warning is printed and ftrace is disabled. Rakib Mullick noted that the code that sets up the nop is a .init section where as the nop itself is in the .text section. This is needed because the nop is used later on after boot up. The problem is that the test of the nop jumps back to the setup code and causes a "section mismatch" warning. Rakib first recommended to convert the nop to .init.text, but as stated above, this would fail since that text is used later. The real solution is to extend Rabik's patch, and to make the ftrace_nop into an array, and just save the code from the assembly to this array. Now the section can stay as an init section, and we have a nop to use later on. Reported-by: Rakib Mullick Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index b1e5e2244eca..50ea0ac8c9bf 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -21,8 +21,7 @@ #include -/* Long is fine, even if it is only 4 bytes ;-) */ -static unsigned long *ftrace_nop; +static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; @@ -40,7 +39,7 @@ static int ftrace_calc_offset(long ip, long addr) unsigned char *ftrace_nop_replace(void) { - return (char *)ftrace_nop; + return ftrace_nop; } unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) @@ -125,9 +124,6 @@ int __init ftrace_dyn_arch_init(void *data) * TODO: check the cpuid to determine the best nop. */ asm volatile ( - "jmp ftrace_test_jmp\n" - /* This code needs to stay around */ - ".section .text, \"ax\"\n" "ftrace_test_jmp:" "jmp ftrace_test_p6nop\n" "nop\n" @@ -138,8 +134,6 @@ int __init ftrace_dyn_arch_init(void *data) "jmp 1f\n" "ftrace_test_nop5:" ".byte 0x66,0x66,0x66,0x66,0x90\n" - "jmp 1f\n" - ".previous\n" "1:" ".section .fixup, \"ax\"\n" "2: movl $1, %0\n" @@ -154,15 +148,15 @@ int __init ftrace_dyn_arch_init(void *data) switch (faulted) { case 0: pr_info("ftrace: converting mcount calls to 0f 1f 44 00 00\n"); - ftrace_nop = (unsigned long *)ftrace_test_p6nop; + memcpy(ftrace_nop, ftrace_test_p6nop, MCOUNT_INSN_SIZE); break; case 1: pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n"); - ftrace_nop = (unsigned long *)ftrace_test_nop5; + memcpy(ftrace_nop, ftrace_test_nop5, MCOUNT_INSN_SIZE); break; case 2: pr_info("ftrace: converting mcount calls to jmp . + 5\n"); - ftrace_nop = (unsigned long *)ftrace_test_jmp; + memcpy(ftrace_nop, ftrace_test_jmp, MCOUNT_INSN_SIZE); break; } -- cgit v1.2.3 From 04d2aac33eb54fd3084140f2db130530d71e97c6 Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sun, 5 Oct 2008 11:08:10 -0700 Subject: x86: corruption-check: fix some style issues Impact: cleanup Before moving the code to it's own file, fix some style issues in the corruption check code. Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 0fa6790c1dd3..4f38e0305b07 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -662,7 +662,7 @@ static void __init setup_bios_corruption_check(void) corruption_check_size = round_up(corruption_check_size, PAGE_SIZE); - while(addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) { + while (addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) { u64 size; addr = find_e820_area_size(addr, &size, PAGE_SIZE); @@ -701,11 +701,11 @@ void check_for_bios_corruption(void) if (!memory_corruption_check) return; - for(i = 0; i < num_scan_areas; i++) { + for (i = 0; i < num_scan_areas; i++) { unsigned long *addr = __va(scan_areas[i].addr); unsigned long size = scan_areas[i].size; - for(; size; addr++, size -= sizeof(unsigned long)) { + for (; size; addr++, size -= sizeof(unsigned long)) { if (!*addr) continue; printk(KERN_ERR "Corrupted low memory at %p (%lx phys) = %08lx\n", @@ -721,7 +721,8 @@ void check_for_bios_corruption(void) static void periodic_check_for_corruption(unsigned long data) { check_for_bios_corruption(); - mod_timer(&periodic_check_timer, round_jiffies(jiffies + corruption_check_period*HZ)); + mod_timer(&periodic_check_timer, + round_jiffies(jiffies + corruption_check_period*HZ)); } void start_periodic_check_for_corruption(void) -- cgit v1.2.3 From 6784f7d0a5016a397d38be1134e63fc784c1ca8e Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sun, 5 Oct 2008 11:33:42 -0700 Subject: x86: corruption check: move the corruption checks into their own file Impact: cleanup The corruption check code is rather sizable and it's likely to grow over time when we add checks for more types of corruptions (there's a few candidates in kerneloops.org that I want to add checks for)... so lets move it to its own file Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 1 + arch/x86/kernel/check.c | 158 +++++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/setup.c | 152 --------------------------------------------- 3 files changed, 159 insertions(+), 152 deletions(-) create mode 100644 arch/x86/kernel/check.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index d7e5a58ee22f..31fbcaf3df70 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -35,6 +35,7 @@ obj-y += bootflag.o e820.o obj-y += pci-dma.o quirks.o i8237.o topology.o kdebugfs.o obj-y += alternative.o i8253.o pci-nommu.o obj-y += tsc.o io_delay.o rtc.o +obj-y += check.o obj-$(CONFIG_X86_TRAMPOLINE) += trampoline.o obj-y += process.o diff --git a/arch/x86/kernel/check.c b/arch/x86/kernel/check.c new file mode 100644 index 000000000000..5056703e1b05 --- /dev/null +++ b/arch/x86/kernel/check.c @@ -0,0 +1,158 @@ +#include +#include + +#include +#include + +/* + * Some BIOSes seem to corrupt the low 64k of memory during events + * like suspend/resume and unplugging an HDMI cable. Reserve all + * remaining free memory in that area and fill it with a distinct + * pattern. + */ +#ifdef CONFIG_X86_CHECK_BIOS_CORRUPTION +#define MAX_SCAN_AREAS 8 + +static int __read_mostly memory_corruption_check = -1; + +static unsigned __read_mostly corruption_check_size = 64*1024; +static unsigned __read_mostly corruption_check_period = 60; /* seconds */ + +static struct e820entry scan_areas[MAX_SCAN_AREAS]; +static int num_scan_areas; + + +static int set_corruption_check(char *arg) +{ + char *end; + + memory_corruption_check = simple_strtol(arg, &end, 10); + + return (*end == 0) ? 0 : -EINVAL; +} +early_param("memory_corruption_check", set_corruption_check); + +static int set_corruption_check_period(char *arg) +{ + char *end; + + corruption_check_period = simple_strtoul(arg, &end, 10); + + return (*end == 0) ? 0 : -EINVAL; +} +early_param("memory_corruption_check_period", set_corruption_check_period); + +static int set_corruption_check_size(char *arg) +{ + char *end; + unsigned size; + + size = memparse(arg, &end); + + if (*end == '\0') + corruption_check_size = size; + + return (size == corruption_check_size) ? 0 : -EINVAL; +} +early_param("memory_corruption_check_size", set_corruption_check_size); + + +void __init setup_bios_corruption_check(void) +{ + u64 addr = PAGE_SIZE; /* assume first page is reserved anyway */ + + if (memory_corruption_check == -1) { + memory_corruption_check = +#ifdef CONFIG_X86_BOOTPARAM_MEMORY_CORRUPTION_CHECK + 1 +#else + 0 +#endif + ; + } + + if (corruption_check_size == 0) + memory_corruption_check = 0; + + if (!memory_corruption_check) + return; + + corruption_check_size = round_up(corruption_check_size, PAGE_SIZE); + + while (addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) { + u64 size; + addr = find_e820_area_size(addr, &size, PAGE_SIZE); + + if (addr == 0) + break; + + if ((addr + size) > corruption_check_size) + size = corruption_check_size - addr; + + if (size == 0) + break; + + e820_update_range(addr, size, E820_RAM, E820_RESERVED); + scan_areas[num_scan_areas].addr = addr; + scan_areas[num_scan_areas].size = size; + num_scan_areas++; + + /* Assume we've already mapped this early memory */ + memset(__va(addr), 0, size); + + addr += size; + } + + printk(KERN_INFO "Scanning %d areas for low memory corruption\n", + num_scan_areas); + update_e820(); +} + +static struct timer_list periodic_check_timer; + +void check_for_bios_corruption(void) +{ + int i; + int corruption = 0; + + if (!memory_corruption_check) + return; + + for (i = 0; i < num_scan_areas; i++) { + unsigned long *addr = __va(scan_areas[i].addr); + unsigned long size = scan_areas[i].size; + + for (; size; addr++, size -= sizeof(unsigned long)) { + if (!*addr) + continue; + printk(KERN_ERR "Corrupted low memory at %p (%lx phys) = %08lx\n", + addr, __pa(addr), *addr); + corruption = 1; + *addr = 0; + } + } + + WARN(corruption, KERN_ERR "Memory corruption detected in low memory\n"); +} + +static void periodic_check_for_corruption(unsigned long data) +{ + check_for_bios_corruption(); + mod_timer(&periodic_check_timer, + round_jiffies(jiffies + corruption_check_period*HZ)); +} + +void start_periodic_check_for_corruption(void) +{ + if (!memory_corruption_check || corruption_check_period == 0) + return; + + printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n", + corruption_check_period); + + init_timer(&periodic_check_timer); + periodic_check_timer.function = &periodic_check_for_corruption; + periodic_check_for_corruption(0); +} +#endif + diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 4f38e0305b07..af690aa593a9 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -587,158 +587,6 @@ static struct x86_quirks default_x86_quirks __initdata; struct x86_quirks *x86_quirks __initdata = &default_x86_quirks; -/* - * Some BIOSes seem to corrupt the low 64k of memory during events - * like suspend/resume and unplugging an HDMI cable. Reserve all - * remaining free memory in that area and fill it with a distinct - * pattern. - */ -#ifdef CONFIG_X86_CHECK_BIOS_CORRUPTION -#define MAX_SCAN_AREAS 8 - -static int __read_mostly memory_corruption_check = -1; - -static unsigned __read_mostly corruption_check_size = 64*1024; -static unsigned __read_mostly corruption_check_period = 60; /* seconds */ - -static struct e820entry scan_areas[MAX_SCAN_AREAS]; -static int num_scan_areas; - - -static int set_corruption_check(char *arg) -{ - char *end; - - memory_corruption_check = simple_strtol(arg, &end, 10); - - return (*end == 0) ? 0 : -EINVAL; -} -early_param("memory_corruption_check", set_corruption_check); - -static int set_corruption_check_period(char *arg) -{ - char *end; - - corruption_check_period = simple_strtoul(arg, &end, 10); - - return (*end == 0) ? 0 : -EINVAL; -} -early_param("memory_corruption_check_period", set_corruption_check_period); - -static int set_corruption_check_size(char *arg) -{ - char *end; - unsigned size; - - size = memparse(arg, &end); - - if (*end == '\0') - corruption_check_size = size; - - return (size == corruption_check_size) ? 0 : -EINVAL; -} -early_param("memory_corruption_check_size", set_corruption_check_size); - - -static void __init setup_bios_corruption_check(void) -{ - u64 addr = PAGE_SIZE; /* assume first page is reserved anyway */ - - if (memory_corruption_check == -1) { - memory_corruption_check = -#ifdef CONFIG_X86_BOOTPARAM_MEMORY_CORRUPTION_CHECK - 1 -#else - 0 -#endif - ; - } - - if (corruption_check_size == 0) - memory_corruption_check = 0; - - if (!memory_corruption_check) - return; - - corruption_check_size = round_up(corruption_check_size, PAGE_SIZE); - - while (addr < corruption_check_size && num_scan_areas < MAX_SCAN_AREAS) { - u64 size; - addr = find_e820_area_size(addr, &size, PAGE_SIZE); - - if (addr == 0) - break; - - if ((addr + size) > corruption_check_size) - size = corruption_check_size - addr; - - if (size == 0) - break; - - e820_update_range(addr, size, E820_RAM, E820_RESERVED); - scan_areas[num_scan_areas].addr = addr; - scan_areas[num_scan_areas].size = size; - num_scan_areas++; - - /* Assume we've already mapped this early memory */ - memset(__va(addr), 0, size); - - addr += size; - } - - printk(KERN_INFO "Scanning %d areas for low memory corruption\n", - num_scan_areas); - update_e820(); -} - -static struct timer_list periodic_check_timer; - -void check_for_bios_corruption(void) -{ - int i; - int corruption = 0; - - if (!memory_corruption_check) - return; - - for (i = 0; i < num_scan_areas; i++) { - unsigned long *addr = __va(scan_areas[i].addr); - unsigned long size = scan_areas[i].size; - - for (; size; addr++, size -= sizeof(unsigned long)) { - if (!*addr) - continue; - printk(KERN_ERR "Corrupted low memory at %p (%lx phys) = %08lx\n", - addr, __pa(addr), *addr); - corruption = 1; - *addr = 0; - } - } - - WARN(corruption, KERN_ERR "Memory corruption detected in low memory\n"); -} - -static void periodic_check_for_corruption(unsigned long data) -{ - check_for_bios_corruption(); - mod_timer(&periodic_check_timer, - round_jiffies(jiffies + corruption_check_period*HZ)); -} - -void start_periodic_check_for_corruption(void) -{ - if (!memory_corruption_check || corruption_check_period == 0) - return; - - printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n", - corruption_check_period); - - init_timer(&periodic_check_timer); - periodic_check_timer.function = &periodic_check_for_corruption; - periodic_check_for_corruption(0); -} -#endif - static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) { printk(KERN_NOTICE -- cgit v1.2.3 From 304e629bf4a3150a0bf6556fc45c52c5c082340f Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sun, 5 Oct 2008 12:09:03 -0700 Subject: x86: corruption check: run the corruption checks from a work queue Impact: change the implementation of the debug feature the periodic corruption checks are better off run from a work queue; there's nothing time critical about them and this way the amount of interrupt-context work is reduced. Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- arch/x86/kernel/check.c | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/check.c b/arch/x86/kernel/check.c index 5056703e1b05..55eed1752b43 100644 --- a/arch/x86/kernel/check.c +++ b/arch/x86/kernel/check.c @@ -1,6 +1,7 @@ #include #include - +#include +#include #include #include @@ -108,13 +109,14 @@ void __init setup_bios_corruption_check(void) update_e820(); } -static struct timer_list periodic_check_timer; void check_for_bios_corruption(void) { int i; int corruption = 0; + printk("dot\n"); + if (!memory_corruption_check) return; @@ -135,24 +137,29 @@ void check_for_bios_corruption(void) WARN(corruption, KERN_ERR "Memory corruption detected in low memory\n"); } -static void periodic_check_for_corruption(unsigned long data) +static void check_corruption(struct work_struct *dummy); +static DECLARE_DELAYED_WORK(bios_check_work, check_corruption); + +static void check_corruption(struct work_struct *dummy) { check_for_bios_corruption(); - mod_timer(&periodic_check_timer, - round_jiffies(jiffies + corruption_check_period*HZ)); + schedule_delayed_work(&bios_check_work, + round_jiffies_relative(corruption_check_period*HZ)); } -void start_periodic_check_for_corruption(void) +static int start_periodic_check_for_corruption(void) { if (!memory_corruption_check || corruption_check_period == 0) - return; + return 0; printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n", corruption_check_period); - init_timer(&periodic_check_timer); - periodic_check_timer.function = &periodic_check_for_corruption; - periodic_check_for_corruption(0); + /* First time we run the checks right away */ + schedule_delayed_work(&bios_check_work, 0); + return 0; } + +module_init(start_periodic_check_for_corruption); #endif -- cgit v1.2.3 From b43d196c4d3fe46d6dda7c987c47792612b80b1b Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sun, 5 Oct 2008 12:21:32 -0700 Subject: x86: corruption-check: some post-move cleanups Impact: cleanup now that the code is moved and converted to a work queue, there's some minor cleanups that can be done. Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 3 ++- arch/x86/kernel/check.c | 12 ++++-------- 2 files changed, 6 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 31fbcaf3df70..f63a8034fb82 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -35,7 +35,6 @@ obj-y += bootflag.o e820.o obj-y += pci-dma.o quirks.o i8237.o topology.o kdebugfs.o obj-y += alternative.o i8253.o pci-nommu.o obj-y += tsc.o io_delay.o rtc.o -obj-y += check.o obj-$(CONFIG_X86_TRAMPOLINE) += trampoline.o obj-y += process.o @@ -105,6 +104,8 @@ microcode-$(CONFIG_MICROCODE_INTEL) += microcode_intel.o microcode-$(CONFIG_MICROCODE_AMD) += microcode_amd.o obj-$(CONFIG_MICROCODE) += microcode.o +obj-$(CONFIG_X86_CHECK_BIOS_CORRUPTION) += check.o + ### # 64 bit specific files ifeq ($(CONFIG_X86_64),y) diff --git a/arch/x86/kernel/check.c b/arch/x86/kernel/check.c index 55eed1752b43..2ac0ab71412a 100644 --- a/arch/x86/kernel/check.c +++ b/arch/x86/kernel/check.c @@ -11,7 +11,6 @@ * remaining free memory in that area and fill it with a distinct * pattern. */ -#ifdef CONFIG_X86_CHECK_BIOS_CORRUPTION #define MAX_SCAN_AREAS 8 static int __read_mostly memory_corruption_check = -1; @@ -23,7 +22,7 @@ static struct e820entry scan_areas[MAX_SCAN_AREAS]; static int num_scan_areas; -static int set_corruption_check(char *arg) +static __init int set_corruption_check(char *arg) { char *end; @@ -33,7 +32,7 @@ static int set_corruption_check(char *arg) } early_param("memory_corruption_check", set_corruption_check); -static int set_corruption_check_period(char *arg) +static __init int set_corruption_check_period(char *arg) { char *end; @@ -43,7 +42,7 @@ static int set_corruption_check_period(char *arg) } early_param("memory_corruption_check_period", set_corruption_check_period); -static int set_corruption_check_size(char *arg) +static __init int set_corruption_check_size(char *arg) { char *end; unsigned size; @@ -115,8 +114,6 @@ void check_for_bios_corruption(void) int i; int corruption = 0; - printk("dot\n"); - if (!memory_corruption_check) return; @@ -134,7 +131,7 @@ void check_for_bios_corruption(void) } } - WARN(corruption, KERN_ERR "Memory corruption detected in low memory\n"); + WARN_ONCE(corruption, KERN_ERR "Memory corruption detected in low memory\n"); } static void check_corruption(struct work_struct *dummy); @@ -161,5 +158,4 @@ static int start_periodic_check_for_corruption(void) } module_init(start_periodic_check_for_corruption); -#endif -- cgit v1.2.3 From 6f290b4e016d6c61511542cf6d9ebdef1965978e Mon Sep 17 00:00:00 2001 From: Aristeu Rozanski Date: Mon, 27 Oct 2008 12:42:34 -0400 Subject: x86, NMI watchdog: add support to enable and disable IOAPIC NMI Impact: change/improve the way /proc/sys/kernel/nmi_watchdog works This patch adds support to enable/disable IOAPIC NMI watchdog in runtime via procfs. Signed-off-by: Aristeu Rozanski Cc: "Maciej W. Rozycki" Signed-off-by: Ingo Molnar --- arch/x86/kernel/nmi.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index 2c97f07f1c2c..2c005fac6171 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -340,6 +340,8 @@ void stop_apic_nmi_watchdog(void *unused) return; if (nmi_watchdog == NMI_LOCAL_APIC) lapic_watchdog_stop(); + else + __acpi_nmi_disable(NULL); __get_cpu_var(wd_enabled) = 0; atomic_dec(&nmi_active); } @@ -465,6 +467,24 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason) #ifdef CONFIG_SYSCTL +static void enable_ioapic_nmi_watchdog_single(void *unused) +{ + __get_cpu_var(wd_enabled) = 1; + atomic_inc(&nmi_active); + __acpi_nmi_enable(NULL); +} + +static void enable_ioapic_nmi_watchdog(void) +{ + on_each_cpu(enable_ioapic_nmi_watchdog_single, NULL, 1); + touch_nmi_watchdog(); +} + +static void disable_ioapic_nmi_watchdog(void) +{ + on_each_cpu(stop_apic_nmi_watchdog, NULL, 1); +} + static int __init setup_unknown_nmi_panic(char *str) { unknown_nmi_panic = 1; @@ -507,6 +527,11 @@ int proc_nmi_enabled(struct ctl_table *table, int write, struct file *file, enable_lapic_nmi_watchdog(); else disable_lapic_nmi_watchdog(); + } else if (nmi_watchdog == NMI_IO_APIC) { + if (nmi_watchdog_enabled) + enable_ioapic_nmi_watchdog(); + else + disable_ioapic_nmi_watchdog(); } else { printk(KERN_WARNING "NMI watchdog doesn't know what hardware to touch\n"); -- cgit v1.2.3 From 7d5a78cd98c3a5eb83bd6a061c5ea6ef1e9b8fcb Mon Sep 17 00:00:00 2001 From: Aristeu Rozanski Date: Mon, 27 Oct 2008 12:42:35 -0400 Subject: x86, NMI watchdog: disable NMIs on LVT0 in case NMI watchdog is not working Impact: change NMI watchdog detection and disabling sequence Currently, if the NMI watchdog fails using IOAPIC method, it'll only disable interrupts on 8259 if the timer is passing thru it. This patch disables NMI delivery on LINT0 if the NMI watchdog initial test fails, just for safety. Signed-off-by: Aristeu Rozanski Cc: "Maciej W. Rozycki" Signed-off-by: Ingo Molnar --- arch/x86/kernel/nmi.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index 2c005fac6171..13316cf57cdb 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -131,6 +131,11 @@ static void report_broken_nmi(int cpu, int *prev_nmi_count) atomic_dec(&nmi_active); } +static void __acpi_nmi_disable(void *__unused) +{ + apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED); +} + int __init check_nmi_watchdog(void) { unsigned int *prev_nmi_count; @@ -179,8 +184,12 @@ int __init check_nmi_watchdog(void) kfree(prev_nmi_count); return 0; error: - if (nmi_watchdog == NMI_IO_APIC && !timer_through_8259) - disable_8259A_irq(0); + if (nmi_watchdog == NMI_IO_APIC) { + if (!timer_through_8259) + disable_8259A_irq(0); + on_each_cpu(__acpi_nmi_disable, NULL, 1); + } + #ifdef CONFIG_X86_32 timer_ack = 0; #endif @@ -285,11 +294,6 @@ void acpi_nmi_enable(void) on_each_cpu(__acpi_nmi_enable, NULL, 1); } -static void __acpi_nmi_disable(void *__unused) -{ - apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED); -} - /* * Disable timer based NMIs on all CPUs: */ -- cgit v1.2.3 From 878719e831d9e076961aa15d4049a57a6668c67a Mon Sep 17 00:00:00 2001 From: Neil Horman Date: Thu, 23 Oct 2008 10:40:06 -0400 Subject: x86: unify appropriate bits from dumpstack_32 and dumpstack_64 Impact: cleanup As promised, now that dumpstack_32 and dumpstack_64 have so many bits in common, we should merge the in-sync bits into a common file, to prevent them from diverging again. This patch removes bits which are common between dumpstack_32.c and dumpstack_64.c and places them in a common dumpstack.c which is built for both 32 and 64 bit arches. Signed-off-by: Neil Horman Acked-by: Alexander van Heukelum Signed-off-by: Ingo Molnar Makefile | 2 arch/x86/kernel/Makefile | 2 arch/x86/kernel/Makefile | 2 arch/x86/kernel/Makefile | 2 arch/x86/kernel/Makefile | 2 arch/x86/kernel/Makefile | 2 arch/x86/kernel/dumpstack.c | 319 +++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/dumpstack.h | 39 +++++ arch/x86/kernel/dumpstack_32.c | 294 ------------------------------------- arch/x86/kernel/dumpstack_64.c | 285 ------------------------------------ 5 files changed, 363 insertions(+), 576 deletions(-) --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/dumpstack.c | 319 +++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/dumpstack.h | 39 +++++ arch/x86/kernel/dumpstack_32.c | 294 +------------------------------------ arch/x86/kernel/dumpstack_64.c | 285 +----------------------------------- 5 files changed, 363 insertions(+), 576 deletions(-) create mode 100644 arch/x86/kernel/dumpstack.c create mode 100644 arch/x86/kernel/dumpstack.h (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index d7e5a58ee22f..db3216a9d2b9 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -24,7 +24,7 @@ CFLAGS_tsc.o := $(nostackp) obj-y := process_$(BITS).o signal_$(BITS).o entry_$(BITS).o obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o -obj-y += time_$(BITS).o ioport.o ldt.o +obj-y += time_$(BITS).o ioport.o ldt.o dumpstack.o obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o obj-$(CONFIG_X86_VISWS) += visws_quirks.o obj-$(CONFIG_X86_32) += probe_roms_32.o diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c new file mode 100644 index 000000000000..5962176dfabb --- /dev/null +++ b/arch/x86/kernel/dumpstack.c @@ -0,0 +1,319 @@ +/* + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "dumpstack.h" + +int panic_on_unrecovered_nmi; +unsigned int code_bytes = 64; +int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; +static int die_counter; + +void printk_address(unsigned long address, int reliable) +{ + printk(" [<%p>] %s%pS\n", (void *) address, + reliable ? "" : "? ", (void *) address); +} + +/* + * x86-64 can have up to three kernel stacks: + * process stack + * interrupt stack + * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack + */ + +static inline int valid_stack_ptr(struct thread_info *tinfo, + void *p, unsigned int size, void *end) +{ + void *t = tinfo; + if (end) { + if (p < end && p >= (end-THREAD_SIZE)) + return 1; + else + return 0; + } + return p > t && p < t + THREAD_SIZE - size; +} + +unsigned long +print_context_stack(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end) +{ + struct stack_frame *frame = (struct stack_frame *)bp; + + while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { + unsigned long addr; + + addr = *stack; + if (__kernel_text_address(addr)) { + if ((unsigned long) stack == bp + sizeof(long)) { + ops->address(data, addr, 1); + frame = frame->next_frame; + bp = (unsigned long) frame; + } else { + ops->address(data, addr, bp == 0); + } + } + stack++; + } + return bp; +} + + +static void +print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) +{ + printk(data); + print_symbol(msg, symbol); + printk("\n"); +} + +static void print_trace_warning(void *data, char *msg) +{ + printk("%s%s\n", (char *)data, msg); +} + +static int print_trace_stack(void *data, char *name) +{ + printk("%s <%s> ", (char *)data, name); + return 0; +} + +/* + * Print one address/symbol entries per line. + */ +static void print_trace_address(void *data, unsigned long addr, int reliable) +{ + touch_nmi_watchdog(); + printk(data); + printk_address(addr, reliable); +} + +static const struct stacktrace_ops print_trace_ops = { + .warning = print_trace_warning, + .warning_symbol = print_trace_warning_symbol, + .stack = print_trace_stack, + .address = print_trace_address, +}; + +void +show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp, char *log_lvl) +{ + printk("%sCall Trace:\n", log_lvl); + dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); +} + +void show_trace(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp) +{ + show_trace_log_lvl(task, regs, stack, bp, ""); +} + +void show_stack(struct task_struct *task, unsigned long *sp) +{ + show_stack_log_lvl(task, NULL, sp, 0, ""); +} + +/* + * The architecture-independent dump_stack generator + */ +void dump_stack(void) +{ + unsigned long bp = 0; + unsigned long stack; + +#ifdef CONFIG_FRAME_POINTER + if (!bp) + get_bp(bp); +#endif + + printk("Pid: %d, comm: %.20s %s %s %.*s\n", + current->pid, current->comm, print_tainted(), + init_utsname()->release, + (int)strcspn(init_utsname()->version, " "), + init_utsname()->version); + show_trace(NULL, NULL, &stack, bp); +} +EXPORT_SYMBOL(dump_stack); + +static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; +static int die_owner = -1; +static unsigned int die_nest_count; + +unsigned __kprobes long oops_begin(void) +{ + int cpu; + unsigned long flags; + + oops_enter(); + + /* racy, but better than risking deadlock. */ + raw_local_irq_save(flags); + cpu = smp_processor_id(); + if (!__raw_spin_trylock(&die_lock)) { + if (cpu == die_owner) + /* nested oops. should stop eventually */; + else + __raw_spin_lock(&die_lock); + } + die_nest_count++; + die_owner = cpu; + console_verbose(); + bust_spinlocks(1); + return flags; +} + +void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) +{ + if (regs && kexec_should_crash(current)) + crash_kexec(regs); + + bust_spinlocks(0); + die_owner = -1; + add_taint(TAINT_DIE); + die_nest_count--; + if (!die_nest_count) + /* Nest count reaches zero, release the lock. */ + __raw_spin_unlock(&die_lock); + raw_local_irq_restore(flags); + oops_exit(); + + if (!signr) + return; + if (in_interrupt()) + panic("Fatal exception in interrupt"); + if (panic_on_oops) + panic("Fatal exception"); + do_exit(signr); +} + +int __kprobes __die(const char *str, struct pt_regs *regs, long err) +{ +#ifdef CONFIG_X86_32 + unsigned short ss; + unsigned long sp; +#endif + printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); +#ifdef CONFIG_PREEMPT + printk("PREEMPT "); +#endif +#ifdef CONFIG_SMP + printk("SMP "); +#endif +#ifdef CONFIG_DEBUG_PAGEALLOC + printk("DEBUG_PAGEALLOC"); +#endif + printk("\n"); + sysfs_printk_last_file(); + if (notify_die(DIE_OOPS, str, regs, err, + current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) + return 1; + + show_registers(regs); +#ifdef CONFIG_X86_32 + sp = (unsigned long) (®s->sp); + savesegment(ss, ss); + if (user_mode(regs)) { + sp = regs->sp; + ss = regs->ss & 0xffff; + } + printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip); + print_symbol("%s", regs->ip); + printk(" SS:ESP %04x:%08lx\n", ss, sp); +#else + /* Executive summary in case the oops scrolled away */ + printk(KERN_ALERT "RIP "); + printk_address(regs->ip, 1); + printk(" RSP <%016lx>\n", regs->sp); +#endif + return 0; +} + +/* + * This is gone through when something in the kernel has done something bad + * and is about to be terminated: + */ +void die(const char *str, struct pt_regs *regs, long err) +{ + unsigned long flags = oops_begin(); + int sig = SIGSEGV; + + if (!user_mode_vm(regs)) + report_bug(regs->ip, regs); + + if (__die(str, regs, err)) + sig = 0; + oops_end(flags, regs, sig); +} + +void notrace __kprobes +die_nmi(char *str, struct pt_regs *regs, int do_panic) +{ + unsigned long flags; + + if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) + return; + + /* + * We are in trouble anyway, lets at least try + * to get a message out. + */ + flags = oops_begin(); + printk(KERN_EMERG "%s", str); + printk(" on CPU%d, ip %08lx, registers:\n", + smp_processor_id(), regs->ip); + show_registers(regs); + oops_end(flags, regs, 0); + if (do_panic || panic_on_oops) + panic("Non maskable interrupt"); + nmi_exit(); + local_irq_enable(); + do_exit(SIGBUS); +} + +static int __init oops_setup(char *s) +{ + if (!s) + return -EINVAL; + if (!strcmp(s, "panic")) + panic_on_oops = 1; + return 0; +} +early_param("oops", oops_setup); + +static int __init kstack_setup(char *s) +{ + if (!s) + return -EINVAL; + kstack_depth_to_print = simple_strtoul(s, NULL, 0); + return 0; +} +early_param("kstack", kstack_setup); + +static int __init code_bytes_setup(char *s) +{ + code_bytes = simple_strtoul(s, NULL, 0); + if (code_bytes > 8192) + code_bytes = 8192; + + return 1; +} +__setup("code_bytes=", code_bytes_setup); diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h new file mode 100644 index 000000000000..3119a801c32b --- /dev/null +++ b/arch/x86/kernel/dumpstack.h @@ -0,0 +1,39 @@ +/* + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs + */ + +#ifndef DUMPSTACK_H +#define DUMPSTACK_H + +#ifdef CONFIG_X86_32 +#define STACKSLOTS_PER_LINE 8 +#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :) +#else +#define STACKSLOTS_PER_LINE 4 +#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) +#endif + +extern unsigned long +print_context_stack(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end); + +extern void +show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp, char *log_lvl); + +extern void +show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *sp, unsigned long bp, char *log_lvl); + +extern unsigned int code_bytes; +extern int kstack_depth_to_print; + +/* The form of the top of the frame on the stack */ +struct stack_frame { + struct stack_frame *next_frame; + unsigned long return_address; +}; +#endif diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index f2046c5752d0..7b031b106ec8 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -17,64 +17,7 @@ #include -#define STACKSLOTS_PER_LINE 8 -#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :) - -int panic_on_unrecovered_nmi; -int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; -static unsigned int code_bytes = 64; -static int die_counter; - -void printk_address(unsigned long address, int reliable) -{ - printk(" [<%p>] %s%pS\n", (void *) address, - reliable ? "" : "? ", (void *) address); -} - -static inline int valid_stack_ptr(struct thread_info *tinfo, - void *p, unsigned int size, void *end) -{ - void *t = tinfo; - if (end) { - if (p < end && p >= (end-THREAD_SIZE)) - return 1; - else - return 0; - } - return p > t && p < t + THREAD_SIZE - size; -} - -/* The form of the top of the frame on the stack */ -struct stack_frame { - struct stack_frame *next_frame; - unsigned long return_address; -}; - -static inline unsigned long -print_context_stack(struct thread_info *tinfo, - unsigned long *stack, unsigned long bp, - const struct stacktrace_ops *ops, void *data, - unsigned long *end) -{ - struct stack_frame *frame = (struct stack_frame *)bp; - - while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { - unsigned long addr; - - addr = *stack; - if (__kernel_text_address(addr)) { - if ((unsigned long) stack == bp + sizeof(long)) { - ops->address(data, addr, 1); - frame = frame->next_frame; - bp = (unsigned long) frame; - } else { - ops->address(data, addr, bp == 0); - } - } - stack++; - } - return bp; -} +#include "dumpstack.h" void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, @@ -119,57 +62,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, } EXPORT_SYMBOL(dump_trace); -static void -print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) -{ - printk(data); - print_symbol(msg, symbol); - printk("\n"); -} - -static void print_trace_warning(void *data, char *msg) -{ - printk("%s%s\n", (char *)data, msg); -} - -static int print_trace_stack(void *data, char *name) -{ - printk("%s <%s> ", (char *)data, name); - return 0; -} - -/* - * Print one address/symbol entries per line. - */ -static void print_trace_address(void *data, unsigned long addr, int reliable) -{ - touch_nmi_watchdog(); - printk(data); - printk_address(addr, reliable); -} - -static const struct stacktrace_ops print_trace_ops = { - .warning = print_trace_warning, - .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, -}; - -static void -show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp, char *log_lvl) -{ - printk("%sCall Trace:\n", log_lvl); - dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); -} - -void show_trace(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp) -{ - show_trace_log_lvl(task, regs, stack, bp, ""); -} - -static void +void show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *sp, unsigned long bp, char *log_lvl) { @@ -196,33 +89,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, show_trace_log_lvl(task, regs, sp, bp, log_lvl); } -void show_stack(struct task_struct *task, unsigned long *sp) -{ - show_stack_log_lvl(task, NULL, sp, 0, ""); -} - -/* - * The architecture-independent dump_stack generator - */ -void dump_stack(void) -{ - unsigned long bp = 0; - unsigned long stack; - -#ifdef CONFIG_FRAME_POINTER - if (!bp) - get_bp(bp); -#endif - - printk("Pid: %d, comm: %.20s %s %s %.*s\n", - current->pid, current->comm, print_tainted(), - init_utsname()->release, - (int)strcspn(init_utsname()->version, " "), - init_utsname()->version); - show_trace(NULL, NULL, &stack, bp); -} - -EXPORT_SYMBOL(dump_stack); void show_registers(struct pt_regs *regs) { @@ -283,159 +149,3 @@ int is_valid_bugaddr(unsigned long ip) return ud2 == 0x0b0f; } -static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; -static int die_owner = -1; -static unsigned int die_nest_count; - -unsigned __kprobes long oops_begin(void) -{ - int cpu; - unsigned long flags; - - oops_enter(); - - /* racy, but better than risking deadlock. */ - raw_local_irq_save(flags); - cpu = smp_processor_id(); - if (!__raw_spin_trylock(&die_lock)) { - if (cpu == die_owner) - /* nested oops. should stop eventually */; - else - __raw_spin_lock(&die_lock); - } - die_nest_count++; - die_owner = cpu; - console_verbose(); - bust_spinlocks(1); - return flags; -} - -void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) -{ - if (regs && kexec_should_crash(current)) - crash_kexec(regs); - - bust_spinlocks(0); - die_owner = -1; - add_taint(TAINT_DIE); - die_nest_count--; - if (!die_nest_count) - /* Nest count reaches zero, release the lock. */ - __raw_spin_unlock(&die_lock); - raw_local_irq_restore(flags); - oops_exit(); - - if (!signr) - return; - if (in_interrupt()) - panic("Fatal exception in interrupt"); - if (panic_on_oops) - panic("Fatal exception"); - do_exit(signr); -} - -int __kprobes __die(const char *str, struct pt_regs *regs, long err) -{ - unsigned short ss; - unsigned long sp; - - printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); -#ifdef CONFIG_PREEMPT - printk("PREEMPT "); -#endif -#ifdef CONFIG_SMP - printk("SMP "); -#endif -#ifdef CONFIG_DEBUG_PAGEALLOC - printk("DEBUG_PAGEALLOC"); -#endif - printk("\n"); - sysfs_printk_last_file(); - if (notify_die(DIE_OOPS, str, regs, err, - current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) - return 1; - - show_registers(regs); - /* Executive summary in case the oops scrolled away */ - sp = (unsigned long) (®s->sp); - savesegment(ss, ss); - if (user_mode(regs)) { - sp = regs->sp; - ss = regs->ss & 0xffff; - } - printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip); - print_symbol("%s", regs->ip); - printk(" SS:ESP %04x:%08lx\n", ss, sp); - return 0; -} - -/* - * This is gone through when something in the kernel has done something bad - * and is about to be terminated: - */ -void die(const char *str, struct pt_regs *regs, long err) -{ - unsigned long flags = oops_begin(); - int sig = SIGSEGV; - - if (!user_mode_vm(regs)) - report_bug(regs->ip, regs); - - if (__die(str, regs, err)) - sig = 0; - oops_end(flags, regs, sig); -} - -void notrace __kprobes -die_nmi(char *str, struct pt_regs *regs, int do_panic) -{ - unsigned long flags; - - if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) - return; - - /* - * We are in trouble anyway, lets at least try - * to get a message out. - */ - flags = oops_begin(); - printk(KERN_EMERG "%s", str); - printk(" on CPU%d, ip %08lx, registers:\n", - smp_processor_id(), regs->ip); - show_registers(regs); - oops_end(flags, regs, 0); - if (do_panic || panic_on_oops) - panic("Non maskable interrupt"); - nmi_exit(); - local_irq_enable(); - do_exit(SIGBUS); -} - -static int __init oops_setup(char *s) -{ - if (!s) - return -EINVAL; - if (!strcmp(s, "panic")) - panic_on_oops = 1; - return 0; -} -early_param("oops", oops_setup); - -static int __init kstack_setup(char *s) -{ - if (!s) - return -EINVAL; - kstack_depth_to_print = simple_strtoul(s, NULL, 0); - return 0; -} -early_param("kstack", kstack_setup); - -static int __init code_bytes_setup(char *s) -{ - code_bytes = simple_strtoul(s, NULL, 0); - if (code_bytes > 8192) - code_bytes = 8192; - - return 1; -} -__setup("code_bytes=", code_bytes_setup); diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 28c67aae5562..33ff10287a5d 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -17,19 +17,7 @@ #include -#define STACKSLOTS_PER_LINE 4 -#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) - -int panic_on_unrecovered_nmi; -int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; -static unsigned int code_bytes = 64; -static int die_counter; - -void printk_address(unsigned long address, int reliable) -{ - printk(" [<%p>] %s%pS\n", (void *) address, - reliable ? "" : "? ", (void *) address); -} +#include "dumpstack.h" static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, unsigned *usedp, char **idp) @@ -113,51 +101,6 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack */ -static inline int valid_stack_ptr(struct thread_info *tinfo, - void *p, unsigned int size, void *end) -{ - void *t = tinfo; - if (end) { - if (p < end && p >= (end-THREAD_SIZE)) - return 1; - else - return 0; - } - return p > t && p < t + THREAD_SIZE - size; -} - -/* The form of the top of the frame on the stack */ -struct stack_frame { - struct stack_frame *next_frame; - unsigned long return_address; -}; - -static inline unsigned long -print_context_stack(struct thread_info *tinfo, - unsigned long *stack, unsigned long bp, - const struct stacktrace_ops *ops, void *data, - unsigned long *end) -{ - struct stack_frame *frame = (struct stack_frame *)bp; - - while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { - unsigned long addr; - - addr = *stack; - if (__kernel_text_address(addr)) { - if ((unsigned long) stack == bp + sizeof(long)) { - ops->address(data, addr, 1); - frame = frame->next_frame; - bp = (unsigned long) frame; - } else { - ops->address(data, addr, bp == 0); - } - } - stack++; - } - return bp; -} - void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data) @@ -248,57 +191,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, } EXPORT_SYMBOL(dump_trace); -static void -print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) -{ - printk(data); - print_symbol(msg, symbol); - printk("\n"); -} - -static void print_trace_warning(void *data, char *msg) -{ - printk("%s%s\n", (char *)data, msg); -} - -static int print_trace_stack(void *data, char *name) -{ - printk("%s <%s> ", (char *)data, name); - return 0; -} - -/* - * Print one address/symbol entries per line. - */ -static void print_trace_address(void *data, unsigned long addr, int reliable) -{ - touch_nmi_watchdog(); - printk(data); - printk_address(addr, reliable); -} - -static const struct stacktrace_ops print_trace_ops = { - .warning = print_trace_warning, - .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, -}; - -static void -show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp, char *log_lvl) -{ - printk("%sCall Trace:\n", log_lvl); - dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); -} - -void show_trace(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp) -{ - show_trace_log_lvl(task, regs, stack, bp, ""); -} - -static void +void show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *sp, unsigned long bp, char *log_lvl) { @@ -342,33 +235,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, show_trace_log_lvl(task, regs, sp, bp, log_lvl); } -void show_stack(struct task_struct *task, unsigned long *sp) -{ - show_stack_log_lvl(task, NULL, sp, 0, ""); -} - -/* - * The architecture-independent dump_stack generator - */ -void dump_stack(void) -{ - unsigned long bp = 0; - unsigned long stack; - -#ifdef CONFIG_FRAME_POINTER - if (!bp) - get_bp(bp); -#endif - - printk("Pid: %d, comm: %.20s %s %s %.*s\n", - current->pid, current->comm, print_tainted(), - init_utsname()->release, - (int)strcspn(init_utsname()->version, " "), - init_utsname()->version); - show_trace(NULL, NULL, &stack, bp); -} -EXPORT_SYMBOL(dump_stack); - void show_registers(struct pt_regs *regs) { int i; @@ -429,150 +295,3 @@ int is_valid_bugaddr(unsigned long ip) return ud2 == 0x0b0f; } -static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; -static int die_owner = -1; -static unsigned int die_nest_count; - -unsigned __kprobes long oops_begin(void) -{ - int cpu; - unsigned long flags; - - oops_enter(); - - /* racy, but better than risking deadlock. */ - raw_local_irq_save(flags); - cpu = smp_processor_id(); - if (!__raw_spin_trylock(&die_lock)) { - if (cpu == die_owner) - /* nested oops. should stop eventually */; - else - __raw_spin_lock(&die_lock); - } - die_nest_count++; - die_owner = cpu; - console_verbose(); - bust_spinlocks(1); - return flags; -} - -void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) -{ - if (regs && kexec_should_crash(current)) - crash_kexec(regs); - - bust_spinlocks(0); - die_owner = -1; - add_taint(TAINT_DIE); - die_nest_count--; - if (!die_nest_count) - /* Nest count reaches zero, release the lock. */ - __raw_spin_unlock(&die_lock); - raw_local_irq_restore(flags); - oops_exit(); - - if (!signr) - return; - if (in_interrupt()) - panic("Fatal exception in interrupt"); - if (panic_on_oops) - panic("Fatal exception"); - do_exit(signr); -} - -int __kprobes __die(const char *str, struct pt_regs *regs, long err) -{ - printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); -#ifdef CONFIG_PREEMPT - printk("PREEMPT "); -#endif -#ifdef CONFIG_SMP - printk("SMP "); -#endif -#ifdef CONFIG_DEBUG_PAGEALLOC - printk("DEBUG_PAGEALLOC"); -#endif - printk("\n"); - sysfs_printk_last_file(); - if (notify_die(DIE_OOPS, str, regs, err, - current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) - return 1; - - show_registers(regs); - /* Executive summary in case the oops scrolled away */ - printk(KERN_ALERT "RIP "); - printk_address(regs->ip, 1); - printk(" RSP <%016lx>\n", regs->sp); - return 0; -} - -/* - * This is gone through when something in the kernel has done something bad - * and is about to be terminated: - */ -void die(const char *str, struct pt_regs *regs, long err) -{ - unsigned long flags = oops_begin(); - int sig = SIGSEGV; - - if (!user_mode_vm(regs)) - report_bug(regs->ip, regs); - - if (__die(str, regs, err)) - sig = 0; - oops_end(flags, regs, sig); -} - -void notrace __kprobes -die_nmi(char *str, struct pt_regs *regs, int do_panic) -{ - unsigned long flags; - - if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) - return; - - /* - * We are in trouble anyway, lets at least try - * to get a message out. - */ - flags = oops_begin(); - printk(KERN_EMERG "%s", str); - printk(" on CPU%d, ip %08lx, registers:\n", - smp_processor_id(), regs->ip); - show_registers(regs); - oops_end(flags, regs, 0); - if (do_panic || panic_on_oops) - panic("Non maskable interrupt"); - nmi_exit(); - local_irq_enable(); - do_exit(SIGBUS); -} - -static int __init oops_setup(char *s) -{ - if (!s) - return -EINVAL; - if (!strcmp(s, "panic")) - panic_on_oops = 1; - return 0; -} -early_param("oops", oops_setup); - -static int __init kstack_setup(char *s) -{ - if (!s) - return -EINVAL; - kstack_depth_to_print = simple_strtoul(s, NULL, 0); - return 0; -} -early_param("kstack", kstack_setup); - -static int __init code_bytes_setup(char *s) -{ - code_bytes = simple_strtoul(s, NULL, 0); - if (code_bytes > 8192) - code_bytes = 8192; - - return 1; -} -__setup("code_bytes=", code_bytes_setup); -- cgit v1.2.3 From 69a72a0e9337aad8c730e8e9942d5aa022bc4c5c Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Mon, 27 Oct 2008 07:51:20 -0700 Subject: x86/uv: update SCIR driver to use the idle_cpu() function Impact: cleanup Change UV heartbeat function to use idle_cpu to determine cpu's "idleness". Realign uv_hub definitions. Signed-of-by: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 84367d84bb10..85fb7dd48f67 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -370,8 +370,8 @@ static void uv_heartbeat(unsigned long ignored) /* flip heartbeat bit */ bits ^= SCIR_CPU_HEARTBEAT; - /* are we the idle thread? */ - if (current->pid == 0) + /* is this cpu idle? */ + if (idle_cpu(raw_smp_processor_id())) bits &= ~SCIR_CPU_ACTIVITY; else bits |= SCIR_CPU_ACTIVITY; -- cgit v1.2.3 From 30604bb410b53efa9c93ee8f03d7aa7494094faa Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 14 Oct 2008 18:59:18 -0700 Subject: x86: break up mtrr_cleanup() into several small functions. Ingo said mtrr_cleanup() is big and ugly. so break it up into more functions and make it more readable. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mtrr/main.c | 346 ++++++++++++++++++++-------------------- 1 file changed, 171 insertions(+), 175 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c index c78c04821ea1..1159e269e596 100644 --- a/arch/x86/kernel/cpu/mtrr/main.c +++ b/arch/x86/kernel/cpu/mtrr/main.c @@ -803,6 +803,7 @@ x86_get_mtrr_mem_range(struct res_range *range, int nr_range, } static struct res_range __initdata range[RANGE_NUM]; +static int __initdata nr_range; #ifdef CONFIG_MTRR_SANITIZER @@ -1206,39 +1207,43 @@ struct mtrr_cleanup_result { #define PSHIFT (PAGE_SHIFT - 10) static struct mtrr_cleanup_result __initdata result[NUM_RESULT]; -static struct res_range __initdata range_new[RANGE_NUM]; static unsigned long __initdata min_loss_pfn[RANGE_NUM]; -static int __init mtrr_cleanup(unsigned address_bits) +static void __init print_out_mtrr_range_state(void) { - unsigned long extra_remove_base, extra_remove_size; - unsigned long base, size, def, dummy; - mtrr_type type; - int nr_range, nr_range_new; - u64 chunk_size, gran_size; - unsigned long range_sums, range_sums_new; - int index_good; - int num_reg_good; int i; + char start_factor = 'K', size_factor = 'K'; + unsigned long start_base, size_base; + mtrr_type type; - /* extra one for all 0 */ - int num[MTRR_NUM_TYPES + 1]; + for (i = 0; i < num_var_ranges; i++) { - if (!is_cpu(INTEL) || enable_mtrr_cleanup < 1) - return 0; - rdmsr(MTRRdefType_MSR, def, dummy); - def &= 0xff; - if (def != MTRR_TYPE_UNCACHABLE) - return 0; + size_base = range_state[i].size_pfn << (PAGE_SHIFT - 10); + if (!size_base) + continue; - /* get it and store it aside */ - memset(range_state, 0, sizeof(range_state)); - for (i = 0; i < num_var_ranges; i++) { - mtrr_if->get(i, &base, &size, &type); - range_state[i].base_pfn = base; - range_state[i].size_pfn = size; - range_state[i].type = type; + size_base = to_size_factor(size_base, &size_factor), + start_base = range_state[i].base_pfn << (PAGE_SHIFT - 10); + start_base = to_size_factor(start_base, &start_factor), + type = range_state[i].type; + + printk(KERN_DEBUG "reg %d, base: %ld%cB, range: %ld%cB, type %s\n", + i, start_base, start_factor, + size_base, size_factor, + (type == MTRR_TYPE_UNCACHABLE) ? "UC" : + ((type == MTRR_TYPE_WRPROT) ? "WP" : + ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other")) + ); } +} + +static int __init mtrr_need_cleanup(void) +{ + int i; + mtrr_type type; + unsigned long size; + /* extra one for all 0 */ + int num[MTRR_NUM_TYPES + 1]; /* check entries number */ memset(num, 0, sizeof(num)); @@ -1263,29 +1268,133 @@ static int __init mtrr_cleanup(unsigned address_bits) num_var_ranges - num[MTRR_NUM_TYPES]) return 0; - /* print original var MTRRs at first, for debugging: */ - printk(KERN_DEBUG "original variable MTRRs\n"); - for (i = 0; i < num_var_ranges; i++) { - char start_factor = 'K', size_factor = 'K'; - unsigned long start_base, size_base; + return 1; +} - size_base = range_state[i].size_pfn << (PAGE_SHIFT - 10); - if (!size_base) - continue; +static unsigned long __initdata range_sums; +static void __init mtrr_calc_range_state(u64 chunk_size, u64 gran_size, + unsigned long extra_remove_base, + unsigned long extra_remove_size, + int i) +{ + int num_reg; + static struct res_range range_new[RANGE_NUM]; + static int nr_range_new; + unsigned long range_sums_new; + + /* convert ranges to var ranges state */ + num_reg = x86_setup_var_mtrrs(range, nr_range, + chunk_size, gran_size); + + /* we got new setting in range_state, check it */ + memset(range_new, 0, sizeof(range_new)); + nr_range_new = x86_get_mtrr_mem_range(range_new, 0, + extra_remove_base, extra_remove_size); + range_sums_new = sum_ranges(range_new, nr_range_new); + + result[i].chunk_sizek = chunk_size >> 10; + result[i].gran_sizek = gran_size >> 10; + result[i].num_reg = num_reg; + if (range_sums < range_sums_new) { + result[i].lose_cover_sizek = + (range_sums_new - range_sums) << PSHIFT; + result[i].bad = 1; + } else + result[i].lose_cover_sizek = + (range_sums - range_sums_new) << PSHIFT; - size_base = to_size_factor(size_base, &size_factor), - start_base = range_state[i].base_pfn << (PAGE_SHIFT - 10); - start_base = to_size_factor(start_base, &start_factor), - type = range_state[i].type; + /* double check it */ + if (!result[i].bad && !result[i].lose_cover_sizek) { + if (nr_range_new != nr_range || + memcmp(range, range_new, sizeof(range))) + result[i].bad = 1; + } - printk(KERN_DEBUG "reg %d, base: %ld%cB, range: %ld%cB, type %s\n", - i, start_base, start_factor, - size_base, size_factor, - (type == MTRR_TYPE_UNCACHABLE) ? "UC" : - ((type == MTRR_TYPE_WRPROT) ? "WP" : - ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other")) - ); + if (!result[i].bad && (range_sums - range_sums_new < + min_loss_pfn[num_reg])) { + min_loss_pfn[num_reg] = + range_sums - range_sums_new; } +} + +static void __init mtrr_print_out_one_result(int i) +{ + char gran_factor, chunk_factor, lose_factor; + unsigned long gran_base, chunk_base, lose_base; + + gran_base = to_size_factor(result[i].gran_sizek, &gran_factor), + chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor), + lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor), + printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t", + result[i].bad ? "*BAD*" : " ", + gran_base, gran_factor, chunk_base, chunk_factor); + printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n", + result[i].num_reg, result[i].bad ? "-" : "", + lose_base, lose_factor); +} + +static int __init mtrr_search_optimal_index(void) +{ + int i; + int num_reg_good; + int index_good; + + if (nr_mtrr_spare_reg >= num_var_ranges) + nr_mtrr_spare_reg = num_var_ranges - 1; + num_reg_good = -1; + for (i = num_var_ranges - nr_mtrr_spare_reg; i > 0; i--) { + if (!min_loss_pfn[i]) + num_reg_good = i; + } + + index_good = -1; + if (num_reg_good != -1) { + for (i = 0; i < NUM_RESULT; i++) { + if (!result[i].bad && + result[i].num_reg == num_reg_good && + !result[i].lose_cover_sizek) { + index_good = i; + break; + } + } + } + + return index_good; +} + + +static int __init mtrr_cleanup(unsigned address_bits) +{ + unsigned long extra_remove_base, extra_remove_size; + unsigned long base, size, def, dummy; + mtrr_type type; + u64 chunk_size, gran_size; + int index_good; + int i; + + if (!is_cpu(INTEL) || enable_mtrr_cleanup < 1) + return 0; + rdmsr(MTRRdefType_MSR, def, dummy); + def &= 0xff; + if (def != MTRR_TYPE_UNCACHABLE) + return 0; + + /* get it and store it aside */ + memset(range_state, 0, sizeof(range_state)); + for (i = 0; i < num_var_ranges; i++) { + mtrr_if->get(i, &base, &size, &type); + range_state[i].base_pfn = base; + range_state[i].size_pfn = size; + range_state[i].type = type; + } + + /* check if we need handle it and can handle it */ + if (!mtrr_need_cleanup()) + return 0; + + /* print original var MTRRs at first, for debugging: */ + printk(KERN_DEBUG "original variable MTRRs\n"); + print_out_mtrr_range_state(); memset(range, 0, sizeof(range)); extra_remove_size = 0; @@ -1309,176 +1418,64 @@ static int __init mtrr_cleanup(unsigned address_bits) range_sums >> (20 - PAGE_SHIFT)); if (mtrr_chunk_size && mtrr_gran_size) { - int num_reg; - char gran_factor, chunk_factor, lose_factor; - unsigned long gran_base, chunk_base, lose_base; - - debug_print++; - /* convert ranges to var ranges state */ - num_reg = x86_setup_var_mtrrs(range, nr_range, mtrr_chunk_size, - mtrr_gran_size); + i = 0; + mtrr_calc_range_state(mtrr_chunk_size, mtrr_gran_size, + extra_remove_base, extra_remove_size, i); - /* we got new setting in range_state, check it */ - memset(range_new, 0, sizeof(range_new)); - nr_range_new = x86_get_mtrr_mem_range(range_new, 0, - extra_remove_base, - extra_remove_size); - range_sums_new = sum_ranges(range_new, nr_range_new); + mtrr_print_out_one_result(i); - i = 0; - result[i].chunk_sizek = mtrr_chunk_size >> 10; - result[i].gran_sizek = mtrr_gran_size >> 10; - result[i].num_reg = num_reg; - if (range_sums < range_sums_new) { - result[i].lose_cover_sizek = - (range_sums_new - range_sums) << PSHIFT; - result[i].bad = 1; - } else - result[i].lose_cover_sizek = - (range_sums - range_sums_new) << PSHIFT; - - gran_base = to_size_factor(result[i].gran_sizek, &gran_factor), - chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor), - lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor), - printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t", - result[i].bad?"*BAD*":" ", - gran_base, gran_factor, chunk_base, chunk_factor); - printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n", - result[i].num_reg, result[i].bad?"-":"", - lose_base, lose_factor); if (!result[i].bad) { set_var_mtrr_all(address_bits); return 1; } printk(KERN_INFO "invalid mtrr_gran_size or mtrr_chunk_size, " "will find optimal one\n"); - debug_print--; - memset(result, 0, sizeof(result[0])); } i = 0; memset(min_loss_pfn, 0xff, sizeof(min_loss_pfn)); memset(result, 0, sizeof(result)); for (gran_size = (1ULL<<16); gran_size < (1ULL<<32); gran_size <<= 1) { - char gran_factor; - unsigned long gran_base; - - if (debug_print) - gran_base = to_size_factor(gran_size >> 10, &gran_factor); for (chunk_size = gran_size; chunk_size < (1ULL<<32); chunk_size <<= 1) { - int num_reg; - if (debug_print) { - char chunk_factor; - unsigned long chunk_base; - - chunk_base = to_size_factor(chunk_size>>10, &chunk_factor), - printk(KERN_INFO "\n"); - printk(KERN_INFO "gran_size: %ld%c chunk_size: %ld%c \n", - gran_base, gran_factor, chunk_base, chunk_factor); - } if (i >= NUM_RESULT) continue; - /* convert ranges to var ranges state */ - num_reg = x86_setup_var_mtrrs(range, nr_range, - chunk_size, gran_size); - - /* we got new setting in range_state, check it */ - memset(range_new, 0, sizeof(range_new)); - nr_range_new = x86_get_mtrr_mem_range(range_new, 0, - extra_remove_base, extra_remove_size); - range_sums_new = sum_ranges(range_new, nr_range_new); - - result[i].chunk_sizek = chunk_size >> 10; - result[i].gran_sizek = gran_size >> 10; - result[i].num_reg = num_reg; - if (range_sums < range_sums_new) { - result[i].lose_cover_sizek = - (range_sums_new - range_sums) << PSHIFT; - result[i].bad = 1; - } else - result[i].lose_cover_sizek = - (range_sums - range_sums_new) << PSHIFT; - - /* double check it */ - if (!result[i].bad && !result[i].lose_cover_sizek) { - if (nr_range_new != nr_range || - memcmp(range, range_new, sizeof(range))) - result[i].bad = 1; + mtrr_calc_range_state(chunk_size, gran_size, + extra_remove_base, extra_remove_size, i); + if (debug_print) { + mtrr_print_out_one_result(i); + printk(KERN_INFO "\n"); } - if (!result[i].bad && (range_sums - range_sums_new < - min_loss_pfn[num_reg])) { - min_loss_pfn[num_reg] = - range_sums - range_sums_new; - } i++; } } - /* print out all */ - for (i = 0; i < NUM_RESULT; i++) { - char gran_factor, chunk_factor, lose_factor; - unsigned long gran_base, chunk_base, lose_base; - - gran_base = to_size_factor(result[i].gran_sizek, &gran_factor), - chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor), - lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor), - printk(KERN_INFO "%sgran_size: %ld%c \tchunk_size: %ld%c \t", - result[i].bad?"*BAD*":" ", - gran_base, gran_factor, chunk_base, chunk_factor); - printk(KERN_CONT "num_reg: %d \tlose cover RAM: %s%ld%c\n", - result[i].num_reg, result[i].bad?"-":"", - lose_base, lose_factor); - } - /* try to find the optimal index */ - if (nr_mtrr_spare_reg >= num_var_ranges) - nr_mtrr_spare_reg = num_var_ranges - 1; - num_reg_good = -1; - for (i = num_var_ranges - nr_mtrr_spare_reg; i > 0; i--) { - if (!min_loss_pfn[i]) - num_reg_good = i; - } - - index_good = -1; - if (num_reg_good != -1) { - for (i = 0; i < NUM_RESULT; i++) { - if (!result[i].bad && - result[i].num_reg == num_reg_good && - !result[i].lose_cover_sizek) { - index_good = i; - break; - } - } - } + index_good = mtrr_search_optimal_index(); if (index_good != -1) { - char gran_factor, chunk_factor, lose_factor; - unsigned long gran_base, chunk_base, lose_base; - printk(KERN_INFO "Found optimal setting for mtrr clean up\n"); i = index_good; - gran_base = to_size_factor(result[i].gran_sizek, &gran_factor), - chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor), - lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor), - printk(KERN_INFO "gran_size: %ld%c \tchunk_size: %ld%c \t", - gran_base, gran_factor, chunk_base, chunk_factor); - printk(KERN_CONT "num_reg: %d \tlose RAM: %ld%c\n", - result[i].num_reg, lose_base, lose_factor); + mtrr_print_out_one_result(i); + /* convert ranges to var ranges state */ chunk_size = result[i].chunk_sizek; chunk_size <<= 10; gran_size = result[i].gran_sizek; gran_size <<= 10; - debug_print++; x86_setup_var_mtrrs(range, nr_range, chunk_size, gran_size); - debug_print--; set_var_mtrr_all(address_bits); + printk(KERN_DEBUG "New variable MTRRs\n"); + print_out_mtrr_range_state(); return 1; + } else { + /* print out all */ + for (i = 0; i < NUM_RESULT; i++) + mtrr_print_out_one_result(i); } printk(KERN_INFO "mtrr_cleanup: can not find optimal value\n"); @@ -1562,7 +1559,6 @@ int __init mtrr_trim_uncached_memory(unsigned long end_pfn) { unsigned long i, base, size, highest_pfn = 0, def, dummy; mtrr_type type; - int nr_range; u64 total_trim_size; /* extra one for all 0 */ -- cgit v1.2.3 From 36b75da27bb51dc34e358d0b7487406132806c46 Mon Sep 17 00:00:00 2001 From: Peter Oruba Date: Fri, 17 Oct 2008 15:30:37 +0200 Subject: x86: microcode patch loader author update Removed one author's email address from module init message. Signed-off-by: Peter Oruba Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c index 936d8d55f230..82fb2809ce32 100644 --- a/arch/x86/kernel/microcode_core.c +++ b/arch/x86/kernel/microcode_core.c @@ -480,8 +480,8 @@ static int __init microcode_init(void) printk(KERN_INFO "Microcode Update Driver: v" MICROCODE_VERSION - " " - " \n"); + " ," + " Peter Oruba\n"); return 0; } -- cgit v1.2.3 From 3c52204bb90834bca8e9e78a3628d886ad6d4db5 Mon Sep 17 00:00:00 2001 From: Peter Oruba Date: Fri, 17 Oct 2008 15:30:38 +0200 Subject: x86: AMD microcode patch loader author update Removed author's email address from MODULE_AUTHOR. Signed-off-by: Peter Oruba Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 7a1f8eeac2c7..5f8e5d75a254 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -39,7 +39,7 @@ #include MODULE_DESCRIPTION("AMD Microcode Update Driver"); -MODULE_AUTHOR("Peter Oruba "); +MODULE_AUTHOR("Peter Oruba"); MODULE_LICENSE("GPL v2"); #define UCODE_MAGIC 0x00414d44 -- cgit v1.2.3 From e7706fc691513b0f06adb3de3d6ac04293180146 Mon Sep 17 00:00:00 2001 From: Ken'ichi Ohmichi Date: Mon, 20 Oct 2008 13:51:52 +0900 Subject: x86, kdump: fix invalid access on i386 sparsemem Impact: fix kdump crash on 32-bit sparsemem kernels Since linux-2.6.27, kdump has failed on i386 sparsemem kernel. 1st-kernel gets a panic just before switching to 2nd-kernel. The cause is that a kernel accesses invalid mem_section by page_to_pfn(image->swap_page) at machine_kexec(). image->swap_page is allocated if kexec for hibernation, but it is not allocated if kdump. So if kdump, a kernel should not access the mem_section corresponding to image->swap_page. The attached patch fixes this invalid access. Signed-off-by: Ken'ichi Ohmichi Cc: kexec-ml Cc: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/kernel/machine_kexec_32.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 0732adba05ca..7a385746509a 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -162,7 +162,10 @@ void machine_kexec(struct kimage *image) page_list[VA_PTE_0] = (unsigned long)kexec_pte0; page_list[PA_PTE_1] = __pa(kexec_pte1); page_list[VA_PTE_1] = (unsigned long)kexec_pte1; - page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) << PAGE_SHIFT); + + if (image->type == KEXEC_TYPE_DEFAULT) + page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) + << PAGE_SHIFT); /* The segment registers are funny things, they have both a * visible and an invisible part. Whenever the visible part is -- cgit v1.2.3 From 87c6f40128f92621698f97a62d2ead5184d1dd97 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Tue, 28 Oct 2008 16:13:54 +0100 Subject: x86, gart: fix gart detection for Fam11h CPUs Impact: fix AMD Family 11h boot hangs / USB device problems The AMD Fam11h CPUs have a K8 northbridge. This northbridge is different from other family's because it lacks GART support (as I just learned). But the kernel implicitly expects a GART if it finds an AMD northbridge. Fix this by removing the Fam11h northbridge id from the scan list of K8 northbridges. This patch also changes the message in the GART driver about missing K8 northbridges to tell that the GART is missing which is the correct information in this case. Reported-by: Jouni Malinen Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/k8.c | 1 - arch/x86/kernel/pci-gart_64.c | 2 +- 2 files changed, 1 insertion(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/k8.c b/arch/x86/kernel/k8.c index 304d8bad6559..cbc4332a77b2 100644 --- a/arch/x86/kernel/k8.c +++ b/arch/x86/kernel/k8.c @@ -18,7 +18,6 @@ static u32 *flush_words; struct pci_device_id k8_nb_ids[] = { { PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_K8_NB_MISC) }, { PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_10H_NB_MISC) }, - { PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_11H_NB_MISC) }, {} }; EXPORT_SYMBOL(k8_nb_ids); diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index e3f75bbcedea..a42b02b4df68 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -744,7 +744,7 @@ void __init gart_iommu_init(void) long i; if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) { - printk(KERN_INFO "PCI-GART: No AMD northbridge found.\n"); + printk(KERN_INFO "PCI-GART: No AMD GART found.\n"); return; } -- cgit v1.2.3 From 9352f5698db2c6d7f2789f6cd37e3996d49ac4b5 Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Tue, 28 Oct 2008 23:05:22 -0700 Subject: x86: two trivial sparse annotations Impact: fewer sparse warnings, no functional changes arch/x86/kernel/vsmp_64.c:87:14: warning: incorrect type in argument 1 (different address spaces) arch/x86/kernel/vsmp_64.c:87:14: expected void const volatile [noderef] *addr arch/x86/kernel/vsmp_64.c:87:14: got void *[assigned] address arch/x86/kernel/vsmp_64.c:88:22: warning: incorrect type in argument 1 (different address spaces) arch/x86/kernel/vsmp_64.c:88:22: expected void const volatile [noderef] *addr arch/x86/kernel/vsmp_64.c:88:22: got void * arch/x86/kernel/vsmp_64.c:100:23: warning: incorrect type in argument 2 (different address spaces) arch/x86/kernel/vsmp_64.c:100:23: expected void volatile [noderef] *addr arch/x86/kernel/vsmp_64.c:100:23: got void * arch/x86/kernel/vsmp_64.c:101:23: warning: incorrect type in argument 1 (different address spaces) arch/x86/kernel/vsmp_64.c:101:23: expected void const volatile [noderef] *addr arch/x86/kernel/vsmp_64.c:101:23: got void * arch/x86/mm/gup.c:235:6: warning: incorrect type in argument 1 (different base types) arch/x86/mm/gup.c:235:6: expected void const volatile [noderef] * arch/x86/mm/gup.c:235:6: got unsigned long [unsigned] [assigned] start Signed-off-by: Harvey Harrison Signed-off-by: Ingo Molnar --- arch/x86/kernel/vsmp_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c index 7766d36983fc..a688f3bfaec2 100644 --- a/arch/x86/kernel/vsmp_64.c +++ b/arch/x86/kernel/vsmp_64.c @@ -78,7 +78,7 @@ static unsigned __init_or_module vsmp_patch(u8 type, u16 clobbers, void *ibuf, static void __init set_vsmp_pv_ops(void) { - void *address; + void __iomem *address; unsigned int cap, ctl, cfg; /* set vSMP magic bits to indicate vSMP capable kernel */ -- cgit v1.2.3 From 96bf84b71255b0ee4fcee91e9acd1b5e73030eaf Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 29 Oct 2008 18:44:08 -0700 Subject: x86: signal: cosmetic unification of signr_convert() Impact: cleanup Make signr_convert() same. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 2 ++ arch/x86/kernel/signal_64.c | 6 ++++++ 2 files changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 85a0d37cdae9..abf0df700fd0 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -503,10 +503,12 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, */ static int signr_convert(int sig) { +#ifdef CONFIG_X86_32 struct thread_info *info = current_thread_info(); if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) return info->exec_domain->signal_invmap[sig]; +#endif /* CONFIG_X86_32 */ return sig; } diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 3d0deb336745..a4b46e6392b1 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -295,6 +295,12 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, */ static int signr_convert(int sig) { +#ifdef CONFIG_X86_32 + struct thread_info *info = current_thread_info(); + + if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) + return info->exec_domain->signal_invmap[sig]; +#endif /* CONFIG_X86_32 */ return sig; } -- cgit v1.2.3 From cabf503588961d202a33b3fd872767e9f6abbef7 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 29 Oct 2008 18:46:07 -0700 Subject: x86: signal: cosmetic unification of macros for setup_rt_frame() Impact: cleanup Add #ifdef directive for unification. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 12 ++++++++++++ arch/x86/kernel/signal_64.c | 14 ++++++++++++-- 2 files changed, 24 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index abf0df700fd0..6f3b9a9cc123 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -512,10 +512,22 @@ static int signr_convert(int sig) return sig; } +#ifdef CONFIG_X86_32 + #define is_ia32 1 #define ia32_setup_frame __setup_frame #define ia32_setup_rt_frame __setup_rt_frame +#else /* !CONFIG_X86_32 */ + +#ifdef CONFIG_IA32_EMULATION +#define is_ia32 test_thread_flag(TIF_IA32) +#else /* !CONFIG_IA32_EMULATION */ +#define is_ia32 0 +#endif /* CONFIG_IA32_EMULATION */ + +#endif /* CONFIG_X86_32 */ + static int setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, sigset_t *set, struct pt_regs *regs) diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index a4b46e6392b1..49df79e05111 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -304,11 +304,21 @@ static int signr_convert(int sig) return sig; } +#ifdef CONFIG_X86_32 + +#define is_ia32 1 +#define ia32_setup_frame __setup_frame +#define ia32_setup_rt_frame __setup_rt_frame + +#else /* !CONFIG_X86_32 */ + #ifdef CONFIG_IA32_EMULATION #define is_ia32 test_thread_flag(TIF_IA32) -#else +#else /* !CONFIG_IA32_EMULATION */ #define is_ia32 0 -#endif +#endif /* CONFIG_IA32_EMULATION */ + +#endif /* CONFIG_X86_32 */ static int setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, -- cgit v1.2.3 From 57917752f51bcead3bb6c83d74137fbe342504ec Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 29 Oct 2008 18:46:40 -0700 Subject: x86: signal: cosmetic unification of NR_restart_syscall Impact: cleanup Add #ifdef directive to unify NR_restart_syscall. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 6 ++++++ arch/x86/kernel/signal_64.c | 5 +++++ 2 files changed, 11 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 6f3b9a9cc123..a0efc1b3c4c9 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -628,7 +628,13 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka, return 0; } +#ifdef CONFIG_X86_32 #define NR_restart_syscall __NR_restart_syscall +#else /* !CONFIG_X86_32 */ +#define NR_restart_syscall \ + test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall +#endif /* CONFIG_X86_32 */ + /* * Note that 'init' is a special process: it doesn't get signals it doesn't * want to handle. Thus you cannot kill init even with a SIGKILL even by diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 49df79e05111..83990db82f74 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -420,8 +420,13 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka, return 0; } +#ifdef CONFIG_X86_32 +#define NR_restart_syscall __NR_restart_syscall +#else /* !CONFIG_X86_32 */ #define NR_restart_syscall \ test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall +#endif /* CONFIG_X86_32 */ + /* * Note that 'init' is a special process: it doesn't get signals it doesn't * want to handle. Thus you cannot kill init even with a SIGKILL even by -- cgit v1.2.3 From ae9b9403644f3ecc76867af042e7e1cfd5c099d0 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 30 Oct 2008 17:43:57 +0100 Subject: AMD IOMMU: fix detection of NP capable IOMMUs This patch changes the code to use IOMMU_CAP_NPCACHE as a shift and not as a mask. Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 3b346c6f5514..38e88d40ab10 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -50,7 +50,7 @@ static int dma_ops_unity_map(struct dma_ops_domain *dma_dom, /* returns !0 if the IOMMU is caching non-present entries in its TLB */ static int iommu_has_npcache(struct amd_iommu *iommu) { - return iommu->cap & IOMMU_CAP_NPCACHE; + return iommu->cap & (1UL << IOMMU_CAP_NPCACHE); } /**************************************************************************** -- cgit v1.2.3 From b062f841b569791d3054e975cd85f48562161565 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 30 Oct 2008 19:16:46 +0300 Subject: x86: nmi - add sensible names to nmi_watchdog boot param Impact: introduce nmi_watchdog=lapic and nmi_watchdog=ioapic aliases Add sensible names as "lapic" and "ioapic" to nmi_watchdog boot parameter. Sometimes it is not that easy to recall what exactly nmi_watchdog=1 does mean so we allow the using of symbolic names here. Old numeric values remain valid. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/nmi.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index 2c97f07f1c2c..c4869e4532a3 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -199,12 +199,17 @@ static int __init setup_nmi_watchdog(char *str) ++str; } - get_option(&str, &nmi); - - if (nmi >= NMI_INVALID) - return 0; + if (!strncmp(str, "lapic", 5)) + nmi_watchdog = NMI_LOCAL_APIC; + else if (!strncmp(str, "ioapic", 6)) + nmi_watchdog = NMI_IO_APIC; + else { + get_option(&str, &nmi); + if (nmi >= NMI_INVALID) + return 0; + nmi_watchdog = nmi; + } - nmi_watchdog = nmi; return 1; } __setup("nmi_watchdog=", setup_nmi_watchdog); -- cgit v1.2.3 From 1cbd8b3fdcf56a3c39a7596512095c9e33221fa1 Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Thu, 30 Oct 2008 10:45:36 +0000 Subject: x86: add two missing unwind annotations Impact: improve debuginfo Signed-off-by: Jan Beulich Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index b86f332c96a6..ddeeb1052583 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -255,6 +255,7 @@ ENTRY(ret_from_fork) call schedule_tail GET_THREAD_INFO(%rcx) testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),TI_flags(%rcx) + CFI_REMEMBER_STATE jnz rff_trace rff_action: RESTORE_REST @@ -264,6 +265,7 @@ rff_action: jnz int_ret_from_sys_call RESTORE_TOP_OF_STACK %rdi,ARGOFFSET jmp ret_from_sys_call + CFI_RESTORE_STATE rff_trace: movq %rsp,%rdi call syscall_trace_leave -- cgit v1.2.3 From 17666f02b118099028522dfc3df00a235700e216 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 30 Oct 2008 16:08:32 -0400 Subject: ftrace: nmi safe code modification Impact: fix crashes that can occur in NMI handlers, if their code is modified Modifying code is something that needs special care. On SMP boxes, if code that is being modified is also being executed on another CPU, that CPU will have undefined results. The dynamic ftrace uses kstop_machine to make the system act like a uniprocessor system. But this does not address NMIs, that can still run on other CPUs. One approach to handle this is to make all code that are used by NMIs not be traced. But NMIs can call notifiers that spread throughout the kernel and this will be very hard to maintain, and the chance of missing a function is very high. The approach that this patch takes is to have the NMIs modify the code if the modification is taking place. The way this works is that just writing to code executing on another CPU is not harmful if what is written is the same as what exists. Two buffers are used: an IP buffer and a "code" buffer. The steps that the patcher takes are: 1) Put in the instruction pointer into the IP buffer and the new code into the "code" buffer. 2) Set a flag that says we are modifying code 3) Wait for any running NMIs to finish. 4) Write the code 5) clear the flag. 6) Wait for any running NMIs to finish. If an NMI is executed, it will also write the pending code. Multiple writes are OK, because what is being written is the same. Then the patcher must wait for all running NMIs to finish before going to the next line that must be patched. This is basically the RCU approach to code modification. Thanks to Ingo Molnar for suggesting the idea, and to Arjan van de Ven for his guidence on what is safe and what is not. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 107 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 106 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 50ea0ac8c9bf..fe5f859130b5 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -56,6 +56,111 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) return calc.code; } +/* + * Modifying code must take extra care. On an SMP machine, if + * the code being modified is also being executed on another CPU + * that CPU will have undefined results and possibly take a GPF. + * We use kstop_machine to stop other CPUS from exectuing code. + * But this does not stop NMIs from happening. We still need + * to protect against that. We separate out the modification of + * the code to take care of this. + * + * Two buffers are added: An IP buffer and a "code" buffer. + * + * 1) Put in the instruction pointer into the IP buffer + * and the new code into the "code" buffer. + * 2) Set a flag that says we are modifying code + * 3) Wait for any running NMIs to finish. + * 4) Write the code + * 5) clear the flag. + * 6) Wait for any running NMIs to finish. + * + * If an NMI is executed, the first thing it does is to call + * "ftrace_nmi_enter". This will check if the flag is set to write + * and if it is, it will write what is in the IP and "code" buffers. + * + * The trick is, it does not matter if everyone is writing the same + * content to the code location. Also, if a CPU is executing code + * it is OK to write to that code location if the contents being written + * are the same as what exists. + */ + +static atomic_t in_nmi; +static int mod_code_status; +static int mod_code_write; +static void *mod_code_ip; +static void *mod_code_newcode; + +static void ftrace_mod_code(void) +{ + /* + * Yes, more than one CPU process can be writing to mod_code_status. + * (and the code itself) + * But if one were to fail, then they all should, and if one were + * to succeed, then they all should. + */ + mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode, + MCOUNT_INSN_SIZE); + +} + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); + /* Must have in_nmi seen before reading write flag */ + smp_mb(); + if (mod_code_write) + ftrace_mod_code(); +} + +void ftrace_nmi_exit(void) +{ + /* Finish all executions before clearing in_nmi */ + smp_wmb(); + atomic_dec(&in_nmi); +} + +static void wait_for_nmi(void) +{ + while (atomic_read(&in_nmi)) + cpu_relax(); +} + +static int +do_ftrace_mod_code(unsigned long ip, void *new_code) +{ + mod_code_ip = (void *)ip; + mod_code_newcode = new_code; + + /* The buffers need to be visible before we let NMIs write them */ + smp_wmb(); + + mod_code_write = 1; + + /* Make sure write bit is visible before we wait on NMIs */ + smp_mb(); + + wait_for_nmi(); + + /* Make sure all running NMIs have finished before we write the code */ + smp_mb(); + + ftrace_mod_code(); + + /* Make sure the write happens before clearing the bit */ + smp_wmb(); + + mod_code_write = 0; + + /* make sure NMIs see the cleared bit */ + smp_mb(); + + wait_for_nmi(); + + return mod_code_status; +} + + int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) @@ -81,7 +186,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return -EINVAL; /* replace the text with the new text */ - if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE)) + if (do_ftrace_mod_code(ip, new_code)) return -EPERM; sync_core(); -- cgit v1.2.3 From b807c3d0f8e39ed7cbbbe6da162650e305e8de15 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 30 Oct 2008 16:08:33 -0400 Subject: ftrace: nmi update statistics Impact: add more debug info to /debugfs/tracing/dyn_ftrace_total_info This patch adds dynamic ftrace NMI update statistics to the /debugfs/tracing/dyn_ftrace_total_info stat file. Signed-off-by: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 26 ++++++++++++++++++++++++-- 1 file changed, 24 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index fe5f859130b5..6685b0fc1b44 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -91,6 +91,19 @@ static int mod_code_write; static void *mod_code_ip; static void *mod_code_newcode; +static int nmi_wait_count; +static atomic_t nmi_update_count; + +int ftrace_arch_read_dyn_info(char *buf, int size) +{ + int r; + + r = snprintf(buf, size, "%u %u", + nmi_wait_count, + atomic_read(&nmi_update_count)); + return r; +} + static void ftrace_mod_code(void) { /* @@ -109,8 +122,10 @@ void ftrace_nmi_enter(void) atomic_inc(&in_nmi); /* Must have in_nmi seen before reading write flag */ smp_mb(); - if (mod_code_write) + if (mod_code_write) { ftrace_mod_code(); + atomic_inc(&nmi_update_count); + } } void ftrace_nmi_exit(void) @@ -122,8 +137,15 @@ void ftrace_nmi_exit(void) static void wait_for_nmi(void) { - while (atomic_read(&in_nmi)) + int waited = 0; + + while (atomic_read(&in_nmi)) { + waited = 1; cpu_relax(); + } + + if (waited) + nmi_wait_count++; } static int -- cgit v1.2.3 From 017d9d20d88cacb0a6a29f343b23c95e203f6645 Mon Sep 17 00:00:00 2001 From: James Bottomley Date: Thu, 30 Oct 2008 16:05:39 -0500 Subject: x86: use CONFIG_X86_SMP instead of CONFIG_SMP Impact: fix x86/Voyager boot CONFIG_SMP is used for features which work on *all* x86 boxes. CONFIG_X86_SMP is used for standard PC like x86 boxes (for things like multi core and apics) Signed-off-by: James Bottomley Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/addon_cpuid_features.c | 2 +- arch/x86/kernel/tsc.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index 0d9c993aa93e..ef8f831af823 100644 --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -69,7 +69,7 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c) */ void __cpuinit detect_extended_topology(struct cpuinfo_x86 *c) { -#ifdef CONFIG_SMP +#ifdef CONFIG_X86_SMP unsigned int eax, ebx, ecx, edx, sub_index; unsigned int ht_mask_width, core_plus_mask_width; unsigned int core_select_mask, core_level_siblings; diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 161bb850fc47..62348e4fd8d1 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -759,7 +759,7 @@ __cpuinit int unsynchronized_tsc(void) if (!cpu_has_tsc || tsc_unstable) return 1; -#ifdef CONFIG_SMP +#ifdef CONFIG_X86_SMP if (apic_is_clustered_box()) return 1; #endif -- cgit v1.2.3 From b3572e361b6b2ac5e724bc4bb932b7774b720b95 Mon Sep 17 00:00:00 2001 From: James Bottomley Date: Thu, 30 Oct 2008 16:00:59 -0500 Subject: x86/voyager: fix compile breakage caused by dc1e35c6e95e8923cf1d3510438b63c600fee1e2 Impact: build fix on x86/Voyager Given commits like this: | Author: Suresh Siddha | Date: Tue Jul 29 10:29:19 2008 -0700 | | x86, xsave: enable xsave/xrstor on cpus with xsave support Which deliberately expose boot cpu dependence to pieces of the system, I think it's time to explicitly have a variable for it to prevent this continual misassumption that the boot CPU is zero. Signed-off-by: James Bottomley Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 25581dcb280e..93e9393ea64a 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1134,7 +1134,7 @@ void __cpuinit cpu_init(void) /* * Boot processor to setup the FP and extended state context info. */ - if (!smp_processor_id()) + if (smp_processor_id() == boot_cpu_id) init_thread_xstate(); xsave_init(); -- cgit v1.2.3 From bfcb4c1becf93b1592f4a03a4d6e00a3ab89d5ec Mon Sep 17 00:00:00 2001 From: James Bottomley Date: Thu, 30 Oct 2008 16:13:37 -0500 Subject: x86/voyager: fix missing cpu_index initialisation Impact: fix /proc/cpuinfo output on x86/Voyager Ever since | commit 92cb7612aee39642d109b8d935ad265e602c0563 | Author: Mike Travis | Date: Fri Oct 19 20:35:04 2007 +0200 | | x86: convert cpuinfo_x86 array to a per_cpu array We've had an extra field in cpuinfo_x86 which is cpu_index. Unfortunately, voyager has never initialised this, although the only noticeable impact seems to be that /proc/cpuinfo shows all zeros for the processor ids. Anyway, fix this by initialising the boot CPU properly and setting the index when the secondaries update. Signed-off-by: James Bottomley Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 93e9393ea64a..da8f15ac7a6d 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -549,6 +549,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c) this_cpu->c_early_init(c); validate_pat_support(c); + + c->cpu_index = boot_cpu_id; } void __init early_cpu_init(void) -- cgit v1.2.3 From 1c4acdb467f8a6704855a5670ff3d82e3c18eb0b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 31 Oct 2008 00:43:03 +0100 Subject: x86: cpu_index build fix fix: arch/x86/kernel/cpu/common.c: In function 'early_identify_cpu': arch/x86/kernel/cpu/common.c:553: error: 'struct cpuinfo_x86' has no member named 'cpu_index' as cpu_index is only available on SMP. Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index da8f15ac7a6d..003a65395bd5 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -550,7 +550,9 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c) validate_pat_support(c); +#ifdef CONFIG_SMP c->cpu_index = boot_cpu_id; +#endif } void __init early_cpu_init(void) -- cgit v1.2.3 From b342797c1e5116a130841527b47dfaa462ed0968 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 31 Oct 2008 09:31:38 +0100 Subject: x86: build fix Impact: build fix on certain UP configs fix: arch/x86/kernel/cpu/common.c: In function 'cpu_init': arch/x86/kernel/cpu/common.c:1141: error: 'boot_cpu_id' undeclared (first use in this function) arch/x86/kernel/cpu/common.c:1141: error: (Each undeclared identifier is reported only once arch/x86/kernel/cpu/common.c:1141: error: for each function it appears in.) Pull in asm/smp.h on UP, so that we get the definition of boot_cpu_id. Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 003a65395bd5..b9c9ea0217a9 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -20,6 +20,7 @@ #include #include #include +#include #ifdef CONFIG_X86_LOCAL_APIC #include #include -- cgit v1.2.3 From 31498a01496ffca3b542bae72b8ec499cd9302db Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 31 Oct 2008 09:48:02 +0800 Subject: kexec/i386: remove PAGE_SIZE alignment from relocate_kernel Impact: save kernel .text by loosening kexec page alignment This patch removes PAGE_SIZE alignment from relocate_kernel(). Before kexec jump patches are merged, control page is mapped to relocate_kernel in kexec page tables, so relocate_kernel must be PAGE_SIZE aligned. Now, control page is mapped to identity mapped address, so relocate_kernel need not to be PAGE_SIZE aligned any more. This can reduce a few KB from kernel text segement. Signed-off-by: Huang Ying Signed-off-by: Ingo Molnar --- arch/x86/kernel/relocate_kernel_32.S | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S index 6f50664b2ba5..377da3f78e8c 100644 --- a/arch/x86/kernel/relocate_kernel_32.S +++ b/arch/x86/kernel/relocate_kernel_32.S @@ -39,7 +39,6 @@ #define CP_PA_BACKUP_PAGES_MAP DATA(0x1c) .text - .align PAGE_SIZE .globl relocate_kernel relocate_kernel: /* Save the CPU context, used for jumping back */ -- cgit v1.2.3 From 92be3d6bdf2cb34972ab50e12ad4da1076e690da Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 31 Oct 2008 09:48:08 +0800 Subject: kexec/i386: allocate page table pages dynamically Impact: save .text size when kexec is built in but not loaded This patch adds an architecture specific struct kimage_arch into struct kimage. The pointers to page table pages used by kexec are added to struct kimage_arch. The page tables pages are dynamically allocated in machine_kexec_prepare instead of statically from BSS segment. This will save up to 20k memory when kexec image is not loaded. Signed-off-by: Huang Ying Signed-off-by: Ingo Molnar --- arch/x86/kernel/machine_kexec_32.c | 67 ++++++++++++++++++++++++++------------ 1 file changed, 46 insertions(+), 21 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 7a385746509a..1100312847a5 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include @@ -25,15 +26,6 @@ #include #include -#define PAGE_ALIGNED __attribute__ ((__aligned__(PAGE_SIZE))) -static u32 kexec_pgd[1024] PAGE_ALIGNED; -#ifdef CONFIG_X86_PAE -static u32 kexec_pmd0[1024] PAGE_ALIGNED; -static u32 kexec_pmd1[1024] PAGE_ALIGNED; -#endif -static u32 kexec_pte0[1024] PAGE_ALIGNED; -static u32 kexec_pte1[1024] PAGE_ALIGNED; - static void set_idt(void *newidt, __u16 limit) { struct desc_ptr curidt; @@ -76,6 +68,37 @@ static void load_segments(void) #undef __STR } +static void machine_kexec_free_page_tables(struct kimage *image) +{ + free_page((unsigned long)image->arch.pgd); +#ifdef CONFIG_X86_PAE + free_page((unsigned long)image->arch.pmd0); + free_page((unsigned long)image->arch.pmd1); +#endif + free_page((unsigned long)image->arch.pte0); + free_page((unsigned long)image->arch.pte1); +} + +static int machine_kexec_alloc_page_tables(struct kimage *image) +{ + image->arch.pgd = (pgd_t *)get_zeroed_page(GFP_KERNEL); +#ifdef CONFIG_X86_PAE + image->arch.pmd0 = (pmd_t *)get_zeroed_page(GFP_KERNEL); + image->arch.pmd1 = (pmd_t *)get_zeroed_page(GFP_KERNEL); +#endif + image->arch.pte0 = (pte_t *)get_zeroed_page(GFP_KERNEL); + image->arch.pte1 = (pte_t *)get_zeroed_page(GFP_KERNEL); + if (!image->arch.pgd || +#ifdef CONFIG_X86_PAE + !image->arch.pmd0 || !image->arch.pmd1 || +#endif + !image->arch.pte0 || !image->arch.pte1) { + machine_kexec_free_page_tables(image); + return -ENOMEM; + } + return 0; +} + /* * A architecture hook called to validate the * proposed image and prepare the control pages @@ -87,13 +110,14 @@ static void load_segments(void) * reboot code buffer to allow us to avoid allocations * later. * - * Make control page executable. + * - Make control page executable. + * - Allocate page tables */ int machine_kexec_prepare(struct kimage *image) { if (nx_enabled) set_pages_x(image->control_code_page, 1); - return 0; + return machine_kexec_alloc_page_tables(image); } /* @@ -104,6 +128,7 @@ void machine_kexec_cleanup(struct kimage *image) { if (nx_enabled) set_pages_nx(image->control_code_page, 1); + machine_kexec_free_page_tables(image); } /* @@ -150,18 +175,18 @@ void machine_kexec(struct kimage *image) relocate_kernel_ptr = control_page; page_list[PA_CONTROL_PAGE] = __pa(control_page); page_list[VA_CONTROL_PAGE] = (unsigned long)control_page; - page_list[PA_PGD] = __pa(kexec_pgd); - page_list[VA_PGD] = (unsigned long)kexec_pgd; + page_list[PA_PGD] = __pa(image->arch.pgd); + page_list[VA_PGD] = (unsigned long)image->arch.pgd; #ifdef CONFIG_X86_PAE - page_list[PA_PMD_0] = __pa(kexec_pmd0); - page_list[VA_PMD_0] = (unsigned long)kexec_pmd0; - page_list[PA_PMD_1] = __pa(kexec_pmd1); - page_list[VA_PMD_1] = (unsigned long)kexec_pmd1; + page_list[PA_PMD_0] = __pa(image->arch.pmd0); + page_list[VA_PMD_0] = (unsigned long)image->arch.pmd0; + page_list[PA_PMD_1] = __pa(image->arch.pmd1); + page_list[VA_PMD_1] = (unsigned long)image->arch.pmd1; #endif - page_list[PA_PTE_0] = __pa(kexec_pte0); - page_list[VA_PTE_0] = (unsigned long)kexec_pte0; - page_list[PA_PTE_1] = __pa(kexec_pte1); - page_list[VA_PTE_1] = (unsigned long)kexec_pte1; + page_list[PA_PTE_0] = __pa(image->arch.pte0); + page_list[VA_PTE_0] = (unsigned long)image->arch.pte0; + page_list[PA_PTE_1] = __pa(image->arch.pte1); + page_list[VA_PTE_1] = (unsigned long)image->arch.pte1; if (image->type == KEXEC_TYPE_DEFAULT) page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) -- cgit v1.2.3 From 9868ee63b896ee4d2ceb8c292e88d7f4e66caaf9 Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 31 Oct 2008 09:48:15 +0800 Subject: kexec/i386: setup kexec page table in C Impact: change the kexec bootstrap code implementation from assembly to C This patch transforms the kexec page tables setup code from assembler code to C code in machine_kexec_prepare. This improves readability and reduces code line number. Signed-off-by: Huang Ying Signed-off-by: Ingo Molnar --- arch/x86/kernel/machine_kexec_32.c | 59 ++++++++++++++---- arch/x86/kernel/relocate_kernel_32.S | 114 ----------------------------------- 2 files changed, 47 insertions(+), 126 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 1100312847a5..37f420018a41 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -99,6 +99,45 @@ static int machine_kexec_alloc_page_tables(struct kimage *image) return 0; } +static void machine_kexec_page_table_set_one( + pgd_t *pgd, pmd_t *pmd, pte_t *pte, + unsigned long vaddr, unsigned long paddr) +{ + pud_t *pud; + + pgd += pgd_index(vaddr); +#ifdef CONFIG_X86_PAE + if (!(pgd_val(*pgd) & _PAGE_PRESENT)) + set_pgd(pgd, __pgd(__pa(pmd) | _PAGE_PRESENT)); +#endif + pud = pud_offset(pgd, vaddr); + pmd = pmd_offset(pud, vaddr); + if (!(pmd_val(*pmd) & _PAGE_PRESENT)) + set_pmd(pmd, __pmd(__pa(pte) | _PAGE_TABLE)); + pte = pte_offset_kernel(pmd, vaddr); + set_pte(pte, pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL_EXEC)); +} + +static void machine_kexec_prepare_page_tables(struct kimage *image) +{ + void *control_page; + pmd_t *pmd = 0; + + control_page = page_address(image->control_code_page); +#ifdef CONFIG_X86_PAE + pmd = image->arch.pmd0; +#endif + machine_kexec_page_table_set_one( + image->arch.pgd, pmd, image->arch.pte0, + (unsigned long)control_page, __pa(control_page)); +#ifdef CONFIG_X86_PAE + pmd = image->arch.pmd1; +#endif + machine_kexec_page_table_set_one( + image->arch.pgd, pmd, image->arch.pte1, + __pa(control_page), __pa(control_page)); +} + /* * A architecture hook called to validate the * proposed image and prepare the control pages @@ -112,12 +151,19 @@ static int machine_kexec_alloc_page_tables(struct kimage *image) * * - Make control page executable. * - Allocate page tables + * - Setup page tables */ int machine_kexec_prepare(struct kimage *image) { + int error; + if (nx_enabled) set_pages_x(image->control_code_page, 1); - return machine_kexec_alloc_page_tables(image); + error = machine_kexec_alloc_page_tables(image); + if (error) + return error; + machine_kexec_prepare_page_tables(image); + return 0; } /* @@ -176,17 +222,6 @@ void machine_kexec(struct kimage *image) page_list[PA_CONTROL_PAGE] = __pa(control_page); page_list[VA_CONTROL_PAGE] = (unsigned long)control_page; page_list[PA_PGD] = __pa(image->arch.pgd); - page_list[VA_PGD] = (unsigned long)image->arch.pgd; -#ifdef CONFIG_X86_PAE - page_list[PA_PMD_0] = __pa(image->arch.pmd0); - page_list[VA_PMD_0] = (unsigned long)image->arch.pmd0; - page_list[PA_PMD_1] = __pa(image->arch.pmd1); - page_list[VA_PMD_1] = (unsigned long)image->arch.pmd1; -#endif - page_list[PA_PTE_0] = __pa(image->arch.pte0); - page_list[VA_PTE_0] = (unsigned long)image->arch.pte0; - page_list[PA_PTE_1] = __pa(image->arch.pte1); - page_list[VA_PTE_1] = (unsigned long)image->arch.pte1; if (image->type == KEXEC_TYPE_DEFAULT) page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S index 377da3f78e8c..a160f3119725 100644 --- a/arch/x86/kernel/relocate_kernel_32.S +++ b/arch/x86/kernel/relocate_kernel_32.S @@ -10,15 +10,12 @@ #include #include #include -#include /* * Must be relocatable PIC code callable as a C function */ #define PTR(x) (x << 2) -#define PAGE_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY) -#define PAE_PGD_ATTR (_PAGE_PRESENT) /* control_page + KEXEC_CONTROL_CODE_MAX_SIZE * ~ control_page + PAGE_SIZE are used as data storage and stack for @@ -59,117 +56,6 @@ relocate_kernel: movl %cr4, %eax movl %eax, CR4(%edi) -#ifdef CONFIG_X86_PAE - /* map the control page at its virtual address */ - - movl PTR(VA_PGD)(%ebp), %edi - movl PTR(VA_CONTROL_PAGE)(%ebp), %eax - andl $0xc0000000, %eax - shrl $27, %eax - addl %edi, %eax - - movl PTR(PA_PMD_0)(%ebp), %edx - orl $PAE_PGD_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PMD_0)(%ebp), %edi - movl PTR(VA_CONTROL_PAGE)(%ebp), %eax - andl $0x3fe00000, %eax - shrl $18, %eax - addl %edi, %eax - - movl PTR(PA_PTE_0)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PTE_0)(%ebp), %edi - movl PTR(VA_CONTROL_PAGE)(%ebp), %eax - andl $0x001ff000, %eax - shrl $9, %eax - addl %edi, %eax - - movl PTR(PA_CONTROL_PAGE)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - /* identity map the control page at its physical address */ - - movl PTR(VA_PGD)(%ebp), %edi - movl PTR(PA_CONTROL_PAGE)(%ebp), %eax - andl $0xc0000000, %eax - shrl $27, %eax - addl %edi, %eax - - movl PTR(PA_PMD_1)(%ebp), %edx - orl $PAE_PGD_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PMD_1)(%ebp), %edi - movl PTR(PA_CONTROL_PAGE)(%ebp), %eax - andl $0x3fe00000, %eax - shrl $18, %eax - addl %edi, %eax - - movl PTR(PA_PTE_1)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PTE_1)(%ebp), %edi - movl PTR(PA_CONTROL_PAGE)(%ebp), %eax - andl $0x001ff000, %eax - shrl $9, %eax - addl %edi, %eax - - movl PTR(PA_CONTROL_PAGE)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) -#else - /* map the control page at its virtual address */ - - movl PTR(VA_PGD)(%ebp), %edi - movl PTR(VA_CONTROL_PAGE)(%ebp), %eax - andl $0xffc00000, %eax - shrl $20, %eax - addl %edi, %eax - - movl PTR(PA_PTE_0)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PTE_0)(%ebp), %edi - movl PTR(VA_CONTROL_PAGE)(%ebp), %eax - andl $0x003ff000, %eax - shrl $10, %eax - addl %edi, %eax - - movl PTR(PA_CONTROL_PAGE)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - /* identity map the control page at its physical address */ - - movl PTR(VA_PGD)(%ebp), %edi - movl PTR(PA_CONTROL_PAGE)(%ebp), %eax - andl $0xffc00000, %eax - shrl $20, %eax - addl %edi, %eax - - movl PTR(PA_PTE_1)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) - - movl PTR(VA_PTE_1)(%ebp), %edi - movl PTR(PA_CONTROL_PAGE)(%ebp), %eax - andl $0x003ff000, %eax - shrl $10, %eax - addl %edi, %eax - - movl PTR(PA_CONTROL_PAGE)(%ebp), %edx - orl $PAGE_ATTR, %edx - movl %edx, (%eax) -#endif - -relocate_new_kernel: /* read the arguments and say goodbye to the stack */ movl 20+4(%esp), %ebx /* page_list */ movl 20+8(%esp), %ebp /* list of pages */ -- cgit v1.2.3 From a26a2a27396c0a0877aa701f8f92d08ba550a6c9 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 31 Oct 2008 00:03:22 -0400 Subject: ftrace: nmi safe code clean ups Impact: cleanup This patch cleans up the NMI safe code for dynamic ftrace as suggested by Andrew Morton. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 6685b0fc1b44..69149337f2fe 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -67,7 +67,7 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) * * Two buffers are added: An IP buffer and a "code" buffer. * - * 1) Put in the instruction pointer into the IP buffer + * 1) Put the instruction pointer into the IP buffer * and the new code into the "code" buffer. * 2) Set a flag that says we are modifying code * 3) Wait for any running NMIs to finish. @@ -85,14 +85,14 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) * are the same as what exists. */ -static atomic_t in_nmi; -static int mod_code_status; -static int mod_code_write; -static void *mod_code_ip; -static void *mod_code_newcode; +static atomic_t in_nmi = ATOMIC_INIT(0); +static int mod_code_status; /* holds return value of text write */ +static int mod_code_write; /* set when NMI should do the write */ +static void *mod_code_ip; /* holds the IP to write to */ +static void *mod_code_newcode; /* holds the text to write to the IP */ -static int nmi_wait_count; -static atomic_t nmi_update_count; +static unsigned nmi_wait_count; +static atomic_t nmi_update_count = ATOMIC_INIT(0); int ftrace_arch_read_dyn_info(char *buf, int size) { -- cgit v1.2.3 From 1f98757776eafe31065be9118db6051afcf8643c Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sat, 1 Nov 2008 10:17:22 -0700 Subject: x86: Clean up late e820 resource allocation This makes the late e820 resources use 'insert_resource_expand_to_fit()' instead of doing a 'reserve_region_with_split()', and also avoids marking them as IORESOURCE_BUSY. This results in us being perfectly happy to use pre-existing PCI resources even if they were marked as being in a reserved region, while still avoiding any _new_ allocations in the reserved regions. It also makes for a simpler and more accurate resource tree. Example resource allocation from Jonathan Corbet, who has firmware that has an e820 reserved entry that covered a big range (e0000000-fed003ff), and that had various PCI resources in it set up by firmware. With old kernels, the reserved range would force us to re-allocate all pre-existing PCI resources, and his reserved range would end up looking like this: e0000000-fed003ff : reserved fec00000-fec00fff : IOAPIC 0 fed00000-fed003ff : HPET 0 where only the pre-allocated special regions (IOAPIC and HPET) were kept around. With 2.6.28-rc2, which uses 'reserve_region_with_split()', Jonathan's resource tree looked like this: e0000000-fe7fffff : reserved fe800000-fe8fffff : PCI Bus 0000:01 fe800000-fe8fffff : reserved fe900000-fe9d9aff : reserved fe9d9b00-fe9d9bff : 0000:00:1f.3 fe9d9b00-fe9d9bff : reserved fe9d9c00-fe9d9fff : 0000:00:1a.7 fe9d9c00-fe9d9fff : reserved fe9da000-fe9dafff : 0000:00:03.3 fe9da000-fe9dafff : reserved fe9db000-fe9dbfff : 0000:00:19.0 fe9db000-fe9dbfff : reserved fe9dc000-fe9dffff : 0000:00:1b.0 fe9dc000-fe9dffff : reserved fe9e0000-fe9fffff : 0000:00:19.0 fe9e0000-fe9fffff : reserved fea00000-fea7ffff : 0000:00:02.0 fea00000-fea7ffff : reserved fea80000-feafffff : 0000:00:02.1 fea80000-feafffff : reserved feb00000-febfffff : 0000:00:02.0 feb00000-febfffff : reserved fec00000-fed003ff : reserved fec00000-fec00fff : IOAPIC 0 fed00000-fed003ff : HPET 0 and because the reserved entry had been split and moved into the individual resources, and because it used the IORESOURCE_BUSY flag, the drivers that actually wanted to _use_ those resources couldn't actually attach to them: e1000e 0000:00:19.0: BAR 0: can't reserve mem region [0xfe9e0000-0xfe9fffff] HDA Intel 0000:00:1b.0: BAR 0: can't reserve mem region [0xfe9dc000-0xfe9dffff] with this patch, the resource tree instead becomes e0000000-fed003ff : reserved fe800000-fe8fffff : PCI Bus 0000:01 fe9d9b00-fe9d9bff : 0000:00:1f.3 fe9d9c00-fe9d9fff : 0000:00:1a.7 fe9d9c00-fe9d9fff : ehci_hcd fe9da000-fe9dafff : 0000:00:03.3 fe9db000-fe9dbfff : 0000:00:19.0 fe9db000-fe9dbfff : e1000e fe9dc000-fe9dffff : 0000:00:1b.0 fe9dc000-fe9dffff : ICH HD audio fe9e0000-fe9fffff : 0000:00:19.0 fe9e0000-fe9fffff : e1000e fea00000-fea7ffff : 0000:00:02.0 fea80000-feafffff : 0000:00:02.1 feb00000-febfffff : 0000:00:02.0 fec00000-fec00fff : IOAPIC 0 fed00000-fed003ff : HPET 0 ie the one reserved region now ends up surrounding all the PCI resources that were allocated inside of it by firmware, and because it is not marked BUSY, drivers have no problem attaching to the pre-allocated resources. Reported-and-tested-by: Jonathan Corbet Cc: Yinghai Lu Cc: Ingo Molnar Cc: Robert Hancock Signed-off-by: Linus Torvalds --- arch/x86/kernel/e820.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index ce97bf3bed12..7aafeb5263ef 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -1290,15 +1290,17 @@ void __init e820_reserve_resources(void) res->start = e820.map[i].addr; res->end = end; - res->flags = IORESOURCE_MEM | IORESOURCE_BUSY; + res->flags = IORESOURCE_MEM; /* * don't register the region that could be conflicted with * pci device BAR resource and insert them later in * pcibios_resource_survey() */ - if (e820.map[i].type != E820_RESERVED || res->start < (1ULL<<20)) + if (e820.map[i].type != E820_RESERVED || res->start < (1ULL<<20)) { + res->flags |= IORESOURCE_BUSY; insert_resource(&iomem_resource, res); + } res++; } @@ -1318,7 +1320,7 @@ void __init e820_reserve_resources_late(void) res = e820_res; for (i = 0; i < e820.nr_map; i++) { if (!res->parent && res->end) - reserve_region_with_split(&iomem_resource, res->start, res->end, res->name); + insert_resource_expand_to_fit(&iomem_resource, res); res++; } } -- cgit v1.2.3 From 88b094fb8d4fe43b7025ea8d487059e8813e02cd Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Mon, 27 Oct 2008 10:41:46 -0700 Subject: x86: Hypervisor detection and get tsc_freq from hypervisor Impact: Changes timebase calibration on Vmware. v3->v2 : Abstract the hypervisor detection and feature (tsc_freq) request behind a hypervisor.c file v2->v1 : Add a x86_hyper_vendor field to the cpuinfo_x86 structure. This avoids multiple calls to the hypervisor detection function. This patch adds function to detect if we are running under VMware. The current way to check if we are on VMware is following, # check if "hypervisor present bit" is set, if so read the 0x40000000 cpuid leaf and check for "VMwareVMware" signature. # if the above fails, check the DMI vendors name for "VMware" string if we find one we query the VMware hypervisor port to check if we are under VMware. The DMI + "VMware hypervisor port check" is needed for older VMware products, which don't implement the hypervisor signature cpuid leaf. Also note that since we are checking for the DMI signature the hypervisor port should never be accessed on native hardware. This patch also adds a hypervisor_get_tsc_freq function, instead of calibrating the frequency which can be error prone in virtualized environment, we ask the hypervisor for it. We get the frequency from the hypervisor by accessing the hypervisor port if we are running on VMware. Other hypervisors too can add code to the generic routine to get frequency on their platform. Signed-off-by: Alok N Kataria Signed-off-by: Dan Hecht Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/Makefile | 1 + arch/x86/kernel/cpu/common.c | 2 + arch/x86/kernel/cpu/hypervisor.c | 48 ++++++++++++++++++++++ arch/x86/kernel/cpu/vmware.c | 88 ++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/setup.c | 7 ++++ arch/x86/kernel/tsc.c | 9 +++- 6 files changed, 154 insertions(+), 1 deletion(-) create mode 100644 arch/x86/kernel/cpu/hypervisor.c create mode 100644 arch/x86/kernel/cpu/vmware.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index 82ec6075c057..a5c04e88777e 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -4,6 +4,7 @@ obj-y := intel_cacheinfo.o addon_cpuid_features.o obj-y += proc.o capflags.o powerflags.o common.o +obj-y += vmware.o hypervisor.o obj-$(CONFIG_X86_32) += bugs.o cmpxchg.o obj-$(CONFIG_X86_64) += bugs_64.o diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index b9c9ea0217a9..b88595c36254 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -36,6 +36,7 @@ #include #include #include +#include #include "cpu.h" @@ -703,6 +704,7 @@ static void __cpuinit identify_cpu(struct cpuinfo_x86 *c) detect_ht(c); #endif + init_hypervisor(c); /* * On SMP, boot_cpu_data holds the common feature set between * all CPUs; so make sure that we indicate which features are diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c new file mode 100644 index 000000000000..7bd55064ffe9 --- /dev/null +++ b/arch/x86/kernel/cpu/hypervisor.c @@ -0,0 +1,48 @@ +/* + * Common hypervisor code + * + * Copyright (C) 2008, VMware, Inc. + * Author : Alok N Kataria + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more + * details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. + * + */ + +#include +#include + +static inline void __cpuinit +detect_hypervisor_vendor(struct cpuinfo_x86 *c) +{ + if (vmware_platform()) { + c->x86_hyper_vendor = X86_HYPER_VENDOR_VMWARE; + } else { + c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE; + } +} + +unsigned long get_hypervisor_tsc_freq(void) +{ + if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) + return vmware_get_tsc_khz(); + return 0; +} + +void __cpuinit init_hypervisor(struct cpuinfo_x86 *c) +{ + detect_hypervisor_vendor(c); +} + diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c new file mode 100644 index 000000000000..d5d1b75a4b77 --- /dev/null +++ b/arch/x86/kernel/cpu/vmware.c @@ -0,0 +1,88 @@ +/* + * VMware Detection code. + * + * Copyright (C) 2008, VMware, Inc. + * Author : Alok N Kataria + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more + * details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. + * + */ + +#include +#include + +#define CPUID_VMWARE_INFO_LEAF 0x40000000 +#define VMWARE_HYPERVISOR_MAGIC 0x564D5868 +#define VMWARE_HYPERVISOR_PORT 0x5658 + +#define VMWARE_PORT_CMD_GETVERSION 10 +#define VMWARE_PORT_CMD_GETHZ 45 + +#define VMWARE_PORT(cmd, eax, ebx, ecx, edx) \ + __asm__("inl (%%dx)" : \ + "=a"(eax), "=c"(ecx), "=d"(edx), "=b"(ebx) : \ + "0"(VMWARE_HYPERVISOR_MAGIC), \ + "1"(VMWARE_PORT_CMD_##cmd), \ + "2"(VMWARE_HYPERVISOR_PORT), "3"(0) : \ + "memory"); + +static inline int __vmware_platform(void) +{ + uint32_t eax, ebx, ecx, edx; + VMWARE_PORT(GETVERSION, eax, ebx, ecx, edx); + return eax != (uint32_t)-1 && ebx == VMWARE_HYPERVISOR_MAGIC; +} + +static unsigned long __vmware_get_tsc_khz(void) +{ + uint64_t tsc_hz; + uint32_t eax, ebx, ecx, edx; + + VMWARE_PORT(GETHZ, eax, ebx, ecx, edx); + + if (eax == (uint32_t)-1) + return 0; + tsc_hz = eax | (((uint64_t)ebx) << 32); + do_div(tsc_hz, 1000); + BUG_ON(tsc_hz >> 32); + return tsc_hz; +} + +int vmware_platform(void) +{ + if (cpu_has_hypervisor) { + unsigned int eax, ebx, ecx, edx; + char hyper_vendor_id[13]; + + cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &ebx, &ecx, &edx); + memcpy(hyper_vendor_id + 0, &ebx, 4); + memcpy(hyper_vendor_id + 4, &ecx, 4); + memcpy(hyper_vendor_id + 8, &edx, 4); + hyper_vendor_id[12] = '\0'; + if (!strcmp(hyper_vendor_id, "VMwareVMware")) + return 1; + } else if (dmi_available && dmi_name_in_vendors("VMware") && + __vmware_platform()) + return 1; + + return 0; +} + +unsigned long vmware_get_tsc_khz(void) +{ + BUG_ON(!vmware_platform()); + return __vmware_get_tsc_khz(); +} diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 0fa6790c1dd3..f44dadfb32cf 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -98,6 +98,7 @@ #include #include +#include #include #include @@ -909,6 +910,12 @@ void __init setup_arch(char **cmdline_p) dmi_check_system(bad_bios_dmi_table); + /* + * VMware detection requires dmi to be available, so this + * needs to be done after dmi_scan_machine, for the BP. + */ + init_hypervisor(&boot_cpu_data); + #ifdef CONFIG_X86_32 probe_roms(); #endif diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 62348e4fd8d1..6dbf0bcb44a8 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -15,6 +15,7 @@ #include #include #include +#include unsigned int cpu_khz; /* TSC clocks / usec, not used here */ EXPORT_SYMBOL(cpu_khz); @@ -352,9 +353,15 @@ unsigned long native_calibrate_tsc(void) { u64 tsc1, tsc2, delta, ref1, ref2; unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX; - unsigned long flags, latch, ms, fast_calibrate; + unsigned long flags, latch, ms, fast_calibrate, tsc_khz; int hpet = is_hpet_enabled(), i, loopmin; + tsc_khz = get_hypervisor_tsc_freq(); + if (tsc_khz) { + printk(KERN_INFO "TSC: Frequency read from the hypervisor\n"); + return tsc_khz; + } + local_irq_save(flags); fast_calibrate = quick_pit_calibrate(); local_irq_restore(flags); -- cgit v1.2.3 From eca0cd028bdf0f6aaceb0d023e9c7501079a7dda Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Fri, 31 Oct 2008 12:01:58 -0700 Subject: x86: Add a synthetic TSC_RELIABLE feature bit. Impact: Changes timebase calibration on Vmware. Use the synthetic TSC_RELIABLE bit to workaround virtualization anomalies. Virtual TSCs can be kept nearly in sync, but because the virtual TSC offset is set by software, it's not perfect. So, the TSC synchronization test can fail. Even then the TSC can be used as a clocksource since the VMware platform exports a reliable TSC to the guest for timekeeping purposes. Use this bit to check if we need to skip the TSC sync checks. Along with this also set the CONSTANT_TSC bit when on VMware, since we still want to use TSC as clocksource on VM running over hardware which has unsynchronized TSC's (opteron's), since the hypervisor will take care of providing consistent TSC to the guest. Signed-off-by: Alok N Kataria Signed-off-by: Dan Hecht Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/hypervisor.c | 11 ++++++++++- arch/x86/kernel/cpu/vmware.c | 18 ++++++++++++++++++ arch/x86/kernel/tsc_sync.c | 8 +++++++- 3 files changed, 35 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c index 7bd55064ffe9..35ae2b75226d 100644 --- a/arch/x86/kernel/cpu/hypervisor.c +++ b/arch/x86/kernel/cpu/hypervisor.c @@ -41,8 +41,17 @@ unsigned long get_hypervisor_tsc_freq(void) return 0; } +static inline void __cpuinit +hypervisor_set_feature_bits(struct cpuinfo_x86 *c) +{ + if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) { + vmware_set_feature_bits(c); + return; + } +} + void __cpuinit init_hypervisor(struct cpuinfo_x86 *c) { detect_hypervisor_vendor(c); + hypervisor_set_feature_bits(c); } - diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c index d5d1b75a4b77..2ac4394fcb90 100644 --- a/arch/x86/kernel/cpu/vmware.c +++ b/arch/x86/kernel/cpu/vmware.c @@ -86,3 +86,21 @@ unsigned long vmware_get_tsc_khz(void) BUG_ON(!vmware_platform()); return __vmware_get_tsc_khz(); } + +/* + * VMware hypervisor takes care of exporting a reliable TSC to the guest. + * Still, due to timing difference when running on virtual cpus, the TSC can + * be marked as unstable in some cases. For example, the TSC sync check at + * bootup can fail due to a marginal offset between vcpus' TSCs (though the + * TSCs do not drift from each other). Also, the ACPI PM timer clocksource + * is not suitable as a watchdog when running on a hypervisor because the + * kernel may miss a wrap of the counter if the vcpu is descheduled for a + * long time. To skip these checks at runtime we set these capability bits, + * so that the kernel could just trust the hypervisor with providing a + * reliable virtual TSC that is suitable for timekeeping. + */ +void __cpuinit vmware_set_feature_bits(struct cpuinfo_x86 *c) +{ + set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE); +} diff --git a/arch/x86/kernel/tsc_sync.c b/arch/x86/kernel/tsc_sync.c index 9ffb01c31c40..5977c40a138f 100644 --- a/arch/x86/kernel/tsc_sync.c +++ b/arch/x86/kernel/tsc_sync.c @@ -108,6 +108,12 @@ void __cpuinit check_tsc_sync_source(int cpu) if (unsynchronized_tsc()) return; + if (boot_cpu_has(X86_FEATURE_TSC_RELIABLE)) { + printk(KERN_INFO + "Skipping synchronization checks as TSC is reliable.\n"); + return; + } + printk(KERN_INFO "checking TSC synchronization [CPU#%d -> CPU#%d]:", smp_processor_id(), cpu); @@ -161,7 +167,7 @@ void __cpuinit check_tsc_sync_target(void) { int cpus = 2; - if (unsynchronized_tsc()) + if (unsynchronized_tsc() || boot_cpu_has(X86_FEATURE_TSC_RELIABLE)) return; /* -- cgit v1.2.3 From 395628ef4ea12ff0748099f145363b5e33c69acb Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Fri, 24 Oct 2008 17:22:01 -0700 Subject: x86: Skip verification by the watchdog for TSC clocksource. Impact: Changes timekeeping on Vmware (or with tsc=reliable). This is achieved by resetting the CLOCKSOURCE_MUST_VERIFY flag. We add a tsc=reliable commandline option to enable this. This enables legacy hardware without HPET, LAPIC, or ACPI timers to enter high-resolution timer mode. Along with that have extended this to be used in virtualization environement too. Now we also set this flag if the X86_FEATURE_TSC_RELIABLE bit is set. This is important since there is a wrap-around problem with the acpi_pm timer. The acpi_pm counter is just 24bits and this can overflow in ~4 seconds. With the NO_HZ kernels in virtualized environment, there can be situations when the guest is descheduled for longer duration, as a result we may miss the wrap of the acpi counter. When TSC is used as a clocksource and acpi_pm timer is being used as the watchdog clocksource this error in acpi_pm results in TSC being marked as unstable, and essentially results in time dropping in chunks of 4 seconds whenever this wrap is missed. Since the virtualized TSC is reliable on VMware, we should always use the TSCs clocksource on VMware, so we skip the verfication at runtime, by checking for the feature bit. Since we reset the flag for mgeode systems too, i have combined the mgeode case with the feature bit check. Signed-off-by: Jeff Hansen Signed-off-by: Alok N Kataria Signed-off-by: Dan Hecht Signed-off-by: H. Peter Anvin --- arch/x86/kernel/tsc.c | 33 +++++++++++++++++++++------------ 1 file changed, 21 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 6dbf0bcb44a8..ee01cd96b5e1 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -32,6 +32,7 @@ static int tsc_unstable; erroneous rdtsc usage on !cpu_has_tsc processors */ static int tsc_disabled = -1; +static int tsc_clocksource_reliable; /* * Scheduler clock - returns current time in nanosec units. */ @@ -99,6 +100,15 @@ int __init notsc_setup(char *str) __setup("notsc", notsc_setup); +static int __init tsc_setup(char *str) +{ + if (!strcmp(str, "reliable")) + tsc_clocksource_reliable = 1; + return 1; +} + +__setup("tsc=", tsc_setup); + #define MAX_RETRIES 5 #define SMI_TRESHOLD 50000 @@ -738,24 +748,21 @@ static struct dmi_system_id __initdata bad_tsc_dmi_table[] = { {} }; -/* - * Geode_LX - the OLPC CPU has a possibly a very reliable TSC - */ +static void __init check_system_tsc_reliable(void) +{ #ifdef CONFIG_MGEODE_LX -/* RTSC counts during suspend */ + /* RTSC counts during suspend */ #define RTSC_SUSP 0x100 - -static void __init check_geode_tsc_reliable(void) -{ unsigned long res_low, res_high; rdmsr_safe(MSR_GEODE_BUSCONT_CONF0, &res_low, &res_high); + /* Geode_LX - the OLPC CPU has a possibly a very reliable TSC */ if (res_low & RTSC_SUSP) - clocksource_tsc.flags &= ~CLOCK_SOURCE_MUST_VERIFY; -} -#else -static inline void check_geode_tsc_reliable(void) { } + tsc_clocksource_reliable = 1; #endif + if (boot_cpu_has(X86_FEATURE_TSC_RELIABLE)) + tsc_clocksource_reliable = 1; +} /* * Make an educated guess if the TSC is trustworthy and synchronized @@ -790,6 +797,8 @@ static void __init init_tsc_clocksource(void) { clocksource_tsc.mult = clocksource_khz2mult(tsc_khz, clocksource_tsc.shift); + if (tsc_clocksource_reliable) + clocksource_tsc.flags &= ~CLOCK_SOURCE_MUST_VERIFY; /* lower the rating if we already know its unstable: */ if (check_tsc_unstable()) { clocksource_tsc.rating = 0; @@ -850,7 +859,7 @@ void __init tsc_init(void) if (unsynchronized_tsc()) mark_tsc_unstable("TSCs unsynchronized"); - check_geode_tsc_reliable(); + check_system_tsc_reliable(); init_tsc_clocksource(); } -- cgit v1.2.3 From 6bdbfe99916398dbb28d83833cc04757110f2738 Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Mon, 3 Nov 2008 11:31:28 -0800 Subject: x86: VMware: Fix vmware_get_tsc code Impact: Fix possible failure to calibrate the TSC on Vmware near 4 GHz The current version of the code to get the tsc frequency from the VMware hypervisor, will be broken on processor with frequency (4G-1) HZ, because on such processors eax will have UINT_MAX and that would be legitimate. We instead check that EBX did change to decide if we were able to read the frequency from the hypervisor. Signed-off-by: Alok N Kataria Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/vmware.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c index 2ac4394fcb90..a0905ecfe7d2 100644 --- a/arch/x86/kernel/cpu/vmware.c +++ b/arch/x86/kernel/cpu/vmware.c @@ -36,7 +36,7 @@ "=a"(eax), "=c"(ecx), "=d"(edx), "=b"(ebx) : \ "0"(VMWARE_HYPERVISOR_MAGIC), \ "1"(VMWARE_PORT_CMD_##cmd), \ - "2"(VMWARE_HYPERVISOR_PORT), "3"(0) : \ + "2"(VMWARE_HYPERVISOR_PORT), "3"(UINT_MAX) : \ "memory"); static inline int __vmware_platform(void) @@ -53,7 +53,7 @@ static unsigned long __vmware_get_tsc_khz(void) VMWARE_PORT(GETHZ, eax, ebx, ecx, edx); - if (eax == (uint32_t)-1) + if (ebx == UINT_MAX) return 0; tsc_hz = eax | (((uint64_t)ebx) << 32); do_div(tsc_hz, 1000); -- cgit v1.2.3 From 70de9a97049e0ba79dc040868564408d5ce697f9 Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Mon, 3 Nov 2008 11:18:47 -0800 Subject: x86: don't use tsc_khz to calculate lpj if notsc is passed Impact: fix udelay when "notsc" boot parameter is passed With notsc passed on commandline, tsc may not be used for udelays, make sure that we do not use tsc_khz to calculate the lpj value in such cases. Reported-by: Bartlomiej Zolnierkiewicz Signed-off-by: Alok N Kataria Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/tsc.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 62348e4fd8d1..2ef80e301925 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -813,10 +813,6 @@ void __init tsc_init(void) cpu_khz = calibrate_cpu(); #endif - lpj = ((u64)tsc_khz * 1000); - do_div(lpj, HZ); - lpj_fine = lpj; - printk("Detected %lu.%03lu MHz processor.\n", (unsigned long)cpu_khz / 1000, (unsigned long)cpu_khz % 1000); @@ -836,6 +832,10 @@ void __init tsc_init(void) /* now allow native_sched_clock() to use rdtsc */ tsc_disabled = 0; + lpj = ((u64)tsc_khz * 1000); + do_div(lpj, HZ); + lpj_fine = lpj; + use_tsc_delay(); /* Check and install the TSC clocksource */ dmi_check_system(bad_tsc_dmi_table); -- cgit v1.2.3 From 124ffe1456d6efea5b32cc6d36e3fa434cdc84d9 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 3 Nov 2008 19:23:01 -0800 Subject: x86: signal_64: remove unused code in __setup_rt_frame() Impact: cleanup sizeof(*set) is always 8 on x86_64. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_64.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 83990db82f74..cfbb60a5f9d2 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -251,11 +251,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); err |= setup_sigcontext(&frame->uc.uc_mcontext, regs, set->sig[0], me); err |= __put_user(fp, &frame->uc.uc_mcontext.fpstate); - if (sizeof(*set) == 16) { - __put_user(set->sig[0], &frame->uc.uc_sigmask.sig[0]); - __put_user(set->sig[1], &frame->uc.uc_sigmask.sig[1]); - } else - err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); + err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); /* Set up to return from userspace. If provided, use a stub already in userspace. */ -- cgit v1.2.3 From 6cf87efbc7a3676e0ad7c9622ec6aec244a593bc Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 4 Nov 2008 10:42:23 +0100 Subject: x86 debug: mark early_printk.o as notrace Impact: do not do function-tracing in the early-printk code this is useful when earlyprintk=vga,keep is used to debug tracer plugins. Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index e489ff9cb3e2..943fe6026c64 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -12,6 +12,7 @@ CFLAGS_REMOVE_tsc.o = -pg CFLAGS_REMOVE_rtc.o = -pg CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg +CFLAGS_REMOVE_early_printk.o = -pg endif # -- cgit v1.2.3 From fd8cd7e1919fc1c27fe2fdccd2a1cd32f791ef0f Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Mon, 3 Nov 2008 15:50:38 -0800 Subject: x86: vmware: look for DMI string in the product serial key Impact: Should permit VMware detection on older platforms where the vendor is changed. Could theoretically cause a regression if some weird serial number scheme contains the string "VMware" by pure chance. Seems unlikely, especially with the mixed case. In some user configured cases, VMware may choose not to put a VMware specific DMI string, but the product serial key is always there and is VMware specific. Add a interface to check the serial key, when checking for VMware in the DMI information. Signed-off-by: Alok N Kataria Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/vmware.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c index a0905ecfe7d2..c034bda842d9 100644 --- a/arch/x86/kernel/cpu/vmware.c +++ b/arch/x86/kernel/cpu/vmware.c @@ -61,6 +61,11 @@ static unsigned long __vmware_get_tsc_khz(void) return tsc_hz; } +/* + * While checking the dmi string infomation, just checking the product + * serial key should be enough, as this will always have a VMware + * specific string when running under VMware hypervisor. + */ int vmware_platform(void) { if (cpu_has_hypervisor) { @@ -74,7 +79,7 @@ int vmware_platform(void) hyper_vendor_id[12] = '\0'; if (!strcmp(hyper_vendor_id, "VMwareVMware")) return 1; - } else if (dmi_available && dmi_name_in_vendors("VMware") && + } else if (dmi_available && dmi_name_in_serial("VMware") && __vmware_platform()) return 1; -- cgit v1.2.3 From 64ccf2f9a70a06ba56cd8cedfa610b4e77181587 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Wed, 5 Nov 2008 22:11:56 -0600 Subject: x86: uv: Add UV watchlist bios call Add UV bios calls to allocate and free watchlists. Signed-off-by: Russ Anderson Signed-off-by: H. Peter Anvin --- arch/x86/kernel/bios_uv.c | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index 7cefb7170e75..4c02b2799216 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -100,6 +100,39 @@ s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher, return ret; } +int +uv_bios_mq_watchlist_alloc(int blade, void *mq, unsigned int mq_size, + unsigned long *intr_mmr_offset) +{ + union uv_watchlist_u size_blade; + unsigned long addr; + u64 watchlist; + s64 ret; + + addr = (unsigned long)mq; + size_blade.size = mq_size; + size_blade.blade = blade; + + /* + * bios returns watchlist number or negative error number. + */ + ret = (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_ALLOC, addr, + size_blade.val, (u64)intr_mmr_offset, + (u64)&watchlist, 0); + if (ret < BIOS_STATUS_SUCCESS) + return ret; + + return watchlist; +} +EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_alloc); + +int +uv_bios_mq_watchlist_free(int blade, int watchlist_num) +{ + return (int)uv_bios_call_irqsave(UV_BIOS_WATCHLIST_FREE, + blade, watchlist_num, 0, 0, 0); +} +EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_free); s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) { -- cgit v1.2.3 From e8929c8a6acbecbd629b8e3f2d1a2546ec4ebdfc Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Wed, 5 Nov 2008 22:13:44 -0600 Subject: x86: uv: Add UV memory protection bios call Add UV bios call to change memory protections. Signed-off-by: Russ Anderson Signed-off-by: H. Peter Anvin --- arch/x86/kernel/bios_uv.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index 4c02b2799216..7cf6fc3d1c10 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -134,6 +134,14 @@ uv_bios_mq_watchlist_free(int blade, int watchlist_num) } EXPORT_SYMBOL_GPL(uv_bios_mq_watchlist_free); +s64 +uv_bios_change_memprotect(u64 paddr, u64 len, enum uv_memprotect perms) +{ + return uv_bios_call_irqsave(UV_BIOS_MEMPROTECT, paddr, len, + perms, 0, 0); +} +EXPORT_SYMBOL_GPL(uv_bios_change_memprotect); + s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) { return uv_bios_call(UV_BIOS_FREQ_BASE, clock_type, -- cgit v1.2.3 From 23c357003b3671cdfb17bc4d5383589e74b71511 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Wed, 5 Nov 2008 22:15:13 -0600 Subject: x86: uv: Add UV reserved page bios call Add UV bios call to get the address of the reserved page. Signed-off-by: Russ Anderson Signed-off-by: H. Peter Anvin --- arch/x86/kernel/bios_uv.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index 7cf6fc3d1c10..d22d0f1bbea0 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -142,6 +142,17 @@ uv_bios_change_memprotect(u64 paddr, u64 len, enum uv_memprotect perms) } EXPORT_SYMBOL_GPL(uv_bios_change_memprotect); +s64 +uv_bios_reserved_page_pa(u64 buf, u64 *cookie, u64 *addr, u64 *len) +{ + s64 ret; + + ret = uv_bios_call_irqsave(UV_BIOS_GET_PARTITION_ADDR, (u64)cookie, + (u64)addr, buf, (u64)len, 0); + return ret; +} +EXPORT_SYMBOL_GPL(uv_bios_reserved_page_pa); + s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) { return uv_bios_call(UV_BIOS_FREQ_BASE, clock_type, -- cgit v1.2.3 From c78d0cf2925bffae8a6f00e7d9b8e971b0392edd Mon Sep 17 00:00:00 2001 From: Ben Hutchings Date: Wed, 5 Nov 2008 12:04:46 +0000 Subject: x86: don't allow nr_irqs > NR_IRQS Impact: fix boot hang on 32-bit systems with more than 224 IO-APIC pins On some 32-bit systems with a lot of IO-APICs probe_nr_irqs() can return a value larger than NR_IRQS. This will lead to probe_irq_on() overrunning the irq_desc array. I hit this when running net-next-2.6 (close to 2.6.28-rc3) on a Supermicro dual Xeon system. NR_IRQS is 224 but probe_nr_irqs() detects 5 IOAPICs and returns 240. Here are the log messages: Tue Nov 4 16:53:47 2008 ACPI: IOAPIC (id[0x01] address[0xfec00000] gsi_base[0]) Tue Nov 4 16:53:47 2008 IOAPIC[0]: apic_id 1, version 32, address 0xfec00000, GSI 0-23 Tue Nov 4 16:53:47 2008 ACPI: IOAPIC (id[0x02] address[0xfec81000] gsi_base[24]) Tue Nov 4 16:53:47 2008 IOAPIC[1]: apic_id 2, version 32, address 0xfec81000, GSI 24-47 Tue Nov 4 16:53:47 2008 ACPI: IOAPIC (id[0x03] address[0xfec81400] gsi_base[48]) Tue Nov 4 16:53:47 2008 IOAPIC[2]: apic_id 3, version 32, address 0xfec81400, GSI 48-71 Tue Nov 4 16:53:47 2008 ACPI: IOAPIC (id[0x04] address[0xfec82000] gsi_base[72]) Tue Nov 4 16:53:47 2008 IOAPIC[3]: apic_id 4, version 32, address 0xfec82000, GSI 72-95 Tue Nov 4 16:53:47 2008 ACPI: IOAPIC (id[0x05] address[0xfec82400] gsi_base[96]) Tue Nov 4 16:53:47 2008 IOAPIC[4]: apic_id 5, version 32, address 0xfec82400, GSI 96-119 Tue Nov 4 16:53:47 2008 ACPI: INT_SRC_OVR (bus 0 bus_irq 0 global_irq 2 high edge) Tue Nov 4 16:53:47 2008 ACPI: INT_SRC_OVR (bus 0 bus_irq 9 global_irq 9 high level) Tue Nov 4 16:53:47 2008 Enabling APIC mode: Flat. Using 5 I/O APICs Signed-off-by: Ben Hutchings Acked-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index b764d7429c61..7a3f2028e2eb 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3611,6 +3611,8 @@ int __init probe_nr_irqs(void) /* something wrong ? */ if (nr < nr_min) nr = nr_min; + if (WARN_ON(nr > NR_IRQS)) + nr = NR_IRQS; return nr; } -- cgit v1.2.3 From 60a7ecf42661f2b22168751298592da6ee210c9e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Nov 2008 16:05:44 -0500 Subject: ftrace: add quick function trace stop Impact: quick start and stop of function tracer This patch adds a way to disable the function tracer quickly without the need to run kstop_machine. It adds a new variable called function_trace_stop which will stop the calls to functions from mcount when set. This is just an on/off switch and does not handle recursion like preempt_disable(). It's main purpose is to help other tracers/debuggers start and stop tracing fuctions without the need to call kstop_machine. The config option HAVE_FUNCTION_TRACE_MCOUNT_TEST is added for archs that implement the testing of the function_trace_stop in the mcount arch dependent code. Otherwise, the test is done in the C code. x86 is the only arch at the moment that supports this. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 6 ++++++ arch/x86/kernel/entry_64.S | 5 +++++ 2 files changed, 11 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 28b597ef9ca1..9134de814c97 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1157,6 +1157,9 @@ ENTRY(mcount) END(mcount) ENTRY(ftrace_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + pushl %eax pushl %ecx pushl %edx @@ -1180,6 +1183,9 @@ END(ftrace_caller) #else /* ! CONFIG_DYNAMIC_FTRACE */ ENTRY(mcount) + cmpl $0, function_trace_stop + jne ftrace_stub + cmpl $ftrace_stub, ftrace_trace_function jnz trace .globl ftrace_stub diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index b86f332c96a6..08aa6b10933c 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -68,6 +68,8 @@ ENTRY(mcount) END(mcount) ENTRY(ftrace_caller) + cmpl $0, function_trace_stop + jne ftrace_stub /* taken from glibc */ subq $0x38, %rsp @@ -103,6 +105,9 @@ END(ftrace_caller) #else /* ! CONFIG_DYNAMIC_FTRACE */ ENTRY(mcount) + cmpl $0, function_trace_stop + jne ftrace_stub + cmpq $ftrace_stub, ftrace_trace_function jnz trace .globl ftrace_stub -- cgit v1.2.3 From 4b33669e817a01dd99ff91df330d504ccfb2e99c Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 5 Nov 2008 18:30:25 -0800 Subject: x86: signal_32: do save_i387_xstate() at get_sigframe() Impact: cleanup move calling save_i387_xstate() into get_sigframe() from setup_sigcontext() like 64bit. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index a0efc1b3c4c9..6a05c74b4084 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -303,11 +303,7 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, err |= __put_user(regs->sp, &sc->sp_at_signal); err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); - tmp = save_i387_xstate(fpstate); - if (tmp < 0) - err = 1; - else - err |= __put_user(tmp ? fpstate : NULL, &sc->fpstate); + err |= __put_user(fpstate, &sc->fpstate); /* non-iBCS2 extensions.. */ err |= __put_user(mask, &sc->oldmask); @@ -350,6 +346,8 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, if (used_math()) { sp = sp - sig_xstate_size; *fpstate = (struct _fpstate *) sp; + if (save_i387_xstate(*fpstate) < 0) + return (void __user *)-1L; } sp -= frame_size; -- cgit v1.2.3 From ee7d523c124a186ce3a886868de9cd1d8bc991f3 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 5 Nov 2008 18:33:35 -0800 Subject: x86: signal_64: setup fpstate in setup_sigcontext() Impact: cleanup set fpstate field of signal context at setup_sigcontext(). Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_64.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index cfbb60a5f9d2..97d26fa62ac1 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -159,8 +159,9 @@ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) */ static inline int -setup_sigcontext(struct sigcontext __user *sc, struct pt_regs *regs, - unsigned long mask, struct task_struct *me) +setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, + struct pt_regs *regs, + unsigned long mask, struct task_struct *me) { int err = 0; @@ -188,6 +189,7 @@ setup_sigcontext(struct sigcontext __user *sc, struct pt_regs *regs, err |= __put_user(me->thread.error_code, &sc->err); err |= __put_user(regs->ip, &sc->ip); err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(fpstate, &sc->fpstate); err |= __put_user(mask, &sc->oldmask); err |= __put_user(me->thread.cr2, &sc->cr2); @@ -249,8 +251,8 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, err |= __put_user(sas_ss_flags(regs->sp), &frame->uc.uc_stack.ss_flags); err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, regs, set->sig[0], me); - err |= __put_user(fp, &frame->uc.uc_mcontext.fpstate); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, + regs, set->sig[0], me); err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); /* Set up to return from userspace. If provided, use a stub -- cgit v1.2.3 From 8735b7d0a2a6246faa406a8cdd1376bd0e689ba3 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 5 Nov 2008 18:34:35 -0800 Subject: x86: signal_64: make setup_sigcontext() similar Impact: cleanup remove passing task struct. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_64.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 97d26fa62ac1..3868c2a21793 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -160,8 +160,7 @@ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) static inline int setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, - struct pt_regs *regs, - unsigned long mask, struct task_struct *me) + struct pt_regs *regs, unsigned long mask) { int err = 0; @@ -185,13 +184,13 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, err |= __put_user(regs->r13, &sc->r13); err |= __put_user(regs->r14, &sc->r14); err |= __put_user(regs->r15, &sc->r15); - err |= __put_user(me->thread.trap_no, &sc->trapno); - err |= __put_user(me->thread.error_code, &sc->err); + err |= __put_user(current->thread.trap_no, &sc->trapno); + err |= __put_user(current->thread.error_code, &sc->err); err |= __put_user(regs->ip, &sc->ip); err |= __put_user(regs->flags, &sc->flags); err |= __put_user(fpstate, &sc->fpstate); err |= __put_user(mask, &sc->oldmask); - err |= __put_user(me->thread.cr2, &sc->cr2); + err |= __put_user(current->thread.cr2, &sc->cr2); return err; } @@ -251,8 +250,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, err |= __put_user(sas_ss_flags(regs->sp), &frame->uc.uc_stack.ss_flags); err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, - regs, set->sig[0], me); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]); err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); /* Set up to return from userspace. If provided, use a stub -- cgit v1.2.3 From d6f0f39b7d05e62b347c4352d070e4afb3ade4b5 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Tue, 4 Nov 2008 13:53:04 -0800 Subject: x86: add smp_mb() before sending INVALIDATE_TLB_VECTOR Impact: fix rare x2apic hang On x86, x2apic mode accesses for sending IPI's don't have serializing semantics. If the IPI receivner refers(in lock-free fashion) to some memory setup by the sender, the need for smp_mb() before sending the IPI becomes critical in x2apic mode. Add the smp_mb() in native_flush_tlb_others() before sending the IPI. Signed-off-by: Suresh Siddha Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_32.c | 6 ++++++ arch/x86/kernel/tlb_64.c | 5 +++++ 2 files changed, 11 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c index e00534b33534..f4049f3513b6 100644 --- a/arch/x86/kernel/tlb_32.c +++ b/arch/x86/kernel/tlb_32.c @@ -154,6 +154,12 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, flush_mm = mm; flush_va = va; cpus_or(flush_cpumask, cpumask, flush_cpumask); + + /* + * Make the above memory operations globally visible before + * sending the IPI. + */ + smp_mb(); /* * We have to send the IPI only to * CPUs affected. diff --git a/arch/x86/kernel/tlb_64.c b/arch/x86/kernel/tlb_64.c index dcbf7a1159ea..8f919ca69494 100644 --- a/arch/x86/kernel/tlb_64.c +++ b/arch/x86/kernel/tlb_64.c @@ -182,6 +182,11 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, f->flush_va = va; cpus_or(f->flush_cpumask, cpumask, f->flush_cpumask); + /* + * Make the above memory operations globally visible before + * sending the IPI. + */ + smp_mb(); /* * We have to send the IPI only to * CPUs affected. -- cgit v1.2.3 From 80be308dfa3798c7bad0fc81760b2faf83870e91 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 6 Nov 2008 14:59:05 +0100 Subject: AMD IOMMU: fix lazy IO/TLB flushing in unmap path Lazy flushing needs to take care of the unmap path too which is not yet implemented and leads to stale IO/TLB entries. This is fixed by this patch. Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 38e88d40ab10..4755bbc7ae5b 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -526,6 +526,9 @@ static void dma_ops_free_addresses(struct dma_ops_domain *dom, { address >>= PAGE_SHIFT; iommu_area_free(dom->bitmap, address, pages); + + if (address + pages >= dom->next_bit) + dom->need_flush = true; } /**************************************************************************** @@ -981,8 +984,10 @@ static void __unmap_single(struct amd_iommu *iommu, dma_ops_free_addresses(dma_dom, dma_addr, pages); - if (amd_iommu_unmap_flush) + if (amd_iommu_unmap_flush || dma_dom->need_flush) { iommu_flush_pages(iommu, dma_dom->domain.id, dma_addr, size); + dma_dom->need_flush = false; + } } /* -- cgit v1.2.3 From 8d00450d296dedec9ada38d43b83e79cca6fd5a3 Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Tue, 4 Nov 2008 12:52:44 -0200 Subject: Revert "x86: default to reboot via ACPI" This reverts commit c7ffa6c26277b403920e2255d10df849bd613380. the assumptio of this change was that this would not break any existing machine. Andrey Borzenkov reported troubles with the ACPI reboot method: the system would hang on reboot, necessiating a power cycle. Probably more systems are affected as well. Also, there are patches queued up for v2.6.29 to disable virtualization on emergency_restart() - which was the original motivation of this change. Reported-by: Andrey Borzenkov Bisected-by: Andrey Borzenkov Signed-off-by: Eduardo Habkost Acked-by: Avi Kivity Signed-off-by: Ingo Molnar --- arch/x86/kernel/reboot.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index f4c93f1cfc19..724adfc63cb9 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -29,11 +29,7 @@ EXPORT_SYMBOL(pm_power_off); static const struct desc_ptr no_idt = {}; static int reboot_mode; -/* - * Keyboard reset and triple fault may result in INIT, not RESET, which - * doesn't work when we're in vmx root mode. Try ACPI first. - */ -enum reboot_type reboot_type = BOOT_ACPI; +enum reboot_type reboot_type = BOOT_KBD; int reboot_force; #if defined(CONFIG_X86_32) && defined(CONFIG_SMP) -- cgit v1.2.3 From 15002fa9bf3a79ac9dcafba7ff308586936088b2 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 7 Nov 2008 19:25:36 -0800 Subject: x86: signal: cosmetic unification of setup_sigcontext() Impact: cleanup Make setup_sigcontext() same. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 32 ++++++++++++++++++++++++++++---- arch/x86/kernel/signal_64.c | 33 ++++++++++++++++++++++++++++----- 2 files changed, 56 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 6a05c74b4084..27a5c8174322 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -279,14 +279,20 @@ static int setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, struct pt_regs *regs, unsigned long mask) { - int tmp, err = 0; + int err = 0; - err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); - savesegment(gs, tmp); - err |= __put_user(tmp, (unsigned int __user *)&sc->gs); +#ifdef CONFIG_X86_32 + { + unsigned int tmp; + savesegment(gs, tmp); + err |= __put_user(tmp, (unsigned int __user *)&sc->gs); + } + err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); err |= __put_user(regs->es, (unsigned int __user *)&sc->es); err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); +#endif /* CONFIG_X86_32 */ + err |= __put_user(regs->di, &sc->di); err |= __put_user(regs->si, &sc->si); err |= __put_user(regs->bp, &sc->bp); @@ -295,13 +301,31 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, err |= __put_user(regs->dx, &sc->dx); err |= __put_user(regs->cx, &sc->cx); err |= __put_user(regs->ax, &sc->ax); +#ifdef CONFIG_X86_64 + err |= __put_user(regs->r8, &sc->r8); + err |= __put_user(regs->r9, &sc->r9); + err |= __put_user(regs->r10, &sc->r10); + err |= __put_user(regs->r11, &sc->r11); + err |= __put_user(regs->r12, &sc->r12); + err |= __put_user(regs->r13, &sc->r13); + err |= __put_user(regs->r14, &sc->r14); + err |= __put_user(regs->r15, &sc->r15); +#endif /* CONFIG_X86_64 */ + err |= __put_user(current->thread.trap_no, &sc->trapno); err |= __put_user(current->thread.error_code, &sc->err); err |= __put_user(regs->ip, &sc->ip); +#ifdef CONFIG_X86_32 err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); err |= __put_user(regs->flags, &sc->flags); err |= __put_user(regs->sp, &sc->sp_at_signal); err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); +#else /* !CONFIG_X86_32 */ + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->cs, &sc->cs); + err |= __put_user(0, &sc->gs); + err |= __put_user(0, &sc->fs); +#endif /* CONFIG_X86_32 */ err |= __put_user(fpstate, &sc->fpstate); diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 3868c2a21793..d2307e41fbdb 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -157,16 +157,23 @@ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) /* * Set up a signal frame. */ - -static inline int +static int setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, struct pt_regs *regs, unsigned long mask) { int err = 0; - err |= __put_user(regs->cs, &sc->cs); - err |= __put_user(0, &sc->gs); - err |= __put_user(0, &sc->fs); +#ifdef CONFIG_X86_32 + { + unsigned int tmp; + + savesegment(gs, tmp); + err |= __put_user(tmp, (unsigned int __user *)&sc->gs); + } + err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); + err |= __put_user(regs->es, (unsigned int __user *)&sc->es); + err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); +#endif /* CONFIG_X86_32 */ err |= __put_user(regs->di, &sc->di); err |= __put_user(regs->si, &sc->si); @@ -176,6 +183,7 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, err |= __put_user(regs->dx, &sc->dx); err |= __put_user(regs->cx, &sc->cx); err |= __put_user(regs->ax, &sc->ax); +#ifdef CONFIG_X86_64 err |= __put_user(regs->r8, &sc->r8); err |= __put_user(regs->r9, &sc->r9); err |= __put_user(regs->r10, &sc->r10); @@ -184,11 +192,26 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, err |= __put_user(regs->r13, &sc->r13); err |= __put_user(regs->r14, &sc->r14); err |= __put_user(regs->r15, &sc->r15); +#endif /* CONFIG_X86_64 */ + err |= __put_user(current->thread.trap_no, &sc->trapno); err |= __put_user(current->thread.error_code, &sc->err); err |= __put_user(regs->ip, &sc->ip); +#ifdef CONFIG_X86_32 + err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->sp, &sc->sp_at_signal); + err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); +#else /* !CONFIG_X86_32 */ + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->cs, &sc->cs); + err |= __put_user(0, &sc->gs); + err |= __put_user(0, &sc->fs); +#endif /* CONFIG_X86_32 */ + err |= __put_user(fpstate, &sc->fpstate); + + /* non-iBCS2 extensions.. */ err |= __put_user(mask, &sc->oldmask); err |= __put_user(current->thread.cr2, &sc->cr2); -- cgit v1.2.3 From 7cbaef9c83e58bbd4bdd534b09052b6c5ec457d5 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sat, 8 Nov 2008 17:05:38 +0100 Subject: sched: optimize sched_clock() a bit sched_clock() uses cycles_2_ns() needlessly - which is an irq-disabling variant of __cycles_2_ns(). Most of the time sched_clock() is called with irqs disabled already. The few places that call it with irqs enabled need to be updated. Signed-off-by: Ingo Molnar --- arch/x86/kernel/tsc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 2ef80e301925..424093b157d3 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -55,7 +55,7 @@ u64 native_sched_clock(void) rdtscll(this_offset); /* return the value in ns */ - return cycles_2_ns(this_offset); + return __cycles_2_ns(this_offset); } /* We need to define a real function for sched_clock, to override the -- cgit v1.2.3 From cb9e35dce94a1b9c59d46224e8a94377d673e204 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sat, 8 Nov 2008 20:27:00 +0100 Subject: x86: clean up rdtsc_barrier() use Impact: cleanup Move rdtsc_barrier() use to vsyscall_64.c where it's relied on, and point out its role in the context of its use. Signed-off-by: Ingo Molnar --- arch/x86/kernel/vsyscall_64.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index 0b8b6690a86d..ebf2f12900f5 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -128,7 +128,16 @@ static __always_inline void do_vgettimeofday(struct timeval * tv) gettimeofday(tv,NULL); return; } + + /* + * Surround the RDTSC by barriers, to make sure it's not + * speculated to outside the seqlock critical section and + * does not cause time warps: + */ + rdtsc_barrier(); now = vread(); + rdtsc_barrier(); + base = __vsyscall_gtod_data.clock.cycle_last; mask = __vsyscall_gtod_data.clock.mask; mult = __vsyscall_gtod_data.clock.mult; -- cgit v1.2.3 From 6c2e94033df5ca11149e52dd179b8dde3172e9bf Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 7 Nov 2008 12:33:49 +0100 Subject: x86: apic honour irq affinity which was set in early boot setup_ioapic_dest() is called after the non boot cpus have been brought up. It sets the irq affinity of all already configured interrupts to all cpus and ignores affinity settings which were done by the early bootup code. If the IRQ_NO_BALANCING or IRQ_AFFINITY_SET flags are set then use the affinity mask from the irq descriptor and not TARGET_CPUS. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 26 +++++++++++++++++++++----- 1 file changed, 21 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 7a3f2028e2eb..988ee89467d3 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3761,7 +3761,9 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) void __init setup_ioapic_dest(void) { int pin, ioapic, irq, irq_entry; + struct irq_desc *desc; struct irq_cfg *cfg; + cpumask_t mask; if (skip_ioapic_setup == 1) return; @@ -3778,16 +3780,30 @@ void __init setup_ioapic_dest(void) * cpu is online. */ cfg = irq_cfg(irq); - if (!cfg->vector) + if (!cfg->vector) { setup_IO_APIC_irq(ioapic, pin, irq, irq_trigger(irq_entry), irq_polarity(irq_entry)); + continue; + + } + + /* + * Honour affinities which have been set in early boot + */ + desc = irq_to_desc(irq); + if (desc->status & + (IRQ_NO_BALANCING | IRQ_AFFINITY_SET)) + mask = desc->affinity; + else + mask = TARGET_CPUS; + #ifdef CONFIG_INTR_REMAP - else if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq(irq, TARGET_CPUS); -#endif + if (intr_remapping_enabled) + set_ir_ioapic_affinity_irq(irq, mask); else - set_ioapic_affinity_irq(irq, TARGET_CPUS); +#endif + set_ioapic_affinity_irq(irq, mask); } } -- cgit v1.2.3 From f4166c54bfe04f64603974058e44fbd7cfef0ccc Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Sun, 9 Nov 2008 14:29:21 +0100 Subject: x86, bts: DS and BTS initialization Impact: widen BTS/PEBS ptrace enablement to more CPU models Move BTS initialisation out of an #ifdef CONFIG_X86_64 guard. Assume core2 BTS and DS layout for future models of family 6 processors. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/intel.c | 3 +-- arch/x86/kernel/ds.c | 9 ++++----- arch/x86/kernel/ptrace.c | 9 ++++----- 3 files changed, 9 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index cce0b6118d55..816f27f289b1 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -307,12 +307,11 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_P4); if (c->x86 == 6) set_cpu_cap(c, X86_FEATURE_P3); +#endif if (cpu_has_bts) ptrace_bts_init_intel(c); -#endif - detect_extended_topology(c); if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) { /* diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 2b69994fd3a8..c570252905a1 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -821,17 +821,16 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) switch (c->x86) { case 0x6: switch (c->x86_model) { + case 0 ... 0xC: + /* sorry, don't know about them */ + break; case 0xD: case 0xE: /* Pentium M */ ds_configure(&ds_cfg_var); break; - case 0xF: /* Core2 */ - case 0x1C: /* Atom */ + default: /* Core2, Atom, ... */ ds_configure(&ds_cfg_64); break; - default: - /* sorry, don't know about them */ - break; } break; case 0xF: diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 0a6d8c12e10d..06180dff5b2e 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -929,17 +929,16 @@ void __cpuinit ptrace_bts_init_intel(struct cpuinfo_x86 *c) switch (c->x86) { case 0x6: switch (c->x86_model) { + case 0 ... 0xC: + /* sorry, don't know about them */ + break; case 0xD: case 0xE: /* Pentium M */ bts_configure(&bts_cfg_pentium_m); break; - case 0xF: /* Core2 */ - case 0x1C: /* Atom */ + default: /* Core2, Atom, ... */ bts_configure(&bts_cfg_core2); break; - default: - /* sorry, don't know about them */ - break; } break; case 0xF: -- cgit v1.2.3 From 4e0304310f5180eee11b4edc72cf4cb78acdc634 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Mon, 10 Nov 2008 09:16:40 +0100 Subject: x86: apic - calibrate_APIC_clock remove redundant irq-enable-disable Impact: cleanup lapic_timer_setup is self-protected with local_irq_save/restore no need to use them in caller and levt is the per-cpu variable so no concurrent access from another cpu. Signed-off-by: Cyrill Gorcunov Acked-by: "Maciej W. Rozycki" Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 04a7f960bbc0..ce90dc184139 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -672,13 +672,9 @@ static int __init calibrate_APIC_clock(void) while (lapic_cal_loops <= LAPIC_CAL_LOOPS) cpu_relax(); - local_irq_disable(); - /* Stop the lapic timer */ lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, levt); - local_irq_enable(); - /* Jiffies delta */ deltaj = lapic_cal_j2 - lapic_cal_j1; apic_printk(APIC_VERBOSE, "... jiffies delta = %lu\n", deltaj); -- cgit v1.2.3 From ba21ebb6abac5c46e1d818d2ceda82420bd099ba Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Mon, 10 Nov 2008 09:16:41 +0100 Subject: x86: apic - use pr_ macros for logging Impact: cleanup It saves us some source lines and shift the code a bit righter. And a multiline comment style is fixed too :-) Signed-off-by: Cyrill Gorcunov Acked-by: "Maciej W. Rozycki" Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 116 ++++++++++++++++++++++--------------------------- 1 file changed, 53 insertions(+), 63 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index ce90dc184139..70879c9e3936 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -559,13 +559,13 @@ static int __init calibrate_by_pmtimer(long deltapm, long *delta) } else { res = (((u64)deltapm) * mult) >> 22; do_div(res, 1000000); - printk(KERN_WARNING "APIC calibration not consistent " + pr_warning("APIC calibration not consistent " "with PM Timer: %ldms instead of 100ms\n", (long)res); /* Correct the lapic counter value */ res = (((u64)(*delta)) * pm_100ms); do_div(res, deltapm); - printk(KERN_INFO "APIC delta adjusted to PM-Timer: " + pr_info("APIC delta adjusted to PM-Timer: " "%lu (%ld)\n", (unsigned long)res, *delta); *delta = (long)res; } @@ -645,8 +645,7 @@ static int __init calibrate_APIC_clock(void) */ if (calibration_result < (1000000 / HZ)) { local_irq_enable(); - printk(KERN_WARNING - "APIC frequency too slow, disabling apic timer\n"); + pr_warning("APIC frequency too slow, disabling apic timer\n"); return -1; } @@ -688,8 +687,7 @@ static int __init calibrate_APIC_clock(void) local_irq_enable(); if (levt->features & CLOCK_EVT_FEAT_DUMMY) { - printk(KERN_WARNING - "APIC timer disabled due to verification failure.\n"); + pr_warning("APIC timer disabled due to verification failure.\n"); return -1; } @@ -710,7 +708,7 @@ void __init setup_boot_APIC_clock(void) * broadcast mechanism is used. On UP systems simply ignore it. */ if (disable_apic_timer) { - printk(KERN_INFO "Disabling APIC timer\n"); + pr_info("Disabling APIC timer\n"); /* No broadcast on UP ! */ if (num_possible_cpus() > 1) { lapic_clockevent.mult = 1; @@ -737,7 +735,7 @@ void __init setup_boot_APIC_clock(void) if (nmi_watchdog != NMI_IO_APIC) lapic_clockevent.features &= ~CLOCK_EVT_FEAT_DUMMY; else - printk(KERN_WARNING "APIC timer registered as dummy," + pr_warning("APIC timer registered as dummy," " due to nmi_watchdog=%d!\n", nmi_watchdog); /* Setup the lapic or request the broadcast */ @@ -769,8 +767,7 @@ static void local_apic_timer_interrupt(void) * spurious. */ if (!evt->event_handler) { - printk(KERN_WARNING - "Spurious LAPIC timer interrupt on cpu %d\n", cpu); + pr_warning("Spurious LAPIC timer interrupt on cpu %d\n", cpu); /* Switch it off */ lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt); return; @@ -1089,7 +1086,7 @@ static void __cpuinit lapic_setup_esr(void) unsigned int oldvalue, value, maxlvt; if (!lapic_is_integrated()) { - printk(KERN_INFO "No ESR for 82489DX.\n"); + pr_info("No ESR for 82489DX.\n"); return; } @@ -1100,7 +1097,7 @@ static void __cpuinit lapic_setup_esr(void) * ESR disabled - we can't do anything useful with the * errors anyway - mbligh */ - printk(KERN_INFO "Leaving ESR disabled.\n"); + pr_info("Leaving ESR disabled.\n"); return; } @@ -1294,7 +1291,7 @@ void check_x2apic(void) rdmsr(MSR_IA32_APICBASE, msr, msr2); if (msr & X2APIC_ENABLE) { - printk("x2apic enabled by BIOS, switching to x2apic ops\n"); + pr_info("x2apic enabled by BIOS, switching to x2apic ops\n"); x2apic_preenabled = x2apic = 1; apic_ops = &x2apic_ops; } @@ -1306,7 +1303,7 @@ void enable_x2apic(void) rdmsr(MSR_IA32_APICBASE, msr, msr2); if (!(msr & X2APIC_ENABLE)) { - printk("Enabling x2apic\n"); + pr_info("Enabling x2apic\n"); wrmsr(MSR_IA32_APICBASE, msr | X2APIC_ENABLE, 0); } } @@ -1321,9 +1318,8 @@ void enable_IR_x2apic(void) return; if (!x2apic_preenabled && disable_x2apic) { - printk(KERN_INFO - "Skipped enabling x2apic and Interrupt-remapping " - "because of nox2apic\n"); + pr_info("Skipped enabling x2apic and Interrupt-remapping " + "because of nox2apic\n"); return; } @@ -1331,22 +1327,19 @@ void enable_IR_x2apic(void) panic("Bios already enabled x2apic, can't enforce nox2apic"); if (!x2apic_preenabled && skip_ioapic_setup) { - printk(KERN_INFO - "Skipped enabling x2apic and Interrupt-remapping " - "because of skipping io-apic setup\n"); + pr_info("Skipped enabling x2apic and Interrupt-remapping " + "because of skipping io-apic setup\n"); return; } ret = dmar_table_init(); if (ret) { - printk(KERN_INFO - "dmar_table_init() failed with %d:\n", ret); + pr_info("dmar_table_init() failed with %d:\n", ret); if (x2apic_preenabled) panic("x2apic enabled by bios. But IR enabling failed"); else - printk(KERN_INFO - "Not enabling x2apic,Intr-remapping\n"); + pr_info("Not enabling x2apic,Intr-remapping\n"); return; } @@ -1355,7 +1348,7 @@ void enable_IR_x2apic(void) ret = save_mask_IO_APIC_setup(); if (ret) { - printk(KERN_INFO "Saving IO-APIC state failed: %d\n", ret); + pr_info("Saving IO-APIC state failed: %d\n", ret); goto end; } @@ -1390,14 +1383,11 @@ end: if (!ret) { if (!x2apic_preenabled) - printk(KERN_INFO - "Enabled x2apic and interrupt-remapping\n"); + pr_info("Enabled x2apic and interrupt-remapping\n"); else - printk(KERN_INFO - "Enabled Interrupt-remapping\n"); + pr_info("Enabled Interrupt-remapping\n"); } else - printk(KERN_ERR - "Failed to enable Interrupt-remapping and x2apic\n"); + pr_err("Failed to enable Interrupt-remapping and x2apic\n"); #else if (!cpu_has_x2apic) return; @@ -1406,8 +1396,8 @@ end: panic("x2apic enabled prior OS handover," " enable CONFIG_INTR_REMAP"); - printk(KERN_INFO "Enable CONFIG_INTR_REMAP for enabling intr-remapping " - " and x2apic\n"); + pr_info("Enable CONFIG_INTR_REMAP for enabling intr-remapping " + " and x2apic\n"); #endif return; @@ -1424,7 +1414,7 @@ end: static int __init detect_init_APIC(void) { if (!cpu_has_apic) { - printk(KERN_INFO "No local APIC present\n"); + pr_info("No local APIC present\n"); return -1; } @@ -1465,8 +1455,8 @@ static int __init detect_init_APIC(void) * "lapic" specified. */ if (!force_enable_local_apic) { - printk(KERN_INFO "Local APIC disabled by BIOS -- " - "you can enable it with \"lapic\"\n"); + pr_info("Local APIC disabled by BIOS -- " + "you can enable it with \"lapic\"\n"); return -1; } /* @@ -1476,8 +1466,7 @@ static int __init detect_init_APIC(void) */ rdmsr(MSR_IA32_APICBASE, l, h); if (!(l & MSR_IA32_APICBASE_ENABLE)) { - printk(KERN_INFO - "Local APIC disabled by BIOS -- reenabling.\n"); + pr_info("Local APIC disabled by BIOS -- reenabling.\n"); l &= ~MSR_IA32_APICBASE_BASE; l |= MSR_IA32_APICBASE_ENABLE | APIC_DEFAULT_PHYS_BASE; wrmsr(MSR_IA32_APICBASE, l, h); @@ -1490,7 +1479,7 @@ static int __init detect_init_APIC(void) */ features = cpuid_edx(1); if (!(features & (1 << X86_FEATURE_APIC))) { - printk(KERN_WARNING "Could not enable APIC!\n"); + pr_warning("Could not enable APIC!\n"); return -1; } set_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); @@ -1501,14 +1490,14 @@ static int __init detect_init_APIC(void) if (l & MSR_IA32_APICBASE_ENABLE) mp_lapic_addr = l & MSR_IA32_APICBASE_BASE; - printk(KERN_INFO "Found and enabled local APIC!\n"); + pr_info("Found and enabled local APIC!\n"); apic_pm_activate(); return 0; no_apic: - printk(KERN_INFO "No local APIC present or hardware disabled\n"); + pr_info("No local APIC present or hardware disabled\n"); return -1; } #endif @@ -1584,12 +1573,12 @@ int __init APIC_init_uniprocessor(void) { #ifdef CONFIG_X86_64 if (disable_apic) { - printk(KERN_INFO "Apic disabled\n"); + pr_info("Apic disabled\n"); return -1; } if (!cpu_has_apic) { disable_apic = 1; - printk(KERN_INFO "Apic disabled by BIOS\n"); + pr_info("Apic disabled by BIOS\n"); return -1; } #else @@ -1601,8 +1590,8 @@ int __init APIC_init_uniprocessor(void) */ if (!cpu_has_apic && APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { - printk(KERN_ERR "BIOS bug, local APIC 0x%x not detected!...\n", - boot_cpu_physical_apicid); + pr_err("BIOS bug, local APIC 0x%x not detected!...\n", + boot_cpu_physical_apicid); clear_cpu_cap(&boot_cpu_data, X86_FEATURE_APIC); return -1; } @@ -1695,8 +1684,8 @@ void smp_spurious_interrupt(struct pt_regs *regs) add_pda(irq_spurious_count, 1); #else /* see sw-dev-man vol 3, chapter 7.4.13.5 */ - printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " - "should never happen.\n", smp_processor_id()); + pr_info("spurious APIC interrupt on CPU#%d, " + "should never happen.\n", smp_processor_id()); __get_cpu_var(irq_stat).irq_spurious_count++; #endif irq_exit(); @@ -1720,17 +1709,18 @@ void smp_error_interrupt(struct pt_regs *regs) ack_APIC_irq(); atomic_inc(&irq_err_count); - /* Here is what the APIC error bits mean: - 0: Send CS error - 1: Receive CS error - 2: Send accept error - 3: Receive accept error - 4: Reserved - 5: Send illegal vector - 6: Received illegal vector - 7: Illegal register address - */ - printk(KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", + /* + * Here is what the APIC error bits mean: + * 0: Send CS error + * 1: Receive CS error + * 2: Send accept error + * 3: Receive accept error + * 4: Reserved + * 5: Send illegal vector + * 6: Received illegal vector + * 7: Illegal register address + */ + pr_debug("APIC error on CPU%d: %02x(%02x)\n", smp_processor_id(), v , v1); irq_exit(); } @@ -1834,15 +1824,15 @@ void __cpuinit generic_processor_info(int apicid, int version) * Validate version */ if (version == 0x0) { - printk(KERN_WARNING "BIOS bug, APIC version is 0 for CPU#%d! " - "fixing up to 0x10. (tell your hw vendor)\n", - version); + pr_warning("BIOS bug, APIC version is 0 for CPU#%d! " + "fixing up to 0x10. (tell your hw vendor)\n", + version); version = 0x10; } apic_version[apicid] = version; if (num_processors >= NR_CPUS) { - printk(KERN_WARNING "WARNING: NR_CPUS limit of %i reached." + pr_warning("WARNING: NR_CPUS limit of %i reached." " Processor ignored.\n", NR_CPUS); return; } @@ -2205,7 +2195,7 @@ static int __init apic_set_verbosity(char *arg) else if (strcmp("verbose", arg) == 0) apic_verbosity = APIC_VERBOSE; else { - printk(KERN_WARNING "APIC Verbosity level %s not recognised" + pr_warning("APIC Verbosity level %s not recognised" " use apic=verbose or apic=debug\n", arg); return -EINVAL; } -- cgit v1.2.3 From 1de5b0854623d30d01d72cd4ea323eb5f39d1f16 Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Sun, 2 Nov 2008 16:04:18 +0000 Subject: x86: HPET: convert WARN_ON to WARN_ON_ONCE It is possible to flood the console with call traces if the WARN_ON condition is true because of the frequency with which this function is called. Signed-off-by: Matt Fleming Cc: mingo@elte.hu Cc: venkatesh.pallipadi@intel.com Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 77017e834cf7..f10f9461a43d 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -322,7 +322,7 @@ static int hpet_next_event(unsigned long delta, * what we wrote hit the chip before we compare it to the * counter. */ - WARN_ON((u32)hpet_readl(HPET_T0_CMP) != cnt); + WARN_ON_ONCE((u32)hpet_readl(HPET_T0_CMP) != cnt); return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } -- cgit v1.2.3 From 89d77a1eb60be916d85d9394bedbfa2037af89c5 Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Sun, 2 Nov 2008 16:04:20 +0000 Subject: x86: HPET: read from HPET_Tn_CMP() not HPET_T0_CMP In hpet_next_event() we check that the value we just wrote to HPET_Tn_CMP(timer) has reached the chip. Currently, we're checking that the value we wrote to HPET_Tn_CMP(timer) is in HPET_T0_CMP, which, if timer is anything other than timer 0, is likely to fail. Signed-off-by: Matt Fleming Cc: mingo@elte.hu Cc: venkatesh.pallipadi@intel.com Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index f10f9461a43d..cfe6aa56f71b 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -322,7 +322,7 @@ static int hpet_next_event(unsigned long delta, * what we wrote hit the chip before we compare it to the * counter. */ - WARN_ON_ONCE((u32)hpet_readl(HPET_T0_CMP) != cnt); + WARN_ON_ONCE((u32)hpet_readl(HPET_Tn_CMP(timer)) != cnt); return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } -- cgit v1.2.3 From 5ceb1a04187553e08c6ab60d30cee7c454ee139a Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Sun, 2 Nov 2008 22:23:13 +0000 Subject: x86: HPET: enter hpet_interrupt_handler with interrupts disabled Some functions that may be called from this handler require that interrupts are disabled. Also, combining IRQF_DISABLED and IRQF_SHARED does not reliably disable interrupts in a handler, so remove IRQF_SHARED from the irq flags (this irq is not shared anyway). Signed-off-by: Matt Fleming Cc: mingo@elte.hu Cc: venkatesh.pallipadi@intel.com Cc: "Will Newton" Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index cfe6aa56f71b..067d8de913f6 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -445,7 +445,7 @@ static int hpet_setup_irq(struct hpet_dev *dev) { if (request_irq(dev->irq, hpet_interrupt_handler, - IRQF_SHARED|IRQF_NOBALANCING, dev->name, dev)) + IRQF_DISABLED|IRQF_NOBALANCING, dev->name, dev)) return -1; disable_irq(dev->irq); -- cgit v1.2.3 From caf4b323b02a16c92fba449952ac6515ddc76d7a Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 11 Nov 2008 07:03:45 +0100 Subject: tracing, x86: add low level support for ftrace return tracing Impact: add infrastructure for function-return tracing Add low level support for ftrace return tracing. This plug-in stores return addresses on the thread_info structure of the current task. The index of the current return address is initialized when the task is the first one (init) and when a process forks (the child). It is not needed when a task does a sys_execve because after this syscall, it still needs to return on the kernel functions it called. Note that the code of return_to_handler has been suggested by Steven Rostedt as almost all of the ideas of improvements in this V3. For purpose of security, arch/x86/kernel/process_32.c is not traced because __switch_to() changes the current task during its execution. That could cause inconsistency in the stored return address of this function even if I didn't have any crash after testing with tracing on this function enabled. Signed-off-by: Frederic Weisbecker Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 6 ++ arch/x86/kernel/entry_32.S | 33 +++++++++ arch/x86/kernel/ftrace.c | 181 +++++++++++++++++++++++++++++++++++++++++++-- 3 files changed, 213 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index e489ff9cb3e2..1d8ed95da846 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -14,6 +14,11 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg endif +ifdef CONFIG_FUNCTION_RET_TRACER +# Don't trace __switch_to() but let it for function tracer +CFLAGS_REMOVE_process_32.o = -pg +endif + # # vsyscalls (which work on the user stack) should have # no stack-protector checks: @@ -65,6 +70,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o +obj-$(CONFIG_FUNCTION_RET_TRACER) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 9134de814c97..9a0ac85946db 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1188,6 +1188,10 @@ ENTRY(mcount) cmpl $ftrace_stub, ftrace_trace_function jnz trace +#ifdef CONFIG_FUNCTION_RET_TRACER + cmpl $ftrace_stub, ftrace_function_return + jnz trace_return +#endif .globl ftrace_stub ftrace_stub: ret @@ -1206,8 +1210,37 @@ trace: popl %edx popl %ecx popl %eax + jmp ftrace_stub +#ifdef CONFIG_FUNCTION_RET_TRACER +trace_return: + pushl %eax + pushl %ecx + pushl %edx + movl 0xc(%esp), %eax + pushl %eax + lea 0x4(%ebp), %eax + pushl %eax + call prepare_ftrace_return + addl $8, %esp + popl %edx + popl %ecx + popl %eax jmp ftrace_stub + +.globl return_to_handler +return_to_handler: + pushl $0 + pushl %eax + pushl %ecx + pushl %edx + call ftrace_return_to_handler + movl %eax, 0xc(%esp) + popl %edx + popl %ecx + popl %eax + ret +#endif /* CONFIG_FUNCTION_RET_TRACER */ END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 69149337f2fe..d68033bba223 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -18,10 +18,173 @@ #include #include +#include #include +#include -static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; + +#ifdef CONFIG_FUNCTION_RET_TRACER + +/* + * These functions are picked from those used on + * this page for dynamic ftrace. They have been + * simplified to ignore all traces in NMI context. + */ +static atomic_t in_nmi; + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); +} + +void ftrace_nmi_exit(void) +{ + atomic_dec(&in_nmi); +} + +/* + * Synchronize accesses to return adresses stack with + * interrupts. + */ +static raw_spinlock_t ret_stack_lock; + +/* Add a function return address to the trace stack on thread info.*/ +static int push_return_trace(unsigned long ret, unsigned long long time, + unsigned long func) +{ + int index; + struct thread_info *ti; + unsigned long flags; + int err = 0; + + raw_local_irq_save(flags); + __raw_spin_lock(&ret_stack_lock); + + ti = current_thread_info(); + /* The return trace stack is full */ + if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { + err = -EBUSY; + goto out; + } + + index = ++ti->curr_ret_stack; + ti->ret_stack[index].ret = ret; + ti->ret_stack[index].func = func; + ti->ret_stack[index].calltime = time; + +out: + __raw_spin_unlock(&ret_stack_lock); + raw_local_irq_restore(flags); + return err; +} + +/* Retrieve a function return address to the trace stack on thread info.*/ +static void pop_return_trace(unsigned long *ret, unsigned long long *time, + unsigned long *func) +{ + struct thread_info *ti; + int index; + unsigned long flags; + + raw_local_irq_save(flags); + __raw_spin_lock(&ret_stack_lock); + + ti = current_thread_info(); + index = ti->curr_ret_stack; + *ret = ti->ret_stack[index].ret; + *func = ti->ret_stack[index].func; + *time = ti->ret_stack[index].calltime; + ti->curr_ret_stack--; + + __raw_spin_unlock(&ret_stack_lock); + raw_local_irq_restore(flags); +} + +/* + * Send the trace to the ring-buffer. + * @return the original return address. + */ +unsigned long ftrace_return_to_handler(void) +{ + struct ftrace_retfunc trace; + pop_return_trace(&trace.ret, &trace.calltime, &trace.func); + trace.rettime = cpu_clock(raw_smp_processor_id()); + ftrace_function_return(&trace); + + return trace.ret; +} + +/* + * Hook the return address and push it in the stack of return addrs + * in current thread info. + */ +asmlinkage +void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) +{ + unsigned long old; + unsigned long long calltime; + int faulted; + unsigned long return_hooker = (unsigned long) + &return_to_handler; + + /* Nmi's are currently unsupported */ + if (atomic_read(&in_nmi)) + return; + + /* + * Protect against fault, even if it shouldn't + * happen. This tool is too much intrusive to + * ignore such a protection. + */ + asm volatile( + "1: movl (%[parent_old]), %[old]\n" + "2: movl %[return_hooker], (%[parent_replaced])\n" + " movl $0, %[faulted]\n" + + ".section .fixup, \"ax\"\n" + "3: movl $1, %[faulted]\n" + ".previous\n" + + ".section __ex_table, \"a\"\n" + " .long 1b, 3b\n" + " .long 2b, 3b\n" + ".previous\n" + + : [parent_replaced] "=rm" (parent), [old] "=r" (old), + [faulted] "=r" (faulted) + : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) + : "memory" + ); + + if (WARN_ON(faulted)) { + unregister_ftrace_return(); + return; + } + + if (WARN_ON(!__kernel_text_address(old))) { + unregister_ftrace_return(); + *parent = old; + return; + } + + calltime = cpu_clock(raw_smp_processor_id()); + + if (push_return_trace(old, calltime, self_addr) == -EBUSY) + *parent = old; +} + +static int __init init_ftrace_function_return(void) +{ + ret_stack_lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; + return 0; +} +device_initcall(init_ftrace_function_return); + + +#endif + +#ifdef CONFIG_DYNAMIC_FTRACE union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; @@ -31,17 +194,11 @@ union ftrace_code_union { } __attribute__((packed)); }; - static int ftrace_calc_offset(long ip, long addr) { return (int)(addr - ip); } -unsigned char *ftrace_nop_replace(void) -{ - return ftrace_nop; -} - unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -183,6 +340,15 @@ do_ftrace_mod_code(unsigned long ip, void *new_code) } + + +static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; + +unsigned char *ftrace_nop_replace(void) +{ + return ftrace_nop; +} + int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) @@ -292,3 +458,4 @@ int __init ftrace_dyn_arch_init(void *data) return 0; } +#endif -- cgit v1.2.3 From 867f7fb3ebb831970847b179e7df5a9ab10da16d Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 11 Nov 2008 11:18:14 +0100 Subject: tracing, x86: function return tracer, fix assembly constraints fix: arch/x86/kernel/ftrace.c: Assembler messages: arch/x86/kernel/ftrace.c:140: Error: missing ')' arch/x86/kernel/ftrace.c:140: Error: junk `(%ebp))' after expression arch/x86/kernel/ftrace.c:141: Error: missing ')' arch/x86/kernel/ftrace.c:141: Error: junk `(%ebp))' after expression the [parent_replaced] is used in an =rm fashion, so that constraint is correct in isolation - but [parent_old] aliases register %0 and uses it in an addressing mode that is only valid with registers - so change the constraint from =rm to =r. This fixes the build failure. Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index d68033bba223..9b2325a4d53c 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -151,7 +151,7 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) " .long 2b, 3b\n" ".previous\n" - : [parent_replaced] "=rm" (parent), [old] "=r" (old), + : [parent_replaced] "=r" (parent), [old] "=r" (old), [faulted] "=r" (faulted) : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) : "memory" -- cgit v1.2.3 From a3d732f93785da17e0137210deadb4616f5536fc Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Mon, 10 Nov 2008 16:16:31 -0600 Subject: x86, UV: fix redundant creation of sgi_uv Impact: fix double entry creation in /proc There is a collision between two UV functions: both uv_ptc_init() and gru_proc_init() try to make /proc/sgi_uv So move it's creation to a single place: uv_system_init() Signed-off-by: Cliff Wickman Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 2 ++ arch/x86/kernel/tlb_uv.c | 4 ---- 2 files changed, 2 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 85fb7dd48f67..d7213a1cb784 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include #include @@ -570,4 +571,5 @@ void __init uv_system_init(void) uv_cpu_init(); uv_scir_register_cpu_notifier(); + proc_mkdir("sgi_uv", NULL); } diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index 04431f34fd16..6a00e5faaa74 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -566,14 +566,10 @@ static int __init uv_ptc_init(void) if (!is_uv_system()) return 0; - if (!proc_mkdir("sgi_uv", NULL)) - return -EINVAL; - proc_uv_ptc = create_proc_entry(UV_PTC_BASENAME, 0444, NULL); if (!proc_uv_ptc) { printk(KERN_ERR "unable to create %s proc entry\n", UV_PTC_BASENAME); - remove_proc_entry("sgi_uv", NULL); return -EINVAL; } proc_uv_ptc->proc_fops = &proc_uv_ptc_operations; -- cgit v1.2.3 From 19b3e9671c5a219b8c34da2cc66e0ce7c3a501ae Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 11 Nov 2008 11:57:02 +0100 Subject: tracing: function return tracer, build fix fix: arch/x86/kernel/ftrace.c: In function 'ftrace_return_to_handler': arch/x86/kernel/ftrace.c:112: error: implicit declaration of function 'cpu_clock' cpu_clock() is implicitly included via a number of ways, but its real location is sched.h. (Build failure is triggerable if enough other kernel components are turned off.) Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 9b2325a4d53c..16a571dea2ef 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -14,6 +14,7 @@ #include #include #include +#include #include #include -- cgit v1.2.3 From d3ec5cae0921611ceae06464ef6291012dd9849f Mon Sep 17 00:00:00 2001 From: Ivan Vecera Date: Tue, 11 Nov 2008 14:33:44 +0100 Subject: x86: call machine_shutdown and stop all CPUs in native_machine_halt Impact: really halt all CPUs on halt Function machine_halt (resp. native_machine_halt) is empty for x86 architectures. When command 'halt -f' is invoked, the message "System halted." is displayed but this is not really true because all CPUs are still running. There are also similar inconsistencies for other arches (some uses power-off for halt or forever-loop with IRQs enabled/disabled). IMO there should be used the same approach for all architectures OR what does the message "System halted" really mean? This patch fixes it for x86. Signed-off-by: Ivan Vecera Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 16 ++++++++++++++++ arch/x86/kernel/reboot.c | 5 +++++ arch/x86/kernel/smp.c | 13 ------------- 3 files changed, 21 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c622772744d8..a4da7c4b3129 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -8,6 +8,7 @@ #include #include #include +#include unsigned long idle_halt; EXPORT_SYMBOL(idle_halt); @@ -122,6 +123,21 @@ void default_idle(void) EXPORT_SYMBOL(default_idle); #endif +void stop_this_cpu(void *dummy) +{ + local_irq_disable(); + /* + * Remove this CPU: + */ + cpu_clear(smp_processor_id(), cpu_online_map); + disable_local_APIC(); + + for (;;) { + if (hlt_works(smp_processor_id())) + halt(); + } +} + static void do_nothing(void *unused) { } diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 724adfc63cb9..34f8d37ae3c5 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -461,6 +461,11 @@ static void native_machine_restart(char *__unused) static void native_machine_halt(void) { + /* stop other cpus and apics */ + machine_shutdown(); + + /* stop this cpu */ + stop_this_cpu(NULL); } static void native_machine_power_off(void) diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c index 18f9b19f5f8f..3f92b134ab90 100644 --- a/arch/x86/kernel/smp.c +++ b/arch/x86/kernel/smp.c @@ -140,19 +140,6 @@ void native_send_call_func_ipi(cpumask_t mask) send_IPI_mask(mask, CALL_FUNCTION_VECTOR); } -static void stop_this_cpu(void *dummy) -{ - local_irq_disable(); - /* - * Remove this CPU: - */ - cpu_clear(smp_processor_id(), cpu_online_map); - disable_local_APIC(); - if (hlt_works(smp_processor_id())) - for (;;) halt(); - for (;;); -} - /* * this function calls the 'stop' function on all other CPUs in the system. */ -- cgit v1.2.3 From a98f8fd24fb24fcb9a359553e64dd6aac5cf4279 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 6 Nov 2008 01:13:39 +0100 Subject: x86: apic reset counter on shutdown Impact: avoid spurious lapic timer events on shutdown The apic timer might be close to firing when it is shutdown. We can not really disable the timer - we just mask the interrupt. That way we can get an extra interrupt when it is reenabled. Set the counter to max on shutdown to avoid this. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 70879c9e3936..1d410ee4b064 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -441,6 +441,7 @@ static void lapic_timer_setup(enum clock_event_mode mode, v = apic_read(APIC_LVTT); v |= (APIC_LVT_MASKED | LOCAL_TIMER_VECTOR); apic_write(APIC_LVTT, v); + apic_write(APIC_TMICT, 0xffffffff); break; case CLOCK_EVT_MODE_RESUME: /* Nothing to do here */ -- cgit v1.2.3 From a29a2af378f3f6362b68e126e2541c8bde885ead Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Wed, 29 Oct 2008 14:13:39 -0700 Subject: x86: KVM guest: fix section mismatch warning in kvmclock.c WARNING: arch/x86/kernel/built-in.o(.text+0x1722c): Section mismatch in reference from the function kvm_setup_secondary_clock() to the function .devinit.text:setup_secondary_APIC_clock() The function kvm_setup_secondary_clock() references the function __devinit setup_secondary_APIC_clock(). This is often because kvm_setup_secondary_clock lacks a __devinit annotation or the annotation of setup_secondary_APIC_clock is wrong. Signed-off-by: Md.Rakib H. Mullick Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Avi Kivity --- arch/x86/kernel/kvmclock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c index 774ac4991568..1c9cc431ea4f 100644 --- a/arch/x86/kernel/kvmclock.c +++ b/arch/x86/kernel/kvmclock.c @@ -128,7 +128,7 @@ static int kvm_register_clock(char *txt) } #ifdef CONFIG_X86_LOCAL_APIC -static void kvm_setup_secondary_clock(void) +static void __devinit kvm_setup_secondary_clock(void) { /* * Now that the first cpu already had this clocksource initialized, -- cgit v1.2.3 From 4687518c4cb7807fbeff21770e309080f9eb7f2f Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Tue, 11 Nov 2008 13:03:07 -0800 Subject: x86: 32 bit: interrupt stub consistency with 64 bit Don't generate interrupt stubs for interrupt vectors below FIRST_EXTERNAL_VECTOR, and make the table of interrupt vectors (interrupt[]) __initconst. Both of these changes both conserve memory and improve consistency with 64 bits. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/entry_32.S | 6 +++--- arch/x86/kernel/irqinit_32.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 28b597ef9ca1..4aea95652cff 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -622,16 +622,16 @@ END(syscall_badsys) * Build the entry stubs and pointer table with * some assembler magic. */ -.section .rodata,"a" +.section .init.rodata,"a" ENTRY(interrupt) .text ENTRY(irq_entries_start) RING0_INT_FRAME -vector=0 +vector=FIRST_EXTERNAL_VECTOR .rept NR_VECTORS ALIGN - .if vector + .if vector != FIRST_EXTERNAL_VECTOR CFI_ADJUST_CFA_OFFSET -4 .endif 1: pushl $~(vector) diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 845aa9803e80..607db63044a5 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -129,7 +129,7 @@ void __init native_init_IRQ(void) for (i = FIRST_EXTERNAL_VECTOR; i < NR_VECTORS; i++) { /* SYSCALL_VECTOR was reserved in trap_init. */ if (i != SYSCALL_VECTOR) - set_intr_gate(i, interrupt[i]); + set_intr_gate(i, interrupt[i-FIRST_EXTERNAL_VECTOR]); } -- cgit v1.2.3 From b7c6244f13d37592003b46e12500a90e9781ad9d Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Tue, 11 Nov 2008 13:24:58 -0800 Subject: x86: 32 bits: shrink and align IRQ stubs Shrink the IRQ stubs on 32 bits down to just over four bytes per (we fit seven into a 32-byte chunk.) This shrinks the total icache consumption of the IRQ stubs down to an even kilobyte, if all of them are in active use. The downside is that we end up with a double jump, which could have a negative effect on some pipelines. The double jump is always inside the same cacheline on any modern chips (the exception being 486/Elan/Geode which have only 16-byte cachelines, but are unlikely to have too many interrupt sources.) To get the most effect, cache-align the IRQ stubs. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/entry_32.S | 34 ++++++++++++++++++++++------------ 1 file changed, 22 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 4aea95652cff..dae81b9fd451 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -619,28 +619,37 @@ END(syscall_badsys) 27:; /* - * Build the entry stubs and pointer table with - * some assembler magic. + * Build the entry stubs and pointer table with some assembler magic. + * We pack 7 stubs into a single 32-byte chunk, which will fit in a + * single cache line on all modern x86 implementations. */ .section .init.rodata,"a" ENTRY(interrupt) .text - + .p2align 5 + .p2align CONFIG_X86_L1_CACHE_SHIFT ENTRY(irq_entries_start) RING0_INT_FRAME vector=FIRST_EXTERNAL_VECTOR -.rept NR_VECTORS - ALIGN - .if vector != FIRST_EXTERNAL_VECTOR +.rept (NR_VECTORS-FIRST_EXTERNAL_VECTOR+6)/7 + .balign 32 + .rept 7 + .if vector < NR_VECTORS + .if vector != FIRST_EXTERNAL_VECTOR CFI_ADJUST_CFA_OFFSET -4 - .endif -1: pushl $~(vector) + .endif +1: pushl $(~vector+0x80) /* Note: always in signed byte range */ CFI_ADJUST_CFA_OFFSET 4 - jmp common_interrupt - .previous + .if ((vector-FIRST_EXTERNAL_VECTOR)%7) != 6 + jmp 2f + .endif + .previous .long 1b - .text + .text vector=vector+1 + .endif + .endr +2: jmp common_interrupt .endr END(irq_entries_start) @@ -652,8 +661,9 @@ END(interrupt) * the CPU automatically disables interrupts when executing an IRQ vector, * so IRQ-flags tracing has to follow that: */ - ALIGN + .p2align CONFIG_X86_L1_CACHE_SHIFT common_interrupt: + addl $-0x80,(%esp) /* Adjust vector into the [-256,-1] range */ SAVE_ALL TRACE_IRQS_OFF movl %esp,%eax -- cgit v1.2.3 From 939b787130bf22887a09d8fd2641a094dcef8c22 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Tue, 11 Nov 2008 13:51:52 -0800 Subject: x86: 64 bits: shrink and align IRQ stubs Move the IRQ stub generation to assembly to simplify it and for consistency with 32 bits. Doing it in a C file with asm() statements doesn't help clarity, and it prevents some optimizations. Shrink the IRQ stubs down to just over four bytes per (we fit seven into a 32-byte chunk.) This shrinks the total icache consumption of the IRQ stubs down to an even kilobyte, if all of them are in active use. The downside is that we end up with a double jump, which could have a negative effect on some pipelines. The double jump is always inside the same cacheline on any modern chips. To get the most effect, cache-align the IRQ stubs. This makes the 64-bit code match changes already done to the 32-bit code, and should open up irqinit*.c for unification. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/entry_64.S | 48 ++++++++++++++++++++++++++++++-- arch/x86/kernel/irqinit_64.c | 66 -------------------------------------------- 2 files changed, 45 insertions(+), 69 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index b86f332c96a6..9b2aeaac9a6b 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -627,6 +627,46 @@ END(stub_rt_sigreturn) vector already pushed) */ #define XCPT_FRAME _frame ORIG_RAX +/* + * Build the entry stubs and pointer table with some assembler magic. + * We pack 7 stubs into a single 32-byte chunk, which will fit in a + * single cache line on all modern x86 implementations. + */ + .section .init.rodata,"a" +ENTRY(interrupt) + .text + .p2align 5 + .p2align CONFIG_X86_L1_CACHE_SHIFT +ENTRY(irq_entries_start) + INTR_FRAME +vector=FIRST_EXTERNAL_VECTOR +.rept (NR_VECTORS-FIRST_EXTERNAL_VECTOR+6)/7 + .balign 32 + .rept 7 + .if vector < NR_VECTORS + .if vector != FIRST_EXTERNAL_VECTOR + CFI_ADJUST_CFA_OFFSET -8 + .endif +1: pushq $(~vector+0x80) /* Note: always in signed byte range */ + CFI_ADJUST_CFA_OFFSET 8 + .if ((vector-FIRST_EXTERNAL_VECTOR)%7) != 6 + jmp 2f + .endif + .previous + .quad 1b + .text +vector=vector+1 + .endif + .endr +2: jmp common_interrupt +.endr + CFI_ENDPROC +END(irq_entries_start) + +.previous +END(interrupt) +.previous + /* * Interrupt entry/exit. * @@ -635,11 +675,12 @@ END(stub_rt_sigreturn) * Entry runs with interrupts off. */ -/* 0(%rsp): interrupt number */ +/* 0(%rsp): ~(interrupt number)+0x80 */ .macro interrupt func + addq $-0x80,(%rsp) /* Adjust vector to [-256,-1] range */ cld SAVE_ARGS - leaq -ARGOFFSET(%rsp),%rdi # arg1 for handler + leaq -ARGOFFSET(%rsp),%rdi /* arg1 for handler */ pushq %rbp /* * Save rbp twice: One is for marking the stack frame, as usual, and the @@ -670,7 +711,8 @@ END(stub_rt_sigreturn) call \func .endm -ENTRY(common_interrupt) + .p2align CONFIG_X86_L1_CACHE_SHIFT +common_interrupt: XCPT_FRAME interrupt do_IRQ /* 0(%rsp): oldrsp-ARGOFFSET */ diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index ff0235391285..8670b3ce626e 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -23,41 +23,6 @@ #include #include -/* - * Common place to define all x86 IRQ vectors - * - * This builds up the IRQ handler stubs using some ugly macros in irq.h - * - * These macros create the low-level assembly IRQ routines that save - * register context and call do_IRQ(). do_IRQ() then does all the - * operations that are needed to keep the AT (or SMP IOAPIC) - * interrupt-controller happy. - */ - -#define IRQ_NAME2(nr) nr##_interrupt(void) -#define IRQ_NAME(nr) IRQ_NAME2(IRQ##nr) - -/* - * SMP has a few special interrupts for IPI messages - */ - -#define BUILD_IRQ(nr) \ - asmlinkage void IRQ_NAME(nr); \ - asm("\n.text\n.p2align\n" \ - "IRQ" #nr "_interrupt:\n\t" \ - "push $~(" #nr ") ; " \ - "jmp common_interrupt\n" \ - ".previous"); - -#define BI(x,y) \ - BUILD_IRQ(x##y) - -#define BUILD_16_IRQS(x) \ - BI(x,0) BI(x,1) BI(x,2) BI(x,3) \ - BI(x,4) BI(x,5) BI(x,6) BI(x,7) \ - BI(x,8) BI(x,9) BI(x,a) BI(x,b) \ - BI(x,c) BI(x,d) BI(x,e) BI(x,f) - /* * ISA PIC or low IO-APIC triggered (INTA-cycle or APIC) interrupts: * (these are usually mapped to vectors 0x30-0x3f) @@ -73,37 +38,6 @@ * * (these are usually mapped into the 0x30-0xff vector range) */ - BUILD_16_IRQS(0x2) BUILD_16_IRQS(0x3) -BUILD_16_IRQS(0x4) BUILD_16_IRQS(0x5) BUILD_16_IRQS(0x6) BUILD_16_IRQS(0x7) -BUILD_16_IRQS(0x8) BUILD_16_IRQS(0x9) BUILD_16_IRQS(0xa) BUILD_16_IRQS(0xb) -BUILD_16_IRQS(0xc) BUILD_16_IRQS(0xd) BUILD_16_IRQS(0xe) BUILD_16_IRQS(0xf) - -#undef BUILD_16_IRQS -#undef BI - - -#define IRQ(x,y) \ - IRQ##x##y##_interrupt - -#define IRQLIST_16(x) \ - IRQ(x,0), IRQ(x,1), IRQ(x,2), IRQ(x,3), \ - IRQ(x,4), IRQ(x,5), IRQ(x,6), IRQ(x,7), \ - IRQ(x,8), IRQ(x,9), IRQ(x,a), IRQ(x,b), \ - IRQ(x,c), IRQ(x,d), IRQ(x,e), IRQ(x,f) - -/* for the irq vectors */ -static void (*__initdata interrupt[NR_VECTORS - FIRST_EXTERNAL_VECTOR])(void) = { - IRQLIST_16(0x2), IRQLIST_16(0x3), - IRQLIST_16(0x4), IRQLIST_16(0x5), IRQLIST_16(0x6), IRQLIST_16(0x7), - IRQLIST_16(0x8), IRQLIST_16(0x9), IRQLIST_16(0xa), IRQLIST_16(0xb), - IRQLIST_16(0xc), IRQLIST_16(0xd), IRQLIST_16(0xe), IRQLIST_16(0xf) -}; - -#undef IRQ -#undef IRQLIST_16 - - - /* * IRQ2 is cascade interrupt to second interrupt controller -- cgit v1.2.3 From 14d7ca5c575853664d8fe4f225a77b8df1b7de7d Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Tue, 11 Nov 2008 16:19:48 -0800 Subject: x86: attempt reboot via port CF9 if we have standard PCI ports Impact: Changes reboot behavior. If port CF9 seems to be safe to touch, attempt it before trying the keyboard controller. Port CF9 is not available on all chipsets (a significant but decreasing number of modern chipsets don't implement it), but port CF9 itself should in general be safe to poke (no ill effects if unimplemented) on any system which has PCI Configuration Method #1 or #2, as it falls inside the PCI configuration port range in both cases. No chipset without PCI is known to have port CF9, either, although an explicit "pci=bios" would mean we miss this and therefore don't use port CF9. An explicit "reboot=pci" can be used to force the use of port CF9. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/reboot.c | 34 +++++++++++++++++++++++++++------- 1 file changed, 27 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 34f8d37ae3c5..ddc93891cdcd 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -29,14 +29,17 @@ EXPORT_SYMBOL(pm_power_off); static const struct desc_ptr no_idt = {}; static int reboot_mode; -enum reboot_type reboot_type = BOOT_KBD; +enum reboot_type reboot_type = BOOT_CF9_COND; int reboot_force; #if defined(CONFIG_X86_32) && defined(CONFIG_SMP) static int reboot_cpu = -1; #endif -/* reboot=b[ios] | s[mp] | t[riple] | k[bd] | e[fi] [, [w]arm | [c]old] +/* This is set by the PCI code if either type 1 or type 2 PCI is detected */ +bool port_cf9_safe = false; + +/* reboot=b[ios] | s[mp] | t[riple] | k[bd] | e[fi] [, [w]arm | [c]old] | p[ci] warm Don't set the cold reboot flag cold Set the cold reboot flag bios Reboot by jumping through the BIOS (only for X86_32) @@ -45,6 +48,7 @@ static int reboot_cpu = -1; kbd Use the keyboard controller. cold reset (default) acpi Use the RESET_REG in the FADT efi Use efi reset_system runtime service + pci Use the so-called "PCI reset register", CF9 force Avoid anything that could hang. */ static int __init reboot_setup(char *str) @@ -79,6 +83,7 @@ static int __init reboot_setup(char *str) case 'k': case 't': case 'e': + case 'p': reboot_type = *str; break; @@ -379,28 +384,43 @@ static void native_machine_emergency_restart(void) load_idt(&no_idt); __asm__ __volatile__("int3"); - reboot_type = BOOT_KBD; + reboot_type = BOOT_CF9_COND; break; #ifdef CONFIG_X86_32 case BOOT_BIOS: machine_real_restart(jump_to_bios, sizeof(jump_to_bios)); - reboot_type = BOOT_KBD; + reboot_type = BOOT_CF9_COND; break; #endif case BOOT_ACPI: acpi_reboot(); - reboot_type = BOOT_KBD; + reboot_type = BOOT_CF9_COND; break; - case BOOT_EFI: if (efi_enabled) - efi.reset_system(reboot_mode ? EFI_RESET_WARM : EFI_RESET_COLD, + efi.reset_system(reboot_mode ? + EFI_RESET_WARM : + EFI_RESET_COLD, EFI_SUCCESS, 0, NULL); + reboot_type = BOOT_CF9_COND; + break; + + case BOOT_CF9: + port_cf9_safe = true; + /* fall through */ + case BOOT_CF9_COND: + if (port_cf9_safe) { + u8 cf9 = inb(0xcf9) & ~6; + outb(cf9|2, 0xcf9); /* Request hard reset */ + udelay(50); + outb(cf9|6, 0xcf9); /* Actually do the reset */ + udelay(50); + } reboot_type = BOOT_KBD; break; } -- cgit v1.2.3 From 32836259ff25ce97010569706cd33ba94de81d62 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Wed, 5 Nov 2008 16:17:52 -0700 Subject: ACPI: pci_link: remove acpi_irq_balance_set() interface This removes the acpi_irq_balance_set() interface from the PCI interrupt link driver. x86 used acpi_irq_balance_set() to tell the PCI interrupt link driver to configure links to minimize IRQ sharing. But the link driver can easily figure out whether to turn on IRQ balancing based on the IRQ model (PIC/IOAPIC/etc), so we can get rid of that external interface. It's better for the driver to figure this out at init-time. If we set it externally via the x86 code, the interface reduces modularity, and we depend on the fact that acpi_process_madt() happens before we process the kernel command line. Signed-off-by: Bjorn Helgaas Signed-off-by: Len Brown --- arch/x86/kernel/acpi/boot.c | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 8c1f76abae9e..4c51a2f8fd31 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -1343,7 +1343,6 @@ static void __init acpi_process_madt(void) error = acpi_parse_madt_ioapic_entries(); if (!error) { acpi_irq_model = ACPI_IRQ_MODEL_IOAPIC; - acpi_irq_balance_set(NULL); acpi_ioapic = 1; smp_found_config = 1; -- cgit v1.2.3 From 1f0d69a9fc815db82f15722bf05227190b1d714d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 12 Nov 2008 00:14:39 -0500 Subject: tracing: profile likely and unlikely annotations Impact: new unlikely/likely profiler Andrew Morton recently suggested having an in-kernel way to profile likely and unlikely macros. This patch achieves that goal. When configured, every(*) likely and unlikely macro gets a counter attached to it. When the condition is hit, the hit and misses of that condition are recorded. These numbers can later be retrieved by: /debugfs/tracing/profile_likely - All likely markers /debugfs/tracing/profile_unlikely - All unlikely markers. # cat /debug/tracing/profile_unlikely | head correct incorrect % Function File Line ------- --------- - -------- ---- ---- 2167 0 0 do_arch_prctl process_64.c 832 0 0 0 do_arch_prctl process_64.c 804 2670 0 0 IS_ERR err.h 34 71230 5693 7 __switch_to process_64.c 673 76919 0 0 __switch_to process_64.c 639 43184 33743 43 __switch_to process_64.c 624 12740 64181 83 __switch_to process_64.c 594 12740 64174 83 __switch_to process_64.c 590 # cat /debug/tracing/profile_unlikely | \ awk '{ if ($3 > 25) print $0; }' |head -20 44963 35259 43 __switch_to process_64.c 624 12762 67454 84 __switch_to process_64.c 594 12762 67447 84 __switch_to process_64.c 590 1478 595 28 syscall_get_error syscall.h 51 0 2821 100 syscall_trace_leave ptrace.c 1567 0 1 100 native_smp_prepare_cpus smpboot.c 1237 86338 265881 75 calc_delta_fair sched_fair.c 408 210410 108540 34 calc_delta_mine sched.c 1267 0 54550 100 sched_info_queued sched_stats.h 222 51899 66435 56 pick_next_task_fair sched_fair.c 1422 6 10 62 yield_task_fair sched_fair.c 982 7325 2692 26 rt_policy sched.c 144 0 1270 100 pre_schedule_rt sched_rt.c 1261 1268 48073 97 pick_next_task_rt sched_rt.c 884 0 45181 100 sched_info_dequeued sched_stats.h 177 0 15 100 sched_move_task sched.c 8700 0 15 100 sched_move_task sched.c 8690 53167 33217 38 schedule sched.c 4457 0 80208 100 sched_info_switch sched_stats.h 270 30585 49631 61 context_switch sched.c 2619 # cat /debug/tracing/profile_likely | awk '{ if ($3 > 25) print $0; }' 39900 36577 47 pick_next_task sched.c 4397 20824 15233 42 switch_mm mmu_context_64.h 18 0 7 100 __cancel_work_timer workqueue.c 560 617 66484 99 clocksource_adjust timekeeping.c 456 0 346340 100 audit_syscall_exit auditsc.c 1570 38 347350 99 audit_get_context auditsc.c 732 0 345244 100 audit_syscall_entry auditsc.c 1541 38 1017 96 audit_free auditsc.c 1446 0 1090 100 audit_alloc auditsc.c 862 2618 1090 29 audit_alloc auditsc.c 858 0 6 100 move_masked_irq migration.c 9 1 198 99 probe_sched_wakeup trace_sched_switch.c 58 2 2 50 probe_wakeup trace_sched_wakeup.c 227 0 2 100 probe_wakeup_sched_switch trace_sched_wakeup.c 144 4514 2090 31 __grab_cache_page filemap.c 2149 12882 228786 94 mapping_unevictable pagemap.h 50 4 11 73 __flush_cpu_slab slub.c 1466 627757 330451 34 slab_free slub.c 1731 2959 61245 95 dentry_lru_del_init dcache.c 153 946 1217 56 load_elf_binary binfmt_elf.c 904 102 82 44 disk_put_part genhd.h 206 1 1 50 dst_gc_task dst.c 82 0 19 100 tcp_mss_split_point tcp_output.c 1126 As you can see by the above, there's a bit of work to do in rethinking the use of some unlikelys and likelys. Note: the unlikely case had 71 hits that were more than 25%. Note: After submitting my first version of this patch, Andrew Morton showed me a version written by Daniel Walker, where I picked up the following ideas from: 1) Using __builtin_constant_p to avoid profiling fixed values. 2) Using __FILE__ instead of instruction pointers. 3) Using the preprocessor to stop all profiling of likely annotations from vsyscall_64.c. Thanks to Andrew Morton, Arjan van de Ven, Theodore Tso and Ingo Molnar for their feed back on this patch. (*) Not ever unlikely is recorded, those that are used by vsyscalls (a few of them) had to have profiling disabled. Signed-off-by: Steven Rostedt Cc: Andrew Morton Cc: Frederic Weisbecker Cc: Theodore Tso Cc: Arjan van de Ven Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/vsyscall_64.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index 0b8b6690a86d..2f90202e59b3 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -17,6 +17,14 @@ * want per guest time just set the kernel.vsyscall64 sysctl to 0. */ +/* Protect userspace from profiling */ +#ifdef CONFIG_TRACE_UNLIKELY_PROFILE +# undef likely +# undef unlikely +# define likely(x) likely_notrace(x) +# define unlikely(x) unlikely_notrace(x) +#endif + #include #include #include -- cgit v1.2.3 From 4a61204856e8b28e9f5489a7875cb3a60afd1e43 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Tue, 11 Nov 2008 19:09:29 -0800 Subject: x86: signal_32: introduce retcode and rt_retcode Impact: cleanup Introduce retcode and rt_retcode to replace setting up frame->retcode. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 30 ++++++++++++++++++++++++------ 1 file changed, 24 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 27a5c8174322..514171ac0d03 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -45,6 +45,28 @@ # define FIX_EFLAGS __FIX_EFLAGS #endif +static const struct { + u16 poplmovl; + u32 val; + u16 int80; +} __attribute__((packed)) retcode = { + 0xb858, /* popl %eax; movl $..., %eax */ + __NR_sigreturn, + 0x80cd, /* int $0x80 */ +}; + +static const struct { + u8 movl; + u32 val; + u16 int80; + u8 pad; +} __attribute__((packed)) rt_retcode = { + 0xb8, /* movl $..., %eax */ + __NR_rt_sigreturn, + 0x80cd, /* int $0x80 */ + 0 +}; + /* * Atomically swap in the new signal mask, and wait for a signal. */ @@ -427,9 +449,7 @@ __setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, * reasons and because gdb uses it as a signature to notice * signal handler stack frames. */ - err |= __put_user(0xb858, (short __user *)(frame->retcode+0)); - err |= __put_user(__NR_sigreturn, (int __user *)(frame->retcode+2)); - err |= __put_user(0x80cd, (short __user *)(frame->retcode+6)); + err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode); if (err) return -EFAULT; @@ -498,9 +518,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, * reasons and because gdb uses it as a signature to notice * signal handler stack frames. */ - err |= __put_user(0xb8, (char __user *)(frame->retcode+0)); - err |= __put_user(__NR_rt_sigreturn, (int __user *)(frame->retcode+1)); - err |= __put_user(0x80cd, (short __user *)(frame->retcode+5)); + err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode); if (err) return -EFAULT; -- cgit v1.2.3 From 2b7d0390a6d6d595f43ea3806639664afe5b9ebe Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 12 Nov 2008 13:17:38 +0100 Subject: tracing: branch tracer, fix vdso crash Impact: fix bootup crash the branch tracer missed arch/x86/vdso/vclock_gettime.c from disabling tracing, which caused such bootup crashes: [ 201.840097] init[1]: segfault at 7fffed3fe7c0 ip 00007fffed3fea2e sp 000077 also clean up the ugly ifdefs in arch/x86/kernel/vsyscall_64.c by creating DISABLE_UNLIKELY_PROFILE facility for code to turn off instrumentation on a per file basis. Signed-off-by: Ingo Molnar --- arch/x86/kernel/vsyscall_64.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index 2f90202e59b3..ece02932ea57 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -17,13 +17,8 @@ * want per guest time just set the kernel.vsyscall64 sysctl to 0. */ -/* Protect userspace from profiling */ -#ifdef CONFIG_TRACE_UNLIKELY_PROFILE -# undef likely -# undef unlikely -# define likely(x) likely_notrace(x) -# define unlikely(x) unlikely_notrace(x) -#endif +/* Disable profiling for userspace code: */ +#define DISABLE_UNLIKELY_PROFILE #include #include -- cgit v1.2.3 From a7d41820f683c35b53af719210a51f6aa0f86a6a Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:37 -0200 Subject: x86 kdump: extract kdump-specific code from crash_nmi_callback() Impact: cleanup The NMI CPU-halting code will be used on non-kdump cases, also (e.g. emergency_reboot when virtualization is enabled). Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 31 ++++++++++++++++++++----------- 1 file changed, 20 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index 268553817909..60475422a51a 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -35,19 +35,34 @@ static int crashing_cpu; #if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) static atomic_t waiting_for_crash_ipi; -static int crash_nmi_callback(struct notifier_block *self, - unsigned long val, void *data) +static void kdump_nmi_callback(int cpu, struct die_args *args) { struct pt_regs *regs; #ifdef CONFIG_X86_32 struct pt_regs fixed_regs; #endif + + regs = args->regs; + +#ifdef CONFIG_X86_32 + if (!user_mode_vm(regs)) { + crash_fixup_ss_esp(&fixed_regs, regs); + regs = &fixed_regs; + } +#endif + crash_save_cpu(regs, cpu); + + disable_local_APIC(); +} + +static int crash_nmi_callback(struct notifier_block *self, + unsigned long val, void *data) +{ int cpu; if (val != DIE_NMI_IPI) return NOTIFY_OK; - regs = ((struct die_args *)data)->regs; cpu = raw_smp_processor_id(); /* Don't do anything if this handler is invoked on crashing cpu. @@ -58,14 +73,8 @@ static int crash_nmi_callback(struct notifier_block *self, return NOTIFY_STOP; local_irq_disable(); -#ifdef CONFIG_X86_32 - if (!user_mode_vm(regs)) { - crash_fixup_ss_esp(&fixed_regs, regs); - regs = &fixed_regs; - } -#endif - crash_save_cpu(regs, cpu); - disable_local_APIC(); + kdump_nmi_callback(cpu, (struct die_args *)data); + atomic_dec(&waiting_for_crash_ipi); /* Assume hlt works */ halt(); -- cgit v1.2.3 From b2bbe71b829564fb65a6bc7e1e25e02d70cffce8 Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:38 -0200 Subject: x86 kdump: move crashing_cpu assignment to nmi_shootdown_cpus() Impact: cleanup This variable will be moved to non-kdump-specific code. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index 60475422a51a..ed2f0f9dc894 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -29,10 +29,11 @@ #include +#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) + /* This keeps a track of which one is crashing cpu. */ static int crashing_cpu; -#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) static atomic_t waiting_for_crash_ipi; static void kdump_nmi_callback(int cpu, struct die_args *args) @@ -100,6 +101,9 @@ static void nmi_shootdown_cpus(void) { unsigned long msecs; + /* Make a note of crashing cpu. Will be used in NMI callback.*/ + crashing_cpu = safe_smp_processor_id(); + atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); /* Would it be better to replace the trap vector here? */ if (register_die_notifier(&crash_nmi_nb)) @@ -140,8 +144,6 @@ void native_machine_crash_shutdown(struct pt_regs *regs) /* The kernel is broken so disable interrupts */ local_irq_disable(); - /* Make a note of crashing cpu. Will be used in NMI callback.*/ - crashing_cpu = safe_smp_processor_id(); nmi_shootdown_cpus(); lapic_shutdown(); #if defined(CONFIG_X86_IO_APIC) -- cgit v1.2.3 From d1e7b91cfaa8fc5ed736dcfb8beb5134a2385228 Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:39 -0200 Subject: x86 kdump: create kdump_nmi_shootdown_cpus() Impact: cleanup For the kdump-specific code that was living on nmi_shootdown_cpus(). Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index ed2f0f9dc894..75c468cc7e59 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -122,10 +122,17 @@ static void nmi_shootdown_cpus(void) } /* Leave the nmi callback set */ +} + +static void kdump_nmi_shootdown_cpus(void) +{ + nmi_shootdown_cpus(); + disable_local_APIC(); } + #else -static void nmi_shootdown_cpus(void) +static void kdump_nmi_shootdown_cpus(void) { /* There are no cpus to shootdown */ } @@ -144,7 +151,7 @@ void native_machine_crash_shutdown(struct pt_regs *regs) /* The kernel is broken so disable interrupts */ local_irq_disable(); - nmi_shootdown_cpus(); + kdump_nmi_shootdown_cpus(); lapic_shutdown(); #if defined(CONFIG_X86_IO_APIC) disable_IO_APIC(); -- cgit v1.2.3 From 8e294786316aca41c66b8b73ba1ee74a4ae7d452 Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:40 -0200 Subject: x86 kdump: make kdump_nmi_callback() a function ptr on crash_nmi_callback() Impact: extend nmi_shootdown_cpus() with a callback The reboot code will use a different function on crash_nmi_callback(). Adding a function pointer parameter to nmi_shootdown_cpus() for that. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index 75c468cc7e59..f23c2beeb37d 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -29,10 +29,13 @@ #include +typedef void (*nmi_shootdown_cb)(int, struct die_args*); + #if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) /* This keeps a track of which one is crashing cpu. */ static int crashing_cpu; +static nmi_shootdown_cb shootdown_callback; static atomic_t waiting_for_crash_ipi; @@ -74,7 +77,7 @@ static int crash_nmi_callback(struct notifier_block *self, return NOTIFY_STOP; local_irq_disable(); - kdump_nmi_callback(cpu, (struct die_args *)data); + shootdown_callback(cpu, (struct die_args *)data); atomic_dec(&waiting_for_crash_ipi); /* Assume hlt works */ @@ -97,13 +100,15 @@ static struct notifier_block crash_nmi_nb = { .notifier_call = crash_nmi_callback, }; -static void nmi_shootdown_cpus(void) +static void nmi_shootdown_cpus(nmi_shootdown_cb callback) { unsigned long msecs; /* Make a note of crashing cpu. Will be used in NMI callback.*/ crashing_cpu = safe_smp_processor_id(); + shootdown_callback = callback; + atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); /* Would it be better to replace the trap vector here? */ if (register_die_notifier(&crash_nmi_nb)) @@ -126,7 +131,7 @@ static void nmi_shootdown_cpus(void) static void kdump_nmi_shootdown_cpus(void) { - nmi_shootdown_cpus(); + nmi_shootdown_cpus(kdump_nmi_callback); disable_local_APIC(); } -- cgit v1.2.3 From c370e5e089adfd5b1b863f3464cccae9ebf33cca Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:41 -0200 Subject: x86 kdump: make nmi_shootdown_cpus() non-static Impact: make API available to the rest of x86 platform code Add prototype to asm/reboot.h. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index f23c2beeb37d..fb298d1daac9 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -29,7 +29,6 @@ #include -typedef void (*nmi_shootdown_cb)(int, struct die_args*); #if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) @@ -100,7 +99,7 @@ static struct notifier_block crash_nmi_nb = { .notifier_call = crash_nmi_callback, }; -static void nmi_shootdown_cpus(nmi_shootdown_cb callback) +void nmi_shootdown_cpus(nmi_shootdown_cb callback) { unsigned long msecs; -- cgit v1.2.3 From 2ddded213895e41b9cfe1c084127e6c01632ac1a Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:42 -0200 Subject: x86: move nmi_shootdown_cpus() to reboot.c Impact: make nmi_shootdown_cpus() available to the rest of the x86 platform Now nmi_shootdown_cpus() is ready to be used by non-kdump code also. Move it to reboot.c. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/crash.c | 76 -------------------------------------------- arch/x86/kernel/reboot.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 83 insertions(+), 76 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index fb298d1daac9..d84a852e4cd7 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -32,12 +32,6 @@ #if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) -/* This keeps a track of which one is crashing cpu. */ -static int crashing_cpu; -static nmi_shootdown_cb shootdown_callback; - -static atomic_t waiting_for_crash_ipi; - static void kdump_nmi_callback(int cpu, struct die_args *args) { struct pt_regs *regs; @@ -58,76 +52,6 @@ static void kdump_nmi_callback(int cpu, struct die_args *args) disable_local_APIC(); } -static int crash_nmi_callback(struct notifier_block *self, - unsigned long val, void *data) -{ - int cpu; - - if (val != DIE_NMI_IPI) - return NOTIFY_OK; - - cpu = raw_smp_processor_id(); - - /* Don't do anything if this handler is invoked on crashing cpu. - * Otherwise, system will completely hang. Crashing cpu can get - * an NMI if system was initially booted with nmi_watchdog parameter. - */ - if (cpu == crashing_cpu) - return NOTIFY_STOP; - local_irq_disable(); - - shootdown_callback(cpu, (struct die_args *)data); - - atomic_dec(&waiting_for_crash_ipi); - /* Assume hlt works */ - halt(); - for (;;) - cpu_relax(); - - return 1; -} - -static void smp_send_nmi_allbutself(void) -{ - cpumask_t mask = cpu_online_map; - cpu_clear(safe_smp_processor_id(), mask); - if (!cpus_empty(mask)) - send_IPI_mask(mask, NMI_VECTOR); -} - -static struct notifier_block crash_nmi_nb = { - .notifier_call = crash_nmi_callback, -}; - -void nmi_shootdown_cpus(nmi_shootdown_cb callback) -{ - unsigned long msecs; - - /* Make a note of crashing cpu. Will be used in NMI callback.*/ - crashing_cpu = safe_smp_processor_id(); - - shootdown_callback = callback; - - atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); - /* Would it be better to replace the trap vector here? */ - if (register_die_notifier(&crash_nmi_nb)) - return; /* return what? */ - /* Ensure the new callback function is set before sending - * out the NMI - */ - wmb(); - - smp_send_nmi_allbutself(); - - msecs = 1000; /* Wait at most a second for the other cpus to stop */ - while ((atomic_read(&waiting_for_crash_ipi) > 0) && msecs) { - mdelay(1); - msecs--; - } - - /* Leave the nmi callback set */ -} - static void kdump_nmi_shootdown_cpus(void) { nmi_shootdown_cpus(kdump_nmi_callback); diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 724adfc63cb9..364edeecc235 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -21,6 +21,9 @@ # include #endif +#include + + /* * Power off function, if any */ @@ -514,3 +517,83 @@ void machine_crash_shutdown(struct pt_regs *regs) machine_ops.crash_shutdown(regs); } #endif + + +#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) + +/* This keeps a track of which one is crashing cpu. */ +static int crashing_cpu; +static nmi_shootdown_cb shootdown_callback; + +static atomic_t waiting_for_crash_ipi; + +static int crash_nmi_callback(struct notifier_block *self, + unsigned long val, void *data) +{ + int cpu; + + if (val != DIE_NMI_IPI) + return NOTIFY_OK; + + cpu = raw_smp_processor_id(); + + /* Don't do anything if this handler is invoked on crashing cpu. + * Otherwise, system will completely hang. Crashing cpu can get + * an NMI if system was initially booted with nmi_watchdog parameter. + */ + if (cpu == crashing_cpu) + return NOTIFY_STOP; + local_irq_disable(); + + shootdown_callback(cpu, (struct die_args *)data); + + atomic_dec(&waiting_for_crash_ipi); + /* Assume hlt works */ + halt(); + for (;;) + cpu_relax(); + + return 1; +} + +static void smp_send_nmi_allbutself(void) +{ + cpumask_t mask = cpu_online_map; + cpu_clear(safe_smp_processor_id(), mask); + if (!cpus_empty(mask)) + send_IPI_mask(mask, NMI_VECTOR); +} + +static struct notifier_block crash_nmi_nb = { + .notifier_call = crash_nmi_callback, +}; + +void nmi_shootdown_cpus(nmi_shootdown_cb callback) +{ + unsigned long msecs; + + /* Make a note of crashing cpu. Will be used in NMI callback.*/ + crashing_cpu = safe_smp_processor_id(); + + shootdown_callback = callback; + + atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); + /* Would it be better to replace the trap vector here? */ + if (register_die_notifier(&crash_nmi_nb)) + return; /* return what? */ + /* Ensure the new callback function is set before sending + * out the NMI + */ + wmb(); + + smp_send_nmi_allbutself(); + + msecs = 1000; /* Wait at most a second for the other cpus to stop */ + while ((atomic_read(&waiting_for_crash_ipi) > 0) && msecs) { + mdelay(1); + msecs--; + } + + /* Leave the nmi callback set */ +} +#endif -- cgit v1.2.3 From bb8dd270e62217e2d2172094c6c352c4ddc0a127 Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:43 -0200 Subject: x86: make nmi_shootdown_cpus() available on !SMP and !X86_LOCAL_APIC Impact: widen nmi_shootdown_cpus() availability The X86_LOCAL_APIC #ifdef was for kdump. For !SMP, the function simply does nothing. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/reboot.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 364edeecc235..17a41e055565 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -519,7 +519,7 @@ void machine_crash_shutdown(struct pt_regs *regs) #endif -#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) +#if defined(CONFIG_SMP) /* This keeps a track of which one is crashing cpu. */ static int crashing_cpu; @@ -568,6 +568,12 @@ static struct notifier_block crash_nmi_nb = { .notifier_call = crash_nmi_callback, }; +/* Halt all other CPUs, calling the specified function on each of them + * + * This function can be used to halt all other CPUs on crash + * or emergency reboot time. The function passed as parameter + * will be called inside a NMI handler on all CPUs. + */ void nmi_shootdown_cpus(nmi_shootdown_cb callback) { unsigned long msecs; @@ -596,4 +602,9 @@ void nmi_shootdown_cpus(nmi_shootdown_cb callback) /* Leave the nmi callback set */ } +#else /* !CONFIG_SMP */ +void nmi_shootdown_cpus(nmi_shootdown_cb callback) +{ + /* No other CPUs to shoot down */ +} #endif -- cgit v1.2.3 From c415b3dce30dfb41234e118662e8720f47343a4f Mon Sep 17 00:00:00 2001 From: Eduardo Habkost Date: Wed, 12 Nov 2008 11:34:44 -0200 Subject: x86: disable IRQs before doing anything on nmi_shootdown_cpus() Impact: make nmi_shootdown_cpus() callable from preemptible context We need to know on which CPU we are running on, and we don't want to be preempted while doing this. Signed-off-by: Eduardo Habkost Signed-off-by: Ingo Molnar --- arch/x86/kernel/reboot.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 17a41e055565..c3cd512484e5 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -577,6 +577,7 @@ static struct notifier_block crash_nmi_nb = { void nmi_shootdown_cpus(nmi_shootdown_cb callback) { unsigned long msecs; + local_irq_disable(); /* Make a note of crashing cpu. Will be used in NMI callback.*/ crashing_cpu = safe_smp_processor_id(); -- cgit v1.2.3 From 8665596ec05498525014436520b316ba174a068a Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Wed, 12 Nov 2008 10:27:35 -0800 Subject: x86: fix up the new IRQ code for older versions of gas Older versions of gas don't implement the C-style != operator, they instead want the Pascal-style <> operator. Change != to <> so we don't break compilation with those old versions of gas. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/entry_32.S | 4 ++-- arch/x86/kernel/entry_64.S | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index dae81b9fd451..bd02ec77edc4 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -635,12 +635,12 @@ vector=FIRST_EXTERNAL_VECTOR .balign 32 .rept 7 .if vector < NR_VECTORS - .if vector != FIRST_EXTERNAL_VECTOR + .if vector <> FIRST_EXTERNAL_VECTOR CFI_ADJUST_CFA_OFFSET -4 .endif 1: pushl $(~vector+0x80) /* Note: always in signed byte range */ CFI_ADJUST_CFA_OFFSET 4 - .if ((vector-FIRST_EXTERNAL_VECTOR)%7) != 6 + .if ((vector-FIRST_EXTERNAL_VECTOR)%7) <> 6 jmp 2f .endif .previous diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 9b2aeaac9a6b..2b42362a85b2 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -644,12 +644,12 @@ vector=FIRST_EXTERNAL_VECTOR .balign 32 .rept 7 .if vector < NR_VECTORS - .if vector != FIRST_EXTERNAL_VECTOR + .if vector <> FIRST_EXTERNAL_VECTOR CFI_ADJUST_CFA_OFFSET -8 .endif 1: pushq $(~vector+0x80) /* Note: always in signed byte range */ CFI_ADJUST_CFA_OFFSET 8 - .if ((vector-FIRST_EXTERNAL_VECTOR)%7) != 6 + .if ((vector-FIRST_EXTERNAL_VECTOR)%7) <> 6 jmp 2f .endif .previous -- cgit v1.2.3 From 8652cb4b0d87accbe78725fd2a13be2787059649 Mon Sep 17 00:00:00 2001 From: Prarit Bhargava Date: Wed, 12 Nov 2008 13:35:00 -0500 Subject: x86: warn of incorrect cpu_khz on AMD systems Impact: add debug check If none of the perfctrs are free when calculating cpu_khz we default to using ctr 3 (ie, we just choose 3). This may lead to an incorrect tsc freq value which can cause the system to be unstable. To aid in future debugging, WARN the user of a potential problem. Signed-off-by: Prarit Bhargava Signed-off-by: Ingo Molnar --- arch/x86/kernel/time_64.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/time_64.c b/arch/x86/kernel/time_64.c index cb19d650c216..418a095c5796 100644 --- a/arch/x86/kernel/time_64.c +++ b/arch/x86/kernel/time_64.c @@ -80,6 +80,8 @@ unsigned long __init calibrate_cpu(void) break; no_ctr_free = (i == 4); if (no_ctr_free) { + WARN(1, KERN_WARNING "Warning: AMD perfctrs busy ... " + "cpu_khz value may be incorrect.\n"); i = 3; rdmsrl(MSR_K7_EVNTSEL3, evntsel3); wrmsrl(MSR_K7_EVNTSEL3, 0); -- cgit v1.2.3 From 2ed84eeb8808cf3c9f039213ca137ffd7d753f0e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 12 Nov 2008 15:24:24 -0500 Subject: trace: rename unlikely profiler to branch profiler Impact: name change of unlikely tracer and profiler Ingo Molnar suggested changing the config from UNLIKELY_PROFILE to BRANCH_PROFILING. I never did like the "unlikely" name so I went one step farther, and renamed all the unlikely configurations to a "BRANCH" variant. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/vsyscall_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index ece02932ea57..6f3d3d4cd973 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -18,7 +18,7 @@ */ /* Disable profiling for userspace code: */ -#define DISABLE_UNLIKELY_PROFILE +#define DISABLE_BRANCH_PROFILING #include #include -- cgit v1.2.3 From 62d59d17a5f98edb48b171742dfa531488802f07 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Wed, 12 Nov 2008 22:47:54 +0100 Subject: tracing/function-return-tracer: make the function return tracer lockless Impact: remove spinlocks and irq disabling in function return tracer. I've tried to figure out all of the race condition that could happen when the tracer pushes or pops a return address trace to/from the current thread_info. Theory: _ One thread can only execute on one cpu at a time. So this code doesn't need to be SMP-safe. Just drop the spinlock. _ The only race could happen between the current thread and an interrupt. If an interrupt is raised, it will increase the index of the return stack storage and then execute until the end of the tracing to finally free the index it used. We don't need to disable irqs. This is theorical. In practice, I've tested it with a two-core SMP and had no problem at all. Perhaps -tip testing could confirm it. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 43 +++++-------------------------------------- 1 file changed, 5 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 16a571dea2ef..1db0e121a3e7 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -44,62 +44,37 @@ void ftrace_nmi_exit(void) atomic_dec(&in_nmi); } -/* - * Synchronize accesses to return adresses stack with - * interrupts. - */ -static raw_spinlock_t ret_stack_lock; - /* Add a function return address to the trace stack on thread info.*/ static int push_return_trace(unsigned long ret, unsigned long long time, unsigned long func) { int index; - struct thread_info *ti; - unsigned long flags; - int err = 0; - - raw_local_irq_save(flags); - __raw_spin_lock(&ret_stack_lock); + struct thread_info *ti = current_thread_info(); - ti = current_thread_info(); /* The return trace stack is full */ - if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { - err = -EBUSY; - goto out; - } + if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) + return -EBUSY; index = ++ti->curr_ret_stack; ti->ret_stack[index].ret = ret; ti->ret_stack[index].func = func; ti->ret_stack[index].calltime = time; -out: - __raw_spin_unlock(&ret_stack_lock); - raw_local_irq_restore(flags); - return err; + return 0; } /* Retrieve a function return address to the trace stack on thread info.*/ static void pop_return_trace(unsigned long *ret, unsigned long long *time, unsigned long *func) { - struct thread_info *ti; int index; - unsigned long flags; - - raw_local_irq_save(flags); - __raw_spin_lock(&ret_stack_lock); - ti = current_thread_info(); + struct thread_info *ti = current_thread_info(); index = ti->curr_ret_stack; *ret = ti->ret_stack[index].ret; *func = ti->ret_stack[index].func; *time = ti->ret_stack[index].calltime; ti->curr_ret_stack--; - - __raw_spin_unlock(&ret_stack_lock); - raw_local_irq_restore(flags); } /* @@ -175,14 +150,6 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) *parent = old; } -static int __init init_ftrace_function_return(void) -{ - ret_stack_lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; - return 0; -} -device_initcall(init_ftrace_function_return); - - #endif #ifdef CONFIG_DYNAMIC_FTRACE -- cgit v1.2.3 From 1dc1c6adf38bc5799d1594681645ced40ced4b6b Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Wed, 12 Nov 2008 22:49:23 +0100 Subject: tracing/function-return-tracer: call prepare_ftrace_return by registers Impact: Optimize a bit the function return tracer This patch changes the calling convention of prepare_ftrace_return to pass its arguments by register. This will optimize it a bit and prepare it to support dynamic tracing. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 5 +---- arch/x86/kernel/ftrace.c | 1 - 2 files changed, 1 insertion(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 9a0ac85946db..f97621149839 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1217,12 +1217,9 @@ trace_return: pushl %eax pushl %ecx pushl %edx - movl 0xc(%esp), %eax - pushl %eax + movl 0xc(%esp), %edx lea 0x4(%ebp), %eax - pushl %eax call prepare_ftrace_return - addl $8, %esp popl %edx popl %ecx popl %eax diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 1db0e121a3e7..fe832738e1e2 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -95,7 +95,6 @@ unsigned long ftrace_return_to_handler(void) * Hook the return address and push it in the stack of return addrs * in current thread info. */ -asmlinkage void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) { unsigned long old; -- cgit v1.2.3 From 722024dbb74f3ea316c285c0a71a4512e113b0c4 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Thu, 13 Nov 2008 13:50:20 +0100 Subject: x86: irq: fix apicinterrupts on 64 bits Impact: Fix interrupt via the apicinterrupt macro Checkin 939b787130bf22887a09d8fd2641a094dcef8c22 changed the "interrupt" macro, but the "interrupt" macro is also invoked indirectly from the "apicinterrupt" macro. The "apicinterrupt" macro probably should have its own collection of systematic stubs for the same reason the main IRQ code does; as is it is a huge amount of replicated code. Signed-off-by: Alexander van Heukelum Signed-off-by: H. Peter Anvin --- arch/x86/kernel/entry_64.S | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 2b42362a85b2..369de6973c58 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -675,9 +675,8 @@ END(interrupt) * Entry runs with interrupts off. */ -/* 0(%rsp): ~(interrupt number)+0x80 */ +/* 0(%rsp): ~(interrupt number) */ .macro interrupt func - addq $-0x80,(%rsp) /* Adjust vector to [-256,-1] range */ cld SAVE_ARGS leaq -ARGOFFSET(%rsp),%rdi /* arg1 for handler */ @@ -711,9 +710,14 @@ END(interrupt) call \func .endm + /* + * The interrupt stubs push (~vector+0x80) onto the stack and + * then jump to common_interrupt. + */ .p2align CONFIG_X86_L1_CACHE_SHIFT common_interrupt: XCPT_FRAME + addq $-0x80,(%rsp) /* Adjust vector to [-256,-1] range */ interrupt do_IRQ /* 0(%rsp): oldrsp-ARGOFFSET */ ret_from_intr: -- cgit v1.2.3 From 52168e60f7d86d83124903098ac8c2dba93cd1c4 Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Fri, 14 Nov 2008 13:47:31 +0000 Subject: Revert "x86: blacklist DMAR on Intel G31/G33 chipsets" This reverts commit e51af6630848406fc97adbd71443818cdcda297b, which was wrongly hoovered up and submitted about a month after a better fix had already been merged. The better fix is commit cbda1ba898647aeb4ee770b803c922f595e97731 ("PCI/iommu: blacklist DMAR on Intel G31/G33 chipsets"), where we do this blacklisting based on the DMI identification for the offending motherboard, since sometimes this chipset (or at least a chipset with the same PCI ID) apparently _does_ actually have an IOMMU. Signed-off-by: David Woodhouse Signed-off-by: Linus Torvalds --- arch/x86/kernel/early-quirks.c | 18 ------------------ 1 file changed, 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 3ce029ffaa55..1b894b72c0f5 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -188,20 +188,6 @@ static void __init ati_bugs_contd(int num, int slot, int func) } #endif -#ifdef CONFIG_DMAR -static void __init intel_g33_dmar(int num, int slot, int func) -{ - struct acpi_table_header *dmar_tbl; - acpi_status status; - - status = acpi_get_table(ACPI_SIG_DMAR, 0, &dmar_tbl); - if (ACPI_SUCCESS(status)) { - printk(KERN_INFO "BIOS BUG: DMAR advertised on Intel G31/G33 chipset -- ignoring\n"); - dmar_disabled = 1; - } -} -#endif - #define QFLAG_APPLY_ONCE 0x1 #define QFLAG_APPLIED 0x2 #define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED) @@ -225,10 +211,6 @@ static struct chipset early_qrk[] __initdata = { PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, -#ifdef CONFIG_DMAR - { PCI_VENDOR_ID_INTEL, 0x29c0, - PCI_CLASS_BRIDGE_HOST, PCI_ANY_ID, 0, intel_g33_dmar }, -#endif {} }; -- cgit v1.2.3 From 31e889098a80ceb3e9e3c555d522b2686a6663c6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 14 Nov 2008 16:21:19 -0800 Subject: ftrace: pass module struct to arch dynamic ftrace functions Impact: allow archs more flexibility on dynamic ftrace implementations Dynamic ftrace has largly been developed on x86. Since x86 does not have the same limitations as other architectures, the ftrace interaction between the generic code and the architecture specific code was not flexible enough to handle some of the issues that other architectures have. Most notably, module trampolines. Due to the limited branch distance that archs make in calling kernel core code from modules, the module load code must create a trampoline to jump to what will make the larger jump into core kernel code. The problem arises when this happens to a call to mcount. Ftrace checks all code before modifying it and makes sure the current code is what it expects. Right now, there is not enough information to handle modifying module trampolines. This patch changes the API between generic dynamic ftrace code and the arch dependent code. There is now two functions for modifying code: ftrace_make_nop(mod, rec, addr) - convert the code at rec->ip into a nop, where the original text is calling addr. (mod is the module struct if called by module init) ftrace_make_caller(rec, addr) - convert the code rec->ip that should be a nop into a caller to addr. The record "rec" now has a new field called "arch" where the architecture can add any special attributes to each call site record. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 29 ++++++++++++++++++++++++++--- 1 file changed, 26 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index fe832738e1e2..762222ad1387 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -166,7 +166,7 @@ static int ftrace_calc_offset(long ip, long addr) return (int)(addr - ip); } -unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) +static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -311,12 +311,12 @@ do_ftrace_mod_code(unsigned long ip, void *new_code) static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; -unsigned char *ftrace_nop_replace(void) +static unsigned char *ftrace_nop_replace(void) { return ftrace_nop; } -int +static int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { @@ -349,6 +349,29 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return 0; } +int ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_call_replace(ip, addr); + new = ftrace_nop_replace(); + + return ftrace_modify_code(rec->ip, old, new); +} + +int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_nop_replace(); + new = ftrace_call_replace(ip, addr); + + return ftrace_modify_code(rec->ip, old, new); +} + int ftrace_update_ftrace_func(ftrace_func_t func) { unsigned long ip = (unsigned long)(&ftrace_call); -- cgit v1.2.3 From b01c746617da5e260803eb10ed64ca043e9a1241 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 15 Nov 2008 02:37:44 +0100 Subject: tracing/function-return-tracer: add a barrier to ensure return stack index is incremented in memory Impact: fix possible race condition in ftrace function return tracer This fixes a possible race condition if index incrementation is not immediately flushed in memory. Thanks for Andi Kleen and Steven Rostedt for pointing out this issue and give me this solution. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 762222ad1387..d98b5a8ecf4c 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -56,6 +56,7 @@ static int push_return_trace(unsigned long ret, unsigned long long time, return -EBUSY; index = ++ti->curr_ret_stack; + barrier(); ti->ret_stack[index].ret = ret; ti->ret_stack[index].func = func; ti->ret_stack[index].calltime = time; -- cgit v1.2.3 From e7d3737ea1b102030f44e96c97754101e41515f0 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 16 Nov 2008 06:02:06 +0100 Subject: tracing/function-return-tracer: support for dynamic ftrace on function return tracer This patch adds the support for dynamic tracing on the function return tracer. The whole difference with normal dynamic function tracing is that we don't need to hook on a particular callback. The only pro that we want is to nop or set dynamically the calls to ftrace_caller (which is ftrace_return_caller here). Some security checks ensure that we are not trying to launch dynamic tracing for return tracing while normal function tracing is already running. An example of trace with getnstimeofday set as a filter: ktime_get_ts+0x22/0x50 -> getnstimeofday (2283 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1396 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1382 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1825 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1426 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1464 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1524 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1382 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1382 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1434 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1464 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1502 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1404 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1397 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1051 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1314 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1344 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1163 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1390 ns) ktime_get_ts+0x22/0x50 -> getnstimeofday (1374 ns) Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 18 ++-- arch/x86/kernel/ftrace.c | 258 +++++++++++++++++++++++---------------------- 2 files changed, 141 insertions(+), 135 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index f97621149839..74defe21ba42 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1190,7 +1190,7 @@ ENTRY(mcount) jnz trace #ifdef CONFIG_FUNCTION_RET_TRACER cmpl $ftrace_stub, ftrace_function_return - jnz trace_return + jnz ftrace_return_caller #endif .globl ftrace_stub ftrace_stub: @@ -1211,9 +1211,15 @@ trace: popl %ecx popl %eax jmp ftrace_stub +END(mcount) +#endif /* CONFIG_DYNAMIC_FTRACE */ +#endif /* CONFIG_FUNCTION_TRACER */ #ifdef CONFIG_FUNCTION_RET_TRACER -trace_return: +ENTRY(ftrace_return_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + pushl %eax pushl %ecx pushl %edx @@ -1223,7 +1229,8 @@ trace_return: popl %edx popl %ecx popl %eax - jmp ftrace_stub + ret +END(ftrace_return_caller) .globl return_to_handler return_to_handler: @@ -1237,10 +1244,7 @@ return_to_handler: popl %ecx popl %eax ret -#endif /* CONFIG_FUNCTION_RET_TRACER */ -END(mcount) -#endif /* CONFIG_DYNAMIC_FTRACE */ -#endif /* CONFIG_FUNCTION_TRACER */ +#endif .section .rodata,"a" #include "syscall_table_32.S" diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index d98b5a8ecf4c..924153edd973 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -24,134 +24,6 @@ #include - -#ifdef CONFIG_FUNCTION_RET_TRACER - -/* - * These functions are picked from those used on - * this page for dynamic ftrace. They have been - * simplified to ignore all traces in NMI context. - */ -static atomic_t in_nmi; - -void ftrace_nmi_enter(void) -{ - atomic_inc(&in_nmi); -} - -void ftrace_nmi_exit(void) -{ - atomic_dec(&in_nmi); -} - -/* Add a function return address to the trace stack on thread info.*/ -static int push_return_trace(unsigned long ret, unsigned long long time, - unsigned long func) -{ - int index; - struct thread_info *ti = current_thread_info(); - - /* The return trace stack is full */ - if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) - return -EBUSY; - - index = ++ti->curr_ret_stack; - barrier(); - ti->ret_stack[index].ret = ret; - ti->ret_stack[index].func = func; - ti->ret_stack[index].calltime = time; - - return 0; -} - -/* Retrieve a function return address to the trace stack on thread info.*/ -static void pop_return_trace(unsigned long *ret, unsigned long long *time, - unsigned long *func) -{ - int index; - - struct thread_info *ti = current_thread_info(); - index = ti->curr_ret_stack; - *ret = ti->ret_stack[index].ret; - *func = ti->ret_stack[index].func; - *time = ti->ret_stack[index].calltime; - ti->curr_ret_stack--; -} - -/* - * Send the trace to the ring-buffer. - * @return the original return address. - */ -unsigned long ftrace_return_to_handler(void) -{ - struct ftrace_retfunc trace; - pop_return_trace(&trace.ret, &trace.calltime, &trace.func); - trace.rettime = cpu_clock(raw_smp_processor_id()); - ftrace_function_return(&trace); - - return trace.ret; -} - -/* - * Hook the return address and push it in the stack of return addrs - * in current thread info. - */ -void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) -{ - unsigned long old; - unsigned long long calltime; - int faulted; - unsigned long return_hooker = (unsigned long) - &return_to_handler; - - /* Nmi's are currently unsupported */ - if (atomic_read(&in_nmi)) - return; - - /* - * Protect against fault, even if it shouldn't - * happen. This tool is too much intrusive to - * ignore such a protection. - */ - asm volatile( - "1: movl (%[parent_old]), %[old]\n" - "2: movl %[return_hooker], (%[parent_replaced])\n" - " movl $0, %[faulted]\n" - - ".section .fixup, \"ax\"\n" - "3: movl $1, %[faulted]\n" - ".previous\n" - - ".section __ex_table, \"a\"\n" - " .long 1b, 3b\n" - " .long 2b, 3b\n" - ".previous\n" - - : [parent_replaced] "=r" (parent), [old] "=r" (old), - [faulted] "=r" (faulted) - : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) - : "memory" - ); - - if (WARN_ON(faulted)) { - unregister_ftrace_return(); - return; - } - - if (WARN_ON(!__kernel_text_address(old))) { - unregister_ftrace_return(); - *parent = old; - return; - } - - calltime = cpu_clock(raw_smp_processor_id()); - - if (push_return_trace(old, calltime, self_addr) == -EBUSY) - *parent = old; -} - -#endif - #ifdef CONFIG_DYNAMIC_FTRACE union ftrace_code_union { @@ -450,3 +322,133 @@ int __init ftrace_dyn_arch_init(void *data) return 0; } #endif + +#ifdef CONFIG_FUNCTION_RET_TRACER + +#ifndef CONFIG_DYNAMIC_FTRACE + +/* + * These functions are picked from those used on + * this page for dynamic ftrace. They have been + * simplified to ignore all traces in NMI context. + */ +static atomic_t in_nmi; + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); +} + +void ftrace_nmi_exit(void) +{ + atomic_dec(&in_nmi); +} +#endif /* !CONFIG_DYNAMIC_FTRACE */ + +/* Add a function return address to the trace stack on thread info.*/ +static int push_return_trace(unsigned long ret, unsigned long long time, + unsigned long func) +{ + int index; + struct thread_info *ti = current_thread_info(); + + /* The return trace stack is full */ + if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) + return -EBUSY; + + index = ++ti->curr_ret_stack; + barrier(); + ti->ret_stack[index].ret = ret; + ti->ret_stack[index].func = func; + ti->ret_stack[index].calltime = time; + + return 0; +} + +/* Retrieve a function return address to the trace stack on thread info.*/ +static void pop_return_trace(unsigned long *ret, unsigned long long *time, + unsigned long *func) +{ + int index; + + struct thread_info *ti = current_thread_info(); + index = ti->curr_ret_stack; + *ret = ti->ret_stack[index].ret; + *func = ti->ret_stack[index].func; + *time = ti->ret_stack[index].calltime; + ti->curr_ret_stack--; +} + +/* + * Send the trace to the ring-buffer. + * @return the original return address. + */ +unsigned long ftrace_return_to_handler(void) +{ + struct ftrace_retfunc trace; + pop_return_trace(&trace.ret, &trace.calltime, &trace.func); + trace.rettime = cpu_clock(raw_smp_processor_id()); + ftrace_function_return(&trace); + + return trace.ret; +} + +/* + * Hook the return address and push it in the stack of return addrs + * in current thread info. + */ +void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) +{ + unsigned long old; + unsigned long long calltime; + int faulted; + unsigned long return_hooker = (unsigned long) + &return_to_handler; + + /* Nmi's are currently unsupported */ + if (atomic_read(&in_nmi)) + return; + + /* + * Protect against fault, even if it shouldn't + * happen. This tool is too much intrusive to + * ignore such a protection. + */ + asm volatile( + "1: movl (%[parent_old]), %[old]\n" + "2: movl %[return_hooker], (%[parent_replaced])\n" + " movl $0, %[faulted]\n" + + ".section .fixup, \"ax\"\n" + "3: movl $1, %[faulted]\n" + ".previous\n" + + ".section __ex_table, \"a\"\n" + " .long 1b, 3b\n" + " .long 2b, 3b\n" + ".previous\n" + + : [parent_replaced] "=r" (parent), [old] "=r" (old), + [faulted] "=r" (faulted) + : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) + : "memory" + ); + + if (WARN_ON(faulted)) { + unregister_ftrace_return(); + return; + } + + if (WARN_ON(!__kernel_text_address(old))) { + unregister_ftrace_return(); + *parent = old; + return; + } + + calltime = cpu_clock(raw_smp_processor_id()); + + if (push_return_trace(old, calltime, self_addr) == -EBUSY) + *parent = old; +} + +#endif /* CONFIG_FUNCTION_RET_TRACER */ -- cgit v1.2.3 From d1f1e9c01006b4b050e090055c75278f80c2a5c5 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Sat, 15 Nov 2008 11:00:17 +0100 Subject: x86, bts: fix unlock problem in ds.c Fix a problem where ds_request() returned an error without releasing the ds lock. Reported-by: Stephane Eranian Signed-off-by: Markus Metzger Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 2b69994fd3a8..ac1d5b0586ba 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -384,8 +384,9 @@ static int ds_request(struct task_struct *task, void *base, size_t size, spin_lock(&ds_lock); + error = -EPERM; if (!check_tracer(task)) - return -EPERM; + goto out_unlock; error = -ENOMEM; context = ds_alloc_context(task); -- cgit v1.2.3 From d3c6aa1e69f705ac3ab64584101b1d38435b1353 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 16 Nov 2008 00:49:31 -0800 Subject: x86: fix es7000 compiling Impact: fix es7000 build CC arch/x86/kernel/es7000_32.o arch/x86/kernel/es7000_32.c: In function find_unisys_acpi_oem_table: arch/x86/kernel/es7000_32.c:255: error: implicit declaration of function acpi_get_table_with_size arch/x86/kernel/es7000_32.c:261: error: implicit declaration of function early_acpi_os_unmap_memory arch/x86/kernel/es7000_32.c: In function unmap_unisys_acpi_oem_table: arch/x86/kernel/es7000_32.c:277: error: implicit declaration of function __acpi_unmap_table make[1]: *** [arch/x86/kernel/es7000_32.o] Error 1 we applied one patch out of order... | commit a73aaedd95703bd49f4c3f9df06fb7b7373ba905 | Author: Yinghai Lu | Date: Sun Sep 14 02:33:14 2008 -0700 | | x86: check dsdt before find oem table for es7000, v2 | | v2: use __acpi_unmap_table() that patch need: x86: use early_ioremap in __acpi_map_table x86: always explicitly map acpi memory acpi: remove final __acpi_map_table mapping before setting acpi_gbl_permanent_mmap acpi/x86: introduce __apci_map_table, v4 submitted to the ACPI tree but not upstream yet. fix it until those patches applied, need to revert this one Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/es7000_32.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/es7000_32.c b/arch/x86/kernel/es7000_32.c index f454c78fcef6..0aa2c443d600 100644 --- a/arch/x86/kernel/es7000_32.c +++ b/arch/x86/kernel/es7000_32.c @@ -250,31 +250,24 @@ int __init find_unisys_acpi_oem_table(unsigned long *oem_addr) { struct acpi_table_header *header = NULL; int i = 0; - acpi_size tbl_size; - while (ACPI_SUCCESS(acpi_get_table_with_size("OEM1", i++, &header, &tbl_size))) { + while (ACPI_SUCCESS(acpi_get_table("OEM1", i++, &header))) { if (!memcmp((char *) &header->oem_id, "UNISYS", 6)) { struct oem_table *t = (struct oem_table *)header; oem_addrX = t->OEMTableAddr; oem_size = t->OEMTableSize; - early_acpi_os_unmap_memory(header, tbl_size); *oem_addr = (unsigned long)__acpi_map_table(oem_addrX, oem_size); return 0; } - early_acpi_os_unmap_memory(header, tbl_size); } return -1; } void __init unmap_unisys_acpi_oem_table(unsigned long oem_addr) { - if (!oem_addr) - return; - - __acpi_unmap_table((char *)oem_addr, oem_size); } #endif -- cgit v1.2.3 From 0bd7b79851d0f74b24a9ce87d088f2e7c718f668 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Sun, 16 Nov 2008 15:29:00 +0100 Subject: x86: entry_64.S: remove whitespace at end of lines Impact: cleanup All blame goes to: color white,red "[^[:graph:]]+$" in .nanorc ;). Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 190 ++++++++++++++++++++++----------------------- 1 file changed, 95 insertions(+), 95 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index b86f332c96a6..54927784bab9 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -11,15 +11,15 @@ * * NOTE: This code handles signal-recognition, which happens every time * after an interrupt and after each system call. - * - * Normal syscalls and interrupts don't save a full stack frame, this is + * + * Normal syscalls and interrupts don't save a full stack frame, this is * only done for syscall tracing, signals or fork/exec et.al. - * - * A note on terminology: - * - top of stack: Architecture defined interrupt frame from SS to RIP - * at the top of the kernel process stack. + * + * A note on terminology: + * - top of stack: Architecture defined interrupt frame from SS to RIP + * at the top of the kernel process stack. * - partial stack frame: partially saved registers upto R11. - * - full stack frame: Like partial stack frame, but all register saved. + * - full stack frame: Like partial stack frame, but all register saved. * * Some macro usage: * - CFI macros are used to generate dwarf2 unwind information for better @@ -142,7 +142,7 @@ END(mcount) #ifndef CONFIG_PREEMPT #define retint_kernel retint_restore_args -#endif +#endif #ifdef CONFIG_PARAVIRT ENTRY(native_usergs_sysret64) @@ -161,14 +161,14 @@ ENTRY(native_usergs_sysret64) .endm /* - * C code is not supposed to know about undefined top of stack. Every time - * a C function with an pt_regs argument is called from the SYSCALL based + * C code is not supposed to know about undefined top of stack. Every time + * a C function with an pt_regs argument is called from the SYSCALL based * fast path FIXUP_TOP_OF_STACK is needed. * RESTORE_TOP_OF_STACK syncs the syscall state after any possible ptregs * manipulation. - */ - - /* %rsp:at FRAMEEND */ + */ + + /* %rsp:at FRAMEEND */ .macro FIXUP_TOP_OF_STACK tmp movq %gs:pda_oldrsp,\tmp movq \tmp,RSP(%rsp) @@ -244,8 +244,8 @@ ENTRY(native_usergs_sysret64) .endm /* * A newly forked process directly context switches into this. - */ -/* rdi: prev */ + */ +/* rdi: prev */ ENTRY(ret_from_fork) CFI_DEFAULT_STACK push kernel_eflags(%rip) @@ -256,7 +256,7 @@ ENTRY(ret_from_fork) GET_THREAD_INFO(%rcx) testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),TI_flags(%rcx) jnz rff_trace -rff_action: +rff_action: RESTORE_REST testl $3,CS-ARGOFFSET(%rsp) # from kernel_thread? je int_ret_from_sys_call @@ -267,7 +267,7 @@ rff_action: rff_trace: movq %rsp,%rdi call syscall_trace_leave - GET_THREAD_INFO(%rcx) + GET_THREAD_INFO(%rcx) jmp rff_action CFI_ENDPROC END(ret_from_fork) @@ -278,20 +278,20 @@ END(ret_from_fork) * SYSCALL does not save anything on the stack and does not change the * stack pointer. */ - + /* - * Register setup: + * Register setup: * rax system call number * rdi arg0 - * rcx return address for syscall/sysret, C arg3 + * rcx return address for syscall/sysret, C arg3 * rsi arg1 - * rdx arg2 + * rdx arg2 * r10 arg3 (--> moved to rcx for C) * r8 arg4 * r9 arg5 * r11 eflags for syscall/sysret, temporary for C - * r12-r15,rbp,rbx saved by C code, not touched. - * + * r12-r15,rbp,rbx saved by C code, not touched. + * * Interrupts are off on entry. * Only called from user space. * @@ -301,7 +301,7 @@ END(ret_from_fork) * When user can change the frames always force IRET. That is because * it deals with uncanonical addresses better. SYSRET has trouble * with them due to bugs in both AMD and Intel CPUs. - */ + */ ENTRY(system_call) CFI_STARTPROC simple @@ -317,7 +317,7 @@ ENTRY(system_call) */ ENTRY(system_call_after_swapgs) - movq %rsp,%gs:pda_oldrsp + movq %rsp,%gs:pda_oldrsp movq %gs:pda_kernelstack,%rsp /* * No need to follow this irqs off/on section - it's straight @@ -325,7 +325,7 @@ ENTRY(system_call_after_swapgs) */ ENABLE_INTERRUPTS(CLBR_NONE) SAVE_ARGS 8,1 - movq %rax,ORIG_RAX-ARGOFFSET(%rsp) + movq %rax,ORIG_RAX-ARGOFFSET(%rsp) movq %rcx,RIP-ARGOFFSET(%rsp) CFI_REL_OFFSET rip,RIP-ARGOFFSET GET_THREAD_INFO(%rcx) @@ -339,19 +339,19 @@ system_call_fastpath: movq %rax,RAX-ARGOFFSET(%rsp) /* * Syscall return path ending with SYSRET (fast path) - * Has incomplete stack frame and undefined top of stack. - */ + * Has incomplete stack frame and undefined top of stack. + */ ret_from_sys_call: movl $_TIF_ALLWORK_MASK,%edi /* edi: flagmask */ -sysret_check: +sysret_check: LOCKDEP_SYS_EXIT GET_THREAD_INFO(%rcx) DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF movl TI_flags(%rcx),%edx andl %edi,%edx - jnz sysret_careful + jnz sysret_careful CFI_REMEMBER_STATE /* * sysretq will re-enable interrupts: @@ -366,7 +366,7 @@ sysret_check: CFI_RESTORE_STATE /* Handle reschedules */ - /* edx: work, edi: workmask */ + /* edx: work, edi: workmask */ sysret_careful: bt $TIF_NEED_RESCHED,%edx jnc sysret_signal @@ -379,7 +379,7 @@ sysret_careful: CFI_ADJUST_CFA_OFFSET -8 jmp sysret_check - /* Handle a signal */ + /* Handle a signal */ sysret_signal: TRACE_IRQS_ON ENABLE_INTERRUPTS(CLBR_NONE) @@ -398,7 +398,7 @@ sysret_signal: DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF jmp int_with_check - + badsys: movq $-ENOSYS,RAX-ARGOFFSET(%rsp) jmp ret_from_sys_call @@ -437,7 +437,7 @@ sysret_audit: #endif /* CONFIG_AUDITSYSCALL */ /* Do syscall tracing */ -tracesys: +tracesys: #ifdef CONFIG_AUDITSYSCALL testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags(%rcx) jz auditsys @@ -460,8 +460,8 @@ tracesys: call *sys_call_table(,%rax,8) movq %rax,RAX-ARGOFFSET(%rsp) /* Use IRET because user could have changed frame */ - -/* + +/* * Syscall return path ending with IRET. * Has correct top of stack, but partial stack frame. */ @@ -505,18 +505,18 @@ int_very_careful: TRACE_IRQS_ON ENABLE_INTERRUPTS(CLBR_NONE) SAVE_REST - /* Check for syscall exit trace */ + /* Check for syscall exit trace */ testl $_TIF_WORK_SYSCALL_EXIT,%edx jz int_signal pushq %rdi CFI_ADJUST_CFA_OFFSET 8 - leaq 8(%rsp),%rdi # &ptregs -> arg1 + leaq 8(%rsp),%rdi # &ptregs -> arg1 call syscall_trace_leave popq %rdi CFI_ADJUST_CFA_OFFSET -8 andl $~(_TIF_WORK_SYSCALL_EXIT|_TIF_SYSCALL_EMU),%edi jmp int_restore_rest - + int_signal: testl $_TIF_DO_NOTIFY_MASK,%edx jz 1f @@ -531,11 +531,11 @@ int_restore_rest: jmp int_with_check CFI_ENDPROC END(system_call) - -/* + +/* * Certain special system calls that need to save a complete full stack frame. - */ - + */ + .macro PTREGSCALL label,func,arg .globl \label \label: @@ -572,7 +572,7 @@ ENTRY(ptregscall_common) ret CFI_ENDPROC END(ptregscall_common) - + ENTRY(stub_execve) CFI_STARTPROC popq %r11 @@ -588,11 +588,11 @@ ENTRY(stub_execve) jmp int_ret_from_sys_call CFI_ENDPROC END(stub_execve) - + /* * sigreturn is special because it needs to restore all registers on return. * This cannot be done with SYSRET, so use the IRET return path instead. - */ + */ ENTRY(stub_rt_sigreturn) CFI_STARTPROC addq $8, %rsp @@ -685,12 +685,12 @@ exit_intr: GET_THREAD_INFO(%rcx) testl $3,CS-ARGOFFSET(%rsp) je retint_kernel - + /* Interrupt came from user space */ /* * Has a correct top of stack, but a partial stack frame * %rcx: thread info. Interrupts off. - */ + */ retint_with_reschedule: movl $_TIF_WORK_MASK,%edi retint_check: @@ -763,20 +763,20 @@ retint_careful: pushq %rdi CFI_ADJUST_CFA_OFFSET 8 call schedule - popq %rdi + popq %rdi CFI_ADJUST_CFA_OFFSET -8 GET_THREAD_INFO(%rcx) DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF jmp retint_check - + retint_signal: testl $_TIF_DO_NOTIFY_MASK,%edx jz retint_swapgs TRACE_IRQS_ON ENABLE_INTERRUPTS(CLBR_NONE) SAVE_REST - movq $-1,ORIG_RAX(%rsp) + movq $-1,ORIG_RAX(%rsp) xorl %esi,%esi # oldset movq %rsp,%rdi # &pt_regs call do_notify_resume @@ -798,14 +798,14 @@ ENTRY(retint_kernel) jnc retint_restore_args call preempt_schedule_irq jmp exit_intr -#endif +#endif CFI_ENDPROC END(common_interrupt) - + /* * APIC interrupts. - */ + */ .macro apicinterrupt num,func INTR_FRAME pushq $~(\num) @@ -823,14 +823,14 @@ ENTRY(threshold_interrupt) apicinterrupt THRESHOLD_APIC_VECTOR,mce_threshold_interrupt END(threshold_interrupt) -#ifdef CONFIG_SMP +#ifdef CONFIG_SMP ENTRY(reschedule_interrupt) apicinterrupt RESCHEDULE_VECTOR,smp_reschedule_interrupt END(reschedule_interrupt) .macro INVALIDATE_ENTRY num ENTRY(invalidate_interrupt\num) - apicinterrupt INVALIDATE_TLB_VECTOR_START+\num,smp_invalidate_interrupt + apicinterrupt INVALIDATE_TLB_VECTOR_START+\num,smp_invalidate_interrupt END(invalidate_interrupt\num) .endm @@ -869,22 +869,22 @@ END(error_interrupt) ENTRY(spurious_interrupt) apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt END(spurious_interrupt) - + /* * Exception entry points. - */ + */ .macro zeroentry sym INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $0 /* push error code/oldrax */ + pushq $0 /* push error code/oldrax */ CFI_ADJUST_CFA_OFFSET 8 - pushq %rax /* push real oldrax to the rdi slot */ + pushq %rax /* push real oldrax to the rdi slot */ CFI_ADJUST_CFA_OFFSET 8 CFI_REL_OFFSET rax,0 leaq \sym(%rip),%rax jmp error_entry CFI_ENDPROC - .endm + .endm .macro errorentry sym XCPT_FRAME @@ -998,13 +998,13 @@ paranoid_schedule\trace: /* * Exception entry point. This expects an error code/orig_rax on the stack - * and the exception handler in %rax. - */ + * and the exception handler in %rax. + */ KPROBE_ENTRY(error_entry) _frame RDI CFI_REL_OFFSET rax,0 /* rdi slot contains rax, oldrax contains error code */ - cld + cld subq $14*8,%rsp CFI_ADJUST_CFA_OFFSET (14*8) movq %rsi,13*8(%rsp) @@ -1015,7 +1015,7 @@ KPROBE_ENTRY(error_entry) CFI_REL_OFFSET rdx,RDX movq %rcx,11*8(%rsp) CFI_REL_OFFSET rcx,RCX - movq %rsi,10*8(%rsp) /* store rax */ + movq %rsi,10*8(%rsp) /* store rax */ CFI_REL_OFFSET rax,RAX movq %r8, 9*8(%rsp) CFI_REL_OFFSET r8,R8 @@ -1025,29 +1025,29 @@ KPROBE_ENTRY(error_entry) CFI_REL_OFFSET r10,R10 movq %r11,6*8(%rsp) CFI_REL_OFFSET r11,R11 - movq %rbx,5*8(%rsp) + movq %rbx,5*8(%rsp) CFI_REL_OFFSET rbx,RBX - movq %rbp,4*8(%rsp) + movq %rbp,4*8(%rsp) CFI_REL_OFFSET rbp,RBP - movq %r12,3*8(%rsp) + movq %r12,3*8(%rsp) CFI_REL_OFFSET r12,R12 - movq %r13,2*8(%rsp) + movq %r13,2*8(%rsp) CFI_REL_OFFSET r13,R13 - movq %r14,1*8(%rsp) + movq %r14,1*8(%rsp) CFI_REL_OFFSET r14,R14 - movq %r15,(%rsp) + movq %r15,(%rsp) CFI_REL_OFFSET r15,R15 - xorl %ebx,%ebx + xorl %ebx,%ebx testl $3,CS(%rsp) je error_kernelspace -error_swapgs: +error_swapgs: SWAPGS error_sti: TRACE_IRQS_OFF - movq %rdi,RDI(%rsp) + movq %rdi,RDI(%rsp) CFI_REL_OFFSET rdi,RDI movq %rsp,%rdi - movq ORIG_RAX(%rsp),%rsi /* get error code */ + movq ORIG_RAX(%rsp),%rsi /* get error code */ movq $-1,ORIG_RAX(%rsp) call *%rax /* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ @@ -1056,7 +1056,7 @@ error_exit: RESTORE_REST DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF - GET_THREAD_INFO(%rcx) + GET_THREAD_INFO(%rcx) testl %eax,%eax jne retint_kernel LOCKDEP_SYS_EXIT_IRQ @@ -1072,7 +1072,7 @@ error_kernelspace: /* There are two places in the kernel that can potentially fault with usergs. Handle them here. The exception handlers after iret run with kernel gs again, so don't set the user space flag. - B stepping K8s sometimes report an truncated RIP for IRET + B stepping K8s sometimes report an truncated RIP for IRET exceptions returning to compat mode. Check for these here too. */ leaq irq_return(%rip),%rcx cmpq %rcx,RIP(%rsp) @@ -1084,17 +1084,17 @@ error_kernelspace: je error_swapgs jmp error_sti KPROBE_END(error_entry) - + /* Reload gs selector with exception handling */ - /* edi: new selector */ + /* edi: new selector */ ENTRY(native_load_gs_index) CFI_STARTPROC pushf CFI_ADJUST_CFA_OFFSET 8 DISABLE_INTERRUPTS(CLBR_ANY | ~(CLBR_RDI)) SWAPGS -gs_change: - movl %edi,%gs +gs_change: + movl %edi,%gs 2: mfence /* workaround */ SWAPGS popf @@ -1102,20 +1102,20 @@ gs_change: ret CFI_ENDPROC ENDPROC(native_load_gs_index) - + .section __ex_table,"a" .align 8 .quad gs_change,bad_gs .previous .section .fixup,"ax" /* running with kernelgs */ -bad_gs: +bad_gs: SWAPGS /* switch back to user gs */ xorl %eax,%eax movl %eax,%gs jmp 2b - .previous - + .previous + /* * Create a kernel thread. * @@ -1138,7 +1138,7 @@ ENTRY(kernel_thread) xorl %r8d,%r8d xorl %r9d,%r9d - + # clone now call do_fork movq %rax,RAX(%rsp) @@ -1149,14 +1149,14 @@ ENTRY(kernel_thread) * so internally to the x86_64 port you can rely on kernel_thread() * not to reschedule the child before returning, this avoids the need * of hacks for example to fork off the per-CPU idle tasks. - * [Hopefully no generic code relies on the reschedule -AK] + * [Hopefully no generic code relies on the reschedule -AK] */ RESTORE_ALL UNFAKE_STACK_FRAME ret CFI_ENDPROC ENDPROC(kernel_thread) - + child_rip: pushq $0 # fake return address CFI_STARTPROC @@ -1191,10 +1191,10 @@ ENDPROC(child_rip) ENTRY(kernel_execve) CFI_STARTPROC FAKE_STACK_FRAME $0 - SAVE_ALL + SAVE_ALL movq %rsp,%rcx call sys_execve - movq %rax, RAX(%rsp) + movq %rax, RAX(%rsp) RESTORE_REST testq %rax,%rax je int_ret_from_sys_call @@ -1213,7 +1213,7 @@ ENTRY(coprocessor_error) END(coprocessor_error) ENTRY(simd_coprocessor_error) - zeroentry do_simd_coprocessor_error + zeroentry do_simd_coprocessor_error END(simd_coprocessor_error) ENTRY(device_not_available) @@ -1225,12 +1225,12 @@ KPROBE_ENTRY(debug) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $0 - CFI_ADJUST_CFA_OFFSET 8 + CFI_ADJUST_CFA_OFFSET 8 paranoidentry do_debug, DEBUG_STACK paranoidexit KPROBE_END(debug) - /* runs on exception stack */ + /* runs on exception stack */ KPROBE_ENTRY(nmi) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME @@ -1264,7 +1264,7 @@ ENTRY(bounds) END(bounds) ENTRY(invalid_op) - zeroentry do_invalid_op + zeroentry do_invalid_op END(invalid_op) ENTRY(coprocessor_segment_overrun) @@ -1319,7 +1319,7 @@ ENTRY(machine_check) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $0 - CFI_ADJUST_CFA_OFFSET 8 + CFI_ADJUST_CFA_OFFSET 8 paranoidentry do_machine_check jmp paranoid_exit1 CFI_ENDPROC -- cgit v1.2.3 From 569712b2b0970fa5b19673544d62ae661d04a220 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 16 Nov 2008 03:12:49 -0800 Subject: x86: fix wakeup_cpu with numaq/es7000, v2 Impact: fix secondary-CPU wakeup/init path with numaq and es7000 While looking at wakeup_secondary_cpu for WAKE_SECONDARY_VIA_NMI: |#ifdef WAKE_SECONDARY_VIA_NMI |/* | * Poke the other CPU in the eye via NMI to wake it up. Remember that the normal | * INIT, INIT, STARTUP sequence will reset the chip hard for us, and this | * won't ... remember to clear down the APIC, etc later. | */ |static int __devinit |wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip) |{ | unsigned long send_status, accept_status = 0; | int maxlvt; |... | if (APIC_INTEGRATED(apic_version[phys_apicid])) { | maxlvt = lapic_get_maxlvt(); I noticed that there is no warning about undefined phys_apicid... because WAKE_SECONDARY_VIA_NMI and WAKE_SECONDARY_VIA_INIT can not be defined at the same time. So NUMAQ is using wrong wakeup_secondary_cpu. WAKE_SECONDARY_VIA_NMI, WAKE_SECONDARY_VIA_INIT and WAKE_SECONDARY_VIA_MIP are variants of a weird and fragile preprocessor-driven "HAL" mechanisms to specify the kind of secondary-CPU wakeup strategy a given x86 kernel will use. The vast majority of systems want to use INIT for secondary wakeup - NUMAQ uses an NMI, (old-style-) ES7000 uses 'MIP' (a firmware driven in-memory flag to let secondaries continue). So convert these mechanisms to x86_quirks and add a ->wakeup_secondary_cpu() method to specify the rare exception to the sane default. Extend genapic accordingly as well, for 32-bit. While looking further, I noticed that functions in wakecup.h for numaq and es7000 are different to the default in mach_wakecpu.h - but smpboot.c will only use default mach_wakecpu.h with smphook.h. So we need to add mach_wakecpu.h for mach_generic, to properly support numaq and es7000, and vectorize the following SMP init methods: int trampoline_phys_low; int trampoline_phys_high; void (*wait_for_init_deassert)(atomic_t *deassert); void (*smp_callin_clear_local_apic)(void); void (*store_NMI_vector)(unsigned short *high, unsigned short *low); void (*restore_NMI_vector)(unsigned short *high, unsigned short *low); void (*inquire_remote_apic)(int apicid); Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/es7000_32.c | 44 ++++++++++++++++++++++++-------------------- arch/x86/kernel/numaq_32.c | 1 + arch/x86/kernel/smpboot.c | 24 +++++++++++++++--------- 3 files changed, 40 insertions(+), 29 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/es7000_32.c b/arch/x86/kernel/es7000_32.c index f454c78fcef6..bed10dddf099 100644 --- a/arch/x86/kernel/es7000_32.c +++ b/arch/x86/kernel/es7000_32.c @@ -40,6 +40,7 @@ #include #include #include +#include /* * ES7000 chipsets @@ -161,6 +162,26 @@ es7000_rename_gsi(int ioapic, int gsi) return gsi; } +#ifdef CONFIG_ES7000_CLUSTERED_APIC +static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip) +{ + unsigned long vect = 0, psaival = 0; + + if (psai == NULL) + return -1; + + vect = ((unsigned long)__pa(eip)/0x1000) << 16; + psaival = (0x1000000 | vect | cpu); + + while (*psai & 0x1000000) + ; + + *psai = psaival; + + return 0; +} +#endif + void __init setup_unisys(void) { @@ -176,6 +197,9 @@ setup_unisys(void) else es7000_plat = ES7000_CLASSIC; ioapic_renumber_irq = es7000_rename_gsi; +#ifdef CONFIG_ES7000_CLUSTERED_APIC + x86_quirks->wakeup_secondary_cpu = wakeup_secondary_cpu_via_mip; +#endif } /* @@ -324,26 +348,6 @@ es7000_mip_write(struct mip_reg *mip_reg) return status; } -int -es7000_start_cpu(int cpu, unsigned long eip) -{ - unsigned long vect = 0, psaival = 0; - - if (psai == NULL) - return -1; - - vect = ((unsigned long)__pa(eip)/0x1000) << 16; - psaival = (0x1000000 | vect | cpu); - - while (*psai & 0x1000000) - ; - - *psai = psaival; - - return 0; - -} - void __init es7000_sw_apic(void) { diff --git a/arch/x86/kernel/numaq_32.c b/arch/x86/kernel/numaq_32.c index 4caff39078e0..745891b7d0fb 100644 --- a/arch/x86/kernel/numaq_32.c +++ b/arch/x86/kernel/numaq_32.c @@ -250,6 +250,7 @@ static struct x86_quirks numaq_x86_quirks __initdata = { .mpc_oem_pci_bus = mpc_oem_pci_bus, .smp_read_mpc_oem = smp_read_mpc_oem, .setup_ioapic_ids = numaq_setup_ioapic_ids, + .wakeup_secondary_cpu = wakeup_secondary_cpu_via_nmi, }; void numaq_mps_oem_check(struct mp_config_table *mpc, char *oem, diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 7b1093397319..498c1ef37fe0 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -62,6 +62,7 @@ #include #include #include +#include #include #include @@ -536,7 +537,7 @@ static void impress_friends(void) pr_debug("Before bogocount - setting activated=1.\n"); } -static inline void __inquire_remote_apic(int apicid) +void __inquire_remote_apic(int apicid) { unsigned i, regs[] = { APIC_ID >> 4, APIC_LVR >> 4, APIC_SPIV >> 4 }; char *names[] = { "ID", "VERSION", "SPIV" }; @@ -575,14 +576,13 @@ static inline void __inquire_remote_apic(int apicid) } } -#ifdef WAKE_SECONDARY_VIA_NMI /* * Poke the other CPU in the eye via NMI to wake it up. Remember that the normal * INIT, INIT, STARTUP sequence will reset the chip hard for us, and this * won't ... remember to clear down the APIC, etc later. */ -static int __devinit -wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip) +int __devinit +wakeup_secondary_cpu_via_nmi(int logical_apicid, unsigned long start_eip) { unsigned long send_status, accept_status = 0; int maxlvt; @@ -599,7 +599,7 @@ wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip) * Give the other CPU some time to accept the IPI. */ udelay(200); - if (APIC_INTEGRATED(apic_version[phys_apicid])) { + if (APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid])) { maxlvt = lapic_get_maxlvt(); if (maxlvt > 3) /* Due to the Pentium erratum 3AP. */ apic_write(APIC_ESR, 0); @@ -614,11 +614,9 @@ wakeup_secondary_cpu(int logical_apicid, unsigned long start_eip) return (send_status | accept_status); } -#endif /* WAKE_SECONDARY_VIA_NMI */ -#ifdef WAKE_SECONDARY_VIA_INIT static int __devinit -wakeup_secondary_cpu(int phys_apicid, unsigned long start_eip) +wakeup_secondary_cpu_via_init(int phys_apicid, unsigned long start_eip) { unsigned long send_status, accept_status = 0; int maxlvt, num_starts, j; @@ -737,7 +735,15 @@ wakeup_secondary_cpu(int phys_apicid, unsigned long start_eip) return (send_status | accept_status); } -#endif /* WAKE_SECONDARY_VIA_INIT */ + +static int __devinit +wakeup_secondary_cpu(int apicid, unsigned long start_eip) +{ + if (x86_quirks->wakeup_secondary_cpu) + return x86_quirks->wakeup_secondary_cpu(apicid, start_eip); + + return wakeup_secondary_cpu_via_init(apicid, start_eip); +} struct create_idle { struct work_struct work; -- cgit v1.2.3 From 93ce99e849433ede4ce8b410b749dc0cad1100b2 Mon Sep 17 00:00:00 2001 From: Venki Pallipadi Date: Mon, 17 Nov 2008 14:43:58 -0800 Subject: x86: add rdtsc barrier to TSC sync check Impact: fix incorrectly marked unstable TSC clock Patch (commit 0d12cdd "sched: improve sched_clock() performance") has a regression on one of the test systems here. With the patch, I see: checking TSC synchronization [CPU#0 -> CPU#1]: Measured 28 cycles TSC warp between CPUs, turning off TSC clock. Marking TSC unstable due to check_tsc_sync_source failed Whereas, without the patch syncs pass fine on all CPUs: checking TSC synchronization [CPU#0 -> CPU#1]: passed. Due to this, TSC is marked unstable, when it is not actually unstable. This is because syncs in check_tsc_wrap() goes away due to this commit. As per the discussion on this thread, correct way to fix this is to add explicit syncs as below? Signed-off-by: Venkatesh Pallipadi Signed-off-by: Ingo Molnar --- arch/x86/kernel/tsc_sync.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc_sync.c b/arch/x86/kernel/tsc_sync.c index 9ffb01c31c40..1c0dfbca87c1 100644 --- a/arch/x86/kernel/tsc_sync.c +++ b/arch/x86/kernel/tsc_sync.c @@ -46,7 +46,9 @@ static __cpuinit void check_tsc_warp(void) cycles_t start, now, prev, end; int i; + rdtsc_barrier(); start = get_cycles(); + rdtsc_barrier(); /* * The measurement runs for 20 msecs: */ @@ -61,7 +63,9 @@ static __cpuinit void check_tsc_warp(void) */ __raw_spin_lock(&sync_lock); prev = last_tsc; + rdtsc_barrier(); now = get_cycles(); + rdtsc_barrier(); last_tsc = now; __raw_spin_unlock(&sync_lock); -- cgit v1.2.3 From 54ac14a8e982ae6c7ac71ee2b0d0173b974509e2 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Mon, 17 Nov 2008 15:19:53 -0800 Subject: x86: fix wakeup_cpu with numaq/es7000, v2, fix Impact: fix wakeup_secondary_cpu with hotplug We can not put that into x86_quirks, because that is __initdata. So try to move that to genapic, and add update_genapic in x86_quirks. later we even could use that stub to: 1. autodetect CONFIG_ES7000_CLUSTERED_APIC 2. more correct inquire_remote_apic with apic_verbosity setting. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/es7000_32.c | 11 ++++++++++- arch/x86/kernel/genapic_64.c | 4 ++++ arch/x86/kernel/numaq_32.c | 11 +++++++++-- arch/x86/kernel/setup.c | 13 ++++++++++++- arch/x86/kernel/smpboot.c | 11 +---------- 5 files changed, 36 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/es7000_32.c b/arch/x86/kernel/es7000_32.c index bed10dddf099..fb3bfe66fbe2 100644 --- a/arch/x86/kernel/es7000_32.c +++ b/arch/x86/kernel/es7000_32.c @@ -40,6 +40,7 @@ #include #include #include +#include #include /* @@ -180,6 +181,13 @@ static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip) return 0; } + +static int __init es7000_update_genapic(void) +{ + genapic->wakeup_cpu = wakeup_secondary_cpu_via_mip; + + return 0; +} #endif void __init @@ -197,8 +205,9 @@ setup_unisys(void) else es7000_plat = ES7000_CLASSIC; ioapic_renumber_irq = es7000_rename_gsi; + #ifdef CONFIG_ES7000_CLUSTERED_APIC - x86_quirks->wakeup_secondary_cpu = wakeup_secondary_cpu_via_mip; + x86_quirks->update_genapic = es7000_update_genapic; #endif } diff --git a/arch/x86/kernel/genapic_64.c b/arch/x86/kernel/genapic_64.c index 6c9bfc9e1e95..2bced78b0b8e 100644 --- a/arch/x86/kernel/genapic_64.c +++ b/arch/x86/kernel/genapic_64.c @@ -21,6 +21,7 @@ #include #include #include +#include extern struct genapic apic_flat; extern struct genapic apic_physflat; @@ -53,6 +54,9 @@ void __init setup_apic_routing(void) genapic = &apic_physflat; printk(KERN_INFO "Setting APIC routing to %s\n", genapic->name); } + + if (x86_quirks->update_genapic) + x86_quirks->update_genapic(); } /* Same for both flat and physical. */ diff --git a/arch/x86/kernel/numaq_32.c b/arch/x86/kernel/numaq_32.c index 745891b7d0fb..0deea37a53cf 100644 --- a/arch/x86/kernel/numaq_32.c +++ b/arch/x86/kernel/numaq_32.c @@ -31,7 +31,7 @@ #include #include #include -#include +#include #include #include @@ -235,6 +235,13 @@ static int __init numaq_setup_ioapic_ids(void) return 1; } +static int __init numaq_update_genapic(void) +{ + genapic->wakeup_cpu = wakeup_secondary_cpu_via_nmi; + + return 0; +} + static struct x86_quirks numaq_x86_quirks __initdata = { .arch_pre_time_init = numaq_pre_time_init, .arch_time_init = NULL, @@ -250,7 +257,7 @@ static struct x86_quirks numaq_x86_quirks __initdata = { .mpc_oem_pci_bus = mpc_oem_pci_bus, .smp_read_mpc_oem = smp_read_mpc_oem, .setup_ioapic_ids = numaq_setup_ioapic_ids, - .wakeup_secondary_cpu = wakeup_secondary_cpu_via_nmi, + .update_genapic = numaq_update_genapic, }; void numaq_mps_oem_check(struct mp_config_table *mpc, char *oem, diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 0fa6790c1dd3..c366e891e10b 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -583,7 +583,18 @@ static int __init setup_elfcorehdr(char *arg) early_param("elfcorehdr", setup_elfcorehdr); #endif -static struct x86_quirks default_x86_quirks __initdata; +static int __init default_update_genapic(void) +{ +#if defined(CONFIG_X86_GENERICARCH) || defined(CONFIG_X86_64) + genapic->wakeup_cpu = wakeup_secondary_cpu_via_nmi; +#endif + + return 0; +} + +static struct x86_quirks default_x86_quirks __initdata = { + .update_genapic = default_update_genapic, +}; struct x86_quirks *x86_quirks __initdata = &default_x86_quirks; diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 498c1ef37fe0..0e9f446269f4 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -615,7 +615,7 @@ wakeup_secondary_cpu_via_nmi(int logical_apicid, unsigned long start_eip) return (send_status | accept_status); } -static int __devinit +int __devinit wakeup_secondary_cpu_via_init(int phys_apicid, unsigned long start_eip) { unsigned long send_status, accept_status = 0; @@ -736,15 +736,6 @@ wakeup_secondary_cpu_via_init(int phys_apicid, unsigned long start_eip) return (send_status | accept_status); } -static int __devinit -wakeup_secondary_cpu(int apicid, unsigned long start_eip) -{ - if (x86_quirks->wakeup_secondary_cpu) - return x86_quirks->wakeup_secondary_cpu(apicid, start_eip); - - return wakeup_secondary_cpu_via_init(apicid, start_eip); -} - struct create_idle { struct work_struct work; struct task_struct *idle; -- cgit v1.2.3 From 0231022cc32d5f2e7f3c06b75691dda0ad6aec33 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Mon, 17 Nov 2008 03:22:41 +0100 Subject: tracing/function-return-tracer: add the overrun field Impact: help to find the better depth of trace We decided to arbitrary define the depth of function return trace as "20". Perhaps this is not enough. To help finding an optimal depth, we measure now the overrun: the number of functions that have been missed for the current thread. By default this is not displayed, we have to do set a particular flag on the return tracer: echo overrun > /debug/tracing/trace_options And the overrun will be printed on the right. As the trace shows below, the current 20 depth is not enough. update_wall_time+0x37f/0x8c0 -> update_xtime_cache (345 ns) (Overruns: 2838) update_wall_time+0x384/0x8c0 -> clocksource_get_next (1141 ns) (Overruns: 2838) do_timer+0x23/0x100 -> update_wall_time (3882 ns) (Overruns: 2838) tick_do_update_jiffies64+0xbf/0x160 -> do_timer (5339 ns) (Overruns: 2838) tick_sched_timer+0x6a/0xf0 -> tick_do_update_jiffies64 (7209 ns) (Overruns: 2838) vgacon_set_cursor_size+0x98/0x120 -> native_io_delay (2613 ns) (Overruns: 274) vgacon_cursor+0x16e/0x1d0 -> vgacon_set_cursor_size (33151 ns) (Overruns: 274) set_cursor+0x5f/0x80 -> vgacon_cursor (36432 ns) (Overruns: 274) con_flush_chars+0x34/0x40 -> set_cursor (38790 ns) (Overruns: 274) release_console_sem+0x1ec/0x230 -> up (721 ns) (Overruns: 274) release_console_sem+0x225/0x230 -> wake_up_klogd (316 ns) (Overruns: 274) con_flush_chars+0x39/0x40 -> release_console_sem (2996 ns) (Overruns: 274) con_write+0x22/0x30 -> con_flush_chars (46067 ns) (Overruns: 274) n_tty_write+0x1cc/0x360 -> con_write (292670 ns) (Overruns: 274) smp_apic_timer_interrupt+0x2a/0x90 -> native_apic_mem_write (330 ns) (Overruns: 274) irq_enter+0x17/0x70 -> idle_cpu (413 ns) (Overruns: 274) smp_apic_timer_interrupt+0x2f/0x90 -> irq_enter (1525 ns) (Overruns: 274) ktime_get_ts+0x40/0x70 -> getnstimeofday (465 ns) (Overruns: 274) ktime_get_ts+0x60/0x70 -> set_normalized_timespec (436 ns) (Overruns: 274) ktime_get+0x16/0x30 -> ktime_get_ts (2501 ns) (Overruns: 274) hrtimer_interrupt+0x77/0x1a0 -> ktime_get (3439 ns) (Overruns: 274) Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 924153edd973..356bb1eb6e9a 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -353,8 +353,10 @@ static int push_return_trace(unsigned long ret, unsigned long long time, struct thread_info *ti = current_thread_info(); /* The return trace stack is full */ - if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) + if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { + atomic_inc(&ti->trace_overrun); return -EBUSY; + } index = ++ti->curr_ret_stack; barrier(); @@ -367,7 +369,7 @@ static int push_return_trace(unsigned long ret, unsigned long long time, /* Retrieve a function return address to the trace stack on thread info.*/ static void pop_return_trace(unsigned long *ret, unsigned long long *time, - unsigned long *func) + unsigned long *func, unsigned long *overrun) { int index; @@ -376,6 +378,7 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, *ret = ti->ret_stack[index].ret; *func = ti->ret_stack[index].func; *time = ti->ret_stack[index].calltime; + *overrun = atomic_read(&ti->trace_overrun); ti->curr_ret_stack--; } @@ -386,7 +389,8 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, unsigned long ftrace_return_to_handler(void) { struct ftrace_retfunc trace; - pop_return_trace(&trace.ret, &trace.calltime, &trace.func); + pop_return_trace(&trace.ret, &trace.calltime, &trace.func, + &trace.overrun); trace.rettime = cpu_clock(raw_smp_processor_id()); ftrace_function_return(&trace); -- cgit v1.2.3 From 10db4ef7b9a65b86e4d047671a1886f4c101a859 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 18 Nov 2008 15:23:08 +0100 Subject: x86, PEBS/DS: fix code flow in ds_request() this compiler warning: arch/x86/kernel/ds.c: In function 'ds_request': arch/x86/kernel/ds.c:368: warning: 'context' may be used uninitialized in this function Shows that the code flow in ds_request() is buggy - it goes into the unlock+release-context path even when the context is not allocated yet. First allocate the context, then do the other checks. Also, take care with GFP allocations under the ds_lock spinlock. Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 26 +++++++++++++++++++++----- 1 file changed, 21 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index ac1d5b0586ba..d1a121443bde 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -236,17 +236,33 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task) struct ds_context *context = *p_context; if (!context) { + spin_unlock(&ds_lock); + context = kzalloc(sizeof(*context), GFP_KERNEL); - if (!context) + if (!context) { + spin_lock(&ds_lock); return NULL; + } context->ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); if (!context->ds) { kfree(context); + spin_lock(&ds_lock); return NULL; } + spin_lock(&ds_lock); + /* + * Check for race - another CPU could have allocated + * it meanwhile: + */ + if (*p_context) { + kfree(context->ds); + kfree(context); + return *p_context; + } + *p_context = context; context->this = p_context; @@ -384,15 +400,15 @@ static int ds_request(struct task_struct *task, void *base, size_t size, spin_lock(&ds_lock); - error = -EPERM; - if (!check_tracer(task)) - goto out_unlock; - error = -ENOMEM; context = ds_alloc_context(task); if (!context) goto out_unlock; + error = -EPERM; + if (!check_tracer(task)) + goto out_unlock; + error = -EALREADY; if (context->owner[qual] == current) goto out_unlock; -- cgit v1.2.3 From e5e1f606ecbf67e52ebe2df5d14f8b94ec6544d0 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 17 Nov 2008 15:07:17 +0100 Subject: AMD IOMMU: add parameter to disable device isolation Impact: add a new AMD IOMMU kernel command line parameter Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu_init.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 0cdcda35a05f..838a2e1d5bb2 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -1213,6 +1213,8 @@ static int __init parse_amd_iommu_options(char *str) for (; *str; ++str) { if (strncmp(str, "isolate", 7) == 0) amd_iommu_isolate = 1; + if (strncmp(str, "share", 5) == 0) + amd_iommu_isolate = 0; if (strncmp(str, "fullflush", 11) == 0) amd_iommu_unmap_flush = true; } -- cgit v1.2.3 From 3ce1f93c6d53c3f91c3846cf66b018276c8ac2e7 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 17 Nov 2008 15:09:20 +0100 Subject: AMD IOMMU: enable device isolation per default Impact: makes device isolation the default for AMD IOMMU Some device drivers showed double-free bugs of DMA memory while testing them with AMD IOMMU. If all devices share the same protection domain this can lead to data corruption and data loss. Prevent this by putting each device into its own protection domain per default. Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu_init.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 838a2e1d5bb2..595edd2befc6 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -121,7 +121,7 @@ u16 amd_iommu_last_bdf; /* largest PCI device id we have LIST_HEAD(amd_iommu_unity_map); /* a list of required unity mappings we find in ACPI */ unsigned amd_iommu_aperture_order = 26; /* size of aperture in power of 2 */ -int amd_iommu_isolate; /* if 1, device isolation is enabled */ +int amd_iommu_isolate = 1; /* if 1, device isolation is enabled */ bool amd_iommu_unmap_flush; /* if true, flush on every unmap */ LIST_HEAD(amd_iommu_list); /* list of all AMD IOMMUs in the -- cgit v1.2.3 From 695b5676c727d80921a2dc8737d5b3322222db85 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 17 Nov 2008 15:16:43 +0100 Subject: AMD IOMMU: fix fullflush comparison length Impact: fix comparison length for 'fullflush' Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu_init.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 595edd2befc6..30ae2701b3df 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -1215,7 +1215,7 @@ static int __init parse_amd_iommu_options(char *str) amd_iommu_isolate = 1; if (strncmp(str, "share", 5) == 0) amd_iommu_isolate = 0; - if (strncmp(str, "fullflush", 11) == 0) + if (strncmp(str, "fullflush", 9) == 0) amd_iommu_unmap_flush = true; } -- cgit v1.2.3 From 8501c45cc32c311ae755a2d5ac8c4a5f04908d42 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 17 Nov 2008 19:11:46 +0100 Subject: AMD IOMMU: check for next_bit also in unmapped area Impact: fix possible use of stale IO/TLB entries Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 331b318304eb..e4899e0e8787 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -537,7 +537,7 @@ static void dma_ops_free_addresses(struct dma_ops_domain *dom, address >>= PAGE_SHIFT; iommu_area_free(dom->bitmap, address, pages); - if (address + pages >= dom->next_bit) + if (address >= dom->next_bit) dom->need_flush = true; } -- cgit v1.2.3 From 0af40a4b1050c050e62eb1dc30b82d5ab22bf221 Mon Sep 17 00:00:00 2001 From: Philipp Kohlbecher Date: Sun, 16 Nov 2008 12:11:01 +0100 Subject: x86: more general identifier for Phoenix BIOS Impact: widen the reach of the low-memory-protect DMI quirk Phoenix BIOSes variously identify their vendor as "Phoenix Technologies, LTD" or "Phoenix Technologies LTD" (without the comma.) This patch makes the identification string in the bad_bios_dmi_table more general (following a suggestion by Ingo Molnar), so that both versions are handled. Again, the patched file compiles cleanly and the patch has been tested successfully on my machine. Signed-off-by: Philipp Kohlbecher Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 0fa6790c1dd3..9d5674f7b6cc 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -764,7 +764,7 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = { .callback = dmi_low_memory_corruption, .ident = "Phoenix BIOS", .matches = { - DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies, LTD"), + DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies"), }, }, #endif -- cgit v1.2.3 From f632ddcc0786149c0e4bef9b6b44c96a75c0d074 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 18 Nov 2008 17:32:26 +0100 Subject: x86: fix wakeup_cpu with numaq/es7000, v2, fix #2 Impact: fix boot crash fix default_update_genapic(). Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index c366e891e10b..31328909456e 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -585,8 +585,10 @@ early_param("elfcorehdr", setup_elfcorehdr); static int __init default_update_genapic(void) { -#if defined(CONFIG_X86_GENERICARCH) || defined(CONFIG_X86_64) - genapic->wakeup_cpu = wakeup_secondary_cpu_via_nmi; +#ifdef CONFIG_X86_SMP +# if defined(CONFIG_X86_GENERICARCH) || defined(CONFIG_X86_64) + genapic->wakeup_cpu = wakeup_secondary_cpu_via_init; +# endif #endif return 0; -- cgit v1.2.3 From b5fe363b7d89577fcfda9b6cf0efc32760bbccc6 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 18 Nov 2008 08:14:14 -0800 Subject: x86: use update_genapic to get rid of ES7000_CLUSTERED_APIC v2 Impact: clean up We can autodetect those system that need cluster apic, and update genapic accordingly. We can also remove wakeup.h for e7000, because it's default one is now the same as overall default mach_wakecpu.h Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/es7000_32.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/es7000_32.c b/arch/x86/kernel/es7000_32.c index fb3bfe66fbe2..71d7be624d46 100644 --- a/arch/x86/kernel/es7000_32.c +++ b/arch/x86/kernel/es7000_32.c @@ -38,6 +38,7 @@ #include #include #include +#include #include #include #include @@ -163,7 +164,6 @@ es7000_rename_gsi(int ioapic, int gsi) return gsi; } -#ifdef CONFIG_ES7000_CLUSTERED_APIC static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip) { unsigned long vect = 0, psaival = 0; @@ -182,13 +182,24 @@ static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip) return 0; } +static void noop_wait_for_deassert(atomic_t *deassert_not_used) +{ +} + static int __init es7000_update_genapic(void) { genapic->wakeup_cpu = wakeup_secondary_cpu_via_mip; + /* MPENTIUMIII */ + if (boot_cpu_data.x86 == 6 && + (boot_cpu_data.x86_model >= 7 || boot_cpu_data.x86_model <= 11)) { + es7000_update_genapic_to_cluster(); + genapic->wait_for_init_deassert = noop_wait_for_deassert; + genapic->wakeup_cpu = wakeup_secondary_cpu_via_mip; + } + return 0; } -#endif void __init setup_unisys(void) @@ -206,9 +217,7 @@ setup_unisys(void) es7000_plat = ES7000_CLASSIC; ioapic_renumber_irq = es7000_rename_gsi; -#ifdef CONFIG_ES7000_CLUSTERED_APIC x86_quirks->update_genapic = es7000_update_genapic; -#endif } /* -- cgit v1.2.3 From 093bac154c142fa1fb31a3ac69ae1bc08930231b Mon Sep 17 00:00:00 2001 From: Steve Conklin Date: Fri, 14 Nov 2008 00:55:51 -0600 Subject: x86: quirk for reboot stalls on a Dell Optiplex 330 Dell Optiplex 330 appears to hang on reboot. This is resolved by adding a quirk to set bios reboot. Signed-off-by: Leann Ogasawara Signed-off-by: Steve Conklin Signed-off-by: Ingo Molnar --- arch/x86/kernel/reboot.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 724adfc63cb9..cc5a2545dd41 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -169,6 +169,15 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = { DMI_MATCH(DMI_BOARD_NAME, "0KW626"), }, }, + { /* Handle problems with rebooting on Dell Optiplex 330 with 0KP561 */ + .callback = set_bios_reboot, + .ident = "Dell OptiPlex 330", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 330"), + DMI_MATCH(DMI_BOARD_NAME, "0KP561"), + }, + }, { /* Handle problems with rebooting on Dell 2400's */ .callback = set_bios_reboot, .ident = "Dell PowerEdge 2400", -- cgit v1.2.3 From 77be80e437fec44f8b7a620314b7d7b605b8d93b Mon Sep 17 00:00:00 2001 From: "Richard A. Holden III" Date: Wed, 19 Nov 2008 16:05:14 -0700 Subject: x86: fix arch/x86/kernel/genx2apic_uv_x.c build warning when !CONFIG_HOTPLUG_CPU Impact: cleanup, reduce size of the kernel image a bit Fix: arch/x86/kernel/genx2apic_uv_x.c:403: warning: 'uv_heartbeat_disable' defined but not used the function is only used when CONFIG_HOTPLUG_CPU is defined. Signed-off-by: Richard A. Holden III Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index f02bbe5d0178..221299f4509f 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -400,6 +400,7 @@ static void __cpuinit uv_heartbeat_enable(int cpu) uv_heartbeat_enable(0); } +#ifdef CONFIG_HOTPLUG_CPU static void __cpuinit uv_heartbeat_disable(int cpu) { if (uv_cpu_hub_info(cpu)->scir.enabled) { @@ -409,7 +410,6 @@ static void __cpuinit uv_heartbeat_disable(int cpu) uv_set_cpu_scir_bits(cpu, 0xff); } -#ifdef CONFIG_HOTPLUG_CPU /* * cpu hotplug notifier */ -- cgit v1.2.3 From bb5574608a8375026510b4f983ffbb06ece33fe2 Mon Sep 17 00:00:00 2001 From: "Richard A. Holden III" Date: Wed, 19 Nov 2008 16:05:15 -0700 Subject: x86: fix arch/x86/kernel/setup.c build warning when !CONFIG_X86_RESERVE_LOW_64K Impact: cleanup Fix: arch/x86/kernel/setup.c:592: warning: 'dmi_low_memory_corruption' defined but not used this is only used if CONFIG_X86_RESERVE_LOW_64K is defined. Signed-off-by: Richard A. Holden III Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index e6c51433247d..13a5f592ac28 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -587,6 +587,7 @@ static struct x86_quirks default_x86_quirks __initdata; struct x86_quirks *x86_quirks __initdata = &default_x86_quirks; +#ifdef CONFIG_X86_RESERVE_LOW_64K static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) { printk(KERN_NOTICE @@ -598,6 +599,7 @@ static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) return 0; } +#endif /* List of systems that have known low memory corruption BIOS problems */ static struct dmi_system_id __initdata bad_bios_dmi_table[] = { -- cgit v1.2.3 From d99015b1abbad743aa049b439c1e1dede6d0fa49 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Wed, 19 Nov 2008 01:18:11 +0100 Subject: x86: move entry_64.S register saving out of the macros Here is a combined patch that moves "save_args" out-of-line for the interrupt macro and moves "error_entry" mostly out-of-line for the zeroentry and errorentry macros. The save_args function becomes really straightforward and easy to understand, with the possible exception of the stack switch code, which now needs to copy the return address of to the calling function. Normal interrupts arrive with ((~vector)-0x80) on the stack, which gets adjusted in common_interrupt: : (5) addq $0xffffffffffffff80,(%rsp) /* -> ~(vector) */ (4) sub $0x50,%rsp /* space for registers */ (5) callq ffffffff80211290 (5) callq ffffffff80214290 : ... An apic interrupt stub now look like this: : (5) pushq $0xffffffffffffff05 /* ~(vector) */ (4) sub $0x50,%rsp /* space for registers */ (5) callq ffffffff80211290 (5) callq ffffffff80212b8f (5) jmpq ffffffff80211f93 Similarly the exception handler register saving function becomes simpler, without the need of any parameter shuffling. The stub for an exception without errorcode looks like this: : (6) callq *0x1cad12(%rip) # ffffffff803dd448 (2) pushq $0xffffffffffffffff /* no syscall */ (4) sub $0x78,%rsp /* space for registers */ (5) callq ffffffff8030e3b0 (3) mov %rsp,%rdi /* pt_regs pointer */ (2) xor %esi,%esi /* no error code */ (5) callq ffffffff80213446 (5) jmpq ffffffff8030e460 And one for an exception with errorcode like this: : (6) callq *0x1cab92(%rip) # ffffffff803dd448 (4) sub $0x78,%rsp /* space for registers */ (5) callq ffffffff8030e3b0 (3) mov %rsp,%rdi /* pt_regs pointer */ (5) mov 0x78(%rsp),%rsi /* load error code */ (9) movq $0xffffffffffffffff,0x78(%rsp) /* no syscall */ (5) callq ffffffff80213209 (5) jmpq ffffffff8030e460 Unfortunately, this last type is more than 32 bytes. But the total space savings due to this patch is about 2500 bytes on an smp-configuration, and I think the code is clearer than it was before. The tested kernels were non-paravirt ones (i.e., without the indirect call at the top of the exception handlers). Anyhow, I tested this patch on top of a recent -tip. The machine was an 2x4-core Xeon at 2333MHz. Measured where the delays between (almost-)adjacent rdtsc instructions. The graphs show how much time is spent outside of the program as a function of the measured delay. The area under the graph represents the total time spent outside the program. Eight instances of the rdtsctest were started, each pinned to a single cpu. The histogams are added. For each kernel two measurements were done: one in mostly idle condition, the other while running "bonnie++ -f", bound to cpu 0. Each measurement took 40 minutes runtime. See the attached graphs for the results. The graphs overlap almost everywhere, but there are small differences. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 300 +++++++++++++++++++++++++-------------------- 1 file changed, 166 insertions(+), 134 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index dbf06a0ef3d5..5a12432ccdf9 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -242,6 +242,78 @@ ENTRY(native_usergs_sysret64) CFI_REL_OFFSET rsp,RSP /*CFI_REL_OFFSET ss,SS*/ .endm + +/* + * initial frame state for interrupts and exceptions + */ + .macro _frame ref + CFI_STARTPROC simple + CFI_SIGNAL_FRAME + CFI_DEF_CFA rsp,SS+8-\ref + /*CFI_REL_OFFSET ss,SS-\ref*/ + CFI_REL_OFFSET rsp,RSP-\ref + /*CFI_REL_OFFSET rflags,EFLAGS-\ref*/ + /*CFI_REL_OFFSET cs,CS-\ref*/ + CFI_REL_OFFSET rip,RIP-\ref + .endm + +/* + * initial frame state for interrupts (and exceptions without error code) + */ +#define INTR_FRAME _frame RIP +/* + * initial frame state for exceptions with error code (and interrupts + * with vector already pushed) + */ +#define XCPT_FRAME _frame ORIG_RAX + +/* save partial stack frame */ +ENTRY(save_args) + XCPT_FRAME + cld + movq %rdi, 8*8+16(%rsp) + CFI_REL_OFFSET rdi, 8*8+16 + movq %rsi, 7*8+16(%rsp) + CFI_REL_OFFSET rsi, 7*8+16 + movq %rdx, 6*8+16(%rsp) + CFI_REL_OFFSET rdx, 6*8+16 + movq %rcx, 5*8+16(%rsp) + CFI_REL_OFFSET rcx, 5*8+16 + movq %rax, 4*8+16(%rsp) + CFI_REL_OFFSET rax, 4*8+16 + movq %r8, 3*8+16(%rsp) + CFI_REL_OFFSET r8, 3*8+16 + movq %r9, 2*8+16(%rsp) + CFI_REL_OFFSET r9, 2*8+16 + movq %r10, 1*8+16(%rsp) + CFI_REL_OFFSET r10, 1*8+16 + movq %r11, 0*8+16(%rsp) + CFI_REL_OFFSET r11, 0*8+16 + leaq -ARGOFFSET+16(%rsp),%rdi /* arg1 for handler */ + movq %rbp, 8(%rsp) /* push %rbp */ + leaq 8(%rsp), %rbp /* mov %rsp, %ebp */ + testl $3, CS(%rdi) + je 1f + SWAPGS + /* + * irqcount is used to check if a CPU is already on an interrupt stack + * or not. While this is essentially redundant with preempt_count it is + * a little cheaper to use a separate counter in the PDA (short of + * moving irq_enter into assembly, which would be too much work) + */ +1: incl %gs:pda_irqcount + jne 2f + pop %rax /* move return address... */ + mov %gs:pda_irqstackptr,%rsp + push %rax /* ... to the new stack */ + /* + * We entered an interrupt context - irqs are off: + */ +2: TRACE_IRQS_OFF + ret + CFI_ENDPROC +END(save_args) + /* * A newly forked process directly context switches into this. */ @@ -607,26 +679,6 @@ ENTRY(stub_rt_sigreturn) CFI_ENDPROC END(stub_rt_sigreturn) -/* - * initial frame state for interrupts and exceptions - */ - .macro _frame ref - CFI_STARTPROC simple - CFI_SIGNAL_FRAME - CFI_DEF_CFA rsp,SS+8-\ref - /*CFI_REL_OFFSET ss,SS-\ref*/ - CFI_REL_OFFSET rsp,RSP-\ref - /*CFI_REL_OFFSET rflags,EFLAGS-\ref*/ - /*CFI_REL_OFFSET cs,CS-\ref*/ - CFI_REL_OFFSET rip,RIP-\ref - .endm - -/* initial frame state for interrupts (and exceptions without error code) */ -#define INTR_FRAME _frame RIP -/* initial frame state for exceptions with error code (and interrupts with - vector already pushed) */ -#define XCPT_FRAME _frame ORIG_RAX - /* * Build the entry stubs and pointer table with some assembler magic. * We pack 7 stubs into a single 32-byte chunk, which will fit in a @@ -667,46 +719,19 @@ END(irq_entries_start) END(interrupt) .previous -/* +/* * Interrupt entry/exit. * * Interrupt entry points save only callee clobbered registers in fast path. - * - * Entry runs with interrupts off. - */ + * + * Entry runs with interrupts off. + */ /* 0(%rsp): ~(interrupt number) */ .macro interrupt func - cld - SAVE_ARGS - leaq -ARGOFFSET(%rsp),%rdi /* arg1 for handler */ - pushq %rbp - /* - * Save rbp twice: One is for marking the stack frame, as usual, and the - * other, to fill pt_regs properly. This is because bx comes right - * before the last saved register in that structure, and not bp. If the - * base pointer were in the place bx is today, this would not be needed. - */ - movq %rbp, -8(%rsp) - CFI_ADJUST_CFA_OFFSET 8 - CFI_REL_OFFSET rbp, 0 - movq %rsp,%rbp - CFI_DEF_CFA_REGISTER rbp - testl $3,CS(%rdi) - je 1f - SWAPGS - /* irqcount is used to check if a CPU is already on an interrupt - stack or not. While this is essentially redundant with preempt_count - it is a little cheaper to use a separate counter in the PDA - (short of moving irq_enter into assembly, which would be too - much work) */ -1: incl %gs:pda_irqcount - cmoveq %gs:pda_irqstackptr,%rsp - push %rbp # backlink for old unwinder - /* - * We entered an interrupt context - irqs are off: - */ - TRACE_IRQS_OFF + subq $10*8, %rsp + CFI_ADJUST_CFA_OFFSET 10*8 + call save_args call \func .endm @@ -852,6 +877,8 @@ END(common_interrupt) /* * APIC interrupts. */ + .p2align 5 + .macro apicinterrupt num,func INTR_FRAME pushq $~(\num) @@ -922,24 +949,29 @@ END(spurious_interrupt) .macro zeroentry sym INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $0 /* push error code/oldrax */ + pushq $-1 /* ORIG_RAX: no syscall to restart */ CFI_ADJUST_CFA_OFFSET 8 - pushq %rax /* push real oldrax to the rdi slot */ - CFI_ADJUST_CFA_OFFSET 8 - CFI_REL_OFFSET rax,0 - leaq \sym(%rip),%rax - jmp error_entry + subq $15*8,%rsp + CFI_ADJUST_CFA_OFFSET 15*8 + call error_entry + movq %rsp,%rdi /* pt_regs pointer */ + xorl %esi,%esi /* no error code */ + call \sym + jmp error_exit /* %ebx: no swapgs flag */ CFI_ENDPROC .endm .macro errorentry sym XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq %rax - CFI_ADJUST_CFA_OFFSET 8 - CFI_REL_OFFSET rax,0 - leaq \sym(%rip),%rax - jmp error_entry + subq $15*8,%rsp + CFI_ADJUST_CFA_OFFSET 15*8 + call error_entry + movq %rsp,%rdi /* pt_regs pointer */ + movq ORIG_RAX(%rsp),%rsi /* get error code */ + movq $-1,ORIG_RAX(%rsp) /* no syscall to restart */ + call \sym + jmp error_exit /* %ebx: no swapgs flag */ CFI_ENDPROC .endm @@ -1043,93 +1075,93 @@ paranoid_schedule\trace: .endm /* - * Exception entry point. This expects an error code/orig_rax on the stack - * and the exception handler in %rax. + * Exception entry point. This expects an error code/orig_rax on the stack. + * returns in "no swapgs flag" in %ebx. */ KPROBE_ENTRY(error_entry) _frame RDI - CFI_REL_OFFSET rax,0 - /* rdi slot contains rax, oldrax contains error code */ + CFI_ADJUST_CFA_OFFSET 15*8 + /* oldrax contains error code */ cld - subq $14*8,%rsp - CFI_ADJUST_CFA_OFFSET (14*8) - movq %rsi,13*8(%rsp) - CFI_REL_OFFSET rsi,RSI - movq 14*8(%rsp),%rsi /* load rax from rdi slot */ - CFI_REGISTER rax,rsi - movq %rdx,12*8(%rsp) - CFI_REL_OFFSET rdx,RDX - movq %rcx,11*8(%rsp) - CFI_REL_OFFSET rcx,RCX - movq %rsi,10*8(%rsp) /* store rax */ - CFI_REL_OFFSET rax,RAX - movq %r8, 9*8(%rsp) - CFI_REL_OFFSET r8,R8 - movq %r9, 8*8(%rsp) - CFI_REL_OFFSET r9,R9 - movq %r10,7*8(%rsp) - CFI_REL_OFFSET r10,R10 - movq %r11,6*8(%rsp) - CFI_REL_OFFSET r11,R11 - movq %rbx,5*8(%rsp) - CFI_REL_OFFSET rbx,RBX - movq %rbp,4*8(%rsp) - CFI_REL_OFFSET rbp,RBP - movq %r12,3*8(%rsp) - CFI_REL_OFFSET r12,R12 - movq %r13,2*8(%rsp) - CFI_REL_OFFSET r13,R13 - movq %r14,1*8(%rsp) - CFI_REL_OFFSET r14,R14 - movq %r15,(%rsp) - CFI_REL_OFFSET r15,R15 + movq %rdi,14*8+8(%rsp) + CFI_REL_OFFSET rdi,RDI+8 + movq %rsi,13*8+8(%rsp) + CFI_REL_OFFSET rsi,RSI+8 + movq %rdx,12*8+8(%rsp) + CFI_REL_OFFSET rdx,RDX+8 + movq %rcx,11*8+8(%rsp) + CFI_REL_OFFSET rcx,RCX+8 + movq %rax,10*8+8(%rsp) + CFI_REL_OFFSET rax,RAX+8 + movq %r8, 9*8+8(%rsp) + CFI_REL_OFFSET r8,R8+8 + movq %r9, 8*8+8(%rsp) + CFI_REL_OFFSET r9,R9+8 + movq %r10,7*8+8(%rsp) + CFI_REL_OFFSET r10,R10+8 + movq %r11,6*8+8(%rsp) + CFI_REL_OFFSET r11,R11+8 + movq %rbx,5*8+8(%rsp) + CFI_REL_OFFSET rbx,RBX+8 + movq %rbp,4*8+8(%rsp) + CFI_REL_OFFSET rbp,RBP+8 + movq %r12,3*8+8(%rsp) + CFI_REL_OFFSET r12,R12+8 + movq %r13,2*8+8(%rsp) + CFI_REL_OFFSET r13,R13+8 + movq %r14,1*8+8(%rsp) + CFI_REL_OFFSET r14,R14+8 + movq %r15,0*8+8(%rsp) + CFI_REL_OFFSET r15,R15+8 xorl %ebx,%ebx - testl $3,CS(%rsp) - je error_kernelspace + testl $3,CS+8(%rsp) + je error_kernelspace error_swapgs: SWAPGS error_sti: TRACE_IRQS_OFF - movq %rdi,RDI(%rsp) - CFI_REL_OFFSET rdi,RDI - movq %rsp,%rdi - movq ORIG_RAX(%rsp),%rsi /* get error code */ - movq $-1,ORIG_RAX(%rsp) - call *%rax - /* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ -error_exit: + ret + CFI_ENDPROC + +/* + * There are two places in the kernel that can potentially fault with + * usergs. Handle them here. The exception handlers after iret run with + * kernel gs again, so don't set the user space flag. B stepping K8s + * sometimes report an truncated RIP for IRET exceptions returning to + * compat mode. Check for these here too. + */ +error_kernelspace: + incl %ebx + leaq irq_return(%rip),%rcx + cmpq %rcx,RIP+8(%rsp) + je error_swapgs + movl %ecx,%ecx /* zero extend */ + cmpq %rcx,RIP+8(%rsp) + je error_swapgs + cmpq $gs_change,RIP+8(%rsp) + je error_swapgs + jmp error_sti +KPROBE_END(error_entry) + + +/* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ +KPROBE_ENTRY(error_exit) + _frame R15 movl %ebx,%eax RESTORE_REST DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF GET_THREAD_INFO(%rcx) testl %eax,%eax - jne retint_kernel + jne retint_kernel LOCKDEP_SYS_EXIT_IRQ - movl TI_flags(%rcx),%edx - movl $_TIF_WORK_MASK,%edi - andl %edi,%edx - jnz retint_careful + movl TI_flags(%rcx),%edx + movl $_TIF_WORK_MASK,%edi + andl %edi,%edx + jnz retint_careful jmp retint_swapgs CFI_ENDPROC - -error_kernelspace: - incl %ebx - /* There are two places in the kernel that can potentially fault with - usergs. Handle them here. The exception handlers after - iret run with kernel gs again, so don't set the user space flag. - B stepping K8s sometimes report an truncated RIP for IRET - exceptions returning to compat mode. Check for these here too. */ - leaq irq_return(%rip),%rcx - cmpq %rcx,RIP(%rsp) - je error_swapgs - movl %ecx,%ecx /* zero extend */ - cmpq %rcx,RIP(%rsp) - je error_swapgs - cmpq $gs_change,RIP(%rsp) - je error_swapgs - jmp error_sti -KPROBE_END(error_entry) +KPROBE_END(error_exit) /* Reload gs selector with exception handling */ /* edi: new selector */ -- cgit v1.2.3 From 9bc646f163b136684390081262fab0fd8f5343ca Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Thu, 20 Nov 2008 19:08:45 +0600 Subject: x86: fix __cpuinit/__init tangle in init_thread_xstate() Impact: fix incorrect __init annotation This patch removes the following section mismatch warning. A patch set was send previously (http://lkml.org/lkml/2008/11/10/407). But introduce some other problem, reported by Rufus (http://lkml.org/lkml/2008/11/11/46). Then Ingo Molnar suggest that, it's best to remove __init from xsave_cntxt_init(void). Which is the second patch in this series. Now, this one removes the following warning. WARNING: arch/x86/kernel/built-in.o(.cpuinit.text+0x2237): Section mismatch in reference from the function cpu_init() to the function .init.text:init_thread_xstate() The function __cpuinit cpu_init() references a function __init init_thread_xstate(). If init_thread_xstate is only used by cpu_init then annotate init_thread_xstate with a matching annotation. Signed-off-by: Rakib Mullick Signed-off-by: Ingo Molnar --- arch/x86/kernel/i387.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c index 1f20608d4ca8..b0f61f0dcd0a 100644 --- a/arch/x86/kernel/i387.c +++ b/arch/x86/kernel/i387.c @@ -58,7 +58,7 @@ void __cpuinit mxcsr_feature_mask_init(void) stts(); } -void __init init_thread_xstate(void) +void __cpuinit init_thread_xstate(void) { if (!HAVE_HWFP) { xstate_size = sizeof(struct i387_soft_struct); -- cgit v1.2.3 From bfe085f62f98a49e1b864e4950389c7205174e4f Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Thu, 20 Nov 2008 19:12:50 +0600 Subject: x86: fixing __cpuinit/__init tangle, xsave_cntxt_init() Annotate xsave_cntxt_init() as "can be called outside of __init". Signed-off-by: Rakib Mullick Signed-off-by: Ingo Molnar --- arch/x86/kernel/xsave.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c index b13acb75e822..15c3e6999182 100644 --- a/arch/x86/kernel/xsave.c +++ b/arch/x86/kernel/xsave.c @@ -310,7 +310,7 @@ static void __init setup_xstate_init(void) /* * Enable and initialize the xsave feature. */ -void __init xsave_cntxt_init(void) +void __ref xsave_cntxt_init(void) { unsigned int eax, ebx, ecx, edx; -- cgit v1.2.3 From dcd072e26055de600cecdc3f7a1e083ecd55c2e4 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Thu, 20 Nov 2008 14:40:11 +0100 Subject: x86: clean up after: move entry_64.S register saving out of the macros This add-on patch to x86: move entry_64.S register saving out of the macros visually cleans up the appearance of the code by introducing some basic helper macro's. It also adds some cfi annotations which were missing. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 220 +++++++++++++++++++++++---------------------- 1 file changed, 112 insertions(+), 108 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 5a12432ccdf9..7a04f696121d 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -60,6 +60,23 @@ #define __AUDIT_ARCH_LE 0x40000000 .code64 +/* + * Some macro's to hide the most frequently occuring CFI annotations. + */ + .macro CFI_PUSHQ reg + pushq \reg + CFI_ADJUST_CFA_OFFSET 8 + .endm + + .macro CFI_POPQ reg + popq \reg + CFI_ADJUST_CFA_OFFSET -8 + .endm + + .macro CFI_MOVQ reg offset=0 + movq %\reg, \offset(%rsp) + CFI_REL_OFFSET \reg, \offset + .endm #ifdef CONFIG_FUNCTION_TRACER #ifdef CONFIG_DYNAMIC_FTRACE @@ -213,84 +230,84 @@ ENTRY(native_usergs_sysret64) CFI_ADJUST_CFA_OFFSET -(6*8) .endm - .macro CFI_DEFAULT_STACK start=1 +/* + * initial frame state for interrupts (and exceptions without error code) + */ + .macro EMPTY_FRAME start=1 offset=0 .if \start - CFI_STARTPROC simple + CFI_STARTPROC simple CFI_SIGNAL_FRAME - CFI_DEF_CFA rsp,SS+8 + CFI_DEF_CFA rsp,8+\offset .else - CFI_DEF_CFA_OFFSET SS+8 + CFI_DEF_CFA_OFFSET 8+\offset .endif - CFI_REL_OFFSET r15,R15 - CFI_REL_OFFSET r14,R14 - CFI_REL_OFFSET r13,R13 - CFI_REL_OFFSET r12,R12 - CFI_REL_OFFSET rbp,RBP - CFI_REL_OFFSET rbx,RBX - CFI_REL_OFFSET r11,R11 - CFI_REL_OFFSET r10,R10 - CFI_REL_OFFSET r9,R9 - CFI_REL_OFFSET r8,R8 - CFI_REL_OFFSET rax,RAX - CFI_REL_OFFSET rcx,RCX - CFI_REL_OFFSET rdx,RDX - CFI_REL_OFFSET rsi,RSI - CFI_REL_OFFSET rdi,RDI - CFI_REL_OFFSET rip,RIP - /*CFI_REL_OFFSET cs,CS*/ - /*CFI_REL_OFFSET rflags,EFLAGS*/ - CFI_REL_OFFSET rsp,RSP - /*CFI_REL_OFFSET ss,SS*/ .endm /* - * initial frame state for interrupts and exceptions + * initial frame state for interrupts (and exceptions without error code) */ - .macro _frame ref - CFI_STARTPROC simple - CFI_SIGNAL_FRAME - CFI_DEF_CFA rsp,SS+8-\ref - /*CFI_REL_OFFSET ss,SS-\ref*/ - CFI_REL_OFFSET rsp,RSP-\ref - /*CFI_REL_OFFSET rflags,EFLAGS-\ref*/ - /*CFI_REL_OFFSET cs,CS-\ref*/ - CFI_REL_OFFSET rip,RIP-\ref + .macro INTR_FRAME start=1 offset=0 + EMPTY_FRAME \start, (SS+8-RIP)+\offset + /*CFI_REL_OFFSET ss, SS-RIP+\offset*/ + CFI_REL_OFFSET rsp, RSP-RIP+\offset + /*CFI_REL_OFFSET rflags, EFLAGS-RIP+\offset*/ + /*CFI_REL_OFFSET cs, CS-RIP+\offset*/ + CFI_REL_OFFSET rip, RIP-RIP+\offset .endm -/* - * initial frame state for interrupts (and exceptions without error code) - */ -#define INTR_FRAME _frame RIP /* * initial frame state for exceptions with error code (and interrupts * with vector already pushed) */ -#define XCPT_FRAME _frame ORIG_RAX + .macro XCPT_FRAME start=1 offset=0 + INTR_FRAME \start, (RIP-ORIG_RAX)+\offset + /*CFI_REL_OFFSET orig_rax, ORIG_RAX-ORIG_RAX*/ + .endm + +/* + * frame that enables calling into C. + */ + .macro PARTIAL_FRAME start=1 offset=0 + XCPT_FRAME \start, (ORIG_RAX-ARGOFFSET)+\offset + CFI_REL_OFFSET rdi, (RDI-ARGOFFSET)+\offset + CFI_REL_OFFSET rsi, (RSI-ARGOFFSET)+\offset + CFI_REL_OFFSET rdx, (RDX-ARGOFFSET)+\offset + CFI_REL_OFFSET rcx, (RCX-ARGOFFSET)+\offset + CFI_REL_OFFSET rax, (RAX-ARGOFFSET)+\offset + CFI_REL_OFFSET r8, (R8-ARGOFFSET)+\offset + CFI_REL_OFFSET r9, (R9-ARGOFFSET)+\offset + CFI_REL_OFFSET r10, (R10-ARGOFFSET)+\offset + CFI_REL_OFFSET r11, (R11-ARGOFFSET)+\offset + .endm + +/* + * frame that enables passing a complete pt_regs to a C function. + */ + .macro DEFAULT_FRAME start=1 offset=0 + PARTIAL_FRAME \start, (R11-R15)+\offset + CFI_REL_OFFSET rbx, RBX+\offset + CFI_REL_OFFSET rbp, RBP+\offset + CFI_REL_OFFSET r12, R12+\offset + CFI_REL_OFFSET r13, R13+\offset + CFI_REL_OFFSET r14, R14+\offset + CFI_REL_OFFSET r15, R15+\offset + .endm /* save partial stack frame */ ENTRY(save_args) XCPT_FRAME cld - movq %rdi, 8*8+16(%rsp) - CFI_REL_OFFSET rdi, 8*8+16 - movq %rsi, 7*8+16(%rsp) - CFI_REL_OFFSET rsi, 7*8+16 - movq %rdx, 6*8+16(%rsp) - CFI_REL_OFFSET rdx, 6*8+16 - movq %rcx, 5*8+16(%rsp) - CFI_REL_OFFSET rcx, 5*8+16 - movq %rax, 4*8+16(%rsp) - CFI_REL_OFFSET rax, 4*8+16 - movq %r8, 3*8+16(%rsp) - CFI_REL_OFFSET r8, 3*8+16 - movq %r9, 2*8+16(%rsp) - CFI_REL_OFFSET r9, 2*8+16 - movq %r10, 1*8+16(%rsp) - CFI_REL_OFFSET r10, 1*8+16 - movq %r11, 0*8+16(%rsp) - CFI_REL_OFFSET r11, 0*8+16 + CFI_MOVQ rdi, (RDI-ARGOFFSET)+16 + CFI_MOVQ rsi, (RSI-ARGOFFSET)+16 + CFI_MOVQ rdx, (RDX-ARGOFFSET)+16 + CFI_MOVQ rcx, (RCX-ARGOFFSET)+16 + CFI_MOVQ rax, (RAX-ARGOFFSET)+16 + CFI_MOVQ r8, (R8-ARGOFFSET)+16 + CFI_MOVQ r9, (R9-ARGOFFSET)+16 + CFI_MOVQ r10, (R10-ARGOFFSET)+16 + CFI_MOVQ r11, (R11-ARGOFFSET)+16 leaq -ARGOFFSET+16(%rsp),%rdi /* arg1 for handler */ - movq %rbp, 8(%rsp) /* push %rbp */ + CFI_MOVQ rbp, 8 /* push %rbp */ leaq 8(%rsp), %rbp /* mov %rsp, %ebp */ testl $3, CS(%rdi) je 1f @@ -303,9 +320,10 @@ ENTRY(save_args) */ 1: incl %gs:pda_irqcount jne 2f - pop %rax /* move return address... */ + CFI_POPQ %rax /* move return address... */ mov %gs:pda_irqstackptr,%rsp - push %rax /* ... to the new stack */ + EMPTY_FRAME 0 + CFI_PUSHQ %rax /* ... to the new stack */ /* * We entered an interrupt context - irqs are off: */ @@ -319,7 +337,7 @@ END(save_args) */ /* rdi: prev */ ENTRY(ret_from_fork) - CFI_DEFAULT_STACK + DEFAULT_FRAME push kernel_eflags(%rip) CFI_ADJUST_CFA_OFFSET 8 popf # reset kernel eflags @@ -732,6 +750,7 @@ END(interrupt) subq $10*8, %rsp CFI_ADJUST_CFA_OFFSET 10*8 call save_args + PARTIAL_FRAME 0 call \func .endm @@ -949,11 +968,11 @@ END(spurious_interrupt) .macro zeroentry sym INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $-1 /* ORIG_RAX: no syscall to restart */ - CFI_ADJUST_CFA_OFFSET 8 + CFI_PUSHQ $-1 /* ORIG_RAX: no syscall to restart */ subq $15*8,%rsp CFI_ADJUST_CFA_OFFSET 15*8 call error_entry + DEFAULT_FRAME 0 movq %rsp,%rdi /* pt_regs pointer */ xorl %esi,%esi /* no error code */ call \sym @@ -967,6 +986,7 @@ END(spurious_interrupt) subq $15*8,%rsp CFI_ADJUST_CFA_OFFSET 15*8 call error_entry + DEFAULT_FRAME 0 movq %rsp,%rdi /* pt_regs pointer */ movq ORIG_RAX(%rsp),%rsi /* get error code */ movq $-1,ORIG_RAX(%rsp) /* no syscall to restart */ @@ -1079,40 +1099,25 @@ paranoid_schedule\trace: * returns in "no swapgs flag" in %ebx. */ KPROBE_ENTRY(error_entry) - _frame RDI + XCPT_FRAME CFI_ADJUST_CFA_OFFSET 15*8 /* oldrax contains error code */ cld - movq %rdi,14*8+8(%rsp) - CFI_REL_OFFSET rdi,RDI+8 - movq %rsi,13*8+8(%rsp) - CFI_REL_OFFSET rsi,RSI+8 - movq %rdx,12*8+8(%rsp) - CFI_REL_OFFSET rdx,RDX+8 - movq %rcx,11*8+8(%rsp) - CFI_REL_OFFSET rcx,RCX+8 - movq %rax,10*8+8(%rsp) - CFI_REL_OFFSET rax,RAX+8 - movq %r8, 9*8+8(%rsp) - CFI_REL_OFFSET r8,R8+8 - movq %r9, 8*8+8(%rsp) - CFI_REL_OFFSET r9,R9+8 - movq %r10,7*8+8(%rsp) - CFI_REL_OFFSET r10,R10+8 - movq %r11,6*8+8(%rsp) - CFI_REL_OFFSET r11,R11+8 - movq %rbx,5*8+8(%rsp) - CFI_REL_OFFSET rbx,RBX+8 - movq %rbp,4*8+8(%rsp) - CFI_REL_OFFSET rbp,RBP+8 - movq %r12,3*8+8(%rsp) - CFI_REL_OFFSET r12,R12+8 - movq %r13,2*8+8(%rsp) - CFI_REL_OFFSET r13,R13+8 - movq %r14,1*8+8(%rsp) - CFI_REL_OFFSET r14,R14+8 - movq %r15,0*8+8(%rsp) - CFI_REL_OFFSET r15,R15+8 + CFI_MOVQ rdi, RDI+8 + CFI_MOVQ rsi, RSI+8 + CFI_MOVQ rdx, RDX+8 + CFI_MOVQ rcx, RCX+8 + CFI_MOVQ rax, RAX+8 + CFI_MOVQ r8, R8+8 + CFI_MOVQ r9, R9+8 + CFI_MOVQ r10, R10+8 + CFI_MOVQ r11, R11+8 + CFI_MOVQ rbx, RBX+8 + CFI_MOVQ rbp, RBP+8 + CFI_MOVQ r12, R12+8 + CFI_MOVQ r13, R13+8 + CFI_MOVQ r14, R14+8 + CFI_MOVQ r15, R15+8 xorl %ebx,%ebx testl $3,CS+8(%rsp) je error_kernelspace @@ -1146,7 +1151,7 @@ KPROBE_END(error_entry) /* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ KPROBE_ENTRY(error_exit) - _frame R15 + DEFAULT_FRAME movl %ebx,%eax RESTORE_REST DISABLE_INTERRUPTS(CLBR_NONE) @@ -1455,7 +1460,7 @@ ENTRY(xen_do_hypervisor_callback) # do_hypervisor_callback(struct *pt_regs) see the correct pointer to the pt_regs */ movq %rdi, %rsp # we don't return, adjust the stack frame CFI_ENDPROC - CFI_DEFAULT_STACK + DEFAULT_FRAME 11: incl %gs:pda_irqcount movq %rsp,%rbp CFI_DEF_CFA_REGISTER rbp @@ -1483,10 +1488,13 @@ END(do_hypervisor_callback) # with its current contents: any discrepancy means we in category 1. */ ENTRY(xen_failsafe_callback) - framesz = (RIP-0x30) /* workaround buggy gas */ - _frame framesz - CFI_REL_OFFSET rcx, 0 - CFI_REL_OFFSET r11, 8 + INTR_FRAME 1 (6*8) + /*CFI_REL_OFFSET gs,GS*/ + /*CFI_REL_OFFSET fs,FS*/ + /*CFI_REL_OFFSET es,ES*/ + /*CFI_REL_OFFSET ds,DS*/ + CFI_REL_OFFSET r11,8 + CFI_REL_OFFSET rcx,0 movw %ds,%cx cmpw %cx,0x10(%rsp) CFI_REMEMBER_STATE @@ -1507,12 +1515,9 @@ ENTRY(xen_failsafe_callback) CFI_RESTORE r11 addq $0x30,%rsp CFI_ADJUST_CFA_OFFSET -0x30 - pushq $0 - CFI_ADJUST_CFA_OFFSET 8 - pushq %r11 - CFI_ADJUST_CFA_OFFSET 8 - pushq %rcx - CFI_ADJUST_CFA_OFFSET 8 + CFI_PUSHQ $0 /* RIP */ + CFI_PUSHQ %r11 + CFI_PUSHQ %rcx jmp general_protection CFI_RESTORE_STATE 1: /* Segment mismatch => Category 1 (Bad segment). Retry the IRET. */ @@ -1522,8 +1527,7 @@ ENTRY(xen_failsafe_callback) CFI_RESTORE r11 addq $0x30,%rsp CFI_ADJUST_CFA_OFFSET -0x30 - pushq $0 - CFI_ADJUST_CFA_OFFSET 8 + CFI_PUSHQ $0 SAVE_ALL jmp error_exit CFI_ENDPROC -- cgit v1.2.3 From 0ca4b6b00113b064c080d26d803d0d7c80fb5dc8 Mon Sep 17 00:00:00 2001 From: Matthew Wilcox Date: Thu, 20 Nov 2008 14:09:33 -0700 Subject: x86: Fix interrupt leak due to migration When we migrate an interrupt from one CPU to another, we set the move_in_progress flag and clean up the vectors later once they're not being used. If you're unlucky and call destroy_irq() before the vectors become un-used, the move_in_progress flag is never cleared, which causes the interrupt to become unusable. This was discovered by Jesse Brandeburg for whom it manifested as an MSI-X device refusing to use MSI-X mode when the driver was unloaded and reloaded repeatedly. Signed-off-by: Matthew Wilcox Signed-off-by: Linus Torvalds --- arch/x86/kernel/io_apic.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 7a3f2028e2eb..c9513e1ff28d 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -1140,6 +1140,20 @@ static void __clear_irq_vector(int irq) cfg->vector = 0; cpus_clear(cfg->domain); + + if (likely(!cfg->move_in_progress)) + return; + cpus_and(mask, cfg->old_domain, cpu_online_map); + for_each_cpu_mask_nr(cpu, mask) { + for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; + vector++) { + if (per_cpu(vector_irq, cpu)[vector] != irq) + continue; + per_cpu(vector_irq, cpu)[vector] = -1; + break; + } + } + cfg->move_in_progress = 0; } void __setup_vector_irq(int cpu) -- cgit v1.2.3 From 3ddd972d970fdabbe6515aa2f95e0ef2c8df903d Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 20 Nov 2008 18:32:17 -0800 Subject: x86: signal: rename COPY_SEG_STRICT to COPY_SEG_CPL3 Impact: cleanup Rename macro COPY_SEG_STRICT to COPY_SEG_CPL3, as suggested by hpa. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 8 ++++---- arch/x86/kernel/signal_64.c | 8 ++++---- 2 files changed, 8 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 514171ac0d03..c2aabeba27a5 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -145,7 +145,7 @@ asmlinkage int sys_sigaltstack(unsigned long bx) regs->seg = tmp; \ } -#define COPY_SEG_STRICT(seg) { \ +#define COPY_SEG_CPL3(seg) { \ unsigned short tmp; \ err |= __get_user(tmp, &sc->seg); \ regs->seg = tmp | 3; \ @@ -193,13 +193,13 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, #endif /* CONFIG_X86_64 */ #ifdef CONFIG_X86_32 - COPY_SEG_STRICT(cs); - COPY_SEG_STRICT(ss); + COPY_SEG_CPL3(cs); + COPY_SEG_CPL3(ss); #else /* !CONFIG_X86_32 */ /* Kernel saves and restores only the CS segment register on signals, * which is the bare minimum needed to allow mixed 32/64-bit code. * App's signal handler can save/restore other segments if needed. */ - COPY_SEG_STRICT(cs); + COPY_SEG_CPL3(cs); #endif /* CONFIG_X86_32 */ err |= __get_user(tmpflags, &sc->flags); diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index d2307e41fbdb..3d54d366ccb2 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -56,7 +56,7 @@ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, err |= __get_user(regs->x, &sc->x); \ } -#define COPY_SEG_STRICT(seg) { \ +#define COPY_SEG_CPL3(seg) { \ unsigned short tmp; \ err |= __get_user(tmp, &sc->seg); \ regs->seg = tmp | 3; \ @@ -98,13 +98,13 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, #endif /* CONFIG_X86_64 */ #ifdef CONFIG_X86_32 - COPY_SEG_STRICT(cs); - COPY_SEG_STRICT(ss); + COPY_SEG_CPL3(cs); + COPY_SEG_CPL3(ss); #else /* !CONFIG_X86_32 */ /* Kernel saves and restores only the CS segment register on signals, * which is the bare minimum needed to allow mixed 32/64-bit code. * App's signal handler can save/restore other segments if needed. */ - COPY_SEG_STRICT(cs); + COPY_SEG_CPL3(cs); #endif /* CONFIG_X86_32 */ err |= __get_user(tmpflags, &sc->flags); -- cgit v1.2.3 From e8a0e27662186f8856a0a6242e7a8386c9a64a53 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 21 Nov 2008 15:11:32 +0100 Subject: x86: clean up after: move entry_64.S register saving out of the macros, fix Impact: build fix The break builds with older binutils (2.16.1): arch/x86/kernel/entry_64.S: Assembler messages: arch/x86/kernel/entry_64.S:282: Error: too many positional arguments arch/x86/kernel/entry_64.S:283: Error: too many positional arguments arch/x86/kernel/entry_64.S:284: Error: too many positional arguments arch/x86/kernel/entry_64.S:285: Error: too many positional arguments arch/x86/kernel/entry_64.S:286: Error: too many positional arguments arch/x86/kernel/entry_64.S:287: Error: too many positional arguments arch/x86/kernel/entry_64.S:288: Error: too many positional arguments arch/x86/kernel/entry_64.S:289: Error: too many positional arguments arch/x86/kernel/entry_64.S:290: Error: too many positional arguments Took some time to figure out the detail that GAS chokes on: it's negative offsets. Rearrange the calculations to make sure we never go negative. Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 54 +++++++++++++++++++++++----------------------- 1 file changed, 27 insertions(+), 27 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 7a04f696121d..4e3d83678f85 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -247,12 +247,12 @@ ENTRY(native_usergs_sysret64) * initial frame state for interrupts (and exceptions without error code) */ .macro INTR_FRAME start=1 offset=0 - EMPTY_FRAME \start, (SS+8-RIP)+\offset - /*CFI_REL_OFFSET ss, SS-RIP+\offset*/ - CFI_REL_OFFSET rsp, RSP-RIP+\offset - /*CFI_REL_OFFSET rflags, EFLAGS-RIP+\offset*/ - /*CFI_REL_OFFSET cs, CS-RIP+\offset*/ - CFI_REL_OFFSET rip, RIP-RIP+\offset + EMPTY_FRAME \start, SS+8+\offset-RIP + /*CFI_REL_OFFSET ss, SS+\offset-RIP*/ + CFI_REL_OFFSET rsp, RSP+\offset-RIP + /*CFI_REL_OFFSET rflags, EFLAGS+\offset-RIP*/ + /*CFI_REL_OFFSET cs, CS+\offset-RIP*/ + CFI_REL_OFFSET rip, RIP+\offset-RIP .endm /* @@ -260,7 +260,7 @@ ENTRY(native_usergs_sysret64) * with vector already pushed) */ .macro XCPT_FRAME start=1 offset=0 - INTR_FRAME \start, (RIP-ORIG_RAX)+\offset + INTR_FRAME \start, RIP+\offset-ORIG_RAX /*CFI_REL_OFFSET orig_rax, ORIG_RAX-ORIG_RAX*/ .endm @@ -268,23 +268,23 @@ ENTRY(native_usergs_sysret64) * frame that enables calling into C. */ .macro PARTIAL_FRAME start=1 offset=0 - XCPT_FRAME \start, (ORIG_RAX-ARGOFFSET)+\offset - CFI_REL_OFFSET rdi, (RDI-ARGOFFSET)+\offset - CFI_REL_OFFSET rsi, (RSI-ARGOFFSET)+\offset - CFI_REL_OFFSET rdx, (RDX-ARGOFFSET)+\offset - CFI_REL_OFFSET rcx, (RCX-ARGOFFSET)+\offset - CFI_REL_OFFSET rax, (RAX-ARGOFFSET)+\offset - CFI_REL_OFFSET r8, (R8-ARGOFFSET)+\offset - CFI_REL_OFFSET r9, (R9-ARGOFFSET)+\offset - CFI_REL_OFFSET r10, (R10-ARGOFFSET)+\offset - CFI_REL_OFFSET r11, (R11-ARGOFFSET)+\offset + XCPT_FRAME \start, ORIG_RAX+\offset-ARGOFFSET + CFI_REL_OFFSET rdi, RDI+\offset-ARGOFFSET + CFI_REL_OFFSET rsi, RSI+\offset-ARGOFFSET + CFI_REL_OFFSET rdx, RDX+\offset-ARGOFFSET + CFI_REL_OFFSET rcx, RCX+\offset-ARGOFFSET + CFI_REL_OFFSET rax, RAX+\offset-ARGOFFSET + CFI_REL_OFFSET r8, R8+\offset-ARGOFFSET + CFI_REL_OFFSET r9, R9+\offset-ARGOFFSET + CFI_REL_OFFSET r10, R10+\offset-ARGOFFSET + CFI_REL_OFFSET r11, R11+\offset-ARGOFFSET .endm /* * frame that enables passing a complete pt_regs to a C function. */ .macro DEFAULT_FRAME start=1 offset=0 - PARTIAL_FRAME \start, (R11-R15)+\offset + PARTIAL_FRAME \start, R11+\offset-R15 CFI_REL_OFFSET rbx, RBX+\offset CFI_REL_OFFSET rbp, RBP+\offset CFI_REL_OFFSET r12, R12+\offset @@ -297,15 +297,15 @@ ENTRY(native_usergs_sysret64) ENTRY(save_args) XCPT_FRAME cld - CFI_MOVQ rdi, (RDI-ARGOFFSET)+16 - CFI_MOVQ rsi, (RSI-ARGOFFSET)+16 - CFI_MOVQ rdx, (RDX-ARGOFFSET)+16 - CFI_MOVQ rcx, (RCX-ARGOFFSET)+16 - CFI_MOVQ rax, (RAX-ARGOFFSET)+16 - CFI_MOVQ r8, (R8-ARGOFFSET)+16 - CFI_MOVQ r9, (R9-ARGOFFSET)+16 - CFI_MOVQ r10, (R10-ARGOFFSET)+16 - CFI_MOVQ r11, (R11-ARGOFFSET)+16 + CFI_MOVQ rdi, RDI+16-ARGOFFSET + CFI_MOVQ rsi, RSI+16-ARGOFFSET + CFI_MOVQ rdx, RDX+16-ARGOFFSET + CFI_MOVQ rcx, RCX+16-ARGOFFSET + CFI_MOVQ rax, RAX+16-ARGOFFSET + CFI_MOVQ r8, R8+16-ARGOFFSET + CFI_MOVQ r9, R9+16-ARGOFFSET + CFI_MOVQ r10, R10+16-ARGOFFSET + CFI_MOVQ r11, R11+16-ARGOFFSET leaq -ARGOFFSET+16(%rsp),%rdi /* arg1 for handler */ CFI_MOVQ rbp, 8 /* push %rbp */ leaq 8(%rsp), %rbp /* mov %rsp, %ebp */ -- cgit v1.2.3 From 14ae22ba2b8bb3d53fb795f9b8074aa39ef7b6cd Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 21 Nov 2008 15:20:47 +0100 Subject: x86: entry_64.S: rename Impact: cleanup Rename: CFI_PUSHQ => pushq_cfi CFI_POPQ => popq_cfi CFI_MOVQ => movq_cfi To make it blend better into regular assembly code. Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 71 +++++++++++++++++++++++----------------------- 1 file changed, 36 insertions(+), 35 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 4e3d83678f85..92c5e18340db 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -63,17 +63,17 @@ /* * Some macro's to hide the most frequently occuring CFI annotations. */ - .macro CFI_PUSHQ reg + .macro pushq_cfi reg pushq \reg CFI_ADJUST_CFA_OFFSET 8 .endm - .macro CFI_POPQ reg + .macro popq_cfi reg popq \reg CFI_ADJUST_CFA_OFFSET -8 .endm - .macro CFI_MOVQ reg offset=0 + .macro movq_cfi reg offset=0 movq %\reg, \offset(%rsp) CFI_REL_OFFSET \reg, \offset .endm @@ -297,17 +297,18 @@ ENTRY(native_usergs_sysret64) ENTRY(save_args) XCPT_FRAME cld - CFI_MOVQ rdi, RDI+16-ARGOFFSET - CFI_MOVQ rsi, RSI+16-ARGOFFSET - CFI_MOVQ rdx, RDX+16-ARGOFFSET - CFI_MOVQ rcx, RCX+16-ARGOFFSET - CFI_MOVQ rax, RAX+16-ARGOFFSET - CFI_MOVQ r8, R8+16-ARGOFFSET - CFI_MOVQ r9, R9+16-ARGOFFSET - CFI_MOVQ r10, R10+16-ARGOFFSET - CFI_MOVQ r11, R11+16-ARGOFFSET + movq_cfi rdi, RDI+16-ARGOFFSET + movq_cfi rsi, RSI+16-ARGOFFSET + movq_cfi rdx, RDX+16-ARGOFFSET + movq_cfi rcx, RCX+16-ARGOFFSET + movq_cfi rax, RAX+16-ARGOFFSET + movq_cfi r8, R8+16-ARGOFFSET + movq_cfi r9, R9+16-ARGOFFSET + movq_cfi r10, R10+16-ARGOFFSET + movq_cfi r11, R11+16-ARGOFFSET + leaq -ARGOFFSET+16(%rsp),%rdi /* arg1 for handler */ - CFI_MOVQ rbp, 8 /* push %rbp */ + movq_cfi rbp, 8 /* push %rbp */ leaq 8(%rsp), %rbp /* mov %rsp, %ebp */ testl $3, CS(%rdi) je 1f @@ -320,10 +321,10 @@ ENTRY(save_args) */ 1: incl %gs:pda_irqcount jne 2f - CFI_POPQ %rax /* move return address... */ + popq_cfi %rax /* move return address... */ mov %gs:pda_irqstackptr,%rsp EMPTY_FRAME 0 - CFI_PUSHQ %rax /* ... to the new stack */ + pushq_cfi %rax /* ... to the new stack */ /* * We entered an interrupt context - irqs are off: */ @@ -968,7 +969,7 @@ END(spurious_interrupt) .macro zeroentry sym INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - CFI_PUSHQ $-1 /* ORIG_RAX: no syscall to restart */ + pushq_cfi $-1 /* ORIG_RAX: no syscall to restart */ subq $15*8,%rsp CFI_ADJUST_CFA_OFFSET 15*8 call error_entry @@ -1103,21 +1104,21 @@ KPROBE_ENTRY(error_entry) CFI_ADJUST_CFA_OFFSET 15*8 /* oldrax contains error code */ cld - CFI_MOVQ rdi, RDI+8 - CFI_MOVQ rsi, RSI+8 - CFI_MOVQ rdx, RDX+8 - CFI_MOVQ rcx, RCX+8 - CFI_MOVQ rax, RAX+8 - CFI_MOVQ r8, R8+8 - CFI_MOVQ r9, R9+8 - CFI_MOVQ r10, R10+8 - CFI_MOVQ r11, R11+8 - CFI_MOVQ rbx, RBX+8 - CFI_MOVQ rbp, RBP+8 - CFI_MOVQ r12, R12+8 - CFI_MOVQ r13, R13+8 - CFI_MOVQ r14, R14+8 - CFI_MOVQ r15, R15+8 + movq_cfi rdi, RDI+8 + movq_cfi rsi, RSI+8 + movq_cfi rdx, RDX+8 + movq_cfi rcx, RCX+8 + movq_cfi rax, RAX+8 + movq_cfi r8, R8+8 + movq_cfi r9, R9+8 + movq_cfi r10, R10+8 + movq_cfi r11, R11+8 + movq_cfi rbx, RBX+8 + movq_cfi rbp, RBP+8 + movq_cfi r12, R12+8 + movq_cfi r13, R13+8 + movq_cfi r14, R14+8 + movq_cfi r15, R15+8 xorl %ebx,%ebx testl $3,CS+8(%rsp) je error_kernelspace @@ -1515,9 +1516,9 @@ ENTRY(xen_failsafe_callback) CFI_RESTORE r11 addq $0x30,%rsp CFI_ADJUST_CFA_OFFSET -0x30 - CFI_PUSHQ $0 /* RIP */ - CFI_PUSHQ %r11 - CFI_PUSHQ %rcx + pushq_cfi $0 /* RIP */ + pushq_cfi %r11 + pushq_cfi %rcx jmp general_protection CFI_RESTORE_STATE 1: /* Segment mismatch => Category 1 (Bad segment). Retry the IRET. */ @@ -1527,7 +1528,7 @@ ENTRY(xen_failsafe_callback) CFI_RESTORE r11 addq $0x30,%rsp CFI_ADJUST_CFA_OFFSET -0x30 - CFI_PUSHQ $0 + pushq_cfi $0 SAVE_ALL jmp error_exit CFI_ENDPROC -- cgit v1.2.3 From c002a1e6b6b6f07ae04e68987054bf1f2150ae48 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Fri, 21 Nov 2008 16:41:55 +0100 Subject: x86: introduce save_rest and restructure the PTREGSCALL macro in entry_64.S Impact: cleanup The save_rest function completes a partial stack frame for use by the PTREGSCALL macro. This also avoids the indirect call in PTREGSCALLs. This adds the macro movq_cfi_restore to hide the CFI_RESTORE annotation when restoring a register from the stack frame. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 91 +++++++++++++++++++++++++++------------------- 1 file changed, 53 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 92c5e18340db..ef95c45b9269 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -78,6 +78,11 @@ CFI_REL_OFFSET \reg, \offset .endm + .macro movq_cfi_restore offset reg + movq \offset(%rsp), %\reg + CFI_RESTORE \reg + .endm + #ifdef CONFIG_FUNCTION_TRACER #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) @@ -186,21 +191,21 @@ ENTRY(native_usergs_sysret64) */ /* %rsp:at FRAMEEND */ - .macro FIXUP_TOP_OF_STACK tmp - movq %gs:pda_oldrsp,\tmp - movq \tmp,RSP(%rsp) - movq $__USER_DS,SS(%rsp) - movq $__USER_CS,CS(%rsp) - movq $-1,RCX(%rsp) - movq R11(%rsp),\tmp /* get eflags */ - movq \tmp,EFLAGS(%rsp) + .macro FIXUP_TOP_OF_STACK tmp offset=0 + movq %gs:pda_oldrsp,\tmp + movq \tmp,RSP+\offset(%rsp) + movq $__USER_DS,SS+\offset(%rsp) + movq $__USER_CS,CS+\offset(%rsp) + movq $-1,RCX+\offset(%rsp) + movq R11+\offset(%rsp),\tmp /* get eflags */ + movq \tmp,EFLAGS+\offset(%rsp) .endm - .macro RESTORE_TOP_OF_STACK tmp,offset=0 - movq RSP-\offset(%rsp),\tmp - movq \tmp,%gs:pda_oldrsp - movq EFLAGS-\offset(%rsp),\tmp - movq \tmp,R11-\offset(%rsp) + .macro RESTORE_TOP_OF_STACK tmp offset=0 + movq RSP+\offset(%rsp),\tmp + movq \tmp,%gs:pda_oldrsp + movq EFLAGS+\offset(%rsp),\tmp + movq \tmp,R11+\offset(%rsp) .endm .macro FAKE_STACK_FRAME child_rip @@ -333,6 +338,21 @@ ENTRY(save_args) CFI_ENDPROC END(save_args) +ENTRY(save_rest) + PARTIAL_FRAME 1 REST_SKIP+8 + movq 5*8+16(%rsp), %r11 /* save return address */ + movq_cfi rbx, RBX+16 + movq_cfi rbp, RBP+16 + movq_cfi r12, R12+16 + movq_cfi r13, R13+16 + movq_cfi r14, R14+16 + movq_cfi r15, R15+16 + movq %r11, 8(%rsp) /* return address */ + FIXUP_TOP_OF_STACK %r11, 16 + ret + CFI_ENDPROC +END(save_rest) + /* * A newly forked process directly context switches into this. */ @@ -353,7 +373,7 @@ rff_action: je int_ret_from_sys_call testl $_TIF_IA32,TI_flags(%rcx) jnz int_ret_from_sys_call - RESTORE_TOP_OF_STACK %rdi,ARGOFFSET + RESTORE_TOP_OF_STACK %rdi, -ARGOFFSET jmp ret_from_sys_call rff_trace: movq %rsp,%rdi @@ -626,18 +646,20 @@ END(system_call) /* * Certain special system calls that need to save a complete full stack frame. */ - .macro PTREGSCALL label,func,arg - .globl \label -\label: - leaq \func(%rip),%rax - leaq -ARGOFFSET+8(%rsp),\arg /* 8 for return address */ - jmp ptregscall_common +ENTRY(\label) + PARTIAL_FRAME 1 8 /* offset 8: return address */ + subq $REST_SKIP, %rsp + CFI_ADJUST_CFA_OFFSET REST_SKIP + call save_rest + DEFAULT_FRAME 0 8 /* offset 8: return address */ + leaq 8(%rsp), \arg /* pt_regs pointer */ + call \func + jmp ptregscall_common + CFI_ENDPROC END(\label) .endm - CFI_STARTPROC - PTREGSCALL stub_clone, sys_clone, %r8 PTREGSCALL stub_fork, sys_fork, %rdi PTREGSCALL stub_vfork, sys_vfork, %rdi @@ -645,22 +667,15 @@ END(\label) PTREGSCALL stub_iopl, sys_iopl, %rsi ENTRY(ptregscall_common) - popq %r11 - CFI_ADJUST_CFA_OFFSET -8 - CFI_REGISTER rip, r11 - SAVE_REST - movq %r11, %r15 - CFI_REGISTER rip, r15 - FIXUP_TOP_OF_STACK %r11 - call *%rax - RESTORE_TOP_OF_STACK %r11 - movq %r15, %r11 - CFI_REGISTER rip, r11 - RESTORE_REST - pushq %r11 - CFI_ADJUST_CFA_OFFSET 8 - CFI_REL_OFFSET rip, 0 - ret + DEFAULT_FRAME 1 8 /* offset 8: return address */ + RESTORE_TOP_OF_STACK %r11, 8 + movq_cfi_restore R15+8, r15 + movq_cfi_restore R14+8, r14 + movq_cfi_restore R13+8, r13 + movq_cfi_restore R12+8, r12 + movq_cfi_restore RBP+8, rbp + movq_cfi_restore RBX+8, rbx + ret $REST_SKIP /* pop extended registers */ CFI_ENDPROC END(ptregscall_common) -- cgit v1.2.3 From e2f6bc25b98dbb10d809ee50262b43fcae67840a Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Fri, 21 Nov 2008 16:43:18 +0100 Subject: x86: entry_64.S: factor out save_paranoid and paranoid_exit Impact: cleanup, shrink kernel image size Also expand the paranoid_exit0 macro into nmi_exit inside the nmi stub in the case of enabled irq-tracing. This gives a few hundred bytes code size reduction. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 151 ++++++++++++++++++++++++++++++--------------- 1 file changed, 102 insertions(+), 49 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index ef95c45b9269..fad777b11366 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -353,6 +353,36 @@ ENTRY(save_rest) CFI_ENDPROC END(save_rest) +/* save complete stack frame */ +ENTRY(save_paranoid) + XCPT_FRAME 1 RDI+8 + cld + movq_cfi rdi, RDI+8 + movq_cfi rsi, RSI+8 + movq_cfi rdx, RDX+8 + movq_cfi rcx, RCX+8 + movq_cfi rax, RAX+8 + movq_cfi r8, R8+8 + movq_cfi r9, R9+8 + movq_cfi r10, R10+8 + movq_cfi r11, R11+8 + movq_cfi rbx, RBX+8 + movq_cfi rbp, RBP+8 + movq_cfi r12, R12+8 + movq_cfi r13, R13+8 + movq_cfi r14, R14+8 + movq_cfi r15, R15+8 + movl $1,%ebx + movl $MSR_GS_BASE,%ecx + rdmsr + testl %edx,%edx + js 1f /* negative -> in kernel */ + SWAPGS + xorl %ebx,%ebx +1: ret + CFI_ENDPROC +END(save_paranoid) + /* * A newly forked process directly context switches into this. */ @@ -1012,24 +1042,15 @@ END(spurious_interrupt) .endm /* error code is on the stack already */ - /* handle NMI like exceptions that can happen everywhere */ - .macro paranoidentry sym, ist=0, irqtrace=1 - SAVE_ALL - cld - movl $1,%ebx - movl $MSR_GS_BASE,%ecx - rdmsr - testl %edx,%edx - js 1f - SWAPGS - xorl %ebx,%ebx -1: + .macro paranoidentry sym ist=0 + subq $15*8, %rsp + CFI_ADJUST_CFA_OFFSET 15*8 + call save_paranoid + DEFAULT_FRAME 0 .if \ist movq %gs:pda_data_offset, %rbp .endif - .if \irqtrace TRACE_IRQS_OFF - .endif movq %rsp,%rdi movq ORIG_RAX(%rsp),%rsi movq $-1,ORIG_RAX(%rsp) @@ -1041,9 +1062,7 @@ END(spurious_interrupt) addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) .endif DISABLE_INTERRUPTS(CLBR_NONE) - .if \irqtrace TRACE_IRQS_OFF - .endif .endm /* @@ -1058,57 +1077,48 @@ END(spurious_interrupt) * is fundamentally NMI-unsafe. (we cannot change the soft and * hard flags at once, atomically) */ - .macro paranoidexit trace=1 + /* ebx: no swapgs flag */ -paranoid_exit\trace: +KPROBE_ENTRY(paranoid_exit) + INTR_FRAME testl %ebx,%ebx /* swapgs needed? */ - jnz paranoid_restore\trace + jnz paranoid_restore testl $3,CS(%rsp) - jnz paranoid_userspace\trace -paranoid_swapgs\trace: - .if \trace + jnz paranoid_userspace +paranoid_swapgs: TRACE_IRQS_IRETQ 0 - .endif SWAPGS_UNSAFE_STACK -paranoid_restore\trace: +paranoid_restore: RESTORE_ALL 8 jmp irq_return -paranoid_userspace\trace: +paranoid_userspace: GET_THREAD_INFO(%rcx) movl TI_flags(%rcx),%ebx andl $_TIF_WORK_MASK,%ebx - jz paranoid_swapgs\trace + jz paranoid_swapgs movq %rsp,%rdi /* &pt_regs */ call sync_regs movq %rax,%rsp /* switch stack for scheduling */ testl $_TIF_NEED_RESCHED,%ebx - jnz paranoid_schedule\trace + jnz paranoid_schedule movl %ebx,%edx /* arg3: thread flags */ - .if \trace TRACE_IRQS_ON - .endif ENABLE_INTERRUPTS(CLBR_NONE) xorl %esi,%esi /* arg2: oldset */ movq %rsp,%rdi /* arg1: &pt_regs */ call do_notify_resume DISABLE_INTERRUPTS(CLBR_NONE) - .if \trace TRACE_IRQS_OFF - .endif - jmp paranoid_userspace\trace -paranoid_schedule\trace: - .if \trace + jmp paranoid_userspace +paranoid_schedule: TRACE_IRQS_ON - .endif ENABLE_INTERRUPTS(CLBR_ANY) call schedule DISABLE_INTERRUPTS(CLBR_ANY) - .if \trace TRACE_IRQS_OFF - .endif - jmp paranoid_userspace\trace + jmp paranoid_userspace CFI_ENDPROC - .endm +END(paranoid_exit) /* * Exception entry point. This expects an error code/orig_rax on the stack. @@ -1326,20 +1336,63 @@ KPROBE_ENTRY(debug) pushq $0 CFI_ADJUST_CFA_OFFSET 8 paranoidentry do_debug, DEBUG_STACK - paranoidexit + jmp paranoid_exit + CFI_ENDPROC KPROBE_END(debug) /* runs on exception stack */ KPROBE_ENTRY(nmi) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $-1 - CFI_ADJUST_CFA_OFFSET 8 - paranoidentry do_nmi, 0, 0 + pushq_cfi $-1 + subq $15*8, %rsp + CFI_ADJUST_CFA_OFFSET 15*8 + call save_paranoid + DEFAULT_FRAME 0 + /* paranoidentry do_nmi, 0; without TRACE_IRQS_OFF */ + movq %rsp,%rdi + movq ORIG_RAX(%rsp),%rsi + movq $-1,ORIG_RAX(%rsp) + call do_nmi + DISABLE_INTERRUPTS(CLBR_NONE) #ifdef CONFIG_TRACE_IRQFLAGS - paranoidexit 0 + /* paranoidexit; without TRACE_IRQS_OFF */ + /* ebx: no swapgs flag */ +nmi_exit: + testl %ebx,%ebx /* swapgs needed? */ + jnz nmi_restore + testl $3,CS(%rsp) + jnz nmi_userspace +nmi_swapgs: + SWAPGS_UNSAFE_STACK +nmi_restore: + RESTORE_ALL 8 + jmp irq_return +nmi_userspace: + GET_THREAD_INFO(%rcx) + movl TI_flags(%rcx),%ebx + andl $_TIF_WORK_MASK,%ebx + jz nmi_swapgs + movq %rsp,%rdi /* &pt_regs */ + call sync_regs + movq %rax,%rsp /* switch stack for scheduling */ + testl $_TIF_NEED_RESCHED,%ebx + jnz nmi_schedule + movl %ebx,%edx /* arg3: thread flags */ + ENABLE_INTERRUPTS(CLBR_NONE) + xorl %esi,%esi /* arg2: oldset */ + movq %rsp,%rdi /* arg1: &pt_regs */ + call do_notify_resume + DISABLE_INTERRUPTS(CLBR_NONE) + jmp nmi_userspace +nmi_schedule: + ENABLE_INTERRUPTS(CLBR_ANY) + call schedule + DISABLE_INTERRUPTS(CLBR_ANY) + jmp nmi_userspace + CFI_ENDPROC #else - jmp paranoid_exit1 + jmp paranoid_exit CFI_ENDPROC #endif KPROBE_END(nmi) @@ -1350,7 +1403,7 @@ KPROBE_ENTRY(int3) pushq $0 CFI_ADJUST_CFA_OFFSET 8 paranoidentry do_int3, DEBUG_STACK - jmp paranoid_exit1 + jmp paranoid_exit CFI_ENDPROC KPROBE_END(int3) @@ -1375,7 +1428,7 @@ ENTRY(double_fault) XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME paranoidentry do_double_fault - jmp paranoid_exit1 + jmp paranoid_exit CFI_ENDPROC END(double_fault) @@ -1392,7 +1445,7 @@ ENTRY(stack_segment) XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME paranoidentry do_stack_segment - jmp paranoid_exit1 + jmp paranoid_exit CFI_ENDPROC END(stack_segment) @@ -1420,7 +1473,7 @@ ENTRY(machine_check) pushq $0 CFI_ADJUST_CFA_OFFSET 8 paranoidentry do_machine_check - jmp paranoid_exit1 + jmp paranoid_exit CFI_ENDPROC END(machine_check) #endif -- cgit v1.2.3 From b8b1d08bf6fe7c09e6cb2294bc0e5e964b361241 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Fri, 21 Nov 2008 16:44:28 +0100 Subject: x86: entry_64.S: split out some macro's and move common code to paranoid_exit Impact: cleanup DISABLE_INTERRUPTS(CLBR_NONE)/TRACE_IRQS_OFF is now always executed just before paranoid_exit. Move it there. Split out paranoidzeroentry, paranoiderrorentry, and paranoidzeroentry_ist to get more readable macro's. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 102 ++++++++++++++++++++++----------------------- 1 file changed, 51 insertions(+), 51 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index fad777b11366..692c1da61905 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1026,6 +1026,39 @@ END(spurious_interrupt) CFI_ENDPROC .endm + .macro paranoidzeroentry sym + INTR_FRAME + PARAVIRT_ADJUST_EXCEPTION_FRAME + pushq $-1 /* ORIG_RAX: no syscall to restart */ + CFI_ADJUST_CFA_OFFSET 8 + subq $15*8, %rsp + call save_paranoid + TRACE_IRQS_OFF + movq %rsp,%rdi /* pt_regs pointer */ + xorl %esi,%esi /* no error code */ + call \sym + jmp paranoid_exit /* %ebx: no swapgs flag */ + CFI_ENDPROC + .endm + + .macro paranoidzeroentry_ist sym ist + INTR_FRAME + PARAVIRT_ADJUST_EXCEPTION_FRAME + pushq $-1 /* ORIG_RAX: no syscall to restart */ + CFI_ADJUST_CFA_OFFSET 8 + subq $15*8, %rsp + call save_paranoid + TRACE_IRQS_OFF + movq %rsp,%rdi /* pt_regs pointer */ + xorl %esi,%esi /* no error code */ + movq %gs:pda_data_offset, %rbp + subq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) + call \sym + addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) + jmp paranoid_exit /* %ebx: no swapgs flag */ + CFI_ENDPROC + .endm + .macro errorentry sym XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME @@ -1042,27 +1075,20 @@ END(spurious_interrupt) .endm /* error code is on the stack already */ - .macro paranoidentry sym ist=0 - subq $15*8, %rsp + .macro paranoiderrorentry sym + XCPT_FRAME + PARAVIRT_ADJUST_EXCEPTION_FRAME + subq $15*8,%rsp CFI_ADJUST_CFA_OFFSET 15*8 call save_paranoid DEFAULT_FRAME 0 - .if \ist - movq %gs:pda_data_offset, %rbp - .endif TRACE_IRQS_OFF - movq %rsp,%rdi - movq ORIG_RAX(%rsp),%rsi - movq $-1,ORIG_RAX(%rsp) - .if \ist - subq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) - .endif + movq %rsp,%rdi /* pt_regs pointer */ + movq ORIG_RAX(%rsp),%rsi /* get error code */ + movq $-1,ORIG_RAX(%rsp) /* no syscall to restart */ call \sym - .if \ist - addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) - .endif - DISABLE_INTERRUPTS(CLBR_NONE) - TRACE_IRQS_OFF + jmp paranoid_exit /* %ebx: no swapgs flag */ + CFI_ENDPROC .endm /* @@ -1081,6 +1107,8 @@ END(spurious_interrupt) /* ebx: no swapgs flag */ KPROBE_ENTRY(paranoid_exit) INTR_FRAME + DISABLE_INTERRUPTS(CLBR_NONE) + TRACE_IRQS_OFF testl %ebx,%ebx /* swapgs needed? */ jnz paranoid_restore testl $3,CS(%rsp) @@ -1331,13 +1359,7 @@ END(device_not_available) /* runs on exception stack */ KPROBE_ENTRY(debug) - INTR_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $0 - CFI_ADJUST_CFA_OFFSET 8 - paranoidentry do_debug, DEBUG_STACK - jmp paranoid_exit - CFI_ENDPROC + paranoidzeroentry_ist do_debug, DEBUG_STACK KPROBE_END(debug) /* runs on exception stack */ @@ -1351,14 +1373,12 @@ KPROBE_ENTRY(nmi) DEFAULT_FRAME 0 /* paranoidentry do_nmi, 0; without TRACE_IRQS_OFF */ movq %rsp,%rdi - movq ORIG_RAX(%rsp),%rsi - movq $-1,ORIG_RAX(%rsp) + movq $-1,%rsi call do_nmi - DISABLE_INTERRUPTS(CLBR_NONE) #ifdef CONFIG_TRACE_IRQFLAGS /* paranoidexit; without TRACE_IRQS_OFF */ /* ebx: no swapgs flag */ -nmi_exit: + DISABLE_INTERRUPTS(CLBR_NONE) testl %ebx,%ebx /* swapgs needed? */ jnz nmi_restore testl $3,CS(%rsp) @@ -1398,13 +1418,7 @@ nmi_schedule: KPROBE_END(nmi) KPROBE_ENTRY(int3) - INTR_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $0 - CFI_ADJUST_CFA_OFFSET 8 - paranoidentry do_int3, DEBUG_STACK - jmp paranoid_exit - CFI_ENDPROC + paranoidzeroentry_ist do_int3, DEBUG_STACK KPROBE_END(int3) ENTRY(overflow) @@ -1425,11 +1439,7 @@ END(coprocessor_segment_overrun) /* runs on exception stack */ ENTRY(double_fault) - XCPT_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - paranoidentry do_double_fault - jmp paranoid_exit - CFI_ENDPROC + paranoiderrorentry do_double_fault END(double_fault) ENTRY(invalid_TSS) @@ -1442,11 +1452,7 @@ END(segment_not_present) /* runs on exception stack */ ENTRY(stack_segment) - XCPT_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - paranoidentry do_stack_segment - jmp paranoid_exit - CFI_ENDPROC + paranoiderrorentry do_stack_segment END(stack_segment) KPROBE_ENTRY(general_protection) @@ -1468,13 +1474,7 @@ END(spurious_interrupt_bug) #ifdef CONFIG_X86_MCE /* runs on exception stack */ ENTRY(machine_check) - INTR_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq $0 - CFI_ADJUST_CFA_OFFSET 8 - paranoidentry do_machine_check - jmp paranoid_exit - CFI_ENDPROC + paranoidzeroentry do_machine_check END(machine_check) #endif -- cgit v1.2.3 From c81084114f6ff957bc6b5a0048350479c1c1f7b3 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Fri, 21 Nov 2008 22:59:52 +0100 Subject: x86: split out some macro's and move common code to paranoid_exit, fix Impact: fix bootup crash Even though it tested fine for me, there was still a bug in the first patch: I have overlooked a call to ptregscall_common. This patch fixes that, I think, but the code is never executed for me while running a debian install... (I tested this by putting an "1:jmp 1b" in there.) Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 692c1da61905..e5ddf573ded2 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -529,10 +529,13 @@ sysret_signal: jc sysret_audit #endif /* edx: work flags (arg3) */ - leaq do_notify_resume(%rip),%rax leaq -ARGOFFSET(%rsp),%rdi # &pt_regs -> arg1 xorl %esi,%esi # oldset -> arg2 - call ptregscall_common + SAVE_REST + FIXUP_TOP_OF_STACK %r11 + call do_notify_resume + RESTORE_TOP_OF_STACK %r11 + RESTORE_REST movl $_TIF_WORK_MASK,%edi /* Use IRET because user could have changed frame. This works because ptregscall_common has called FIXUP_TOP_OF_STACK. */ -- cgit v1.2.3 From 3889d0cea2b73049bdca062d9ff1e5d33468289c Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Sat, 22 Nov 2008 23:39:23 -0800 Subject: x86: revert default reboot method to REBOOT_KBD Impact: Reverts default reboot method. Checkin 14d7ca5c575853664d8fe4f225a77b8df1b7de7d changed the default reboot method to "pci", a.k.a. port CF9. Unfortunately this has been shown to cause lockups on at least two systems for which REBOOT_KBD worked, both Thinkpads with Intel chipsets. This reverts the default to REBOOT_KBD, while leaving the option to have "reboot=pci" specified explicitly or via a DMI match. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/reboot.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index ddc93891cdcd..790b09fbadcb 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -29,7 +29,7 @@ EXPORT_SYMBOL(pm_power_off); static const struct desc_ptr no_idt = {}; static int reboot_mode; -enum reboot_type reboot_type = BOOT_CF9_COND; +enum reboot_type reboot_type = BOOT_KBD; int reboot_force; #if defined(CONFIG_X86_32) && defined(CONFIG_SMP) -- cgit v1.2.3 From 3aeb95d5b7839708a8d8e11aa274ee4d0d4042cc Mon Sep 17 00:00:00 2001 From: jia zhang Date: Sun, 23 Nov 2008 09:51:41 +0800 Subject: x86_64: fix the check in stack_overflow_check Impact: make stack overflow debug check and printout narrower stack_overflow_check() should consider the stack usage of pt_regs, and thus it could warn us in advance. Additionally, it looks better for the warning time to start at INITIAL_JIFFIES. Assuming that rsp gets close to the check point before interrupt arrives: when interrupt really happens, thread_info will be partly overrode. Signed-off-by: jia zhang Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_64.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 60eb84eb77a0..b842fc82be15 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -29,11 +29,12 @@ static inline void stack_overflow_check(struct pt_regs *regs) { u64 curbase = (u64)task_stack_page(current); - static unsigned long warned = -60*HZ; + static unsigned long warned = INITIAL_JIFFIES - 60*HZ; if (regs->sp >= curbase && regs->sp <= curbase + THREAD_SIZE && - regs->sp < curbase + sizeof(struct thread_info) + 128 && - time_after(jiffies, warned + 60*HZ)) { + regs->sp < curbase + sizeof(struct thread_info) + + sizeof(struct pt_regs) + 128 && + time_after(jiffies, warned + 60*HZ)) { printk("do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n", current->comm, curbase, regs->sp); show_stack(NULL,NULL); -- cgit v1.2.3 From f377fa123d0ec621e8e361ecc3f2a8ee70e81a2e Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sun, 23 Nov 2008 09:02:26 +0100 Subject: x86: clean up stack overflow debug check Impact: cleanup Simplify the irq-sampled stack overflow debug check: - eliminate an #idef - use WARN_ONCE() to emit a single warning (all bets are off after the first such warning anyway) Signed-off-by: Ingo Molnar --- arch/x86/kernel/irq_64.c | 25 ++++++++++--------------- 1 file changed, 10 insertions(+), 15 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index b842fc82be15..1d3d0e71b044 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -18,7 +18,6 @@ #include #include -#ifdef CONFIG_DEBUG_STACKOVERFLOW /* * Probabilistic stack overflow check: * @@ -28,20 +27,18 @@ */ static inline void stack_overflow_check(struct pt_regs *regs) { +#ifdef CONFIG_DEBUG_STACKOVERFLOW u64 curbase = (u64)task_stack_page(current); - static unsigned long warned = INITIAL_JIFFIES - 60*HZ; - - if (regs->sp >= curbase && regs->sp <= curbase + THREAD_SIZE && - regs->sp < curbase + sizeof(struct thread_info) + - sizeof(struct pt_regs) + 128 && - time_after(jiffies, warned + 60*HZ)) { - printk("do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n", - current->comm, curbase, regs->sp); - show_stack(NULL,NULL); - warned = jiffies; - } -} + + WARN_ONCE(regs->sp >= curbase && + regs->sp <= curbase + THREAD_SIZE && + regs->sp < curbase + sizeof(struct thread_info) + + sizeof(struct pt_regs) + 128, + + "do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n", + current->comm, curbase, regs->sp); #endif +} /* * do_IRQ handles all normal device IRQ's (the special @@ -61,9 +58,7 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) irq_enter(); irq = __get_cpu_var(vector_irq)[vector]; -#ifdef CONFIG_DEBUG_STACKOVERFLOW stack_overflow_check(regs); -#endif desc = irq_to_desc(irq); if (likely(desc)) -- cgit v1.2.3 From f201ae2356c74bcae130b2177b3dca903ea98071 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 23 Nov 2008 06:22:56 +0100 Subject: tracing/function-return-tracer: store return stack into task_struct and allocate it dynamically Impact: use deeper function tracing depth safely Some tests showed that function return tracing needed a more deeper depth of function calls. But it could be unsafe to store these return addresses to the stack. So these arrays will now be allocated dynamically into task_struct of current only when the tracer is activated. Typical scheme when tracer is activated: - allocate a return stack for each task in global list. - fork: allocate the return stack for the newly created task - exit: free return stack of current - idle init: same as fork I chose a default depth of 50. I don't have overruns anymore. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 29 +++++++++++++++-------------- 1 file changed, 15 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 356bb1eb6e9a..bb137f7297ed 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -350,19 +350,21 @@ static int push_return_trace(unsigned long ret, unsigned long long time, unsigned long func) { int index; - struct thread_info *ti = current_thread_info(); + + if (!current->ret_stack) + return -EBUSY; /* The return trace stack is full */ - if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { - atomic_inc(&ti->trace_overrun); + if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) { + atomic_inc(¤t->trace_overrun); return -EBUSY; } - index = ++ti->curr_ret_stack; + index = ++current->curr_ret_stack; barrier(); - ti->ret_stack[index].ret = ret; - ti->ret_stack[index].func = func; - ti->ret_stack[index].calltime = time; + current->ret_stack[index].ret = ret; + current->ret_stack[index].func = func; + current->ret_stack[index].calltime = time; return 0; } @@ -373,13 +375,12 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, { int index; - struct thread_info *ti = current_thread_info(); - index = ti->curr_ret_stack; - *ret = ti->ret_stack[index].ret; - *func = ti->ret_stack[index].func; - *time = ti->ret_stack[index].calltime; - *overrun = atomic_read(&ti->trace_overrun); - ti->curr_ret_stack--; + index = current->curr_ret_stack; + *ret = current->ret_stack[index].ret; + *func = current->ret_stack[index].func; + *time = current->ret_stack[index].calltime; + *overrun = atomic_read(¤t->trace_overrun); + current->curr_ret_stack--; } /* -- cgit v1.2.3 From 02b67518e2b1c490787dac7f35e1204e74fe21ba Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?T=C3=B6r=C3=B6k=20Edwin?= Date: Sat, 22 Nov 2008 13:28:47 +0200 Subject: tracing: add support for userspace stacktraces in tracing/iter_ctrl MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Impact: add new (default-off) tracing visualization feature Usage example: mount -t debugfs nodev /sys/kernel/debug cd /sys/kernel/debug/tracing echo userstacktrace >iter_ctrl echo sched_switch >current_tracer echo 1 >tracing_enabled .... run application ... echo 0 >tracing_enabled Then read one of 'trace','latency_trace','trace_pipe'. To get the best output you can compile your userspace programs with frame pointers (at least glibc + the app you are tracing). Signed-off-by: Török Edwin Signed-off-by: Ingo Molnar --- arch/x86/kernel/stacktrace.c | 57 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 57 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/stacktrace.c b/arch/x86/kernel/stacktrace.c index a03e7f6d90c3..b15153060417 100644 --- a/arch/x86/kernel/stacktrace.c +++ b/arch/x86/kernel/stacktrace.c @@ -6,6 +6,7 @@ #include #include #include +#include #include static void save_stack_warning(void *data, char *msg) @@ -83,3 +84,59 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace) trace->entries[trace->nr_entries++] = ULONG_MAX; } EXPORT_SYMBOL_GPL(save_stack_trace_tsk); + +/* Userspace stacktrace - based on kernel/trace/trace_sysprof.c */ + +struct stack_frame { + const void __user *next_fp; + unsigned long return_address; +}; + +static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) +{ + int ret; + + if (!access_ok(VERIFY_READ, fp, sizeof(*frame))) + return 0; + + ret = 1; + pagefault_disable(); + if (__copy_from_user_inatomic(frame, fp, sizeof(*frame))) + ret = 0; + pagefault_enable(); + + return ret; +} + +void save_stack_trace_user(struct stack_trace *trace) +{ + /* + * Trace user stack if we are not a kernel thread + */ + if (current->mm) { + const struct pt_regs *regs = task_pt_regs(current); + const void __user *fp = (const void __user *)regs->bp; + + if (trace->nr_entries < trace->max_entries) + trace->entries[trace->nr_entries++] = regs->ip; + + while (trace->nr_entries < trace->max_entries) { + struct stack_frame frame; + frame.next_fp = NULL; + frame.return_address = 0; + if (!copy_stack_frame(fp, &frame)) + break; + if ((unsigned long)fp < regs->sp) + break; + if (frame.return_address) + trace->entries[trace->nr_entries++] = + frame.return_address; + if (fp == frame.next_fp) + break; + fp = frame.next_fp; + } + } + if (trace->nr_entries < trace->max_entries) + trace->entries[trace->nr_entries++] = ULONG_MAX; +} + -- cgit v1.2.3 From 5c9b3a0c7b8be3cdef3d7418f0a49127e7cdc998 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 21 Nov 2008 17:36:41 -0800 Subject: x86: signal: cosmetic unification of including headers Impact: cleanup Make the headers portion of signal_32.c and signal_64.c the same. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 27 +++++++++++++++------------ arch/x86/kernel/signal_64.c | 7 ++++++- 2 files changed, 21 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index c2aabeba27a5..0ff8d8750a7d 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -4,29 +4,32 @@ * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes */ -#include -#include -#include -#include +#include +#include +#include #include -#include #include -#include -#include #include -#include #include +#include #include -#include -#include -#include +#include +#include +#include +#include #include #include -#include #include #include + +#ifdef CONFIG_X86_64 +#include +#include +#include +#endif /* CONFIG_X86_64 */ + #include #include diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 3d54d366ccb2..c52244ac19fc 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -19,17 +19,22 @@ #include #include #include -#include #include #include #include #include +#include + +#ifdef CONFIG_X86_64 #include #include #include +#endif /* CONFIG_X86_64 */ + #include #include + #include "sigframe.h" #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -- cgit v1.2.3 From 666ac7be049ec290625e65d5922ff59f7bdec527 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 21 Nov 2008 17:38:25 -0800 Subject: x86: signal: cosmetic unification of sys_sigaltstack() Impact: cleanup Add #ifdef directive for unification. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 9 +++++++++ arch/x86/kernel/signal_64.c | 15 +++++++++++++++ 2 files changed, 24 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index 0ff8d8750a7d..d9909881ac66 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -125,6 +125,7 @@ sys_sigaction(int sig, const struct old_sigaction __user *act, return ret; } +#ifdef CONFIG_X86_32 asmlinkage int sys_sigaltstack(unsigned long bx) { /* @@ -137,6 +138,14 @@ asmlinkage int sys_sigaltstack(unsigned long bx) return do_sigaltstack(uss, uoss, regs->sp); } +#else /* !CONFIG_X86_32 */ +asmlinkage long +sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, + struct pt_regs *regs) +{ + return do_sigaltstack(uss, uoss, regs->sp); +} +#endif /* CONFIG_X86_32 */ #define COPY(x) { \ err |= __get_user(regs->x, &sc->x); \ diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index c52244ac19fc..b6e4fe03a36b 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -50,12 +50,27 @@ # define FIX_EFLAGS __FIX_EFLAGS #endif +#ifdef CONFIG_X86_32 +asmlinkage int sys_sigaltstack(unsigned long bx) +{ + /* + * This is needed to make gcc realize it doesn't own the + * "struct pt_regs" + */ + struct pt_regs *regs = (struct pt_regs *)&bx; + const stack_t __user *uss = (const stack_t __user *)bx; + stack_t __user *uoss = (stack_t __user *)regs->cx; + + return do_sigaltstack(uss, uoss, regs->sp); +} +#else /* !CONFIG_X86_32 */ asmlinkage long sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, struct pt_regs *regs) { return do_sigaltstack(uss, uoss, regs->sp); } +#endif /* CONFIG_X86_32 */ #define COPY(x) { \ err |= __get_user(regs->x, &sc->x); \ -- cgit v1.2.3 From 2456d738ef051f85170bf018faef63f83fa84eb5 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 21 Nov 2008 17:38:57 -0800 Subject: x86: signal: cosmetic unification of sys_rt_sigreturn() Impact: cleanup Add #ifdef directive for unification. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 7 +++++++ arch/x86/kernel/signal_64.c | 9 +++++++++ 2 files changed, 16 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index d9909881ac66..f7dd6c44c042 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -299,12 +299,19 @@ badframe: return 0; } +#ifdef CONFIG_X86_32 asmlinkage int sys_rt_sigreturn(unsigned long __unused) { struct pt_regs *regs = (struct pt_regs *)&__unused; return do_rt_sigreturn(regs); } +#else /* !CONFIG_X86_32 */ +asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) +{ + return do_rt_sigreturn(regs); +} +#endif /* CONFIG_X86_32 */ /* * Set up a signal frame. diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index b6e4fe03a36b..32718f5e4f61 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -169,10 +169,19 @@ badframe: return 0; } +#ifdef CONFIG_X86_32 +asmlinkage int sys_rt_sigreturn(unsigned long __unused) +{ + struct pt_regs *regs = (struct pt_regs *)&__unused; + + return do_rt_sigreturn(regs); +} +#else /* !CONFIG_X86_32 */ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) { return do_rt_sigreturn(regs); } +#endif /* CONFIG_X86_32 */ /* * Set up a signal frame. -- cgit v1.2.3 From c450d7805b2c5cac8846c5f490fddfd9030d2207 Mon Sep 17 00:00:00 2001 From: Hannes Eder Date: Fri, 21 Nov 2008 23:17:09 +0100 Subject: x86: vmware - fix sparse warnings Impact: fix sparse build warning Fix the following sparse warnings: arch/x86/kernel/cpu/vmware.c:69:5: warning: symbol 'vmware_platform' was not declared. Should it be static? arch/x86/kernel/cpu/vmware.c:89:15: warning: symbol 'vmware_get_tsc_khz' was not declared. Should it be static? arch/x86/kernel/cpu/vmware.c:107:16: warning: symbol 'vmware_set_feature_bits' was not declared. Should it be static? Signed-off-by: Hannes Eder Cc: "Alok N Kataria" Cc: "Dan Hecht" Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/vmware.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c index c034bda842d9..284c399e3234 100644 --- a/arch/x86/kernel/cpu/vmware.c +++ b/arch/x86/kernel/cpu/vmware.c @@ -23,6 +23,7 @@ #include #include +#include #define CPUID_VMWARE_INFO_LEAF 0x40000000 #define VMWARE_HYPERVISOR_MAGIC 0x564D5868 -- cgit v1.2.3 From 4e42ebd57b2e727b28bf5f6068e95cd19b0e807b Mon Sep 17 00:00:00 2001 From: Hannes Eder Date: Fri, 21 Nov 2008 22:56:17 +0100 Subject: x86: hypervisor - fix sparse warnings Impact: fix sparse build warning Fix the following sparse warnings: arch/x86/kernel/cpu/hypervisor.c:37:15: warning: symbol 'get_hypervisor_tsc_freq' was not declared. Should it be static? arch/x86/kernel/cpu/hypervisor.c:53:16: warning: symbol 'init_hypervisor' was not declared. Should it be static? Signed-off-by: Hannes Eder Cc: "Alok N Kataria" Cc: "Dan Hecht" Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/hypervisor.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c index 35ae2b75226d..fb5b86af0b01 100644 --- a/arch/x86/kernel/cpu/hypervisor.c +++ b/arch/x86/kernel/cpu/hypervisor.c @@ -23,6 +23,7 @@ #include #include +#include static inline void __cpuinit detect_hypervisor_vendor(struct cpuinfo_x86 *c) -- cgit v1.2.3 From 8d7c6a96164651dbbab449ef0b5c20ae1f76a3a1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?T=C3=B6r=C3=B6k=20Edwin?= Date: Sun, 23 Nov 2008 12:39:06 +0200 Subject: tracing/stack-tracer: fix style issues MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Impact: cleanup Signed-off-by: Török Edwin Signed-off-by: Ingo Molnar --- arch/x86/kernel/stacktrace.c | 51 +++++++++++++++++++++++++------------------- 1 file changed, 29 insertions(+), 22 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/stacktrace.c b/arch/x86/kernel/stacktrace.c index b15153060417..10786af95545 100644 --- a/arch/x86/kernel/stacktrace.c +++ b/arch/x86/kernel/stacktrace.c @@ -89,7 +89,7 @@ EXPORT_SYMBOL_GPL(save_stack_trace_tsk); struct stack_frame { const void __user *next_fp; - unsigned long return_address; + unsigned long ret_addr; }; static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) @@ -108,33 +108,40 @@ static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) return ret; } +static inline void __save_stack_trace_user(struct stack_trace *trace) +{ + const struct pt_regs *regs = task_pt_regs(current); + const void __user *fp = (const void __user *)regs->bp; + + if (trace->nr_entries < trace->max_entries) + trace->entries[trace->nr_entries++] = regs->ip; + + while (trace->nr_entries < trace->max_entries) { + struct stack_frame frame; + + frame.next_fp = NULL; + frame.ret_addr = 0; + if (!copy_stack_frame(fp, &frame)) + break; + if ((unsigned long)fp < regs->sp) + break; + if (frame.ret_addr) { + trace->entries[trace->nr_entries++] = + frame.ret_addr; + } + if (fp == frame.next_fp) + break; + fp = frame.next_fp; + } +} + void save_stack_trace_user(struct stack_trace *trace) { /* * Trace user stack if we are not a kernel thread */ if (current->mm) { - const struct pt_regs *regs = task_pt_regs(current); - const void __user *fp = (const void __user *)regs->bp; - - if (trace->nr_entries < trace->max_entries) - trace->entries[trace->nr_entries++] = regs->ip; - - while (trace->nr_entries < trace->max_entries) { - struct stack_frame frame; - frame.next_fp = NULL; - frame.return_address = 0; - if (!copy_stack_frame(fp, &frame)) - break; - if ((unsigned long)fp < regs->sp) - break; - if (frame.return_address) - trace->entries[trace->nr_entries++] = - frame.return_address; - if (fp == frame.next_fp) - break; - fp = frame.next_fp; - } + __save_stack_trace_user(trace); } if (trace->nr_entries < trace->max_entries) trace->entries[trace->nr_entries++] = ULONG_MAX; -- cgit v1.2.3 From a1967d64414dab500e86cbbddf8eae6ad2047903 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 21 Nov 2008 11:16:48 -0800 Subject: x86: revert irq number limitation Impact: fix MSIx not enough irq numbers available regression The manual revert of the sparse_irq patches missed to bring the number of possible irqs back to the .27 status. This resulted in a regression when two multichannel network cards were placed in a system with only one IO_APIC - causing the networking driver to not have the right IRQ and the device not coming up. Remove the dynamic allocation logic leftovers and simply return NR_IRQS in probe_nr_irqs() for now. Fixes: http://lkml.org/lkml/2008/11/19/354 Reported-by: Jesper Dangaard Brouer Signed-off-by: Thomas Gleixner Tested-by: Jesper Dangaard Brouer Acked-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 22 +--------------------- 1 file changed, 1 insertion(+), 21 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index c9513e1ff28d..1fec0f9b1508 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -3608,27 +3608,7 @@ int __init io_apic_get_redir_entries (int ioapic) int __init probe_nr_irqs(void) { - int idx; - int nr = 0; -#ifndef CONFIG_XEN - int nr_min = 32; -#else - int nr_min = NR_IRQS; -#endif - - for (idx = 0; idx < nr_ioapics; idx++) - nr += io_apic_get_redir_entries(idx) + 1; - - /* double it for hotplug and msi and nmi */ - nr <<= 1; - - /* something wrong ? */ - if (nr < nr_min) - nr = nr_min; - if (WARN_ON(nr > NR_IRQS)) - nr = NR_IRQS; - - return nr; + return NR_IRQS; } /* -------------------------------------------------------------------------- -- cgit v1.2.3 From 8a2503fa4a6fae8ee42140b339f37373fc6acaae Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 23 Nov 2008 14:53:43 +0300 Subject: x86: move dwarf2 related macro to dwarf2.h Impact: cleanup Move recently introduced dwarf2 macros to dwarf2.h file. It allow us to not duplicate them in assembly files. Active usage of _cfi macros don't make assembly files more obvious to understand but we already have a lot of macros there which requires to search the definitions of them *anyway*. But at least it make every cfi usage one line shorter. Also some code alignment is done. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 23 ----------------------- 1 file changed, 23 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index e5ddf573ded2..249eb604e71b 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -60,29 +60,6 @@ #define __AUDIT_ARCH_LE 0x40000000 .code64 -/* - * Some macro's to hide the most frequently occuring CFI annotations. - */ - .macro pushq_cfi reg - pushq \reg - CFI_ADJUST_CFA_OFFSET 8 - .endm - - .macro popq_cfi reg - popq \reg - CFI_ADJUST_CFA_OFFSET -8 - .endm - - .macro movq_cfi reg offset=0 - movq %\reg, \offset(%rsp) - CFI_REL_OFFSET \reg, \offset - .endm - - .macro movq_cfi_restore offset reg - movq \offset(%rsp), %\reg - CFI_RESTORE \reg - .endm - #ifdef CONFIG_FUNCTION_TRACER #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) -- cgit v1.2.3 From 322648d1ba75280d62f114d47048beb0b35f5047 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Sun, 23 Nov 2008 10:08:28 +0100 Subject: x86: include ENTRY/END in entry handlers in entry_64.S Impact: cleanup of entry_64.S Except for the order and the place of the functions, this patch should not change the generated code. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 259 +++++++++++++++++++-------------------------- 1 file changed, 109 insertions(+), 150 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 249eb604e71b..1a856c0b21a8 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -922,76 +922,70 @@ END(common_interrupt) /* * APIC interrupts. */ - .p2align 5 - - .macro apicinterrupt num,func +.macro apicinterrupt num sym do_sym +ENTRY(\sym) INTR_FRAME pushq $~(\num) CFI_ADJUST_CFA_OFFSET 8 - interrupt \func + interrupt \do_sym jmp ret_from_intr CFI_ENDPROC - .endm +END(\sym) +.endm -ENTRY(thermal_interrupt) - apicinterrupt THERMAL_APIC_VECTOR,smp_thermal_interrupt -END(thermal_interrupt) +#ifdef CONFIG_SMP +apicinterrupt IRQ_MOVE_CLEANUP_VECTOR \ + irq_move_cleanup_interrupt smp_irq_move_cleanup_interrupt +#endif -ENTRY(threshold_interrupt) - apicinterrupt THRESHOLD_APIC_VECTOR,mce_threshold_interrupt -END(threshold_interrupt) +apicinterrupt 220 \ + uv_bau_message_intr1 uv_bau_message_interrupt +apicinterrupt LOCAL_TIMER_VECTOR \ + apic_timer_interrupt smp_apic_timer_interrupt #ifdef CONFIG_SMP -ENTRY(reschedule_interrupt) - apicinterrupt RESCHEDULE_VECTOR,smp_reschedule_interrupt -END(reschedule_interrupt) - - .macro INVALIDATE_ENTRY num -ENTRY(invalidate_interrupt\num) - apicinterrupt INVALIDATE_TLB_VECTOR_START+\num,smp_invalidate_interrupt -END(invalidate_interrupt\num) - .endm - - INVALIDATE_ENTRY 0 - INVALIDATE_ENTRY 1 - INVALIDATE_ENTRY 2 - INVALIDATE_ENTRY 3 - INVALIDATE_ENTRY 4 - INVALIDATE_ENTRY 5 - INVALIDATE_ENTRY 6 - INVALIDATE_ENTRY 7 - -ENTRY(call_function_interrupt) - apicinterrupt CALL_FUNCTION_VECTOR,smp_call_function_interrupt -END(call_function_interrupt) -ENTRY(call_function_single_interrupt) - apicinterrupt CALL_FUNCTION_SINGLE_VECTOR,smp_call_function_single_interrupt -END(call_function_single_interrupt) -ENTRY(irq_move_cleanup_interrupt) - apicinterrupt IRQ_MOVE_CLEANUP_VECTOR,smp_irq_move_cleanup_interrupt -END(irq_move_cleanup_interrupt) +apicinterrupt INVALIDATE_TLB_VECTOR_START+0 \ + invalidate_interrupt0 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+1 \ + invalidate_interrupt1 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+2 \ + invalidate_interrupt2 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+3 \ + invalidate_interrupt3 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+4 \ + invalidate_interrupt4 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+5 \ + invalidate_interrupt5 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+6 \ + invalidate_interrupt6 smp_invalidate_interrupt +apicinterrupt INVALIDATE_TLB_VECTOR_START+7 \ + invalidate_interrupt7 smp_invalidate_interrupt #endif -ENTRY(apic_timer_interrupt) - apicinterrupt LOCAL_TIMER_VECTOR,smp_apic_timer_interrupt -END(apic_timer_interrupt) - -ENTRY(uv_bau_message_intr1) - apicinterrupt 220,uv_bau_message_interrupt -END(uv_bau_message_intr1) +apicinterrupt THRESHOLD_APIC_VECTOR \ + threshold_interrupt mce_threshold_interrupt +apicinterrupt THERMAL_APIC_VECTOR \ + thermal_interrupt smp_thermal_interrupt -ENTRY(error_interrupt) - apicinterrupt ERROR_APIC_VECTOR,smp_error_interrupt -END(error_interrupt) +#ifdef CONFIG_SMP +apicinterrupt CALL_FUNCTION_SINGLE_VECTOR \ + call_function_single_interrupt smp_call_function_single_interrupt +apicinterrupt CALL_FUNCTION_VECTOR \ + call_function_interrupt smp_call_function_interrupt +apicinterrupt RESCHEDULE_VECTOR \ + reschedule_interrupt smp_reschedule_interrupt +#endif -ENTRY(spurious_interrupt) - apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt -END(spurious_interrupt) +apicinterrupt ERROR_APIC_VECTOR \ + error_interrupt smp_error_interrupt +apicinterrupt SPURIOUS_APIC_VECTOR \ + spurious_interrupt smp_spurious_interrupt /* * Exception entry points. */ - .macro zeroentry sym +.macro zeroentry sym do_sym +ENTRY(\sym) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq_cfi $-1 /* ORIG_RAX: no syscall to restart */ @@ -1001,12 +995,14 @@ END(spurious_interrupt) DEFAULT_FRAME 0 movq %rsp,%rdi /* pt_regs pointer */ xorl %esi,%esi /* no error code */ - call \sym + call \do_sym jmp error_exit /* %ebx: no swapgs flag */ CFI_ENDPROC - .endm +END(\sym) +.endm - .macro paranoidzeroentry sym +.macro paranoidzeroentry sym do_sym +KPROBE_ENTRY(\sym) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $-1 /* ORIG_RAX: no syscall to restart */ @@ -1016,12 +1012,14 @@ END(spurious_interrupt) TRACE_IRQS_OFF movq %rsp,%rdi /* pt_regs pointer */ xorl %esi,%esi /* no error code */ - call \sym + call \do_sym jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC - .endm +KPROBE_END(\sym) +.endm - .macro paranoidzeroentry_ist sym ist +.macro paranoidzeroentry_ist sym do_sym ist +KPROBE_ENTRY(\sym) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $-1 /* ORIG_RAX: no syscall to restart */ @@ -1033,13 +1031,19 @@ END(spurious_interrupt) xorl %esi,%esi /* no error code */ movq %gs:pda_data_offset, %rbp subq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) - call \sym + call \do_sym addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC - .endm +KPROBE_END(\sym) +.endm - .macro errorentry sym +.macro errorentry sym do_sym entry=0 +.if \entry +KPROBE_ENTRY(\sym) +.else +ENTRY(\sym) +.endif XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME subq $15*8,%rsp @@ -1049,13 +1053,23 @@ END(spurious_interrupt) movq %rsp,%rdi /* pt_regs pointer */ movq ORIG_RAX(%rsp),%rsi /* get error code */ movq $-1,ORIG_RAX(%rsp) /* no syscall to restart */ - call \sym + call \do_sym jmp error_exit /* %ebx: no swapgs flag */ CFI_ENDPROC - .endm +.if \entry +KPROBE_END(\sym) +.else +END(\sym) +.endif +.endm /* error code is on the stack already */ - .macro paranoiderrorentry sym +.macro paranoiderrorentry sym do_sym entry=1 +.if \entry +KPROBE_ENTRY(\sym) +.else +ENTRY(\sym) +.endif XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME subq $15*8,%rsp @@ -1066,10 +1080,37 @@ END(spurious_interrupt) movq %rsp,%rdi /* pt_regs pointer */ movq ORIG_RAX(%rsp),%rsi /* get error code */ movq $-1,ORIG_RAX(%rsp) /* no syscall to restart */ - call \sym + call \do_sym jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC - .endm +.if \entry +KPROBE_END(\sym) +.else +END(\sym) +.endif +.endm + +zeroentry divide_error do_divide_error +paranoidzeroentry_ist debug do_debug DEBUG_STACK +paranoidzeroentry_ist int3 do_int3 DEBUG_STACK +zeroentry overflow do_overflow +zeroentry bounds do_bounds +zeroentry invalid_op do_invalid_op +zeroentry device_not_available do_device_not_available +paranoiderrorentry double_fault do_double_fault 0 +zeroentry coprocessor_segment_overrun do_coprocessor_segment_overrun +errorentry invalid_TSS do_invalid_TSS +errorentry segment_not_present do_segment_not_present +paranoiderrorentry stack_segment do_stack_segment +errorentry general_protection do_general_protection 1 +errorentry page_fault do_page_fault 1 +zeroentry spurious_interrupt_bug do_spurious_interrupt_bug +zeroentry coprocessor_error do_coprocessor_error +errorentry alignment_check do_alignment_check +#ifdef CONFIG_X86_MCE +paranoidzeroentry machine_check do_machine_check +#endif +zeroentry simd_coprocessor_error do_simd_coprocessor_error /* * "Paranoid" exit path from exception stack. @@ -1321,26 +1362,7 @@ ENTRY(kernel_execve) CFI_ENDPROC ENDPROC(kernel_execve) -KPROBE_ENTRY(page_fault) - errorentry do_page_fault -KPROBE_END(page_fault) -ENTRY(coprocessor_error) - zeroentry do_coprocessor_error -END(coprocessor_error) - -ENTRY(simd_coprocessor_error) - zeroentry do_simd_coprocessor_error -END(simd_coprocessor_error) - -ENTRY(device_not_available) - zeroentry do_device_not_available -END(device_not_available) - - /* runs on exception stack */ -KPROBE_ENTRY(debug) - paranoidzeroentry_ist do_debug, DEBUG_STACK -KPROBE_END(debug) /* runs on exception stack */ KPROBE_ENTRY(nmi) @@ -1397,67 +1419,6 @@ nmi_schedule: #endif KPROBE_END(nmi) -KPROBE_ENTRY(int3) - paranoidzeroentry_ist do_int3, DEBUG_STACK -KPROBE_END(int3) - -ENTRY(overflow) - zeroentry do_overflow -END(overflow) - -ENTRY(bounds) - zeroentry do_bounds -END(bounds) - -ENTRY(invalid_op) - zeroentry do_invalid_op -END(invalid_op) - -ENTRY(coprocessor_segment_overrun) - zeroentry do_coprocessor_segment_overrun -END(coprocessor_segment_overrun) - - /* runs on exception stack */ -ENTRY(double_fault) - paranoiderrorentry do_double_fault -END(double_fault) - -ENTRY(invalid_TSS) - errorentry do_invalid_TSS -END(invalid_TSS) - -ENTRY(segment_not_present) - errorentry do_segment_not_present -END(segment_not_present) - - /* runs on exception stack */ -ENTRY(stack_segment) - paranoiderrorentry do_stack_segment -END(stack_segment) - -KPROBE_ENTRY(general_protection) - errorentry do_general_protection -KPROBE_END(general_protection) - -ENTRY(alignment_check) - errorentry do_alignment_check -END(alignment_check) - -ENTRY(divide_error) - zeroentry do_divide_error -END(divide_error) - -ENTRY(spurious_interrupt_bug) - zeroentry do_spurious_interrupt_bug -END(spurious_interrupt_bug) - -#ifdef CONFIG_X86_MCE - /* runs on exception stack */ -ENTRY(machine_check) - paranoidzeroentry do_machine_check -END(machine_check) -#endif - /* Call softirq on interrupt stack. Interrupts are off. */ ENTRY(call_softirq) CFI_STARTPROC @@ -1486,9 +1447,7 @@ KPROBE_ENTRY(ignore_sysret) ENDPROC(ignore_sysret) #ifdef CONFIG_XEN -ENTRY(xen_hypervisor_callback) - zeroentry xen_do_hypervisor_callback -END(xen_hypervisor_callback) +zeroentry xen_hypervisor_callback xen_do_hypervisor_callback /* # A note on the "critical region" in our callback handler. -- cgit v1.2.3 From 6efdcfaf16cc4fc76651603e083cf3ec4bd1e6de Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Sun, 23 Nov 2008 10:15:32 +0100 Subject: x86: KPROBE_ENTRY should be paired wth KPROBE_END Impact: move some code out of .kprobes.text KPROBE_ENTRY switches code generation to .kprobes.text, and KPROBE_END uses .popsection to get back to the previous section (.text, normally). Also replace ENDPROC by END, for consistency. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 1a856c0b21a8..f2d546e16354 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1167,7 +1167,7 @@ paranoid_schedule: TRACE_IRQS_OFF jmp paranoid_userspace CFI_ENDPROC -END(paranoid_exit) +KPROBE_END(paranoid_exit) /* * Exception entry point. This expects an error code/orig_rax on the stack. @@ -1259,7 +1259,7 @@ gs_change: CFI_ADJUST_CFA_OFFSET -8 ret CFI_ENDPROC -ENDPROC(native_load_gs_index) +END(native_load_gs_index) .section __ex_table,"a" .align 8 @@ -1313,7 +1313,7 @@ ENTRY(kernel_thread) UNFAKE_STACK_FRAME ret CFI_ENDPROC -ENDPROC(kernel_thread) +END(kernel_thread) child_rip: pushq $0 # fake return address @@ -1329,7 +1329,7 @@ child_rip: mov %eax, %edi call do_exit CFI_ENDPROC -ENDPROC(child_rip) +END(child_rip) /* * execve(). This function needs to use IRET, not SYSRET, to set up all state properly. @@ -1360,9 +1360,7 @@ ENTRY(kernel_execve) UNFAKE_STACK_FRAME ret CFI_ENDPROC -ENDPROC(kernel_execve) - - +END(kernel_execve) /* runs on exception stack */ KPROBE_ENTRY(nmi) @@ -1437,14 +1435,14 @@ ENTRY(call_softirq) decl %gs:pda_irqcount ret CFI_ENDPROC -ENDPROC(call_softirq) +END(call_softirq) KPROBE_ENTRY(ignore_sysret) CFI_STARTPROC mov $-ENOSYS,%eax sysret CFI_ENDPROC -ENDPROC(ignore_sysret) +KPROBE_END(ignore_sysret) #ifdef CONFIG_XEN zeroentry xen_hypervisor_callback xen_do_hypervisor_callback -- cgit v1.2.3 From 5f5db591326779a80cfe490c5d6b6ce9fac08b31 Mon Sep 17 00:00:00 2001 From: jia zhang Date: Sun, 23 Nov 2008 22:47:10 +0800 Subject: x86, debug: remove the confusing entry in call trace Impact: improve backtrace quality avoid the confusion in call trace because of the lack of padding at the tail of function. When do_exit gets called, the return address behind call instruction is pushed into stack. If something get wrong in do_exit, for x86_64, the entry "kernel_execve +0x00/0xXX" rather than "child_rip +0xYY/0xZZ" is in the call trace. That looks confusing, so add a u2d to make the return address still part of the original call site. (This also catches any instances of us returning from that function somehow.) Signed-off-by: jia zhang Acked-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 1 + arch/x86/kernel/entry_64.S | 1 + 2 files changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 28b597ef9ca1..f6402c4ba10d 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1051,6 +1051,7 @@ ENTRY(kernel_thread_helper) push %eax CFI_ADJUST_CFA_OFFSET 4 call do_exit + ud2 # padding for call trace CFI_ENDPROC ENDPROC(kernel_thread_helper) diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index ddeeb1052583..4a16bf31c783 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1172,6 +1172,7 @@ child_rip: # exit mov %eax, %edi call do_exit + ud2 # padding for call trace CFI_ENDPROC ENDPROC(child_rip) -- cgit v1.2.3 From 3b71e9e307b3406aa29960a7428247f8a48b810c Mon Sep 17 00:00:00 2001 From: Hannes Eder Date: Sun, 23 Nov 2008 20:19:33 +0100 Subject: x86: HPET: fix sparse warning Impact: make global variable static Fix this sparse warning: arch/x86/kernel/hpet.c:36:18: warning: symbol 'hpet_num_timers' was not declared. Should it be static? Signed-off-by: Hannes Eder Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 067d8de913f6..15fcaacc1f84 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -33,7 +33,7 @@ * HPET address is set in acpi/boot.c, when an ACPI entry exists */ unsigned long hpet_address; -unsigned long hpet_num_timers; +static unsigned long hpet_num_timers; static void __iomem *hpet_virt_address; struct hpet_dev { -- cgit v1.2.3 From b47b92884212008b4bd044ba6b48b93c00b10ec6 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 24 Nov 2008 00:50:09 -0800 Subject: x86: drop REBOOT_CF9_COND from reboot fallback chain Impact: Reverts sequence of reboot fallbacks Checkin 14d7ca5c575853664d8fe4f225a77b8df1b7de7d changed the default reboot method to "pci", a.k.a. port CF9. Unfortunately this has been shown to cause lockups on at least two systems for which REBOOT_KBD worked, both Thinkpads with Intel chipsets. Checkin 3889d0cea2b73049bdca062d9ff1e5d33468289c reverted the default, but did not revert the fallback chain. This checkin reverts the fallback chain; port CF9 is now only done by explicit "reboot=pci" or a future potential DMI key. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/reboot.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 790b09fbadcb..bb387ab0eea8 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -384,20 +384,20 @@ static void native_machine_emergency_restart(void) load_idt(&no_idt); __asm__ __volatile__("int3"); - reboot_type = BOOT_CF9_COND; + reboot_type = BOOT_KBD; break; #ifdef CONFIG_X86_32 case BOOT_BIOS: machine_real_restart(jump_to_bios, sizeof(jump_to_bios)); - reboot_type = BOOT_CF9_COND; + reboot_type = BOOT_KBD; break; #endif case BOOT_ACPI: acpi_reboot(); - reboot_type = BOOT_CF9_COND; + reboot_type = BOOT_KBD; break; case BOOT_EFI: @@ -406,7 +406,7 @@ static void native_machine_emergency_restart(void) EFI_RESET_WARM : EFI_RESET_COLD, EFI_SUCCESS, 0, NULL); - reboot_type = BOOT_CF9_COND; + reboot_type = BOOT_KBD; break; case BOOT_CF9: -- cgit v1.2.3 From e951e4af2e399c46891004d4931333d2d8d520ab Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 25 Nov 2008 08:42:01 +0100 Subject: x86: fix unused variable warning in arch/x86/kernel/hpet.c MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Impact: fix build warning this warning: arch/x86/kernel/hpet.c:36: warning: ‘hpet_num_timers’ defined but not used Triggers because hpet_num_timers is unused in the !CONFIG_PCI_MSI case. Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 15fcaacc1f84..3f0a3edf0a57 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -33,7 +33,9 @@ * HPET address is set in acpi/boot.c, when an ACPI entry exists */ unsigned long hpet_address; +#ifdef CONFIG_PCI_MSI static unsigned long hpet_num_timers; +#endif static void __iomem *hpet_virt_address; struct hpet_dev { -- cgit v1.2.3 From eff79aee91dd07e944df65fa448c8baeee7709d8 Mon Sep 17 00:00:00 2001 From: Julia Lawall Date: Tue, 25 Nov 2008 14:13:03 +0100 Subject: arch/x86/kernel/pci-calgary_64.c: change simple_strtol to simple_strtoul Impact: fix theoretical option string parsing overflow Since bridge is unsigned, it would seem better to use simple_strtoul that simple_strtol. A simplified version of the semantic patch that makes this change is as follows: (http://www.emn.fr/x-info/coccinelle/) // @r2@ long e; position p; @@ e = simple_strtol@p(...) @@ position p != r2.p; type T; T e; @@ e = - simple_strtol@p + simple_strtoul (...) // Signed-off-by: Julia Lawall Cc: muli@il.ibm.com Cc: jdmason@kudzu.us Cc: discuss@x86-64.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-calgary_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index e1e731d78f38..d28bbdc35e4e 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -1567,7 +1567,7 @@ static int __init calgary_parse_options(char *p) ++p; if (*p == '\0') break; - bridge = simple_strtol(p, &endp, 0); + bridge = simple_strtoul(p, &endp, 0); if (p == endp) break; -- cgit v1.2.3 From 292c669cd7087a090d6420e223eb1072f3e3c50b Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 25 Nov 2008 08:45:13 +0100 Subject: x86, bts: exclude ds.c from build when disabled Impact: cleanup Move the CONFIG guard from the .c file into the makefile. Reported-by: Andi Kleen Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/ds.c | 3 --- 2 files changed, 1 insertion(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index e489ff9cb3e2..b62a7667828e 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -41,7 +41,7 @@ obj-$(CONFIG_X86_TRAMPOLINE) += trampoline.o obj-y += process.o obj-y += i387.o xsave.o obj-y += ptrace.o -obj-y += ds.o +obj-$(CONFIG_X86_DS) += ds.o obj-$(CONFIG_X86_32) += tls.o obj-$(CONFIG_IA32_EMULATION) += tls.o obj-y += step.o diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index d1a121443bde..4c8d57ec9663 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -21,8 +21,6 @@ */ -#ifdef CONFIG_X86_DS - #include #include @@ -878,4 +876,3 @@ void ds_free(struct ds_context *context) while (leftovers--) ds_put_context(context); } -#endif /* CONFIG_X86_DS */ -- cgit v1.2.3 From c4858ffc8f2dc850cb1f609c679b1ac1ad36ef0c Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 25 Nov 2008 08:49:06 +0100 Subject: x86, pebs: fix PEBS record size configuration Impact: fix DS hw enablement on 64-bit x86 Fix the PEBS record size in the DS configuration. Reported-by: Stephane Eranian Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 4c8d57ec9663..04e38ef646af 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -816,13 +816,21 @@ static const struct ds_configuration ds_cfg_var = { .sizeof_ds = sizeof(long) * 12, .sizeof_field = sizeof(long), .sizeof_rec[ds_bts] = sizeof(long) * 3, +#ifdef __i386__ .sizeof_rec[ds_pebs] = sizeof(long) * 10 +#else + .sizeof_rec[ds_pebs] = sizeof(long) * 18 +#endif }; static const struct ds_configuration ds_cfg_64 = { .sizeof_ds = 8 * 12, .sizeof_field = 8, .sizeof_rec[ds_bts] = 8 * 3, +#ifdef __i386__ .sizeof_rec[ds_pebs] = 8 * 10 +#else + .sizeof_rec[ds_pebs] = 8 * 18 +#endif }; static inline void -- cgit v1.2.3 From de90add30e79261c3b5be68bb0f22d2ef98e8113 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 25 Nov 2008 08:52:56 +0100 Subject: x86, bts: fix wrmsr and spinlock over kmalloc Impact: fix sleeping-with-spinlock-held bugs/crashes - Turn a wrmsr to write the DS_AREA MSR into a wrmsrl. - Use irqsave variants of spinlocks. - Do not allocate memory while holding spinlocks. Reported-by: Stephane Eranian Reported-by: Ingo Molnar Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 77 +++++++++++++++++++++++++++------------------------- 1 file changed, 40 insertions(+), 37 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 04e38ef646af..a2d1176c38ee 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -209,14 +209,15 @@ static DEFINE_PER_CPU(struct ds_context *, system_context); static inline struct ds_context *ds_get_context(struct task_struct *task) { struct ds_context *context; + unsigned long irq; - spin_lock(&ds_lock); + spin_lock_irqsave(&ds_lock, irq); context = (task ? task->thread.ds_ctx : this_system_context); if (context) context->count++; - spin_unlock(&ds_lock); + spin_unlock_irqrestore(&ds_lock, irq); return context; } @@ -224,55 +225,46 @@ static inline struct ds_context *ds_get_context(struct task_struct *task) /* * Same as ds_get_context, but allocates the context and it's DS * structure, if necessary; returns NULL; if out of memory. - * - * pre: requires ds_lock to be held */ static inline struct ds_context *ds_alloc_context(struct task_struct *task) { struct ds_context **p_context = (task ? &task->thread.ds_ctx : &this_system_context); struct ds_context *context = *p_context; + unsigned long irq; if (!context) { - spin_unlock(&ds_lock); - context = kzalloc(sizeof(*context), GFP_KERNEL); - - if (!context) { - spin_lock(&ds_lock); + if (!context) return NULL; - } context->ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); if (!context->ds) { kfree(context); - spin_lock(&ds_lock); return NULL; } - spin_lock(&ds_lock); - /* - * Check for race - another CPU could have allocated - * it meanwhile: - */ + spin_lock_irqsave(&ds_lock, irq); + if (*p_context) { kfree(context->ds); kfree(context); - return *p_context; - } - - *p_context = context; - context->this = p_context; - context->task = task; + context = *p_context; + } else { + *p_context = context; - if (task) - set_tsk_thread_flag(task, TIF_DS_AREA_MSR); + context->this = p_context; + context->task = task; - if (!task || (task == current)) - wrmsr(MSR_IA32_DS_AREA, (unsigned long)context->ds, 0); + if (task) + set_tsk_thread_flag(task, TIF_DS_AREA_MSR); - get_tracer(task); + if (!task || (task == current)) + wrmsrl(MSR_IA32_DS_AREA, + (unsigned long)context->ds); + } + spin_unlock_irqrestore(&ds_lock, irq); } context->count++; @@ -286,10 +278,12 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task) */ static inline void ds_put_context(struct ds_context *context) { + unsigned long irq; + if (!context) return; - spin_lock(&ds_lock); + spin_lock_irqsave(&ds_lock, irq); if (--context->count) goto out; @@ -311,7 +305,7 @@ static inline void ds_put_context(struct ds_context *context) kfree(context->ds); kfree(context); out: - spin_unlock(&ds_lock); + spin_unlock_irqrestore(&ds_lock, irq); } @@ -382,6 +376,7 @@ static int ds_request(struct task_struct *task, void *base, size_t size, struct ds_context *context; unsigned long buffer, adj; const unsigned long alignment = (1 << 3); + unsigned long irq; int error = 0; if (!ds_cfg.sizeof_ds) @@ -396,26 +391,27 @@ static int ds_request(struct task_struct *task, void *base, size_t size, return -EOPNOTSUPP; - spin_lock(&ds_lock); - - error = -ENOMEM; context = ds_alloc_context(task); if (!context) - goto out_unlock; + return -ENOMEM; + + spin_lock_irqsave(&ds_lock, irq); error = -EPERM; if (!check_tracer(task)) goto out_unlock; + get_tracer(task); + error = -EALREADY; if (context->owner[qual] == current) - goto out_unlock; + goto out_put_tracer; error = -EPERM; if (context->owner[qual] != NULL) - goto out_unlock; + goto out_put_tracer; context->owner[qual] = current; - spin_unlock(&ds_lock); + spin_unlock_irqrestore(&ds_lock, irq); error = -ENOMEM; @@ -463,10 +459,17 @@ static int ds_request(struct task_struct *task, void *base, size_t size, out_release: context->owner[qual] = NULL; ds_put_context(context); + put_tracer(task); + return error; + + out_put_tracer: + spin_unlock_irqrestore(&ds_lock, irq); + ds_put_context(context); + put_tracer(task); return error; out_unlock: - spin_unlock(&ds_lock); + spin_unlock_irqrestore(&ds_lock, irq); ds_put_context(context); return error; } -- cgit v1.2.3 From ca0002a179bfa532d009a9272d619732872c49bd Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 25 Nov 2008 09:01:25 +0100 Subject: x86, bts: base in-kernel ds interface on handles Impact: generalize the DS code to shared buffers Change the in-kernel ds.h interface to identify the tracer via a handle returned on ds_request_~(). Tracers used to be identified via their task_struct. The changes are required to allow DS to be shared between different tasks, which is needed for perfmon2 and for ftrace. For ptrace, the handle is stored in the traced task's task_struct. This should probably go into a (arch-specific) ptrace context some time. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 679 ++++++++++++++++++++++++----------------------- arch/x86/kernel/ptrace.c | 73 ++--- 2 files changed, 380 insertions(+), 372 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index d6938d9351cf..96768e9cce99 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -28,6 +28,7 @@ #include #include #include +#include /* @@ -44,6 +45,35 @@ struct ds_configuration { }; static struct ds_configuration ds_cfg; +/* + * A BTS or PEBS tracer. + * + * This holds the configuration of the tracer and serves as a handle + * to identify tracers. + */ +struct ds_tracer { + /* the DS context (partially) owned by this tracer */ + struct ds_context *context; + /* the buffer provided on ds_request() and its size in bytes */ + void *buffer; + size_t size; + /* the number of allocated pages for on-request allocated buffers */ + unsigned int pages; +}; + +struct bts_tracer { + /* the common DS part */ + struct ds_tracer ds; + /* buffer overflow notification function */ + bts_ovfl_callback_t ovfl; +}; + +struct pebs_tracer { + /* the common DS part */ + struct ds_tracer ds; + /* buffer overflow notification function */ + pebs_ovfl_callback_t ovfl; +}; /* * Debug Store (DS) save area configuration (see Intel64 and IA32 @@ -107,35 +137,15 @@ static inline void ds_set(unsigned char *base, enum ds_qualifier qual, (*(unsigned long *)base) = value; } +#define DS_ALIGNMENT (1 << 3) /* BTS and PEBS buffer alignment */ + /* * Locking is done only for allocating BTS or PEBS resources and for * guarding context and buffer memory allocation. - * - * Most functions require the current task to own the ds context part - * they are going to access. All the locking is done when validating - * access to the context. */ static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); -/* - * Validate that the current task is allowed to access the BTS/PEBS - * buffer of the parameter task. - * - * Returns 0, if access is granted; -Eerrno, otherwise. - */ -static inline int ds_validate_access(struct ds_context *context, - enum ds_qualifier qual) -{ - if (!context) - return -EPERM; - - if (context->owner[qual] == current) - return 0; - - return -EPERM; -} - /* * We either support (system-wide) per-cpu or per-thread allocation. @@ -183,50 +193,12 @@ static inline int check_tracer(struct task_struct *task) * * Contexts are use-counted. They are allocated on first access and * deallocated when the last user puts the context. - * - * We distinguish between an allocating and a non-allocating get of a - * context: - * - the allocating get is used for requesting BTS/PEBS resources. It - * requires the caller to hold the global ds_lock. - * - the non-allocating get is used for all other cases. A - * non-existing context indicates an error. It acquires and releases - * the ds_lock itself for obtaining the context. - * - * A context and its DS configuration are allocated and deallocated - * together. A context always has a DS configuration of the - * appropriate size. */ static DEFINE_PER_CPU(struct ds_context *, system_context); #define this_system_context per_cpu(system_context, smp_processor_id()) -/* - * Returns the pointer to the parameter task's context or to the - * system-wide context, if task is NULL. - * - * Increases the use count of the returned context, if not NULL. - */ static inline struct ds_context *ds_get_context(struct task_struct *task) -{ - struct ds_context *context; - unsigned long irq; - - spin_lock_irqsave(&ds_lock, irq); - - context = (task ? task->thread.ds_ctx : this_system_context); - if (context) - context->count++; - - spin_unlock_irqrestore(&ds_lock, irq); - - return context; -} - -/* - * Same as ds_get_context, but allocates the context and it's DS - * structure, if necessary; returns NULL; if out of memory. - */ -static inline struct ds_context *ds_alloc_context(struct task_struct *task) { struct ds_context **p_context = (task ? &task->thread.ds_ctx : &this_system_context); @@ -238,16 +210,9 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task) if (!context) return NULL; - context->ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); - if (!context->ds) { - kfree(context); - return NULL; - } - spin_lock_irqsave(&ds_lock, irq); if (*p_context) { - kfree(context->ds); kfree(context); context = *p_context; @@ -272,10 +237,6 @@ static inline struct ds_context *ds_alloc_context(struct task_struct *task) return context; } -/* - * Decreases the use count of the parameter context, if not NULL. - * Deallocates the context, if the use count reaches zero. - */ static inline void ds_put_context(struct ds_context *context) { unsigned long irq; @@ -296,13 +257,6 @@ static inline void ds_put_context(struct ds_context *context) if (!context->task || (context->task == current)) wrmsrl(MSR_IA32_DS_AREA, 0); - put_tracer(context->task); - - /* free any leftover buffers from tracers that did not - * deallocate them properly. */ - kfree(context->buffer[ds_bts]); - kfree(context->buffer[ds_pebs]); - kfree(context->ds); kfree(context); out: spin_unlock_irqrestore(&ds_lock, irq); @@ -312,21 +266,29 @@ static inline void ds_put_context(struct ds_context *context) /* * Handle a buffer overflow * - * task: the task whose buffers are overflowing; - * NULL for a buffer overflow on the current cpu * context: the ds context * qual: the buffer type */ -static void ds_overflow(struct task_struct *task, struct ds_context *context, - enum ds_qualifier qual) -{ - if (!context) - return; - - if (context->callback[qual]) - (*context->callback[qual])(task); - - /* todo: do some more overflow handling */ +static void ds_overflow(struct ds_context *context, enum ds_qualifier qual) +{ + switch (qual) { + case ds_bts: { + struct bts_tracer *tracer = + container_of(context->owner[qual], + struct bts_tracer, ds); + if (tracer->ovfl) + tracer->ovfl(tracer); + } + break; + case ds_pebs: { + struct pebs_tracer *tracer = + container_of(context->owner[qual], + struct pebs_tracer, ds); + if (tracer->ovfl) + tracer->ovfl(tracer); + } + break; + } } @@ -343,23 +305,25 @@ static void ds_overflow(struct task_struct *task, struct ds_context *context, static inline void *ds_allocate_buffer(size_t size, unsigned int *pages) { unsigned long rlim, vm, pgsz; - void *buffer; + void *buffer = NULL; pgsz = PAGE_ALIGN(size) >> PAGE_SHIFT; + down_write(¤t->mm->mmap_sem); + rlim = current->signal->rlim[RLIMIT_AS].rlim_cur >> PAGE_SHIFT; vm = current->mm->total_vm + pgsz; if (rlim < vm) - return NULL; + goto out; rlim = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT; vm = current->mm->locked_vm + pgsz; if (rlim < vm) - return NULL; + goto out; buffer = kzalloc(size, GFP_KERNEL); if (!buffer) - return NULL; + goto out; current->mm->total_vm += pgsz; current->mm->locked_vm += pgsz; @@ -367,290 +331,337 @@ static inline void *ds_allocate_buffer(size_t size, unsigned int *pages) if (pages) *pages = pgsz; + out: + up_write(¤t->mm->mmap_sem); return buffer; } -static int ds_request(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl, enum ds_qualifier qual) +static void ds_install_ds_config(struct ds_context *context, + enum ds_qualifier qual, + void *base, size_t size, size_t ith) { - struct ds_context *context; unsigned long buffer, adj; - const unsigned long alignment = (1 << 3); + + /* adjust the buffer address and size to meet alignment + * constraints: + * - buffer is double-word aligned + * - size is multiple of record size + * + * We checked the size at the very beginning; we have enough + * space to do the adjustment. + */ + buffer = (unsigned long)base; + + adj = ALIGN(buffer, DS_ALIGNMENT) - buffer; + buffer += adj; + size -= adj; + + size /= ds_cfg.sizeof_rec[qual]; + size *= ds_cfg.sizeof_rec[qual]; + + ds_set(context->ds, qual, ds_buffer_base, buffer); + ds_set(context->ds, qual, ds_index, buffer); + ds_set(context->ds, qual, ds_absolute_maximum, buffer + size); + + /* The value for 'no threshold' is -1, which will set the + * threshold outside of the buffer, just like we want it. + */ + ds_set(context->ds, qual, + ds_interrupt_threshold, buffer + size - ith); +} + +static int ds_request(struct ds_tracer *tracer, enum ds_qualifier qual, + struct task_struct *task, + void *base, size_t size, size_t th) +{ + struct ds_context *context; unsigned long irq; - int error = 0; + int error; + error = -EOPNOTSUPP; if (!ds_cfg.sizeof_ds) - return -EOPNOTSUPP; + goto out; /* we require some space to do alignment adjustments below */ - if (size < (alignment + ds_cfg.sizeof_rec[qual])) - return -EINVAL; + error = -EINVAL; + if (size < (DS_ALIGNMENT + ds_cfg.sizeof_rec[qual])) + goto out; - /* buffer overflow notification is not yet implemented */ - if (ovfl) - return -EOPNOTSUPP; + if (th != (size_t)-1) { + th *= ds_cfg.sizeof_rec[qual]; + + error = -EINVAL; + if (size <= th) + goto out; + } + + error = -ENOMEM; + if (!base) { + base = ds_allocate_buffer(size, &tracer->pages); + if (!base) + goto out; + } + tracer->buffer = base; + tracer->size = size; - context = ds_alloc_context(task); + error = -ENOMEM; + context = ds_get_context(task); if (!context) - return -ENOMEM; + goto out; + tracer->context = context; + spin_lock_irqsave(&ds_lock, irq); error = -EPERM; if (!check_tracer(task)) goto out_unlock; - get_tracer(task); - error = -EALREADY; - if (context->owner[qual] == current) - goto out_put_tracer; error = -EPERM; - if (context->owner[qual] != NULL) + if (context->owner[qual]) goto out_put_tracer; - context->owner[qual] = current; + context->owner[qual] = tracer; spin_unlock_irqrestore(&ds_lock, irq); - error = -ENOMEM; - if (!base) { - base = ds_allocate_buffer(size, &context->pages[qual]); - if (!base) - goto out_release; - - context->buffer[qual] = base; - } - error = 0; + ds_install_ds_config(context, qual, base, size, th); - context->callback[qual] = ovfl; - - /* adjust the buffer address and size to meet alignment - * constraints: - * - buffer is double-word aligned - * - size is multiple of record size - * - * We checked the size at the very beginning; we have enough - * space to do the adjustment. - */ - buffer = (unsigned long)base; - - adj = ALIGN(buffer, alignment) - buffer; - buffer += adj; - size -= adj; - - size /= ds_cfg.sizeof_rec[qual]; - size *= ds_cfg.sizeof_rec[qual]; - - ds_set(context->ds, qual, ds_buffer_base, buffer); - ds_set(context->ds, qual, ds_index, buffer); - ds_set(context->ds, qual, ds_absolute_maximum, buffer + size); - - if (ovfl) { - /* todo: select a suitable interrupt threshold */ - } else - ds_set(context->ds, qual, - ds_interrupt_threshold, buffer + size + 1); - - /* we keep the context until ds_release */ - return error; - - out_release: - context->owner[qual] = NULL; - ds_put_context(context); - put_tracer(task); - return error; + return 0; out_put_tracer: - spin_unlock_irqrestore(&ds_lock, irq); - ds_put_context(context); put_tracer(task); - return error; - out_unlock: spin_unlock_irqrestore(&ds_lock, irq); ds_put_context(context); + tracer->context = NULL; + out: return error; } -int ds_request_bts(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl) +struct bts_tracer *ds_request_bts(struct task_struct *task, + void *base, size_t size, + bts_ovfl_callback_t ovfl, size_t th) { - return ds_request(task, base, size, ovfl, ds_bts); -} + struct bts_tracer *tracer; + int error; -int ds_request_pebs(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl) -{ - return ds_request(task, base, size, ovfl, ds_pebs); + /* buffer overflow notification is not yet implemented */ + error = -EOPNOTSUPP; + if (ovfl) + goto out; + + error = -ENOMEM; + tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); + if (!tracer) + goto out; + tracer->ovfl = ovfl; + + error = ds_request(&tracer->ds, ds_bts, task, base, size, th); + if (error < 0) + goto out_tracer; + + return tracer; + + out_tracer: + (void)ds_release_bts(tracer); + out: + return ERR_PTR(error); } -static int ds_release(struct task_struct *task, enum ds_qualifier qual) +struct pebs_tracer *ds_request_pebs(struct task_struct *task, + void *base, size_t size, + pebs_ovfl_callback_t ovfl, size_t th) { - struct ds_context *context; + struct pebs_tracer *tracer; int error; - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) + /* buffer overflow notification is not yet implemented */ + error = -EOPNOTSUPP; + if (ovfl) goto out; - kfree(context->buffer[qual]); - context->buffer[qual] = NULL; + error = -ENOMEM; + tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); + if (!tracer) + goto out; + tracer->ovfl = ovfl; - current->mm->total_vm -= context->pages[qual]; - current->mm->locked_vm -= context->pages[qual]; - context->pages[qual] = 0; - context->owner[qual] = NULL; + error = ds_request(&tracer->ds, ds_pebs, task, base, size, th); + if (error < 0) + goto out_tracer; - /* - * we put the context twice: - * once for the ds_get_context - * once for the corresponding ds_request - */ - ds_put_context(context); + return tracer; + + out_tracer: + (void)ds_release_pebs(tracer); out: - ds_put_context(context); - return error; + return ERR_PTR(error); +} + +static void ds_release(struct ds_tracer *tracer, enum ds_qualifier qual) +{ + if (tracer->context) { + BUG_ON(tracer->context->owner[qual] != tracer); + tracer->context->owner[qual] = NULL; + + put_tracer(tracer->context->task); + ds_put_context(tracer->context); + } + + if (tracer->pages) { + kfree(tracer->buffer); + + down_write(¤t->mm->mmap_sem); + + current->mm->total_vm -= tracer->pages; + current->mm->locked_vm -= tracer->pages; + + up_write(¤t->mm->mmap_sem); + } } -int ds_release_bts(struct task_struct *task) +int ds_release_bts(struct bts_tracer *tracer) { - return ds_release(task, ds_bts); + if (!tracer) + return -EINVAL; + + ds_release(&tracer->ds, ds_bts); + kfree(tracer); + + return 0; } -int ds_release_pebs(struct task_struct *task) +int ds_release_pebs(struct pebs_tracer *tracer) { - return ds_release(task, ds_pebs); + if (!tracer) + return -EINVAL; + + ds_release(&tracer->ds, ds_pebs); + kfree(tracer); + + return 0; } -static int ds_get_index(struct task_struct *task, size_t *pos, - enum ds_qualifier qual) +static size_t ds_get_index(struct ds_context *context, enum ds_qualifier qual) { - struct ds_context *context; unsigned long base, index; - int error; - - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; base = ds_get(context->ds, qual, ds_buffer_base); index = ds_get(context->ds, qual, ds_index); - error = ((index - base) / ds_cfg.sizeof_rec[qual]); - if (pos) - *pos = error; - out: - ds_put_context(context); - return error; + return (index - base) / ds_cfg.sizeof_rec[qual]; } -int ds_get_bts_index(struct task_struct *task, size_t *pos) +int ds_get_bts_index(struct bts_tracer *tracer, size_t *pos) { - return ds_get_index(task, pos, ds_bts); + if (!tracer) + return -EINVAL; + + if (!pos) + return -EINVAL; + + *pos = ds_get_index(tracer->ds.context, ds_bts); + + return 0; } -int ds_get_pebs_index(struct task_struct *task, size_t *pos) +int ds_get_pebs_index(struct pebs_tracer *tracer, size_t *pos) { - return ds_get_index(task, pos, ds_pebs); + if (!tracer) + return -EINVAL; + + if (!pos) + return -EINVAL; + + *pos = ds_get_index(tracer->ds.context, ds_pebs); + + return 0; } -static int ds_get_end(struct task_struct *task, size_t *pos, - enum ds_qualifier qual) +static size_t ds_get_end(struct ds_context *context, enum ds_qualifier qual) { - struct ds_context *context; - unsigned long base, end; - int error; - - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; + unsigned long base, max; base = ds_get(context->ds, qual, ds_buffer_base); - end = ds_get(context->ds, qual, ds_absolute_maximum); + max = ds_get(context->ds, qual, ds_absolute_maximum); - error = ((end - base) / ds_cfg.sizeof_rec[qual]); - if (pos) - *pos = error; - out: - ds_put_context(context); - return error; + return (max - base) / ds_cfg.sizeof_rec[qual]; } -int ds_get_bts_end(struct task_struct *task, size_t *pos) +int ds_get_bts_end(struct bts_tracer *tracer, size_t *pos) { - return ds_get_end(task, pos, ds_bts); + if (!tracer) + return -EINVAL; + + if (!pos) + return -EINVAL; + + *pos = ds_get_end(tracer->ds.context, ds_bts); + + return 0; } -int ds_get_pebs_end(struct task_struct *task, size_t *pos) +int ds_get_pebs_end(struct pebs_tracer *tracer, size_t *pos) { - return ds_get_end(task, pos, ds_pebs); + if (!tracer) + return -EINVAL; + + if (!pos) + return -EINVAL; + + *pos = ds_get_end(tracer->ds.context, ds_pebs); + + return 0; } -static int ds_access(struct task_struct *task, size_t index, - const void **record, enum ds_qualifier qual) +static int ds_access(struct ds_context *context, enum ds_qualifier qual, + size_t index, const void **record) { - struct ds_context *context; unsigned long base, idx; - int error; if (!record) return -EINVAL; - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; - base = ds_get(context->ds, qual, ds_buffer_base); idx = base + (index * ds_cfg.sizeof_rec[qual]); - error = -EINVAL; if (idx > ds_get(context->ds, qual, ds_absolute_maximum)) - goto out; + return -EINVAL; *record = (const void *)idx; - error = ds_cfg.sizeof_rec[qual]; - out: - ds_put_context(context); - return error; + + return ds_cfg.sizeof_rec[qual]; } -int ds_access_bts(struct task_struct *task, size_t index, const void **record) +int ds_access_bts(struct bts_tracer *tracer, size_t index, + const void **record) { - return ds_access(task, index, record, ds_bts); + if (!tracer) + return -EINVAL; + + return ds_access(tracer->ds.context, ds_bts, index, record); } -int ds_access_pebs(struct task_struct *task, size_t index, const void **record) +int ds_access_pebs(struct pebs_tracer *tracer, size_t index, + const void **record) { - return ds_access(task, index, record, ds_pebs); + if (!tracer) + return -EINVAL; + + return ds_access(tracer->ds.context, ds_pebs, index, record); } -static int ds_write(struct task_struct *task, const void *record, size_t size, - enum ds_qualifier qual, int force) +static int ds_write(struct ds_context *context, enum ds_qualifier qual, + const void *record, size_t size) { - struct ds_context *context; - int error; + int bytes_written = 0; if (!record) return -EINVAL; - error = -EPERM; - context = ds_get_context(task); - if (!context) - goto out; - - if (!force) { - error = ds_validate_access(context, qual); - if (error < 0) - goto out; - } - - error = 0; while (size) { unsigned long base, index, end, write_end, int_th; unsigned long write_size, adj_write_size; @@ -678,14 +689,14 @@ static int ds_write(struct task_struct *task, const void *record, size_t size, write_end = end; if (write_end <= index) - goto out; + break; write_size = min((unsigned long) size, write_end - index); memcpy((void *)index, record, write_size); record = (const char *)record + write_size; - size -= write_size; - error += write_size; + size -= write_size; + bytes_written += write_size; adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; adj_write_size *= ds_cfg.sizeof_rec[qual]; @@ -700,47 +711,32 @@ static int ds_write(struct task_struct *task, const void *record, size_t size, ds_set(context->ds, qual, ds_index, index); if (index >= int_th) - ds_overflow(task, context, qual); + ds_overflow(context, qual); } - out: - ds_put_context(context); - return error; + return bytes_written; } -int ds_write_bts(struct task_struct *task, const void *record, size_t size) +int ds_write_bts(struct bts_tracer *tracer, const void *record, size_t size) { - return ds_write(task, record, size, ds_bts, /* force = */ 0); -} + if (!tracer) + return -EINVAL; -int ds_write_pebs(struct task_struct *task, const void *record, size_t size) -{ - return ds_write(task, record, size, ds_pebs, /* force = */ 0); + return ds_write(tracer->ds.context, ds_bts, record, size); } -int ds_unchecked_write_bts(struct task_struct *task, - const void *record, size_t size) +int ds_write_pebs(struct pebs_tracer *tracer, const void *record, size_t size) { - return ds_write(task, record, size, ds_bts, /* force = */ 1); -} + if (!tracer) + return -EINVAL; -int ds_unchecked_write_pebs(struct task_struct *task, - const void *record, size_t size) -{ - return ds_write(task, record, size, ds_pebs, /* force = */ 1); + return ds_write(tracer->ds.context, ds_pebs, record, size); } -static int ds_reset_or_clear(struct task_struct *task, - enum ds_qualifier qual, int clear) +static void ds_reset_or_clear(struct ds_context *context, + enum ds_qualifier qual, int clear) { - struct ds_context *context; unsigned long base, end; - int error; - - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; base = ds_get(context->ds, qual, ds_buffer_base); end = ds_get(context->ds, qual, ds_absolute_maximum); @@ -749,70 +745,69 @@ static int ds_reset_or_clear(struct task_struct *task, memset((void *)base, 0, end - base); ds_set(context->ds, qual, ds_index, base); - - error = 0; - out: - ds_put_context(context); - return error; } -int ds_reset_bts(struct task_struct *task) +int ds_reset_bts(struct bts_tracer *tracer) { - return ds_reset_or_clear(task, ds_bts, /* clear = */ 0); + if (!tracer) + return -EINVAL; + + ds_reset_or_clear(tracer->ds.context, ds_bts, /* clear = */ 0); + + return 0; } -int ds_reset_pebs(struct task_struct *task) +int ds_reset_pebs(struct pebs_tracer *tracer) { - return ds_reset_or_clear(task, ds_pebs, /* clear = */ 0); + if (!tracer) + return -EINVAL; + + ds_reset_or_clear(tracer->ds.context, ds_pebs, /* clear = */ 0); + + return 0; } -int ds_clear_bts(struct task_struct *task) +int ds_clear_bts(struct bts_tracer *tracer) { - return ds_reset_or_clear(task, ds_bts, /* clear = */ 1); + if (!tracer) + return -EINVAL; + + ds_reset_or_clear(tracer->ds.context, ds_bts, /* clear = */ 1); + + return 0; } -int ds_clear_pebs(struct task_struct *task) +int ds_clear_pebs(struct pebs_tracer *tracer) { - return ds_reset_or_clear(task, ds_pebs, /* clear = */ 1); + if (!tracer) + return -EINVAL; + + ds_reset_or_clear(tracer->ds.context, ds_pebs, /* clear = */ 1); + + return 0; } -int ds_get_pebs_reset(struct task_struct *task, u64 *value) +int ds_get_pebs_reset(struct pebs_tracer *tracer, u64 *value) { - struct ds_context *context; - int error; + if (!tracer) + return -EINVAL; if (!value) return -EINVAL; - context = ds_get_context(task); - error = ds_validate_access(context, ds_pebs); - if (error < 0) - goto out; - - *value = *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)); + *value = *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)); - error = 0; - out: - ds_put_context(context); - return error; + return 0; } -int ds_set_pebs_reset(struct task_struct *task, u64 value) +int ds_set_pebs_reset(struct pebs_tracer *tracer, u64 value) { - struct ds_context *context; - int error; - - context = ds_get_context(task); - error = ds_validate_access(context, ds_pebs); - if (error < 0) - goto out; + if (!tracer) + return -EINVAL; - *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)) = value; + *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)) = value; - error = 0; - out: - ds_put_context(context); - return error; + return 0; } static const struct ds_configuration ds_cfg_var = { @@ -840,6 +835,10 @@ static inline void ds_configure(const struct ds_configuration *cfg) { ds_cfg = *cfg; + + printk(KERN_INFO "DS available\n"); + + BUG_ON(MAX_SIZEOF_DS < ds_cfg.sizeof_ds); } void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) @@ -883,6 +882,8 @@ void ds_free(struct ds_context *context) * is dying. There should not be any user of that context left * to disturb us, anymore. */ unsigned long leftovers = context->count; - while (leftovers--) + while (leftovers--) { + put_tracer(context->task); ds_put_context(context); + } } diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 06180dff5b2e..76adf5b640ff 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -668,14 +668,14 @@ static int ptrace_bts_read_record(struct task_struct *child, size_t index, size_t bts_index, bts_end; int error; - error = ds_get_bts_end(child, &bts_end); + error = ds_get_bts_end(child->bts, &bts_end); if (error < 0) return error; if (bts_end <= index) return -EINVAL; - error = ds_get_bts_index(child, &bts_index); + error = ds_get_bts_index(child->bts, &bts_index); if (error < 0) return error; @@ -684,7 +684,7 @@ static int ptrace_bts_read_record(struct task_struct *child, size_t index, if (bts_end <= bts_index) bts_index -= bts_end; - error = ds_access_bts(child, bts_index, &bts_record); + error = ds_access_bts(child->bts, bts_index, &bts_record); if (error < 0) return error; @@ -705,14 +705,14 @@ static int ptrace_bts_drain(struct task_struct *child, size_t end, i; int error; - error = ds_get_bts_index(child, &end); + error = ds_get_bts_index(child->bts, &end); if (error < 0) return error; if (size < (end * sizeof(struct bts_struct))) return -EIO; - error = ds_access_bts(child, 0, (const void **)&raw); + error = ds_access_bts(child->bts, 0, (const void **)&raw); if (error < 0) return error; @@ -723,18 +723,13 @@ static int ptrace_bts_drain(struct task_struct *child, return -EFAULT; } - error = ds_clear_bts(child); + error = ds_clear_bts(child->bts); if (error < 0) return error; return end; } -static void ptrace_bts_ovfl(struct task_struct *child) -{ - send_sig(child->thread.bts_ovfl_signal, child, 0); -} - static int ptrace_bts_config(struct task_struct *child, long cfg_size, const struct ptrace_bts_config __user *ucfg) @@ -760,23 +755,29 @@ static int ptrace_bts_config(struct task_struct *child, goto errout; if (cfg.flags & PTRACE_BTS_O_ALLOC) { - ds_ovfl_callback_t ovfl = NULL; + bts_ovfl_callback_t ovfl = NULL; unsigned int sig = 0; - /* we ignore the error in case we were not tracing child */ - (void)ds_release_bts(child); - if (cfg.flags & PTRACE_BTS_O_SIGNAL) { if (!cfg.signal) goto errout; + error = -EOPNOTSUPP; + goto errout; + sig = cfg.signal; - ovfl = ptrace_bts_ovfl; } - error = ds_request_bts(child, /* base = */ NULL, cfg.size, ovfl); - if (error < 0) + if (child->bts) + (void)ds_release_bts(child->bts); + + child->bts = ds_request_bts(child, /* base = */ NULL, cfg.size, + ovfl, /* th = */ (size_t)-1); + if (IS_ERR(child->bts)) { + error = PTR_ERR(child->bts); + child->bts = NULL; goto errout; + } child->thread.bts_ovfl_signal = sig; } @@ -823,15 +824,15 @@ static int ptrace_bts_status(struct task_struct *child, if (cfg_size < sizeof(cfg)) return -EIO; - error = ds_get_bts_end(child, &end); + error = ds_get_bts_end(child->bts, &end); if (error < 0) return error; - error = ds_access_bts(child, /* index = */ 0, &base); + error = ds_access_bts(child->bts, /* index = */ 0, &base); if (error < 0) return error; - error = ds_access_bts(child, /* index = */ end, &max); + error = ds_access_bts(child->bts, /* index = */ end, &max); if (error < 0) return error; @@ -884,10 +885,7 @@ static int ptrace_bts_write_record(struct task_struct *child, return -EINVAL; } - /* The writing task will be the switched-to task on a context - * switch. It needs to write into the switched-from task's BTS - * buffer. */ - return ds_unchecked_write_bts(child, bts_record, bts_cfg.sizeof_bts); + return ds_write_bts(child->bts, bts_record, bts_cfg.sizeof_bts); } void ptrace_bts_take_timestamp(struct task_struct *tsk, @@ -972,13 +970,15 @@ void ptrace_disable(struct task_struct *child) clear_tsk_thread_flag(child, TIF_SYSCALL_EMU); #endif #ifdef CONFIG_X86_PTRACE_BTS - (void)ds_release_bts(child); + if (child->bts) { + (void)ds_release_bts(child->bts); - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; + if (!child->thread.debugctlmsr) + clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); + clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); + } #endif /* CONFIG_X86_PTRACE_BTS */ } @@ -1110,9 +1110,16 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) (child, data, (struct ptrace_bts_config __user *)addr); break; - case PTRACE_BTS_SIZE: - ret = ds_get_bts_index(child, /* pos = */ NULL); + case PTRACE_BTS_SIZE: { + size_t size; + + ret = ds_get_bts_index(child->bts, &size); + if (ret == 0) { + BUG_ON(size != (int) size); + ret = (int) size; + } break; + } case PTRACE_BTS_GET: ret = ptrace_bts_read_record @@ -1120,7 +1127,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_CLEAR: - ret = ds_clear_bts(child); + ret = ds_clear_bts(child->bts); break; case PTRACE_BTS_DRAIN: -- cgit v1.2.3 From 6abb11aecd888d1da6276399380b7355f127c006 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 25 Nov 2008 09:05:27 +0100 Subject: x86, bts, ptrace: move BTS buffer allocation from ds.c into ptrace.c Impact: restructure DS memory allocation to be done by the usage site of DS Require pre-allocated buffers in ds.h. Move the BTS buffer allocation for ptrace into ptrace.c. The pointer to the allocated buffer is stored in the traced task's task_struct together with the handle returned by ds_request_bts(). Removes memory accounting code. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 92 ++++++++---------------------------------------- arch/x86/kernel/ptrace.c | 22 ++++++++++-- 2 files changed, 34 insertions(+), 80 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 96768e9cce99..19a8c2c0389f 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -7,13 +7,12 @@ * * It manages: * - per-thread and per-cpu allocation of BTS and PEBS - * - buffer memory allocation (optional) - * - buffer overflow handling + * - buffer overflow handling (to be done) * - buffer access * * It assumes: - * - get_task_struct on all parameter tasks - * - current is allowed to trace parameter tasks + * - get_task_struct on all traced tasks + * - current is allowed to trace tasks * * * Copyright (C) 2007-2008 Intel Corporation. @@ -57,8 +56,6 @@ struct ds_tracer { /* the buffer provided on ds_request() and its size in bytes */ void *buffer; size_t size; - /* the number of allocated pages for on-request allocated buffers */ - unsigned int pages; }; struct bts_tracer { @@ -141,8 +138,7 @@ static inline void ds_set(unsigned char *base, enum ds_qualifier qual, /* - * Locking is done only for allocating BTS or PEBS resources and for - * guarding context and buffer memory allocation. + * Locking is done only for allocating BTS or PEBS resources. */ static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); @@ -292,50 +288,6 @@ static void ds_overflow(struct ds_context *context, enum ds_qualifier qual) } -/* - * Allocate a non-pageable buffer of the parameter size. - * Checks the memory and the locked memory rlimit. - * - * Returns the buffer, if successful; - * NULL, if out of memory or rlimit exceeded. - * - * size: the requested buffer size in bytes - * pages (out): if not NULL, contains the number of pages reserved - */ -static inline void *ds_allocate_buffer(size_t size, unsigned int *pages) -{ - unsigned long rlim, vm, pgsz; - void *buffer = NULL; - - pgsz = PAGE_ALIGN(size) >> PAGE_SHIFT; - - down_write(¤t->mm->mmap_sem); - - rlim = current->signal->rlim[RLIMIT_AS].rlim_cur >> PAGE_SHIFT; - vm = current->mm->total_vm + pgsz; - if (rlim < vm) - goto out; - - rlim = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT; - vm = current->mm->locked_vm + pgsz; - if (rlim < vm) - goto out; - - buffer = kzalloc(size, GFP_KERNEL); - if (!buffer) - goto out; - - current->mm->total_vm += pgsz; - current->mm->locked_vm += pgsz; - - if (pages) - *pages = pgsz; - - out: - up_write(¤t->mm->mmap_sem); - return buffer; -} - static void ds_install_ds_config(struct ds_context *context, enum ds_qualifier qual, void *base, size_t size, size_t ith) @@ -382,6 +334,10 @@ static int ds_request(struct ds_tracer *tracer, enum ds_qualifier qual, if (!ds_cfg.sizeof_ds) goto out; + error = -EINVAL; + if (!base) + goto out; + /* we require some space to do alignment adjustments below */ error = -EINVAL; if (size < (DS_ALIGNMENT + ds_cfg.sizeof_rec[qual])) @@ -395,13 +351,6 @@ static int ds_request(struct ds_tracer *tracer, enum ds_qualifier qual, goto out; } - error = -ENOMEM; - if (!base) { - base = ds_allocate_buffer(size, &tracer->pages); - if (!base) - goto out; - } - tracer->buffer = base; tracer->size = size; @@ -466,7 +415,7 @@ struct bts_tracer *ds_request_bts(struct task_struct *task, return tracer; out_tracer: - (void)ds_release_bts(tracer); + kfree(tracer); out: return ERR_PTR(error); } @@ -496,31 +445,18 @@ struct pebs_tracer *ds_request_pebs(struct task_struct *task, return tracer; out_tracer: - (void)ds_release_pebs(tracer); + kfree(tracer); out: return ERR_PTR(error); } static void ds_release(struct ds_tracer *tracer, enum ds_qualifier qual) { - if (tracer->context) { - BUG_ON(tracer->context->owner[qual] != tracer); - tracer->context->owner[qual] = NULL; - - put_tracer(tracer->context->task); - ds_put_context(tracer->context); - } + BUG_ON(tracer->context->owner[qual] != tracer); + tracer->context->owner[qual] = NULL; - if (tracer->pages) { - kfree(tracer->buffer); - - down_write(¤t->mm->mmap_sem); - - current->mm->total_vm -= tracer->pages; - current->mm->locked_vm -= tracer->pages; - - up_write(¤t->mm->mmap_sem); - } + put_tracer(tracer->context->task); + ds_put_context(tracer->context); } int ds_release_bts(struct bts_tracer *tracer) diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 76adf5b640ff..2c8ec1ba75e6 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -758,6 +758,10 @@ static int ptrace_bts_config(struct task_struct *child, bts_ovfl_callback_t ovfl = NULL; unsigned int sig = 0; + error = -EINVAL; + if (cfg.size < (10 * bts_cfg.sizeof_bts)) + goto errout; + if (cfg.flags & PTRACE_BTS_O_SIGNAL) { if (!cfg.signal) goto errout; @@ -768,14 +772,26 @@ static int ptrace_bts_config(struct task_struct *child, sig = cfg.signal; } - if (child->bts) + if (child->bts) { (void)ds_release_bts(child->bts); + kfree(child->bts_buffer); + + child->bts = NULL; + child->bts_buffer = NULL; + } + + error = -ENOMEM; + child->bts_buffer = kzalloc(cfg.size, GFP_KERNEL); + if (!child->bts_buffer) + goto errout; - child->bts = ds_request_bts(child, /* base = */ NULL, cfg.size, + child->bts = ds_request_bts(child, child->bts_buffer, cfg.size, ovfl, /* th = */ (size_t)-1); if (IS_ERR(child->bts)) { error = PTR_ERR(child->bts); + kfree(child->bts_buffer); child->bts = NULL; + child->bts_buffer = NULL; goto errout; } @@ -972,6 +988,8 @@ void ptrace_disable(struct task_struct *child) #ifdef CONFIG_X86_PTRACE_BTS if (child->bts) { (void)ds_release_bts(child->bts); + kfree(child->bts_buffer); + child->bts_buffer = NULL; child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; if (!child->thread.debugctlmsr) -- cgit v1.2.3 From a266d9f1253a38ec2d5655ebcd6846298b0554f4 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Fri, 21 Nov 2008 14:49:25 +0100 Subject: [CPUFREQ] powernow-k8: ignore out-of-range PstateStatus value A workaround for AMD CPU family 11h erratum 311 might cause that the P-state Status Register shows a "current P-state" which is larger than the "current P-state limit" in P-state Current Limit Register. For the wrong P-state value there is no ACPI _PSS object defined and powernow-k8/cpufreq can't determine the proper CPU frequency for that state. As a consequence this can cause a panic during boot (potentially with all recent kernel versions -- at least I have reproduced it with various 2.6.27 kernels and with the current .28 series), as an example: powernow-k8: Found 1 AMD Turion(tm)X2 Ultra DualCore Mobile ZM-82 processors (2 \ ) powernow-k8: 0 : pstate 0 (2200 MHz) powernow-k8: 1 : pstate 1 (1100 MHz) powernow-k8: 2 : pstate 2 (600 MHz) BUG: unable to handle kernel paging request at ffff88086e7528b8 IP: [] cpufreq_stats_update+0x4a/0x5f PGD 202063 PUD 0 Oops: 0002 [#1] SMP last sysfs file: CPU 1 Modules linked in: Pid: 1, comm: swapper Not tainted 2.6.28-rc3-dirty #16 RIP: 0010:[] [] cpufreq_stats_update+0x4a/0\ f Synaptics claims to have extended capabilities, but I'm not able to read them.<6\ 6 RAX: 0000000000000000 RBX: 0000000000000001 RCX: ffff88006e7528c0 RDX: 00000000ffffffff RSI: ffff88006e54af00 RDI: ffffffff808f056c RBP: 00000000fffee697 R08: 0000000000000003 R09: ffff88006e73f080 R10: 0000000000000001 R11: 00000000002191c0 R12: ffff88006fb83c10 R13: 00000000ffffffff R14: 0000000000000001 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffff88006fb50740(0000) knlGS:0000000000000000 Unable to initialize Synaptics hardware. CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b CR2: ffff88086e7528b8 CR3: 0000000000201000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process swapper (pid: 1, threadinfo ffff88006fb82000, task ffff88006fb816d0) Stack: ffff88006e74da50 0000000000000000 ffff88006e54af00 ffffffff804863c7 ffff88006e74da50 0000000000000000 00000000ffffffff 0000000000000000 ffff88006fb83c10 ffffffff8024b46c ffffffff808f0560 ffff88006fb83c10 Call Trace: [] ? cpufreq_stat_notifier_trans+0x51/0x83 [] ? notifier_call_chain+0x29/0x4c [] ? __srcu_notifier_call_chain+0x46/0x61 [] ? cpufreq_notify_transition+0x93/0xa9 [] ? powernowk8_target+0x1e8/0x5f3 [] ? cpufreq_governor_performance+0x1b/0x20 [] ? __cpufreq_governor+0x71/0xa8 [] ? __cpufreq_set_policy+0x101/0x13e [] ? cpufreq_add_dev+0x3f0/0x4cd [] ? handle_update+0x0/0x8 [] ? sysdev_driver_register+0xb6/0x10d [] ? powernowk8_init+0x0/0x7e [] ? cpufreq_register_driver+0x8f/0x140 [] ? _stext+0x56/0x14f [] ? proc_register+0x122/0x17d [] ? create_proc_entry+0x73/0x8a [] ? register_irq_proc+0x92/0xaa [] ? init_irq_proc+0x57/0x69 [] ? kernel_init+0x116/0x169 [] ? child_rip+0xa/0x11 [] ? kernel_init+0x0/0x169 [] ? child_rip+0x0/0x11 Code: 05 c5 83 36 00 48 c7 c2 48 5d 86 80 48 8b 04 d8 48 8b 40 08 48 8b 34 02 48\ RIP [] cpufreq_stats_update+0x4a/0x5f RSP CR2: ffff88086e7528b8 ---[ end trace 0678bac75e67a2f7 ]--- Kernel panic - not syncing: Attempted to kill init! In short, aftereffect of the wrong P-state is that cpufreq_stats_update() uses "-1" as index for some array in cpufreq_stats_update (unsigned int cpu) { ... if (stat->time_in_state) stat->time_in_state[stat->last_index] = cputime64_add(stat->time_in_state[stat->last_index], cputime_sub(cur_time, stat->last_time)); ... } Fortunately, the wrong P-state value is returned only if the core is in P-state 0. This fix solves the problem by detecting the out-of-range P-state, ignoring it, and using "0" instead. Cc: Mark Langsdorf Signed-off-by: Andreas Herrmann Signed-off-by: Dave Jones --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 18 +++++++++++++++--- arch/x86/kernel/cpu/cpufreq/powernow-k8.h | 17 ++++++++++++++++- 2 files changed, 31 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index d3dcd58b87cd..7f05f44b97e9 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -115,9 +115,20 @@ static int query_current_values_with_pending_wait(struct powernow_k8_data *data) u32 i = 0; if (cpu_family == CPU_HW_PSTATE) { - rdmsr(MSR_PSTATE_STATUS, lo, hi); - i = lo & HW_PSTATE_MASK; - data->currpstate = i; + if (data->currpstate == HW_PSTATE_INVALID) { + /* read (initial) hw pstate if not yet set */ + rdmsr(MSR_PSTATE_STATUS, lo, hi); + i = lo & HW_PSTATE_MASK; + + /* + * a workaround for family 11h erratum 311 might cause + * an "out-of-range Pstate if the core is in Pstate-0 + */ + if (i >= data->numps) + data->currpstate = HW_PSTATE_0; + else + data->currpstate = i; + } return 0; } do { @@ -1121,6 +1132,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) } data->cpu = pol->cpu; + data->currpstate = HW_PSTATE_INVALID; if (powernow_k8_cpu_init_acpi(data)) { /* diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h index ab48cfed4d96..65cfb5d7f77f 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h @@ -5,6 +5,19 @@ * http://www.gnu.org/licenses/gpl.html */ + +enum pstate { + HW_PSTATE_INVALID = 0xff, + HW_PSTATE_0 = 0, + HW_PSTATE_1 = 1, + HW_PSTATE_2 = 2, + HW_PSTATE_3 = 3, + HW_PSTATE_4 = 4, + HW_PSTATE_5 = 5, + HW_PSTATE_6 = 6, + HW_PSTATE_7 = 7, +}; + struct powernow_k8_data { unsigned int cpu; @@ -23,7 +36,9 @@ struct powernow_k8_data { u32 exttype; /* extended interface = 1 */ /* keep track of the current fid / vid or pstate */ - u32 currvid, currfid, currpstate; + u32 currvid; + u32 currfid; + enum pstate currpstate; /* the powernow_table includes all frequency and vid/fid pairings: * fid are the lower 8 bits of the index, vid are the upper 8 bits. -- cgit v1.2.3 From fb52607afcd0629776f1dc9e657647ceae81dd50 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 25 Nov 2008 21:07:04 +0100 Subject: tracing/function-return-tracer: change the name into function-graph-tracer Impact: cleanup This patch changes the name of the "return function tracer" into function-graph-tracer which is a more suitable name for a tracing which makes one able to retrieve the ordered call stack during the code flow. Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 4 ++-- arch/x86/kernel/entry_32.S | 12 ++++++------ arch/x86/kernel/ftrace.c | 12 ++++++------ 3 files changed, 14 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index af2bc36ca1c4..64939a0c3986 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -14,7 +14,7 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg endif -ifdef CONFIG_FUNCTION_RET_TRACER +ifdef CONFIG_FUNCTION_GRAPH_TRACER # Don't trace __switch_to() but let it for function tracer CFLAGS_REMOVE_process_32.o = -pg endif @@ -70,7 +70,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o -obj-$(CONFIG_FUNCTION_RET_TRACER) += ftrace.o +obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 74defe21ba42..2b1f0f081a6b 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1188,9 +1188,9 @@ ENTRY(mcount) cmpl $ftrace_stub, ftrace_trace_function jnz trace -#ifdef CONFIG_FUNCTION_RET_TRACER - cmpl $ftrace_stub, ftrace_function_return - jnz ftrace_return_caller +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + cmpl $ftrace_stub, ftrace_graph_function + jnz ftrace_graph_caller #endif .globl ftrace_stub ftrace_stub: @@ -1215,8 +1215,8 @@ END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ -#ifdef CONFIG_FUNCTION_RET_TRACER -ENTRY(ftrace_return_caller) +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +ENTRY(ftrace_graph_caller) cmpl $0, function_trace_stop jne ftrace_stub @@ -1230,7 +1230,7 @@ ENTRY(ftrace_return_caller) popl %ecx popl %eax ret -END(ftrace_return_caller) +END(ftrace_graph_caller) .globl return_to_handler return_to_handler: diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index bb137f7297ed..3595a4c14aba 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -323,7 +323,7 @@ int __init ftrace_dyn_arch_init(void *data) } #endif -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER #ifndef CONFIG_DYNAMIC_FTRACE @@ -389,11 +389,11 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, */ unsigned long ftrace_return_to_handler(void) { - struct ftrace_retfunc trace; + struct ftrace_graph_ret trace; pop_return_trace(&trace.ret, &trace.calltime, &trace.func, &trace.overrun); trace.rettime = cpu_clock(raw_smp_processor_id()); - ftrace_function_return(&trace); + ftrace_graph_function(&trace); return trace.ret; } @@ -440,12 +440,12 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) ); if (WARN_ON(faulted)) { - unregister_ftrace_return(); + unregister_ftrace_graph(); return; } if (WARN_ON(!__kernel_text_address(old))) { - unregister_ftrace_return(); + unregister_ftrace_graph(); *parent = old; return; } @@ -456,4 +456,4 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) *parent = old; } -#endif /* CONFIG_FUNCTION_RET_TRACER */ +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ -- cgit v1.2.3 From 287b6e68ca7209caec40b2f44f837c580a413bae Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Wed, 26 Nov 2008 00:57:25 +0100 Subject: tracing/function-return-tracer: set a more human readable output Impact: feature This patch sets a C-like output for the function graph tracing. For this aim, we now call two handler for each function: one on the entry and one other on return. This way we can draw a well-ordered call stack. The pid of the previous trace is loosely stored to be compared against the one of the current trace to see if there were a context switch. Without this little feature, the call tree would seem broken at some locations. We could use the sched_tracer to capture these sched_events but this way of processing is much more simpler. 2 spaces have been chosen for indentation to fit the screen while deep calls. The time of execution in nanosecs is printed just after closed braces, it seems more easy this way to find the corresponding function. If the time was printed as a first column, it would be not so easy to find the corresponding function if it is called on a deep depth. I plan to output the return value but on 32 bits CPU, the return value can be 32 or 64, and its difficult to guess on which case we are. I don't know what would be the better solution on X86-32: only print eax (low-part) or even edx (high-part). Actually it's thee same problem when a function return a 8 bits value, the high part of eax could contain junk values... Here is an example of trace: sys_read() { fget_light() { } 526 vfs_read() { rw_verify_area() { security_file_permission() { cap_file_permission() { } 519 } 1564 } 2640 do_sync_read() { pipe_read() { __might_sleep() { } 511 pipe_wait() { prepare_to_wait() { } 760 deactivate_task() { dequeue_task() { dequeue_task_fair() { dequeue_entity() { update_curr() { update_min_vruntime() { } 504 } 1587 clear_buddies() { } 512 add_cfs_task_weight() { } 519 update_min_vruntime() { } 511 } 5602 dequeue_entity() { update_curr() { update_min_vruntime() { } 496 } 1631 clear_buddies() { } 496 update_min_vruntime() { } 527 } 4580 hrtick_update() { hrtick_start_fair() { } 488 } 1489 } 13700 } 14949 } 16016 msecs_to_jiffies() { } 496 put_prev_task_fair() { } 504 pick_next_task_fair() { } 489 pick_next_task_rt() { } 496 pick_next_task_fair() { } 489 pick_next_task_idle() { } 489 ------------8<---------- thread 4 ------------8<---------- finish_task_switch() { } 1203 do_softirq() { __do_softirq() { __local_bh_disable() { } 669 rcu_process_callbacks() { __rcu_process_callbacks() { cpu_quiet() { rcu_start_batch() { } 503 } 1647 } 3128 __rcu_process_callbacks() { } 542 } 5362 _local_bh_enable() { } 587 } 8880 } 9986 kthread_should_stop() { } 669 deactivate_task() { dequeue_task() { dequeue_task_fair() { dequeue_entity() { update_curr() { calc_delta_mine() { } 511 update_min_vruntime() { } 511 } 2813 Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 32 +++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 3595a4c14aba..26b2d92d48b3 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -347,7 +347,7 @@ void ftrace_nmi_exit(void) /* Add a function return address to the trace stack on thread info.*/ static int push_return_trace(unsigned long ret, unsigned long long time, - unsigned long func) + unsigned long func, int *depth) { int index; @@ -365,21 +365,22 @@ static int push_return_trace(unsigned long ret, unsigned long long time, current->ret_stack[index].ret = ret; current->ret_stack[index].func = func; current->ret_stack[index].calltime = time; + *depth = index; return 0; } /* Retrieve a function return address to the trace stack on thread info.*/ -static void pop_return_trace(unsigned long *ret, unsigned long long *time, - unsigned long *func, unsigned long *overrun) +static void pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret) { int index; index = current->curr_ret_stack; *ret = current->ret_stack[index].ret; - *func = current->ret_stack[index].func; - *time = current->ret_stack[index].calltime; - *overrun = atomic_read(¤t->trace_overrun); + trace->func = current->ret_stack[index].func; + trace->calltime = current->ret_stack[index].calltime; + trace->overrun = atomic_read(¤t->trace_overrun); + trace->depth = index; current->curr_ret_stack--; } @@ -390,12 +391,13 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, unsigned long ftrace_return_to_handler(void) { struct ftrace_graph_ret trace; - pop_return_trace(&trace.ret, &trace.calltime, &trace.func, - &trace.overrun); + unsigned long ret; + + pop_return_trace(&trace, &ret); trace.rettime = cpu_clock(raw_smp_processor_id()); - ftrace_graph_function(&trace); + ftrace_graph_return(&trace); - return trace.ret; + return ret; } /* @@ -407,6 +409,7 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) unsigned long old; unsigned long long calltime; int faulted; + struct ftrace_graph_ent trace; unsigned long return_hooker = (unsigned long) &return_to_handler; @@ -452,8 +455,15 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) calltime = cpu_clock(raw_smp_processor_id()); - if (push_return_trace(old, calltime, self_addr) == -EBUSY) + if (push_return_trace(old, calltime, + self_addr, &trace.depth) == -EBUSY) { *parent = old; + return; + } + + trace.func = self_addr; + ftrace_graph_entry(&trace); + } #endif /* CONFIG_FUNCTION_GRAPH_TRACER */ -- cgit v1.2.3 From c2324b694fa8ffee382a124198c68754088e483c Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 26 Nov 2008 03:10:01 +0100 Subject: tracing: function graph tracer, fix fix return-tracer => graph-tracer namespace rename fallout. Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 2b1f0f081a6b..7def9fd5c1e6 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1189,7 +1189,7 @@ ENTRY(mcount) cmpl $ftrace_stub, ftrace_trace_function jnz trace #ifdef CONFIG_FUNCTION_GRAPH_TRACER - cmpl $ftrace_stub, ftrace_graph_function + cmpl $ftrace_stub, ftrace_graph_return jnz ftrace_graph_caller #endif .globl ftrace_stub -- cgit v1.2.3 From 2601657d223d82053d4e1fe1063091401e6b860a Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 24 Nov 2008 18:21:37 -0800 Subject: x86: signal: move {setup|restore}_sigcontext() Impact: cleanup Move {setup|restore}_sigcontext() declaration onto head of file. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 271 ++++++++++++++++++++++---------------------- arch/x86/kernel/signal_64.c | 148 ++++++++++++------------ 2 files changed, 210 insertions(+), 209 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index f7dd6c44c042..b3f30d2a2178 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -70,6 +70,142 @@ static const struct { 0 }; +#define COPY(x) { \ + err |= __get_user(regs->x, &sc->x); \ +} + +#define COPY_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + regs->seg = tmp; \ +} + +#define COPY_SEG_CPL3(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + regs->seg = tmp | 3; \ +} + +#define GET_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + loadsegment(seg, tmp); \ +} + +static int +restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, + unsigned long *pax) +{ + void __user *buf; + unsigned int tmpflags; + unsigned int err = 0; + + /* Always make any pending restarted system calls return -EINTR */ + current_thread_info()->restart_block.fn = do_no_restart_syscall; + +#ifdef CONFIG_X86_32 + GET_SEG(gs); + COPY_SEG(fs); + COPY_SEG(es); + COPY_SEG(ds); +#endif /* CONFIG_X86_32 */ + + COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); + COPY(dx); COPY(cx); COPY(ip); + +#ifdef CONFIG_X86_64 + COPY(r8); + COPY(r9); + COPY(r10); + COPY(r11); + COPY(r12); + COPY(r13); + COPY(r14); + COPY(r15); +#endif /* CONFIG_X86_64 */ + +#ifdef CONFIG_X86_32 + COPY_SEG_CPL3(cs); + COPY_SEG_CPL3(ss); +#else /* !CONFIG_X86_32 */ + /* Kernel saves and restores only the CS segment register on signals, + * which is the bare minimum needed to allow mixed 32/64-bit code. + * App's signal handler can save/restore other segments if needed. */ + COPY_SEG_CPL3(cs); +#endif /* CONFIG_X86_32 */ + + err |= __get_user(tmpflags, &sc->flags); + regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); + regs->orig_ax = -1; /* disable syscall checks */ + + err |= __get_user(buf, &sc->fpstate); + err |= restore_i387_xstate(buf); + + err |= __get_user(*pax, &sc->ax); + return err; +} + +static int +setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, + struct pt_regs *regs, unsigned long mask) +{ + int err = 0; + +#ifdef CONFIG_X86_32 + { + unsigned int tmp; + + savesegment(gs, tmp); + err |= __put_user(tmp, (unsigned int __user *)&sc->gs); + } + err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); + err |= __put_user(regs->es, (unsigned int __user *)&sc->es); + err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); +#endif /* CONFIG_X86_32 */ + + err |= __put_user(regs->di, &sc->di); + err |= __put_user(regs->si, &sc->si); + err |= __put_user(regs->bp, &sc->bp); + err |= __put_user(regs->sp, &sc->sp); + err |= __put_user(regs->bx, &sc->bx); + err |= __put_user(regs->dx, &sc->dx); + err |= __put_user(regs->cx, &sc->cx); + err |= __put_user(regs->ax, &sc->ax); +#ifdef CONFIG_X86_64 + err |= __put_user(regs->r8, &sc->r8); + err |= __put_user(regs->r9, &sc->r9); + err |= __put_user(regs->r10, &sc->r10); + err |= __put_user(regs->r11, &sc->r11); + err |= __put_user(regs->r12, &sc->r12); + err |= __put_user(regs->r13, &sc->r13); + err |= __put_user(regs->r14, &sc->r14); + err |= __put_user(regs->r15, &sc->r15); +#endif /* CONFIG_X86_64 */ + + err |= __put_user(current->thread.trap_no, &sc->trapno); + err |= __put_user(current->thread.error_code, &sc->err); + err |= __put_user(regs->ip, &sc->ip); +#ifdef CONFIG_X86_32 + err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->sp, &sc->sp_at_signal); + err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); +#else /* !CONFIG_X86_32 */ + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->cs, &sc->cs); + err |= __put_user(0, &sc->gs); + err |= __put_user(0, &sc->fs); +#endif /* CONFIG_X86_32 */ + + err |= __put_user(fpstate, &sc->fpstate); + + /* non-iBCS2 extensions.. */ + err |= __put_user(mask, &sc->oldmask); + err |= __put_user(current->thread.cr2, &sc->cr2); + + return err; +} + /* * Atomically swap in the new signal mask, and wait for a signal. */ @@ -147,84 +283,9 @@ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, } #endif /* CONFIG_X86_32 */ -#define COPY(x) { \ - err |= __get_user(regs->x, &sc->x); \ -} - -#define COPY_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp; \ -} - -#define COPY_SEG_CPL3(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp | 3; \ -} - -#define GET_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - loadsegment(seg, tmp); \ -} - /* * Do a signal return; undo the signal stack. */ -static int -restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, - unsigned long *pax) -{ - void __user *buf; - unsigned int tmpflags; - unsigned int err = 0; - - /* Always make any pending restarted system calls return -EINTR */ - current_thread_info()->restart_block.fn = do_no_restart_syscall; - -#ifdef CONFIG_X86_32 - GET_SEG(gs); - COPY_SEG(fs); - COPY_SEG(es); - COPY_SEG(ds); -#endif /* CONFIG_X86_32 */ - - COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); - COPY(dx); COPY(cx); COPY(ip); - -#ifdef CONFIG_X86_64 - COPY(r8); - COPY(r9); - COPY(r10); - COPY(r11); - COPY(r12); - COPY(r13); - COPY(r14); - COPY(r15); -#endif /* CONFIG_X86_64 */ - -#ifdef CONFIG_X86_32 - COPY_SEG_CPL3(cs); - COPY_SEG_CPL3(ss); -#else /* !CONFIG_X86_32 */ - /* Kernel saves and restores only the CS segment register on signals, - * which is the bare minimum needed to allow mixed 32/64-bit code. - * App's signal handler can save/restore other segments if needed. */ - COPY_SEG_CPL3(cs); -#endif /* CONFIG_X86_32 */ - - err |= __get_user(tmpflags, &sc->flags); - regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); - regs->orig_ax = -1; /* disable syscall checks */ - - err |= __get_user(buf, &sc->fpstate); - err |= restore_i387_xstate(buf); - - err |= __get_user(*pax, &sc->ax); - return err; -} - asmlinkage unsigned long sys_sigreturn(unsigned long __unused) { struct sigframe __user *frame; @@ -316,66 +377,6 @@ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) /* * Set up a signal frame. */ -static int -setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, - struct pt_regs *regs, unsigned long mask) -{ - int err = 0; - -#ifdef CONFIG_X86_32 - { - unsigned int tmp; - - savesegment(gs, tmp); - err |= __put_user(tmp, (unsigned int __user *)&sc->gs); - } - err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); - err |= __put_user(regs->es, (unsigned int __user *)&sc->es); - err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(regs->di, &sc->di); - err |= __put_user(regs->si, &sc->si); - err |= __put_user(regs->bp, &sc->bp); - err |= __put_user(regs->sp, &sc->sp); - err |= __put_user(regs->bx, &sc->bx); - err |= __put_user(regs->dx, &sc->dx); - err |= __put_user(regs->cx, &sc->cx); - err |= __put_user(regs->ax, &sc->ax); -#ifdef CONFIG_X86_64 - err |= __put_user(regs->r8, &sc->r8); - err |= __put_user(regs->r9, &sc->r9); - err |= __put_user(regs->r10, &sc->r10); - err |= __put_user(regs->r11, &sc->r11); - err |= __put_user(regs->r12, &sc->r12); - err |= __put_user(regs->r13, &sc->r13); - err |= __put_user(regs->r14, &sc->r14); - err |= __put_user(regs->r15, &sc->r15); -#endif /* CONFIG_X86_64 */ - - err |= __put_user(current->thread.trap_no, &sc->trapno); - err |= __put_user(current->thread.error_code, &sc->err); - err |= __put_user(regs->ip, &sc->ip); -#ifdef CONFIG_X86_32 - err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->sp, &sc->sp_at_signal); - err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); -#else /* !CONFIG_X86_32 */ - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->cs, &sc->cs); - err |= __put_user(0, &sc->gs); - err |= __put_user(0, &sc->fs); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(fpstate, &sc->fpstate); - - /* non-iBCS2 extensions.. */ - err |= __put_user(mask, &sc->oldmask); - err |= __put_user(current->thread.cr2, &sc->cr2); - - return err; -} /* * Determine which stack to use.. diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 32718f5e4f61..771c8fcc8b0d 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -50,28 +50,6 @@ # define FIX_EFLAGS __FIX_EFLAGS #endif -#ifdef CONFIG_X86_32 -asmlinkage int sys_sigaltstack(unsigned long bx) -{ - /* - * This is needed to make gcc realize it doesn't own the - * "struct pt_regs" - */ - struct pt_regs *regs = (struct pt_regs *)&bx; - const stack_t __user *uss = (const stack_t __user *)bx; - stack_t __user *uoss = (stack_t __user *)regs->cx; - - return do_sigaltstack(uss, uoss, regs->sp); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long -sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, - struct pt_regs *regs) -{ - return do_sigaltstack(uss, uoss, regs->sp); -} -#endif /* CONFIG_X86_32 */ - #define COPY(x) { \ err |= __get_user(regs->x, &sc->x); \ } @@ -82,9 +60,6 @@ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, regs->seg = tmp | 3; \ } -/* - * Do a signal return; undo the signal stack. - */ static int restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, unsigned long *pax) @@ -138,54 +113,6 @@ restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, return err; } -static long do_rt_sigreturn(struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - unsigned long ax; - sigset_t set; - - frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) - goto badframe; - - if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) - goto badframe; - - return ax; - -badframe: - signal_fault(regs, frame, "rt_sigreturn"); - return 0; -} - -#ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) -{ - struct pt_regs *regs = (struct pt_regs *)&__unused; - - return do_rt_sigreturn(regs); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -{ - return do_rt_sigreturn(regs); -} -#endif /* CONFIG_X86_32 */ - -/* - * Set up a signal frame. - */ static int setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, struct pt_regs *regs, unsigned long mask) @@ -247,10 +174,83 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, return err; } +#ifdef CONFIG_X86_32 +asmlinkage int sys_sigaltstack(unsigned long bx) +{ + /* + * This is needed to make gcc realize it doesn't own the + * "struct pt_regs" + */ + struct pt_regs *regs = (struct pt_regs *)&bx; + const stack_t __user *uss = (const stack_t __user *)bx; + stack_t __user *uoss = (stack_t __user *)regs->cx; + + return do_sigaltstack(uss, uoss, regs->sp); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long +sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, + struct pt_regs *regs) +{ + return do_sigaltstack(uss, uoss, regs->sp); +} +#endif /* CONFIG_X86_32 */ + /* - * Determine which stack to use.. + * Do a signal return; undo the signal stack. + */ +static long do_rt_sigreturn(struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + unsigned long ax; + sigset_t set; + + frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) + goto badframe; + + if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) + goto badframe; + + return ax; + +badframe: + signal_fault(regs, frame, "rt_sigreturn"); + return 0; +} + +#ifdef CONFIG_X86_32 +asmlinkage int sys_rt_sigreturn(unsigned long __unused) +{ + struct pt_regs *regs = (struct pt_regs *)&__unused; + + return do_rt_sigreturn(regs); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) +{ + return do_rt_sigreturn(regs); +} +#endif /* CONFIG_X86_32 */ + +/* + * Set up a signal frame. */ +/* + * Determine which stack to use.. + */ static void __user * get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) { -- cgit v1.2.3 From bfeb91a9435889ef4fe7bfbb4b673f625e69e790 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 24 Nov 2008 18:23:12 -0800 Subject: x86: signal: cosmetic unification of __setup_sigframe() and __setup_rt_sigframe() Impact: cleanup Add #ifdef directive to unify __setup_sigframe() and __setup_rt_sigframe(). Move them after {setup|restore}_sigcontext() declaration. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 469 ++++++++++++++++++++++++++------------------ arch/x86/kernel/signal_64.c | 309 ++++++++++++++++++++++++----- 2 files changed, 536 insertions(+), 242 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index b3f30d2a2178..e9f71298e746 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -48,28 +48,6 @@ # define FIX_EFLAGS __FIX_EFLAGS #endif -static const struct { - u16 poplmovl; - u32 val; - u16 int80; -} __attribute__((packed)) retcode = { - 0xb858, /* popl %eax; movl $..., %eax */ - __NR_sigreturn, - 0x80cd, /* int $0x80 */ -}; - -static const struct { - u8 movl; - u32 val; - u16 int80; - u8 pad; -} __attribute__((packed)) rt_retcode = { - 0xb8, /* movl $..., %eax */ - __NR_rt_sigreturn, - 0x80cd, /* int $0x80 */ - 0 -}; - #define COPY(x) { \ err |= __get_user(regs->x, &sc->x); \ } @@ -207,176 +185,30 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, } /* - * Atomically swap in the new signal mask, and wait for a signal. + * Set up a signal frame. */ -asmlinkage int -sys_sigsuspend(int history0, int history1, old_sigset_t mask) -{ - mask &= _BLOCKABLE; - spin_lock_irq(¤t->sighand->siglock); - current->saved_sigmask = current->blocked; - siginitset(¤t->blocked, mask); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - current->state = TASK_INTERRUPTIBLE; - schedule(); - set_restore_sigmask(); - - return -ERESTARTNOHAND; -} - -asmlinkage int -sys_sigaction(int sig, const struct old_sigaction __user *act, - struct old_sigaction __user *oact) -{ - struct k_sigaction new_ka, old_ka; - int ret; - - if (act) { - old_sigset_t mask; - - if (!access_ok(VERIFY_READ, act, sizeof(*act)) || - __get_user(new_ka.sa.sa_handler, &act->sa_handler) || - __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) - return -EFAULT; - - __get_user(new_ka.sa.sa_flags, &act->sa_flags); - __get_user(mask, &act->sa_mask); - siginitset(&new_ka.sa.sa_mask, mask); - } - - ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); - - if (!ret && oact) { - if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || - __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || - __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) - return -EFAULT; - - __put_user(old_ka.sa.sa_flags, &oact->sa_flags); - __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); - } - - return ret; -} - #ifdef CONFIG_X86_32 -asmlinkage int sys_sigaltstack(unsigned long bx) -{ - /* - * This is needed to make gcc realize it doesn't own the - * "struct pt_regs" - */ - struct pt_regs *regs = (struct pt_regs *)&bx; - const stack_t __user *uss = (const stack_t __user *)bx; - stack_t __user *uoss = (stack_t __user *)regs->cx; - - return do_sigaltstack(uss, uoss, regs->sp); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long -sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, - struct pt_regs *regs) -{ - return do_sigaltstack(uss, uoss, regs->sp); -} -#endif /* CONFIG_X86_32 */ - -/* - * Do a signal return; undo the signal stack. - */ -asmlinkage unsigned long sys_sigreturn(unsigned long __unused) -{ - struct sigframe __user *frame; - struct pt_regs *regs; - unsigned long ax; - sigset_t set; - - regs = (struct pt_regs *) &__unused; - frame = (struct sigframe __user *)(regs->sp - 8); - - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 - && __copy_from_user(&set.sig[1], &frame->extramask, - sizeof(frame->extramask)))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->sc, &ax)) - goto badframe; - return ax; - -badframe: - if (show_unhandled_signals && printk_ratelimit()) { - printk("%s%s[%d] bad frame in sigreturn frame:" - "%p ip:%lx sp:%lx oeax:%lx", - task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, - current->comm, task_pid_nr(current), frame, regs->ip, - regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, current); - - return 0; -} - -static long do_rt_sigreturn(struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - unsigned long ax; - sigset_t set; - - frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) - goto badframe; - - if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) - goto badframe; - - return ax; - -badframe: - signal_fault(regs, frame, "rt_sigreturn"); - return 0; -} - -#ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) -{ - struct pt_regs *regs = (struct pt_regs *)&__unused; - - return do_rt_sigreturn(regs); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -{ - return do_rt_sigreturn(regs); -} -#endif /* CONFIG_X86_32 */ +static const struct { + u16 poplmovl; + u32 val; + u16 int80; +} __attribute__((packed)) retcode = { + 0xb858, /* popl %eax; movl $..., %eax */ + __NR_sigreturn, + 0x80cd, /* int $0x80 */ +}; -/* - * Set up a signal frame. - */ +static const struct { + u8 movl; + u32 val; + u16 int80; + u8 pad; +} __attribute__((packed)) rt_retcode = { + 0xb8, /* movl $..., %eax */ + __NR_rt_sigreturn, + 0x80cd, /* int $0x80 */ + 0 +}; /* * Determine which stack to use.. @@ -557,6 +389,265 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, return 0; } +#else /* !CONFIG_X86_32 */ +/* + * Determine which stack to use.. + */ +static void __user * +get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) +{ + /* Default to using normal stack - redzone*/ + sp -= 128; + + /* This is the X/Open sanctioned signal stack switching. */ + if (ka->sa.sa_flags & SA_ONSTACK) { + if (sas_ss_flags(sp) == 0) + sp = current->sas_ss_sp + current->sas_ss_size; + } + + return (void __user *)round_down(sp - size, 64); +} + +static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + void __user *fp = NULL; + int err = 0; + struct task_struct *me = current; + + if (used_math()) { + fp = get_stack(ka, regs->sp, sig_xstate_size); + frame = (void __user *)round_down( + (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8; + + if (save_i387_xstate(fp) < 0) + return -EFAULT; + } else + frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8; + + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; + + if (ka->sa.sa_flags & SA_SIGINFO) { + if (copy_siginfo_to_user(&frame->info, info)) + return -EFAULT; + } + + /* Create the ucontext. */ + if (cpu_has_xsave) + err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); + else + err |= __put_user(0, &frame->uc.uc_flags); + err |= __put_user(0, &frame->uc.uc_link); + err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp); + err |= __put_user(sas_ss_flags(regs->sp), + &frame->uc.uc_stack.ss_flags); + err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]); + err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); + + /* Set up to return from userspace. If provided, use a stub + already in userspace. */ + /* x86-64 should always use SA_RESTORER. */ + if (ka->sa.sa_flags & SA_RESTORER) { + err |= __put_user(ka->sa.sa_restorer, &frame->pretcode); + } else { + /* could use a vstub here */ + return -EFAULT; + } + + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->di = sig; + /* In case the signal handler was declared without prototypes */ + regs->ax = 0; + + /* This also works for non SA_SIGINFO handlers because they expect the + next argument after the signal number on the stack. */ + regs->si = (unsigned long)&frame->info; + regs->dx = (unsigned long)&frame->uc; + regs->ip = (unsigned long) ka->sa.sa_handler; + + regs->sp = (unsigned long)frame; + + /* Set up the CS register to run signal handlers in 64-bit mode, + even if the handler happens to be interrupting 32-bit code. */ + regs->cs = __USER_CS; + + return 0; +} +#endif /* CONFIG_X86_32 */ + +/* + * Atomically swap in the new signal mask, and wait for a signal. + */ +asmlinkage int +sys_sigsuspend(int history0, int history1, old_sigset_t mask) +{ + mask &= _BLOCKABLE; + spin_lock_irq(¤t->sighand->siglock); + current->saved_sigmask = current->blocked; + siginitset(¤t->blocked, mask); + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + current->state = TASK_INTERRUPTIBLE; + schedule(); + set_restore_sigmask(); + + return -ERESTARTNOHAND; +} + +asmlinkage int +sys_sigaction(int sig, const struct old_sigaction __user *act, + struct old_sigaction __user *oact) +{ + struct k_sigaction new_ka, old_ka; + int ret; + + if (act) { + old_sigset_t mask; + + if (!access_ok(VERIFY_READ, act, sizeof(*act)) || + __get_user(new_ka.sa.sa_handler, &act->sa_handler) || + __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) + return -EFAULT; + + __get_user(new_ka.sa.sa_flags, &act->sa_flags); + __get_user(mask, &act->sa_mask); + siginitset(&new_ka.sa.sa_mask, mask); + } + + ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); + + if (!ret && oact) { + if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || + __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || + __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) + return -EFAULT; + + __put_user(old_ka.sa.sa_flags, &oact->sa_flags); + __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); + } + + return ret; +} + +#ifdef CONFIG_X86_32 +asmlinkage int sys_sigaltstack(unsigned long bx) +{ + /* + * This is needed to make gcc realize it doesn't own the + * "struct pt_regs" + */ + struct pt_regs *regs = (struct pt_regs *)&bx; + const stack_t __user *uss = (const stack_t __user *)bx; + stack_t __user *uoss = (stack_t __user *)regs->cx; + + return do_sigaltstack(uss, uoss, regs->sp); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long +sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, + struct pt_regs *regs) +{ + return do_sigaltstack(uss, uoss, regs->sp); +} +#endif /* CONFIG_X86_32 */ + +/* + * Do a signal return; undo the signal stack. + */ +asmlinkage unsigned long sys_sigreturn(unsigned long __unused) +{ + struct sigframe __user *frame; + struct pt_regs *regs; + unsigned long ax; + sigset_t set; + + regs = (struct pt_regs *) &__unused; + frame = (struct sigframe __user *)(regs->sp - 8); + + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 + && __copy_from_user(&set.sig[1], &frame->extramask, + sizeof(frame->extramask)))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->sc, &ax)) + goto badframe; + return ax; + +badframe: + if (show_unhandled_signals && printk_ratelimit()) { + printk("%s%s[%d] bad frame in sigreturn frame:" + "%p ip:%lx sp:%lx oeax:%lx", + task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, + current->comm, task_pid_nr(current), frame, regs->ip, + regs->sp, regs->orig_ax); + print_vma_addr(" in ", regs->ip); + printk(KERN_CONT "\n"); + } + + force_sig(SIGSEGV, current); + + return 0; +} + +static long do_rt_sigreturn(struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + unsigned long ax; + sigset_t set; + + frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) + goto badframe; + + if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) + goto badframe; + + return ax; + +badframe: + signal_fault(regs, frame, "rt_sigreturn"); + return 0; +} + +#ifdef CONFIG_X86_32 +asmlinkage int sys_rt_sigreturn(unsigned long __unused) +{ + struct pt_regs *regs = (struct pt_regs *)&__unused; + + return do_rt_sigreturn(regs); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) +{ + return do_rt_sigreturn(regs); +} +#endif /* CONFIG_X86_32 */ /* * OK, we're invoking a handler: diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 771c8fcc8b0d..2da7e6e60807 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -174,80 +174,212 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, return err; } +/* + * Set up a signal frame. + */ #ifdef CONFIG_X86_32 -asmlinkage int sys_sigaltstack(unsigned long bx) +static const struct { + u16 poplmovl; + u32 val; + u16 int80; +} __attribute__((packed)) retcode = { + 0xb858, /* popl %eax; movl $..., %eax */ + __NR_sigreturn, + 0x80cd, /* int $0x80 */ +}; + +static const struct { + u8 movl; + u32 val; + u16 int80; + u8 pad; +} __attribute__((packed)) rt_retcode = { + 0xb8, /* movl $..., %eax */ + __NR_rt_sigreturn, + 0x80cd, /* int $0x80 */ + 0 +}; + +/* + * Determine which stack to use.. + */ +static inline void __user * +get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, + void **fpstate) { + unsigned long sp; + + /* Default to using normal stack */ + sp = regs->sp; + /* - * This is needed to make gcc realize it doesn't own the - * "struct pt_regs" + * If we are on the alternate signal stack and would overflow it, don't. + * Return an always-bogus address instead so we will die with SIGSEGV. */ - struct pt_regs *regs = (struct pt_regs *)&bx; - const stack_t __user *uss = (const stack_t __user *)bx; - stack_t __user *uoss = (stack_t __user *)regs->cx; + if (on_sig_stack(sp) && !likely(on_sig_stack(sp - frame_size))) + return (void __user *) -1L; - return do_sigaltstack(uss, uoss, regs->sp); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long -sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, - struct pt_regs *regs) -{ - return do_sigaltstack(uss, uoss, regs->sp); + /* This is the X/Open sanctioned signal stack switching. */ + if (ka->sa.sa_flags & SA_ONSTACK) { + if (sas_ss_flags(sp) == 0) + sp = current->sas_ss_sp + current->sas_ss_size; + } else { + /* This is the legacy signal stack switching. */ + if ((regs->ss & 0xffff) != __USER_DS && + !(ka->sa.sa_flags & SA_RESTORER) && + ka->sa.sa_restorer) + sp = (unsigned long) ka->sa.sa_restorer; + } + + if (used_math()) { + sp = sp - sig_xstate_size; + *fpstate = (struct _fpstate *) sp; + if (save_i387_xstate(*fpstate) < 0) + return (void __user *)-1L; + } + + sp -= frame_size; + /* + * Align the stack pointer according to the i386 ABI, + * i.e. so that on function entry ((sp + 4) & 15) == 0. + */ + sp = ((sp + 4) & -16ul) - 4; + + return (void __user *) sp; } -#endif /* CONFIG_X86_32 */ -/* - * Do a signal return; undo the signal stack. - */ -static long do_rt_sigreturn(struct pt_regs *regs) +static int +__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, + struct pt_regs *regs) { - struct rt_sigframe __user *frame; - unsigned long ax; - sigset_t set; + struct sigframe __user *frame; + void __user *restorer; + int err = 0; + void __user *fpstate = NULL; - frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) - goto badframe; + frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; - if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) - goto badframe; + if (__put_user(sig, &frame->sig)) + return -EFAULT; - if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) - goto badframe; + if (setup_sigcontext(&frame->sc, fpstate, regs, set->sig[0])) + return -EFAULT; - return ax; + if (_NSIG_WORDS > 1) { + if (__copy_to_user(&frame->extramask, &set->sig[1], + sizeof(frame->extramask))) + return -EFAULT; + } + + if (current->mm->context.vdso) + restorer = VDSO32_SYMBOL(current->mm->context.vdso, sigreturn); + else + restorer = &frame->retcode; + if (ka->sa.sa_flags & SA_RESTORER) + restorer = ka->sa.sa_restorer; + + /* Set up to return from userspace. */ + err |= __put_user(restorer, &frame->pretcode); + + /* + * This is popl %eax ; movl $__NR_sigreturn, %eax ; int $0x80 + * + * WE DO NOT USE IT ANY MORE! It's only left here for historical + * reasons and because gdb uses it as a signature to notice + * signal handler stack frames. + */ + err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode); + + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->sp = (unsigned long)frame; + regs->ip = (unsigned long)ka->sa.sa_handler; + regs->ax = (unsigned long)sig; + regs->dx = 0; + regs->cx = 0; + + regs->ds = __USER_DS; + regs->es = __USER_DS; + regs->ss = __USER_DS; + regs->cs = __USER_CS; -badframe: - signal_fault(regs, frame, "rt_sigreturn"); return 0; } -#ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) +static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs) { - struct pt_regs *regs = (struct pt_regs *)&__unused; + struct rt_sigframe __user *frame; + void __user *restorer; + int err = 0; + void __user *fpstate = NULL; - return do_rt_sigreturn(regs); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -{ - return do_rt_sigreturn(regs); -} -#endif /* CONFIG_X86_32 */ + frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); -/* - * Set up a signal frame. - */ + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; + + err |= __put_user(sig, &frame->sig); + err |= __put_user(&frame->info, &frame->pinfo); + err |= __put_user(&frame->uc, &frame->puc); + err |= copy_siginfo_to_user(&frame->info, info); + if (err) + return -EFAULT; + + /* Create the ucontext. */ + if (cpu_has_xsave) + err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); + else + err |= __put_user(0, &frame->uc.uc_flags); + err |= __put_user(0, &frame->uc.uc_link); + err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp); + err |= __put_user(sas_ss_flags(regs->sp), + &frame->uc.uc_stack.ss_flags); + err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fpstate, + regs, set->sig[0]); + err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); + if (err) + return -EFAULT; + + /* Set up to return from userspace. */ + restorer = VDSO32_SYMBOL(current->mm->context.vdso, rt_sigreturn); + if (ka->sa.sa_flags & SA_RESTORER) + restorer = ka->sa.sa_restorer; + err |= __put_user(restorer, &frame->pretcode); + + /* + * This is movl $__NR_rt_sigreturn, %ax ; int $0x80 + * + * WE DO NOT USE IT ANY MORE! It's only left here for historical + * reasons and because gdb uses it as a signature to notice + * signal handler stack frames. + */ + err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode); + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->sp = (unsigned long)frame; + regs->ip = (unsigned long)ka->sa.sa_handler; + regs->ax = (unsigned long)sig; + regs->dx = (unsigned long)&frame->info; + regs->cx = (unsigned long)&frame->uc; + + regs->ds = __USER_DS; + regs->es = __USER_DS; + regs->ss = __USER_DS; + regs->cs = __USER_CS; + + return 0; +} +#else /* !CONFIG_X86_32 */ /* * Determine which stack to use.. */ @@ -337,6 +469,77 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, return 0; } +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_X86_32 +asmlinkage int sys_sigaltstack(unsigned long bx) +{ + /* + * This is needed to make gcc realize it doesn't own the + * "struct pt_regs" + */ + struct pt_regs *regs = (struct pt_regs *)&bx; + const stack_t __user *uss = (const stack_t __user *)bx; + stack_t __user *uoss = (stack_t __user *)regs->cx; + + return do_sigaltstack(uss, uoss, regs->sp); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long +sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, + struct pt_regs *regs) +{ + return do_sigaltstack(uss, uoss, regs->sp); +} +#endif /* CONFIG_X86_32 */ + +/* + * Do a signal return; undo the signal stack. + */ +static long do_rt_sigreturn(struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + unsigned long ax; + sigset_t set; + + frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) + goto badframe; + + if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) + goto badframe; + + return ax; + +badframe: + signal_fault(regs, frame, "rt_sigreturn"); + return 0; +} + +#ifdef CONFIG_X86_32 +asmlinkage int sys_rt_sigreturn(unsigned long __unused) +{ + struct pt_regs *regs = (struct pt_regs *)&__unused; + + return do_rt_sigreturn(regs); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) +{ + return do_rt_sigreturn(regs); +} +#endif /* CONFIG_X86_32 */ /* * OK, we're invoking a handler -- cgit v1.2.3 From e5fa2d063cf2ca38eae5fb3469315db669d5c041 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 24 Nov 2008 18:24:11 -0800 Subject: x86: signal: unify signal_{32|64}.c, prepare Impact: cleanup Add #ifdef directive for 32-bit only code. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_32.c | 6 +++ arch/x86/kernel/signal_64.c | 116 +++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 121 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c index e9f71298e746..b1f4d34e0a38 100644 --- a/arch/x86/kernel/signal_32.c +++ b/arch/x86/kernel/signal_32.c @@ -1,8 +1,10 @@ /* * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs * * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes + * 2000-2002 x86-64 support by Andi Kleen */ #include @@ -481,6 +483,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, } #endif /* CONFIG_X86_32 */ +#ifdef CONFIG_X86_32 /* * Atomically swap in the new signal mask, and wait for a signal. */ @@ -535,6 +538,7 @@ sys_sigaction(int sig, const struct old_sigaction __user *act, return ret; } +#endif /* CONFIG_X86_32 */ #ifdef CONFIG_X86_32 asmlinkage int sys_sigaltstack(unsigned long bx) @@ -561,6 +565,7 @@ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, /* * Do a signal return; undo the signal stack. */ +#ifdef CONFIG_X86_32 asmlinkage unsigned long sys_sigreturn(unsigned long __unused) { struct sigframe __user *frame; @@ -603,6 +608,7 @@ badframe: return 0; } +#endif /* CONFIG_X86_32 */ static long do_rt_sigreturn(struct pt_regs *regs) { diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 2da7e6e60807..b1f4d34e0a38 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -54,12 +54,24 @@ err |= __get_user(regs->x, &sc->x); \ } +#define COPY_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + regs->seg = tmp; \ +} + #define COPY_SEG_CPL3(seg) { \ unsigned short tmp; \ err |= __get_user(tmp, &sc->seg); \ regs->seg = tmp | 3; \ } +#define GET_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + loadsegment(seg, tmp); \ +} + static int restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, unsigned long *pax) @@ -471,6 +483,63 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, } #endif /* CONFIG_X86_32 */ +#ifdef CONFIG_X86_32 +/* + * Atomically swap in the new signal mask, and wait for a signal. + */ +asmlinkage int +sys_sigsuspend(int history0, int history1, old_sigset_t mask) +{ + mask &= _BLOCKABLE; + spin_lock_irq(¤t->sighand->siglock); + current->saved_sigmask = current->blocked; + siginitset(¤t->blocked, mask); + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + current->state = TASK_INTERRUPTIBLE; + schedule(); + set_restore_sigmask(); + + return -ERESTARTNOHAND; +} + +asmlinkage int +sys_sigaction(int sig, const struct old_sigaction __user *act, + struct old_sigaction __user *oact) +{ + struct k_sigaction new_ka, old_ka; + int ret; + + if (act) { + old_sigset_t mask; + + if (!access_ok(VERIFY_READ, act, sizeof(*act)) || + __get_user(new_ka.sa.sa_handler, &act->sa_handler) || + __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) + return -EFAULT; + + __get_user(new_ka.sa.sa_flags, &act->sa_flags); + __get_user(mask, &act->sa_mask); + siginitset(&new_ka.sa.sa_mask, mask); + } + + ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); + + if (!ret && oact) { + if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || + __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || + __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) + return -EFAULT; + + __put_user(old_ka.sa.sa_flags, &oact->sa_flags); + __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); + } + + return ret; +} +#endif /* CONFIG_X86_32 */ + #ifdef CONFIG_X86_32 asmlinkage int sys_sigaltstack(unsigned long bx) { @@ -496,6 +565,51 @@ sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, /* * Do a signal return; undo the signal stack. */ +#ifdef CONFIG_X86_32 +asmlinkage unsigned long sys_sigreturn(unsigned long __unused) +{ + struct sigframe __user *frame; + struct pt_regs *regs; + unsigned long ax; + sigset_t set; + + regs = (struct pt_regs *) &__unused; + frame = (struct sigframe __user *)(regs->sp - 8); + + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 + && __copy_from_user(&set.sig[1], &frame->extramask, + sizeof(frame->extramask)))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->sc, &ax)) + goto badframe; + return ax; + +badframe: + if (show_unhandled_signals && printk_ratelimit()) { + printk("%s%s[%d] bad frame in sigreturn frame:" + "%p ip:%lx sp:%lx oeax:%lx", + task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, + current->comm, task_pid_nr(current), frame, regs->ip, + regs->sp, regs->orig_ax); + print_vma_addr(" in ", regs->ip); + printk(KERN_CONT "\n"); + } + + force_sig(SIGSEGV, current); + + return 0; +} +#endif /* CONFIG_X86_32 */ + static long do_rt_sigreturn(struct pt_regs *regs) { struct rt_sigframe __user *frame; @@ -542,7 +656,7 @@ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) #endif /* CONFIG_X86_32 */ /* - * OK, we're invoking a handler + * OK, we're invoking a handler: */ static int signr_convert(int sig) { -- cgit v1.2.3 From 5ceb40da9bacc8b056805d72efb1a52502d56b6b Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 24 Nov 2008 18:24:11 -0800 Subject: x86: signal: unify signal_{32|64}.c Impact: cleanup Unify signal_{32|64}.c! Mechanic unification - the two files are the same. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 2 +- arch/x86/kernel/signal.c | 915 ++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/signal_32.c | 915 -------------------------------------------- arch/x86/kernel/signal_64.c | 915 -------------------------------------------- 4 files changed, 916 insertions(+), 1831 deletions(-) create mode 100644 arch/x86/kernel/signal.c delete mode 100644 arch/x86/kernel/signal_32.c delete mode 100644 arch/x86/kernel/signal_64.c (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index d7e5a58ee22f..ef28c210ebf8 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -22,7 +22,7 @@ CFLAGS_vsyscall_64.o := $(PROFILING) -g0 $(nostackp) CFLAGS_hpet.o := $(nostackp) CFLAGS_tsc.o := $(nostackp) -obj-y := process_$(BITS).o signal_$(BITS).o entry_$(BITS).o +obj-y := process_$(BITS).o signal.o entry_$(BITS).o obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o obj-y += time_$(BITS).o ioport.o ldt.o obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c new file mode 100644 index 000000000000..b1f4d34e0a38 --- /dev/null +++ b/arch/x86/kernel/signal.c @@ -0,0 +1,915 @@ +/* + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs + * + * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson + * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes + * 2000-2002 x86-64 support by Andi Kleen + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#ifdef CONFIG_X86_64 +#include +#include +#include +#endif /* CONFIG_X86_64 */ + +#include +#include + +#include "sigframe.h" + +#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) + +#define __FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \ + X86_EFLAGS_DF | X86_EFLAGS_TF | X86_EFLAGS_SF | \ + X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \ + X86_EFLAGS_CF) + +#ifdef CONFIG_X86_32 +# define FIX_EFLAGS (__FIX_EFLAGS | X86_EFLAGS_RF) +#else +# define FIX_EFLAGS __FIX_EFLAGS +#endif + +#define COPY(x) { \ + err |= __get_user(regs->x, &sc->x); \ +} + +#define COPY_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + regs->seg = tmp; \ +} + +#define COPY_SEG_CPL3(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + regs->seg = tmp | 3; \ +} + +#define GET_SEG(seg) { \ + unsigned short tmp; \ + err |= __get_user(tmp, &sc->seg); \ + loadsegment(seg, tmp); \ +} + +static int +restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, + unsigned long *pax) +{ + void __user *buf; + unsigned int tmpflags; + unsigned int err = 0; + + /* Always make any pending restarted system calls return -EINTR */ + current_thread_info()->restart_block.fn = do_no_restart_syscall; + +#ifdef CONFIG_X86_32 + GET_SEG(gs); + COPY_SEG(fs); + COPY_SEG(es); + COPY_SEG(ds); +#endif /* CONFIG_X86_32 */ + + COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); + COPY(dx); COPY(cx); COPY(ip); + +#ifdef CONFIG_X86_64 + COPY(r8); + COPY(r9); + COPY(r10); + COPY(r11); + COPY(r12); + COPY(r13); + COPY(r14); + COPY(r15); +#endif /* CONFIG_X86_64 */ + +#ifdef CONFIG_X86_32 + COPY_SEG_CPL3(cs); + COPY_SEG_CPL3(ss); +#else /* !CONFIG_X86_32 */ + /* Kernel saves and restores only the CS segment register on signals, + * which is the bare minimum needed to allow mixed 32/64-bit code. + * App's signal handler can save/restore other segments if needed. */ + COPY_SEG_CPL3(cs); +#endif /* CONFIG_X86_32 */ + + err |= __get_user(tmpflags, &sc->flags); + regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); + regs->orig_ax = -1; /* disable syscall checks */ + + err |= __get_user(buf, &sc->fpstate); + err |= restore_i387_xstate(buf); + + err |= __get_user(*pax, &sc->ax); + return err; +} + +static int +setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, + struct pt_regs *regs, unsigned long mask) +{ + int err = 0; + +#ifdef CONFIG_X86_32 + { + unsigned int tmp; + + savesegment(gs, tmp); + err |= __put_user(tmp, (unsigned int __user *)&sc->gs); + } + err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); + err |= __put_user(regs->es, (unsigned int __user *)&sc->es); + err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); +#endif /* CONFIG_X86_32 */ + + err |= __put_user(regs->di, &sc->di); + err |= __put_user(regs->si, &sc->si); + err |= __put_user(regs->bp, &sc->bp); + err |= __put_user(regs->sp, &sc->sp); + err |= __put_user(regs->bx, &sc->bx); + err |= __put_user(regs->dx, &sc->dx); + err |= __put_user(regs->cx, &sc->cx); + err |= __put_user(regs->ax, &sc->ax); +#ifdef CONFIG_X86_64 + err |= __put_user(regs->r8, &sc->r8); + err |= __put_user(regs->r9, &sc->r9); + err |= __put_user(regs->r10, &sc->r10); + err |= __put_user(regs->r11, &sc->r11); + err |= __put_user(regs->r12, &sc->r12); + err |= __put_user(regs->r13, &sc->r13); + err |= __put_user(regs->r14, &sc->r14); + err |= __put_user(regs->r15, &sc->r15); +#endif /* CONFIG_X86_64 */ + + err |= __put_user(current->thread.trap_no, &sc->trapno); + err |= __put_user(current->thread.error_code, &sc->err); + err |= __put_user(regs->ip, &sc->ip); +#ifdef CONFIG_X86_32 + err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->sp, &sc->sp_at_signal); + err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); +#else /* !CONFIG_X86_32 */ + err |= __put_user(regs->flags, &sc->flags); + err |= __put_user(regs->cs, &sc->cs); + err |= __put_user(0, &sc->gs); + err |= __put_user(0, &sc->fs); +#endif /* CONFIG_X86_32 */ + + err |= __put_user(fpstate, &sc->fpstate); + + /* non-iBCS2 extensions.. */ + err |= __put_user(mask, &sc->oldmask); + err |= __put_user(current->thread.cr2, &sc->cr2); + + return err; +} + +/* + * Set up a signal frame. + */ +#ifdef CONFIG_X86_32 +static const struct { + u16 poplmovl; + u32 val; + u16 int80; +} __attribute__((packed)) retcode = { + 0xb858, /* popl %eax; movl $..., %eax */ + __NR_sigreturn, + 0x80cd, /* int $0x80 */ +}; + +static const struct { + u8 movl; + u32 val; + u16 int80; + u8 pad; +} __attribute__((packed)) rt_retcode = { + 0xb8, /* movl $..., %eax */ + __NR_rt_sigreturn, + 0x80cd, /* int $0x80 */ + 0 +}; + +/* + * Determine which stack to use.. + */ +static inline void __user * +get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, + void **fpstate) +{ + unsigned long sp; + + /* Default to using normal stack */ + sp = regs->sp; + + /* + * If we are on the alternate signal stack and would overflow it, don't. + * Return an always-bogus address instead so we will die with SIGSEGV. + */ + if (on_sig_stack(sp) && !likely(on_sig_stack(sp - frame_size))) + return (void __user *) -1L; + + /* This is the X/Open sanctioned signal stack switching. */ + if (ka->sa.sa_flags & SA_ONSTACK) { + if (sas_ss_flags(sp) == 0) + sp = current->sas_ss_sp + current->sas_ss_size; + } else { + /* This is the legacy signal stack switching. */ + if ((regs->ss & 0xffff) != __USER_DS && + !(ka->sa.sa_flags & SA_RESTORER) && + ka->sa.sa_restorer) + sp = (unsigned long) ka->sa.sa_restorer; + } + + if (used_math()) { + sp = sp - sig_xstate_size; + *fpstate = (struct _fpstate *) sp; + if (save_i387_xstate(*fpstate) < 0) + return (void __user *)-1L; + } + + sp -= frame_size; + /* + * Align the stack pointer according to the i386 ABI, + * i.e. so that on function entry ((sp + 4) & 15) == 0. + */ + sp = ((sp + 4) & -16ul) - 4; + + return (void __user *) sp; +} + +static int +__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, + struct pt_regs *regs) +{ + struct sigframe __user *frame; + void __user *restorer; + int err = 0; + void __user *fpstate = NULL; + + frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); + + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; + + if (__put_user(sig, &frame->sig)) + return -EFAULT; + + if (setup_sigcontext(&frame->sc, fpstate, regs, set->sig[0])) + return -EFAULT; + + if (_NSIG_WORDS > 1) { + if (__copy_to_user(&frame->extramask, &set->sig[1], + sizeof(frame->extramask))) + return -EFAULT; + } + + if (current->mm->context.vdso) + restorer = VDSO32_SYMBOL(current->mm->context.vdso, sigreturn); + else + restorer = &frame->retcode; + if (ka->sa.sa_flags & SA_RESTORER) + restorer = ka->sa.sa_restorer; + + /* Set up to return from userspace. */ + err |= __put_user(restorer, &frame->pretcode); + + /* + * This is popl %eax ; movl $__NR_sigreturn, %eax ; int $0x80 + * + * WE DO NOT USE IT ANY MORE! It's only left here for historical + * reasons and because gdb uses it as a signature to notice + * signal handler stack frames. + */ + err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode); + + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->sp = (unsigned long)frame; + regs->ip = (unsigned long)ka->sa.sa_handler; + regs->ax = (unsigned long)sig; + regs->dx = 0; + regs->cx = 0; + + regs->ds = __USER_DS; + regs->es = __USER_DS; + regs->ss = __USER_DS; + regs->cs = __USER_CS; + + return 0; +} + +static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + void __user *restorer; + int err = 0; + void __user *fpstate = NULL; + + frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); + + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; + + err |= __put_user(sig, &frame->sig); + err |= __put_user(&frame->info, &frame->pinfo); + err |= __put_user(&frame->uc, &frame->puc); + err |= copy_siginfo_to_user(&frame->info, info); + if (err) + return -EFAULT; + + /* Create the ucontext. */ + if (cpu_has_xsave) + err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); + else + err |= __put_user(0, &frame->uc.uc_flags); + err |= __put_user(0, &frame->uc.uc_link); + err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp); + err |= __put_user(sas_ss_flags(regs->sp), + &frame->uc.uc_stack.ss_flags); + err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fpstate, + regs, set->sig[0]); + err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); + if (err) + return -EFAULT; + + /* Set up to return from userspace. */ + restorer = VDSO32_SYMBOL(current->mm->context.vdso, rt_sigreturn); + if (ka->sa.sa_flags & SA_RESTORER) + restorer = ka->sa.sa_restorer; + err |= __put_user(restorer, &frame->pretcode); + + /* + * This is movl $__NR_rt_sigreturn, %ax ; int $0x80 + * + * WE DO NOT USE IT ANY MORE! It's only left here for historical + * reasons and because gdb uses it as a signature to notice + * signal handler stack frames. + */ + err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode); + + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->sp = (unsigned long)frame; + regs->ip = (unsigned long)ka->sa.sa_handler; + regs->ax = (unsigned long)sig; + regs->dx = (unsigned long)&frame->info; + regs->cx = (unsigned long)&frame->uc; + + regs->ds = __USER_DS; + regs->es = __USER_DS; + regs->ss = __USER_DS; + regs->cs = __USER_CS; + + return 0; +} +#else /* !CONFIG_X86_32 */ +/* + * Determine which stack to use.. + */ +static void __user * +get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) +{ + /* Default to using normal stack - redzone*/ + sp -= 128; + + /* This is the X/Open sanctioned signal stack switching. */ + if (ka->sa.sa_flags & SA_ONSTACK) { + if (sas_ss_flags(sp) == 0) + sp = current->sas_ss_sp + current->sas_ss_size; + } + + return (void __user *)round_down(sp - size, 64); +} + +static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + void __user *fp = NULL; + int err = 0; + struct task_struct *me = current; + + if (used_math()) { + fp = get_stack(ka, regs->sp, sig_xstate_size); + frame = (void __user *)round_down( + (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8; + + if (save_i387_xstate(fp) < 0) + return -EFAULT; + } else + frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8; + + if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) + return -EFAULT; + + if (ka->sa.sa_flags & SA_SIGINFO) { + if (copy_siginfo_to_user(&frame->info, info)) + return -EFAULT; + } + + /* Create the ucontext. */ + if (cpu_has_xsave) + err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); + else + err |= __put_user(0, &frame->uc.uc_flags); + err |= __put_user(0, &frame->uc.uc_link); + err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp); + err |= __put_user(sas_ss_flags(regs->sp), + &frame->uc.uc_stack.ss_flags); + err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); + err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]); + err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); + + /* Set up to return from userspace. If provided, use a stub + already in userspace. */ + /* x86-64 should always use SA_RESTORER. */ + if (ka->sa.sa_flags & SA_RESTORER) { + err |= __put_user(ka->sa.sa_restorer, &frame->pretcode); + } else { + /* could use a vstub here */ + return -EFAULT; + } + + if (err) + return -EFAULT; + + /* Set up registers for signal handler */ + regs->di = sig; + /* In case the signal handler was declared without prototypes */ + regs->ax = 0; + + /* This also works for non SA_SIGINFO handlers because they expect the + next argument after the signal number on the stack. */ + regs->si = (unsigned long)&frame->info; + regs->dx = (unsigned long)&frame->uc; + regs->ip = (unsigned long) ka->sa.sa_handler; + + regs->sp = (unsigned long)frame; + + /* Set up the CS register to run signal handlers in 64-bit mode, + even if the handler happens to be interrupting 32-bit code. */ + regs->cs = __USER_CS; + + return 0; +} +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_X86_32 +/* + * Atomically swap in the new signal mask, and wait for a signal. + */ +asmlinkage int +sys_sigsuspend(int history0, int history1, old_sigset_t mask) +{ + mask &= _BLOCKABLE; + spin_lock_irq(¤t->sighand->siglock); + current->saved_sigmask = current->blocked; + siginitset(¤t->blocked, mask); + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + current->state = TASK_INTERRUPTIBLE; + schedule(); + set_restore_sigmask(); + + return -ERESTARTNOHAND; +} + +asmlinkage int +sys_sigaction(int sig, const struct old_sigaction __user *act, + struct old_sigaction __user *oact) +{ + struct k_sigaction new_ka, old_ka; + int ret; + + if (act) { + old_sigset_t mask; + + if (!access_ok(VERIFY_READ, act, sizeof(*act)) || + __get_user(new_ka.sa.sa_handler, &act->sa_handler) || + __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) + return -EFAULT; + + __get_user(new_ka.sa.sa_flags, &act->sa_flags); + __get_user(mask, &act->sa_mask); + siginitset(&new_ka.sa.sa_mask, mask); + } + + ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); + + if (!ret && oact) { + if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || + __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || + __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) + return -EFAULT; + + __put_user(old_ka.sa.sa_flags, &oact->sa_flags); + __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); + } + + return ret; +} +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_X86_32 +asmlinkage int sys_sigaltstack(unsigned long bx) +{ + /* + * This is needed to make gcc realize it doesn't own the + * "struct pt_regs" + */ + struct pt_regs *regs = (struct pt_regs *)&bx; + const stack_t __user *uss = (const stack_t __user *)bx; + stack_t __user *uoss = (stack_t __user *)regs->cx; + + return do_sigaltstack(uss, uoss, regs->sp); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long +sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, + struct pt_regs *regs) +{ + return do_sigaltstack(uss, uoss, regs->sp); +} +#endif /* CONFIG_X86_32 */ + +/* + * Do a signal return; undo the signal stack. + */ +#ifdef CONFIG_X86_32 +asmlinkage unsigned long sys_sigreturn(unsigned long __unused) +{ + struct sigframe __user *frame; + struct pt_regs *regs; + unsigned long ax; + sigset_t set; + + regs = (struct pt_regs *) &__unused; + frame = (struct sigframe __user *)(regs->sp - 8); + + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 + && __copy_from_user(&set.sig[1], &frame->extramask, + sizeof(frame->extramask)))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->sc, &ax)) + goto badframe; + return ax; + +badframe: + if (show_unhandled_signals && printk_ratelimit()) { + printk("%s%s[%d] bad frame in sigreturn frame:" + "%p ip:%lx sp:%lx oeax:%lx", + task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, + current->comm, task_pid_nr(current), frame, regs->ip, + regs->sp, regs->orig_ax); + print_vma_addr(" in ", regs->ip); + printk(KERN_CONT "\n"); + } + + force_sig(SIGSEGV, current); + + return 0; +} +#endif /* CONFIG_X86_32 */ + +static long do_rt_sigreturn(struct pt_regs *regs) +{ + struct rt_sigframe __user *frame; + unsigned long ax; + sigset_t set; + + frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); + if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) + goto badframe; + if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) + goto badframe; + + sigdelsetmask(&set, ~_BLOCKABLE); + spin_lock_irq(¤t->sighand->siglock); + current->blocked = set; + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) + goto badframe; + + if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) + goto badframe; + + return ax; + +badframe: + signal_fault(regs, frame, "rt_sigreturn"); + return 0; +} + +#ifdef CONFIG_X86_32 +asmlinkage int sys_rt_sigreturn(unsigned long __unused) +{ + struct pt_regs *regs = (struct pt_regs *)&__unused; + + return do_rt_sigreturn(regs); +} +#else /* !CONFIG_X86_32 */ +asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) +{ + return do_rt_sigreturn(regs); +} +#endif /* CONFIG_X86_32 */ + +/* + * OK, we're invoking a handler: + */ +static int signr_convert(int sig) +{ +#ifdef CONFIG_X86_32 + struct thread_info *info = current_thread_info(); + + if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) + return info->exec_domain->signal_invmap[sig]; +#endif /* CONFIG_X86_32 */ + return sig; +} + +#ifdef CONFIG_X86_32 + +#define is_ia32 1 +#define ia32_setup_frame __setup_frame +#define ia32_setup_rt_frame __setup_rt_frame + +#else /* !CONFIG_X86_32 */ + +#ifdef CONFIG_IA32_EMULATION +#define is_ia32 test_thread_flag(TIF_IA32) +#else /* !CONFIG_IA32_EMULATION */ +#define is_ia32 0 +#endif /* CONFIG_IA32_EMULATION */ + +#endif /* CONFIG_X86_32 */ + +static int +setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs) +{ + int usig = signr_convert(sig); + int ret; + + /* Set up the stack frame */ + if (is_ia32) { + if (ka->sa.sa_flags & SA_SIGINFO) + ret = ia32_setup_rt_frame(usig, ka, info, set, regs); + else + ret = ia32_setup_frame(usig, ka, set, regs); + } else + ret = __setup_rt_frame(sig, ka, info, set, regs); + + if (ret) { + force_sigsegv(sig, current); + return -EFAULT; + } + + return ret; +} + +static int +handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka, + sigset_t *oldset, struct pt_regs *regs) +{ + int ret; + + /* Are we from a system call? */ + if (syscall_get_nr(current, regs) >= 0) { + /* If so, check system call restarting.. */ + switch (syscall_get_error(current, regs)) { + case -ERESTART_RESTARTBLOCK: + case -ERESTARTNOHAND: + regs->ax = -EINTR; + break; + + case -ERESTARTSYS: + if (!(ka->sa.sa_flags & SA_RESTART)) { + regs->ax = -EINTR; + break; + } + /* fallthrough */ + case -ERESTARTNOINTR: + regs->ax = regs->orig_ax; + regs->ip -= 2; + break; + } + } + + /* + * If TF is set due to a debugger (TIF_FORCED_TF), clear the TF + * flag so that register information in the sigcontext is correct. + */ + if (unlikely(regs->flags & X86_EFLAGS_TF) && + likely(test_and_clear_thread_flag(TIF_FORCED_TF))) + regs->flags &= ~X86_EFLAGS_TF; + + ret = setup_rt_frame(sig, ka, info, oldset, regs); + + if (ret) + return ret; + +#ifdef CONFIG_X86_64 + /* + * This has nothing to do with segment registers, + * despite the name. This magic affects uaccess.h + * macros' behavior. Reset it to the normal setting. + */ + set_fs(USER_DS); +#endif + + /* + * Clear the direction flag as per the ABI for function entry. + */ + regs->flags &= ~X86_EFLAGS_DF; + + /* + * Clear TF when entering the signal handler, but + * notify any tracer that was single-stepping it. + * The tracer may want to single-step inside the + * handler too. + */ + regs->flags &= ~X86_EFLAGS_TF; + + spin_lock_irq(¤t->sighand->siglock); + sigorsets(¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); + if (!(ka->sa.sa_flags & SA_NODEFER)) + sigaddset(¤t->blocked, sig); + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); + + tracehook_signal_handler(sig, info, ka, regs, + test_thread_flag(TIF_SINGLESTEP)); + + return 0; +} + +#ifdef CONFIG_X86_32 +#define NR_restart_syscall __NR_restart_syscall +#else /* !CONFIG_X86_32 */ +#define NR_restart_syscall \ + test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall +#endif /* CONFIG_X86_32 */ + +/* + * Note that 'init' is a special process: it doesn't get signals it doesn't + * want to handle. Thus you cannot kill init even with a SIGKILL even by + * mistake. + */ +static void do_signal(struct pt_regs *regs) +{ + struct k_sigaction ka; + siginfo_t info; + int signr; + sigset_t *oldset; + + /* + * We want the common case to go fast, which is why we may in certain + * cases get here from kernel mode. Just return without doing anything + * if so. + * X86_32: vm86 regs switched out by assembly code before reaching + * here, so testing against kernel CS suffices. + */ + if (!user_mode(regs)) + return; + + if (current_thread_info()->status & TS_RESTORE_SIGMASK) + oldset = ¤t->saved_sigmask; + else + oldset = ¤t->blocked; + + signr = get_signal_to_deliver(&info, &ka, regs, NULL); + if (signr > 0) { + /* + * Re-enable any watchpoints before delivering the + * signal to user space. The processor register will + * have been cleared if the watchpoint triggered + * inside the kernel. + */ + if (current->thread.debugreg7) + set_debugreg(current->thread.debugreg7, 7); + + /* Whee! Actually deliver the signal. */ + if (handle_signal(signr, &info, &ka, oldset, regs) == 0) { + /* + * A signal was successfully delivered; the saved + * sigmask will have been stored in the signal frame, + * and will be restored by sigreturn, so we can simply + * clear the TS_RESTORE_SIGMASK flag. + */ + current_thread_info()->status &= ~TS_RESTORE_SIGMASK; + } + return; + } + + /* Did we come from a system call? */ + if (syscall_get_nr(current, regs) >= 0) { + /* Restart the system call - no handlers present */ + switch (syscall_get_error(current, regs)) { + case -ERESTARTNOHAND: + case -ERESTARTSYS: + case -ERESTARTNOINTR: + regs->ax = regs->orig_ax; + regs->ip -= 2; + break; + + case -ERESTART_RESTARTBLOCK: + regs->ax = NR_restart_syscall; + regs->ip -= 2; + break; + } + } + + /* + * If there's no signal to deliver, we just put the saved sigmask + * back. + */ + if (current_thread_info()->status & TS_RESTORE_SIGMASK) { + current_thread_info()->status &= ~TS_RESTORE_SIGMASK; + sigprocmask(SIG_SETMASK, ¤t->saved_sigmask, NULL); + } +} + +/* + * notification of userspace execution resumption + * - triggered by the TIF_WORK_MASK flags + */ +void +do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) +{ +#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE) + /* notify userspace of pending MCEs */ + if (thread_info_flags & _TIF_MCE_NOTIFY) + mce_notify_user(); +#endif /* CONFIG_X86_64 && CONFIG_X86_MCE */ + + /* deal with pending signal delivery */ + if (thread_info_flags & _TIF_SIGPENDING) + do_signal(regs); + + if (thread_info_flags & _TIF_NOTIFY_RESUME) { + clear_thread_flag(TIF_NOTIFY_RESUME); + tracehook_notify_resume(regs); + } + +#ifdef CONFIG_X86_32 + clear_thread_flag(TIF_IRET); +#endif /* CONFIG_X86_32 */ +} + +void signal_fault(struct pt_regs *regs, void __user *frame, char *where) +{ + struct task_struct *me = current; + + if (show_unhandled_signals && printk_ratelimit()) { + printk(KERN_INFO + "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx", + me->comm, me->pid, where, frame, + regs->ip, regs->sp, regs->orig_ax); + print_vma_addr(" in ", regs->ip); + printk(KERN_CONT "\n"); + } + + force_sig(SIGSEGV, me); +} diff --git a/arch/x86/kernel/signal_32.c b/arch/x86/kernel/signal_32.c deleted file mode 100644 index b1f4d34e0a38..000000000000 --- a/arch/x86/kernel/signal_32.c +++ /dev/null @@ -1,915 +0,0 @@ -/* - * Copyright (C) 1991, 1992 Linus Torvalds - * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs - * - * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson - * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes - * 2000-2002 x86-64 support by Andi Kleen - */ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#ifdef CONFIG_X86_64 -#include -#include -#include -#endif /* CONFIG_X86_64 */ - -#include -#include - -#include "sigframe.h" - -#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) - -#define __FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \ - X86_EFLAGS_DF | X86_EFLAGS_TF | X86_EFLAGS_SF | \ - X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \ - X86_EFLAGS_CF) - -#ifdef CONFIG_X86_32 -# define FIX_EFLAGS (__FIX_EFLAGS | X86_EFLAGS_RF) -#else -# define FIX_EFLAGS __FIX_EFLAGS -#endif - -#define COPY(x) { \ - err |= __get_user(regs->x, &sc->x); \ -} - -#define COPY_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp; \ -} - -#define COPY_SEG_CPL3(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp | 3; \ -} - -#define GET_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - loadsegment(seg, tmp); \ -} - -static int -restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, - unsigned long *pax) -{ - void __user *buf; - unsigned int tmpflags; - unsigned int err = 0; - - /* Always make any pending restarted system calls return -EINTR */ - current_thread_info()->restart_block.fn = do_no_restart_syscall; - -#ifdef CONFIG_X86_32 - GET_SEG(gs); - COPY_SEG(fs); - COPY_SEG(es); - COPY_SEG(ds); -#endif /* CONFIG_X86_32 */ - - COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); - COPY(dx); COPY(cx); COPY(ip); - -#ifdef CONFIG_X86_64 - COPY(r8); - COPY(r9); - COPY(r10); - COPY(r11); - COPY(r12); - COPY(r13); - COPY(r14); - COPY(r15); -#endif /* CONFIG_X86_64 */ - -#ifdef CONFIG_X86_32 - COPY_SEG_CPL3(cs); - COPY_SEG_CPL3(ss); -#else /* !CONFIG_X86_32 */ - /* Kernel saves and restores only the CS segment register on signals, - * which is the bare minimum needed to allow mixed 32/64-bit code. - * App's signal handler can save/restore other segments if needed. */ - COPY_SEG_CPL3(cs); -#endif /* CONFIG_X86_32 */ - - err |= __get_user(tmpflags, &sc->flags); - regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); - regs->orig_ax = -1; /* disable syscall checks */ - - err |= __get_user(buf, &sc->fpstate); - err |= restore_i387_xstate(buf); - - err |= __get_user(*pax, &sc->ax); - return err; -} - -static int -setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, - struct pt_regs *regs, unsigned long mask) -{ - int err = 0; - -#ifdef CONFIG_X86_32 - { - unsigned int tmp; - - savesegment(gs, tmp); - err |= __put_user(tmp, (unsigned int __user *)&sc->gs); - } - err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); - err |= __put_user(regs->es, (unsigned int __user *)&sc->es); - err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(regs->di, &sc->di); - err |= __put_user(regs->si, &sc->si); - err |= __put_user(regs->bp, &sc->bp); - err |= __put_user(regs->sp, &sc->sp); - err |= __put_user(regs->bx, &sc->bx); - err |= __put_user(regs->dx, &sc->dx); - err |= __put_user(regs->cx, &sc->cx); - err |= __put_user(regs->ax, &sc->ax); -#ifdef CONFIG_X86_64 - err |= __put_user(regs->r8, &sc->r8); - err |= __put_user(regs->r9, &sc->r9); - err |= __put_user(regs->r10, &sc->r10); - err |= __put_user(regs->r11, &sc->r11); - err |= __put_user(regs->r12, &sc->r12); - err |= __put_user(regs->r13, &sc->r13); - err |= __put_user(regs->r14, &sc->r14); - err |= __put_user(regs->r15, &sc->r15); -#endif /* CONFIG_X86_64 */ - - err |= __put_user(current->thread.trap_no, &sc->trapno); - err |= __put_user(current->thread.error_code, &sc->err); - err |= __put_user(regs->ip, &sc->ip); -#ifdef CONFIG_X86_32 - err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->sp, &sc->sp_at_signal); - err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); -#else /* !CONFIG_X86_32 */ - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->cs, &sc->cs); - err |= __put_user(0, &sc->gs); - err |= __put_user(0, &sc->fs); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(fpstate, &sc->fpstate); - - /* non-iBCS2 extensions.. */ - err |= __put_user(mask, &sc->oldmask); - err |= __put_user(current->thread.cr2, &sc->cr2); - - return err; -} - -/* - * Set up a signal frame. - */ -#ifdef CONFIG_X86_32 -static const struct { - u16 poplmovl; - u32 val; - u16 int80; -} __attribute__((packed)) retcode = { - 0xb858, /* popl %eax; movl $..., %eax */ - __NR_sigreturn, - 0x80cd, /* int $0x80 */ -}; - -static const struct { - u8 movl; - u32 val; - u16 int80; - u8 pad; -} __attribute__((packed)) rt_retcode = { - 0xb8, /* movl $..., %eax */ - __NR_rt_sigreturn, - 0x80cd, /* int $0x80 */ - 0 -}; - -/* - * Determine which stack to use.. - */ -static inline void __user * -get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, - void **fpstate) -{ - unsigned long sp; - - /* Default to using normal stack */ - sp = regs->sp; - - /* - * If we are on the alternate signal stack and would overflow it, don't. - * Return an always-bogus address instead so we will die with SIGSEGV. - */ - if (on_sig_stack(sp) && !likely(on_sig_stack(sp - frame_size))) - return (void __user *) -1L; - - /* This is the X/Open sanctioned signal stack switching. */ - if (ka->sa.sa_flags & SA_ONSTACK) { - if (sas_ss_flags(sp) == 0) - sp = current->sas_ss_sp + current->sas_ss_size; - } else { - /* This is the legacy signal stack switching. */ - if ((regs->ss & 0xffff) != __USER_DS && - !(ka->sa.sa_flags & SA_RESTORER) && - ka->sa.sa_restorer) - sp = (unsigned long) ka->sa.sa_restorer; - } - - if (used_math()) { - sp = sp - sig_xstate_size; - *fpstate = (struct _fpstate *) sp; - if (save_i387_xstate(*fpstate) < 0) - return (void __user *)-1L; - } - - sp -= frame_size; - /* - * Align the stack pointer according to the i386 ABI, - * i.e. so that on function entry ((sp + 4) & 15) == 0. - */ - sp = ((sp + 4) & -16ul) - 4; - - return (void __user *) sp; -} - -static int -__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, - struct pt_regs *regs) -{ - struct sigframe __user *frame; - void __user *restorer; - int err = 0; - void __user *fpstate = NULL; - - frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - if (__put_user(sig, &frame->sig)) - return -EFAULT; - - if (setup_sigcontext(&frame->sc, fpstate, regs, set->sig[0])) - return -EFAULT; - - if (_NSIG_WORDS > 1) { - if (__copy_to_user(&frame->extramask, &set->sig[1], - sizeof(frame->extramask))) - return -EFAULT; - } - - if (current->mm->context.vdso) - restorer = VDSO32_SYMBOL(current->mm->context.vdso, sigreturn); - else - restorer = &frame->retcode; - if (ka->sa.sa_flags & SA_RESTORER) - restorer = ka->sa.sa_restorer; - - /* Set up to return from userspace. */ - err |= __put_user(restorer, &frame->pretcode); - - /* - * This is popl %eax ; movl $__NR_sigreturn, %eax ; int $0x80 - * - * WE DO NOT USE IT ANY MORE! It's only left here for historical - * reasons and because gdb uses it as a signature to notice - * signal handler stack frames. - */ - err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode); - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->sp = (unsigned long)frame; - regs->ip = (unsigned long)ka->sa.sa_handler; - regs->ax = (unsigned long)sig; - regs->dx = 0; - regs->cx = 0; - - regs->ds = __USER_DS; - regs->es = __USER_DS; - regs->ss = __USER_DS; - regs->cs = __USER_CS; - - return 0; -} - -static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - void __user *restorer; - int err = 0; - void __user *fpstate = NULL; - - frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - err |= __put_user(sig, &frame->sig); - err |= __put_user(&frame->info, &frame->pinfo); - err |= __put_user(&frame->uc, &frame->puc); - err |= copy_siginfo_to_user(&frame->info, info); - if (err) - return -EFAULT; - - /* Create the ucontext. */ - if (cpu_has_xsave) - err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); - else - err |= __put_user(0, &frame->uc.uc_flags); - err |= __put_user(0, &frame->uc.uc_link); - err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp); - err |= __put_user(sas_ss_flags(regs->sp), - &frame->uc.uc_stack.ss_flags); - err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, fpstate, - regs, set->sig[0]); - err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); - if (err) - return -EFAULT; - - /* Set up to return from userspace. */ - restorer = VDSO32_SYMBOL(current->mm->context.vdso, rt_sigreturn); - if (ka->sa.sa_flags & SA_RESTORER) - restorer = ka->sa.sa_restorer; - err |= __put_user(restorer, &frame->pretcode); - - /* - * This is movl $__NR_rt_sigreturn, %ax ; int $0x80 - * - * WE DO NOT USE IT ANY MORE! It's only left here for historical - * reasons and because gdb uses it as a signature to notice - * signal handler stack frames. - */ - err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode); - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->sp = (unsigned long)frame; - regs->ip = (unsigned long)ka->sa.sa_handler; - regs->ax = (unsigned long)sig; - regs->dx = (unsigned long)&frame->info; - regs->cx = (unsigned long)&frame->uc; - - regs->ds = __USER_DS; - regs->es = __USER_DS; - regs->ss = __USER_DS; - regs->cs = __USER_CS; - - return 0; -} -#else /* !CONFIG_X86_32 */ -/* - * Determine which stack to use.. - */ -static void __user * -get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) -{ - /* Default to using normal stack - redzone*/ - sp -= 128; - - /* This is the X/Open sanctioned signal stack switching. */ - if (ka->sa.sa_flags & SA_ONSTACK) { - if (sas_ss_flags(sp) == 0) - sp = current->sas_ss_sp + current->sas_ss_size; - } - - return (void __user *)round_down(sp - size, 64); -} - -static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - void __user *fp = NULL; - int err = 0; - struct task_struct *me = current; - - if (used_math()) { - fp = get_stack(ka, regs->sp, sig_xstate_size); - frame = (void __user *)round_down( - (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8; - - if (save_i387_xstate(fp) < 0) - return -EFAULT; - } else - frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8; - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - if (ka->sa.sa_flags & SA_SIGINFO) { - if (copy_siginfo_to_user(&frame->info, info)) - return -EFAULT; - } - - /* Create the ucontext. */ - if (cpu_has_xsave) - err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); - else - err |= __put_user(0, &frame->uc.uc_flags); - err |= __put_user(0, &frame->uc.uc_link); - err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp); - err |= __put_user(sas_ss_flags(regs->sp), - &frame->uc.uc_stack.ss_flags); - err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]); - err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); - - /* Set up to return from userspace. If provided, use a stub - already in userspace. */ - /* x86-64 should always use SA_RESTORER. */ - if (ka->sa.sa_flags & SA_RESTORER) { - err |= __put_user(ka->sa.sa_restorer, &frame->pretcode); - } else { - /* could use a vstub here */ - return -EFAULT; - } - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->di = sig; - /* In case the signal handler was declared without prototypes */ - regs->ax = 0; - - /* This also works for non SA_SIGINFO handlers because they expect the - next argument after the signal number on the stack. */ - regs->si = (unsigned long)&frame->info; - regs->dx = (unsigned long)&frame->uc; - regs->ip = (unsigned long) ka->sa.sa_handler; - - regs->sp = (unsigned long)frame; - - /* Set up the CS register to run signal handlers in 64-bit mode, - even if the handler happens to be interrupting 32-bit code. */ - regs->cs = __USER_CS; - - return 0; -} -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_X86_32 -/* - * Atomically swap in the new signal mask, and wait for a signal. - */ -asmlinkage int -sys_sigsuspend(int history0, int history1, old_sigset_t mask) -{ - mask &= _BLOCKABLE; - spin_lock_irq(¤t->sighand->siglock); - current->saved_sigmask = current->blocked; - siginitset(¤t->blocked, mask); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - current->state = TASK_INTERRUPTIBLE; - schedule(); - set_restore_sigmask(); - - return -ERESTARTNOHAND; -} - -asmlinkage int -sys_sigaction(int sig, const struct old_sigaction __user *act, - struct old_sigaction __user *oact) -{ - struct k_sigaction new_ka, old_ka; - int ret; - - if (act) { - old_sigset_t mask; - - if (!access_ok(VERIFY_READ, act, sizeof(*act)) || - __get_user(new_ka.sa.sa_handler, &act->sa_handler) || - __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) - return -EFAULT; - - __get_user(new_ka.sa.sa_flags, &act->sa_flags); - __get_user(mask, &act->sa_mask); - siginitset(&new_ka.sa.sa_mask, mask); - } - - ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); - - if (!ret && oact) { - if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || - __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || - __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) - return -EFAULT; - - __put_user(old_ka.sa.sa_flags, &oact->sa_flags); - __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); - } - - return ret; -} -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_X86_32 -asmlinkage int sys_sigaltstack(unsigned long bx) -{ - /* - * This is needed to make gcc realize it doesn't own the - * "struct pt_regs" - */ - struct pt_regs *regs = (struct pt_regs *)&bx; - const stack_t __user *uss = (const stack_t __user *)bx; - stack_t __user *uoss = (stack_t __user *)regs->cx; - - return do_sigaltstack(uss, uoss, regs->sp); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long -sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, - struct pt_regs *regs) -{ - return do_sigaltstack(uss, uoss, regs->sp); -} -#endif /* CONFIG_X86_32 */ - -/* - * Do a signal return; undo the signal stack. - */ -#ifdef CONFIG_X86_32 -asmlinkage unsigned long sys_sigreturn(unsigned long __unused) -{ - struct sigframe __user *frame; - struct pt_regs *regs; - unsigned long ax; - sigset_t set; - - regs = (struct pt_regs *) &__unused; - frame = (struct sigframe __user *)(regs->sp - 8); - - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 - && __copy_from_user(&set.sig[1], &frame->extramask, - sizeof(frame->extramask)))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->sc, &ax)) - goto badframe; - return ax; - -badframe: - if (show_unhandled_signals && printk_ratelimit()) { - printk("%s%s[%d] bad frame in sigreturn frame:" - "%p ip:%lx sp:%lx oeax:%lx", - task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, - current->comm, task_pid_nr(current), frame, regs->ip, - regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, current); - - return 0; -} -#endif /* CONFIG_X86_32 */ - -static long do_rt_sigreturn(struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - unsigned long ax; - sigset_t set; - - frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) - goto badframe; - - if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) - goto badframe; - - return ax; - -badframe: - signal_fault(regs, frame, "rt_sigreturn"); - return 0; -} - -#ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) -{ - struct pt_regs *regs = (struct pt_regs *)&__unused; - - return do_rt_sigreturn(regs); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -{ - return do_rt_sigreturn(regs); -} -#endif /* CONFIG_X86_32 */ - -/* - * OK, we're invoking a handler: - */ -static int signr_convert(int sig) -{ -#ifdef CONFIG_X86_32 - struct thread_info *info = current_thread_info(); - - if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) - return info->exec_domain->signal_invmap[sig]; -#endif /* CONFIG_X86_32 */ - return sig; -} - -#ifdef CONFIG_X86_32 - -#define is_ia32 1 -#define ia32_setup_frame __setup_frame -#define ia32_setup_rt_frame __setup_rt_frame - -#else /* !CONFIG_X86_32 */ - -#ifdef CONFIG_IA32_EMULATION -#define is_ia32 test_thread_flag(TIF_IA32) -#else /* !CONFIG_IA32_EMULATION */ -#define is_ia32 0 -#endif /* CONFIG_IA32_EMULATION */ - -#endif /* CONFIG_X86_32 */ - -static int -setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - int usig = signr_convert(sig); - int ret; - - /* Set up the stack frame */ - if (is_ia32) { - if (ka->sa.sa_flags & SA_SIGINFO) - ret = ia32_setup_rt_frame(usig, ka, info, set, regs); - else - ret = ia32_setup_frame(usig, ka, set, regs); - } else - ret = __setup_rt_frame(sig, ka, info, set, regs); - - if (ret) { - force_sigsegv(sig, current); - return -EFAULT; - } - - return ret; -} - -static int -handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka, - sigset_t *oldset, struct pt_regs *regs) -{ - int ret; - - /* Are we from a system call? */ - if (syscall_get_nr(current, regs) >= 0) { - /* If so, check system call restarting.. */ - switch (syscall_get_error(current, regs)) { - case -ERESTART_RESTARTBLOCK: - case -ERESTARTNOHAND: - regs->ax = -EINTR; - break; - - case -ERESTARTSYS: - if (!(ka->sa.sa_flags & SA_RESTART)) { - regs->ax = -EINTR; - break; - } - /* fallthrough */ - case -ERESTARTNOINTR: - regs->ax = regs->orig_ax; - regs->ip -= 2; - break; - } - } - - /* - * If TF is set due to a debugger (TIF_FORCED_TF), clear the TF - * flag so that register information in the sigcontext is correct. - */ - if (unlikely(regs->flags & X86_EFLAGS_TF) && - likely(test_and_clear_thread_flag(TIF_FORCED_TF))) - regs->flags &= ~X86_EFLAGS_TF; - - ret = setup_rt_frame(sig, ka, info, oldset, regs); - - if (ret) - return ret; - -#ifdef CONFIG_X86_64 - /* - * This has nothing to do with segment registers, - * despite the name. This magic affects uaccess.h - * macros' behavior. Reset it to the normal setting. - */ - set_fs(USER_DS); -#endif - - /* - * Clear the direction flag as per the ABI for function entry. - */ - regs->flags &= ~X86_EFLAGS_DF; - - /* - * Clear TF when entering the signal handler, but - * notify any tracer that was single-stepping it. - * The tracer may want to single-step inside the - * handler too. - */ - regs->flags &= ~X86_EFLAGS_TF; - - spin_lock_irq(¤t->sighand->siglock); - sigorsets(¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); - if (!(ka->sa.sa_flags & SA_NODEFER)) - sigaddset(¤t->blocked, sig); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - tracehook_signal_handler(sig, info, ka, regs, - test_thread_flag(TIF_SINGLESTEP)); - - return 0; -} - -#ifdef CONFIG_X86_32 -#define NR_restart_syscall __NR_restart_syscall -#else /* !CONFIG_X86_32 */ -#define NR_restart_syscall \ - test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall -#endif /* CONFIG_X86_32 */ - -/* - * Note that 'init' is a special process: it doesn't get signals it doesn't - * want to handle. Thus you cannot kill init even with a SIGKILL even by - * mistake. - */ -static void do_signal(struct pt_regs *regs) -{ - struct k_sigaction ka; - siginfo_t info; - int signr; - sigset_t *oldset; - - /* - * We want the common case to go fast, which is why we may in certain - * cases get here from kernel mode. Just return without doing anything - * if so. - * X86_32: vm86 regs switched out by assembly code before reaching - * here, so testing against kernel CS suffices. - */ - if (!user_mode(regs)) - return; - - if (current_thread_info()->status & TS_RESTORE_SIGMASK) - oldset = ¤t->saved_sigmask; - else - oldset = ¤t->blocked; - - signr = get_signal_to_deliver(&info, &ka, regs, NULL); - if (signr > 0) { - /* - * Re-enable any watchpoints before delivering the - * signal to user space. The processor register will - * have been cleared if the watchpoint triggered - * inside the kernel. - */ - if (current->thread.debugreg7) - set_debugreg(current->thread.debugreg7, 7); - - /* Whee! Actually deliver the signal. */ - if (handle_signal(signr, &info, &ka, oldset, regs) == 0) { - /* - * A signal was successfully delivered; the saved - * sigmask will have been stored in the signal frame, - * and will be restored by sigreturn, so we can simply - * clear the TS_RESTORE_SIGMASK flag. - */ - current_thread_info()->status &= ~TS_RESTORE_SIGMASK; - } - return; - } - - /* Did we come from a system call? */ - if (syscall_get_nr(current, regs) >= 0) { - /* Restart the system call - no handlers present */ - switch (syscall_get_error(current, regs)) { - case -ERESTARTNOHAND: - case -ERESTARTSYS: - case -ERESTARTNOINTR: - regs->ax = regs->orig_ax; - regs->ip -= 2; - break; - - case -ERESTART_RESTARTBLOCK: - regs->ax = NR_restart_syscall; - regs->ip -= 2; - break; - } - } - - /* - * If there's no signal to deliver, we just put the saved sigmask - * back. - */ - if (current_thread_info()->status & TS_RESTORE_SIGMASK) { - current_thread_info()->status &= ~TS_RESTORE_SIGMASK; - sigprocmask(SIG_SETMASK, ¤t->saved_sigmask, NULL); - } -} - -/* - * notification of userspace execution resumption - * - triggered by the TIF_WORK_MASK flags - */ -void -do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) -{ -#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE) - /* notify userspace of pending MCEs */ - if (thread_info_flags & _TIF_MCE_NOTIFY) - mce_notify_user(); -#endif /* CONFIG_X86_64 && CONFIG_X86_MCE */ - - /* deal with pending signal delivery */ - if (thread_info_flags & _TIF_SIGPENDING) - do_signal(regs); - - if (thread_info_flags & _TIF_NOTIFY_RESUME) { - clear_thread_flag(TIF_NOTIFY_RESUME); - tracehook_notify_resume(regs); - } - -#ifdef CONFIG_X86_32 - clear_thread_flag(TIF_IRET); -#endif /* CONFIG_X86_32 */ -} - -void signal_fault(struct pt_regs *regs, void __user *frame, char *where) -{ - struct task_struct *me = current; - - if (show_unhandled_signals && printk_ratelimit()) { - printk(KERN_INFO - "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx", - me->comm, me->pid, where, frame, - regs->ip, regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, me); -} diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c deleted file mode 100644 index b1f4d34e0a38..000000000000 --- a/arch/x86/kernel/signal_64.c +++ /dev/null @@ -1,915 +0,0 @@ -/* - * Copyright (C) 1991, 1992 Linus Torvalds - * Copyright (C) 2000, 2001, 2002 Andi Kleen SuSE Labs - * - * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson - * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes - * 2000-2002 x86-64 support by Andi Kleen - */ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#ifdef CONFIG_X86_64 -#include -#include -#include -#endif /* CONFIG_X86_64 */ - -#include -#include - -#include "sigframe.h" - -#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) - -#define __FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \ - X86_EFLAGS_DF | X86_EFLAGS_TF | X86_EFLAGS_SF | \ - X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \ - X86_EFLAGS_CF) - -#ifdef CONFIG_X86_32 -# define FIX_EFLAGS (__FIX_EFLAGS | X86_EFLAGS_RF) -#else -# define FIX_EFLAGS __FIX_EFLAGS -#endif - -#define COPY(x) { \ - err |= __get_user(regs->x, &sc->x); \ -} - -#define COPY_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp; \ -} - -#define COPY_SEG_CPL3(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - regs->seg = tmp | 3; \ -} - -#define GET_SEG(seg) { \ - unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ - loadsegment(seg, tmp); \ -} - -static int -restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, - unsigned long *pax) -{ - void __user *buf; - unsigned int tmpflags; - unsigned int err = 0; - - /* Always make any pending restarted system calls return -EINTR */ - current_thread_info()->restart_block.fn = do_no_restart_syscall; - -#ifdef CONFIG_X86_32 - GET_SEG(gs); - COPY_SEG(fs); - COPY_SEG(es); - COPY_SEG(ds); -#endif /* CONFIG_X86_32 */ - - COPY(di); COPY(si); COPY(bp); COPY(sp); COPY(bx); - COPY(dx); COPY(cx); COPY(ip); - -#ifdef CONFIG_X86_64 - COPY(r8); - COPY(r9); - COPY(r10); - COPY(r11); - COPY(r12); - COPY(r13); - COPY(r14); - COPY(r15); -#endif /* CONFIG_X86_64 */ - -#ifdef CONFIG_X86_32 - COPY_SEG_CPL3(cs); - COPY_SEG_CPL3(ss); -#else /* !CONFIG_X86_32 */ - /* Kernel saves and restores only the CS segment register on signals, - * which is the bare minimum needed to allow mixed 32/64-bit code. - * App's signal handler can save/restore other segments if needed. */ - COPY_SEG_CPL3(cs); -#endif /* CONFIG_X86_32 */ - - err |= __get_user(tmpflags, &sc->flags); - regs->flags = (regs->flags & ~FIX_EFLAGS) | (tmpflags & FIX_EFLAGS); - regs->orig_ax = -1; /* disable syscall checks */ - - err |= __get_user(buf, &sc->fpstate); - err |= restore_i387_xstate(buf); - - err |= __get_user(*pax, &sc->ax); - return err; -} - -static int -setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate, - struct pt_regs *regs, unsigned long mask) -{ - int err = 0; - -#ifdef CONFIG_X86_32 - { - unsigned int tmp; - - savesegment(gs, tmp); - err |= __put_user(tmp, (unsigned int __user *)&sc->gs); - } - err |= __put_user(regs->fs, (unsigned int __user *)&sc->fs); - err |= __put_user(regs->es, (unsigned int __user *)&sc->es); - err |= __put_user(regs->ds, (unsigned int __user *)&sc->ds); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(regs->di, &sc->di); - err |= __put_user(regs->si, &sc->si); - err |= __put_user(regs->bp, &sc->bp); - err |= __put_user(regs->sp, &sc->sp); - err |= __put_user(regs->bx, &sc->bx); - err |= __put_user(regs->dx, &sc->dx); - err |= __put_user(regs->cx, &sc->cx); - err |= __put_user(regs->ax, &sc->ax); -#ifdef CONFIG_X86_64 - err |= __put_user(regs->r8, &sc->r8); - err |= __put_user(regs->r9, &sc->r9); - err |= __put_user(regs->r10, &sc->r10); - err |= __put_user(regs->r11, &sc->r11); - err |= __put_user(regs->r12, &sc->r12); - err |= __put_user(regs->r13, &sc->r13); - err |= __put_user(regs->r14, &sc->r14); - err |= __put_user(regs->r15, &sc->r15); -#endif /* CONFIG_X86_64 */ - - err |= __put_user(current->thread.trap_no, &sc->trapno); - err |= __put_user(current->thread.error_code, &sc->err); - err |= __put_user(regs->ip, &sc->ip); -#ifdef CONFIG_X86_32 - err |= __put_user(regs->cs, (unsigned int __user *)&sc->cs); - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->sp, &sc->sp_at_signal); - err |= __put_user(regs->ss, (unsigned int __user *)&sc->ss); -#else /* !CONFIG_X86_32 */ - err |= __put_user(regs->flags, &sc->flags); - err |= __put_user(regs->cs, &sc->cs); - err |= __put_user(0, &sc->gs); - err |= __put_user(0, &sc->fs); -#endif /* CONFIG_X86_32 */ - - err |= __put_user(fpstate, &sc->fpstate); - - /* non-iBCS2 extensions.. */ - err |= __put_user(mask, &sc->oldmask); - err |= __put_user(current->thread.cr2, &sc->cr2); - - return err; -} - -/* - * Set up a signal frame. - */ -#ifdef CONFIG_X86_32 -static const struct { - u16 poplmovl; - u32 val; - u16 int80; -} __attribute__((packed)) retcode = { - 0xb858, /* popl %eax; movl $..., %eax */ - __NR_sigreturn, - 0x80cd, /* int $0x80 */ -}; - -static const struct { - u8 movl; - u32 val; - u16 int80; - u8 pad; -} __attribute__((packed)) rt_retcode = { - 0xb8, /* movl $..., %eax */ - __NR_rt_sigreturn, - 0x80cd, /* int $0x80 */ - 0 -}; - -/* - * Determine which stack to use.. - */ -static inline void __user * -get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, - void **fpstate) -{ - unsigned long sp; - - /* Default to using normal stack */ - sp = regs->sp; - - /* - * If we are on the alternate signal stack and would overflow it, don't. - * Return an always-bogus address instead so we will die with SIGSEGV. - */ - if (on_sig_stack(sp) && !likely(on_sig_stack(sp - frame_size))) - return (void __user *) -1L; - - /* This is the X/Open sanctioned signal stack switching. */ - if (ka->sa.sa_flags & SA_ONSTACK) { - if (sas_ss_flags(sp) == 0) - sp = current->sas_ss_sp + current->sas_ss_size; - } else { - /* This is the legacy signal stack switching. */ - if ((regs->ss & 0xffff) != __USER_DS && - !(ka->sa.sa_flags & SA_RESTORER) && - ka->sa.sa_restorer) - sp = (unsigned long) ka->sa.sa_restorer; - } - - if (used_math()) { - sp = sp - sig_xstate_size; - *fpstate = (struct _fpstate *) sp; - if (save_i387_xstate(*fpstate) < 0) - return (void __user *)-1L; - } - - sp -= frame_size; - /* - * Align the stack pointer according to the i386 ABI, - * i.e. so that on function entry ((sp + 4) & 15) == 0. - */ - sp = ((sp + 4) & -16ul) - 4; - - return (void __user *) sp; -} - -static int -__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, - struct pt_regs *regs) -{ - struct sigframe __user *frame; - void __user *restorer; - int err = 0; - void __user *fpstate = NULL; - - frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - if (__put_user(sig, &frame->sig)) - return -EFAULT; - - if (setup_sigcontext(&frame->sc, fpstate, regs, set->sig[0])) - return -EFAULT; - - if (_NSIG_WORDS > 1) { - if (__copy_to_user(&frame->extramask, &set->sig[1], - sizeof(frame->extramask))) - return -EFAULT; - } - - if (current->mm->context.vdso) - restorer = VDSO32_SYMBOL(current->mm->context.vdso, sigreturn); - else - restorer = &frame->retcode; - if (ka->sa.sa_flags & SA_RESTORER) - restorer = ka->sa.sa_restorer; - - /* Set up to return from userspace. */ - err |= __put_user(restorer, &frame->pretcode); - - /* - * This is popl %eax ; movl $__NR_sigreturn, %eax ; int $0x80 - * - * WE DO NOT USE IT ANY MORE! It's only left here for historical - * reasons and because gdb uses it as a signature to notice - * signal handler stack frames. - */ - err |= __put_user(*((u64 *)&retcode), (u64 *)frame->retcode); - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->sp = (unsigned long)frame; - regs->ip = (unsigned long)ka->sa.sa_handler; - regs->ax = (unsigned long)sig; - regs->dx = 0; - regs->cx = 0; - - regs->ds = __USER_DS; - regs->es = __USER_DS; - regs->ss = __USER_DS; - regs->cs = __USER_CS; - - return 0; -} - -static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - void __user *restorer; - int err = 0; - void __user *fpstate = NULL; - - frame = get_sigframe(ka, regs, sizeof(*frame), &fpstate); - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - err |= __put_user(sig, &frame->sig); - err |= __put_user(&frame->info, &frame->pinfo); - err |= __put_user(&frame->uc, &frame->puc); - err |= copy_siginfo_to_user(&frame->info, info); - if (err) - return -EFAULT; - - /* Create the ucontext. */ - if (cpu_has_xsave) - err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); - else - err |= __put_user(0, &frame->uc.uc_flags); - err |= __put_user(0, &frame->uc.uc_link); - err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp); - err |= __put_user(sas_ss_flags(regs->sp), - &frame->uc.uc_stack.ss_flags); - err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, fpstate, - regs, set->sig[0]); - err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); - if (err) - return -EFAULT; - - /* Set up to return from userspace. */ - restorer = VDSO32_SYMBOL(current->mm->context.vdso, rt_sigreturn); - if (ka->sa.sa_flags & SA_RESTORER) - restorer = ka->sa.sa_restorer; - err |= __put_user(restorer, &frame->pretcode); - - /* - * This is movl $__NR_rt_sigreturn, %ax ; int $0x80 - * - * WE DO NOT USE IT ANY MORE! It's only left here for historical - * reasons and because gdb uses it as a signature to notice - * signal handler stack frames. - */ - err |= __put_user(*((u64 *)&rt_retcode), (u64 *)frame->retcode); - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->sp = (unsigned long)frame; - regs->ip = (unsigned long)ka->sa.sa_handler; - regs->ax = (unsigned long)sig; - regs->dx = (unsigned long)&frame->info; - regs->cx = (unsigned long)&frame->uc; - - regs->ds = __USER_DS; - regs->es = __USER_DS; - regs->ss = __USER_DS; - regs->cs = __USER_CS; - - return 0; -} -#else /* !CONFIG_X86_32 */ -/* - * Determine which stack to use.. - */ -static void __user * -get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size) -{ - /* Default to using normal stack - redzone*/ - sp -= 128; - - /* This is the X/Open sanctioned signal stack switching. */ - if (ka->sa.sa_flags & SA_ONSTACK) { - if (sas_ss_flags(sp) == 0) - sp = current->sas_ss_sp + current->sas_ss_size; - } - - return (void __user *)round_down(sp - size, 64); -} - -static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - void __user *fp = NULL; - int err = 0; - struct task_struct *me = current; - - if (used_math()) { - fp = get_stack(ka, regs->sp, sig_xstate_size); - frame = (void __user *)round_down( - (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8; - - if (save_i387_xstate(fp) < 0) - return -EFAULT; - } else - frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8; - - if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) - return -EFAULT; - - if (ka->sa.sa_flags & SA_SIGINFO) { - if (copy_siginfo_to_user(&frame->info, info)) - return -EFAULT; - } - - /* Create the ucontext. */ - if (cpu_has_xsave) - err |= __put_user(UC_FP_XSTATE, &frame->uc.uc_flags); - else - err |= __put_user(0, &frame->uc.uc_flags); - err |= __put_user(0, &frame->uc.uc_link); - err |= __put_user(me->sas_ss_sp, &frame->uc.uc_stack.ss_sp); - err |= __put_user(sas_ss_flags(regs->sp), - &frame->uc.uc_stack.ss_flags); - err |= __put_user(me->sas_ss_size, &frame->uc.uc_stack.ss_size); - err |= setup_sigcontext(&frame->uc.uc_mcontext, fp, regs, set->sig[0]); - err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set)); - - /* Set up to return from userspace. If provided, use a stub - already in userspace. */ - /* x86-64 should always use SA_RESTORER. */ - if (ka->sa.sa_flags & SA_RESTORER) { - err |= __put_user(ka->sa.sa_restorer, &frame->pretcode); - } else { - /* could use a vstub here */ - return -EFAULT; - } - - if (err) - return -EFAULT; - - /* Set up registers for signal handler */ - regs->di = sig; - /* In case the signal handler was declared without prototypes */ - regs->ax = 0; - - /* This also works for non SA_SIGINFO handlers because they expect the - next argument after the signal number on the stack. */ - regs->si = (unsigned long)&frame->info; - regs->dx = (unsigned long)&frame->uc; - regs->ip = (unsigned long) ka->sa.sa_handler; - - regs->sp = (unsigned long)frame; - - /* Set up the CS register to run signal handlers in 64-bit mode, - even if the handler happens to be interrupting 32-bit code. */ - regs->cs = __USER_CS; - - return 0; -} -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_X86_32 -/* - * Atomically swap in the new signal mask, and wait for a signal. - */ -asmlinkage int -sys_sigsuspend(int history0, int history1, old_sigset_t mask) -{ - mask &= _BLOCKABLE; - spin_lock_irq(¤t->sighand->siglock); - current->saved_sigmask = current->blocked; - siginitset(¤t->blocked, mask); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - current->state = TASK_INTERRUPTIBLE; - schedule(); - set_restore_sigmask(); - - return -ERESTARTNOHAND; -} - -asmlinkage int -sys_sigaction(int sig, const struct old_sigaction __user *act, - struct old_sigaction __user *oact) -{ - struct k_sigaction new_ka, old_ka; - int ret; - - if (act) { - old_sigset_t mask; - - if (!access_ok(VERIFY_READ, act, sizeof(*act)) || - __get_user(new_ka.sa.sa_handler, &act->sa_handler) || - __get_user(new_ka.sa.sa_restorer, &act->sa_restorer)) - return -EFAULT; - - __get_user(new_ka.sa.sa_flags, &act->sa_flags); - __get_user(mask, &act->sa_mask); - siginitset(&new_ka.sa.sa_mask, mask); - } - - ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL); - - if (!ret && oact) { - if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) || - __put_user(old_ka.sa.sa_handler, &oact->sa_handler) || - __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer)) - return -EFAULT; - - __put_user(old_ka.sa.sa_flags, &oact->sa_flags); - __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask); - } - - return ret; -} -#endif /* CONFIG_X86_32 */ - -#ifdef CONFIG_X86_32 -asmlinkage int sys_sigaltstack(unsigned long bx) -{ - /* - * This is needed to make gcc realize it doesn't own the - * "struct pt_regs" - */ - struct pt_regs *regs = (struct pt_regs *)&bx; - const stack_t __user *uss = (const stack_t __user *)bx; - stack_t __user *uoss = (stack_t __user *)regs->cx; - - return do_sigaltstack(uss, uoss, regs->sp); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long -sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, - struct pt_regs *regs) -{ - return do_sigaltstack(uss, uoss, regs->sp); -} -#endif /* CONFIG_X86_32 */ - -/* - * Do a signal return; undo the signal stack. - */ -#ifdef CONFIG_X86_32 -asmlinkage unsigned long sys_sigreturn(unsigned long __unused) -{ - struct sigframe __user *frame; - struct pt_regs *regs; - unsigned long ax; - sigset_t set; - - regs = (struct pt_regs *) &__unused; - frame = (struct sigframe __user *)(regs->sp - 8); - - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__get_user(set.sig[0], &frame->sc.oldmask) || (_NSIG_WORDS > 1 - && __copy_from_user(&set.sig[1], &frame->extramask, - sizeof(frame->extramask)))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->sc, &ax)) - goto badframe; - return ax; - -badframe: - if (show_unhandled_signals && printk_ratelimit()) { - printk("%s%s[%d] bad frame in sigreturn frame:" - "%p ip:%lx sp:%lx oeax:%lx", - task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, - current->comm, task_pid_nr(current), frame, regs->ip, - regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, current); - - return 0; -} -#endif /* CONFIG_X86_32 */ - -static long do_rt_sigreturn(struct pt_regs *regs) -{ - struct rt_sigframe __user *frame; - unsigned long ax; - sigset_t set; - - frame = (struct rt_sigframe __user *)(regs->sp - sizeof(long)); - if (!access_ok(VERIFY_READ, frame, sizeof(*frame))) - goto badframe; - if (__copy_from_user(&set, &frame->uc.uc_sigmask, sizeof(set))) - goto badframe; - - sigdelsetmask(&set, ~_BLOCKABLE); - spin_lock_irq(¤t->sighand->siglock); - current->blocked = set; - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - if (restore_sigcontext(regs, &frame->uc.uc_mcontext, &ax)) - goto badframe; - - if (do_sigaltstack(&frame->uc.uc_stack, NULL, regs->sp) == -EFAULT) - goto badframe; - - return ax; - -badframe: - signal_fault(regs, frame, "rt_sigreturn"); - return 0; -} - -#ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) -{ - struct pt_regs *regs = (struct pt_regs *)&__unused; - - return do_rt_sigreturn(regs); -} -#else /* !CONFIG_X86_32 */ -asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -{ - return do_rt_sigreturn(regs); -} -#endif /* CONFIG_X86_32 */ - -/* - * OK, we're invoking a handler: - */ -static int signr_convert(int sig) -{ -#ifdef CONFIG_X86_32 - struct thread_info *info = current_thread_info(); - - if (info->exec_domain && info->exec_domain->signal_invmap && sig < 32) - return info->exec_domain->signal_invmap[sig]; -#endif /* CONFIG_X86_32 */ - return sig; -} - -#ifdef CONFIG_X86_32 - -#define is_ia32 1 -#define ia32_setup_frame __setup_frame -#define ia32_setup_rt_frame __setup_rt_frame - -#else /* !CONFIG_X86_32 */ - -#ifdef CONFIG_IA32_EMULATION -#define is_ia32 test_thread_flag(TIF_IA32) -#else /* !CONFIG_IA32_EMULATION */ -#define is_ia32 0 -#endif /* CONFIG_IA32_EMULATION */ - -#endif /* CONFIG_X86_32 */ - -static int -setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs) -{ - int usig = signr_convert(sig); - int ret; - - /* Set up the stack frame */ - if (is_ia32) { - if (ka->sa.sa_flags & SA_SIGINFO) - ret = ia32_setup_rt_frame(usig, ka, info, set, regs); - else - ret = ia32_setup_frame(usig, ka, set, regs); - } else - ret = __setup_rt_frame(sig, ka, info, set, regs); - - if (ret) { - force_sigsegv(sig, current); - return -EFAULT; - } - - return ret; -} - -static int -handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka, - sigset_t *oldset, struct pt_regs *regs) -{ - int ret; - - /* Are we from a system call? */ - if (syscall_get_nr(current, regs) >= 0) { - /* If so, check system call restarting.. */ - switch (syscall_get_error(current, regs)) { - case -ERESTART_RESTARTBLOCK: - case -ERESTARTNOHAND: - regs->ax = -EINTR; - break; - - case -ERESTARTSYS: - if (!(ka->sa.sa_flags & SA_RESTART)) { - regs->ax = -EINTR; - break; - } - /* fallthrough */ - case -ERESTARTNOINTR: - regs->ax = regs->orig_ax; - regs->ip -= 2; - break; - } - } - - /* - * If TF is set due to a debugger (TIF_FORCED_TF), clear the TF - * flag so that register information in the sigcontext is correct. - */ - if (unlikely(regs->flags & X86_EFLAGS_TF) && - likely(test_and_clear_thread_flag(TIF_FORCED_TF))) - regs->flags &= ~X86_EFLAGS_TF; - - ret = setup_rt_frame(sig, ka, info, oldset, regs); - - if (ret) - return ret; - -#ifdef CONFIG_X86_64 - /* - * This has nothing to do with segment registers, - * despite the name. This magic affects uaccess.h - * macros' behavior. Reset it to the normal setting. - */ - set_fs(USER_DS); -#endif - - /* - * Clear the direction flag as per the ABI for function entry. - */ - regs->flags &= ~X86_EFLAGS_DF; - - /* - * Clear TF when entering the signal handler, but - * notify any tracer that was single-stepping it. - * The tracer may want to single-step inside the - * handler too. - */ - regs->flags &= ~X86_EFLAGS_TF; - - spin_lock_irq(¤t->sighand->siglock); - sigorsets(¤t->blocked, ¤t->blocked, &ka->sa.sa_mask); - if (!(ka->sa.sa_flags & SA_NODEFER)) - sigaddset(¤t->blocked, sig); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - - tracehook_signal_handler(sig, info, ka, regs, - test_thread_flag(TIF_SINGLESTEP)); - - return 0; -} - -#ifdef CONFIG_X86_32 -#define NR_restart_syscall __NR_restart_syscall -#else /* !CONFIG_X86_32 */ -#define NR_restart_syscall \ - test_thread_flag(TIF_IA32) ? __NR_ia32_restart_syscall : __NR_restart_syscall -#endif /* CONFIG_X86_32 */ - -/* - * Note that 'init' is a special process: it doesn't get signals it doesn't - * want to handle. Thus you cannot kill init even with a SIGKILL even by - * mistake. - */ -static void do_signal(struct pt_regs *regs) -{ - struct k_sigaction ka; - siginfo_t info; - int signr; - sigset_t *oldset; - - /* - * We want the common case to go fast, which is why we may in certain - * cases get here from kernel mode. Just return without doing anything - * if so. - * X86_32: vm86 regs switched out by assembly code before reaching - * here, so testing against kernel CS suffices. - */ - if (!user_mode(regs)) - return; - - if (current_thread_info()->status & TS_RESTORE_SIGMASK) - oldset = ¤t->saved_sigmask; - else - oldset = ¤t->blocked; - - signr = get_signal_to_deliver(&info, &ka, regs, NULL); - if (signr > 0) { - /* - * Re-enable any watchpoints before delivering the - * signal to user space. The processor register will - * have been cleared if the watchpoint triggered - * inside the kernel. - */ - if (current->thread.debugreg7) - set_debugreg(current->thread.debugreg7, 7); - - /* Whee! Actually deliver the signal. */ - if (handle_signal(signr, &info, &ka, oldset, regs) == 0) { - /* - * A signal was successfully delivered; the saved - * sigmask will have been stored in the signal frame, - * and will be restored by sigreturn, so we can simply - * clear the TS_RESTORE_SIGMASK flag. - */ - current_thread_info()->status &= ~TS_RESTORE_SIGMASK; - } - return; - } - - /* Did we come from a system call? */ - if (syscall_get_nr(current, regs) >= 0) { - /* Restart the system call - no handlers present */ - switch (syscall_get_error(current, regs)) { - case -ERESTARTNOHAND: - case -ERESTARTSYS: - case -ERESTARTNOINTR: - regs->ax = regs->orig_ax; - regs->ip -= 2; - break; - - case -ERESTART_RESTARTBLOCK: - regs->ax = NR_restart_syscall; - regs->ip -= 2; - break; - } - } - - /* - * If there's no signal to deliver, we just put the saved sigmask - * back. - */ - if (current_thread_info()->status & TS_RESTORE_SIGMASK) { - current_thread_info()->status &= ~TS_RESTORE_SIGMASK; - sigprocmask(SIG_SETMASK, ¤t->saved_sigmask, NULL); - } -} - -/* - * notification of userspace execution resumption - * - triggered by the TIF_WORK_MASK flags - */ -void -do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) -{ -#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE) - /* notify userspace of pending MCEs */ - if (thread_info_flags & _TIF_MCE_NOTIFY) - mce_notify_user(); -#endif /* CONFIG_X86_64 && CONFIG_X86_MCE */ - - /* deal with pending signal delivery */ - if (thread_info_flags & _TIF_SIGPENDING) - do_signal(regs); - - if (thread_info_flags & _TIF_NOTIFY_RESUME) { - clear_thread_flag(TIF_NOTIFY_RESUME); - tracehook_notify_resume(regs); - } - -#ifdef CONFIG_X86_32 - clear_thread_flag(TIF_IRET); -#endif /* CONFIG_X86_32 */ -} - -void signal_fault(struct pt_regs *regs, void __user *frame, char *where) -{ - struct task_struct *me = current; - - if (show_unhandled_signals && printk_ratelimit()) { - printk(KERN_INFO - "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx", - me->comm, me->pid, where, frame, - regs->ip, regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, me); -} -- cgit v1.2.3 From 5a45cfe1c64862e8cd3b0d79d7c4ba71c3118915 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 26 Nov 2008 00:16:24 -0500 Subject: ftrace: use code patching for ftrace graph tracer Impact: more efficient code for ftrace graph tracer This patch uses the dynamic patching, when available, to patch the function graph code into the kernel. This patch will ease the way for letting both function tracing and function graph tracing run together. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 5 +++++ arch/x86/kernel/ftrace.c | 48 ++++++++++++++++++++++++++++++++++++++++++++-- 2 files changed, 51 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 7def9fd5c1e6..958af86186c4 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1174,6 +1174,11 @@ ftrace_call: popl %edx popl %ecx popl %eax +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +.globl ftrace_graph_call +ftrace_graph_call: + jmp ftrace_stub +#endif .globl ftrace_stub ftrace_stub: diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 26b2d92d48b3..7ef914e6a2f6 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -111,7 +111,6 @@ static void ftrace_mod_code(void) */ mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode, MCOUNT_INSN_SIZE); - } void ftrace_nmi_enter(void) @@ -325,7 +324,51 @@ int __init ftrace_dyn_arch_init(void *data) #ifdef CONFIG_FUNCTION_GRAPH_TRACER -#ifndef CONFIG_DYNAMIC_FTRACE +#ifdef CONFIG_DYNAMIC_FTRACE +extern void ftrace_graph_call(void); + +static int ftrace_mod_jmp(unsigned long ip, + int old_offset, int new_offset) +{ + unsigned char code[MCOUNT_INSN_SIZE]; + + if (probe_kernel_read(code, (void *)ip, MCOUNT_INSN_SIZE)) + return -EFAULT; + + if (code[0] != 0xe9 || old_offset != *(int *)(&code[1])) + return -EINVAL; + + *(int *)(&code[1]) = new_offset; + + if (do_ftrace_mod_code(ip, &code)) + return -EPERM; + + return 0; +} + +int ftrace_enable_ftrace_graph_caller(void) +{ + unsigned long ip = (unsigned long)(&ftrace_graph_call); + int old_offset, new_offset; + + old_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE); + new_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE); + + return ftrace_mod_jmp(ip, old_offset, new_offset); +} + +int ftrace_disable_ftrace_graph_caller(void) +{ + unsigned long ip = (unsigned long)(&ftrace_graph_call); + int old_offset, new_offset; + + old_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE); + new_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE); + + return ftrace_mod_jmp(ip, old_offset, new_offset); +} + +#else /* CONFIG_DYNAMIC_FTRACE */ /* * These functions are picked from those used on @@ -343,6 +386,7 @@ void ftrace_nmi_exit(void) { atomic_dec(&in_nmi); } + #endif /* !CONFIG_DYNAMIC_FTRACE */ /* Add a function return address to the trace stack on thread info.*/ -- cgit v1.2.3 From f3f47a6768a29448866da4422b6f6bee485c947f Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sun, 23 Nov 2008 16:49:58 -0800 Subject: tracing: add "power-tracer": C/P state tracer to help power optimization Impact: new "power-tracer" ftrace plugin This patch adds a C/P-state ftrace plugin that will generate detailed statistics about the C/P-states that are being used, so that we can look at detailed decisions that the C/P-state code is making, rather than the too high level "average" that we have today. An example way of using this is: mount -t debugfs none /sys/kernel/debug echo cstate > /sys/kernel/debug/tracing/current_tracer echo 1 > /sys/kernel/debug/tracing/tracing_enabled sleep 1 echo 0 > /sys/kernel/debug/tracing/tracing_enabled cat /sys/kernel/debug/tracing/trace | perl scripts/trace/cstate.pl > out.svg Signed-off-by: Arjan van de Ven Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c | 4 ++++ arch/x86/kernel/process.c | 16 ++++++++++++++++ 2 files changed, 20 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 8e48c5d4467d..88ea02dcb622 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -33,6 +33,7 @@ #include #include #include +#include #include #include @@ -391,6 +392,7 @@ static int acpi_cpufreq_target(struct cpufreq_policy *policy, unsigned int next_perf_state = 0; /* Index into perf table */ unsigned int i; int result = 0; + struct power_trace it; dprintk("acpi_cpufreq_target %d (%d)\n", target_freq, policy->cpu); @@ -427,6 +429,8 @@ static int acpi_cpufreq_target(struct cpufreq_policy *policy, } } + trace_power_mark(&it, POWER_PSTATE, next_perf_state); + switch (data->cpu_feature) { case SYSTEM_INTEL_MSR_CAPABLE: cmd.type = SYSTEM_INTEL_MSR_CAPABLE; diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c622772744d8..c27af49a4ede 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -7,6 +7,7 @@ #include #include #include +#include #include unsigned long idle_halt; @@ -100,6 +101,9 @@ static inline int hlt_use_halt(void) void default_idle(void) { if (hlt_use_halt()) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, 1); current_thread_info()->status &= ~TS_POLLING; /* * TS_POLLING-cleared state must be visible before we @@ -112,6 +116,7 @@ void default_idle(void) else local_irq_enable(); current_thread_info()->status |= TS_POLLING; + trace_power_end(&it); } else { local_irq_enable(); /* loop is done by the caller */ @@ -154,24 +159,31 @@ EXPORT_SYMBOL_GPL(cpu_idle_wait); */ void mwait_idle_with_hints(unsigned long ax, unsigned long cx) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, (ax>>4)+1); if (!need_resched()) { __monitor((void *)¤t_thread_info()->flags, 0, 0); smp_mb(); if (!need_resched()) __mwait(ax, cx); } + trace_power_end(&it); } /* Default MONITOR/MWAIT with no hints, used for default C1 state */ static void mwait_idle(void) { + struct power_trace it; if (!need_resched()) { + trace_power_start(&it, POWER_CSTATE, 1); __monitor((void *)¤t_thread_info()->flags, 0, 0); smp_mb(); if (!need_resched()) __sti_mwait(0, 0); else local_irq_enable(); + trace_power_end(&it); } else local_irq_enable(); } @@ -183,9 +195,13 @@ static void mwait_idle(void) */ static void poll_idle(void) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, 0); local_irq_enable(); while (!need_resched()) cpu_relax(); + trace_power_end(&it); } /* -- cgit v1.2.3 From 4db646b1af8fdcf01d690d29eeae44cd937edb0d Mon Sep 17 00:00:00 2001 From: Hannes Eder Date: Sun, 23 Nov 2008 20:49:52 +0100 Subject: x86: microcode: fix sparse warnings Impact: make global variables and a function static Fix following sparse warnings: arch/x86/kernel/microcode_core.c:102:22: warning: symbol 'microcode_ops' was not declared. Should it be static? arch/x86/kernel/microcode_core.c:206:24: warning: symbol 'microcode_pdev' was not declared. Should it be static? arch/x86/kernel/microcode_core.c:322:6: warning: symbol 'microcode_update_cpu' was not declared. Should it be static? arch/x86/kernel/microcode_intel.c:468:22: warning: symbol 'microcode_intel_ops' was not declared. Should it be static? Signed-off-by: Hannes Eder Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_core.c | 6 +++--- arch/x86/kernel/microcode_intel.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c index 82fb2809ce32..5b711a534495 100644 --- a/arch/x86/kernel/microcode_core.c +++ b/arch/x86/kernel/microcode_core.c @@ -99,7 +99,7 @@ MODULE_LICENSE("GPL"); #define MICROCODE_VERSION "2.00" -struct microcode_ops *microcode_ops; +static struct microcode_ops *microcode_ops; /* no concurrent ->write()s are allowed on /dev/cpu/microcode */ static DEFINE_MUTEX(microcode_mutex); @@ -203,7 +203,7 @@ MODULE_ALIAS_MISCDEV(MICROCODE_MINOR); #endif /* fake device for request_firmware */ -struct platform_device *microcode_pdev; +static struct platform_device *microcode_pdev; static ssize_t reload_store(struct sys_device *dev, struct sysdev_attribute *attr, @@ -319,7 +319,7 @@ static int microcode_resume_cpu(int cpu) return 0; } -void microcode_update_cpu(int cpu) +static void microcode_update_cpu(int cpu) { struct ucode_cpu_info *uci = ucode_cpu_info + cpu; int err = 0; diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c index 622dc4a21784..c34c820ee486 100644 --- a/arch/x86/kernel/microcode_intel.c +++ b/arch/x86/kernel/microcode_intel.c @@ -465,7 +465,7 @@ static void microcode_fini_cpu(int cpu) uci->mc = NULL; } -struct microcode_ops microcode_intel_ops = { +static struct microcode_ops microcode_intel_ops = { .request_microcode_user = request_microcode_user, .request_microcode_fw = request_microcode_fw, .collect_cpu_info = collect_cpu_info, -- cgit v1.2.3 From ddeb8f2149de280d54f0c8910cead42e6042b2cb Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Mon, 24 Nov 2008 13:24:28 +0100 Subject: x86_64: get rid of the use of KPROBE_ENTRY / KPROBE_END Impact: clean up assembly macros and annotations - with some object impact entry_64.S is the only user of KPROBE_ENTRY / KPROBE_END on x86_64. This patch reorders entry_64.S and explicitly generates a separate section for functions that need the protection. The generated code before and after the patch is equal. Implicitly changing sections in assembly files makes it more difficult to follow why the assembler is doing certain things. For example, .p2align 5 KPROBE_ENTRY(...) was not doing what you would expect. Other section changes (__ex_table, .fixup, .init.rodata) are done explicitly already. Signed-off-by: Alexander van Heukelum Acked-by: Jan Beulich Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 444 ++++++++++++++++++++++----------------------- 1 file changed, 220 insertions(+), 224 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index f2d546e16354..38fcd0517c31 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1002,7 +1002,7 @@ END(\sym) .endm .macro paranoidzeroentry sym do_sym -KPROBE_ENTRY(\sym) +ENTRY(\sym) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $-1 /* ORIG_RAX: no syscall to restart */ @@ -1015,11 +1015,11 @@ KPROBE_ENTRY(\sym) call \do_sym jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC -KPROBE_END(\sym) +END(\sym) .endm .macro paranoidzeroentry_ist sym do_sym ist -KPROBE_ENTRY(\sym) +ENTRY(\sym) INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $-1 /* ORIG_RAX: no syscall to restart */ @@ -1035,15 +1035,11 @@ KPROBE_ENTRY(\sym) addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC -KPROBE_END(\sym) +END(\sym) .endm -.macro errorentry sym do_sym entry=0 -.if \entry -KPROBE_ENTRY(\sym) -.else +.macro errorentry sym do_sym ENTRY(\sym) -.endif XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME subq $15*8,%rsp @@ -1056,20 +1052,12 @@ ENTRY(\sym) call \do_sym jmp error_exit /* %ebx: no swapgs flag */ CFI_ENDPROC -.if \entry -KPROBE_END(\sym) -.else END(\sym) -.endif .endm /* error code is on the stack already */ -.macro paranoiderrorentry sym do_sym entry=1 -.if \entry -KPROBE_ENTRY(\sym) -.else +.macro paranoiderrorentry sym do_sym ENTRY(\sym) -.endif XCPT_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME subq $15*8,%rsp @@ -1083,166 +1071,23 @@ ENTRY(\sym) call \do_sym jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC -.if \entry -KPROBE_END(\sym) -.else END(\sym) -.endif .endm zeroentry divide_error do_divide_error -paranoidzeroentry_ist debug do_debug DEBUG_STACK -paranoidzeroentry_ist int3 do_int3 DEBUG_STACK zeroentry overflow do_overflow zeroentry bounds do_bounds zeroentry invalid_op do_invalid_op zeroentry device_not_available do_device_not_available -paranoiderrorentry double_fault do_double_fault 0 +paranoiderrorentry double_fault do_double_fault zeroentry coprocessor_segment_overrun do_coprocessor_segment_overrun errorentry invalid_TSS do_invalid_TSS errorentry segment_not_present do_segment_not_present -paranoiderrorentry stack_segment do_stack_segment -errorentry general_protection do_general_protection 1 -errorentry page_fault do_page_fault 1 zeroentry spurious_interrupt_bug do_spurious_interrupt_bug zeroentry coprocessor_error do_coprocessor_error errorentry alignment_check do_alignment_check -#ifdef CONFIG_X86_MCE -paranoidzeroentry machine_check do_machine_check -#endif zeroentry simd_coprocessor_error do_simd_coprocessor_error - /* - * "Paranoid" exit path from exception stack. - * Paranoid because this is used by NMIs and cannot take - * any kernel state for granted. - * We don't do kernel preemption checks here, because only - * NMI should be common and it does not enable IRQs and - * cannot get reschedule ticks. - * - * "trace" is 0 for the NMI handler only, because irq-tracing - * is fundamentally NMI-unsafe. (we cannot change the soft and - * hard flags at once, atomically) - */ - - /* ebx: no swapgs flag */ -KPROBE_ENTRY(paranoid_exit) - INTR_FRAME - DISABLE_INTERRUPTS(CLBR_NONE) - TRACE_IRQS_OFF - testl %ebx,%ebx /* swapgs needed? */ - jnz paranoid_restore - testl $3,CS(%rsp) - jnz paranoid_userspace -paranoid_swapgs: - TRACE_IRQS_IRETQ 0 - SWAPGS_UNSAFE_STACK -paranoid_restore: - RESTORE_ALL 8 - jmp irq_return -paranoid_userspace: - GET_THREAD_INFO(%rcx) - movl TI_flags(%rcx),%ebx - andl $_TIF_WORK_MASK,%ebx - jz paranoid_swapgs - movq %rsp,%rdi /* &pt_regs */ - call sync_regs - movq %rax,%rsp /* switch stack for scheduling */ - testl $_TIF_NEED_RESCHED,%ebx - jnz paranoid_schedule - movl %ebx,%edx /* arg3: thread flags */ - TRACE_IRQS_ON - ENABLE_INTERRUPTS(CLBR_NONE) - xorl %esi,%esi /* arg2: oldset */ - movq %rsp,%rdi /* arg1: &pt_regs */ - call do_notify_resume - DISABLE_INTERRUPTS(CLBR_NONE) - TRACE_IRQS_OFF - jmp paranoid_userspace -paranoid_schedule: - TRACE_IRQS_ON - ENABLE_INTERRUPTS(CLBR_ANY) - call schedule - DISABLE_INTERRUPTS(CLBR_ANY) - TRACE_IRQS_OFF - jmp paranoid_userspace - CFI_ENDPROC -KPROBE_END(paranoid_exit) - -/* - * Exception entry point. This expects an error code/orig_rax on the stack. - * returns in "no swapgs flag" in %ebx. - */ -KPROBE_ENTRY(error_entry) - XCPT_FRAME - CFI_ADJUST_CFA_OFFSET 15*8 - /* oldrax contains error code */ - cld - movq_cfi rdi, RDI+8 - movq_cfi rsi, RSI+8 - movq_cfi rdx, RDX+8 - movq_cfi rcx, RCX+8 - movq_cfi rax, RAX+8 - movq_cfi r8, R8+8 - movq_cfi r9, R9+8 - movq_cfi r10, R10+8 - movq_cfi r11, R11+8 - movq_cfi rbx, RBX+8 - movq_cfi rbp, RBP+8 - movq_cfi r12, R12+8 - movq_cfi r13, R13+8 - movq_cfi r14, R14+8 - movq_cfi r15, R15+8 - xorl %ebx,%ebx - testl $3,CS+8(%rsp) - je error_kernelspace -error_swapgs: - SWAPGS -error_sti: - TRACE_IRQS_OFF - ret - CFI_ENDPROC - -/* - * There are two places in the kernel that can potentially fault with - * usergs. Handle them here. The exception handlers after iret run with - * kernel gs again, so don't set the user space flag. B stepping K8s - * sometimes report an truncated RIP for IRET exceptions returning to - * compat mode. Check for these here too. - */ -error_kernelspace: - incl %ebx - leaq irq_return(%rip),%rcx - cmpq %rcx,RIP+8(%rsp) - je error_swapgs - movl %ecx,%ecx /* zero extend */ - cmpq %rcx,RIP+8(%rsp) - je error_swapgs - cmpq $gs_change,RIP+8(%rsp) - je error_swapgs - jmp error_sti -KPROBE_END(error_entry) - - -/* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ -KPROBE_ENTRY(error_exit) - DEFAULT_FRAME - movl %ebx,%eax - RESTORE_REST - DISABLE_INTERRUPTS(CLBR_NONE) - TRACE_IRQS_OFF - GET_THREAD_INFO(%rcx) - testl %eax,%eax - jne retint_kernel - LOCKDEP_SYS_EXIT_IRQ - movl TI_flags(%rcx),%edx - movl $_TIF_WORK_MASK,%edi - andl %edi,%edx - jnz retint_careful - jmp retint_swapgs - CFI_ENDPROC -KPROBE_END(error_exit) - /* Reload gs selector with exception handling */ /* edi: new selector */ ENTRY(native_load_gs_index) @@ -1362,61 +1207,6 @@ ENTRY(kernel_execve) CFI_ENDPROC END(kernel_execve) - /* runs on exception stack */ -KPROBE_ENTRY(nmi) - INTR_FRAME - PARAVIRT_ADJUST_EXCEPTION_FRAME - pushq_cfi $-1 - subq $15*8, %rsp - CFI_ADJUST_CFA_OFFSET 15*8 - call save_paranoid - DEFAULT_FRAME 0 - /* paranoidentry do_nmi, 0; without TRACE_IRQS_OFF */ - movq %rsp,%rdi - movq $-1,%rsi - call do_nmi -#ifdef CONFIG_TRACE_IRQFLAGS - /* paranoidexit; without TRACE_IRQS_OFF */ - /* ebx: no swapgs flag */ - DISABLE_INTERRUPTS(CLBR_NONE) - testl %ebx,%ebx /* swapgs needed? */ - jnz nmi_restore - testl $3,CS(%rsp) - jnz nmi_userspace -nmi_swapgs: - SWAPGS_UNSAFE_STACK -nmi_restore: - RESTORE_ALL 8 - jmp irq_return -nmi_userspace: - GET_THREAD_INFO(%rcx) - movl TI_flags(%rcx),%ebx - andl $_TIF_WORK_MASK,%ebx - jz nmi_swapgs - movq %rsp,%rdi /* &pt_regs */ - call sync_regs - movq %rax,%rsp /* switch stack for scheduling */ - testl $_TIF_NEED_RESCHED,%ebx - jnz nmi_schedule - movl %ebx,%edx /* arg3: thread flags */ - ENABLE_INTERRUPTS(CLBR_NONE) - xorl %esi,%esi /* arg2: oldset */ - movq %rsp,%rdi /* arg1: &pt_regs */ - call do_notify_resume - DISABLE_INTERRUPTS(CLBR_NONE) - jmp nmi_userspace -nmi_schedule: - ENABLE_INTERRUPTS(CLBR_ANY) - call schedule - DISABLE_INTERRUPTS(CLBR_ANY) - jmp nmi_userspace - CFI_ENDPROC -#else - jmp paranoid_exit - CFI_ENDPROC -#endif -KPROBE_END(nmi) - /* Call softirq on interrupt stack. Interrupts are off. */ ENTRY(call_softirq) CFI_STARTPROC @@ -1437,13 +1227,6 @@ ENTRY(call_softirq) CFI_ENDPROC END(call_softirq) -KPROBE_ENTRY(ignore_sysret) - CFI_STARTPROC - mov $-ENOSYS,%eax - sysret - CFI_ENDPROC -KPROBE_END(ignore_sysret) - #ifdef CONFIG_XEN zeroentry xen_hypervisor_callback xen_do_hypervisor_callback @@ -1540,3 +1323,216 @@ ENTRY(xen_failsafe_callback) END(xen_failsafe_callback) #endif /* CONFIG_XEN */ + +/* + * Some functions should be protected against kprobes + */ + .pushsection .kprobes.text, "ax" + +paranoidzeroentry_ist debug do_debug DEBUG_STACK +paranoidzeroentry_ist int3 do_int3 DEBUG_STACK +paranoiderrorentry stack_segment do_stack_segment +errorentry general_protection do_general_protection +errorentry page_fault do_page_fault +#ifdef CONFIG_X86_MCE +paranoidzeroentry machine_check do_machine_check +#endif + + /* + * "Paranoid" exit path from exception stack. + * Paranoid because this is used by NMIs and cannot take + * any kernel state for granted. + * We don't do kernel preemption checks here, because only + * NMI should be common and it does not enable IRQs and + * cannot get reschedule ticks. + * + * "trace" is 0 for the NMI handler only, because irq-tracing + * is fundamentally NMI-unsafe. (we cannot change the soft and + * hard flags at once, atomically) + */ + + /* ebx: no swapgs flag */ +ENTRY(paranoid_exit) + INTR_FRAME + DISABLE_INTERRUPTS(CLBR_NONE) + TRACE_IRQS_OFF + testl %ebx,%ebx /* swapgs needed? */ + jnz paranoid_restore + testl $3,CS(%rsp) + jnz paranoid_userspace +paranoid_swapgs: + TRACE_IRQS_IRETQ 0 + SWAPGS_UNSAFE_STACK +paranoid_restore: + RESTORE_ALL 8 + jmp irq_return +paranoid_userspace: + GET_THREAD_INFO(%rcx) + movl TI_flags(%rcx),%ebx + andl $_TIF_WORK_MASK,%ebx + jz paranoid_swapgs + movq %rsp,%rdi /* &pt_regs */ + call sync_regs + movq %rax,%rsp /* switch stack for scheduling */ + testl $_TIF_NEED_RESCHED,%ebx + jnz paranoid_schedule + movl %ebx,%edx /* arg3: thread flags */ + TRACE_IRQS_ON + ENABLE_INTERRUPTS(CLBR_NONE) + xorl %esi,%esi /* arg2: oldset */ + movq %rsp,%rdi /* arg1: &pt_regs */ + call do_notify_resume + DISABLE_INTERRUPTS(CLBR_NONE) + TRACE_IRQS_OFF + jmp paranoid_userspace +paranoid_schedule: + TRACE_IRQS_ON + ENABLE_INTERRUPTS(CLBR_ANY) + call schedule + DISABLE_INTERRUPTS(CLBR_ANY) + TRACE_IRQS_OFF + jmp paranoid_userspace + CFI_ENDPROC +END(paranoid_exit) + +/* + * Exception entry point. This expects an error code/orig_rax on the stack. + * returns in "no swapgs flag" in %ebx. + */ +ENTRY(error_entry) + XCPT_FRAME + CFI_ADJUST_CFA_OFFSET 15*8 + /* oldrax contains error code */ + cld + movq_cfi rdi, RDI+8 + movq_cfi rsi, RSI+8 + movq_cfi rdx, RDX+8 + movq_cfi rcx, RCX+8 + movq_cfi rax, RAX+8 + movq_cfi r8, R8+8 + movq_cfi r9, R9+8 + movq_cfi r10, R10+8 + movq_cfi r11, R11+8 + movq_cfi rbx, RBX+8 + movq_cfi rbp, RBP+8 + movq_cfi r12, R12+8 + movq_cfi r13, R13+8 + movq_cfi r14, R14+8 + movq_cfi r15, R15+8 + xorl %ebx,%ebx + testl $3,CS+8(%rsp) + je error_kernelspace +error_swapgs: + SWAPGS +error_sti: + TRACE_IRQS_OFF + ret + CFI_ENDPROC + +/* + * There are two places in the kernel that can potentially fault with + * usergs. Handle them here. The exception handlers after iret run with + * kernel gs again, so don't set the user space flag. B stepping K8s + * sometimes report an truncated RIP for IRET exceptions returning to + * compat mode. Check for these here too. + */ +error_kernelspace: + incl %ebx + leaq irq_return(%rip),%rcx + cmpq %rcx,RIP+8(%rsp) + je error_swapgs + movl %ecx,%ecx /* zero extend */ + cmpq %rcx,RIP+8(%rsp) + je error_swapgs + cmpq $gs_change,RIP+8(%rsp) + je error_swapgs + jmp error_sti +END(error_entry) + + +/* ebx: no swapgs flag (1: don't need swapgs, 0: need it) */ +ENTRY(error_exit) + DEFAULT_FRAME + movl %ebx,%eax + RESTORE_REST + DISABLE_INTERRUPTS(CLBR_NONE) + TRACE_IRQS_OFF + GET_THREAD_INFO(%rcx) + testl %eax,%eax + jne retint_kernel + LOCKDEP_SYS_EXIT_IRQ + movl TI_flags(%rcx),%edx + movl $_TIF_WORK_MASK,%edi + andl %edi,%edx + jnz retint_careful + jmp retint_swapgs + CFI_ENDPROC +END(error_exit) + + + /* runs on exception stack */ +ENTRY(nmi) + INTR_FRAME + PARAVIRT_ADJUST_EXCEPTION_FRAME + pushq_cfi $-1 + subq $15*8, %rsp + CFI_ADJUST_CFA_OFFSET 15*8 + call save_paranoid + DEFAULT_FRAME 0 + /* paranoidentry do_nmi, 0; without TRACE_IRQS_OFF */ + movq %rsp,%rdi + movq $-1,%rsi + call do_nmi +#ifdef CONFIG_TRACE_IRQFLAGS + /* paranoidexit; without TRACE_IRQS_OFF */ + /* ebx: no swapgs flag */ + DISABLE_INTERRUPTS(CLBR_NONE) + testl %ebx,%ebx /* swapgs needed? */ + jnz nmi_restore + testl $3,CS(%rsp) + jnz nmi_userspace +nmi_swapgs: + SWAPGS_UNSAFE_STACK +nmi_restore: + RESTORE_ALL 8 + jmp irq_return +nmi_userspace: + GET_THREAD_INFO(%rcx) + movl TI_flags(%rcx),%ebx + andl $_TIF_WORK_MASK,%ebx + jz nmi_swapgs + movq %rsp,%rdi /* &pt_regs */ + call sync_regs + movq %rax,%rsp /* switch stack for scheduling */ + testl $_TIF_NEED_RESCHED,%ebx + jnz nmi_schedule + movl %ebx,%edx /* arg3: thread flags */ + ENABLE_INTERRUPTS(CLBR_NONE) + xorl %esi,%esi /* arg2: oldset */ + movq %rsp,%rdi /* arg1: &pt_regs */ + call do_notify_resume + DISABLE_INTERRUPTS(CLBR_NONE) + jmp nmi_userspace +nmi_schedule: + ENABLE_INTERRUPTS(CLBR_ANY) + call schedule + DISABLE_INTERRUPTS(CLBR_ANY) + jmp nmi_userspace + CFI_ENDPROC +#else + jmp paranoid_exit + CFI_ENDPROC +#endif +END(nmi) + +ENTRY(ignore_sysret) + CFI_STARTPROC + mov $-ENOSYS,%eax + sysret + CFI_ENDPROC +END(ignore_sysret) + +/* + * End of kprobes section + */ + .popsection -- cgit v1.2.3 From d211af055d0c12dc3416c2886e6fbdc6eb74a381 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Mon, 24 Nov 2008 15:38:45 +0100 Subject: i386: get rid of the use of KPROBE_ENTRY / KPROBE_END entry_32.S is now the only user of KPROBE_ENTRY / KPROBE_END, treewide. This patch reorders entry_64.S and explicitly generates a separate section for functions that need the protection. The generated code before and after the patch is equal. The KPROBE_ENTRY and KPROBE_END macro's are removed too. Signed-off-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 438 +++++++++++++++++++++++---------------------- 1 file changed, 224 insertions(+), 214 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index bd02ec77edc4..6e96028d1a9c 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -688,65 +688,6 @@ ENDPROC(name) /* The include is where all of the SMP etc. interrupts come from */ #include "entry_arch.h" -KPROBE_ENTRY(page_fault) - RING0_EC_FRAME - pushl $do_page_fault - CFI_ADJUST_CFA_OFFSET 4 - ALIGN -error_code: - /* the function address is in %fs's slot on the stack */ - pushl %es - CFI_ADJUST_CFA_OFFSET 4 - /*CFI_REL_OFFSET es, 0*/ - pushl %ds - CFI_ADJUST_CFA_OFFSET 4 - /*CFI_REL_OFFSET ds, 0*/ - pushl %eax - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET eax, 0 - pushl %ebp - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET ebp, 0 - pushl %edi - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET edi, 0 - pushl %esi - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET esi, 0 - pushl %edx - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET edx, 0 - pushl %ecx - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET ecx, 0 - pushl %ebx - CFI_ADJUST_CFA_OFFSET 4 - CFI_REL_OFFSET ebx, 0 - cld - pushl %fs - CFI_ADJUST_CFA_OFFSET 4 - /*CFI_REL_OFFSET fs, 0*/ - movl $(__KERNEL_PERCPU), %ecx - movl %ecx, %fs - UNWIND_ESPFIX_STACK - popl %ecx - CFI_ADJUST_CFA_OFFSET -4 - /*CFI_REGISTER es, ecx*/ - movl PT_FS(%esp), %edi # get the function address - movl PT_ORIG_EAX(%esp), %edx # get the error code - movl $-1, PT_ORIG_EAX(%esp) # no syscall to restart - mov %ecx, PT_FS(%esp) - /*CFI_REL_OFFSET fs, ES*/ - movl $(__USER_DS), %ecx - movl %ecx, %ds - movl %ecx, %es - TRACE_IRQS_OFF - movl %esp,%eax # pt_regs pointer - call *%edi - jmp ret_from_exception - CFI_ENDPROC -KPROBE_END(page_fault) - ENTRY(coprocessor_error) RING0_INT_FRAME pushl $0 @@ -777,140 +718,6 @@ ENTRY(device_not_available) CFI_ENDPROC END(device_not_available) -/* - * Debug traps and NMI can happen at the one SYSENTER instruction - * that sets up the real kernel stack. Check here, since we can't - * allow the wrong stack to be used. - * - * "TSS_sysenter_sp0+12" is because the NMI/debug handler will have - * already pushed 3 words if it hits on the sysenter instruction: - * eflags, cs and eip. - * - * We just load the right stack, and push the three (known) values - * by hand onto the new stack - while updating the return eip past - * the instruction that would have done it for sysenter. - */ -#define FIX_STACK(offset, ok, label) \ - cmpw $__KERNEL_CS,4(%esp); \ - jne ok; \ -label: \ - movl TSS_sysenter_sp0+offset(%esp),%esp; \ - CFI_DEF_CFA esp, 0; \ - CFI_UNDEFINED eip; \ - pushfl; \ - CFI_ADJUST_CFA_OFFSET 4; \ - pushl $__KERNEL_CS; \ - CFI_ADJUST_CFA_OFFSET 4; \ - pushl $sysenter_past_esp; \ - CFI_ADJUST_CFA_OFFSET 4; \ - CFI_REL_OFFSET eip, 0 - -KPROBE_ENTRY(debug) - RING0_INT_FRAME - cmpl $ia32_sysenter_target,(%esp) - jne debug_stack_correct - FIX_STACK(12, debug_stack_correct, debug_esp_fix_insn) -debug_stack_correct: - pushl $-1 # mark this as an int - CFI_ADJUST_CFA_OFFSET 4 - SAVE_ALL - TRACE_IRQS_OFF - xorl %edx,%edx # error code 0 - movl %esp,%eax # pt_regs pointer - call do_debug - jmp ret_from_exception - CFI_ENDPROC -KPROBE_END(debug) - -/* - * NMI is doubly nasty. It can happen _while_ we're handling - * a debug fault, and the debug fault hasn't yet been able to - * clear up the stack. So we first check whether we got an - * NMI on the sysenter entry path, but after that we need to - * check whether we got an NMI on the debug path where the debug - * fault happened on the sysenter path. - */ -KPROBE_ENTRY(nmi) - RING0_INT_FRAME - pushl %eax - CFI_ADJUST_CFA_OFFSET 4 - movl %ss, %eax - cmpw $__ESPFIX_SS, %ax - popl %eax - CFI_ADJUST_CFA_OFFSET -4 - je nmi_espfix_stack - cmpl $ia32_sysenter_target,(%esp) - je nmi_stack_fixup - pushl %eax - CFI_ADJUST_CFA_OFFSET 4 - movl %esp,%eax - /* Do not access memory above the end of our stack page, - * it might not exist. - */ - andl $(THREAD_SIZE-1),%eax - cmpl $(THREAD_SIZE-20),%eax - popl %eax - CFI_ADJUST_CFA_OFFSET -4 - jae nmi_stack_correct - cmpl $ia32_sysenter_target,12(%esp) - je nmi_debug_stack_check -nmi_stack_correct: - /* We have a RING0_INT_FRAME here */ - pushl %eax - CFI_ADJUST_CFA_OFFSET 4 - SAVE_ALL - TRACE_IRQS_OFF - xorl %edx,%edx # zero error code - movl %esp,%eax # pt_regs pointer - call do_nmi - jmp restore_nocheck_notrace - CFI_ENDPROC - -nmi_stack_fixup: - RING0_INT_FRAME - FIX_STACK(12,nmi_stack_correct, 1) - jmp nmi_stack_correct - -nmi_debug_stack_check: - /* We have a RING0_INT_FRAME here */ - cmpw $__KERNEL_CS,16(%esp) - jne nmi_stack_correct - cmpl $debug,(%esp) - jb nmi_stack_correct - cmpl $debug_esp_fix_insn,(%esp) - ja nmi_stack_correct - FIX_STACK(24,nmi_stack_correct, 1) - jmp nmi_stack_correct - -nmi_espfix_stack: - /* We have a RING0_INT_FRAME here. - * - * create the pointer to lss back - */ - pushl %ss - CFI_ADJUST_CFA_OFFSET 4 - pushl %esp - CFI_ADJUST_CFA_OFFSET 4 - addw $4, (%esp) - /* copy the iret frame of 12 bytes */ - .rept 3 - pushl 16(%esp) - CFI_ADJUST_CFA_OFFSET 4 - .endr - pushl %eax - CFI_ADJUST_CFA_OFFSET 4 - SAVE_ALL - TRACE_IRQS_OFF - FIXUP_ESPFIX_STACK # %eax == %esp - xorl %edx,%edx # zero error code - call do_nmi - RESTORE_REGS - lss 12+4(%esp), %esp # back to espfix stack - CFI_ADJUST_CFA_OFFSET -24 - jmp irq_return - CFI_ENDPROC -KPROBE_END(nmi) - #ifdef CONFIG_PARAVIRT ENTRY(native_iret) iret @@ -926,19 +733,6 @@ ENTRY(native_irq_enable_sysexit) END(native_irq_enable_sysexit) #endif -KPROBE_ENTRY(int3) - RING0_INT_FRAME - pushl $-1 # mark this as an int - CFI_ADJUST_CFA_OFFSET 4 - SAVE_ALL - TRACE_IRQS_OFF - xorl %edx,%edx # zero error code - movl %esp,%eax # pt_regs pointer - call do_int3 - jmp ret_from_exception - CFI_ENDPROC -KPROBE_END(int3) - ENTRY(overflow) RING0_INT_FRAME pushl $0 @@ -1003,14 +797,6 @@ ENTRY(stack_segment) CFI_ENDPROC END(stack_segment) -KPROBE_ENTRY(general_protection) - RING0_EC_FRAME - pushl $do_general_protection - CFI_ADJUST_CFA_OFFSET 4 - jmp error_code - CFI_ENDPROC -KPROBE_END(general_protection) - ENTRY(alignment_check) RING0_EC_FRAME pushl $do_alignment_check @@ -1220,3 +1006,227 @@ END(mcount) #include "syscall_table_32.S" syscall_table_size=(.-sys_call_table) + +/* + * Some functions should be protected against kprobes + */ + .pushsection .kprobes.text, "ax" + +ENTRY(page_fault) + RING0_EC_FRAME + pushl $do_page_fault + CFI_ADJUST_CFA_OFFSET 4 + ALIGN +error_code: + /* the function address is in %fs's slot on the stack */ + pushl %es + CFI_ADJUST_CFA_OFFSET 4 + /*CFI_REL_OFFSET es, 0*/ + pushl %ds + CFI_ADJUST_CFA_OFFSET 4 + /*CFI_REL_OFFSET ds, 0*/ + pushl %eax + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET eax, 0 + pushl %ebp + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET ebp, 0 + pushl %edi + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET edi, 0 + pushl %esi + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET esi, 0 + pushl %edx + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET edx, 0 + pushl %ecx + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET ecx, 0 + pushl %ebx + CFI_ADJUST_CFA_OFFSET 4 + CFI_REL_OFFSET ebx, 0 + cld + pushl %fs + CFI_ADJUST_CFA_OFFSET 4 + /*CFI_REL_OFFSET fs, 0*/ + movl $(__KERNEL_PERCPU), %ecx + movl %ecx, %fs + UNWIND_ESPFIX_STACK + popl %ecx + CFI_ADJUST_CFA_OFFSET -4 + /*CFI_REGISTER es, ecx*/ + movl PT_FS(%esp), %edi # get the function address + movl PT_ORIG_EAX(%esp), %edx # get the error code + movl $-1, PT_ORIG_EAX(%esp) # no syscall to restart + mov %ecx, PT_FS(%esp) + /*CFI_REL_OFFSET fs, ES*/ + movl $(__USER_DS), %ecx + movl %ecx, %ds + movl %ecx, %es + TRACE_IRQS_OFF + movl %esp,%eax # pt_regs pointer + call *%edi + jmp ret_from_exception + CFI_ENDPROC +END(page_fault) + +/* + * Debug traps and NMI can happen at the one SYSENTER instruction + * that sets up the real kernel stack. Check here, since we can't + * allow the wrong stack to be used. + * + * "TSS_sysenter_sp0+12" is because the NMI/debug handler will have + * already pushed 3 words if it hits on the sysenter instruction: + * eflags, cs and eip. + * + * We just load the right stack, and push the three (known) values + * by hand onto the new stack - while updating the return eip past + * the instruction that would have done it for sysenter. + */ +#define FIX_STACK(offset, ok, label) \ + cmpw $__KERNEL_CS,4(%esp); \ + jne ok; \ +label: \ + movl TSS_sysenter_sp0+offset(%esp),%esp; \ + CFI_DEF_CFA esp, 0; \ + CFI_UNDEFINED eip; \ + pushfl; \ + CFI_ADJUST_CFA_OFFSET 4; \ + pushl $__KERNEL_CS; \ + CFI_ADJUST_CFA_OFFSET 4; \ + pushl $sysenter_past_esp; \ + CFI_ADJUST_CFA_OFFSET 4; \ + CFI_REL_OFFSET eip, 0 + +ENTRY(debug) + RING0_INT_FRAME + cmpl $ia32_sysenter_target,(%esp) + jne debug_stack_correct + FIX_STACK(12, debug_stack_correct, debug_esp_fix_insn) +debug_stack_correct: + pushl $-1 # mark this as an int + CFI_ADJUST_CFA_OFFSET 4 + SAVE_ALL + TRACE_IRQS_OFF + xorl %edx,%edx # error code 0 + movl %esp,%eax # pt_regs pointer + call do_debug + jmp ret_from_exception + CFI_ENDPROC +END(debug) + +/* + * NMI is doubly nasty. It can happen _while_ we're handling + * a debug fault, and the debug fault hasn't yet been able to + * clear up the stack. So we first check whether we got an + * NMI on the sysenter entry path, but after that we need to + * check whether we got an NMI on the debug path where the debug + * fault happened on the sysenter path. + */ +ENTRY(nmi) + RING0_INT_FRAME + pushl %eax + CFI_ADJUST_CFA_OFFSET 4 + movl %ss, %eax + cmpw $__ESPFIX_SS, %ax + popl %eax + CFI_ADJUST_CFA_OFFSET -4 + je nmi_espfix_stack + cmpl $ia32_sysenter_target,(%esp) + je nmi_stack_fixup + pushl %eax + CFI_ADJUST_CFA_OFFSET 4 + movl %esp,%eax + /* Do not access memory above the end of our stack page, + * it might not exist. + */ + andl $(THREAD_SIZE-1),%eax + cmpl $(THREAD_SIZE-20),%eax + popl %eax + CFI_ADJUST_CFA_OFFSET -4 + jae nmi_stack_correct + cmpl $ia32_sysenter_target,12(%esp) + je nmi_debug_stack_check +nmi_stack_correct: + /* We have a RING0_INT_FRAME here */ + pushl %eax + CFI_ADJUST_CFA_OFFSET 4 + SAVE_ALL + TRACE_IRQS_OFF + xorl %edx,%edx # zero error code + movl %esp,%eax # pt_regs pointer + call do_nmi + jmp restore_nocheck_notrace + CFI_ENDPROC + +nmi_stack_fixup: + RING0_INT_FRAME + FIX_STACK(12,nmi_stack_correct, 1) + jmp nmi_stack_correct + +nmi_debug_stack_check: + /* We have a RING0_INT_FRAME here */ + cmpw $__KERNEL_CS,16(%esp) + jne nmi_stack_correct + cmpl $debug,(%esp) + jb nmi_stack_correct + cmpl $debug_esp_fix_insn,(%esp) + ja nmi_stack_correct + FIX_STACK(24,nmi_stack_correct, 1) + jmp nmi_stack_correct + +nmi_espfix_stack: + /* We have a RING0_INT_FRAME here. + * + * create the pointer to lss back + */ + pushl %ss + CFI_ADJUST_CFA_OFFSET 4 + pushl %esp + CFI_ADJUST_CFA_OFFSET 4 + addw $4, (%esp) + /* copy the iret frame of 12 bytes */ + .rept 3 + pushl 16(%esp) + CFI_ADJUST_CFA_OFFSET 4 + .endr + pushl %eax + CFI_ADJUST_CFA_OFFSET 4 + SAVE_ALL + TRACE_IRQS_OFF + FIXUP_ESPFIX_STACK # %eax == %esp + xorl %edx,%edx # zero error code + call do_nmi + RESTORE_REGS + lss 12+4(%esp), %esp # back to espfix stack + CFI_ADJUST_CFA_OFFSET -24 + jmp irq_return + CFI_ENDPROC +END(nmi) + +ENTRY(int3) + RING0_INT_FRAME + pushl $-1 # mark this as an int + CFI_ADJUST_CFA_OFFSET 4 + SAVE_ALL + TRACE_IRQS_OFF + xorl %edx,%edx # zero error code + movl %esp,%eax # pt_regs pointer + call do_int3 + jmp ret_from_exception + CFI_ENDPROC +END(int3) + +ENTRY(general_protection) + RING0_EC_FRAME + pushl $do_general_protection + CFI_ADJUST_CFA_OFFSET 4 + jmp error_code + CFI_ENDPROC +END(general_protection) + +/* + * End of kprobes section + */ + .popsection -- cgit v1.2.3 From 33454539f386a2beb38269bea5fff82b3d56b0e9 Mon Sep 17 00:00:00 2001 From: "gorcunov@gmail.com" Date: Wed, 26 Nov 2008 22:17:02 +0300 Subject: x86: entry_64.S - use X86_EFLAGS_IF instead of hardcoded number Impact: cleanup Signed-off-by: Cyrill Gorcunov Acked-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 38fcd0517c31..1c309d546518 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -194,7 +194,7 @@ ENTRY(native_usergs_sysret64) pushq %rax /* rsp */ CFI_ADJUST_CFA_OFFSET 8 CFI_REL_OFFSET rsp,0 - pushq $(1<<9) /* eflags - interrupts on */ + pushq $X86_EFLAGS_IF /* eflags - interrupts on */ CFI_ADJUST_CFA_OFFSET 8 /*CFI_REL_OFFSET rflags,0*/ pushq $__KERNEL_CS /* cs */ -- cgit v1.2.3 From c2c631e318091118587f3b766347d259c9265b8b Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Wed, 26 Nov 2008 22:17:00 +0300 Subject: x86: entry_64.S - use ENTRY to define child_rip child_rip is called not by its name but indirectly rather so make it global and aligned. Signed-off-by: Cyrill Gorcunov Acked-by: Alexander van Heukelum Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 1c309d546518..0a910a7f85f5 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1160,7 +1160,7 @@ ENTRY(kernel_thread) CFI_ENDPROC END(kernel_thread) -child_rip: +ENTRY(child_rip) pushq $0 # fake return address CFI_STARTPROC /* -- cgit v1.2.3 From 1d9b16d1690fe5edb1c907fe4746681cf026cdf3 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 27 Nov 2008 18:39:15 +0100 Subject: x86: move GART specific stuff from iommu.h to gart.h Impact: cleanup Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu.c | 1 + arch/x86/kernel/amd_iommu_init.c | 1 + arch/x86/kernel/early-quirks.c | 1 + arch/x86/kernel/pci-dma.c | 1 + arch/x86/kernel/setup.c | 1 + 5 files changed, 5 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 331b318304eb..172e0dc4641e 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 0cdcda35a05f..7685f0774a8f 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -28,6 +28,7 @@ #include #include #include +#include /* * definitions for the ACPI scanning code diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index 1b894b72c0f5..744aa7fc49d5 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -17,6 +17,7 @@ #include #include #include +#include static void __init fix_hypertransport_config(int num, int slot, int func) { diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 192624820217..12eeb4bfcdeb 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -6,6 +6,7 @@ #include #include #include +#include #include #include diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 0fa6790c1dd3..67d5979e654e 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -93,6 +93,7 @@ #include #include #include +#include #include #include -- cgit v1.2.3 From 5ae3a139cf4fc2349f1dfa1993a66c1dcc119468 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 27 Nov 2008 00:02:10 +0300 Subject: x86: uv bau interrupt -- use proper interrupt number Signed-off-by: Cyrill Gorcunov Acked-by: Cliff Wickman Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 0a910a7f85f5..57d7f7a5ad2f 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -938,7 +938,7 @@ apicinterrupt IRQ_MOVE_CLEANUP_VECTOR \ irq_move_cleanup_interrupt smp_irq_move_cleanup_interrupt #endif -apicinterrupt 220 \ +apicinterrupt UV_BAU_MESSAGE \ uv_bau_message_intr1 uv_bau_message_interrupt apicinterrupt LOCAL_TIMER_VECTOR \ apic_timer_interrupt smp_apic_timer_interrupt -- cgit v1.2.3 From 9f1e87ea3ecb3c46c21f6a1a202ec82f99ed2473 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Thu, 27 Nov 2008 21:10:08 +0300 Subject: x86: entry_64.S - trivial: space, comments fixup Impact: cleanup Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 94 +++++++++++++++++++++++----------------------- 1 file changed, 48 insertions(+), 46 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 57d7f7a5ad2f..08c0c9777a09 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -1020,7 +1020,7 @@ END(\sym) .macro paranoidzeroentry_ist sym do_sym ist ENTRY(\sym) - INTR_FRAME + INTR_FRAME PARAVIRT_ADJUST_EXCEPTION_FRAME pushq $-1 /* ORIG_RAX: no syscall to restart */ CFI_ADJUST_CFA_OFFSET 8 @@ -1088,36 +1088,36 @@ zeroentry coprocessor_error do_coprocessor_error errorentry alignment_check do_alignment_check zeroentry simd_coprocessor_error do_simd_coprocessor_error - /* Reload gs selector with exception handling */ - /* edi: new selector */ + /* Reload gs selector with exception handling */ + /* edi: new selector */ ENTRY(native_load_gs_index) CFI_STARTPROC pushf CFI_ADJUST_CFA_OFFSET 8 DISABLE_INTERRUPTS(CLBR_ANY | ~(CLBR_RDI)) - SWAPGS + SWAPGS gs_change: - movl %edi,%gs + movl %edi,%gs 2: mfence /* workaround */ SWAPGS - popf + popf CFI_ADJUST_CFA_OFFSET -8 - ret + ret CFI_ENDPROC END(native_load_gs_index) - .section __ex_table,"a" - .align 8 - .quad gs_change,bad_gs - .previous - .section .fixup,"ax" + .section __ex_table,"a" + .align 8 + .quad gs_change,bad_gs + .previous + .section .fixup,"ax" /* running with kernelgs */ bad_gs: SWAPGS /* switch back to user gs */ xorl %eax,%eax - movl %eax,%gs - jmp 2b - .previous + movl %eax,%gs + jmp 2b + .previous /* * Create a kernel thread. @@ -1152,7 +1152,7 @@ ENTRY(kernel_thread) * so internally to the x86_64 port you can rely on kernel_thread() * not to reschedule the child before returning, this avoids the need * of hacks for example to fork off the per-CPU idle tasks. - * [Hopefully no generic code relies on the reschedule -AK] + * [Hopefully no generic code relies on the reschedule -AK] */ RESTORE_ALL UNFAKE_STACK_FRAME @@ -1231,22 +1231,24 @@ END(call_softirq) zeroentry xen_hypervisor_callback xen_do_hypervisor_callback /* -# A note on the "critical region" in our callback handler. -# We want to avoid stacking callback handlers due to events occurring -# during handling of the last event. To do this, we keep events disabled -# until we've done all processing. HOWEVER, we must enable events before -# popping the stack frame (can't be done atomically) and so it would still -# be possible to get enough handler activations to overflow the stack. -# Although unlikely, bugs of that kind are hard to track down, so we'd -# like to avoid the possibility. -# So, on entry to the handler we detect whether we interrupted an -# existing activation in its critical region -- if so, we pop the current -# activation and restart the handler using the previous one. -*/ + * A note on the "critical region" in our callback handler. + * We want to avoid stacking callback handlers due to events occurring + * during handling of the last event. To do this, we keep events disabled + * until we've done all processing. HOWEVER, we must enable events before + * popping the stack frame (can't be done atomically) and so it would still + * be possible to get enough handler activations to overflow the stack. + * Although unlikely, bugs of that kind are hard to track down, so we'd + * like to avoid the possibility. + * So, on entry to the handler we detect whether we interrupted an + * existing activation in its critical region -- if so, we pop the current + * activation and restart the handler using the previous one. + */ ENTRY(xen_do_hypervisor_callback) # do_hypervisor_callback(struct *pt_regs) CFI_STARTPROC -/* Since we don't modify %rdi, evtchn_do_upall(struct *pt_regs) will - see the correct pointer to the pt_regs */ +/* + * Since we don't modify %rdi, evtchn_do_upall(struct *pt_regs) will + * see the correct pointer to the pt_regs + */ movq %rdi, %rsp # we don't return, adjust the stack frame CFI_ENDPROC DEFAULT_FRAME @@ -1264,18 +1266,18 @@ ENTRY(xen_do_hypervisor_callback) # do_hypervisor_callback(struct *pt_regs) END(do_hypervisor_callback) /* -# Hypervisor uses this for application faults while it executes. -# We get here for two reasons: -# 1. Fault while reloading DS, ES, FS or GS -# 2. Fault while executing IRET -# Category 1 we do not need to fix up as Xen has already reloaded all segment -# registers that could be reloaded and zeroed the others. -# Category 2 we fix up by killing the current process. We cannot use the -# normal Linux return path in this case because if we use the IRET hypercall -# to pop the stack frame we end up in an infinite loop of failsafe callbacks. -# We distinguish between categories by comparing each saved segment register -# with its current contents: any discrepancy means we in category 1. -*/ + * Hypervisor uses this for application faults while it executes. + * We get here for two reasons: + * 1. Fault while reloading DS, ES, FS or GS + * 2. Fault while executing IRET + * Category 1 we do not need to fix up as Xen has already reloaded all segment + * registers that could be reloaded and zeroed the others. + * Category 2 we fix up by killing the current process. We cannot use the + * normal Linux return path in this case because if we use the IRET hypercall + * to pop the stack frame we end up in an infinite loop of failsafe callbacks. + * We distinguish between categories by comparing each saved segment register + * with its current contents: any discrepancy means we in category 1. + */ ENTRY(xen_failsafe_callback) INTR_FRAME 1 (6*8) /*CFI_REL_OFFSET gs,GS*/ @@ -1339,8 +1341,8 @@ paranoidzeroentry machine_check do_machine_check #endif /* - * "Paranoid" exit path from exception stack. - * Paranoid because this is used by NMIs and cannot take + * "Paranoid" exit path from exception stack. + * Paranoid because this is used by NMIs and cannot take * any kernel state for granted. * We don't do kernel preemption checks here, because only * NMI should be common and it does not enable IRQs and @@ -1445,7 +1447,7 @@ error_kernelspace: cmpq %rcx,RIP+8(%rsp) je error_swapgs cmpq $gs_change,RIP+8(%rsp) - je error_swapgs + je error_swapgs jmp error_sti END(error_entry) @@ -1521,7 +1523,7 @@ nmi_schedule: CFI_ENDPROC #else jmp paranoid_exit - CFI_ENDPROC + CFI_ENDPROC #endif END(nmi) -- cgit v1.2.3 From 5b3eec0c80038c8739ccd465b897a35c0dff1cc4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 27 Nov 2008 14:41:21 +0100 Subject: x86: ret_from_fork - get rid of jump back Impact: remove dead code If we take a closer look at the rff_trace/rff_action ret_from_fork code, we have to realize that it does all the wrong things: for example it checks the TIF flag - while later on jumping back to the ret-from-syscall path - duplicating the check needlessly. But checking for _TIF_SYSCALL_TRACE is completely unnecessary here because we clear that flag for every freshly forked task. So the whole "tracing" code here, for which there is a out of line jump optimization that makes it even harder to read, is in reality completely dead code ... Reported-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar Tested-by: Cyrill Gorcunov --- arch/x86/kernel/entry_64.S | 31 ++++++++++++++++--------------- 1 file changed, 16 insertions(+), 15 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index e41734a537bd..3194636a4293 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -361,34 +361,35 @@ ENTRY(save_paranoid) END(save_paranoid) /* - * A newly forked process directly context switches into this. + * A newly forked process directly context switches into this address. + * + * rdi: prev task we switched from */ -/* rdi: prev */ ENTRY(ret_from_fork) DEFAULT_FRAME + push kernel_eflags(%rip) CFI_ADJUST_CFA_OFFSET 8 - popf # reset kernel eflags + popf # reset kernel eflags CFI_ADJUST_CFA_OFFSET -8 - call schedule_tail + + call schedule_tail # rdi: 'prev' task parameter + GET_THREAD_INFO(%rcx) - testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),TI_flags(%rcx) + CFI_REMEMBER_STATE - jnz rff_trace -rff_action: RESTORE_REST - testl $3,CS-ARGOFFSET(%rsp) # from kernel_thread? + + testl $3, CS-ARGOFFSET(%rsp) # from kernel_thread? je int_ret_from_sys_call - testl $_TIF_IA32,TI_flags(%rcx) + + testl $_TIF_IA32, TI_flags(%rcx) # 32-bit compat task needs IRET jnz int_ret_from_sys_call + RESTORE_TOP_OF_STACK %rdi, -ARGOFFSET - jmp ret_from_sys_call + jmp ret_from_sys_call # go to the SYSRET fastpath + CFI_RESTORE_STATE -rff_trace: - movq %rsp,%rdi - call syscall_trace_leave - GET_THREAD_INFO(%rcx) - jmp rff_action CFI_ENDPROC END(ret_from_fork) -- cgit v1.2.3 From 8caac56305cef98f9357b060a77939d17699937d Mon Sep 17 00:00:00 2001 From: Pavel Machek Date: Wed, 26 Nov 2008 17:15:27 +0100 Subject: aperture_64.c: clarify that too small aperture is valid reason for this code Impact: update comment Clarify that too small aperture is valid reason for this code. Signed-off-by: Pavel Machek Signed-off-by: Ingo Molnar --- arch/x86/kernel/aperture_64.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/aperture_64.c b/arch/x86/kernel/aperture_64.c index 9a32b37ee2ee..676debfc1702 100644 --- a/arch/x86/kernel/aperture_64.c +++ b/arch/x86/kernel/aperture_64.c @@ -1,8 +1,9 @@ /* * Firmware replacement code. * - * Work around broken BIOSes that don't set an aperture or only set the - * aperture in the AGP bridge. + * Work around broken BIOSes that don't set an aperture, only set the + * aperture in the AGP bridge, or set too small aperture. + * * If all fails map the aperture over some low memory. This is cheaper than * doing bounce buffering. The memory is lost. This is done at early boot * because only the bootmem allocator can allocate 32+MB. -- cgit v1.2.3 From 4385cecf1f5866fb33fc95e2ee26a44e9b6f6be2 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Sat, 29 Nov 2008 22:33:16 +0100 Subject: x86: intel_cacheinfo, minor show_type cleanup Impact: cleanup Signed-off-by: Jiri Slaby Cc: Jiri Slaby Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/intel_cacheinfo.c | 17 +++++++---------- 1 file changed, 7 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index 3f46afbb1cf1..68b5d8681cbb 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -644,20 +644,17 @@ static inline ssize_t show_shared_cpu_list(struct _cpuid4_info *leaf, char *buf) return show_shared_cpu_map_func(leaf, 1, buf); } -static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf) { - switch(this_leaf->eax.split.type) { - case CACHE_TYPE_DATA: +static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf) +{ + switch (this_leaf->eax.split.type) { + case CACHE_TYPE_DATA: return sprintf(buf, "Data\n"); - break; - case CACHE_TYPE_INST: + case CACHE_TYPE_INST: return sprintf(buf, "Instruction\n"); - break; - case CACHE_TYPE_UNIFIED: + case CACHE_TYPE_UNIFIED: return sprintf(buf, "Unified\n"); - break; - default: + default: return sprintf(buf, "Unknown\n"); - break; } } -- cgit v1.2.3 From 2236d252e001ea57d53cec1954f680e503f3b8bc Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 22 Nov 2008 17:37:34 +0000 Subject: enable_IR_x2apic() needs to be __init calls __init, called only from __init Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- arch/x86/kernel/apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 04a7f960bbc0..16f94879b525 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1315,7 +1315,7 @@ void enable_x2apic(void) } } -void enable_IR_x2apic(void) +void __init enable_IR_x2apic(void) { #ifdef CONFIG_INTR_REMAP int ret; -- cgit v1.2.3 From 23a14b9e9db49ed5f7683857557c26c874d4abb6 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 22 Nov 2008 17:37:44 +0000 Subject: kvm_setup_secondary_clock() is cpuinit Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- arch/x86/kernel/kvmclock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c index 1c9cc431ea4f..e169ae9b6a62 100644 --- a/arch/x86/kernel/kvmclock.c +++ b/arch/x86/kernel/kvmclock.c @@ -128,7 +128,7 @@ static int kvm_register_clock(char *txt) } #ifdef CONFIG_X86_LOCAL_APIC -static void __devinit kvm_setup_secondary_clock(void) +static void __cpuinit kvm_setup_secondary_clock(void) { /* * Now that the first cpu already had this clocksource initialized, -- cgit v1.2.3 From 48d68b20d00865035b8b65e69af343d0f53fac9d Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 2 Dec 2008 00:20:39 +0100 Subject: tracing/function-graph-tracer: support for x86-64 Impact: extend and enable the function graph tracer to 64-bit x86 This patch implements the support for function graph tracer under x86-64. Both static and dynamic tracing are supported. This causes some small CPP conditional asm on arch/x86/kernel/ftrace.c I wanted to use probe_kernel_read/write to make the return address saving/patching code more generic but it causes tracing recursion. That would be perhaps useful to implement a notrace version of these function for other archs ports. Note that arch/x86/process_64.c is not traced, as in X86-32. I first thought __switch_to() was responsible of crashes during tracing because I believed current task were changed inside but that's actually not the case (actually yes, but not the "current" pointer). So I will have to investigate to find the functions that harm here, to enable tracing of the other functions inside (but there is no issue at this time, while process_64.c stays out of -pg flags). A little possible race condition is fixed inside this patch too. When the tracer allocate a return stack dynamically, the current depth is not initialized before but after. An interrupt could occur at this time and, after seeing that the return stack is allocated, the tracer could try to trace it with a random uninitialized depth. It's a prevention, even if I hadn't problems with it. Signed-off-by: Frederic Weisbecker Cc: Steven Rostedt Cc: Tim Bird Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 1 + arch/x86/kernel/entry_64.S | 74 ++++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kernel/ftrace.c | 11 ++++++- 3 files changed, 85 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 64939a0c3986..d274425fb076 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -17,6 +17,7 @@ endif ifdef CONFIG_FUNCTION_GRAPH_TRACER # Don't trace __switch_to() but let it for function tracer CFLAGS_REMOVE_process_32.o = -pg +CFLAGS_REMOVE_process_64.o = -pg endif # diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 08aa6b10933c..2aa0526ac30e 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -98,6 +98,12 @@ ftrace_call: movq (%rsp), %rax addq $0x38, %rsp +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +.globl ftrace_graph_call +ftrace_graph_call: + jmp ftrace_stub +#endif + .globl ftrace_stub ftrace_stub: retq @@ -110,6 +116,12 @@ ENTRY(mcount) cmpq $ftrace_stub, ftrace_trace_function jnz trace + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + cmpq $ftrace_stub, ftrace_graph_return + jnz ftrace_graph_caller +#endif + .globl ftrace_stub ftrace_stub: retq @@ -145,6 +157,68 @@ END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +ENTRY(ftrace_graph_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + + subq $0x38, %rsp + movq %rax, (%rsp) + movq %rcx, 8(%rsp) + movq %rdx, 16(%rsp) + movq %rsi, 24(%rsp) + movq %rdi, 32(%rsp) + movq %r8, 40(%rsp) + movq %r9, 48(%rsp) + + leaq 8(%rbp), %rdi + movq 0x38(%rsp), %rsi + + call prepare_ftrace_return + + movq 48(%rsp), %r9 + movq 40(%rsp), %r8 + movq 32(%rsp), %rdi + movq 24(%rsp), %rsi + movq 16(%rsp), %rdx + movq 8(%rsp), %rcx + movq (%rsp), %rax + addq $0x38, %rsp + retq +END(ftrace_graph_caller) + + +.globl return_to_handler +return_to_handler: + subq $80, %rsp + + movq %rax, (%rsp) + movq %rcx, 8(%rsp) + movq %rdx, 16(%rsp) + movq %rsi, 24(%rsp) + movq %rdi, 32(%rsp) + movq %r8, 40(%rsp) + movq %r9, 48(%rsp) + movq %r10, 56(%rsp) + movq %r11, 64(%rsp) + + call ftrace_return_to_handler + + movq %rax, 72(%rsp) + movq 64(%rsp), %r11 + movq 56(%rsp), %r10 + movq 48(%rsp), %r9 + movq 40(%rsp), %r8 + movq 32(%rsp), %rdi + movq 24(%rsp), %rsi + movq 16(%rsp), %rdx + movq 8(%rsp), %rcx + movq (%rsp), %rax + addq $72, %rsp + retq +#endif + + #ifndef CONFIG_PREEMPT #define retint_kernel retint_restore_args #endif diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 7ef914e6a2f6..58832478b94e 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -467,8 +467,13 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) * ignore such a protection. */ asm volatile( +#ifdef CONFIG_X86_64 + "1: movq (%[parent_old]), %[old]\n" + "2: movq %[return_hooker], (%[parent_replaced])\n" +#else "1: movl (%[parent_old]), %[old]\n" "2: movl %[return_hooker], (%[parent_replaced])\n" +#endif " movl $0, %[faulted]\n" ".section .fixup, \"ax\"\n" @@ -476,8 +481,13 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) ".previous\n" ".section __ex_table, \"a\"\n" +#ifdef CONFIG_X86_64 + " .quad 1b, 3b\n" + " .quad 2b, 3b\n" +#else " .long 1b, 3b\n" " .long 2b, 3b\n" +#endif ".previous\n" : [parent_replaced] "=r" (parent), [old] "=r" (old), @@ -509,5 +519,4 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) ftrace_graph_entry(&trace); } - #endif /* CONFIG_FUNCTION_GRAPH_TRACER */ -- cgit v1.2.3 From 8daa19051e1c7369c89ace7b18e74fe1f55dfa29 Mon Sep 17 00:00:00 2001 From: Niels de Vos Date: Mon, 1 Dec 2008 14:13:53 -0800 Subject: x86, apm: remove CONFIG_APM_REAL_MODE_POWER_OFF in favor of a kernel parameter Remove CONFIG_APM_REAL_MODE_POWER_OFF like CONFIG_APM_POWER_OFF which has been done for linux-2.2.14pre8 (http://lkml.org/lkml/1999/11/23/3). Re-introducing CONFIG_APM_POWER_OFF got nack-ed. Stephen didn't bother to remove CONFIG_APM_REAL_MODE_POWER_OFF, let's get rid of it now. Reference: http://lkml.org/lkml/2008/5/7/97 Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/kernel/apm_32.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c index 5145a6e72bbb..3a26525a3f31 100644 --- a/arch/x86/kernel/apm_32.c +++ b/arch/x86/kernel/apm_32.c @@ -391,11 +391,7 @@ static int power_off; #else static int power_off = 1; #endif -#ifdef CONFIG_APM_REAL_MODE_POWER_OFF -static int realmode_power_off = 1; -#else static int realmode_power_off; -#endif #ifdef CONFIG_APM_ALLOW_INTS static int allow_ints = 1; #else -- cgit v1.2.3 From dcb7731a185efbf3d800618d874af99895df5afb Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Tue, 2 Dec 2008 20:16:03 +0100 Subject: x86: fix broken flushing in GART nofullflush path Impact: remove stale IOTLB entries In the non-default nofullflush case the GART is only flushed when next_bit wraps around. But it can happen that an unmap operation unmaps memory which is behind the current next_bit location. If these addresses are reused it may result in stale GART IO/TLB entries. Fix this by setting the GART next_bit always behind an unmapped location. Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-gart_64.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index a42b02b4df68..ba7ad83e20a8 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -123,6 +123,8 @@ static void free_iommu(unsigned long offset, int size) spin_lock_irqsave(&iommu_bitmap_lock, flags); iommu_area_free(iommu_gart_bitmap, offset, size); + if (offset >= next_bit) + next_bit = offset + size; spin_unlock_irqrestore(&iommu_bitmap_lock, flags); } -- cgit v1.2.3 From 181de82ee3ffda1175f89d50c991dae31b79280c Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Wed, 3 Dec 2008 14:53:04 +0900 Subject: x86: remove dead BIO_VMERGE_BOUNDARY definition Impact: cleanup, remove dead code The block layer dropped the virtual merge feature (b8b3e16cfe6435d961f6aaebcfd52a1ff2a988c5). BIO_VMERGE_BOUNDARY definition is meaningless now. Signed-off-by: FUJITA Tomonori Acked-by: Jens Axboe Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-dma.c | 6 ------ 1 file changed, 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 12eeb4bfcdeb..da93c65f8f0b 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -31,11 +31,6 @@ int no_iommu __read_mostly; /* Set this to 1 if there is a HW IOMMU in the system */ int iommu_detected __read_mostly = 0; -/* This tells the BIO block layer to assume merging. Default to off - because we cannot guarantee merging later. */ -int iommu_bio_merge __read_mostly = 0; -EXPORT_SYMBOL(iommu_bio_merge); - dma_addr_t bad_dma_address __read_mostly = 0; EXPORT_SYMBOL(bad_dma_address); @@ -189,7 +184,6 @@ static __init int iommu_setup(char *p) } if (!strncmp(p, "biomerge", 8)) { - iommu_bio_merge = 4096; iommu_merge = 1; force_iommu = 1; } -- cgit v1.2.3 From 347fdd9dd4e5d3f3a4e415925c35bdff1d59c3a9 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 15:34:08 -0500 Subject: ftrace: clean up function graph asm Impact: clean up There exists macros for x86 asm to handle x86_64 and i386. This patch updates function graph asm to use them. Signed-off-by: Steven Rostedt Acked-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 20 ++++---------------- 1 file changed, 4 insertions(+), 16 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 58832478b94e..1a5b8f8cb3cc 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -467,28 +467,16 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) * ignore such a protection. */ asm volatile( -#ifdef CONFIG_X86_64 - "1: movq (%[parent_old]), %[old]\n" - "2: movq %[return_hooker], (%[parent_replaced])\n" -#else - "1: movl (%[parent_old]), %[old]\n" - "2: movl %[return_hooker], (%[parent_replaced])\n" -#endif + "1: " _ASM_MOV " (%[parent_old]), %[old]\n" + "2: " _ASM_MOV " %[return_hooker], (%[parent_replaced])\n" " movl $0, %[faulted]\n" ".section .fixup, \"ax\"\n" "3: movl $1, %[faulted]\n" ".previous\n" - ".section __ex_table, \"a\"\n" -#ifdef CONFIG_X86_64 - " .quad 1b, 3b\n" - " .quad 2b, 3b\n" -#else - " .long 1b, 3b\n" - " .long 2b, 3b\n" -#endif - ".previous\n" + _ASM_EXTABLE(1b, 3b) + _ASM_EXTABLE(2b, 3b) : [parent_replaced] "=r" (parent), [old] "=r" (old), [faulted] "=r" (faulted) -- cgit v1.2.3 From bb4304c71c97bf727ec43cd2f195c2c237c27fd3 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 15:34:09 -0500 Subject: ftrace: have function graph use mcount caller address Impact: consistency change for function graph This patch makes function graph record the mcount caller address the same way the function tracer does. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 1 + arch/x86/kernel/entry_64.S | 1 + 2 files changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 958af86186c4..826682abed1d 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1230,6 +1230,7 @@ ENTRY(ftrace_graph_caller) pushl %edx movl 0xc(%esp), %edx lea 0x4(%ebp), %eax + subl $MCOUNT_INSN_SIZE, %edx call prepare_ftrace_return popl %edx popl %ecx diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 2aa0526ac30e..9060ba6497e2 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -173,6 +173,7 @@ ENTRY(ftrace_graph_caller) leaq 8(%rbp), %rdi movq 0x38(%rsp), %rsi + subq $MCOUNT_INSN_SIZE, %rsi call prepare_ftrace_return -- cgit v1.2.3 From 14a866c567e040ccf6240d68b083dd1dbbde63e6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 23:50:02 -0500 Subject: ftrace: add ftrace_graph_stop() Impact: new ftrace_graph_stop function While developing more features of function graph, I hit a bug that caused the WARN_ON to trigger in the prepare_ftrace_return function. Well, it was hard for me to find out that was happening because the bug would not print, it would just cause a hard lockup or reboot. The reason is that it is not safe to call printk from this function. Looking further, I also found that it calls unregister_ftrace_graph, which grabs a mutex and calls kstop machine. This would definitely lock the box up if it were to trigger. This patch adds a fast and safe ftrace_graph_stop() which will stop the function tracer. Then it is safe to call the WARN ON. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 1a5b8f8cb3cc..adba8e9a427c 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -484,14 +484,16 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) : "memory" ); - if (WARN_ON(faulted)) { - unregister_ftrace_graph(); + if (unlikely(faulted)) { + ftrace_graph_stop(); + WARN_ON(1); return; } - if (WARN_ON(!__kernel_text_address(old))) { - unregister_ftrace_graph(); + if (unlikely(!__kernel_text_address(old))) { + ftrace_graph_stop(); *parent = old; + WARN_ON(1); return; } -- cgit v1.2.3 From 7ee991fbc6f947e9b04f29c9c6c1d057d0671a16 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 23:50:04 -0500 Subject: ftrace: print real return in dumpstack for function graph Impact: better dumpstack output I noticed in my crash dumps and even in the stack tracer that a lot of functions listed in the stack trace are simply return_to_handler which is ftrace graphs way to insert its own call into the return of a function. But we lose out where the actually function was called from. This patch adds in hooks to the dumpstack mechanism that detects this and finds the real function to print. Both are printed to let the user know that a hook is still in place. This does give a funny side effect in the stack tracer output: Depth Size Location (80 entries) ----- ---- -------- 0) 4144 48 save_stack_trace+0x2f/0x4d 1) 4096 128 ftrace_call+0x5/0x2b 2) 3968 16 mempool_alloc_slab+0x16/0x18 3) 3952 384 return_to_handler+0x0/0x73 4) 3568 -240 stack_trace_call+0x11d/0x209 5) 3808 144 return_to_handler+0x0/0x73 6) 3664 -128 mempool_alloc+0x4d/0xfe 7) 3792 128 return_to_handler+0x0/0x73 8) 3664 -32 scsi_sg_alloc+0x48/0x4a [scsi_mod] As you can see, the real functions are now negative. This is due to them not being found inside the stack. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/dumpstack.c | 34 +++++++++++++++++++++++++++++++++- arch/x86/kernel/dumpstack.h | 2 +- arch/x86/kernel/dumpstack_32.c | 5 ++++- arch/x86/kernel/dumpstack_64.c | 7 ++++--- 4 files changed, 42 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c index 5962176dfabb..6b1f6f6f8661 100644 --- a/arch/x86/kernel/dumpstack.c +++ b/arch/x86/kernel/dumpstack.c @@ -30,6 +30,37 @@ void printk_address(unsigned long address, int reliable) reliable ? "" : "? ", (void *) address); } +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +static void +print_ftrace_graph_addr(unsigned long addr, void *data, + const struct stacktrace_ops *ops, + struct thread_info *tinfo, int *graph) +{ + struct task_struct *task = tinfo->task; + unsigned long ret_addr; + int index = task->curr_ret_stack; + + if (addr != (unsigned long)return_to_handler) + return; + + if (!task->ret_stack || index < *graph) + return; + + index -= *graph; + ret_addr = task->ret_stack[index].ret; + + ops->address(data, ret_addr, 1); + + (*graph)++; +} +#else +static inline void +print_ftrace_graph_addr(unsigned long addr, void *data, + const struct stacktrace_ops *ops, + struct thread_info *tinfo, int *graph) +{ } +#endif + /* * x86-64 can have up to three kernel stacks: * process stack @@ -54,7 +85,7 @@ unsigned long print_context_stack(struct thread_info *tinfo, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data, - unsigned long *end) + unsigned long *end, int *graph) { struct stack_frame *frame = (struct stack_frame *)bp; @@ -70,6 +101,7 @@ print_context_stack(struct thread_info *tinfo, } else { ops->address(data, addr, bp == 0); } + print_ftrace_graph_addr(addr, data, ops, tinfo, graph); } stack++; } diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h index 3119a801c32b..da87590b8698 100644 --- a/arch/x86/kernel/dumpstack.h +++ b/arch/x86/kernel/dumpstack.h @@ -18,7 +18,7 @@ extern unsigned long print_context_stack(struct thread_info *tinfo, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data, - unsigned long *end); + unsigned long *end, int *graph); extern void show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index 7b031b106ec8..d593cd1f58dc 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -23,6 +23,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data) { + int graph = 0; + if (!task) task = current; @@ -50,7 +52,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, context = (struct thread_info *) ((unsigned long)stack & (~(THREAD_SIZE - 1))); - bp = print_context_stack(context, stack, bp, ops, data, NULL); + bp = print_context_stack(context, stack, bp, ops, + data, NULL, &graph); stack = (unsigned long *)context->previous_esp; if (!stack) diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 33ff10287a5d..c302d0707048 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -109,6 +109,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *irqstack_end = (unsigned long *)cpu_pda(cpu)->irqstackptr; unsigned used = 0; struct thread_info *tinfo; + int graph = 0; if (!task) task = current; @@ -149,7 +150,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, break; bp = print_context_stack(tinfo, stack, bp, ops, - data, estack_end); + data, estack_end, &graph); ops->stack(data, ""); /* * We link to the next stack via the @@ -168,7 +169,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, if (ops->stack(data, "IRQ") < 0) break; bp = print_context_stack(tinfo, stack, bp, - ops, data, irqstack_end); + ops, data, irqstack_end, &graph); /* * We link to the next stack (which would be * the process stack normally) the last @@ -186,7 +187,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, /* * This handles the process stack: */ - bp = print_context_stack(tinfo, stack, bp, ops, data, NULL); + bp = print_context_stack(tinfo, stack, bp, ops, data, NULL, &graph); put_cpu(); } EXPORT_SYMBOL(dump_trace); -- cgit v1.2.3 From e49dc19c6a19ea112fcb94b7c62ec62cdd5c08aa Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 23:50:05 -0500 Subject: ftrace: function graph return for function entry Impact: feature, let entry function decide to trace or not This patch lets the graph tracer entry function decide if the tracing should be done at the end as well. This requires all function graph entry functions return 1 if it should trace, or 0 if the return should not be traced. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 3 +++ arch/x86/kernel/entry_64.S | 3 +++ arch/x86/kernel/ftrace.c | 7 ++++++- 3 files changed, 12 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 826682abed1d..43ceb3f454bf 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1196,6 +1196,9 @@ ENTRY(mcount) #ifdef CONFIG_FUNCTION_GRAPH_TRACER cmpl $ftrace_stub, ftrace_graph_return jnz ftrace_graph_caller + + cmpl $ftrace_graph_entry_stub, ftrace_graph_entry + jnz ftrace_graph_caller #endif .globl ftrace_stub ftrace_stub: diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 9060ba6497e2..54e0bbdccb99 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -120,6 +120,9 @@ ENTRY(mcount) #ifdef CONFIG_FUNCTION_GRAPH_TRACER cmpq $ftrace_stub, ftrace_graph_return jnz ftrace_graph_caller + + cmpq $ftrace_graph_entry_stub, ftrace_graph_entry + jnz ftrace_graph_caller #endif .globl ftrace_stub diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index adba8e9a427c..d278ad2ebda2 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -425,6 +425,7 @@ static void pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret) trace->calltime = current->ret_stack[index].calltime; trace->overrun = atomic_read(¤t->trace_overrun); trace->depth = index; + barrier(); current->curr_ret_stack--; } @@ -506,7 +507,11 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) } trace.func = self_addr; - ftrace_graph_entry(&trace); + /* Only trace if the calling function expects to */ + if (!ftrace_graph_entry(&trace)) { + current->curr_ret_stack--; + *parent = old; + } } #endif /* CONFIG_FUNCTION_GRAPH_TRACER */ -- cgit v1.2.3 From 62679efe0a5f02987a621942afc5979a80a6ca5a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 2 Dec 2008 23:50:06 -0500 Subject: ftrace: add checks on ret stack in function graph Import: robustness checks Add more checks in the function graph code to detect errors and perhaps print out better information if a bug happens. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index d278ad2ebda2..f98c4076a170 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -420,6 +420,15 @@ static void pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret) int index; index = current->curr_ret_stack; + + if (unlikely(index < 0)) { + ftrace_graph_stop(); + WARN_ON(1); + /* Might as well panic, otherwise we have no where to go */ + *ret = (unsigned long)panic; + return; + } + *ret = current->ret_stack[index].ret; trace->func = current->ret_stack[index].func; trace->calltime = current->ret_stack[index].calltime; @@ -427,6 +436,7 @@ static void pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret) trace->depth = index; barrier(); current->curr_ret_stack--; + } /* @@ -442,6 +452,13 @@ unsigned long ftrace_return_to_handler(void) trace.rettime = cpu_clock(raw_smp_processor_id()); ftrace_graph_return(&trace); + if (unlikely(!ret)) { + ftrace_graph_stop(); + WARN_ON(1); + /* Might as well panic. What else to do? */ + ret = (unsigned long)panic; + } + return ret; } -- cgit v1.2.3 From 70d7d357578245f1993fd2d3ccd26088bcd38941 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Tue, 2 Dec 2008 20:16:03 +0100 Subject: x86: fix broken flushing in GART nofullflush path Impact: remove stale IOTLB entries In the non-default nofullflush case the GART is only flushed when next_bit wraps around. But it can happen that an unmap operation unmaps memory which is behind the current next_bit location. If these addresses are reused it may result in stale GART IO/TLB entries. Fix this by setting the GART next_bit always behind an unmapped location. Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-gart_64.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index a42b02b4df68..ba7ad83e20a8 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -123,6 +123,8 @@ static void free_iommu(unsigned long offset, int size) spin_lock_irqsave(&iommu_bitmap_lock, flags); iommu_area_free(iommu_gart_bitmap, offset, size); + if (offset >= next_bit) + next_bit = offset + size; spin_unlock_irqrestore(&iommu_bitmap_lock, flags); } -- cgit v1.2.3 From f91ba190648be4ff127d6aaf3993ac19d66dc2c2 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Tue, 25 Nov 2008 12:56:12 +0100 Subject: AMD IOMMU: set device table entry for aliased devices In some rare cases a request can arrive an IOMMU with its originial requestor id even it is aliased. Handle this by setting the device table entry to the same protection domain for the original and the aliased requestor id. Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index e4899e0e8787..a232e5a85d48 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -858,6 +858,9 @@ static int get_device_resources(struct device *dev, print_devid(_bdf, 1); } + if (domain_for_device(_bdf) == NULL) + set_device_domain(*iommu, *domain, _bdf); + return 1; } -- cgit v1.2.3 From 09ee17eb8ea89514c13980c4010bdbbaea8630c2 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 3 Dec 2008 12:19:27 +0100 Subject: AMD IOMMU: fix possible race while accessing iommu->need_sync The access to the iommu->need_sync member needs to be protected by the iommu->lock. Otherwise this is a possible race condition. Fix it with this patch. Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 33 +++++++++++++-------------------- 1 file changed, 13 insertions(+), 20 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index a232e5a85d48..5662e226b0c9 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -187,6 +187,8 @@ static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd) spin_lock_irqsave(&iommu->lock, flags); ret = __iommu_queue_command(iommu, cmd); + if (!ret) + iommu->need_sync = 1; spin_unlock_irqrestore(&iommu->lock, flags); return ret; @@ -210,10 +212,13 @@ static int iommu_completion_wait(struct amd_iommu *iommu) cmd.data[0] = CMD_COMPL_WAIT_INT_MASK; CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT); - iommu->need_sync = 0; - spin_lock_irqsave(&iommu->lock, flags); + if (!iommu->need_sync) + goto out; + + iommu->need_sync = 0; + ret = __iommu_queue_command(iommu, &cmd); if (ret) @@ -254,8 +259,6 @@ static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid) ret = iommu_queue_command(iommu, &cmd); - iommu->need_sync = 1; - return ret; } @@ -281,8 +284,6 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu, ret = iommu_queue_command(iommu, &cmd); - iommu->need_sync = 1; - return ret; } @@ -762,8 +763,6 @@ static void set_device_domain(struct amd_iommu *iommu, write_unlock_irqrestore(&amd_iommu_devtable_lock, flags); iommu_queue_inv_dev_entry(iommu, devid); - - iommu->need_sync = 1; } /***************************************************************************** @@ -1034,8 +1033,7 @@ static dma_addr_t map_single(struct device *dev, phys_addr_t paddr, if (addr == bad_dma_address) goto out; - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); out: spin_unlock_irqrestore(&domain->lock, flags); @@ -1063,8 +1061,7 @@ static void unmap_single(struct device *dev, dma_addr_t dma_addr, __unmap_single(iommu, domain->priv, dma_addr, size, dir); - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); spin_unlock_irqrestore(&domain->lock, flags); } @@ -1130,8 +1127,7 @@ static int map_sg(struct device *dev, struct scatterlist *sglist, goto unmap; } - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); out: spin_unlock_irqrestore(&domain->lock, flags); @@ -1176,8 +1172,7 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist, s->dma_address = s->dma_length = 0; } - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); spin_unlock_irqrestore(&domain->lock, flags); } @@ -1228,8 +1223,7 @@ static void *alloc_coherent(struct device *dev, size_t size, goto out; } - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); out: spin_unlock_irqrestore(&domain->lock, flags); @@ -1260,8 +1254,7 @@ static void free_coherent(struct device *dev, size_t size, __unmap_single(iommu, domain->priv, dma_addr, size, DMA_BIDIRECTIONAL); - if (unlikely(iommu->need_sync)) - iommu_completion_wait(iommu); + iommu_completion_wait(iommu); spin_unlock_irqrestore(&domain->lock, flags); -- cgit v1.2.3 From 9adc13867ec5fe0cd35434f92954d90e42381f0b Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Thu, 4 Dec 2008 13:33:35 +0100 Subject: x86: fix early panic with boot option "nosmp" Impact: fix boot crash with numcpus=0 on certain systems Fix early exception in __get_smp_config with nosmp. Bail out early when there is no MP table. Reported-by: Wu Fengguang Signed-off-by: Andi Kleen Tested-by: Wu Fengguang Signed-off-by: Ingo Molnar --- arch/x86/kernel/mpparse.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index f98f4e1dba09..0f4c1fd5a1f4 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -604,6 +604,9 @@ static void __init __get_smp_config(unsigned int early) printk(KERN_INFO "Using ACPI for processor (LAPIC) " "configuration information\n"); + if (!mpf) + return; + printk(KERN_INFO "Intel MultiProcessor Specification v1.%d\n", mpf->mpf_specification); #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86_32) -- cgit v1.2.3 From 55c395b47042e12d5c25aa07f271f56ffe44f793 Mon Sep 17 00:00:00 2001 From: Michael Tokarev Date: Fri, 5 Dec 2008 14:42:20 +0300 Subject: x86: fix missing space in printk Just come across this when booting on an old hw.. Looks somewhat ugly, that single missing space ;) Signed-off-by: Michael Tokarev Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 7b1093397319..1a3c3253f0ed 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1086,8 +1086,10 @@ static int __init smp_sanity_check(unsigned max_cpus) #endif if (!physid_isset(hard_smp_processor_id(), phys_cpu_present_map)) { - printk(KERN_WARNING "weird, boot CPU (#%d) not listed" - "by the BIOS.\n", hard_smp_processor_id()); + printk(KERN_WARNING + "weird, boot CPU (#%d) not listed by the BIOS.\n", + hard_smp_processor_id()); + physid_set(hard_smp_processor_id(), phys_cpu_present_map); } -- cgit v1.2.3 From a0286c94f07636380082608196d41dd725a83229 Mon Sep 17 00:00:00 2001 From: Michael Tokarev Date: Fri, 5 Dec 2008 15:47:29 +0300 Subject: x86: fix missing space in printk, #2 Impact: clean up printk Signed-off-by: Michael Tokarev Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-dma.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 192624820217..dc572994703d 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -300,8 +300,8 @@ fs_initcall(pci_iommu_init); static __devinit void via_no_dac(struct pci_dev *dev) { if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI && forbid_dac == 0) { - printk(KERN_INFO "PCI: VIA PCI bridge detected." - "Disabling DAC.\n"); + printk(KERN_INFO + "PCI: VIA PCI bridge detected. Disabling DAC.\n"); forbid_dac = 1; } } -- cgit v1.2.3 From 3e1e9002aa8b32bd4c95ac6c8fad376b7a8127fb Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 8 Dec 2008 00:50:22 +0100 Subject: x86: change static allocation of trampoline area Impact: fix trampoline sizing bug, save space While debugging a suspend-to-RAM related issue it occured to me that if the trampoline code had grown past 4 KB, we would have been allocating too little memory for it, since the 4 KB size of the trampoline is hardcoded into arch/x86/kernel/e820.c . Change that by making the kernel compute the trampoline size and allocate as much memory as necessary. Signed-off-by: Rafael J. Wysocki Signed-off-by: Ingo Molnar --- arch/x86/kernel/e820.c | 16 ---------------- arch/x86/kernel/head32.c | 3 +++ arch/x86/kernel/head64.c | 3 +++ arch/x86/kernel/trampoline.c | 19 +++++++++++++++++-- 4 files changed, 23 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 7aafeb5263ef..65a13943e098 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -677,22 +677,6 @@ struct early_res { }; static struct early_res early_res[MAX_EARLY_RES] __initdata = { { 0, PAGE_SIZE, "BIOS data page" }, /* BIOS data page */ -#if defined(CONFIG_X86_64) && defined(CONFIG_X86_TRAMPOLINE) - { TRAMPOLINE_BASE, TRAMPOLINE_BASE + 2 * PAGE_SIZE, "TRAMPOLINE" }, -#endif -#if defined(CONFIG_X86_32) && defined(CONFIG_SMP) - /* - * But first pinch a few for the stack/trampoline stuff - * FIXME: Don't need the extra page at 4K, but need to fix - * trampoline before removing it. (see the GDT stuff) - */ - { PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, "EX TRAMPOLINE" }, - /* - * Has to be in very low memory so we can execute - * real-mode AP code. - */ - { TRAMPOLINE_BASE, TRAMPOLINE_BASE + PAGE_SIZE, "TRAMPOLINE" }, -#endif {} }; diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c index fa1d25dd83e3..ac108d1fe182 100644 --- a/arch/x86/kernel/head32.c +++ b/arch/x86/kernel/head32.c @@ -12,9 +12,12 @@ #include #include #include +#include void __init i386_start_kernel(void) { + reserve_trampoline_memory(); + reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS"); #ifdef CONFIG_BLK_DEV_INITRD diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c index d16084f90649..388e05a5fc17 100644 --- a/arch/x86/kernel/head64.c +++ b/arch/x86/kernel/head64.c @@ -24,6 +24,7 @@ #include #include #include +#include /* boot cpu pda */ static struct x8664_pda _boot_cpu_pda __read_mostly; @@ -120,6 +121,8 @@ void __init x86_64_start_reservations(char *real_mode_data) { copy_bootdata(__va(real_mode_data)); + reserve_trampoline_memory(); + reserve_early(__pa_symbol(&_text), __pa_symbol(&_end), "TEXT DATA BSS"); #ifdef CONFIG_BLK_DEV_INITRD diff --git a/arch/x86/kernel/trampoline.c b/arch/x86/kernel/trampoline.c index 1106fac6024d..808031a5ba19 100644 --- a/arch/x86/kernel/trampoline.c +++ b/arch/x86/kernel/trampoline.c @@ -1,10 +1,26 @@ #include #include +#include /* ready for x86_64 and x86 */ unsigned char *trampoline_base = __va(TRAMPOLINE_BASE); +void __init reserve_trampoline_memory(void) +{ +#ifdef CONFIG_X86_32 + /* + * But first pinch a few for the stack/trampoline stuff + * FIXME: Don't need the extra page at 4K, but need to fix + * trampoline before removing it. (see the GDT stuff) + */ + reserve_early(PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, "EX TRAMPOLINE"); +#endif + /* Has to be in very low memory so we can execute real-mode AP code. */ + reserve_early(TRAMPOLINE_BASE, TRAMPOLINE_BASE + TRAMPOLINE_SIZE, + "TRAMPOLINE"); +} + /* * Currently trivial. Write the real->protected mode * bootstrap into the page concerned. The caller @@ -12,7 +28,6 @@ unsigned char *trampoline_base = __va(TRAMPOLINE_BASE); */ unsigned long setup_trampoline(void) { - memcpy(trampoline_base, trampoline_data, - trampoline_end - trampoline_data); + memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE); return virt_to_phys(trampoline_base); } -- cgit v1.2.3 From 0b8f1efad30bd58f89961b82dfe68b9edf8fd2ac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:31 -0800 Subject: sparse irq_desc[] array: core kernel and x86 changes Impact: new feature Problem on distro kernels: irq_desc[NR_IRQS] takes megabytes of RAM with NR_CPUS set to large values. The goal is to be able to scale up to much larger NR_IRQS value without impacting the (important) common case. To solve this, we generalize irq_desc[NR_IRQS] to an (optional) array of irq_desc pointers. When CONFIG_SPARSE_IRQ=y is used, we use kzalloc_node to get irq_desc, this also makes the IRQ descriptors NUMA-local (to the site that calls request_irq()). This gets rid of the irq_cfg[] static array on x86 as well: irq_cfg now uses desc->chip_data for x86 to store irq_cfg. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 301 +++++++++++++++++++++++++++---------------- arch/x86/kernel/irq.c | 3 + arch/x86/kernel/irq_32.c | 2 + arch/x86/kernel/irq_64.c | 2 + arch/x86/kernel/irqinit_32.c | 1 - arch/x86/kernel/irqinit_64.c | 1 - 6 files changed, 197 insertions(+), 113 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9043251210fb..9de17f5c1125 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -108,8 +108,33 @@ static int __init parse_noapic(char *str) early_param("noapic", parse_noapic); struct irq_pin_list; + +/* + * This is performance-critical, we want to do it O(1) + * + * the indexing order of this array favors 1:1 mappings + * between pins and IRQs. + */ + +struct irq_pin_list { + int apic, pin; + struct irq_pin_list *next; +}; + +static struct irq_pin_list *get_one_free_irq_2_pin(int cpu) +{ + struct irq_pin_list *pin; + int node; + + node = cpu_to_node(cpu); + + pin = kzalloc_node(sizeof(*pin), GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc irq_2_pin on cpu %d node %d\n", cpu, node); + + return pin; +} + struct irq_cfg { - unsigned int irq; struct irq_pin_list *irq_2_pin; cpumask_t domain; cpumask_t old_domain; @@ -119,83 +144,93 @@ struct irq_cfg { }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ +#ifdef CONFIG_SPARSE_IRQ +static struct irq_cfg irq_cfgx[] = { +#else static struct irq_cfg irq_cfgx[NR_IRQS] = { - [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, +#endif + [0] = { .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, + [1] = { .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, + [2] = { .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, + [3] = { .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, + [4] = { .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, + [5] = { .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, + [6] = { .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, + [7] = { .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, + [8] = { .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, + [9] = { .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, + [10] = { .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, + [11] = { .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, + [12] = { .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, + [13] = { .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, + [14] = { .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, + [15] = { .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, }; -#define for_each_irq_cfg(irq, cfg) \ - for (irq = 0, cfg = irq_cfgx; irq < nr_irqs; irq++, cfg++) - -static struct irq_cfg *irq_cfg(unsigned int irq) +void __init arch_early_irq_init(void) { - return irq < nr_irqs ? irq_cfgx + irq : NULL; -} + struct irq_cfg *cfg; + struct irq_desc *desc; + int count; + int i; -static struct irq_cfg *irq_cfg_alloc(unsigned int irq) -{ - return irq_cfg(irq); -} + cfg = irq_cfgx; + count = ARRAY_SIZE(irq_cfgx); -/* - * Rough estimation of how many shared IRQs there are, can be changed - * anytime. - */ -#define MAX_PLUS_SHARED_IRQS NR_IRQS -#define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) + for (i = 0; i < count; i++) { + desc = irq_to_desc(i); + desc->chip_data = &cfg[i]; + } +} -/* - * This is performance-critical, we want to do it O(1) - * - * the indexing order of this array favors 1:1 mappings - * between pins and IRQs. - */ +#ifdef CONFIG_SPARSE_IRQ +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + struct irq_cfg *cfg = NULL; + struct irq_desc *desc; -struct irq_pin_list { - int apic, pin; - struct irq_pin_list *next; -}; + desc = irq_to_desc(irq); + if (desc) + cfg = desc->chip_data; -static struct irq_pin_list irq_2_pin_head[PIN_MAP_SIZE]; -static struct irq_pin_list *irq_2_pin_ptr; + return cfg; +} -static void __init irq_2_pin_init(void) +static struct irq_cfg *get_one_free_irq_cfg(int cpu) { - struct irq_pin_list *pin = irq_2_pin_head; - int i; + struct irq_cfg *cfg; + int node; + + node = cpu_to_node(cpu); - for (i = 1; i < PIN_MAP_SIZE; i++) - pin[i-1].next = &pin[i]; + cfg = kzalloc_node(sizeof(*cfg), GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc irq_cfg on cpu %d node %d\n", cpu, node); - irq_2_pin_ptr = &pin[0]; + return cfg; } -static struct irq_pin_list *get_one_free_irq_2_pin(void) +void arch_init_chip_data(struct irq_desc *desc, int cpu) { - struct irq_pin_list *pin = irq_2_pin_ptr; + struct irq_cfg *cfg; - if (!pin) - panic("can not get more irq_2_pin\n"); + cfg = desc->chip_data; + if (!cfg) { + desc->chip_data = get_one_free_irq_cfg(cpu); + if (!desc->chip_data) { + printk(KERN_ERR "can not alloc irq_cfg\n"); + BUG_ON(1); + } + } +} - irq_2_pin_ptr = pin->next; - pin->next = NULL; - return pin; +#else +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + return irq < nr_irqs ? irq_cfgx + irq : NULL; } +#endif + struct io_apic { unsigned int index; unsigned int unused[3]; @@ -397,16 +432,19 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -static void add_pin_to_irq(unsigned int irq, int apic, int pin) +static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) { - struct irq_cfg *cfg; struct irq_pin_list *entry; + struct irq_cfg *cfg = irq_cfg(irq); - /* first time to refer irq_cfg, so with new */ - cfg = irq_cfg_alloc(irq); entry = cfg->irq_2_pin; if (!entry) { - entry = get_one_free_irq_2_pin(); + entry = get_one_free_irq_2_pin(cpu); + if (!entry) { + printk(KERN_ERR "can not alloc irq_2_pin to add %d - %d\n", + apic, pin); + return; + } cfg->irq_2_pin = entry; entry->apic = apic; entry->pin = pin; @@ -421,7 +459,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) entry = entry->next; } - entry->next = get_one_free_irq_2_pin(); + entry->next = get_one_free_irq_2_pin(cpu); entry = entry->next; entry->apic = apic; entry->pin = pin; @@ -430,7 +468,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) /* * Reroute an IRQ to a different pin. */ -static void __init replace_pin_at_irq(unsigned int irq, +static void __init replace_pin_at_irq(unsigned int irq, int cpu, int oldapic, int oldpin, int newapic, int newpin) { @@ -451,7 +489,7 @@ static void __init replace_pin_at_irq(unsigned int irq, /* why? call replace before add? */ if (!replaced) - add_pin_to_irq(irq, newapic, newpin); + add_pin_to_irq_cpu(irq, cpu, newapic, newpin); } static inline void io_apic_modify_irq(unsigned int irq, @@ -1162,9 +1200,13 @@ void __setup_vector_irq(int cpu) /* This function must be called with vector_lock held */ int irq, vector; struct irq_cfg *cfg; + struct irq_desc *desc; /* Mark the inuse vectors */ - for_each_irq_cfg(irq, cfg) { + for_each_irq_desc(irq, desc) { + if (!desc) + continue; + cfg = desc->chip_data; if (!cpu_isset(cpu, cfg->domain)) continue; vector = cfg->vector; @@ -1356,6 +1398,8 @@ static void __init setup_IO_APIC_irqs(void) { int apic, pin, idx, irq; int notcon = 0; + struct irq_desc *desc; + int cpu = boot_cpu_id; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); @@ -1387,7 +1431,12 @@ static void __init setup_IO_APIC_irqs(void) if (multi_timer_check(apic, irq)) continue; #endif - add_pin_to_irq(irq, apic, pin); + desc = irq_to_desc_alloc_cpu(irq, cpu); + if (!desc) { + printk(KERN_INFO "can not get irq_desc for %d\n", irq); + continue; + } + add_pin_to_irq_cpu(irq, cpu, apic, pin); setup_IO_APIC_irq(apic, pin, irq, irq_trigger(idx), irq_polarity(idx)); @@ -1448,6 +1497,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_03 reg_03; unsigned long flags; struct irq_cfg *cfg; + struct irq_desc *desc; unsigned int irq; if (apic_verbosity == APIC_QUIET) @@ -1537,8 +1587,13 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for_each_irq_cfg(irq, cfg) { - struct irq_pin_list *entry = cfg->irq_2_pin; + for_each_irq_desc(irq, desc) { + struct irq_pin_list *entry; + + if (!desc) + continue; + cfg = desc->chip_data; + entry = cfg->irq_2_pin; if (!entry) continue; printk(KERN_DEBUG "IRQ%d ", irq); @@ -2022,6 +2077,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) { int was_pending = 0; unsigned long flags; + struct irq_cfg *cfg; spin_lock_irqsave(&ioapic_lock, flags); if (irq < 16) { @@ -2029,6 +2085,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) if (i8259A_irq_pending(irq)) was_pending = 1; } + cfg = irq_cfg(irq); __unmask_IO_APIC_irq(irq); spin_unlock_irqrestore(&ioapic_lock, flags); @@ -2178,6 +2235,9 @@ static void ir_irq_migration(struct work_struct *work) struct irq_desc *desc; for_each_irq_desc(irq, desc) { + if (!desc) + continue; + if (desc->status & IRQ_MOVE_PENDING) { unsigned long flags; @@ -2229,6 +2289,9 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) struct irq_cfg *cfg; irq = __get_cpu_var(vector_irq)[vector]; + if (irq == -1) + continue; + desc = irq_to_desc(irq); if (!desc) continue; @@ -2430,8 +2493,12 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for_each_irq_cfg(irq, cfg) { - if (IO_APIC_IRQ(irq) && !cfg->vector) { + for_each_irq_desc(irq, desc) { + if (!desc) + continue; + + cfg = desc->chip_data; + if (IO_APIC_IRQ(irq) && cfg && !cfg->vector) { /* * Hmm.. We don't have an entry for this, * so default to an old-fashioned 8259 @@ -2439,11 +2506,9 @@ static inline void init_IO_APIC_traps(void) */ if (irq < 16) make_8259A_irq(irq); - else { - desc = irq_to_desc(irq); + else /* Strange. Oh, well.. */ desc->chip = &no_irq_chip; - } } } } @@ -2654,7 +2719,7 @@ static inline void __init check_timer(void) * Ok, does IRQ0 through the IOAPIC work? */ if (no_pin1) { - add_pin_to_irq(0, apic1, pin1); + add_pin_to_irq_cpu(0, boot_cpu_id, apic1, pin1); setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } unmask_IO_APIC_irq(0); @@ -2683,7 +2748,7 @@ static inline void __init check_timer(void) /* * legacy devices should be connected to IO APIC #0 */ - replace_pin_at_irq(0, apic1, pin1, apic2, pin2); + replace_pin_at_irq(0, boot_cpu_id, apic1, pin1, apic2, pin2); setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); unmask_IO_APIC_irq(0); enable_8259A_irq(0); @@ -2902,21 +2967,25 @@ unsigned int create_irq_nr(unsigned int irq_want) unsigned int irq; unsigned int new; unsigned long flags; - struct irq_cfg *cfg_new; - - irq_want = nr_irqs - 1; + struct irq_cfg *cfg_new = NULL; + int cpu = boot_cpu_id; + struct irq_desc *desc_new = NULL; irq = 0; spin_lock_irqsave(&vector_lock, flags); for (new = irq_want; new > 0; new--) { if (platform_legacy_irq(new)) continue; - cfg_new = irq_cfg(new); - if (cfg_new && cfg_new->vector != 0) + + desc_new = irq_to_desc_alloc_cpu(new, cpu); + if (!desc_new) { + printk(KERN_INFO "can not get irq_desc for %d\n", new); + continue; + } + cfg_new = desc_new->chip_data; + + if (cfg_new->vector != 0) continue; - /* check if need to create one */ - if (!cfg_new) - cfg_new = irq_cfg_alloc(new); if (__assign_irq_vector(new, TARGET_CPUS) == 0) irq = new; break; @@ -2925,6 +2994,9 @@ unsigned int create_irq_nr(unsigned int irq_want) if (irq > 0) { dynamic_irq_init(irq); + /* restore it, in case dynamic_irq_init clear it */ + if (desc_new) + desc_new->chip_data = cfg_new; } return irq; } @@ -2944,8 +3016,16 @@ int create_irq(void) void destroy_irq(unsigned int irq) { unsigned long flags; + struct irq_cfg *cfg; + struct irq_desc *desc; + /* store it, in case dynamic_irq_cleanup clear it */ + desc = irq_to_desc(irq); + cfg = desc->chip_data; dynamic_irq_cleanup(irq); + /* connect back irq_cfg */ + if (desc) + desc->chip_data = cfg; #ifdef CONFIG_INTR_REMAP free_irte(irq); @@ -3195,26 +3275,13 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) return 0; } -static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) -{ - unsigned int irq; - - irq = dev->bus->number; - irq <<= 8; - irq |= dev->devfn; - irq <<= 12; - - return irq; -} - -int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) +int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc) { unsigned int irq; int ret; unsigned int irq_want; - irq_want = build_irq_for_pci_dev(dev) + 0x100; - + irq_want = nr_irqs - 1; irq = create_irq_nr(irq_want); if (irq == 0) return -1; @@ -3228,7 +3295,7 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) goto error; no_ir: #endif - ret = setup_msi_irq(dev, desc, irq); + ret = setup_msi_irq(dev, msidesc, irq); if (ret < 0) { destroy_irq(irq); return ret; @@ -3246,7 +3313,7 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) { unsigned int irq; int ret, sub_handle; - struct msi_desc *desc; + struct msi_desc *msidesc; unsigned int irq_want; #ifdef CONFIG_INTR_REMAP @@ -3254,10 +3321,11 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) int index = 0; #endif - irq_want = build_irq_for_pci_dev(dev) + 0x100; + irq_want = nr_irqs - 1; sub_handle = 0; - list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq_nr(irq_want--); + list_for_each_entry(msidesc, &dev->msi_list, list) { + irq = create_irq_nr(irq_want); + irq_want--; if (irq == 0) return -1; #ifdef CONFIG_INTR_REMAP @@ -3289,7 +3357,7 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) } no_ir: #endif - ret = setup_msi_irq(dev, desc, irq); + ret = setup_msi_irq(dev, msidesc, irq); if (ret < 0) goto error; sub_handle++; @@ -3707,17 +3775,29 @@ int __init io_apic_get_version(int ioapic) int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) { + struct irq_desc *desc; + struct irq_cfg *cfg; + int cpu = boot_cpu_id; + if (!IO_APIC_IRQ(irq)) { apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", ioapic); return -EINVAL; } + desc = irq_to_desc_alloc_cpu(irq, cpu); + if (!desc) { + printk(KERN_INFO "can not get irq_desc %d\n", irq); + return 0; + } + /* * IRQs < 16 are already in the irq_2_pin[] map */ - if (irq >= 16) - add_pin_to_irq(irq, ioapic, pin); + if (irq >= 16) { + cfg = desc->chip_data; + add_pin_to_irq_cpu(irq, cpu, ioapic, pin); + } setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); @@ -3773,7 +3853,8 @@ void __init setup_ioapic_dest(void) * when you have too many devices, because at that time only boot * cpu is online. */ - cfg = irq_cfg(irq); + desc = irq_to_desc(irq); + cfg = desc->chip_data; if (!cfg->vector) { setup_IO_APIC_irq(ioapic, pin, irq, irq_trigger(irq_entry), @@ -3785,7 +3866,6 @@ void __init setup_ioapic_dest(void) /* * Honour affinities which have been set in early boot */ - desc = irq_to_desc(irq); if (desc->status & (IRQ_NO_BALANCING | IRQ_AFFINITY_SET)) mask = desc->affinity; @@ -3846,7 +3926,6 @@ void __init ioapic_init_mappings(void) struct resource *ioapic_res; int i; - irq_2_pin_init(); ioapic_res = ioapic_setup_resources(); for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index d1d4dc52f649..3f1d9d18df67 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -118,6 +118,9 @@ int show_interrupts(struct seq_file *p, void *v) } desc = irq_to_desc(i); + if (!desc) + return 0; + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP any_count = kstat_irqs(i); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index a51382672de0..119fc9c8ff7f 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -242,6 +242,8 @@ void fixup_irqs(cpumask_t map) for_each_irq_desc(irq, desc) { cpumask_t mask; + if (!desc) + continue; if (irq == 2) continue; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 60eb84eb77a0..900009c70591 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -94,6 +94,8 @@ void fixup_irqs(cpumask_t map) int break_affinity = 0; int set_affinity = 1; + if (!desc) + continue; if (irq == 2) continue; diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 845aa9803e80..5a5651b7f9e6 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -69,7 +69,6 @@ void __init init_ISA_irqs (void) * 16 old-style INTA-cycle interrupts: */ for (i = 0; i < 16; i++) { - /* first time call this irq_desc */ struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index ff0235391285..cd9f42d028d9 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -143,7 +143,6 @@ void __init init_ISA_irqs(void) init_8259A(0); for (i = 0; i < 16; i++) { - /* first time call this irq_desc */ struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; -- cgit v1.2.3 From 99d093d12897562a253540a902bbf65ec16042ac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:32 -0800 Subject: x86: use NR_IRQS_LEGACY Impact: cleanup Introduce NR_IRQS_LEGACY instead of hard coded number. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 10 +++++----- arch/x86/kernel/irqinit_32.c | 2 +- arch/x86/kernel/irqinit_64.c | 2 +- 3 files changed, 7 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9de17f5c1125..9870461ae933 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -847,7 +847,7 @@ EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); */ static int EISA_ELCR(unsigned int irq) { - if (irq < 16) { + if (irq < NR_IRQS_LEGACY) { unsigned int port = 0x4d0 + (irq >> 3); return (inb(port) >> (irq & 7)) & 1; } @@ -1388,7 +1388,7 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, } ioapic_register_intr(irq, trigger); - if (irq < 16) + if (irq < NR_IRQS_LEGACY) disable_8259A_irq(irq); ioapic_write_entry(apic, pin, entry); @@ -2080,7 +2080,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) struct irq_cfg *cfg; spin_lock_irqsave(&ioapic_lock, flags); - if (irq < 16) { + if (irq < NR_IRQS_LEGACY) { disable_8259A_irq(irq); if (i8259A_irq_pending(irq)) was_pending = 1; @@ -2504,7 +2504,7 @@ static inline void init_IO_APIC_traps(void) * so default to an old-fashioned 8259 * interrupt if we can.. */ - if (irq < 16) + if (irq < NR_IRQS_LEGACY) make_8259A_irq(irq); else /* Strange. Oh, well.. */ @@ -3794,7 +3794,7 @@ int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int p /* * IRQs < 16 are already in the irq_2_pin[] map */ - if (irq >= 16) { + if (irq >= NR_IRQS_LEGACY) { cfg = desc->chip_data; add_pin_to_irq_cpu(irq, cpu, ioapic, pin); } diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 5a5651b7f9e6..6a92f47c52e7 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -68,7 +68,7 @@ void __init init_ISA_irqs (void) /* * 16 old-style INTA-cycle interrupts: */ - for (i = 0; i < 16; i++) { + for (i = 0; i < NR_IRQS_LEGACY; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index cd9f42d028d9..40c1e62ec785 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -142,7 +142,7 @@ void __init init_ISA_irqs(void) init_bsp_APIC(); init_8259A(0); - for (i = 0; i < 16; i++) { + for (i = 0; i < NR_IRQS_LEGACY; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; -- cgit v1.2.3 From be5d5350a937cd8513b258739f1099420129e96f Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:33 -0800 Subject: x86: MSI start irq numbering from nr_irqs_gsi Impact: sanitize MSI irq number ordering from top-down to bottom-up Increase new MSI IRQs starting from nr_irqs_gsi (which is somewhere below 256), instead of decreasing from NR_IRQS. (The latter method can result in confusingly high IRQ numbers - if NR_CPUS is set to a high value and NR_IRQS scales up to a high value.) Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 24 +++++++++++++++++------- arch/x86/kernel/setup.c | 2 +- 2 files changed, 18 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9870461ae933..0dcde74abd1d 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2973,7 +2973,7 @@ unsigned int create_irq_nr(unsigned int irq_want) irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = irq_want; new > 0; new--) { + for (new = irq_want; new < NR_IRQS; new++) { if (platform_legacy_irq(new)) continue; @@ -3001,11 +3001,14 @@ unsigned int create_irq_nr(unsigned int irq_want) return irq; } +static int nr_irqs_gsi = NR_IRQS_LEGACY; int create_irq(void) { + unsigned int irq_want; int irq; - irq = create_irq_nr(nr_irqs - 1); + irq_want = nr_irqs_gsi; + irq = create_irq_nr(irq_want); if (irq == 0) irq = -1; @@ -3281,7 +3284,7 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc) int ret; unsigned int irq_want; - irq_want = nr_irqs - 1; + irq_want = nr_irqs_gsi; irq = create_irq_nr(irq_want); if (irq == 0) return -1; @@ -3321,11 +3324,11 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) int index = 0; #endif - irq_want = nr_irqs - 1; + irq_want = nr_irqs_gsi; sub_handle = 0; list_for_each_entry(msidesc, &dev->msi_list, list) { irq = create_irq_nr(irq_want); - irq_want--; + irq_want++; if (irq == 0) return -1; #ifdef CONFIG_INTR_REMAP @@ -3674,9 +3677,16 @@ int __init io_apic_get_redir_entries (int ioapic) return reg_01.bits.entries; } -int __init probe_nr_irqs(void) +void __init probe_nr_irqs_gsi(void) { - return NR_IRQS; + int idx; + int nr = 0; + + for (idx = 0; idx < nr_ioapics; idx++) + nr += io_apic_get_redir_entries(idx) + 1; + + if (nr > nr_irqs_gsi) + nr_irqs_gsi = nr; } /* -------------------------------------------------------------------------- diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 9d5674f7b6cc..a3834f123206 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1082,7 +1082,7 @@ void __init setup_arch(char **cmdline_p) ioapic_init_mappings(); /* need to wait for io_apic is mapped */ - nr_irqs = probe_nr_irqs(); + probe_nr_irqs_gsi(); kvm_guest_init(); -- cgit v1.2.3 From 3145e941fcfe2548fa2270afb1a05bab3a6bc418 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:34 -0800 Subject: x86, MSI: pass irq_cfg and irq_desc Impact: simplify code Pass irq_desc and cfg around, instead of raw IRQ numbers - this way we dont have to look it up again and again. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 318 ++++++++++++++++++++++++++-------------------- 1 file changed, 181 insertions(+), 137 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 0dcde74abd1d..a1a2e070f31a 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -231,6 +231,10 @@ static struct irq_cfg *irq_cfg(unsigned int irq) #endif +static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) +{ +} + struct io_apic { unsigned int index; unsigned int unused[3]; @@ -272,11 +276,10 @@ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned writel(value, &io_apic->data); } -static bool io_apic_level_ack_pending(unsigned int irq) +static bool io_apic_level_ack_pending(struct irq_cfg *cfg) { struct irq_pin_list *entry; unsigned long flags; - struct irq_cfg *cfg = irq_cfg(irq); spin_lock_irqsave(&ioapic_lock, flags); entry = cfg->irq_2_pin; @@ -358,13 +361,12 @@ static void ioapic_mask_entry(int apic, int pin) } #ifdef CONFIG_SMP -static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) +static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, struct irq_cfg *cfg) { int apic, pin; - struct irq_cfg *cfg; struct irq_pin_list *entry; + u8 vector = cfg->vector; - cfg = irq_cfg(irq); entry = cfg->irq_2_pin; for (;;) { unsigned int reg; @@ -394,24 +396,27 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) } } -static int assign_irq_vector(int irq, cpumask_t mask); +static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask); -static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +static void set_ioapic_affinity_irq_desc(struct irq_desc *desc, cpumask_t mask) { struct irq_cfg *cfg; unsigned long flags; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; + unsigned int irq; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - cfg = irq_cfg(irq); - if (assign_irq_vector(irq, mask)) + irq = desc->irq; + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); /* @@ -419,12 +424,20 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) */ dest = SET_APIC_LOGICAL_ID(dest); - desc = irq_to_desc(irq); spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); + __target_IO_APIC_irq(irq, dest, cfg); desc->affinity = mask; spin_unlock_irqrestore(&ioapic_lock, flags); } + +static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc; + + desc = irq_to_desc(irq); + + set_ioapic_affinity_irq_desc(desc, mask); +} #endif /* CONFIG_SMP */ /* @@ -432,10 +445,9 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) +static void add_pin_to_irq_cpu(struct irq_cfg *cfg, int cpu, int apic, int pin) { struct irq_pin_list *entry; - struct irq_cfg *cfg = irq_cfg(irq); entry = cfg->irq_2_pin; if (!entry) { @@ -468,11 +480,10 @@ static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) /* * Reroute an IRQ to a different pin. */ -static void __init replace_pin_at_irq(unsigned int irq, int cpu, +static void __init replace_pin_at_irq_cpu(struct irq_cfg *cfg, int cpu, int oldapic, int oldpin, int newapic, int newpin) { - struct irq_cfg *cfg = irq_cfg(irq); struct irq_pin_list *entry = cfg->irq_2_pin; int replaced = 0; @@ -489,18 +500,16 @@ static void __init replace_pin_at_irq(unsigned int irq, int cpu, /* why? call replace before add? */ if (!replaced) - add_pin_to_irq_cpu(irq, cpu, newapic, newpin); + add_pin_to_irq_cpu(cfg, cpu, newapic, newpin); } -static inline void io_apic_modify_irq(unsigned int irq, +static inline void io_apic_modify_irq(struct irq_cfg *cfg, int mask_and, int mask_or, void (*final)(struct irq_pin_list *entry)) { int pin; - struct irq_cfg *cfg; struct irq_pin_list *entry; - cfg = irq_cfg(irq); for (entry = cfg->irq_2_pin; entry != NULL; entry = entry->next) { unsigned int reg; pin = entry->pin; @@ -513,9 +522,9 @@ static inline void io_apic_modify_irq(unsigned int irq, } } -static void __unmask_IO_APIC_irq(unsigned int irq) +static void __unmask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, 0, NULL); + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_MASKED, 0, NULL); } #ifdef CONFIG_X86_64 @@ -530,47 +539,64 @@ void io_apic_sync(struct irq_pin_list *entry) readl(&io_apic->data); } -static void __mask_IO_APIC_irq(unsigned int irq) +static void __mask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, &io_apic_sync); + io_apic_modify_irq(cfg, ~0, IO_APIC_REDIR_MASKED, &io_apic_sync); } #else /* CONFIG_X86_32 */ -static void __mask_IO_APIC_irq(unsigned int irq) +static void __mask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, NULL); + io_apic_modify_irq(cfg, ~0, IO_APIC_REDIR_MASKED, NULL); } -static void __mask_and_edge_IO_APIC_irq(unsigned int irq) +static void __mask_and_edge_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_LEVEL_TRIGGER, + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_LEVEL_TRIGGER, IO_APIC_REDIR_MASKED, NULL); } -static void __unmask_and_level_IO_APIC_irq(unsigned int irq) +static void __unmask_and_level_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_MASKED, IO_APIC_REDIR_LEVEL_TRIGGER, NULL); } #endif /* CONFIG_X86_32 */ -static void mask_IO_APIC_irq (unsigned int irq) +static void mask_IO_APIC_irq_desc(struct irq_desc *desc) { + struct irq_cfg *cfg = desc->chip_data; unsigned long flags; + BUG_ON(!cfg); + spin_lock_irqsave(&ioapic_lock, flags); - __mask_IO_APIC_irq(irq); + __mask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } -static void unmask_IO_APIC_irq (unsigned int irq) +static void unmask_IO_APIC_irq_desc(struct irq_desc *desc) { + struct irq_cfg *cfg = desc->chip_data; unsigned long flags; spin_lock_irqsave(&ioapic_lock, flags); - __unmask_IO_APIC_irq(irq); + __unmask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } +static void mask_IO_APIC_irq(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + + mask_IO_APIC_irq_desc(desc); +} +static void unmask_IO_APIC_irq(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + + unmask_IO_APIC_irq_desc(desc); +} + static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) { struct IO_APIC_route_entry entry; @@ -1072,7 +1098,7 @@ void unlock_vector_lock(void) spin_unlock(&vector_lock); } -static int __assign_irq_vector(int irq, cpumask_t mask) +static int __assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) { /* * NOTE! The local APIC isn't very good at handling @@ -1088,16 +1114,13 @@ static int __assign_irq_vector(int irq, cpumask_t mask) static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; unsigned int old_vector; int cpu; - struct irq_cfg *cfg; - cfg = irq_cfg(irq); + if ((cfg->move_in_progress) || cfg->move_cleanup_count) + return -EBUSY; /* Only try and allocate irqs on cpus that are present */ cpus_and(mask, mask, cpu_online_map); - if ((cfg->move_in_progress) || cfg->move_cleanup_count) - return -EBUSY; - old_vector = cfg->vector; if (old_vector) { cpumask_t tmp; @@ -1151,24 +1174,22 @@ next: return -ENOSPC; } -static int assign_irq_vector(int irq, cpumask_t mask) +static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) { int err; unsigned long flags; spin_lock_irqsave(&vector_lock, flags); - err = __assign_irq_vector(irq, mask); + err = __assign_irq_vector(irq, cfg, mask); spin_unlock_irqrestore(&vector_lock, flags); return err; } -static void __clear_irq_vector(int irq) +static void __clear_irq_vector(int irq, struct irq_cfg *cfg) { - struct irq_cfg *cfg; cpumask_t mask; int cpu, vector; - cfg = irq_cfg(irq); BUG_ON(!cfg->vector); vector = cfg->vector; @@ -1257,11 +1278,8 @@ static inline int IO_APIC_irq_trigger(int irq) } #endif -static void ioapic_register_intr(int irq, unsigned long trigger) +static void ioapic_register_intr(int irq, struct irq_desc *desc, unsigned long trigger) { - struct irq_desc *desc; - - desc = irq_to_desc(irq); if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) @@ -1353,7 +1371,7 @@ static int setup_ioapic_entry(int apic, int irq, return 0; } -static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, +static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_desc *desc, int trigger, int polarity) { struct irq_cfg *cfg; @@ -1363,10 +1381,10 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, if (!IO_APIC_IRQ(irq)) return; - cfg = irq_cfg(irq); + cfg = desc->chip_data; mask = TARGET_CPUS; - if (assign_irq_vector(irq, mask)) + if (assign_irq_vector(irq, cfg, mask)) return; cpus_and(mask, cfg->domain, mask); @@ -1383,11 +1401,11 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, cfg->vector)) { printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", mp_ioapics[apic].mp_apicid, pin); - __clear_irq_vector(irq); + __clear_irq_vector(irq, cfg); return; } - ioapic_register_intr(irq, trigger); + ioapic_register_intr(irq, desc, trigger); if (irq < NR_IRQS_LEGACY) disable_8259A_irq(irq); @@ -1399,6 +1417,7 @@ static void __init setup_IO_APIC_irqs(void) int apic, pin, idx, irq; int notcon = 0; struct irq_desc *desc; + struct irq_cfg *cfg; int cpu = boot_cpu_id; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); @@ -1436,9 +1455,10 @@ static void __init setup_IO_APIC_irqs(void) printk(KERN_INFO "can not get irq_desc for %d\n", irq); continue; } - add_pin_to_irq_cpu(irq, cpu, apic, pin); + cfg = desc->chip_data; + add_pin_to_irq_cpu(cfg, cpu, apic, pin); - setup_IO_APIC_irq(apic, pin, irq, + setup_IO_APIC_irq(apic, pin, irq, desc, irq_trigger(idx), irq_polarity(idx)); } } @@ -2086,7 +2106,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) was_pending = 1; } cfg = irq_cfg(irq); - __unmask_IO_APIC_irq(irq); + __unmask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); return was_pending; @@ -2149,35 +2169,37 @@ static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); * as simple as edge triggered migration and we can do the irq migration * with a simple atomic update to IO-APIC RTE. */ -static void migrate_ioapic_irq(int irq, cpumask_t mask) +static void migrate_ioapic_irq_desc(struct irq_desc *desc, cpumask_t mask) { struct irq_cfg *cfg; - struct irq_desc *desc; cpumask_t tmp, cleanup_mask; struct irte irte; int modify_ioapic_rte; unsigned int dest; unsigned long flags; + unsigned int irq; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; + irq = desc->irq; if (get_irte(irq, &irte)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); - desc = irq_to_desc(irq); modify_ioapic_rte = desc->status & IRQ_LEVEL; if (modify_ioapic_rte) { spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); + __target_IO_APIC_irq(irq, dest, cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } @@ -2199,14 +2221,14 @@ static void migrate_ioapic_irq(int irq, cpumask_t mask) desc->affinity = mask; } -static int migrate_irq_remapped_level(int irq) +static int migrate_irq_remapped_level_desc(struct irq_desc *desc) { int ret = -1; - struct irq_desc *desc = irq_to_desc(irq); + struct irq_cfg *cfg = desc->chip_data; - mask_IO_APIC_irq(irq); + mask_IO_APIC_irq_desc(desc); - if (io_apic_level_ack_pending(irq)) { + if (io_apic_level_ack_pending(cfg)) { /* * Interrupt in progress. Migrating irq now will change the * vector information in the IO-APIC RTE and that will confuse @@ -2218,14 +2240,15 @@ static int migrate_irq_remapped_level(int irq) } /* everthing is clear. we have right of way */ - migrate_ioapic_irq(irq, desc->pending_mask); + migrate_ioapic_irq_desc(desc, desc->pending_mask); ret = 0; desc->status &= ~IRQ_MOVE_PENDING; cpus_clear(desc->pending_mask); unmask: - unmask_IO_APIC_irq(irq); + unmask_IO_APIC_irq_desc(desc); + return ret; } @@ -2258,18 +2281,22 @@ static void ir_irq_migration(struct work_struct *work) /* * Migrates the IRQ destination in the process context. */ -static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +static void set_ir_ioapic_affinity_irq_desc(struct irq_desc *desc, cpumask_t mask) { - struct irq_desc *desc = irq_to_desc(irq); - if (desc->status & IRQ_LEVEL) { desc->status |= IRQ_MOVE_PENDING; desc->pending_mask = mask; - migrate_irq_remapped_level(irq); + migrate_irq_remapped_level_desc(desc); return; } - migrate_ioapic_irq(irq, mask); + migrate_ioapic_irq_desc(desc, mask); +} +static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc = irq_to_desc(irq); + + set_ir_ioapic_affinity_irq_desc(desc, mask); } #endif @@ -2313,9 +2340,10 @@ unlock: irq_exit(); } -static void irq_complete_move(unsigned int irq) +static void irq_complete_move(struct irq_desc **descp) { - struct irq_cfg *cfg = irq_cfg(irq); + struct irq_desc *desc = *descp; + struct irq_cfg *cfg = desc->chip_data; unsigned vector, me; if (likely(!cfg->move_in_progress)) @@ -2333,8 +2361,9 @@ static void irq_complete_move(unsigned int irq) } } #else -static inline void irq_complete_move(unsigned int irq) {} +static inline void irq_complete_move(struct irq_desc **descp) {} #endif + #ifdef CONFIG_INTR_REMAP static void ack_x2apic_level(unsigned int irq) { @@ -2345,11 +2374,14 @@ static void ack_x2apic_edge(unsigned int irq) { ack_x2APIC_irq(); } + #endif static void ack_apic_edge(unsigned int irq) { - irq_complete_move(irq); + struct irq_desc *desc = irq_to_desc(irq); + + irq_complete_move(&desc); move_native_irq(irq); ack_APIC_irq(); } @@ -2358,18 +2390,21 @@ atomic_t irq_mis_count; static void ack_apic_level(unsigned int irq) { + struct irq_desc *desc = irq_to_desc(irq); + #ifdef CONFIG_X86_32 unsigned long v; int i; #endif + struct irq_cfg *cfg; int do_unmask_irq = 0; - irq_complete_move(irq); + irq_complete_move(&desc); #ifdef CONFIG_GENERIC_PENDING_IRQ /* If we are moving the irq we need to mask it */ - if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { + if (unlikely(desc->status & IRQ_MOVE_PENDING)) { do_unmask_irq = 1; - mask_IO_APIC_irq(irq); + mask_IO_APIC_irq_desc(desc); } #endif @@ -2393,7 +2428,8 @@ static void ack_apic_level(unsigned int irq) * operation to prevent an edge-triggered interrupt escaping meanwhile. * The idea is from Manfred Spraul. --macro */ - i = irq_cfg(irq)->vector; + cfg = desc->chip_data; + i = cfg->vector; v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); #endif @@ -2432,17 +2468,18 @@ static void ack_apic_level(unsigned int irq) * accurate and is causing problems then it is a hardware bug * and you can go talk to the chipset vendor about it. */ - if (!io_apic_level_ack_pending(irq)) + cfg = desc->chip_data; + if (!io_apic_level_ack_pending(cfg)) move_masked_irq(irq); - unmask_IO_APIC_irq(irq); + unmask_IO_APIC_irq_desc(desc); } #ifdef CONFIG_X86_32 if (!(v & (1 << (i & 0x1f)))) { atomic_inc(&irq_mis_count); spin_lock(&ioapic_lock); - __mask_and_edge_IO_APIC_irq(irq); - __unmask_and_level_IO_APIC_irq(irq); + __mask_and_edge_IO_APIC_irq(cfg); + __unmask_and_level_IO_APIC_irq(cfg); spin_unlock(&ioapic_lock); } #endif @@ -2533,7 +2570,7 @@ static void unmask_lapic_irq(unsigned int irq) apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); } -static void ack_lapic_irq (unsigned int irq) +static void ack_lapic_irq(unsigned int irq) { ack_APIC_irq(); } @@ -2545,11 +2582,8 @@ static struct irq_chip lapic_chip __read_mostly = { .ack = ack_lapic_irq, }; -static void lapic_register_intr(int irq) +static void lapic_register_intr(int irq, struct irq_desc *desc) { - struct irq_desc *desc; - - desc = irq_to_desc(irq); desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, "edge"); @@ -2653,7 +2687,9 @@ int timer_through_8259 __initdata; */ static inline void __init check_timer(void) { - struct irq_cfg *cfg = irq_cfg(0); + struct irq_desc *desc = irq_to_desc(0); + struct irq_cfg *cfg = desc->chip_data; + int cpu = boot_cpu_id; int apic1, pin1, apic2, pin2; unsigned long flags; unsigned int ver; @@ -2668,7 +2704,7 @@ static inline void __init check_timer(void) * get/set the timer IRQ vector: */ disable_8259A_irq(0); - assign_irq_vector(0, TARGET_CPUS); + assign_irq_vector(0, cfg, TARGET_CPUS); /* * As IRQ0 is to be enabled in the 8259A, the virtual @@ -2719,10 +2755,10 @@ static inline void __init check_timer(void) * Ok, does IRQ0 through the IOAPIC work? */ if (no_pin1) { - add_pin_to_irq_cpu(0, boot_cpu_id, apic1, pin1); + add_pin_to_irq_cpu(cfg, cpu, apic1, pin1); setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } - unmask_IO_APIC_irq(0); + unmask_IO_APIC_irq_desc(desc); if (timer_irq_works()) { if (nmi_watchdog == NMI_IO_APIC) { setup_nmi(); @@ -2748,9 +2784,9 @@ static inline void __init check_timer(void) /* * legacy devices should be connected to IO APIC #0 */ - replace_pin_at_irq(0, boot_cpu_id, apic1, pin1, apic2, pin2); + replace_pin_at_irq_cpu(cfg, cpu, apic1, pin1, apic2, pin2); setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); - unmask_IO_APIC_irq(0); + unmask_IO_APIC_irq_desc(desc); enable_8259A_irq(0); if (timer_irq_works()) { apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); @@ -2782,7 +2818,7 @@ static inline void __init check_timer(void) apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as Virtual Wire IRQ...\n"); - lapic_register_intr(0); + lapic_register_intr(0, desc); apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ enable_8259A_irq(0); @@ -2986,7 +3022,7 @@ unsigned int create_irq_nr(unsigned int irq_want) if (cfg_new->vector != 0) continue; - if (__assign_irq_vector(new, TARGET_CPUS) == 0) + if (__assign_irq_vector(new, cfg_new, TARGET_CPUS) == 0) irq = new; break; } @@ -3034,7 +3070,7 @@ void destroy_irq(unsigned int irq) free_irte(irq); #endif spin_lock_irqsave(&vector_lock, flags); - __clear_irq_vector(irq); + __clear_irq_vector(irq, cfg); spin_unlock_irqrestore(&vector_lock, flags); } @@ -3049,12 +3085,12 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms unsigned dest; cpumask_t tmp; + cfg = irq_cfg(irq); tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); + err = assign_irq_vector(irq, cfg, tmp); if (err) return err; - cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -3112,35 +3148,35 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms #ifdef CONFIG_SMP static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); - read_msi_msg(irq, &msg); + read_msi_msg_desc(desc, &msg); msg.data &= ~MSI_DATA_VECTOR_MASK; msg.data |= MSI_DATA_VECTOR(cfg->vector); msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; msg.address_lo |= MSI_ADDR_DEST_ID(dest); - write_msi_msg(irq, &msg); - desc = irq_to_desc(irq); + write_msi_msg_desc(desc, &msg); desc->affinity = mask; } - #ifdef CONFIG_INTR_REMAP /* * Migrate the MSI irq to another cpumask. This migration is @@ -3148,11 +3184,11 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) */ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp, cleanup_mask; struct irte irte; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -3161,10 +3197,12 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) if (get_irte(irq, &irte)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3188,9 +3226,9 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) cfg->move_in_progress = 0; } - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif #endif /* CONFIG_SMP */ @@ -3249,7 +3287,7 @@ static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) } #endif -static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) +static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc, int irq) { int ret; struct msi_msg msg; @@ -3258,7 +3296,7 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) if (ret < 0) return ret; - set_irq_msi(irq, desc); + set_irq_msi(irq, msidesc); write_msi_msg(irq, &msg); #ifdef CONFIG_INTR_REMAP @@ -3381,20 +3419,22 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_SMP static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3406,9 +3446,9 @@ static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); dmar_msi_write(irq, &msg); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif /* CONFIG_SMP */ struct irq_chip dmar_msi_type = { @@ -3442,8 +3482,8 @@ int arch_setup_dmar_msi(unsigned int irq) #ifdef CONFIG_SMP static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; - struct irq_desc *desc; struct msi_msg msg; unsigned int dest; cpumask_t tmp; @@ -3452,10 +3492,12 @@ static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3467,9 +3509,9 @@ static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); hpet_msi_write(irq, &msg); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif /* CONFIG_SMP */ struct irq_chip hpet_msi_type = { @@ -3524,26 +3566,28 @@ static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); target_ht_irq(irq, dest, cfg->vector); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif static struct irq_chip ht_irq_chip = { @@ -3563,13 +3607,13 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) int err; cpumask_t tmp; + cfg = irq_cfg(irq); tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); + err = assign_irq_vector(irq, cfg, tmp); if (!err) { struct ht_irq_msg msg; unsigned dest; - cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -3615,7 +3659,9 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, unsigned long flags; int err; - err = assign_irq_vector(irq, *eligible_cpu); + cfg = irq_cfg(irq); + + err = assign_irq_vector(irq, cfg, *eligible_cpu); if (err != 0) return err; @@ -3624,8 +3670,6 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, irq_name); spin_unlock_irqrestore(&vector_lock, flags); - cfg = irq_cfg(irq); - mmr_value = 0; entry = (struct uv_IO_APIC_route_entry *)&mmr_value; BUG_ON(sizeof(struct uv_IO_APIC_route_entry) != sizeof(unsigned long)); @@ -3806,10 +3850,10 @@ int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int p */ if (irq >= NR_IRQS_LEGACY) { cfg = desc->chip_data; - add_pin_to_irq_cpu(irq, cpu, ioapic, pin); + add_pin_to_irq_cpu(cfg, cpu, ioapic, pin); } - setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); + setup_IO_APIC_irq(ioapic, pin, irq, desc, triggering, polarity); return 0; } @@ -3866,7 +3910,7 @@ void __init setup_ioapic_dest(void) desc = irq_to_desc(irq); cfg = desc->chip_data; if (!cfg->vector) { - setup_IO_APIC_irq(ioapic, pin, irq, + setup_IO_APIC_irq(ioapic, pin, irq, desc, irq_trigger(irq_entry), irq_polarity(irq_entry)); continue; @@ -3884,10 +3928,10 @@ void __init setup_ioapic_dest(void) #ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq(irq, mask); + set_ir_ioapic_affinity_irq_desc(desc, mask); else #endif - set_ioapic_affinity_irq(irq, mask); + set_ioapic_affinity_irq_desc(desc, mask); } } -- cgit v1.2.3 From bb9d4ff80bc032d7961815c2ff5eaf458ae3adff Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 4 Dec 2008 15:59:48 +0100 Subject: AMD IOMMU: fix iommu_map_page function Impact: bugfix in iommu_map_page function Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 5662e226b0c9..67970a8c2ee9 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -344,7 +344,7 @@ static int iommu_map(struct protection_domain *dom, u64 __pte, *pte, *page; bus_addr = PAGE_ALIGN(bus_addr); - phys_addr = PAGE_ALIGN(bus_addr); + phys_addr = PAGE_ALIGN(phys_addr); /* only support 512GB address spaces for now */ if (bus_addr > IOMMU_MAP_SIZE_L3 || !(prot & IOMMU_PROT_MASK)) -- cgit v1.2.3 From 3cc3d84bffbd93bdb671ac7961b12cd98fbb9266 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 4 Dec 2008 16:44:31 +0100 Subject: AMD IOMMU: fix loop counter in free_pagetable function Impact: bugfix Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 67970a8c2ee9..0637e65ca0c7 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -600,7 +600,7 @@ static void dma_ops_free_pagetable(struct dma_ops_domain *dma_dom) continue; p2 = IOMMU_PTE_PAGE(p1[i]); - for (j = 0; j < 512; ++i) { + for (j = 0; j < 512; ++j) { if (!IOMMU_PTE_PRESENT(p2[j])) continue; p3 = IOMMU_PTE_PAGE(p2[j]); -- cgit v1.2.3 From 24f811603e8ef4b9173f47fa263230168e237128 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 8 Dec 2008 14:25:39 +0100 Subject: AMD IOMMU: fix typo in comment Impact: cleanup Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 0637e65ca0c7..2fde073b6be6 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -922,8 +922,8 @@ static void dma_ops_domain_unmap(struct amd_iommu *iommu, /* * This function contains common code for mapping of a physically - * contiguous memory region into DMA address space. It is uses by all - * mapping functions provided by this IOMMU driver. + * contiguous memory region into DMA address space. It is used by all + * mapping functions provided with this IOMMU driver. * Must be called with the domain lock held. */ static dma_addr_t __map_single(struct device *dev, -- cgit v1.2.3 From 8ad909c4c1b91bd35ba6a2af5e7ab3fc8d9fe283 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 8 Dec 2008 14:37:20 +0100 Subject: AMD IOMMU: fix WARN_ON in dma_ops unmap path Impact: minor fix Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 2fde073b6be6..3133a0ea09ff 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -910,7 +910,7 @@ static void dma_ops_domain_unmap(struct amd_iommu *iommu, if (address >= dom->aperture_size) return; - WARN_ON(address & 0xfffULL || address > dom->aperture_size); + WARN_ON(address & ~PAGE_MASK || address >= dom->aperture_size); pte = dom->pte_pages[IOMMU_PTE_L1_INDEX(address)]; pte += IOMMU_PTE_L0_INDEX(address); -- cgit v1.2.3 From b8d9905d025d80a2357e8ce4704fde2923f6a1bd Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 8 Dec 2008 14:40:26 +0100 Subject: AMD IOMMU: __unmap_single: check for bad_dma_address instead of 0 Impact: minor fix Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 3133a0ea09ff..a7b6dec6fc3f 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -983,7 +983,8 @@ static void __unmap_single(struct amd_iommu *iommu, dma_addr_t i, start; unsigned int pages; - if ((dma_addr == 0) || (dma_addr + size > dma_dom->aperture_size)) + if ((dma_addr == bad_dma_address) || + (dma_addr + size > dma_dom->aperture_size)) return; pages = iommu_num_pages(dma_addr, size, PAGE_SIZE); -- cgit v1.2.3 From 69b88afa8d114a43a3c0431722b79e31d9920692 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 22:45:50 -0800 Subject: x86: clean up get_smp_config() Impact: cleanup reorder exit path in __get_smp_config(). also move two print outs to acpi_process_madt Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 11 +++++++++++ arch/x86/kernel/mpparse.c | 25 +++++++++++-------------- 2 files changed, 22 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 4c51a2f8fd31..65d0b72777ea 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -1360,6 +1360,17 @@ static void __init acpi_process_madt(void) disable_acpi(); } } + + /* + * ACPI supports both logical (e.g. Hyper-Threading) and physical + * processors, where MPS only supports physical. + */ + if (acpi_lapic && acpi_ioapic) + printk(KERN_INFO "Using ACPI (MADT) for SMP configuration " + "information\n"); + else if (acpi_lapic) + printk(KERN_INFO "Using ACPI for processor (LAPIC) " + "configuration information\n"); #endif return; } diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 0f4c1fd5a1f4..45e3b69808ba 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -586,26 +586,23 @@ static void __init __get_smp_config(unsigned int early) { struct intel_mp_floating *mpf = mpf_found; - if (x86_quirks->mach_get_smp_config) { - if (x86_quirks->mach_get_smp_config(early)) - return; - } + if (!mpf) + return; + if (acpi_lapic && early) return; + /* - * ACPI supports both logical (e.g. Hyper-Threading) and physical - * processors, where MPS only supports physical. + * MPS doesn't support hyperthreading, aka only have + * thread 0 apic id in MPS table */ - if (acpi_lapic && acpi_ioapic) { - printk(KERN_INFO "Using ACPI (MADT) for SMP configuration " - "information\n"); + if (acpi_lapic && acpi_ioapic) return; - } else if (acpi_lapic) - printk(KERN_INFO "Using ACPI for processor (LAPIC) " - "configuration information\n"); - if (!mpf) - return; + if (x86_quirks->mach_get_smp_config) { + if (x86_quirks->mach_get_smp_config(early)) + return; + } printk(KERN_INFO "Intel MultiProcessor Specification v1.%d\n", mpf->mpf_specification); -- cgit v1.2.3 From 8b96f0119818964e4944fd1c423bf6770027d3ac Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 6 Dec 2008 03:40:00 +0100 Subject: tracing/function-graph-tracer: introduce __notrace_funcgraph to filter special functions Impact: trace more functions When the function graph tracer is configured, three more files are not traced to prevent only four functions to be traced. And this impacts the normal function tracer too. arch/x86/kernel/process_64/32.c: I had crashes when I let this file traced. After some debugging, I saw that the "current" task point was changed inside__swtich_to(), ie: "write_pda(pcurrent, next_p);" inside process_64.c Since the tracer store the original return address of the function inside current, we had crashes. Only __switch_to() has to be excluded from tracing. kernel/module.c and kernel/extable.c: Because of a function used internally by the function graph tracer: __kernel_text_address() To let the other functions inside these files to be traced, this patch introduces the __notrace_funcgraph function prefix which is __notrace if function graph tracer is configured and nothing if not. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 6 ------ arch/x86/kernel/process_32.c | 4 +++- arch/x86/kernel/process_64.c | 4 +++- 3 files changed, 6 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index a3049da61985..1cad9318d217 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -14,12 +14,6 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg endif -ifdef CONFIG_FUNCTION_GRAPH_TRACER -# Don't trace __switch_to() but let it for function tracer -CFLAGS_REMOVE_process_32.o = -pg -CFLAGS_REMOVE_process_64.o = -pg -endif - # # vsyscalls (which work on the user stack) should have # no stack-protector checks: diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 0a1302fe6d45..24c2276aa453 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -38,6 +38,7 @@ #include #include #include +#include #include #include @@ -548,7 +549,8 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, * the task-switch, and shows up in ret_from_fork in entry.S, * for example. */ -struct task_struct * __switch_to(struct task_struct *prev_p, struct task_struct *next_p) +__notrace_funcgraph struct task_struct * +__switch_to(struct task_struct *prev_p, struct task_struct *next_p) { struct thread_struct *prev = &prev_p->thread, *next = &next_p->thread; diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index c958120fb1b6..fbb321d53d34 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -39,6 +39,7 @@ #include #include #include +#include #include #include @@ -551,8 +552,9 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, * - could test fs/gs bitsliced * * Kprobes not supported here. Set the probe on schedule instead. + * Function graph tracer not supported too. */ -struct task_struct * +__notrace_funcgraph struct task_struct * __switch_to(struct task_struct *prev_p, struct task_struct *next_p) { struct thread_struct *prev = &prev_p->thread; -- cgit v1.2.3 From 380c4b1411ccd6885f92b2c8ceb08433a720f44e Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 6 Dec 2008 03:43:41 +0100 Subject: tracing/function-graph-tracer: append the tracing_graph_flag Impact: Provide a way to pause the function graph tracer As suggested by Steven Rostedt, the previous patch that prevented from spinlock function tracing shouldn't use the raw_spinlock to fix it. It's much better to follow lockdep with normal spinlock, so this patch adds a new flag for each task to make the function graph tracer able to be paused. We also can send an ftrace_printk whithout worrying of the irrelevant traced spinlock during insertion. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index f98c4076a170..1b43086b097a 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -476,7 +476,10 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) &return_to_handler; /* Nmi's are currently unsupported */ - if (atomic_read(&in_nmi)) + if (unlikely(atomic_read(&in_nmi))) + return; + + if (unlikely(atomic_read(¤t->tracing_graph_pause))) return; /* -- cgit v1.2.3 From 4217458dafaa57d8e26a46f5d05ab8c53cf64191 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 5 Dec 2008 17:17:09 -0800 Subject: x86: signal: change type of paramter for sys_rt_sigreturn() Impact: cleanup on 32-bit Peter pointed this parameter can be changed. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index b1f4d34e0a38..b1cc6da64208 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -642,11 +642,9 @@ badframe: } #ifdef CONFIG_X86_32 -asmlinkage int sys_rt_sigreturn(unsigned long __unused) +asmlinkage int sys_rt_sigreturn(struct pt_regs regs) { - struct pt_regs *regs = (struct pt_regs *)&__unused; - - return do_rt_sigreturn(regs); + return do_rt_sigreturn(®s); } #else /* !CONFIG_X86_32 */ asmlinkage long sys_rt_sigreturn(struct pt_regs *regs) -- cgit v1.2.3 From 087052b02f42b50316c6e4d7f2d8dfba3de6fc2e Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 17 Oct 2008 16:09:57 +0200 Subject: x86: fix default_spin_lock_flags() prototype MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit these warnings: arch/x86/kernel/paravirt-spinlocks.c: In function ‘default_spin_lock_flags’: arch/x86/kernel/paravirt-spinlocks.c:12: warning: passing argument 1 of ‘__raw_spin_lock’ from incompatible pointer type arch/x86/kernel/paravirt-spinlocks.c: At top level: arch/x86/kernel/paravirt-spinlocks.c:11: warning: ‘default_spin_lock_flags’ defined but not used showed that the prototype of default_spin_lock_flags() was confused about what type spinlocks have. the proper type on UP is raw_spinlock_t. Signed-off-by: Ingo Molnar --- arch/x86/kernel/paravirt-spinlocks.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/paravirt-spinlocks.c b/arch/x86/kernel/paravirt-spinlocks.c index 0e9f1982b1dd..95777b0faa73 100644 --- a/arch/x86/kernel/paravirt-spinlocks.c +++ b/arch/x86/kernel/paravirt-spinlocks.c @@ -7,7 +7,8 @@ #include -static void default_spin_lock_flags(struct raw_spinlock *lock, unsigned long flags) +static inline void +default_spin_lock_flags(raw_spinlock_t *lock, unsigned long flags) { __raw_spin_lock(lock); } -- cgit v1.2.3 From b0884e25fe361f2ca228808fb5fd1b74cb04e711 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Thu, 11 Dec 2008 13:45:23 +0100 Subject: x86, bts: turn BUG_ON into WARN_ON_ONCE Impact: make the ds code more debuggable Turn BUG_ON's into WARN_ON_ONCE. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 4 ++-- arch/x86/kernel/ptrace.c | 5 +++-- 2 files changed, 5 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 19a8c2c0389f..095306988667 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -452,7 +452,7 @@ struct pebs_tracer *ds_request_pebs(struct task_struct *task, static void ds_release(struct ds_tracer *tracer, enum ds_qualifier qual) { - BUG_ON(tracer->context->owner[qual] != tracer); + WARN_ON_ONCE(tracer->context->owner[qual] != tracer); tracer->context->owner[qual] = NULL; put_tracer(tracer->context->task); @@ -774,7 +774,7 @@ ds_configure(const struct ds_configuration *cfg) printk(KERN_INFO "DS available\n"); - BUG_ON(MAX_SIZEOF_DS < ds_cfg.sizeof_ds); + WARN_ON_ONCE(MAX_SIZEOF_DS < ds_cfg.sizeof_ds); } void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 2c8ec1ba75e6..b2998fe1166b 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -878,7 +878,8 @@ static int ptrace_bts_write_record(struct task_struct *child, { unsigned char bts_record[BTS_MAX_RECORD_SIZE]; - BUG_ON(BTS_MAX_RECORD_SIZE < bts_cfg.sizeof_bts); + if (BTS_MAX_RECORD_SIZE < bts_cfg.sizeof_bts) + return -EOVERFLOW; memset(bts_record, 0, bts_cfg.sizeof_bts); switch (in->qualifier) { @@ -1133,7 +1134,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) ret = ds_get_bts_index(child->bts, &size); if (ret == 0) { - BUG_ON(size != (int) size); + WARN_ON_ONCE(size != (int) size); ret = (int) size; } break; -- cgit v1.2.3 From c2724775ce57c98b8af9694857b941dc61056516 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Thu, 11 Dec 2008 13:49:59 +0100 Subject: x86, bts: provide in-kernel branch-trace interface Impact: cleanup Move the BTS bits from ptrace.c into ds.c. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/intel.c | 4 - arch/x86/kernel/ds.c | 857 ++++++++++++++++++++++++++----------------- arch/x86/kernel/process_32.c | 59 +-- arch/x86/kernel/process_64.c | 50 +-- arch/x86/kernel/ptrace.c | 416 ++++++--------------- 5 files changed, 657 insertions(+), 729 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 816f27f289b1..cd413d9a0218 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -11,7 +11,6 @@ #include #include #include -#include #include #include @@ -309,9 +308,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_P3); #endif - if (cpu_has_bts) - ptrace_bts_init_intel(c); - detect_extended_topology(c); if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) { /* diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 095306988667..f0583005b75e 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -6,13 +6,13 @@ * precise-event based sampling (PEBS). * * It manages: - * - per-thread and per-cpu allocation of BTS and PEBS + * - DS and BTS hardware configuration * - buffer overflow handling (to be done) * - buffer access * - * It assumes: - * - get_task_struct on all traced tasks - * - current is allowed to trace tasks + * It does not do: + * - security checking (is the caller allowed to trace the task) + * - buffer allocation (memory accounting) * * * Copyright (C) 2007-2008 Intel Corporation. @@ -34,15 +34,30 @@ * The configuration for a particular DS hardware implementation. */ struct ds_configuration { - /* the size of the DS structure in bytes */ - unsigned char sizeof_ds; - /* the size of one pointer-typed field in the DS structure in bytes; - this covers the first 8 fields related to buffer management. */ + /* the name of the configuration */ + const char *name; + /* the size of one pointer-typed field in the DS structure and + in the BTS and PEBS buffers in bytes; + this covers the first 8 DS fields related to buffer management. */ unsigned char sizeof_field; /* the size of a BTS/PEBS record in bytes */ unsigned char sizeof_rec[2]; + /* a series of bit-masks to control various features indexed + * by enum ds_feature */ + unsigned long ctl[dsf_ctl_max]; }; -static struct ds_configuration ds_cfg; +static DEFINE_PER_CPU(struct ds_configuration, ds_cfg_array); + +#define ds_cfg per_cpu(ds_cfg_array, smp_processor_id()) + +#define MAX_SIZEOF_DS (12 * 8) /* maximal size of a DS configuration */ +#define MAX_SIZEOF_BTS (3 * 8) /* maximal size of a BTS record */ +#define DS_ALIGNMENT (1 << 3) /* BTS and PEBS buffer alignment */ + +#define BTS_CONTROL \ + (ds_cfg.ctl[dsf_bts] | ds_cfg.ctl[dsf_bts_kernel] | ds_cfg.ctl[dsf_bts_user] |\ + ds_cfg.ctl[dsf_bts_overflow]) + /* * A BTS or PEBS tracer. @@ -61,6 +76,8 @@ struct ds_tracer { struct bts_tracer { /* the common DS part */ struct ds_tracer ds; + /* the trace including the DS configuration */ + struct bts_trace trace; /* buffer overflow notification function */ bts_ovfl_callback_t ovfl; }; @@ -68,6 +85,8 @@ struct bts_tracer { struct pebs_tracer { /* the common DS part */ struct ds_tracer ds; + /* the trace including the DS configuration */ + struct pebs_trace trace; /* buffer overflow notification function */ pebs_ovfl_callback_t ovfl; }; @@ -134,13 +153,11 @@ static inline void ds_set(unsigned char *base, enum ds_qualifier qual, (*(unsigned long *)base) = value; } -#define DS_ALIGNMENT (1 << 3) /* BTS and PEBS buffer alignment */ - /* * Locking is done only for allocating BTS or PEBS resources. */ -static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); +static DEFINE_SPINLOCK(ds_lock); /* @@ -156,27 +173,32 @@ static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); * >0 number of per-thread tracers * <0 number of per-cpu tracers * - * The below functions to get and put tracers and to check the - * allocation type require the ds_lock to be held by the caller. - * * Tracers essentially gives the number of ds contexts for a certain * type of allocation. */ -static long tracers; +static atomic_t tracers = ATOMIC_INIT(0); static inline void get_tracer(struct task_struct *task) { - tracers += (task ? 1 : -1); + if (task) + atomic_inc(&tracers); + else + atomic_dec(&tracers); } static inline void put_tracer(struct task_struct *task) { - tracers -= (task ? 1 : -1); + if (task) + atomic_dec(&tracers); + else + atomic_inc(&tracers); } static inline int check_tracer(struct task_struct *task) { - return (task ? (tracers >= 0) : (tracers <= 0)); + return task ? + (atomic_read(&tracers) >= 0) : + (atomic_read(&tracers) <= 0); } @@ -190,14 +212,30 @@ static inline int check_tracer(struct task_struct *task) * Contexts are use-counted. They are allocated on first access and * deallocated when the last user puts the context. */ -static DEFINE_PER_CPU(struct ds_context *, system_context); +struct ds_context { + /* pointer to the DS configuration; goes into MSR_IA32_DS_AREA */ + unsigned char ds[MAX_SIZEOF_DS]; + /* the owner of the BTS and PEBS configuration, respectively */ + struct bts_tracer *bts_master; + struct pebs_tracer *pebs_master; + /* use count */ + unsigned long count; + /* a pointer to the context location inside the thread_struct + * or the per_cpu context array */ + struct ds_context **this; + /* a pointer to the task owning this context, or NULL, if the + * context is owned by a cpu */ + struct task_struct *task; +}; + +static DEFINE_PER_CPU(struct ds_context *, system_context_array); -#define this_system_context per_cpu(system_context, smp_processor_id()) +#define system_context per_cpu(system_context_array, smp_processor_id()) static inline struct ds_context *ds_get_context(struct task_struct *task) { struct ds_context **p_context = - (task ? &task->thread.ds_ctx : &this_system_context); + (task ? &task->thread.ds_ctx : &system_context); struct ds_context *context = *p_context; unsigned long irq; @@ -225,10 +263,22 @@ static inline struct ds_context *ds_get_context(struct task_struct *task) wrmsrl(MSR_IA32_DS_AREA, (unsigned long)context->ds); } + + context->count++; + + spin_unlock_irqrestore(&ds_lock, irq); + } else { + spin_lock_irqsave(&ds_lock, irq); + + context = *p_context; + if (context) + context->count++; + spin_unlock_irqrestore(&ds_lock, irq); - } - context->count++; + if (!context) + context = ds_get_context(task); + } return context; } @@ -242,8 +292,10 @@ static inline void ds_put_context(struct ds_context *context) spin_lock_irqsave(&ds_lock, irq); - if (--context->count) - goto out; + if (--context->count) { + spin_unlock_irqrestore(&ds_lock, irq); + return; + } *(context->this) = NULL; @@ -253,14 +305,14 @@ static inline void ds_put_context(struct ds_context *context) if (!context->task || (context->task == current)) wrmsrl(MSR_IA32_DS_AREA, 0); - kfree(context); - out: spin_unlock_irqrestore(&ds_lock, irq); + + kfree(context); } /* - * Handle a buffer overflow + * Call the tracer's callback on a buffer overflow. * * context: the ds context * qual: the buffer type @@ -268,30 +320,244 @@ static inline void ds_put_context(struct ds_context *context) static void ds_overflow(struct ds_context *context, enum ds_qualifier qual) { switch (qual) { - case ds_bts: { - struct bts_tracer *tracer = - container_of(context->owner[qual], - struct bts_tracer, ds); - if (tracer->ovfl) - tracer->ovfl(tracer); - } + case ds_bts: + if (context->bts_master && + context->bts_master->ovfl) + context->bts_master->ovfl(context->bts_master); + break; + case ds_pebs: + if (context->pebs_master && + context->pebs_master->ovfl) + context->pebs_master->ovfl(context->pebs_master); break; - case ds_pebs: { - struct pebs_tracer *tracer = - container_of(context->owner[qual], - struct pebs_tracer, ds); - if (tracer->ovfl) - tracer->ovfl(tracer); } +} + + +/* + * Write raw data into the BTS or PEBS buffer. + * + * The remainder of any partially written record is zeroed out. + * + * context: the DS context + * qual: the buffer type + * record: the data to write + * size: the size of the data + */ +static int ds_write(struct ds_context *context, enum ds_qualifier qual, + const void *record, size_t size) +{ + int bytes_written = 0; + + if (!record) + return -EINVAL; + + while (size) { + unsigned long base, index, end, write_end, int_th; + unsigned long write_size, adj_write_size; + + /* + * write as much as possible without producing an + * overflow interrupt. + * + * interrupt_threshold must either be + * - bigger than absolute_maximum or + * - point to a record between buffer_base and absolute_maximum + * + * index points to a valid record. + */ + base = ds_get(context->ds, qual, ds_buffer_base); + index = ds_get(context->ds, qual, ds_index); + end = ds_get(context->ds, qual, ds_absolute_maximum); + int_th = ds_get(context->ds, qual, ds_interrupt_threshold); + + write_end = min(end, int_th); + + /* if we are already beyond the interrupt threshold, + * we fill the entire buffer */ + if (write_end <= index) + write_end = end; + + if (write_end <= index) + break; + + write_size = min((unsigned long) size, write_end - index); + memcpy((void *)index, record, write_size); + + record = (const char *)record + write_size; + size -= write_size; + bytes_written += write_size; + + adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; + adj_write_size *= ds_cfg.sizeof_rec[qual]; + + /* zero out trailing bytes */ + memset((char *)index + write_size, 0, + adj_write_size - write_size); + index += adj_write_size; + + if (index >= end) + index = base; + ds_set(context->ds, qual, ds_index, index); + + if (index >= int_th) + ds_overflow(context, qual); + } + + return bytes_written; +} + + +/* + * Branch Trace Store (BTS) uses the following format. Different + * architectures vary in the size of those fields. + * - source linear address + * - destination linear address + * - flags + * + * Later architectures use 64bit pointers throughout, whereas earlier + * architectures use 32bit pointers in 32bit mode. + * + * We compute the base address for the first 8 fields based on: + * - the field size stored in the DS configuration + * - the relative field position + * + * In order to store additional information in the BTS buffer, we use + * a special source address to indicate that the record requires + * special interpretation. + * + * Netburst indicated via a bit in the flags field whether the branch + * was predicted; this is ignored. + * + * We use two levels of abstraction: + * - the raw data level defined here + * - an arch-independent level defined in ds.h + */ + +enum bts_field { + bts_from, + bts_to, + bts_flags, + + bts_qual = bts_from, + bts_jiffies = bts_to, + bts_pid = bts_flags, + + bts_qual_mask = (bts_qual_max - 1), + bts_escape = ((unsigned long)-1 & ~bts_qual_mask) +}; + +static inline unsigned long bts_get(const char *base, enum bts_field field) +{ + base += (ds_cfg.sizeof_field * field); + return *(unsigned long *)base; +} + +static inline void bts_set(char *base, enum bts_field field, unsigned long val) +{ + base += (ds_cfg.sizeof_field * field);; + (*(unsigned long *)base) = val; +} + + +/* + * The raw BTS data is architecture dependent. + * + * For higher-level users, we give an arch-independent view. + * - ds.h defines struct bts_struct + * - bts_read translates one raw bts record into a bts_struct + * - bts_write translates one bts_struct into the raw format and + * writes it into the top of the parameter tracer's buffer. + * + * return: bytes read/written on success; -Eerrno, otherwise + */ +static int bts_read(struct bts_tracer *tracer, const void *at, + struct bts_struct *out) +{ + if (!tracer) + return -EINVAL; + + if (at < tracer->trace.ds.begin) + return -EINVAL; + + if (tracer->trace.ds.end < (at + tracer->trace.ds.size)) + return -EINVAL; + + memset(out, 0, sizeof(*out)); + if ((bts_get(at, bts_qual) & ~bts_qual_mask) == bts_escape) { + out->qualifier = (bts_get(at, bts_qual) & bts_qual_mask); + out->variant.timestamp.jiffies = bts_get(at, bts_jiffies); + out->variant.timestamp.pid = bts_get(at, bts_pid); + } else { + out->qualifier = bts_branch; + out->variant.lbr.from = bts_get(at, bts_from); + out->variant.lbr.to = bts_get(at, bts_to); + } + + return ds_cfg.sizeof_rec[ds_bts]; +} + +static int bts_write(struct bts_tracer *tracer, const struct bts_struct *in) +{ + unsigned char raw[MAX_SIZEOF_BTS]; + + if (!tracer) + return -EINVAL; + + if (MAX_SIZEOF_BTS < ds_cfg.sizeof_rec[ds_bts]) + return -EOVERFLOW; + + switch (in->qualifier) { + case bts_invalid: + bts_set(raw, bts_from, 0); + bts_set(raw, bts_to, 0); + bts_set(raw, bts_flags, 0); + break; + case bts_branch: + bts_set(raw, bts_from, in->variant.lbr.from); + bts_set(raw, bts_to, in->variant.lbr.to); + bts_set(raw, bts_flags, 0); + break; + case bts_task_arrives: + case bts_task_departs: + bts_set(raw, bts_qual, (bts_escape | in->qualifier)); + bts_set(raw, bts_jiffies, in->variant.timestamp.jiffies); + bts_set(raw, bts_pid, in->variant.timestamp.pid); break; + default: + return -EINVAL; } + + return ds_write(tracer->ds.context, ds_bts, raw, + ds_cfg.sizeof_rec[ds_bts]); } -static void ds_install_ds_config(struct ds_context *context, - enum ds_qualifier qual, - void *base, size_t size, size_t ith) +static void ds_write_config(struct ds_context *context, + struct ds_trace *cfg, enum ds_qualifier qual) +{ + unsigned char *ds = context->ds; + + ds_set(ds, qual, ds_buffer_base, (unsigned long)cfg->begin); + ds_set(ds, qual, ds_index, (unsigned long)cfg->top); + ds_set(ds, qual, ds_absolute_maximum, (unsigned long)cfg->end); + ds_set(ds, qual, ds_interrupt_threshold, (unsigned long)cfg->ith); +} + +static void ds_read_config(struct ds_context *context, + struct ds_trace *cfg, enum ds_qualifier qual) { + unsigned char *ds = context->ds; + + cfg->begin = (void *)ds_get(ds, qual, ds_buffer_base); + cfg->top = (void *)ds_get(ds, qual, ds_index); + cfg->end = (void *)ds_get(ds, qual, ds_absolute_maximum); + cfg->ith = (void *)ds_get(ds, qual, ds_interrupt_threshold); +} + +static void ds_init_ds_trace(struct ds_trace *trace, enum ds_qualifier qual, + void *base, size_t size, size_t ith, + unsigned int flags) { unsigned long buffer, adj; /* adjust the buffer address and size to meet alignment @@ -308,32 +574,30 @@ static void ds_install_ds_config(struct ds_context *context, buffer += adj; size -= adj; - size /= ds_cfg.sizeof_rec[qual]; - size *= ds_cfg.sizeof_rec[qual]; + trace->n = size / ds_cfg.sizeof_rec[qual]; + trace->size = ds_cfg.sizeof_rec[qual]; - ds_set(context->ds, qual, ds_buffer_base, buffer); - ds_set(context->ds, qual, ds_index, buffer); - ds_set(context->ds, qual, ds_absolute_maximum, buffer + size); + size = (trace->n * trace->size); + trace->begin = (void *)buffer; + trace->top = trace->begin; + trace->end = (void *)(buffer + size); /* The value for 'no threshold' is -1, which will set the * threshold outside of the buffer, just like we want it. */ - ds_set(context->ds, qual, - ds_interrupt_threshold, buffer + size - ith); + trace->ith = (void *)(buffer + size - ith); + + trace->flags = flags; } -static int ds_request(struct ds_tracer *tracer, enum ds_qualifier qual, - struct task_struct *task, - void *base, size_t size, size_t th) + +static int ds_request(struct ds_tracer *tracer, struct ds_trace *trace, + enum ds_qualifier qual, struct task_struct *task, + void *base, size_t size, size_t th, unsigned int flags) { struct ds_context *context; - unsigned long irq; int error; - error = -EOPNOTSUPP; - if (!ds_cfg.sizeof_ds) - goto out; - error = -EINVAL; if (!base) goto out; @@ -360,43 +624,26 @@ static int ds_request(struct ds_tracer *tracer, enum ds_qualifier qual, goto out; tracer->context = context; + ds_init_ds_trace(trace, qual, base, size, th, flags); - spin_lock_irqsave(&ds_lock, irq); - - error = -EPERM; - if (!check_tracer(task)) - goto out_unlock; - get_tracer(task); - - error = -EPERM; - if (context->owner[qual]) - goto out_put_tracer; - context->owner[qual] = tracer; - - spin_unlock_irqrestore(&ds_lock, irq); - - - ds_install_ds_config(context, qual, base, size, th); - - return 0; - - out_put_tracer: - put_tracer(task); - out_unlock: - spin_unlock_irqrestore(&ds_lock, irq); - ds_put_context(context); - tracer->context = NULL; + error = 0; out: return error; } struct bts_tracer *ds_request_bts(struct task_struct *task, void *base, size_t size, - bts_ovfl_callback_t ovfl, size_t th) + bts_ovfl_callback_t ovfl, size_t th, + unsigned int flags) { struct bts_tracer *tracer; + unsigned long irq; int error; + error = -EOPNOTSUPP; + if (!ds_cfg.ctl[dsf_bts]) + goto out; + /* buffer overflow notification is not yet implemented */ error = -EOPNOTSUPP; if (ovfl) @@ -408,12 +655,40 @@ struct bts_tracer *ds_request_bts(struct task_struct *task, goto out; tracer->ovfl = ovfl; - error = ds_request(&tracer->ds, ds_bts, task, base, size, th); + error = ds_request(&tracer->ds, &tracer->trace.ds, + ds_bts, task, base, size, th, flags); if (error < 0) goto out_tracer; + + spin_lock_irqsave(&ds_lock, irq); + + error = -EPERM; + if (!check_tracer(task)) + goto out_unlock; + get_tracer(task); + + error = -EPERM; + if (tracer->ds.context->bts_master) + goto out_put_tracer; + tracer->ds.context->bts_master = tracer; + + spin_unlock_irqrestore(&ds_lock, irq); + + + tracer->trace.read = bts_read; + tracer->trace.write = bts_write; + + ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + ds_resume_bts(tracer); + return tracer; + out_put_tracer: + put_tracer(task); + out_unlock: + spin_unlock_irqrestore(&ds_lock, irq); + ds_put_context(tracer->ds.context); out_tracer: kfree(tracer); out: @@ -422,9 +697,11 @@ struct bts_tracer *ds_request_bts(struct task_struct *task, struct pebs_tracer *ds_request_pebs(struct task_struct *task, void *base, size_t size, - pebs_ovfl_callback_t ovfl, size_t th) + pebs_ovfl_callback_t ovfl, size_t th, + unsigned int flags) { struct pebs_tracer *tracer; + unsigned long irq; int error; /* buffer overflow notification is not yet implemented */ @@ -438,300 +715,171 @@ struct pebs_tracer *ds_request_pebs(struct task_struct *task, goto out; tracer->ovfl = ovfl; - error = ds_request(&tracer->ds, ds_pebs, task, base, size, th); + error = ds_request(&tracer->ds, &tracer->trace.ds, + ds_pebs, task, base, size, th, flags); if (error < 0) goto out_tracer; + spin_lock_irqsave(&ds_lock, irq); + + error = -EPERM; + if (!check_tracer(task)) + goto out_unlock; + get_tracer(task); + + error = -EPERM; + if (tracer->ds.context->pebs_master) + goto out_put_tracer; + tracer->ds.context->pebs_master = tracer; + + spin_unlock_irqrestore(&ds_lock, irq); + + ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + ds_resume_pebs(tracer); + return tracer; + out_put_tracer: + put_tracer(task); + out_unlock: + spin_unlock_irqrestore(&ds_lock, irq); + ds_put_context(tracer->ds.context); out_tracer: kfree(tracer); out: return ERR_PTR(error); } -static void ds_release(struct ds_tracer *tracer, enum ds_qualifier qual) -{ - WARN_ON_ONCE(tracer->context->owner[qual] != tracer); - tracer->context->owner[qual] = NULL; - - put_tracer(tracer->context->task); - ds_put_context(tracer->context); -} - -int ds_release_bts(struct bts_tracer *tracer) +void ds_release_bts(struct bts_tracer *tracer) { if (!tracer) - return -EINVAL; + return; - ds_release(&tracer->ds, ds_bts); - kfree(tracer); + ds_suspend_bts(tracer); - return 0; -} + WARN_ON_ONCE(tracer->ds.context->bts_master != tracer); + tracer->ds.context->bts_master = NULL; -int ds_release_pebs(struct pebs_tracer *tracer) -{ - if (!tracer) - return -EINVAL; + put_tracer(tracer->ds.context->task); + ds_put_context(tracer->ds.context); - ds_release(&tracer->ds, ds_pebs); kfree(tracer); - - return 0; -} - -static size_t ds_get_index(struct ds_context *context, enum ds_qualifier qual) -{ - unsigned long base, index; - - base = ds_get(context->ds, qual, ds_buffer_base); - index = ds_get(context->ds, qual, ds_index); - - return (index - base) / ds_cfg.sizeof_rec[qual]; } -int ds_get_bts_index(struct bts_tracer *tracer, size_t *pos) +void ds_suspend_bts(struct bts_tracer *tracer) { - if (!tracer) - return -EINVAL; + struct task_struct *task; - if (!pos) - return -EINVAL; - - *pos = ds_get_index(tracer->ds.context, ds_bts); - - return 0; -} - -int ds_get_pebs_index(struct pebs_tracer *tracer, size_t *pos) -{ if (!tracer) - return -EINVAL; + return; - if (!pos) - return -EINVAL; + task = tracer->ds.context->task; - *pos = ds_get_index(tracer->ds.context, ds_pebs); + if (!task || (task == current)) + update_debugctlmsr(get_debugctlmsr() & ~BTS_CONTROL); - return 0; -} + if (task) { + task->thread.debugctlmsr &= ~BTS_CONTROL; -static size_t ds_get_end(struct ds_context *context, enum ds_qualifier qual) -{ - unsigned long base, max; - - base = ds_get(context->ds, qual, ds_buffer_base); - max = ds_get(context->ds, qual, ds_absolute_maximum); - - return (max - base) / ds_cfg.sizeof_rec[qual]; + if (!task->thread.debugctlmsr) + clear_tsk_thread_flag(task, TIF_DEBUGCTLMSR); + } } -int ds_get_bts_end(struct bts_tracer *tracer, size_t *pos) +void ds_resume_bts(struct bts_tracer *tracer) { - if (!tracer) - return -EINVAL; - - if (!pos) - return -EINVAL; - - *pos = ds_get_end(tracer->ds.context, ds_bts); - - return 0; -} + struct task_struct *task; + unsigned long control; -int ds_get_pebs_end(struct pebs_tracer *tracer, size_t *pos) -{ if (!tracer) - return -EINVAL; - - if (!pos) - return -EINVAL; - - *pos = ds_get_end(tracer->ds.context, ds_pebs); - - return 0; -} - -static int ds_access(struct ds_context *context, enum ds_qualifier qual, - size_t index, const void **record) -{ - unsigned long base, idx; - - if (!record) - return -EINVAL; - - base = ds_get(context->ds, qual, ds_buffer_base); - idx = base + (index * ds_cfg.sizeof_rec[qual]); - - if (idx > ds_get(context->ds, qual, ds_absolute_maximum)) - return -EINVAL; + return; - *record = (const void *)idx; + task = tracer->ds.context->task; - return ds_cfg.sizeof_rec[qual]; -} + control = ds_cfg.ctl[dsf_bts]; + if (!(tracer->trace.ds.flags & BTS_KERNEL)) + control |= ds_cfg.ctl[dsf_bts_kernel]; + if (!(tracer->trace.ds.flags & BTS_USER)) + control |= ds_cfg.ctl[dsf_bts_user]; -int ds_access_bts(struct bts_tracer *tracer, size_t index, - const void **record) -{ - if (!tracer) - return -EINVAL; + if (task) { + task->thread.debugctlmsr |= control; + set_tsk_thread_flag(task, TIF_DEBUGCTLMSR); + } - return ds_access(tracer->ds.context, ds_bts, index, record); + if (!task || (task == current)) + update_debugctlmsr(get_debugctlmsr() | control); } -int ds_access_pebs(struct pebs_tracer *tracer, size_t index, - const void **record) +void ds_release_pebs(struct pebs_tracer *tracer) { if (!tracer) - return -EINVAL; - - return ds_access(tracer->ds.context, ds_pebs, index, record); -} - -static int ds_write(struct ds_context *context, enum ds_qualifier qual, - const void *record, size_t size) -{ - int bytes_written = 0; - - if (!record) - return -EINVAL; - - while (size) { - unsigned long base, index, end, write_end, int_th; - unsigned long write_size, adj_write_size; - - /* - * write as much as possible without producing an - * overflow interrupt. - * - * interrupt_threshold must either be - * - bigger than absolute_maximum or - * - point to a record between buffer_base and absolute_maximum - * - * index points to a valid record. - */ - base = ds_get(context->ds, qual, ds_buffer_base); - index = ds_get(context->ds, qual, ds_index); - end = ds_get(context->ds, qual, ds_absolute_maximum); - int_th = ds_get(context->ds, qual, ds_interrupt_threshold); - - write_end = min(end, int_th); - - /* if we are already beyond the interrupt threshold, - * we fill the entire buffer */ - if (write_end <= index) - write_end = end; - - if (write_end <= index) - break; - - write_size = min((unsigned long) size, write_end - index); - memcpy((void *)index, record, write_size); - - record = (const char *)record + write_size; - size -= write_size; - bytes_written += write_size; - - adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; - adj_write_size *= ds_cfg.sizeof_rec[qual]; - - /* zero out trailing bytes */ - memset((char *)index + write_size, 0, - adj_write_size - write_size); - index += adj_write_size; + return; - if (index >= end) - index = base; - ds_set(context->ds, qual, ds_index, index); + ds_suspend_pebs(tracer); - if (index >= int_th) - ds_overflow(context, qual); - } + WARN_ON_ONCE(tracer->ds.context->pebs_master != tracer); + tracer->ds.context->pebs_master = NULL; - return bytes_written; -} + put_tracer(tracer->ds.context->task); + ds_put_context(tracer->ds.context); -int ds_write_bts(struct bts_tracer *tracer, const void *record, size_t size) -{ - if (!tracer) - return -EINVAL; - - return ds_write(tracer->ds.context, ds_bts, record, size); + kfree(tracer); } -int ds_write_pebs(struct pebs_tracer *tracer, const void *record, size_t size) +void ds_suspend_pebs(struct pebs_tracer *tracer) { - if (!tracer) - return -EINVAL; - return ds_write(tracer->ds.context, ds_pebs, record, size); } -static void ds_reset_or_clear(struct ds_context *context, - enum ds_qualifier qual, int clear) +void ds_resume_pebs(struct pebs_tracer *tracer) { - unsigned long base, end; - - base = ds_get(context->ds, qual, ds_buffer_base); - end = ds_get(context->ds, qual, ds_absolute_maximum); - - if (clear) - memset((void *)base, 0, end - base); - ds_set(context->ds, qual, ds_index, base); } -int ds_reset_bts(struct bts_tracer *tracer) +const struct bts_trace *ds_read_bts(struct bts_tracer *tracer) { if (!tracer) - return -EINVAL; - - ds_reset_or_clear(tracer->ds.context, ds_bts, /* clear = */ 0); + return NULL; - return 0; + ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + return &tracer->trace; } -int ds_reset_pebs(struct pebs_tracer *tracer) +const struct pebs_trace *ds_read_pebs(struct pebs_tracer *tracer) { if (!tracer) - return -EINVAL; + return NULL; - ds_reset_or_clear(tracer->ds.context, ds_pebs, /* clear = */ 0); + ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_pebs); + tracer->trace.reset_value = + *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)); - return 0; + return &tracer->trace; } -int ds_clear_bts(struct bts_tracer *tracer) +int ds_reset_bts(struct bts_tracer *tracer) { if (!tracer) return -EINVAL; - ds_reset_or_clear(tracer->ds.context, ds_bts, /* clear = */ 1); - - return 0; -} - -int ds_clear_pebs(struct pebs_tracer *tracer) -{ - if (!tracer) - return -EINVAL; + tracer->trace.ds.top = tracer->trace.ds.begin; - ds_reset_or_clear(tracer->ds.context, ds_pebs, /* clear = */ 1); + ds_set(tracer->ds.context->ds, ds_bts, ds_index, + (unsigned long)tracer->trace.ds.top); return 0; } -int ds_get_pebs_reset(struct pebs_tracer *tracer, u64 *value) +int ds_reset_pebs(struct pebs_tracer *tracer) { if (!tracer) return -EINVAL; - if (!value) - return -EINVAL; + tracer->trace.ds.top = tracer->trace.ds.begin; - *value = *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)); + ds_set(tracer->ds.context->ds, ds_bts, ds_index, + (unsigned long)tracer->trace.ds.top); return 0; } @@ -746,35 +894,59 @@ int ds_set_pebs_reset(struct pebs_tracer *tracer, u64 value) return 0; } -static const struct ds_configuration ds_cfg_var = { - .sizeof_ds = sizeof(long) * 12, - .sizeof_field = sizeof(long), - .sizeof_rec[ds_bts] = sizeof(long) * 3, +static const struct ds_configuration ds_cfg_netburst = { + .name = "netburst", + .ctl[dsf_bts] = (1 << 2) | (1 << 3), + .ctl[dsf_bts_kernel] = (1 << 5), + .ctl[dsf_bts_user] = (1 << 6), + + .sizeof_field = sizeof(long), + .sizeof_rec[ds_bts] = sizeof(long) * 3, #ifdef __i386__ - .sizeof_rec[ds_pebs] = sizeof(long) * 10 + .sizeof_rec[ds_pebs] = sizeof(long) * 10, #else - .sizeof_rec[ds_pebs] = sizeof(long) * 18 + .sizeof_rec[ds_pebs] = sizeof(long) * 18, #endif }; -static const struct ds_configuration ds_cfg_64 = { - .sizeof_ds = 8 * 12, - .sizeof_field = 8, - .sizeof_rec[ds_bts] = 8 * 3, +static const struct ds_configuration ds_cfg_pentium_m = { + .name = "pentium m", + .ctl[dsf_bts] = (1 << 6) | (1 << 7), + + .sizeof_field = sizeof(long), + .sizeof_rec[ds_bts] = sizeof(long) * 3, #ifdef __i386__ - .sizeof_rec[ds_pebs] = 8 * 10 + .sizeof_rec[ds_pebs] = sizeof(long) * 10, #else - .sizeof_rec[ds_pebs] = 8 * 18 + .sizeof_rec[ds_pebs] = sizeof(long) * 18, #endif }; +static const struct ds_configuration ds_cfg_core2 = { + .name = "core 2", + .ctl[dsf_bts] = (1 << 6) | (1 << 7), + .ctl[dsf_bts_kernel] = (1 << 9), + .ctl[dsf_bts_user] = (1 << 10), + + .sizeof_field = 8, + .sizeof_rec[ds_bts] = 8 * 3, + .sizeof_rec[ds_pebs] = 8 * 18, +}; -static inline void +static void ds_configure(const struct ds_configuration *cfg) { + memset(&ds_cfg, 0, sizeof(ds_cfg)); ds_cfg = *cfg; - printk(KERN_INFO "DS available\n"); + printk(KERN_INFO "[ds] using %s configuration\n", ds_cfg.name); + + if (!cpu_has_bts) { + ds_cfg.ctl[dsf_bts] = 0; + printk(KERN_INFO "[ds] bts not available\n"); + } + if (!cpu_has_pebs) + printk(KERN_INFO "[ds] pebs not available\n"); - WARN_ON_ONCE(MAX_SIZEOF_DS < ds_cfg.sizeof_ds); + WARN_ON_ONCE(MAX_SIZEOF_DS < (12 * ds_cfg.sizeof_field)); } void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) @@ -787,10 +959,10 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) break; case 0xD: case 0xE: /* Pentium M */ - ds_configure(&ds_cfg_var); + ds_configure(&ds_cfg_pentium_m); break; default: /* Core2, Atom, ... */ - ds_configure(&ds_cfg_64); + ds_configure(&ds_cfg_core2); break; } break; @@ -799,7 +971,7 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) case 0x0: case 0x1: case 0x2: /* Netburst */ - ds_configure(&ds_cfg_var); + ds_configure(&ds_cfg_netburst); break; default: /* sorry, don't know about them */ @@ -812,14 +984,41 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) } } -void ds_free(struct ds_context *context) +/* + * Change the DS configuration from tracing prev to tracing next. + */ +void ds_switch_to(struct task_struct *prev, struct task_struct *next) { - /* This is called when the task owning the parameter context - * is dying. There should not be any user of that context left - * to disturb us, anymore. */ - unsigned long leftovers = context->count; - while (leftovers--) { - put_tracer(context->task); - ds_put_context(context); + struct ds_context *prev_ctx = prev->thread.ds_ctx; + struct ds_context *next_ctx = next->thread.ds_ctx; + + if (prev_ctx) { + update_debugctlmsr(0); + + if (prev_ctx->bts_master && + (prev_ctx->bts_master->trace.ds.flags & BTS_TIMESTAMPS)) { + struct bts_struct ts = { + .qualifier = bts_task_departs, + .variant.timestamp.jiffies = jiffies_64, + .variant.timestamp.pid = prev->pid + }; + bts_write(prev_ctx->bts_master, &ts); + } + } + + if (next_ctx) { + if (next_ctx->bts_master && + (next_ctx->bts_master->trace.ds.flags & BTS_TIMESTAMPS)) { + struct bts_struct ts = { + .qualifier = bts_task_arrives, + .variant.timestamp.jiffies = jiffies_64, + .variant.timestamp.pid = next->pid + }; + bts_write(next_ctx->bts_master, &ts); + } + + wrmsrl(MSR_IA32_DS_AREA, (unsigned long)next_ctx->ds); } + + update_debugctlmsr(next->thread.debugctlmsr); } diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 24c2276aa453..605eff9a8ac0 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -252,11 +252,14 @@ void exit_thread(void) put_cpu(); } #ifdef CONFIG_X86_DS - /* Free any DS contexts that have not been properly released. */ - if (unlikely(current->thread.ds_ctx)) { - /* we clear debugctl to make sure DS is not used. */ - update_debugctlmsr(0); - ds_free(current->thread.ds_ctx); + /* Free any BTS tracers that have not been properly released. */ + if (unlikely(current->bts)) { + ds_release_bts(current->bts); + current->bts = NULL; + + kfree(current->bts_buffer); + current->bts_buffer = NULL; + current->bts_size = 0; } #endif /* CONFIG_X86_DS */ } @@ -420,48 +423,19 @@ int set_tsc_mode(unsigned int val) return 0; } -#ifdef CONFIG_X86_DS -static int update_debugctl(struct thread_struct *prev, - struct thread_struct *next, unsigned long debugctl) -{ - unsigned long ds_prev = 0; - unsigned long ds_next = 0; - - if (prev->ds_ctx) - ds_prev = (unsigned long)prev->ds_ctx->ds; - if (next->ds_ctx) - ds_next = (unsigned long)next->ds_ctx->ds; - - if (ds_next != ds_prev) { - /* we clear debugctl to make sure DS - * is not in use when we change it */ - debugctl = 0; - update_debugctlmsr(0); - wrmsr(MSR_IA32_DS_AREA, ds_next, 0); - } - return debugctl; -} -#else -static int update_debugctl(struct thread_struct *prev, - struct thread_struct *next, unsigned long debugctl) -{ - return debugctl; -} -#endif /* CONFIG_X86_DS */ - static noinline void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, struct tss_struct *tss) { struct thread_struct *prev, *next; - unsigned long debugctl; prev = &prev_p->thread; next = &next_p->thread; - debugctl = update_debugctl(prev, next, prev->debugctlmsr); - - if (next->debugctlmsr != debugctl) + if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) || + test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR)) + ds_switch_to(prev_p, next_p); + else if (next->debugctlmsr != prev->debugctlmsr) update_debugctlmsr(next->debugctlmsr); if (test_tsk_thread_flag(next_p, TIF_DEBUG)) { @@ -483,15 +457,6 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, hard_enable_TSC(); } -#ifdef CONFIG_X86_PTRACE_BTS - if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); - - if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif /* CONFIG_X86_PTRACE_BTS */ - - if (!test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) { /* * Disable the bitmap via an invalid offset. We still cache diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index fbb321d53d34..1cfd2a4bf853 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -237,11 +237,14 @@ void exit_thread(void) put_cpu(); } #ifdef CONFIG_X86_DS - /* Free any DS contexts that have not been properly released. */ - if (unlikely(t->ds_ctx)) { - /* we clear debugctl to make sure DS is not used. */ - update_debugctlmsr(0); - ds_free(t->ds_ctx); + /* Free any BTS tracers that have not been properly released. */ + if (unlikely(current->bts)) { + ds_release_bts(current->bts); + current->bts = NULL; + + kfree(current->bts_buffer); + current->bts_buffer = NULL; + current->bts_size = 0; } #endif /* CONFIG_X86_DS */ } @@ -471,35 +474,14 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, struct tss_struct *tss) { struct thread_struct *prev, *next; - unsigned long debugctl; prev = &prev_p->thread, next = &next_p->thread; - debugctl = prev->debugctlmsr; - -#ifdef CONFIG_X86_DS - { - unsigned long ds_prev = 0, ds_next = 0; - - if (prev->ds_ctx) - ds_prev = (unsigned long)prev->ds_ctx->ds; - if (next->ds_ctx) - ds_next = (unsigned long)next->ds_ctx->ds; - - if (ds_next != ds_prev) { - /* - * We clear debugctl to make sure DS - * is not in use when we change it: - */ - debugctl = 0; - update_debugctlmsr(0); - wrmsrl(MSR_IA32_DS_AREA, ds_next); - } - } -#endif /* CONFIG_X86_DS */ - - if (next->debugctlmsr != debugctl) + if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) || + test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR)) + ds_switch_to(prev_p, next_p); + else if (next->debugctlmsr != prev->debugctlmsr) update_debugctlmsr(next->debugctlmsr); if (test_tsk_thread_flag(next_p, TIF_DEBUG)) { @@ -534,14 +516,6 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, */ memset(tss->io_bitmap, 0xff, prev->io_bitmap_max); } - -#ifdef CONFIG_X86_PTRACE_BTS - if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); - - if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif /* CONFIG_X86_PTRACE_BTS */ } /* diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index b2998fe1166b..45e9855da2d2 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -581,153 +581,73 @@ static int ioperm_get(struct task_struct *target, } #ifdef CONFIG_X86_PTRACE_BTS -/* - * The configuration for a particular BTS hardware implementation. - */ -struct bts_configuration { - /* the size of a BTS record in bytes; at most BTS_MAX_RECORD_SIZE */ - unsigned char sizeof_bts; - /* the size of a field in the BTS record in bytes */ - unsigned char sizeof_field; - /* a bitmask to enable/disable BTS in DEBUGCTL MSR */ - unsigned long debugctl_mask; -}; -static struct bts_configuration bts_cfg; - -#define BTS_MAX_RECORD_SIZE (8 * 3) - - -/* - * Branch Trace Store (BTS) uses the following format. Different - * architectures vary in the size of those fields. - * - source linear address - * - destination linear address - * - flags - * - * Later architectures use 64bit pointers throughout, whereas earlier - * architectures use 32bit pointers in 32bit mode. - * - * We compute the base address for the first 8 fields based on: - * - the field size stored in the DS configuration - * - the relative field position - * - * In order to store additional information in the BTS buffer, we use - * a special source address to indicate that the record requires - * special interpretation. - * - * Netburst indicated via a bit in the flags field whether the branch - * was predicted; this is ignored. - */ - -enum bts_field { - bts_from = 0, - bts_to, - bts_flags, - - bts_escape = (unsigned long)-1, - bts_qual = bts_to, - bts_jiffies = bts_flags -}; - -static inline unsigned long bts_get(const char *base, enum bts_field field) -{ - base += (bts_cfg.sizeof_field * field); - return *(unsigned long *)base; -} - -static inline void bts_set(char *base, enum bts_field field, unsigned long val) -{ - base += (bts_cfg.sizeof_field * field);; - (*(unsigned long *)base) = val; -} - -/* - * Translate a BTS record from the raw format into the bts_struct format - * - * out (out): bts_struct interpretation - * raw: raw BTS record - */ -static void ptrace_bts_translate_record(struct bts_struct *out, const void *raw) -{ - memset(out, 0, sizeof(*out)); - if (bts_get(raw, bts_from) == bts_escape) { - out->qualifier = bts_get(raw, bts_qual); - out->variant.jiffies = bts_get(raw, bts_jiffies); - } else { - out->qualifier = BTS_BRANCH; - out->variant.lbr.from_ip = bts_get(raw, bts_from); - out->variant.lbr.to_ip = bts_get(raw, bts_to); - } -} - static int ptrace_bts_read_record(struct task_struct *child, size_t index, struct bts_struct __user *out) { - struct bts_struct ret; - const void *bts_record; - size_t bts_index, bts_end; + const struct bts_trace *trace; + struct bts_struct bts; + const unsigned char *at; int error; - error = ds_get_bts_end(child->bts, &bts_end); - if (error < 0) - return error; - - if (bts_end <= index) - return -EINVAL; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - error = ds_get_bts_index(child->bts, &bts_index); - if (error < 0) - return error; + at = trace->ds.top - ((index + 1) * trace->ds.size); + if ((void *)at < trace->ds.begin) + at += (trace->ds.n * trace->ds.size); - /* translate the ptrace bts index into the ds bts index */ - bts_index += bts_end - (index + 1); - if (bts_end <= bts_index) - bts_index -= bts_end; + if (!trace->read) + return -EOPNOTSUPP; - error = ds_access_bts(child->bts, bts_index, &bts_record); + error = trace->read(child->bts, at, &bts); if (error < 0) return error; - ptrace_bts_translate_record(&ret, bts_record); - - if (copy_to_user(out, &ret, sizeof(ret))) + if (copy_to_user(out, &bts, sizeof(bts))) return -EFAULT; - return sizeof(ret); + return sizeof(bts); } static int ptrace_bts_drain(struct task_struct *child, long size, struct bts_struct __user *out) { - struct bts_struct ret; - const unsigned char *raw; - size_t end, i; - int error; + const struct bts_trace *trace; + const unsigned char *at; + int error, drained = 0; - error = ds_get_bts_index(child->bts, &end); - if (error < 0) - return error; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - if (size < (end * sizeof(struct bts_struct))) + if (!trace->read) + return -EOPNOTSUPP; + + if (size < (trace->ds.top - trace->ds.begin)) return -EIO; - error = ds_access_bts(child->bts, 0, (const void **)&raw); - if (error < 0) - return error; + for (at = trace->ds.begin; (void *)at < trace->ds.top; + out++, drained++, at += trace->ds.size) { + struct bts_struct bts; + int error; - for (i = 0; i < end; i++, out++, raw += bts_cfg.sizeof_bts) { - ptrace_bts_translate_record(&ret, raw); + error = trace->read(child->bts, at, &bts); + if (error < 0) + return error; - if (copy_to_user(out, &ret, sizeof(ret))) + if (copy_to_user(out, &bts, sizeof(bts))) return -EFAULT; } - error = ds_clear_bts(child->bts); + memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); + + error = ds_reset_bts(child->bts); if (error < 0) return error; - return end; + return drained; } static int ptrace_bts_config(struct task_struct *child, @@ -735,136 +655,89 @@ static int ptrace_bts_config(struct task_struct *child, const struct ptrace_bts_config __user *ucfg) { struct ptrace_bts_config cfg; - int error = 0; - - error = -EOPNOTSUPP; - if (!bts_cfg.sizeof_bts) - goto errout; + unsigned int flags = 0; - error = -EIO; if (cfg_size < sizeof(cfg)) - goto errout; + return -EIO; - error = -EFAULT; if (copy_from_user(&cfg, ucfg, sizeof(cfg))) - goto errout; - - error = -EINVAL; - if ((cfg.flags & PTRACE_BTS_O_SIGNAL) && - !(cfg.flags & PTRACE_BTS_O_ALLOC)) - goto errout; - - if (cfg.flags & PTRACE_BTS_O_ALLOC) { - bts_ovfl_callback_t ovfl = NULL; - unsigned int sig = 0; - - error = -EINVAL; - if (cfg.size < (10 * bts_cfg.sizeof_bts)) - goto errout; + return -EFAULT; - if (cfg.flags & PTRACE_BTS_O_SIGNAL) { - if (!cfg.signal) - goto errout; + if (child->bts) { + ds_release_bts(child->bts); + child->bts = NULL; + } - error = -EOPNOTSUPP; - goto errout; + if (cfg.flags & PTRACE_BTS_O_SIGNAL) { + if (!cfg.signal) + return -EINVAL; - sig = cfg.signal; - } + return -EOPNOTSUPP; - if (child->bts) { - (void)ds_release_bts(child->bts); - kfree(child->bts_buffer); + child->thread.bts_ovfl_signal = cfg.signal; + } - child->bts = NULL; - child->bts_buffer = NULL; - } + if ((cfg.flags & PTRACE_BTS_O_ALLOC) && + (cfg.size != child->bts_size)) { + kfree(child->bts_buffer); - error = -ENOMEM; + child->bts_size = cfg.size; child->bts_buffer = kzalloc(cfg.size, GFP_KERNEL); - if (!child->bts_buffer) - goto errout; - - child->bts = ds_request_bts(child, child->bts_buffer, cfg.size, - ovfl, /* th = */ (size_t)-1); - if (IS_ERR(child->bts)) { - error = PTR_ERR(child->bts); - kfree(child->bts_buffer); - child->bts = NULL; - child->bts_buffer = NULL; - goto errout; + if (!child->bts_buffer) { + child->bts_size = 0; + return -ENOMEM; } - - child->thread.bts_ovfl_signal = sig; } - error = -EINVAL; - if (!child->thread.ds_ctx && cfg.flags) - goto errout; - if (cfg.flags & PTRACE_BTS_O_TRACE) - child->thread.debugctlmsr |= bts_cfg.debugctl_mask; - else - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; + flags |= BTS_USER; if (cfg.flags & PTRACE_BTS_O_SCHED) - set_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - else - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); + flags |= BTS_TIMESTAMPS; - error = sizeof(cfg); + child->bts = ds_request_bts(child, child->bts_buffer, child->bts_size, + /* ovfl = */ NULL, /* th = */ (size_t)-1, + flags); + if (IS_ERR(child->bts)) { + int error = PTR_ERR(child->bts); -out: - if (child->thread.debugctlmsr) - set_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - else - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + kfree(child->bts_buffer); + child->bts = NULL; + child->bts_buffer = NULL; + child->bts_size = 0; - return error; + return error; + } -errout: - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - goto out; + return sizeof(cfg); } static int ptrace_bts_status(struct task_struct *child, long cfg_size, struct ptrace_bts_config __user *ucfg) { + const struct bts_trace *trace; struct ptrace_bts_config cfg; - size_t end; - const void *base, *max; - int error; if (cfg_size < sizeof(cfg)) return -EIO; - error = ds_get_bts_end(child->bts, &end); - if (error < 0) - return error; - - error = ds_access_bts(child->bts, /* index = */ 0, &base); - if (error < 0) - return error; - - error = ds_access_bts(child->bts, /* index = */ end, &max); - if (error < 0) - return error; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; memset(&cfg, 0, sizeof(cfg)); - cfg.size = (max - base); + cfg.size = trace->ds.end - trace->ds.begin; cfg.signal = child->thread.bts_ovfl_signal; cfg.bts_size = sizeof(struct bts_struct); if (cfg.signal) cfg.flags |= PTRACE_BTS_O_SIGNAL; - if (test_tsk_thread_flag(child, TIF_DEBUGCTLMSR) && - child->thread.debugctlmsr & bts_cfg.debugctl_mask) + if (trace->ds.flags & BTS_USER) cfg.flags |= PTRACE_BTS_O_TRACE; - if (test_tsk_thread_flag(child, TIF_BTS_TRACE_TS)) + if (trace->ds.flags & BTS_TIMESTAMPS) cfg.flags |= PTRACE_BTS_O_SCHED; if (copy_to_user(ucfg, &cfg, sizeof(cfg))) @@ -873,105 +746,28 @@ static int ptrace_bts_status(struct task_struct *child, return sizeof(cfg); } -static int ptrace_bts_write_record(struct task_struct *child, - const struct bts_struct *in) +static int ptrace_bts_clear(struct task_struct *child) { - unsigned char bts_record[BTS_MAX_RECORD_SIZE]; + const struct bts_trace *trace; - if (BTS_MAX_RECORD_SIZE < bts_cfg.sizeof_bts) - return -EOVERFLOW; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - memset(bts_record, 0, bts_cfg.sizeof_bts); - switch (in->qualifier) { - case BTS_INVALID: - break; + memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); - case BTS_BRANCH: - bts_set(bts_record, bts_from, in->variant.lbr.from_ip); - bts_set(bts_record, bts_to, in->variant.lbr.to_ip); - break; - - case BTS_TASK_ARRIVES: - case BTS_TASK_DEPARTS: - bts_set(bts_record, bts_from, bts_escape); - bts_set(bts_record, bts_qual, in->qualifier); - bts_set(bts_record, bts_jiffies, in->variant.jiffies); - break; - - default: - return -EINVAL; - } - - return ds_write_bts(child->bts, bts_record, bts_cfg.sizeof_bts); + return ds_reset_bts(child->bts); } -void ptrace_bts_take_timestamp(struct task_struct *tsk, - enum bts_qualifier qualifier) +static int ptrace_bts_size(struct task_struct *child) { - struct bts_struct rec = { - .qualifier = qualifier, - .variant.jiffies = jiffies_64 - }; - - ptrace_bts_write_record(tsk, &rec); -} - -static const struct bts_configuration bts_cfg_netburst = { - .sizeof_bts = sizeof(long) * 3, - .sizeof_field = sizeof(long), - .debugctl_mask = (1<<2)|(1<<3)|(1<<5) -}; + const struct bts_trace *trace; -static const struct bts_configuration bts_cfg_pentium_m = { - .sizeof_bts = sizeof(long) * 3, - .sizeof_field = sizeof(long), - .debugctl_mask = (1<<6)|(1<<7) -}; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; -static const struct bts_configuration bts_cfg_core2 = { - .sizeof_bts = 8 * 3, - .sizeof_field = 8, - .debugctl_mask = (1<<6)|(1<<7)|(1<<9) -}; - -static inline void bts_configure(const struct bts_configuration *cfg) -{ - bts_cfg = *cfg; -} - -void __cpuinit ptrace_bts_init_intel(struct cpuinfo_x86 *c) -{ - switch (c->x86) { - case 0x6: - switch (c->x86_model) { - case 0 ... 0xC: - /* sorry, don't know about them */ - break; - case 0xD: - case 0xE: /* Pentium M */ - bts_configure(&bts_cfg_pentium_m); - break; - default: /* Core2, Atom, ... */ - bts_configure(&bts_cfg_core2); - break; - } - break; - case 0xF: - switch (c->x86_model) { - case 0x0: - case 0x1: - case 0x2: /* Netburst */ - bts_configure(&bts_cfg_netburst); - break; - default: - /* sorry, don't know about them */ - break; - } - break; - default: - /* sorry, don't know about them */ - break; - } + return (trace->ds.top - trace->ds.begin) / trace->ds.size; } #endif /* CONFIG_X86_PTRACE_BTS */ @@ -988,15 +784,12 @@ void ptrace_disable(struct task_struct *child) #endif #ifdef CONFIG_X86_PTRACE_BTS if (child->bts) { - (void)ds_release_bts(child->bts); + ds_release_bts(child->bts); + child->bts = NULL; + kfree(child->bts_buffer); child->bts_buffer = NULL; - - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); + child->bts_size = 0; } #endif /* CONFIG_X86_PTRACE_BTS */ } @@ -1129,16 +922,9 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) (child, data, (struct ptrace_bts_config __user *)addr); break; - case PTRACE_BTS_SIZE: { - size_t size; - - ret = ds_get_bts_index(child->bts, &size); - if (ret == 0) { - WARN_ON_ONCE(size != (int) size); - ret = (int) size; - } + case PTRACE_BTS_SIZE: + ret = ptrace_bts_size(child); break; - } case PTRACE_BTS_GET: ret = ptrace_bts_read_record @@ -1146,7 +932,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_CLEAR: - ret = ds_clear_bts(child->bts); + ret = ptrace_bts_clear(child); break; case PTRACE_BTS_DRAIN: @@ -1409,6 +1195,14 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request, case PTRACE_GET_THREAD_AREA: case PTRACE_SET_THREAD_AREA: +#ifdef CONFIG_X86_PTRACE_BTS + case PTRACE_BTS_CONFIG: + case PTRACE_BTS_STATUS: + case PTRACE_BTS_SIZE: + case PTRACE_BTS_GET: + case PTRACE_BTS_CLEAR: + case PTRACE_BTS_DRAIN: +#endif /* CONFIG_X86_PTRACE_BTS */ return arch_ptrace(child, request, addr, data); default: -- cgit v1.2.3 From ffc2238af8431d930d2c15f16feecf1fd6d75642 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 12 Dec 2008 08:21:19 +0100 Subject: x86, bts: fix build error Impact: build fix arch/x86/kernel/ds.c: In function 'ds_request': arch/x86/kernel/ds.c:236: sorry, unimplemented: inlining failed in call to 'ds_get_context': recursive inlining but the recursion here is scary ... Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index f0583005b75e..dc1e7123ea4e 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -232,7 +232,7 @@ static DEFINE_PER_CPU(struct ds_context *, system_context_array); #define system_context per_cpu(system_context_array, smp_processor_id()) -static inline struct ds_context *ds_get_context(struct task_struct *task) +static struct ds_context *ds_get_context(struct task_struct *task) { struct ds_context **p_context = (task ? &task->thread.ds_ctx : &system_context); -- cgit v1.2.3 From 85072bd55219231b8ca5d9d3fa3492eb4fa6635f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 12 Dec 2008 11:08:42 +0100 Subject: x86, debug: remove EBDA debug printk Remove leftover EBDA debug message. Reported-by: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/kernel/head.c | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/head.c b/arch/x86/kernel/head.c index 1dcb0f13897e..3e66bd364a9d 100644 --- a/arch/x86/kernel/head.c +++ b/arch/x86/kernel/head.c @@ -35,7 +35,6 @@ void __init reserve_ebda_region(void) /* start of EBDA area */ ebda_addr = get_bios_ebda(); - printk(KERN_INFO "BIOS EBDA/lowmem at: %08x/%08x\n", ebda_addr, lowmem); /* Fixup: bios puts an EBDA in the top 64K segment */ /* of conventional memory, but does not adjust lowmem. */ -- cgit v1.2.3 From a0343e823184070f55364d8359f832dcb33c57c7 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 9 Dec 2008 23:53:16 +0100 Subject: tracing/function-graph-tracer: add a new .irqentry.text section Impact: let the function-graph-tracer be aware of the irq entrypoints Add a new .irqentry.text section to store the irq entrypoints functions inside the same section. This way, the tracer will be able to signal an interrupts triggering on output by recognizing these entrypoints. Also, make this section recordable for dynamic tracing. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/vmlinux_64.lds.S | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 46e05447405b..1a614c0e6bef 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -35,6 +35,7 @@ SECTIONS SCHED_TEXT LOCK_TEXT KPROBES_TEXT + IRQENTRY_TEXT *(.fixup) *(.gnu.warning) _etext = .; /* End of text section */ -- cgit v1.2.3 From bcbc4f20b52c2c40c43a4d2337707dcdfe81bc3a Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 9 Dec 2008 23:54:20 +0100 Subject: tracing/function-graph-tracer: annotate do_IRQ and smp_apic_timer_interrupt Impact: move most important x86 irq entry-points to a separate subsection Annotate do_IRQ and smp_apic_timer_interrupt to put them into the .irqentry.text subsection. These function will so be recognized as hardirq entrypoints for the function-graph-tracer. We could also annotate other irq entries but the others are far less important but they can be added on request. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 3 ++- arch/x86/kernel/irq_64.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 16f94879b525..b946ac19753b 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -30,6 +30,7 @@ #include #include #include +#include #include #include @@ -800,7 +801,7 @@ static void local_apic_timer_interrupt(void) * [ if a single-CPU system runs an SMP kernel then we call the local * interrupt as well. Thus we cannot inline the local irq ... ] */ -void smp_apic_timer_interrupt(struct pt_regs *regs) +void __irq_entry smp_apic_timer_interrupt(struct pt_regs *regs) { struct pt_regs *old_regs = set_irq_regs(regs); diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 60eb84eb77a0..11c65e811ffe 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include #include @@ -47,7 +48,7 @@ static inline void stack_overflow_check(struct pt_regs *regs) * SMP cross-CPU interrupts have their own specific * handlers). */ -asmlinkage unsigned int do_IRQ(struct pt_regs *regs) +asmlinkage unsigned int __irq_entry do_IRQ(struct pt_regs *regs) { struct pt_regs *old_regs = set_irq_regs(regs); struct irq_desc *desc; -- cgit v1.2.3 From 8f2466f45f75e3cbe3aa2b69d33fd9d6e343b9cc Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 8 Dec 2008 19:19:07 -0800 Subject: x86: kill #ifdef for exit_idle() Impact: cleanup Introduce helper inline function in arch/x86/include/asm/idle.h to remove #ifdefs around exit_idle(). Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 6 ------ arch/x86/kernel/io_apic.c | 3 +-- 2 files changed, 1 insertion(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 16f94879b525..0fd083713f62 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -814,9 +814,7 @@ void smp_apic_timer_interrupt(struct pt_regs *regs) * Besides, if we don't timer interrupts ignore the global * interrupt lock, which is the WrongThing (tm) to do. */ -#ifdef CONFIG_X86_64 exit_idle(); -#endif irq_enter(); local_apic_timer_interrupt(); irq_exit(); @@ -1682,9 +1680,7 @@ void smp_spurious_interrupt(struct pt_regs *regs) { u32 v; -#ifdef CONFIG_X86_64 exit_idle(); -#endif irq_enter(); /* * Check if this really is a spurious interrupt and ACK it @@ -1713,9 +1709,7 @@ void smp_error_interrupt(struct pt_regs *regs) { u32 v, v1; -#ifdef CONFIG_X86_64 exit_idle(); -#endif irq_enter(); /* First tickle the hardware, only then report what went on. -- REW */ v = apic_read(APIC_ESR); diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9043251210fb..679e7bbbbcd6 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2216,10 +2216,9 @@ static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) asmlinkage void smp_irq_move_cleanup_interrupt(void) { unsigned vector, me; + ack_APIC_irq(); -#ifdef CONFIG_X86_64 exit_idle(); -#endif irq_enter(); me = smp_processor_id(); -- cgit v1.2.3 From 915b0d0104b72fd36af088ba4b11b5690bc96a6c Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Mon, 8 Dec 2008 19:19:26 -0800 Subject: x86: hardirq: introduce inc_irq_stat() Impact: cleanup Introduce inc_irq_stat() macro and unify irq_stat accounting code. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic.c | 13 +++---------- arch/x86/kernel/smp.c | 18 +++--------------- arch/x86/kernel/traps.c | 6 +----- 3 files changed, 7 insertions(+), 30 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 16f94879b525..1771dd746811 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -783,11 +783,7 @@ static void local_apic_timer_interrupt(void) /* * the NMI deadlock-detector uses this. */ -#ifdef CONFIG_X86_64 - add_pda(apic_timer_irqs, 1); -#else - per_cpu(irq_stat, cpu).apic_timer_irqs++; -#endif + inc_irq_stat(apic_timer_irqs); evt->event_handler(evt); } @@ -1695,14 +1691,11 @@ void smp_spurious_interrupt(struct pt_regs *regs) if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) ack_APIC_irq(); -#ifdef CONFIG_X86_64 - add_pda(irq_spurious_count, 1); -#else + inc_irq_stat(irq_spurious_count); + /* see sw-dev-man vol 3, chapter 7.4.13.5 */ printk(KERN_INFO "spurious APIC interrupt on CPU#%d, " "should never happen.\n", smp_processor_id()); - __get_cpu_var(irq_stat).irq_spurious_count++; -#endif irq_exit(); } diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c index 18f9b19f5f8f..d18537ce2c79 100644 --- a/arch/x86/kernel/smp.c +++ b/arch/x86/kernel/smp.c @@ -178,11 +178,7 @@ static void native_smp_send_stop(void) void smp_reschedule_interrupt(struct pt_regs *regs) { ack_APIC_irq(); -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_resched_count++; -#else - add_pda(irq_resched_count, 1); -#endif + inc_irq_stat(irq_resched_count); } void smp_call_function_interrupt(struct pt_regs *regs) @@ -190,11 +186,7 @@ void smp_call_function_interrupt(struct pt_regs *regs) ack_APIC_irq(); irq_enter(); generic_smp_call_function_interrupt(); -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_call_count++; -#else - add_pda(irq_call_count, 1); -#endif + inc_irq_stat(irq_call_count); irq_exit(); } @@ -203,11 +195,7 @@ void smp_call_function_single_interrupt(struct pt_regs *regs) ack_APIC_irq(); irq_enter(); generic_smp_call_function_single_interrupt(); -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_call_count++; -#else - add_pda(irq_call_count, 1); -#endif + inc_irq_stat(irq_call_count); irq_exit(); } diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index 04d242ab0161..d815293e6d94 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -481,11 +481,7 @@ do_nmi(struct pt_regs *regs, long error_code) { nmi_enter(); -#ifdef CONFIG_X86_32 - { int cpu; cpu = smp_processor_id(); ++nmi_count(cpu); } -#else - add_pda(__nmi_count, 1); -#endif + inc_irq_stat(__nmi_count); if (!ignore_nmis) default_do_nmi(regs); -- cgit v1.2.3 From 2bed8446819a7c5033aa1da138d9f230ae212edc Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 12 Dec 2008 12:13:36 +0100 Subject: tracing/function-graph-tracer: add a new .irqentry.text section, fix Impact: build fix 32-bit x86 needs this section too. Signed-off-by: Ingo Molnar --- arch/x86/kernel/vmlinux_32.lds.S | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index a9b8560adbc2..82c67559dde7 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -44,6 +44,7 @@ SECTIONS SCHED_TEXT LOCK_TEXT KPROBES_TEXT + IRQENTRY_TEXT *(.fixup) *(.gnu.warning) _etext = .; /* End of text section */ -- cgit v1.2.3 From 9470565579f29486f4ed0ffa50774268b64994b0 Mon Sep 17 00:00:00 2001 From: Dave Jones Date: Mon, 1 Dec 2008 14:13:50 -0800 Subject: x86: remove init_mm export as planned for 2.6.26 Impact: remove deprecated export Signed-off-by: Dave Jones Signed-off-by: Andrew Morton Signed-off-by: Thomas Gleixner --- arch/x86/kernel/init_task.c | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/init_task.c b/arch/x86/kernel/init_task.c index a4f93b4120c1..d39918076bb4 100644 --- a/arch/x86/kernel/init_task.c +++ b/arch/x86/kernel/init_task.c @@ -14,7 +14,6 @@ static struct fs_struct init_fs = INIT_FS; static struct signal_struct init_signals = INIT_SIGNALS(init_signals); static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand); struct mm_struct init_mm = INIT_MM(init_mm); -EXPORT_UNUSED_SYMBOL(init_mm); /* will be removed in 2.6.26 */ /* * Initial thread structure. -- cgit v1.2.3 From fd28a5b58dddf5cb5df162ae5c8797a63171c31d Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Tue, 21 Oct 2008 14:05:00 +0200 Subject: x86: remove simnow earlyprintk support Impact: remove obsolete code The later versions of SimNow! actually all have serial console emulation, so the direct interface isn't needed anymore. So remove the undocumented simnow earlyprintk console. Signed-off-by: Andi Kleen Signed-off-by: Thomas Gleixner --- arch/x86/kernel/early_printk.c | 47 ------------------------------------------ 1 file changed, 47 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/early_printk.c b/arch/x86/kernel/early_printk.c index 34ad997d3834..23b138e31e9c 100644 --- a/arch/x86/kernel/early_printk.c +++ b/arch/x86/kernel/early_printk.c @@ -875,49 +875,6 @@ static struct console early_dbgp_console = { }; #endif -/* Console interface to a host file on AMD's SimNow! */ - -static int simnow_fd; - -enum { - MAGIC1 = 0xBACCD00A, - MAGIC2 = 0xCA110000, - XOPEN = 5, - XWRITE = 4, -}; - -static noinline long simnow(long cmd, long a, long b, long c) -{ - long ret; - - asm volatile("cpuid" : - "=a" (ret) : - "b" (a), "c" (b), "d" (c), "0" (MAGIC1), "D" (cmd + MAGIC2)); - return ret; -} - -static void __init simnow_init(char *str) -{ - char *fn = "klog"; - - if (*str == '=') - fn = ++str; - /* error ignored */ - simnow_fd = simnow(XOPEN, (unsigned long)fn, O_WRONLY|O_APPEND|O_CREAT, 0644); -} - -static void simnow_write(struct console *con, const char *s, unsigned n) -{ - simnow(XWRITE, simnow_fd, (unsigned long)s, n); -} - -static struct console simnow_console = { - .name = "simnow", - .write = simnow_write, - .flags = CON_PRINTBUFFER, - .index = -1, -}; - /* Direct interface for emergencies */ static struct console *early_console = &early_vga_console; static int __initdata early_console_initialized; @@ -960,10 +917,6 @@ static int __init setup_early_printk(char *buf) max_ypos = boot_params.screen_info.orig_video_lines; current_ypos = boot_params.screen_info.orig_y; early_console = &early_vga_console; - } else if (!strncmp(buf, "simnow", 6)) { - simnow_init(buf + 6); - early_console = &simnow_console; - keep_early = 1; #ifdef CONFIG_EARLY_PRINTK_DBGP } else if (!strncmp(buf, "dbgp", 4)) { if (early_dbgp_init(buf+4) < 0) -- cgit v1.2.3 From ae8d04e2ecbb233926860e9ce145eac19c7835dc Mon Sep 17 00:00:00 2001 From: Zachary Amsden Date: Sat, 13 Dec 2008 12:36:58 -0800 Subject: x86 Fix VMI crash on boot in 2.6.28-rc8 VMI initialiation can relocate the fixmap, causing early_ioremap to malfunction if it is initialized before the relocation. To fix this, VMI activation is split into two phases; the detection, which must happen before setting up ioremap, and the activation, which must happen after parsing early boot parameters. This fixes a crash on boot when VMI is enabled under VMware. Signed-off-by: Zachary Amsden Signed-off-by: Linus Torvalds --- arch/x86/kernel/setup.c | 12 +++++------- arch/x86/kernel/smpboot.c | 2 -- arch/x86/kernel/vmi_32.c | 16 +++++++++++----- 3 files changed, 16 insertions(+), 14 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 9d5674f7b6cc..bdec76e55594 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -794,6 +794,9 @@ void __init setup_arch(char **cmdline_p) printk(KERN_INFO "Command line: %s\n", boot_command_line); #endif + /* VMI may relocate the fixmap; do this before touching ioremap area */ + vmi_init(); + early_cpu_init(); early_ioremap_init(); @@ -880,13 +883,8 @@ void __init setup_arch(char **cmdline_p) check_efer(); #endif -#if defined(CONFIG_VMI) && defined(CONFIG_X86_32) - /* - * Must be before kernel pagetables are setup - * or fixmap area is touched. - */ - vmi_init(); -#endif + /* Must be before kernel pagetables are setup */ + vmi_activate(); /* after early param, so could get panic from serial */ reserve_early_setup_data(); diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 7b1093397319..f71f96fc9e62 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -294,9 +294,7 @@ static void __cpuinit start_secondary(void *unused) * fragile that we want to limit the things done here to the * most necessary things. */ -#ifdef CONFIG_VMI vmi_bringup(); -#endif cpu_init(); preempt_disable(); smp_callin(); diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c index 8b6c393ab9fd..22fd6577156a 100644 --- a/arch/x86/kernel/vmi_32.c +++ b/arch/x86/kernel/vmi_32.c @@ -960,8 +960,6 @@ static inline int __init activate_vmi(void) void __init vmi_init(void) { - unsigned long flags; - if (!vmi_rom) probe_vmi_rom(); else @@ -973,13 +971,21 @@ void __init vmi_init(void) reserve_top_address(-vmi_rom->virtual_top); - local_irq_save(flags); - activate_vmi(); - #ifdef CONFIG_X86_IO_APIC /* This is virtual hardware; timer routing is wired correctly */ no_timer_check = 1; #endif +} + +void vmi_activate(void) +{ + unsigned long flags; + + if (!vmi_rom) + return; + + local_irq_save(flags); + activate_vmi(); local_irq_restore(flags & X86_EFLAGS_IF); } -- cgit v1.2.3 From cc1dc6d039ced64c2f8b8457bf1cccf4ecfc5942 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 16 Dec 2008 15:51:03 +0100 Subject: x86, bts: remove recursion from get_context Impact: cleanup Optimistically allocate a DS context. It is extremely unlikely that one already existed. This simplifies the code a lot. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 58 ++++++++++++++++++++++------------------------------ 1 file changed, 25 insertions(+), 33 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index dc1e7123ea4e..0dc795951d73 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -232,53 +232,45 @@ static DEFINE_PER_CPU(struct ds_context *, system_context_array); #define system_context per_cpu(system_context_array, smp_processor_id()) -static struct ds_context *ds_get_context(struct task_struct *task) + +static inline struct ds_context *ds_get_context(struct task_struct *task) { struct ds_context **p_context = (task ? &task->thread.ds_ctx : &system_context); - struct ds_context *context = *p_context; + struct ds_context *context = NULL; + struct ds_context *new_context = NULL; unsigned long irq; - if (!context) { - context = kzalloc(sizeof(*context), GFP_KERNEL); - if (!context) - return NULL; - - spin_lock_irqsave(&ds_lock, irq); - - if (*p_context) { - kfree(context); + /* Chances are small that we already have a context. */ + new_context = kzalloc(sizeof(*new_context), GFP_KERNEL); + if (!new_context) + return NULL; - context = *p_context; - } else { - *p_context = context; + spin_lock_irqsave(&ds_lock, irq); - context->this = p_context; - context->task = task; + context = *p_context; + if (!context) { + context = new_context; - if (task) - set_tsk_thread_flag(task, TIF_DS_AREA_MSR); + context->this = p_context; + context->task = task; + context->count = 0; - if (!task || (task == current)) - wrmsrl(MSR_IA32_DS_AREA, - (unsigned long)context->ds); - } + if (task) + set_tsk_thread_flag(task, TIF_DS_AREA_MSR); - context->count++; + if (!task || (task == current)) + wrmsrl(MSR_IA32_DS_AREA, (unsigned long)context->ds); - spin_unlock_irqrestore(&ds_lock, irq); - } else { - spin_lock_irqsave(&ds_lock, irq); + *p_context = context; + } - context = *p_context; - if (context) - context->count++; + context->count++; - spin_unlock_irqrestore(&ds_lock, irq); + spin_unlock_irqrestore(&ds_lock, irq); - if (!context) - context = ds_get_context(task); - } + if (context != new_context) + kfree(new_context); return context; } -- cgit v1.2.3 From d072c25f531c6513994960401d2c7f059434c0d2 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Tue, 16 Dec 2008 15:53:11 +0100 Subject: x86, bts: correctly report invalid bts records Impact: change the reporting of empty BTS records Correctly report a cleared BTS record as invalid. Used to be reported as branch from 0 to 0. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 0dc795951d73..98d271e60e08 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -484,6 +484,9 @@ static int bts_read(struct bts_tracer *tracer, const void *at, out->qualifier = bts_branch; out->variant.lbr.from = bts_get(at, bts_from); out->variant.lbr.to = bts_get(at, bts_to); + + if (!out->variant.lbr.from && !out->variant.lbr.to) + out->qualifier = bts_invalid; } return ds_cfg.sizeof_rec[ds_bts]; -- cgit v1.2.3 From cfc319833b5b359bf3bce99564dbac00af7925ac Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Tue, 16 Dec 2008 11:46:58 +0000 Subject: x86, 32-bit: improve lazy TLB handling code Impact: micro-optimize the 32-bit TLB flush code Use the faster x86_{read,write}_percpu() accessors here. Signed-off-by: Jan Beulich Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_32.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c index f4049f3513b6..4290d918b58a 100644 --- a/arch/x86/kernel/tlb_32.c +++ b/arch/x86/kernel/tlb_32.c @@ -34,9 +34,8 @@ static DEFINE_SPINLOCK(tlbstate_lock); */ void leave_mm(int cpu) { - if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) - BUG(); - cpu_clear(cpu, per_cpu(cpu_tlbstate, cpu).active_mm->cpu_vm_mask); + BUG_ON(x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK); + cpu_clear(cpu, x86_read_percpu(cpu_tlbstate.active_mm)->cpu_vm_mask); load_cr3(swapper_pg_dir); } EXPORT_SYMBOL_GPL(leave_mm); @@ -104,8 +103,8 @@ void smp_invalidate_interrupt(struct pt_regs *regs) * BUG(); */ - if (flush_mm == per_cpu(cpu_tlbstate, cpu).active_mm) { - if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) { + if (flush_mm == x86_read_percpu(cpu_tlbstate.active_mm)) { + if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK) { if (flush_va == TLB_FLUSH_ALL) local_flush_tlb(); else @@ -238,7 +237,7 @@ static void do_flush_tlb_all(void *info) unsigned long cpu = smp_processor_id(); __flush_tlb_all(); - if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_LAZY) + if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_LAZY) leave_mm(cpu); } -- cgit v1.2.3 From 83fd5cc6481c6b7fa8b45f8a7e0aa7120213430b Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Tue, 16 Dec 2008 19:17:11 +0100 Subject: AMD IOMMU: allocate rlookup_table with __GFP_ZERO Impact: fix bug which can lead to panic in prealloc_protection_domains() Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu_init.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index 30ae2701b3df..c90a15eba5c5 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -1074,7 +1074,8 @@ int __init amd_iommu_init(void) goto free; /* IOMMU rlookup table - find the IOMMU for a specific device */ - amd_iommu_rlookup_table = (void *)__get_free_pages(GFP_KERNEL, + amd_iommu_rlookup_table = (void *)__get_free_pages( + GFP_KERNEL | __GFP_ZERO, get_order(rlookup_table_size)); if (amd_iommu_rlookup_table == NULL) goto free; -- cgit v1.2.3 From 3c763fd77e66e55d029052da31df0abd9920cb1e Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:07:47 +0100 Subject: x86: microcode_amd: fix wrong handling of equivalent CPU id Impact: fix bug resulting in non-loaded AMD microcode mc_header->processor_rev_id is a 2 byte value. Similar is true for equiv_cpu in an equiv_cpu_entry -- only 2 bytes are of interest. Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 23 ++++++----------------- 1 file changed, 6 insertions(+), 17 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 5f8e5d75a254..b5bc81470bcf 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -62,7 +62,7 @@ struct microcode_header_amd { unsigned int mc_patch_data_checksum; unsigned int nb_dev_id; unsigned int sb_dev_id; - unsigned char processor_rev_id[2]; + u16 processor_rev_id; unsigned char nb_rev_id; unsigned char sb_rev_id; unsigned char bios_api_rev; @@ -125,7 +125,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) while (equiv_cpu_table[i].installed_cpu != 0) { if (current_cpu_id == equiv_cpu_table[i].installed_cpu) { - equiv_cpu_id = equiv_cpu_table[i].equiv_cpu; + equiv_cpu_id = equiv_cpu_table[i].equiv_cpu & 0xffff; break; } i++; @@ -137,21 +137,10 @@ static int get_matching_microcode(int cpu, void *mc, int rev) return 0; } - if ((mc_header->processor_rev_id[0]) != (equiv_cpu_id & 0xff)) { - printk(KERN_ERR - "microcode: CPU%d patch does not match " - "(patch is %x, cpu extended is %x) \n", - cpu, mc_header->processor_rev_id[0], - (equiv_cpu_id & 0xff)); - return 0; - } - - if ((mc_header->processor_rev_id[1]) != ((equiv_cpu_id >> 16) & 0xff)) { - printk(KERN_ERR "microcode: CPU%d patch does not match " - "(patch is %x, cpu base id is %x) \n", - cpu, mc_header->processor_rev_id[1], - ((equiv_cpu_id >> 16) & 0xff)); - + if (mc_header->processor_rev_id != equiv_cpu_id) { + printk(KERN_ERR "microcode: CPU%d patch does not match " + "(processor_rev_id: %x, eqiv_cpu_id: %x)\n", + cpu, mc_header->processor_rev_id, equiv_cpu_id); return 0; } -- cgit v1.2.3 From 2a3282a77b02fb47576ffbdb4867c8c6eeb83ed5 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:08:53 +0100 Subject: x86: microcode_amd: fix typos and trailing whitespaces in log messages Impact: fix printk typos Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index b5bc81470bcf..83a9fa321d9b 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -10,7 +10,7 @@ * This driver allows to upgrade microcode on AMD * family 0x10 and 0x11 processors. * - * Licensed unter the terms of the GNU General Public + * Licensed under the terms of the GNU General Public * License version 2. See file COPYING for details. */ @@ -133,7 +133,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) if (!equiv_cpu_id) { printk(KERN_ERR "microcode: CPU%d cpu_id " - "not found in equivalent cpu table \n", cpu); + "not found in equivalent cpu table\n", cpu); return 0; } @@ -151,7 +151,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) NULL); if ((!nb_pci_dev) || (mc_header->nb_rev_id != nb_pci_dev->revision)) { - printk(KERN_ERR "microcode: CPU%d NB mismatch \n", cpu); + printk(KERN_ERR "microcode: CPU%d NB mismatch\n", cpu); pci_dev_put(nb_pci_dev); return 0; } @@ -165,7 +165,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) NULL); if ((!sb_pci_dev) || (mc_header->sb_rev_id != sb_pci_dev->revision)) { - printk(KERN_ERR "microcode: CPU%d SB mismatch \n", cpu); + printk(KERN_ERR "microcode: CPU%d SB mismatch\n", cpu); pci_dev_put(sb_pci_dev); return 0; } @@ -219,7 +219,7 @@ static void apply_microcode_amd(int cpu) } printk(KERN_INFO "microcode: CPU%d updated from revision " - "0x%x to 0x%x \n", + "0x%x to 0x%x\n", cpu_num, uci->cpu_sig.rev, mc_amd->hdr.patch_id); uci->cpu_sig.rev = rev; @@ -282,7 +282,7 @@ static int install_equiv_cpu_table(u8 *buf, if (buf_pos[1] != UCODE_EQUIV_CPU_TABLE_TYPE || !size) { printk(KERN_ERR "microcode: error! " - "Wrong microcode equivalnet cpu table\n"); + "Wrong microcode equivalent cpu table\n"); return 0; } -- cgit v1.2.3 From be957763b01905d33b53cdd25c8df110f94f499a Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:11:23 +0100 Subject: x86: microcode_amd: fix checkpatch warnings/errors Impact: cleanup Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 83a9fa321d9b..a8a0ec600554 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -32,9 +32,9 @@ #include #include #include +#include #include -#include #include #include @@ -225,7 +225,7 @@ static void apply_microcode_amd(int cpu) uci->cpu_sig.rev = rev; } -static void * get_next_ucode(u8 *buf, unsigned int size, +static void *get_next_ucode(u8 *buf, unsigned int size, int (*get_ucode_data)(void *, const void *, size_t), unsigned int *mc_size) { @@ -256,7 +256,8 @@ static void * get_next_ucode(u8 *buf, unsigned int size, mc = vmalloc(UCODE_MAX_SIZE); if (mc) { memset(mc, 0, UCODE_MAX_SIZE); - if (get_ucode_data(mc, buf + UCODE_CONTAINER_SECTION_HDR, total_size)) { + if (get_ucode_data(mc, buf + UCODE_CONTAINER_SECTION_HDR, + total_size)) { vfree(mc); mc = NULL; } else @@ -332,7 +333,8 @@ static int generic_load_microcode(int cpu, void *data, size_t size, unsigned int uninitialized_var(mc_size); struct microcode_header_amd *mc_header; - mc = get_next_ucode(ucode_ptr, leftover, get_ucode_data, &mc_size); + mc = get_next_ucode(ucode_ptr, leftover, get_ucode_data, + &mc_size); if (!mc) break; @@ -342,7 +344,7 @@ static int generic_load_microcode(int cpu, void *data, size_t size, vfree(new_mc); new_rev = mc_header->patch_id; new_mc = mc; - } else + } else vfree(mc); ucode_ptr += mc_size; @@ -354,9 +356,9 @@ static int generic_load_microcode(int cpu, void *data, size_t size, if (uci->mc) vfree(uci->mc); uci->mc = new_mc; - pr_debug("microcode: CPU%d found a matching microcode update with" - " version 0x%x (current=0x%x)\n", - cpu, new_rev, uci->cpu_sig.rev); + pr_debug("microcode: CPU%d found a matching microcode " + "update with version 0x%x (current=0x%x)\n", + cpu, new_rev, uci->cpu_sig.rev); } else vfree(new_mc); } @@ -383,7 +385,8 @@ static int request_microcode_fw(int cpu, struct device *device) ret = request_firmware(&firmware, fw_name, device); if (ret) { - printk(KERN_ERR "microcode: ucode data file %s load failed\n", fw_name); + printk(KERN_ERR "microcode: ucode data file %s load failed\n", + fw_name); return ret; } @@ -397,8 +400,8 @@ static int request_microcode_fw(int cpu, struct device *device) static int request_microcode_user(int cpu, const void __user *buf, size_t size) { - printk(KERN_WARNING "microcode: AMD microcode update via /dev/cpu/microcode" - "is not supported\n"); + printk(KERN_WARNING "microcode: AMD microcode update via " + "/dev/cpu/microcode is not supported\n"); return -1; } -- cgit v1.2.3 From 8c135206c826095c852c16d94a0a74eeaf05c90d Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:13:00 +0100 Subject: x86: microcode_amd: fix compile warning MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Impact: fix build warning CC arch/x86/kernel/microcode_amd.o arch/x86/kernel/microcode_amd.c: In function ‘request_microcode_fw’: arch/x86/kernel/microcode_amd.c:393: warning: passing argument 2 of ‘generic_load_microcode’ discards qualifiers from pointer target type (Respect "const" qualifier of firmware->data.) Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index a8a0ec600554..89b386c901fd 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -225,8 +225,8 @@ static void apply_microcode_amd(int cpu) uci->cpu_sig.rev = rev; } -static void *get_next_ucode(u8 *buf, unsigned int size, - int (*get_ucode_data)(void *, const void *, size_t), +static void *get_next_ucode(const u8 *buf, unsigned int size, + int (*get_ucode_data)(void *, const u8 *, size_t), unsigned int *mc_size) { unsigned int total_size; @@ -268,8 +268,8 @@ static void *get_next_ucode(u8 *buf, unsigned int size, } -static int install_equiv_cpu_table(u8 *buf, - int (*get_ucode_data)(void *, const void *, size_t)) +static int install_equiv_cpu_table(const u8 *buf, + int (*get_ucode_data)(void *, const u8 *, size_t)) { #define UCODE_CONTAINER_HEADER_SIZE 12 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE]; @@ -311,11 +311,13 @@ static void free_equiv_cpu_table(void) } } -static int generic_load_microcode(int cpu, void *data, size_t size, - int (*get_ucode_data)(void *, const void *, size_t)) +static int generic_load_microcode(int cpu, const u8 *data, size_t size, + int (*get_ucode_data)(void *, const u8 *, size_t)) { struct ucode_cpu_info *uci = ucode_cpu_info + cpu; - u8 *ucode_ptr = data, *new_mc = NULL, *mc; + const u8 *ucode_ptr = data; + void *new_mc = NULL; + void *mc; int new_rev = uci->cpu_sig.rev; unsigned int leftover; unsigned long offset; @@ -368,7 +370,7 @@ static int generic_load_microcode(int cpu, void *data, size_t size, return (int)leftover; } -static int get_ucode_fw(void *to, const void *from, size_t n) +static int get_ucode_fw(void *to, const u8 *from, size_t n) { memcpy(to, from, n); return 0; @@ -390,7 +392,7 @@ static int request_microcode_fw(int cpu, struct device *device) return ret; } - ret = generic_load_microcode(cpu, (void*)firmware->data, firmware->size, + ret = generic_load_microcode(cpu, firmware->data, firmware->size, &get_ucode_fw); release_firmware(firmware); -- cgit v1.2.3 From 0657d9ebff186dcdb17e582dcb909028775a7707 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:14:05 +0100 Subject: x86: microcode_amd: don't pass superfluous function pointer for get_ucode_data Impact: cleanup Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 29 ++++++++++++----------------- 1 file changed, 12 insertions(+), 17 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 89b386c901fd..c7f225c7e481 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -225,9 +225,14 @@ static void apply_microcode_amd(int cpu) uci->cpu_sig.rev = rev; } +static int get_ucode_data(void *to, const u8 *from, size_t n) +{ + memcpy(to, from, n); + return 0; +} + static void *get_next_ucode(const u8 *buf, unsigned int size, - int (*get_ucode_data)(void *, const u8 *, size_t), - unsigned int *mc_size) + unsigned int *mc_size) { unsigned int total_size; #define UCODE_CONTAINER_SECTION_HDR 8 @@ -268,8 +273,7 @@ static void *get_next_ucode(const u8 *buf, unsigned int size, } -static int install_equiv_cpu_table(const u8 *buf, - int (*get_ucode_data)(void *, const u8 *, size_t)) +static int install_equiv_cpu_table(const u8 *buf) { #define UCODE_CONTAINER_HEADER_SIZE 12 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE]; @@ -311,8 +315,7 @@ static void free_equiv_cpu_table(void) } } -static int generic_load_microcode(int cpu, const u8 *data, size_t size, - int (*get_ucode_data)(void *, const u8 *, size_t)) +static int generic_load_microcode(int cpu, const u8 *data, size_t size) { struct ucode_cpu_info *uci = ucode_cpu_info + cpu; const u8 *ucode_ptr = data; @@ -322,7 +325,7 @@ static int generic_load_microcode(int cpu, const u8 *data, size_t size, unsigned int leftover; unsigned long offset; - offset = install_equiv_cpu_table(ucode_ptr, get_ucode_data); + offset = install_equiv_cpu_table(ucode_ptr); if (!offset) { printk(KERN_ERR "microcode: installing equivalent cpu table failed\n"); return -EINVAL; @@ -335,8 +338,7 @@ static int generic_load_microcode(int cpu, const u8 *data, size_t size, unsigned int uninitialized_var(mc_size); struct microcode_header_amd *mc_header; - mc = get_next_ucode(ucode_ptr, leftover, get_ucode_data, - &mc_size); + mc = get_next_ucode(ucode_ptr, leftover, &mc_size); if (!mc) break; @@ -370,12 +372,6 @@ static int generic_load_microcode(int cpu, const u8 *data, size_t size, return (int)leftover; } -static int get_ucode_fw(void *to, const u8 *from, size_t n) -{ - memcpy(to, from, n); - return 0; -} - static int request_microcode_fw(int cpu, struct device *device) { const char *fw_name = "amd-ucode/microcode_amd.bin"; @@ -392,8 +388,7 @@ static int request_microcode_fw(int cpu, struct device *device) return ret; } - ret = generic_load_microcode(cpu, firmware->data, firmware->size, - &get_ucode_fw); + ret = generic_load_microcode(cpu, firmware->data, firmware->size); release_firmware(firmware); -- cgit v1.2.3 From 29d0887ffd084cde9d6a1286cb82b71701a974dd Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:16:34 +0100 Subject: x86: microcode_amd: replace inline asm by common rdmsr/wrmsr functions Impact: cleanup Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 23 +++++------------------ 1 file changed, 5 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index c7f225c7e481..2856955ddab1 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -93,6 +93,7 @@ static struct equiv_cpu_entry *equiv_cpu_table; static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig) { struct cpuinfo_x86 *c = &cpu_data(cpu); + u32 dummy; memset(csig, 0, sizeof(*csig)); @@ -102,9 +103,7 @@ static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig) return -1; } - asm volatile("movl %1, %%ecx; rdmsr" - : "=a" (csig->rev) - : "i" (0x0000008B) : "ecx"); + rdmsr(MSR_AMD64_PATCH_LEVEL, csig->rev, dummy); printk(KERN_INFO "microcode: collect_cpu_info_amd : patch_id=0x%x\n", csig->rev); @@ -181,12 +180,10 @@ static int get_matching_microcode(int cpu, void *mc, int rev) static void apply_microcode_amd(int cpu) { unsigned long flags; - unsigned int eax, edx; - unsigned int rev; + u32 rev, dummy; int cpu_num = raw_smp_processor_id(); struct ucode_cpu_info *uci = ucode_cpu_info + cpu_num; struct microcode_amd *mc_amd = uci->mc; - unsigned long addr; /* We should bind the task to the CPU */ BUG_ON(cpu_num != cpu); @@ -195,19 +192,9 @@ static void apply_microcode_amd(int cpu) return; spin_lock_irqsave(µcode_update_lock, flags); - - addr = (unsigned long)&mc_amd->hdr.data_code; - edx = (unsigned int)(((unsigned long)upper_32_bits(addr))); - eax = (unsigned int)(((unsigned long)lower_32_bits(addr))); - - asm volatile("movl %0, %%ecx; wrmsr" : - : "i" (0xc0010020), "a" (eax), "d" (edx) : "ecx"); - + wrmsrl(MSR_AMD64_PATCH_LOADER, &mc_amd->hdr.data_code); /* get patch id after patching */ - asm volatile("movl %1, %%ecx; rdmsr" - : "=a" (rev) - : "i" (0x0000008B) : "ecx"); - + rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy); spin_unlock_irqrestore(µcode_update_lock, flags); /* check current patch id and patch's id for match */ -- cgit v1.2.3 From 6cc9b6d94b6fee23b0671970f67d297fa76b68b3 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:17:45 +0100 Subject: x86: microcode_amd: consolidate macro definitions Impact: cleanup Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 2856955ddab1..e68e723490a3 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -75,15 +75,9 @@ struct microcode_amd { unsigned int mpb[0]; }; -#define UCODE_MAX_SIZE (2048) -#define DEFAULT_UCODE_DATASIZE (896) -#define MC_HEADER_SIZE (sizeof(struct microcode_header_amd)) -#define DEFAULT_UCODE_TOTALSIZE (DEFAULT_UCODE_DATASIZE + MC_HEADER_SIZE) -#define DWSIZE (sizeof(u32)) -/* For now we support a fixed ucode total size only */ -#define get_totalsize(mc) \ - ((((struct microcode_amd *)mc)->hdr.mc_patch_data_len * 28) \ - + MC_HEADER_SIZE) +#define UCODE_MAX_SIZE 2048 +#define UCODE_CONTAINER_SECTION_HDR 8 +#define UCODE_CONTAINER_HEADER_SIZE 12 /* serialize access to the physical write */ static DEFINE_SPINLOCK(microcode_update_lock); @@ -222,7 +216,6 @@ static void *get_next_ucode(const u8 *buf, unsigned int size, unsigned int *mc_size) { unsigned int total_size; -#define UCODE_CONTAINER_SECTION_HDR 8 u8 section_hdr[UCODE_CONTAINER_SECTION_HDR]; void *mc; @@ -255,14 +248,12 @@ static void *get_next_ucode(const u8 *buf, unsigned int size, } else *mc_size = total_size + UCODE_CONTAINER_SECTION_HDR; } -#undef UCODE_CONTAINER_SECTION_HDR return mc; } static int install_equiv_cpu_table(const u8 *buf) { -#define UCODE_CONTAINER_HEADER_SIZE 12 u8 *container_hdr[UCODE_CONTAINER_HEADER_SIZE]; unsigned int *buf_pos = (unsigned int *)container_hdr; unsigned long size; @@ -291,7 +282,6 @@ static int install_equiv_cpu_table(const u8 *buf) } return size + UCODE_CONTAINER_HEADER_SIZE; /* add header length */ -#undef UCODE_CONTAINER_HEADER_SIZE } static void free_equiv_cpu_table(void) -- cgit v1.2.3 From 98415301ea2dd389539ab429bcfa9da07219eabc Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:20:21 +0100 Subject: x86: microcode_amd: remove (wrong) chipset deivce ID checks Impact: remove dead/incorrect code Currently there is no chipset specific ucode. The checks are incorrect anyway (e.g. pci device IDs are 16 bit and not 8 bit). Thus I remove the stuff for the time being and will reintroduce it if it's foreseeable that it is really needed. Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 32 +++++--------------------------- 1 file changed, 5 insertions(+), 27 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index e68e723490a3..2e8af6ef3da9 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -108,7 +108,6 @@ static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig) static int get_matching_microcode(int cpu, void *mc, int rev) { struct microcode_header_amd *mc_header = mc; - struct pci_dev *nb_pci_dev, *sb_pci_dev; unsigned int current_cpu_id; unsigned int equiv_cpu_id = 0x00; unsigned int i = 0; @@ -137,32 +136,11 @@ static int get_matching_microcode(int cpu, void *mc, int rev) return 0; } - /* ucode may be northbridge specific */ - if (mc_header->nb_dev_id) { - nb_pci_dev = pci_get_device(PCI_VENDOR_ID_AMD, - (mc_header->nb_dev_id & 0xff), - NULL); - if ((!nb_pci_dev) || - (mc_header->nb_rev_id != nb_pci_dev->revision)) { - printk(KERN_ERR "microcode: CPU%d NB mismatch\n", cpu); - pci_dev_put(nb_pci_dev); - return 0; - } - pci_dev_put(nb_pci_dev); - } - - /* ucode may be southbridge specific */ - if (mc_header->sb_dev_id) { - sb_pci_dev = pci_get_device(PCI_VENDOR_ID_AMD, - (mc_header->sb_dev_id & 0xff), - NULL); - if ((!sb_pci_dev) || - (mc_header->sb_rev_id != sb_pci_dev->revision)) { - printk(KERN_ERR "microcode: CPU%d SB mismatch\n", cpu); - pci_dev_put(sb_pci_dev); - return 0; - } - pci_dev_put(sb_pci_dev); + /* ucode might be chipset specific -- currently we don't support this */ + if (mc_header->nb_dev_id || mc_header->sb_dev_id) { + printk(KERN_WARNING "microcode: CPU%d loading of chipset " + "specific code not yet supported\n", cpu); + return 0; } if (mc_header->patch_id <= rev) -- cgit v1.2.3 From 5549b94bc74c3e7edd44e0aeb7d9f773e82d2d20 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:21:30 +0100 Subject: x86: microcode_amd: use 'packed' attribute for structs Impact: cleanup Signed-off-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 45 +++++++++++++++++++++-------------------- 1 file changed, 23 insertions(+), 22 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 2e8af6ef3da9..e1ce650f276b 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -47,28 +47,29 @@ MODULE_LICENSE("GPL v2"); #define UCODE_UCODE_TYPE 0x00000001 struct equiv_cpu_entry { - unsigned int installed_cpu; - unsigned int fixed_errata_mask; - unsigned int fixed_errata_compare; - unsigned int equiv_cpu; -}; + u32 installed_cpu; + u32 fixed_errata_mask; + u32 fixed_errata_compare; + u16 equiv_cpu; + u16 res; +} __attribute__((packed)); struct microcode_header_amd { - unsigned int data_code; - unsigned int patch_id; - unsigned char mc_patch_data_id[2]; - unsigned char mc_patch_data_len; - unsigned char init_flag; - unsigned int mc_patch_data_checksum; - unsigned int nb_dev_id; - unsigned int sb_dev_id; - u16 processor_rev_id; - unsigned char nb_rev_id; - unsigned char sb_rev_id; - unsigned char bios_api_rev; - unsigned char reserved1[3]; - unsigned int match_reg[8]; -}; + u32 data_code; + u32 patch_id; + u16 mc_patch_data_id; + u8 mc_patch_data_len; + u8 init_flag; + u32 mc_patch_data_checksum; + u32 nb_dev_id; + u32 sb_dev_id; + u16 processor_rev_id; + u8 nb_rev_id; + u8 sb_rev_id; + u8 bios_api_rev; + u8 reserved1[3]; + u32 match_reg[8]; +} __attribute__((packed)); struct microcode_amd { struct microcode_header_amd hdr; @@ -109,7 +110,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) { struct microcode_header_amd *mc_header = mc; unsigned int current_cpu_id; - unsigned int equiv_cpu_id = 0x00; + u16 equiv_cpu_id = 0; unsigned int i = 0; BUG_ON(equiv_cpu_table == NULL); @@ -117,7 +118,7 @@ static int get_matching_microcode(int cpu, void *mc, int rev) while (equiv_cpu_table[i].installed_cpu != 0) { if (current_cpu_id == equiv_cpu_table[i].installed_cpu) { - equiv_cpu_id = equiv_cpu_table[i].equiv_cpu & 0xffff; + equiv_cpu_id = equiv_cpu_table[i].equiv_cpu; break; } i++; -- cgit v1.2.3 From df23cab563912ba43f7e9bc8ac517e5a2ddc9cd2 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Tue, 16 Dec 2008 19:22:36 +0100 Subject: x86: microcode_amd: modify log messages Impact: change microcode printk content Change log level and provide (at least I tried to;-) consistent, short, meaningful content. Signed-off-by: Andreas Herrmann Cc: Dmitry Adamushko Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 58 ++++++++++++++++++----------------------- 1 file changed, 26 insertions(+), 32 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index e1ce650f276b..24c256f4e50a 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -91,18 +91,13 @@ static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig) u32 dummy; memset(csig, 0, sizeof(*csig)); - if (c->x86_vendor != X86_VENDOR_AMD || c->x86 < 0x10) { - printk(KERN_ERR "microcode: CPU%d not a capable AMD processor\n", - cpu); + printk(KERN_WARNING "microcode: CPU%d: AMD CPU family 0x%x not " + "supported\n", cpu, c->x86); return -1; } - rdmsr(MSR_AMD64_PATCH_LEVEL, csig->rev, dummy); - - printk(KERN_INFO "microcode: collect_cpu_info_amd : patch_id=0x%x\n", - csig->rev); - + printk(KERN_INFO "microcode: CPU%d: patch_level=0x%x\n", cpu, csig->rev); return 0; } @@ -125,21 +120,21 @@ static int get_matching_microcode(int cpu, void *mc, int rev) } if (!equiv_cpu_id) { - printk(KERN_ERR "microcode: CPU%d cpu_id " - "not found in equivalent cpu table\n", cpu); + printk(KERN_WARNING "microcode: CPU%d: cpu revision " + "not listed in equivalent cpu table\n", cpu); return 0; } if (mc_header->processor_rev_id != equiv_cpu_id) { - printk(KERN_ERR "microcode: CPU%d patch does not match " - "(processor_rev_id: %x, eqiv_cpu_id: %x)\n", + printk(KERN_ERR "microcode: CPU%d: patch mismatch " + "(processor_rev_id: %x, equiv_cpu_id: %x)\n", cpu, mc_header->processor_rev_id, equiv_cpu_id); return 0; } /* ucode might be chipset specific -- currently we don't support this */ if (mc_header->nb_dev_id || mc_header->sb_dev_id) { - printk(KERN_WARNING "microcode: CPU%d loading of chipset " + printk(KERN_ERR "microcode: CPU%d: loading of chipset " "specific code not yet supported\n", cpu); return 0; } @@ -172,15 +167,13 @@ static void apply_microcode_amd(int cpu) /* check current patch id and patch's id for match */ if (rev != mc_amd->hdr.patch_id) { - printk(KERN_ERR "microcode: CPU%d update from revision " - "0x%x to 0x%x failed\n", cpu_num, - mc_amd->hdr.patch_id, rev); + printk(KERN_ERR "microcode: CPU%d: update failed " + "(for patch_level=0x%x)\n", cpu, mc_amd->hdr.patch_id); return; } - printk(KERN_INFO "microcode: CPU%d updated from revision " - "0x%x to 0x%x\n", - cpu_num, uci->cpu_sig.rev, mc_amd->hdr.patch_id); + printk(KERN_INFO "microcode: CPU%d: updated (new patch_level=0x%x)\n", + cpu, rev); uci->cpu_sig.rev = rev; } @@ -202,18 +195,18 @@ static void *get_next_ucode(const u8 *buf, unsigned int size, return NULL; if (section_hdr[0] != UCODE_UCODE_TYPE) { - printk(KERN_ERR "microcode: error! " - "Wrong microcode payload type field\n"); + printk(KERN_ERR "microcode: error: invalid type field in " + "container file section header\n"); return NULL; } total_size = (unsigned long) (section_hdr[4] + (section_hdr[5] << 8)); - printk(KERN_INFO "microcode: size %u, total_size %u\n", - size, total_size); + printk(KERN_DEBUG "microcode: size %u, total_size %u\n", + size, total_size); if (total_size > size || total_size > UCODE_MAX_SIZE) { - printk(KERN_ERR "microcode: error! Bad data in microcode data file\n"); + printk(KERN_ERR "microcode: error: size mismatch\n"); return NULL; } @@ -243,14 +236,15 @@ static int install_equiv_cpu_table(const u8 *buf) size = buf_pos[2]; if (buf_pos[1] != UCODE_EQUIV_CPU_TABLE_TYPE || !size) { - printk(KERN_ERR "microcode: error! " - "Wrong microcode equivalent cpu table\n"); + printk(KERN_ERR "microcode: error: invalid type field in " + "container file section header\n"); return 0; } equiv_cpu_table = (struct equiv_cpu_entry *) vmalloc(size); if (!equiv_cpu_table) { - printk(KERN_ERR "microcode: error, can't allocate memory for equiv CPU table\n"); + printk(KERN_ERR "microcode: failed to allocate " + "equivalent CPU table\n"); return 0; } @@ -283,7 +277,8 @@ static int generic_load_microcode(int cpu, const u8 *data, size_t size) offset = install_equiv_cpu_table(ucode_ptr); if (!offset) { - printk(KERN_ERR "microcode: installing equivalent cpu table failed\n"); + printk(KERN_ERR "microcode: failed to create " + "equivalent cpu table\n"); return -EINVAL; } @@ -339,8 +334,7 @@ static int request_microcode_fw(int cpu, struct device *device) ret = request_firmware(&firmware, fw_name, device); if (ret) { - printk(KERN_ERR "microcode: ucode data file %s load failed\n", - fw_name); + printk(KERN_ERR "microcode: failed to load file %s\n", fw_name); return ret; } @@ -353,8 +347,8 @@ static int request_microcode_fw(int cpu, struct device *device) static int request_microcode_user(int cpu, const void __user *buf, size_t size) { - printk(KERN_WARNING "microcode: AMD microcode update via " - "/dev/cpu/microcode is not supported\n"); + printk(KERN_INFO "microcode: AMD microcode update via " + "/dev/cpu/microcode not supported\n"); return -1; } -- cgit v1.2.3 From bacbe9994541c70aa3abd1a013ac738e58d4bfb2 Mon Sep 17 00:00:00 2001 From: Janne Kulmala Date: Tue, 16 Dec 2008 13:39:57 +0200 Subject: x86: enable HPET on Fujitsu u9200 Impact: auto-enable HPET on Fujitsu u9200 HPET timer is listed in the ACPI table, but needs a quirk entry in order to work. Unfortunately, the quirk code runs after first HPET hpet_enable() which has already determined that the timer doesn't work (reads 0xFFFFFFFF). This patch allows hpet_enable() to be called again after running the quirk code. Signed-off-by: Janne Kulmala Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 2 +- arch/x86/kernel/quirks.c | 2 ++ 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 067d8de913f6..84089dc8fd1d 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -811,7 +811,7 @@ int __init hpet_enable(void) out_nohpet: hpet_clear_mapping(); - boot_hpet_disable = 1; + hpet_address = 0; return 0; } diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c index 67465ed89310..309949e9e1c1 100644 --- a/arch/x86/kernel/quirks.c +++ b/arch/x86/kernel/quirks.c @@ -168,6 +168,8 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_31, ich_force_enable_hpet); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_1, ich_force_enable_hpet); +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_4, + ich_force_enable_hpet); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH9_7, ich_force_enable_hpet); -- cgit v1.2.3 From 40fb17152c50a69dc304dd632131c2f41281ce44 Mon Sep 17 00:00:00 2001 From: Venki Pallipadi Date: Mon, 17 Nov 2008 16:11:37 -0800 Subject: x86: support always running TSC on Intel CPUs Impact: reward non-stop TSCs with good TSC-based clocksources, etc. Add support for CPUID_0x80000007_Bit8 on Intel CPUs as well. This bit means that the TSC is invariant with C/P/T states and always runs at constant frequency. With Intel CPUs, we have 3 classes * CPUs where TSC runs at constant rate and does not stop n C-states * CPUs where TSC runs at constant rate, but will stop in deep C-states * CPUs where TSC rate will vary based on P/T-states and TSC will stop in deep C-states. To cover these 3, one feature bit (CONSTANT_TSC) is not enough. So, add a second bit (NONSTOP_TSC). CONSTANT_TSC indicates that the TSC runs at constant frequency irrespective of P/T-states, and NONSTOP_TSC indicates that TSC does not stop in deep C-states. CPUID_0x8000000_Bit8 indicates both these feature bit can be set. We still have CONSTANT_TSC _set_ and NONSTOP_TSC _not_set_ on some older Intel CPUs, based on model checks. We can use TSC on such CPUs for time, as long as those CPUs do not support/enter deep C-states. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/amd.c | 9 +++++++-- arch/x86/kernel/cpu/intel.c | 10 ++++++++++ arch/x86/kernel/process.c | 2 +- 3 files changed, 18 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 8f1e31db2ad5..7c878f6aa919 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -283,9 +283,14 @@ static void __cpuinit early_init_amd(struct cpuinfo_x86 *c) { early_init_amd_mc(c); - /* c->x86_power is 8000_0007 edx. Bit 8 is constant TSC */ - if (c->x86_power & (1<<8)) + /* + * c->x86_power is 8000_0007 edx. Bit 8 is TSC runs at constant rate + * with P/T states and does not stop in deep C-states + */ + if (c->x86_power & (1 << 8)) { set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); + } #ifdef CONFIG_X86_64 set_cpu_cap(c, X86_FEATURE_SYSCALL32); diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index cce0b6118d55..caec59437a22 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -41,6 +41,16 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c) if (c->x86 == 15 && c->x86_cache_alignment == 64) c->x86_cache_alignment = 128; #endif + + /* + * c->x86_power is 8000_0007 edx. Bit 8 is TSC runs at constant rate + * with P/T states and does not stop in deep C-states + */ + if (c->x86_power & (1 << 8)) { + set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); + } + } #ifdef CONFIG_X86_32 diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c622772744d8..18c70fedba32 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -270,7 +270,7 @@ static void c1e_idle(void) rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi); if (lo & K8_INTP_C1E_ACTIVE_MASK) { c1e_detected = 1; - if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) + if (!boot_cpu_has(X86_FEATURE_NONSTOP_TSC)) mark_tsc_unstable("TSC halt in AMD C1E"); printk(KERN_INFO "System has AMD C1E enabled\n"); set_cpu_cap(&boot_cpu_data, X86_FEATURE_AMDC1E); -- cgit v1.2.3 From a9b43c7d9890066709609df849959009645c1a19 Mon Sep 17 00:00:00 2001 From: Jaswinder Singh Date: Mon, 15 Dec 2008 23:11:10 +0530 Subject: x86: setup.c find_and_reserve_crashkernel should be static Impact: cleanup, reduce kernel size a bit Signed-off-by: Jaswinder Singh Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 9d5674f7b6cc..81f5d22747ae 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -448,6 +448,7 @@ static void __init reserve_early_setup_data(void) * @size: Size of the crashkernel memory to reserve. * Returns the base address on success, and -1ULL on failure. */ +static unsigned long long __init find_and_reserve_crashkernel(unsigned long long size) { const unsigned long long alignment = 16<<20; /* 16M */ -- cgit v1.2.3 From a79b7a2a758c39315344f0d86b5adb21d90d786e Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:25 -0800 Subject: x86: remove unused iommu_nr_pages Impact: cleanup, remove dead code The last usage was removed by the patch set culminating in | commit e3c449f526cebb8d287241c7e82faafd9709668b | Author: Joerg Roedel | Date: Wed Oct 15 22:02:11 2008 -0700 | | x86, AMD IOMMU: convert driver to generic iommu_num_pages function Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-dma.c | 7 ------- 1 file changed, 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 192624820217..e150ad4f0ccc 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -125,13 +125,6 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } -unsigned long iommu_nr_pages(unsigned long addr, unsigned long len) -{ - unsigned long size = roundup((addr & ~PAGE_MASK) + len, PAGE_SIZE); - - return size >> PAGE_SHIFT; -} -EXPORT_SYMBOL(iommu_nr_pages); #endif void *dma_generic_alloc_coherent(struct device *dev, size_t size, -- cgit v1.2.3 From 39c04b55240342d0742ac48538d3d8c71bfc0a94 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:32:23 -0800 Subject: x86: make sure we really have an hpet mapping before using it Impact: prepare the hpet code for Xen dom0 booting When booting in Xen dom0, the hpet isn't really accessible, so make sure the mapping is non-NULL before use. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 84089dc8fd1d..a1f6ed5e1a05 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -834,10 +834,11 @@ static __init int hpet_late_init(void) hpet_address = force_hpet_address; hpet_enable(); - if (!hpet_virt_address) - return -ENODEV; } + if (!hpet_virt_address) + return -ENODEV; + hpet_reserve_platform_timers(hpet_readl(HPET_ID)); for_each_online_cpu(cpu) { -- cgit v1.2.3 From 8ae936690972dfcad73d0dde1095b9f32af5ee95 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Fri, 12 Dec 2008 15:52:26 -0800 Subject: x86: hardirq: use inc_irq_stat() in non-unified functions Impact: cleanup Replace incrementing irq stat with inc_irq_stat() in non-unified functions. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mcheck/mce_amd_64.c | 2 +- arch/x86/kernel/cpu/mcheck/mce_intel_64.c | 2 +- arch/x86/kernel/time_32.c | 2 +- arch/x86/kernel/time_64.c | 2 +- arch/x86/kernel/tlb_32.c | 2 +- arch/x86/kernel/tlb_64.c | 2 +- 6 files changed, 6 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c index 5eb390a4b2e9..748c8f9e7a05 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c @@ -237,7 +237,7 @@ asmlinkage void mce_threshold_interrupt(void) } } out: - add_pda(irq_threshold_count, 1); + inc_irq_stat(irq_threshold_count); irq_exit(); } diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c index c17eaf5dd6dd..4b48f251fd39 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c @@ -26,7 +26,7 @@ asmlinkage void smp_thermal_interrupt(void) if (therm_throt_process(msr_val & 1)) mce_log_therm_throt_event(smp_processor_id(), msr_val); - add_pda(irq_thermal_count, 1); + inc_irq_stat(irq_thermal_count); irq_exit(); } diff --git a/arch/x86/kernel/time_32.c b/arch/x86/kernel/time_32.c index 77b400f06ea2..65309e4cb1c0 100644 --- a/arch/x86/kernel/time_32.c +++ b/arch/x86/kernel/time_32.c @@ -75,7 +75,7 @@ EXPORT_SYMBOL(profile_pc); irqreturn_t timer_interrupt(int irq, void *dev_id) { /* Keep nmi watchdog up to date */ - per_cpu(irq_stat, smp_processor_id()).irq0_irqs++; + inc_irq_stat(irq0_irqs); #ifdef CONFIG_X86_IO_APIC if (timer_ack) { diff --git a/arch/x86/kernel/time_64.c b/arch/x86/kernel/time_64.c index 418a095c5796..1749cacde8b9 100644 --- a/arch/x86/kernel/time_64.c +++ b/arch/x86/kernel/time_64.c @@ -51,7 +51,7 @@ EXPORT_SYMBOL(profile_pc); irqreturn_t timer_interrupt(int irq, void *dev_id) { - add_pda(irq0_irqs, 1); + inc_irq_stat(irq0_irqs); global_clock_event->event_handler(global_clock_event); diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c index f4049f3513b6..f374f83fca42 100644 --- a/arch/x86/kernel/tlb_32.c +++ b/arch/x86/kernel/tlb_32.c @@ -119,7 +119,7 @@ void smp_invalidate_interrupt(struct pt_regs *regs) smp_mb__after_clear_bit(); out: put_cpu_no_resched(); - __get_cpu_var(irq_stat).irq_tlb_count++; + inc_irq_stat(irq_tlb_count); } void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, diff --git a/arch/x86/kernel/tlb_64.c b/arch/x86/kernel/tlb_64.c index 8f919ca69494..29887d7081a9 100644 --- a/arch/x86/kernel/tlb_64.c +++ b/arch/x86/kernel/tlb_64.c @@ -154,7 +154,7 @@ asmlinkage void smp_invalidate_interrupt(struct pt_regs *regs) out: ack_APIC_irq(); cpu_clear(cpu, f->flush_cpumask); - add_pda(irq_tlb_count, 1); + inc_irq_stat(irq_tlb_count); } void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, -- cgit v1.2.3 From fde9071167c4624281553b23232aa8b81e71c790 Mon Sep 17 00:00:00 2001 From: Zachary Amsden Date: Fri, 12 Dec 2008 11:26:35 -0800 Subject: x86: clean up dead code in vmi_32.c Impact: cleanup, remove dead debug code I ran across some old debugging code in vmi paravirt-ops code that was already dead, but still potentially useful. After reviewing recent changes to the way kernel page tables are allocated and initialized, and the lack of bugs caught by this debugging code, I've concluded it is now totally useless to have around, and it's already been #if 0'd for quite some time. There's no rush to get this in mainline, but it's also totally harmless, so I'll let the x86 maintainers decide where it should be tucked. I've been out of the mainstream dev loop for a couple months, so apologies if I haven't got any protocol changes in order. Remove mummified remains found in vmi_32.c Signed-off-by: Zachary Amsden Signed-off-by: Ingo Molnar --- arch/x86/kernel/vmi_32.c | 119 ----------------------------------------------- 1 file changed, 119 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c index 8b6c393ab9fd..8087e0cd877d 100644 --- a/arch/x86/kernel/vmi_32.c +++ b/arch/x86/kernel/vmi_32.c @@ -266,109 +266,6 @@ static void vmi_nop(void) { } -#ifdef CONFIG_DEBUG_PAGE_TYPE - -#ifdef CONFIG_X86_PAE -#define MAX_BOOT_PTS (2048+4+1) -#else -#define MAX_BOOT_PTS (1024+1) -#endif - -/* - * During boot, mem_map is not yet available in paging_init, so stash - * all the boot page allocations here. - */ -static struct { - u32 pfn; - int type; -} boot_page_allocations[MAX_BOOT_PTS]; -static int num_boot_page_allocations; -static int boot_allocations_applied; - -void vmi_apply_boot_page_allocations(void) -{ - int i; - BUG_ON(!mem_map); - for (i = 0; i < num_boot_page_allocations; i++) { - struct page *page = pfn_to_page(boot_page_allocations[i].pfn); - page->type = boot_page_allocations[i].type; - page->type = boot_page_allocations[i].type & - ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE); - } - boot_allocations_applied = 1; -} - -static void record_page_type(u32 pfn, int type) -{ - BUG_ON(num_boot_page_allocations >= MAX_BOOT_PTS); - boot_page_allocations[num_boot_page_allocations].pfn = pfn; - boot_page_allocations[num_boot_page_allocations].type = type; - num_boot_page_allocations++; -} - -static void check_zeroed_page(u32 pfn, int type, struct page *page) -{ - u32 *ptr; - int i; - int limit = PAGE_SIZE / sizeof(int); - - if (page_address(page)) - ptr = (u32 *)page_address(page); - else - ptr = (u32 *)__va(pfn << PAGE_SHIFT); - /* - * When cloning the root in non-PAE mode, only the userspace - * pdes need to be zeroed. - */ - if (type & VMI_PAGE_CLONE) - limit = KERNEL_PGD_BOUNDARY; - for (i = 0; i < limit; i++) - BUG_ON(ptr[i]); -} - -/* - * We stash the page type into struct page so we can verify the page - * types are used properly. - */ -static void vmi_set_page_type(u32 pfn, int type) -{ - /* PAE can have multiple roots per page - don't track */ - if (PTRS_PER_PMD > 1 && (type & VMI_PAGE_PDP)) - return; - - if (boot_allocations_applied) { - struct page *page = pfn_to_page(pfn); - if (type != VMI_PAGE_NORMAL) - BUG_ON(page->type); - else - BUG_ON(page->type == VMI_PAGE_NORMAL); - page->type = type & ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE); - if (type & VMI_PAGE_ZEROED) - check_zeroed_page(pfn, type, page); - } else { - record_page_type(pfn, type); - } -} - -static void vmi_check_page_type(u32 pfn, int type) -{ - /* PAE can have multiple roots per page - skip checks */ - if (PTRS_PER_PMD > 1 && (type & VMI_PAGE_PDP)) - return; - - type &= ~(VMI_PAGE_ZEROED | VMI_PAGE_CLONE); - if (boot_allocations_applied) { - struct page *page = pfn_to_page(pfn); - BUG_ON((page->type ^ type) & VMI_PAGE_PAE); - BUG_ON(type == VMI_PAGE_NORMAL && page->type); - BUG_ON((type & page->type) == 0); - } -} -#else -#define vmi_set_page_type(p,t) do { } while (0) -#define vmi_check_page_type(p,t) do { } while (0) -#endif - #ifdef CONFIG_HIGHPTE static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type) { @@ -395,7 +292,6 @@ static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type) static void vmi_allocate_pte(struct mm_struct *mm, unsigned long pfn) { - vmi_set_page_type(pfn, VMI_PAGE_L1); vmi_ops.allocate_page(pfn, VMI_PAGE_L1, 0, 0, 0); } @@ -406,27 +302,22 @@ static void vmi_allocate_pmd(struct mm_struct *mm, unsigned long pfn) * It is called only for swapper_pg_dir, which already has * data on it. */ - vmi_set_page_type(pfn, VMI_PAGE_L2); vmi_ops.allocate_page(pfn, VMI_PAGE_L2, 0, 0, 0); } static void vmi_allocate_pmd_clone(unsigned long pfn, unsigned long clonepfn, unsigned long start, unsigned long count) { - vmi_set_page_type(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE); - vmi_check_page_type(clonepfn, VMI_PAGE_L2); vmi_ops.allocate_page(pfn, VMI_PAGE_L2 | VMI_PAGE_CLONE, clonepfn, start, count); } static void vmi_release_pte(unsigned long pfn) { vmi_ops.release_page(pfn, VMI_PAGE_L1); - vmi_set_page_type(pfn, VMI_PAGE_NORMAL); } static void vmi_release_pmd(unsigned long pfn) { vmi_ops.release_page(pfn, VMI_PAGE_L2); - vmi_set_page_type(pfn, VMI_PAGE_NORMAL); } /* @@ -450,26 +341,22 @@ static void vmi_release_pmd(unsigned long pfn) static void vmi_update_pte(struct mm_struct *mm, unsigned long addr, pte_t *ptep) { - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE); vmi_ops.update_pte(ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); } static void vmi_update_pte_defer(struct mm_struct *mm, unsigned long addr, pte_t *ptep) { - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE); vmi_ops.update_pte(ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 0)); } static void vmi_set_pte(pte_t *ptep, pte_t pte) { /* XXX because of set_pmd_pte, this can be called on PT or PD layers */ - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE | VMI_PAGE_PD); vmi_ops.set_pte(pte, ptep, VMI_PAGE_PT); } static void vmi_set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte) { - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE); vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); } @@ -477,10 +364,8 @@ static void vmi_set_pmd(pmd_t *pmdp, pmd_t pmdval) { #ifdef CONFIG_X86_PAE const pte_t pte = { .pte = pmdval.pmd }; - vmi_check_page_type(__pa(pmdp) >> PAGE_SHIFT, VMI_PAGE_PMD); #else const pte_t pte = { pmdval.pud.pgd.pgd }; - vmi_check_page_type(__pa(pmdp) >> PAGE_SHIFT, VMI_PAGE_PGD); #endif vmi_ops.set_pte(pte, (pte_t *)pmdp, VMI_PAGE_PD); } @@ -502,7 +387,6 @@ static void vmi_set_pte_atomic(pte_t *ptep, pte_t pteval) static void vmi_set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte) { - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE); vmi_ops.set_pte(pte, ptep, vmi_flags_addr_defer(mm, addr, VMI_PAGE_PT, 1)); } @@ -510,21 +394,18 @@ static void vmi_set_pud(pud_t *pudp, pud_t pudval) { /* Um, eww */ const pte_t pte = { .pte = pudval.pgd.pgd }; - vmi_check_page_type(__pa(pudp) >> PAGE_SHIFT, VMI_PAGE_PGD); vmi_ops.set_pte(pte, (pte_t *)pudp, VMI_PAGE_PDP); } static void vmi_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep) { const pte_t pte = { .pte = 0 }; - vmi_check_page_type(__pa(ptep) >> PAGE_SHIFT, VMI_PAGE_PTE); vmi_ops.set_pte(pte, ptep, vmi_flags_addr(mm, addr, VMI_PAGE_PT, 0)); } static void vmi_pmd_clear(pmd_t *pmd) { const pte_t pte = { .pte = 0 }; - vmi_check_page_type(__pa(pmd) >> PAGE_SHIFT, VMI_PAGE_PMD); vmi_ops.set_pte(pte, (pte_t *)pmd, VMI_PAGE_PD); } #endif -- cgit v1.2.3 From 189f67c4408806563a1f061f5c8bf184a6658477 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Fri, 12 Dec 2008 14:50:40 -0600 Subject: x86: UV fix for global physical addresses Impact: fix UV boot crash This fixes a UV bug related to generating global memory addresses on partitioned systems. Partition systems do not have physical memory at address 0. Instead, a chunk of high memory is remapped by the chipset so that it appears to be at address 0. This remapping is INVISIBLE to most of the OS. The only OS functions that need to be aware of the remaping are functions that directly interface to the chipset. The GRU is one example. Also, delete a couple of unused macros related to global memory addresses. Signed-off-by: Jack Steiner Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 221299f4509f..dece17289731 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -540,8 +540,7 @@ void __init uv_system_init(void) uv_blade_info[blade].nr_possible_cpus++; uv_cpu_hub_info(cpu)->lowmem_remap_base = lowmem_redir_base; - uv_cpu_hub_info(cpu)->lowmem_remap_top = - lowmem_redir_base + lowmem_redir_size; + uv_cpu_hub_info(cpu)->lowmem_remap_top = lowmem_redir_size; uv_cpu_hub_info(cpu)->m_val = m_val; uv_cpu_hub_info(cpu)->n_val = m_val; uv_cpu_hub_info(cpu)->numa_blade_id = blade; -- cgit v1.2.3 From cf9b303e55da810255638c0b616b1a3f7eda9320 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Mon, 15 Dec 2008 23:33:10 +0100 Subject: x86: re-enable MCE on secondary CPUS after suspend/resume Impact: fix disabled MCE after resume Don't prevent multiple initialization of MCEs. Back from early prehistory mcheck_init() has a reentry check. Presumably that was needed in very old kernels to prevent it entering twice. But as Andreas points out this prevents CPU hotplug (and therefore resume) to correctly reinitialize MCEs when a AP boots again after being offlined. Just drop the check. Reported-by: Andreas Herrmann Signed-off-by: Andi Kleen Tested-by: Andreas Herrmann Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mcheck/mce_64.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c index 4b031a4ac856..1c838032fd37 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_64.c @@ -510,12 +510,9 @@ static void __cpuinit mce_cpu_features(struct cpuinfo_x86 *c) */ void __cpuinit mcheck_init(struct cpuinfo_x86 *c) { - static cpumask_t mce_cpus = CPU_MASK_NONE; - mce_cpu_quirks(c); if (mce_dont_init || - cpu_test_and_set(smp_processor_id(), mce_cpus) || !mce_available(c)) return; -- cgit v1.2.3 From c8182f0016fb65a721c4fbe487909a2d56178135 Mon Sep 17 00:00:00 2001 From: Russ Anderson Date: Fri, 12 Dec 2008 11:07:00 -0600 Subject: sgi-xp: xpc needs to pass the physical address, not virtual Impact: fix crash xpc needs to pass the physical address, not virtual. Testing uncovered this problem. The virtual address happens to work most of the time due to the way bios was masking off the node bits. Passing the physical address makes it work all of the time. Signed-off-by: Russ Anderson Acked-by: Dean Nelson Signed-off-by: Ingo Molnar --- arch/x86/kernel/bios_uv.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index d22d0f1bbea0..2a0a2a3cac26 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -101,15 +101,13 @@ s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher, } int -uv_bios_mq_watchlist_alloc(int blade, void *mq, unsigned int mq_size, +uv_bios_mq_watchlist_alloc(int blade, unsigned long addr, unsigned int mq_size, unsigned long *intr_mmr_offset) { union uv_watchlist_u size_blade; - unsigned long addr; u64 watchlist; s64 ret; - addr = (unsigned long)mq; size_blade.size = mq_size; size_blade.blade = blade; -- cgit v1.2.3 From ae417bb487e3bb88dc862b83b4bf00d87ba67ec8 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Tue, 16 Dec 2008 14:02:16 -0800 Subject: x86: signal: use signal_fault() in sys_sigreturn() Impact: cleanup Call signal_fault() in error route of sys_sigreturn(). Change log level to KERN_EMERG if current is init. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal.c | 15 +++------------ 1 file changed, 3 insertions(+), 12 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index b1cc6da64208..2725a294d734 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -594,17 +594,7 @@ asmlinkage unsigned long sys_sigreturn(unsigned long __unused) return ax; badframe: - if (show_unhandled_signals && printk_ratelimit()) { - printk("%s%s[%d] bad frame in sigreturn frame:" - "%p ip:%lx sp:%lx oeax:%lx", - task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, - current->comm, task_pid_nr(current), frame, regs->ip, - regs->sp, regs->orig_ax); - print_vma_addr(" in ", regs->ip); - printk(KERN_CONT "\n"); - } - - force_sig(SIGSEGV, current); + signal_fault(regs, frame, "sigreturn"); return 0; } @@ -901,8 +891,9 @@ void signal_fault(struct pt_regs *regs, void __user *frame, char *where) struct task_struct *me = current; if (show_unhandled_signals && printk_ratelimit()) { - printk(KERN_INFO + printk("%s" "%s[%d] bad frame in %s frame:%p ip:%lx sp:%lx orax:%lx", + task_pid_nr(current) > 1 ? KERN_INFO : KERN_EMERG, me->comm, me->pid, where, frame, regs->ip, regs->sp, regs->orig_ax); print_vma_addr(" in ", regs->ip); -- cgit v1.2.3 From 48a1b10aff588833b73994704c47bbd0deb73e9c Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 11 Dec 2008 00:15:01 -0800 Subject: x86, sparseirq: move irq_desc according to smp_affinity, v7 Impact: improve NUMA handling by migrating irq_desc on smp_affinity changes if CONFIG_NUMA_MIGRATE_IRQ_DESC is set: - make irq_desc to go with affinity aka irq_desc moving etc - call move_irq_desc in irq_complete_move() - legacy irq_desc is not moved, because they are allocated via static array for logical apic mode, need to add move_desc_in_progress_in_same_domain, otherwise it will not be moved ==> also could need two phases to get irq_desc moved. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 142 +++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 141 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index a1a2e070f31a..bfe1245b1a3e 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -141,6 +141,9 @@ struct irq_cfg { unsigned move_cleanup_count; u8 vector; u8 move_in_progress : 1; +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + u8 move_desc_pending : 1; +#endif }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ @@ -223,6 +226,121 @@ void arch_init_chip_data(struct irq_desc *desc, int cpu) } } +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + +static void +init_copy_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg, int cpu) +{ + struct irq_pin_list *old_entry, *head, *tail, *entry; + + cfg->irq_2_pin = NULL; + old_entry = old_cfg->irq_2_pin; + if (!old_entry) + return; + + entry = get_one_free_irq_2_pin(cpu); + if (!entry) + return; + + entry->apic = old_entry->apic; + entry->pin = old_entry->pin; + head = entry; + tail = entry; + old_entry = old_entry->next; + while (old_entry) { + entry = get_one_free_irq_2_pin(cpu); + if (!entry) { + entry = head; + while (entry) { + head = entry->next; + kfree(entry); + entry = head; + } + /* still use the old one */ + return; + } + entry->apic = old_entry->apic; + entry->pin = old_entry->pin; + tail->next = entry; + tail = entry; + old_entry = old_entry->next; + } + + tail->next = NULL; + cfg->irq_2_pin = head; +} + +static void free_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg) +{ + struct irq_pin_list *entry, *next; + + if (old_cfg->irq_2_pin == cfg->irq_2_pin) + return; + + entry = old_cfg->irq_2_pin; + + while (entry) { + next = entry->next; + kfree(entry); + entry = next; + } + old_cfg->irq_2_pin = NULL; +} + +void arch_init_copy_chip_data(struct irq_desc *old_desc, + struct irq_desc *desc, int cpu) +{ + struct irq_cfg *cfg; + struct irq_cfg *old_cfg; + + cfg = get_one_free_irq_cfg(cpu); + + if (!cfg) + return; + + desc->chip_data = cfg; + + old_cfg = old_desc->chip_data; + + memcpy(cfg, old_cfg, sizeof(struct irq_cfg)); + + init_copy_irq_2_pin(old_cfg, cfg, cpu); +} + +static void free_irq_cfg(struct irq_cfg *old_cfg) +{ + kfree(old_cfg); +} + +void arch_free_chip_data(struct irq_desc *old_desc, struct irq_desc *desc) +{ + struct irq_cfg *old_cfg, *cfg; + + old_cfg = old_desc->chip_data; + cfg = desc->chip_data; + + if (old_cfg == cfg) + return; + + if (old_cfg) { + free_irq_2_pin(old_cfg, cfg); + free_irq_cfg(old_cfg); + old_desc->chip_data = NULL; + } +} + +static void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) +{ + struct irq_cfg *cfg = desc->chip_data; + + if (!cfg->move_in_progress) { + /* it means that domain is not changed */ + if (!cpus_intersects(desc->affinity, mask)) + cfg->move_desc_pending = 1; + } +} +#endif + #else static struct irq_cfg *irq_cfg(unsigned int irq) { @@ -231,9 +349,11 @@ static struct irq_cfg *irq_cfg(unsigned int irq) #endif +#ifndef CONFIG_NUMA_MIGRATE_IRQ_DESC static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) { } +#endif struct io_apic { unsigned int index; @@ -2346,14 +2466,34 @@ static void irq_complete_move(struct irq_desc **descp) struct irq_cfg *cfg = desc->chip_data; unsigned vector, me; - if (likely(!cfg->move_in_progress)) + if (likely(!cfg->move_in_progress)) { +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + if (likely(!cfg->move_desc_pending)) + return; + + /* domain is not change, but affinity is changed */ + me = smp_processor_id(); + if (cpu_isset(me, desc->affinity)) { + *descp = desc = move_irq_desc(desc, me); + /* get the new one */ + cfg = desc->chip_data; + cfg->move_desc_pending = 0; + } +#endif return; + } vector = ~get_irq_regs()->orig_ax; me = smp_processor_id(); if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { cpumask_t cleanup_mask; +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + *descp = desc = move_irq_desc(desc, me); + /* get the new one */ + cfg = desc->chip_data; +#endif + cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); cfg->move_cleanup_count = cpus_weight(cleanup_mask); send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); -- cgit v1.2.3 From d680fe44775ed17a80035462d9898f5e77bfd7dd Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sat, 13 Dec 2008 00:09:08 +0300 Subject: x86: entry_64 - introduce FTRACE_ frame macro v2 Impact: clean up Itroduce MCOUNT_SAVE/RESTORE_FRAME which allow us to save a number of lines on source level. Also fix a comment in ftrace.h. Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_64.S | 57 ++++++---------------------------------------- 1 file changed, 7 insertions(+), 50 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 54e0bbdccb99..303dd84d2a98 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -71,15 +71,7 @@ ENTRY(ftrace_caller) cmpl $0, function_trace_stop jne ftrace_stub - /* taken from glibc */ - subq $0x38, %rsp - movq %rax, (%rsp) - movq %rcx, 8(%rsp) - movq %rdx, 16(%rsp) - movq %rsi, 24(%rsp) - movq %rdi, 32(%rsp) - movq %r8, 40(%rsp) - movq %r9, 48(%rsp) + MCOUNT_SAVE_FRAME movq 0x38(%rsp), %rdi movq 8(%rbp), %rsi @@ -89,14 +81,7 @@ ENTRY(ftrace_caller) ftrace_call: call ftrace_stub - movq 48(%rsp), %r9 - movq 40(%rsp), %r8 - movq 32(%rsp), %rdi - movq 24(%rsp), %rsi - movq 16(%rsp), %rdx - movq 8(%rsp), %rcx - movq (%rsp), %rax - addq $0x38, %rsp + MCOUNT_RESTORE_FRAME #ifdef CONFIG_FUNCTION_GRAPH_TRACER .globl ftrace_graph_call @@ -130,15 +115,7 @@ ftrace_stub: retq trace: - /* taken from glibc */ - subq $0x38, %rsp - movq %rax, (%rsp) - movq %rcx, 8(%rsp) - movq %rdx, 16(%rsp) - movq %rsi, 24(%rsp) - movq %rdi, 32(%rsp) - movq %r8, 40(%rsp) - movq %r9, 48(%rsp) + MCOUNT_SAVE_FRAME movq 0x38(%rsp), %rdi movq 8(%rbp), %rsi @@ -146,14 +123,7 @@ trace: call *ftrace_trace_function - movq 48(%rsp), %r9 - movq 40(%rsp), %r8 - movq 32(%rsp), %rdi - movq 24(%rsp), %rsi - movq 16(%rsp), %rdx - movq 8(%rsp), %rcx - movq (%rsp), %rax - addq $0x38, %rsp + MCOUNT_RESTORE_FRAME jmp ftrace_stub END(mcount) @@ -165,14 +135,7 @@ ENTRY(ftrace_graph_caller) cmpl $0, function_trace_stop jne ftrace_stub - subq $0x38, %rsp - movq %rax, (%rsp) - movq %rcx, 8(%rsp) - movq %rdx, 16(%rsp) - movq %rsi, 24(%rsp) - movq %rdi, 32(%rsp) - movq %r8, 40(%rsp) - movq %r9, 48(%rsp) + MCOUNT_SAVE_FRAME leaq 8(%rbp), %rdi movq 0x38(%rsp), %rsi @@ -180,14 +143,8 @@ ENTRY(ftrace_graph_caller) call prepare_ftrace_return - movq 48(%rsp), %r9 - movq 40(%rsp), %r8 - movq 32(%rsp), %rdi - movq 24(%rsp), %rsi - movq 16(%rsp), %rdx - movq 8(%rsp), %rcx - movq (%rsp), %rax - addq $0x38, %rsp + MCOUNT_RESTORE_FRAME + retq END(ftrace_graph_caller) -- cgit v1.2.3 From cf558d25e5c9f70fa0279c9b7b8b4aed7cae9bd4 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 17 Dec 2008 15:06:01 +0100 Subject: AMD IOMMU: set cmd buffer pointers to zero manually Impact: set cmd buffer head and tail pointers to zero in case nobody else did Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu_init.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index c90a15eba5c5..c6cc22815d35 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -427,6 +427,10 @@ static u8 * __init alloc_command_buffer(struct amd_iommu *iommu) memcpy_toio(iommu->mmio_base + MMIO_CMD_BUF_OFFSET, &entry, sizeof(entry)); + /* set head and tail to zero manually */ + writel(0x00, iommu->mmio_base + MMIO_CMD_HEAD_OFFSET); + writel(0x00, iommu->mmio_base + MMIO_CMD_TAIL_OFFSET); + iommu_feature_enable(iommu, CONTROL_CMDBUF_EN); return cmd_buf; -- cgit v1.2.3 From 84df81759590ad16b0024cf46b3423cca76b2e07 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Wed, 17 Dec 2008 16:36:44 +0100 Subject: AMD IOMMU: panic if completion wait loop fails Impact: prevents data corruption after a failed completion wait loop Signed-off-by: Joerg Roedel --- arch/x86/kernel/amd_iommu.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index a7b6dec6fc3f..0a60d60ed036 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -235,8 +235,9 @@ static int iommu_completion_wait(struct amd_iommu *iommu) status &= ~MMIO_STATUS_COM_WAIT_INT_MASK; writel(status, iommu->mmio_base + MMIO_STATUS_OFFSET); - if (unlikely((i == EXIT_LOOP_COUNT) && printk_ratelimit())) - printk(KERN_WARNING "AMD IOMMU: Completion wait loop failed\n"); + if (unlikely(i == EXIT_LOOP_COUNT)) + panic("AMD IOMMU: Completion wait loop failed\n"); + out: spin_unlock_irqrestore(&iommu->lock, flags); -- cgit v1.2.3 From 8ce7996009bab7b2d23e7af7ad831fed7eb6faa1 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:35 -0800 Subject: x86: add swiotlb allocation functions Add x86-specific swiotlb allocation functions. These are purely default for the moment. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index 3c539d111abb..f47a097a135b 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -3,6 +3,8 @@ #include #include #include +#include +#include #include #include @@ -11,6 +13,16 @@ int swiotlb __read_mostly; +void *swiotlb_alloc_boot(size_t size, unsigned long nslabs) +{ + return alloc_bootmem_low_pages(size); +} + +void *swiotlb_alloc(unsigned order, unsigned long nslabs) +{ + return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order); +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From cfb80c9eae8c7ed8f2ee81090062d15ead51cbe8 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:36 -0800 Subject: x86: unify pci iommu setup and allow swiotlb to compile for 32 bit swiotlb on 32 bit will be used by Xen domain 0 support. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/Makefile | 3 ++- arch/x86/kernel/pci-dma.c | 6 ++++-- arch/x86/kernel/pci-swiotlb_64.c | 2 ++ 3 files changed, 8 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index b62a7667828e..a9c656f2d661 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -105,6 +105,8 @@ microcode-$(CONFIG_MICROCODE_INTEL) += microcode_intel.o microcode-$(CONFIG_MICROCODE_AMD) += microcode_amd.o obj-$(CONFIG_MICROCODE) += microcode.o +obj-$(CONFIG_SWIOTLB) += pci-swiotlb_64.o # NB rename without _64 + ### # 64 bit specific files ifeq ($(CONFIG_X86_64),y) @@ -118,7 +120,6 @@ ifeq ($(CONFIG_X86_64),y) obj-$(CONFIG_GART_IOMMU) += pci-gart_64.o aperture_64.o obj-$(CONFIG_CALGARY_IOMMU) += pci-calgary_64.o tce_64.o obj-$(CONFIG_AMD_IOMMU) += amd_iommu_init.o amd_iommu.o - obj-$(CONFIG_SWIOTLB) += pci-swiotlb_64.o obj-$(CONFIG_PCI_MMCONFIG) += mmconf-fam10h_64.o endif diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index e150ad4f0ccc..00e07447a5bd 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -105,11 +105,15 @@ static void __init dma32_free_bootmem(void) dma32_bootmem_ptr = NULL; dma32_bootmem_size = 0; } +#endif void __init pci_iommu_alloc(void) { +#ifdef CONFIG_X86_64 /* free the range so iommu could get some range less than 4G */ dma32_free_bootmem(); +#endif + /* * The order of these functions is important for * fall-back/fail-over reasons @@ -125,8 +129,6 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } -#endif - void *dma_generic_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_addr, gfp_t flag) { diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index f47a097a135b..a991afea6700 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -62,8 +62,10 @@ struct dma_mapping_ops swiotlb_dma_ops = { void __init pci_swiotlb_init(void) { /* don't initialize swiotlb if iommu=off (no_iommu=1) */ +#ifdef CONFIG_X86_64 if (!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN) swiotlb = 1; +#endif if (swiotlb_force) swiotlb = 1; if (swiotlb) { -- cgit v1.2.3 From 1d32251e846ccbcf9d2da041dffd1199f94b2a3b Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Tue, 16 Dec 2008 12:17:37 -0800 Subject: x86/swiotlb: add default phys<->bus conversion Xen will override these later on. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index a991afea6700..93a8371f2c22 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -23,6 +23,16 @@ void *swiotlb_alloc(unsigned order, unsigned long nslabs) return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order); } +dma_addr_t swiotlb_phys_to_bus(phys_addr_t paddr) +{ + return paddr; +} + +phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr) +{ + return baddr; +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From a08636690d06b2e36cfb4c2b3ee133a81c47e1e0 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Tue, 16 Dec 2008 12:17:38 -0800 Subject: x86/swiotlb: add default swiotlb_arch_range_needs_mapping Xen will override these later on. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index 93a8371f2c22..242c3440687f 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -33,6 +33,11 @@ phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr) return baddr; } +int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size) +{ + return 0; +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From f5223763a664da16771211f9d293e18cb242b246 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 17 Dec 2008 18:47:17 -0800 Subject: x86: signal: move ia32 func declarations into arch/x86/kernel/signal.c Impact: cleanup Move declarations of ia32_setup_rt_frame() and ia32_setup_frame() into arch/x86/kernel/signal.c. This is for future use of sigframe.h. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/sigframe.h | 5 ----- arch/x86/kernel/signal.c | 5 +++++ 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/sigframe.h b/arch/x86/kernel/sigframe.h index cc673aa55ce4..6dd7e2b70a4b 100644 --- a/arch/x86/kernel/sigframe.h +++ b/arch/x86/kernel/sigframe.h @@ -34,9 +34,4 @@ struct rt_sigframe { struct siginfo info; /* fp state follows here */ }; - -int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, - sigset_t *set, struct pt_regs *regs); -int ia32_setup_frame(int sig, struct k_sigaction *ka, - sigset_t *set, struct pt_regs *regs); #endif diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 2725a294d734..848c2d64a289 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -671,6 +671,11 @@ static int signr_convert(int sig) #define is_ia32 0 #endif /* CONFIG_IA32_EMULATION */ +int ia32_setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, + sigset_t *set, struct pt_regs *regs); +int ia32_setup_frame(int sig, struct k_sigaction *ka, + sigset_t *set, struct pt_regs *regs); + #endif /* CONFIG_X86_32 */ static int -- cgit v1.2.3 From a5c56eb36f999ae0ecac278e51fd1cf8feb16c2f Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 17 Dec 2008 18:49:55 -0800 Subject: x86: signal: rename sigframe and rt_sigframe on 32-bit Impact: cleanup, prepare to move sigframe.h On 32-bit, rename struct sigrame to struct sigframe_ia32, struct rt_sigframe to struct rt_sigframe_ia32 and several structures. And add helper macros to access the above data in arch/x86/kernel/signal.c. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/sigframe.h | 28 +++++++++++++++++----------- 1 file changed, 17 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/sigframe.h b/arch/x86/kernel/sigframe.h index 6dd7e2b70a4b..6718ed04b05b 100644 --- a/arch/x86/kernel/sigframe.h +++ b/arch/x86/kernel/sigframe.h @@ -1,8 +1,14 @@ #ifdef CONFIG_X86_32 -struct sigframe { - char __user *pretcode; +#define sigframe_ia32 sigframe +#define rt_sigframe_ia32 rt_sigframe +#define sigcontext_ia32 sigcontext +#define _fpstate_ia32 _fpstate +#define ucontext_ia32 ucontext + +struct sigframe_ia32 { + u32 pretcode; int sig; - struct sigcontext sc; + struct sigcontext_ia32 sc; /* * fpstate is unused. fpstate is moved/allocated after * retcode[] below. This movement allows to have the FP state and the @@ -11,27 +17,27 @@ struct sigframe { * the offset of extramask[] in the sigframe and thus prevent any * legacy application accessing/modifying it. */ - struct _fpstate fpstate_unused; + struct _fpstate_ia32 fpstate_unused; unsigned long extramask[_NSIG_WORDS-1]; char retcode[8]; /* fp state follows here */ }; -struct rt_sigframe { - char __user *pretcode; +struct rt_sigframe_ia32 { + u32 pretcode; int sig; - struct siginfo __user *pinfo; - void __user *puc; + u32 pinfo; + u32 puc; struct siginfo info; - struct ucontext uc; + struct ucontext_ia32 uc; char retcode[8]; /* fp state follows here */ }; -#else +#else /* !CONFIG_X86_32 */ struct rt_sigframe { char __user *pretcode; struct ucontext uc; struct siginfo info; /* fp state follows here */ }; -#endif +#endif /* CONFIG_X86_32 */ -- cgit v1.2.3 From 41af86fad3c40646b9748279e3862781e937a5d2 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 17 Dec 2008 18:50:32 -0800 Subject: x86: signal: move sigframe.h to arch/x86/include/asm Impact: cleanup, move header file Move arch/x86/kernel/sigframe.h to arch/x86/include/asm/sigframe.h. It will be used in arch/x86/ia32/ia32_signal.c. Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- arch/x86/kernel/asm-offsets_32.c | 2 +- arch/x86/kernel/sigframe.h | 43 ---------------------------------------- arch/x86/kernel/signal.c | 2 +- 3 files changed, 2 insertions(+), 45 deletions(-) delete mode 100644 arch/x86/kernel/sigframe.h (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/asm-offsets_32.c b/arch/x86/kernel/asm-offsets_32.c index 6649d09ad88f..ee4df08feee6 100644 --- a/arch/x86/kernel/asm-offsets_32.c +++ b/arch/x86/kernel/asm-offsets_32.c @@ -11,7 +11,7 @@ #include #include #include -#include "sigframe.h" +#include #include #include #include diff --git a/arch/x86/kernel/sigframe.h b/arch/x86/kernel/sigframe.h deleted file mode 100644 index 6718ed04b05b..000000000000 --- a/arch/x86/kernel/sigframe.h +++ /dev/null @@ -1,43 +0,0 @@ -#ifdef CONFIG_X86_32 -#define sigframe_ia32 sigframe -#define rt_sigframe_ia32 rt_sigframe -#define sigcontext_ia32 sigcontext -#define _fpstate_ia32 _fpstate -#define ucontext_ia32 ucontext - -struct sigframe_ia32 { - u32 pretcode; - int sig; - struct sigcontext_ia32 sc; - /* - * fpstate is unused. fpstate is moved/allocated after - * retcode[] below. This movement allows to have the FP state and the - * future state extensions (xsave) stay together. - * And at the same time retaining the unused fpstate, prevents changing - * the offset of extramask[] in the sigframe and thus prevent any - * legacy application accessing/modifying it. - */ - struct _fpstate_ia32 fpstate_unused; - unsigned long extramask[_NSIG_WORDS-1]; - char retcode[8]; - /* fp state follows here */ -}; - -struct rt_sigframe_ia32 { - u32 pretcode; - int sig; - u32 pinfo; - u32 puc; - struct siginfo info; - struct ucontext_ia32 uc; - char retcode[8]; - /* fp state follows here */ -}; -#else /* !CONFIG_X86_32 */ -struct rt_sigframe { - char __user *pretcode; - struct ucontext uc; - struct siginfo info; - /* fp state follows here */ -}; -#endif /* CONFIG_X86_32 */ diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 848c2d64a289..89bb7668041d 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -35,7 +35,7 @@ #include #include -#include "sigframe.h" +#include #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -- cgit v1.2.3 From 55aab5f49e384a361668d112eefdb33e90779af9 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Wed, 17 Dec 2008 12:52:34 -0700 Subject: x86 gart: don't complain if no AMD GART found Impact: remove annoying bootup printk It's perfectly normal for no AMD GART to be present, e.g., if you have Intel CPUs. None of the other iommu_init() functions makes noise when it finds nothing. Signed-off-by: Bjorn Helgaas Acked-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-gart_64.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index ba7ad83e20a8..a35eaa379ff6 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -745,10 +745,8 @@ void __init gart_iommu_init(void) unsigned long scratch; long i; - if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) { - printk(KERN_INFO "PCI-GART: No AMD GART found.\n"); + if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) return; - } #ifndef CONFIG_AGP_AMD64 no_agp = 1; -- cgit v1.2.3 From 57a37505d19f4dfeee26f0fd7ea38ed6f1d10cbe Mon Sep 17 00:00:00 2001 From: Jaswinder Singh Date: Wed, 17 Dec 2008 23:17:21 +0530 Subject: x86: time_64.c timer_interrupt() should be static Impact: cleanup, reduce kernel size a bit Signed-off-by: Jaswinder Singh Signed-off-by: Ingo Molnar --- arch/x86/kernel/time_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/time_64.c b/arch/x86/kernel/time_64.c index cb19d650c216..083a4a5bb00f 100644 --- a/arch/x86/kernel/time_64.c +++ b/arch/x86/kernel/time_64.c @@ -49,7 +49,7 @@ unsigned long profile_pc(struct pt_regs *regs) } EXPORT_SYMBOL(profile_pc); -irqreturn_t timer_interrupt(int irq, void *dev_id) +static irqreturn_t timer_interrupt(int irq, void *dev_id) { add_pda(irq0_irqs, 1); -- cgit v1.2.3 From f0bc2202e0373eb8e9b1ddbec930e2e681357db8 Mon Sep 17 00:00:00 2001 From: Jaswinder Singh Date: Wed, 17 Dec 2008 23:20:05 +0530 Subject: x86: process.c declare c1e_remove_cpu before they get used Impact: cleanup, avoid sparse warning Included asm/idle.h for c1e_remove_cpu() declaration. Fixes this sparse warning: CHECK arch/x86/kernel/process.c arch/x86/kernel/process.c:284:6: warning: symbol 'c1e_remove_cpu' was not declared. Should it be static? Signed-off-by: Jaswinder Singh Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c622772744d8..b06100f1d612 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -1,6 +1,7 @@ #include #include #include +#include #include #include #include -- cgit v1.2.3 From 8869a2e5d3a66d5b63b948052d60cd13ede8b735 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 18 Dec 2008 14:46:52 -0800 Subject: x86: asm-offset_64: use rt_sigframe_ia32 Impact: cleanup Use rt_sigframe_ia32 instead of rt_sigframe32. Signed-off-by: Hiroshi Shimamoto Signed-off-by: H. Peter Anvin --- arch/x86/kernel/asm-offsets_64.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c index 7fcf63d22f8b..1d41d3f1edbc 100644 --- a/arch/x86/kernel/asm-offsets_64.c +++ b/arch/x86/kernel/asm-offsets_64.c @@ -20,6 +20,8 @@ #include +#include + #define __NO_STUBS 1 #undef __SYSCALL #undef _ASM_X86_UNISTD_64_H @@ -87,7 +89,7 @@ int main(void) BLANK(); #undef ENTRY DEFINE(IA32_RT_SIGFRAME_sigcontext, - offsetof (struct rt_sigframe32, uc.uc_mcontext)); + offsetof (struct rt_sigframe_ia32, uc.uc_mcontext)); BLANK(); #endif DEFINE(pbe_address, offsetof(struct pbe, address)); -- cgit v1.2.3 From f34a10bd9f8cc95ebdc69a079db195636b2e22e0 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 19 Dec 2008 01:36:14 +0100 Subject: x86: fix warning in arch/x86/kernel/microcode_amd.c MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit this warning: arch/x86/kernel/microcode_amd.c: In function ‘apply_microcode_amd’: arch/x86/kernel/microcode_amd.c:163: warning: cast from pointer to integer of different size arch/x86/kernel/microcode_amd.c:163: warning: cast from pointer to integer of different size triggers because we want to pass the address to the microcode MSR, which is 64-bit even on 32-bit. Cast it explicitly to express this. Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_amd.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c index 24c256f4e50a..c25fdb382292 100644 --- a/arch/x86/kernel/microcode_amd.c +++ b/arch/x86/kernel/microcode_amd.c @@ -160,7 +160,7 @@ static void apply_microcode_amd(int cpu) return; spin_lock_irqsave(µcode_update_lock, flags); - wrmsrl(MSR_AMD64_PATCH_LOADER, &mc_amd->hdr.data_code); + wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code); /* get patch id after patching */ rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy); spin_unlock_irqrestore(µcode_update_lock, flags); @@ -372,3 +372,4 @@ struct microcode_ops * __init init_amd_microcode(void) { return µcode_amd_ops; } + -- cgit v1.2.3 From 345077cd98ff5532b2d1158013c3fec7b1ae85ec Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Thu, 18 Dec 2008 18:09:21 -0800 Subject: x86: fix intel x86_64 llc_shared_map/cpu_llc_id anomolies Impact: fix wrong cache sharing detection on platforms supporting > 8 bit apicid's In the presence of extended topology eumeration leaf 0xb provided by cpuid, 32bit extended initial_apicid in cpuinfo_x86 struct will be updated by detect_extended_topology(). At this instance, we should also reinit the apicid (which could also potentially be extended to 32bit). With out this there will potentially be duplicate apicid's populated in the per cpu's cpuinfo_x86 struct, resulting in wrong cache sharing topology etc detected by init_intel_cacheinfo(). Reported-by: Dimitri Sivanich Signed-off-by: Suresh Siddha Acked-by: Dimitri Sivanich Signed-off-by: Ingo Molnar Cc: --- arch/x86/kernel/cpu/addon_cpuid_features.c | 8 ++++++++ arch/x86/kernel/cpu/intel.c | 8 +++++++- 2 files changed, 15 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index ef8f831af823..2cf23634b6d9 100644 --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -120,9 +120,17 @@ void __cpuinit detect_extended_topology(struct cpuinfo_x86 *c) c->cpu_core_id = phys_pkg_id(c->initial_apicid, ht_mask_width) & core_select_mask; c->phys_proc_id = phys_pkg_id(c->initial_apicid, core_plus_mask_width); + /* + * Reinit the apicid, now that we have extended initial_apicid. + */ + c->apicid = phys_pkg_id(c->initial_apicid, 0); #else c->cpu_core_id = phys_pkg_id(ht_mask_width) & core_select_mask; c->phys_proc_id = phys_pkg_id(core_plus_mask_width); + /* + * Reinit the apicid, now that we have extended initial_apicid. + */ + c->apicid = phys_pkg_id(0); #endif c->x86_max_cores = (core_level_siblings / smp_num_siblings); diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index caec59437a22..b21c37c060a2 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -252,6 +252,13 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) intel_workarounds(c); + /* + * Detect the extended topology information if available. This + * will reinitialise the initial_apicid which will be used + * in init_intel_cacheinfo() + */ + detect_extended_topology(c); + l2 = init_intel_cacheinfo(c); if (c->cpuid_level > 9) { unsigned eax = cpuid_eax(10); @@ -323,7 +330,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) #endif - detect_extended_topology(c); if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) { /* * let's use the legacy cpuid vector 0x1 and 0x4 for topology -- cgit v1.2.3 From b909895739427874c089bc0e03dc119f99cab2dd Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 19 Dec 2008 13:48:34 -0800 Subject: sparseirq: fix numa_migrate_irq_desc dependency and comments Impact: reduce kconfig variable scope and clean up Bartlomiej pointed out that the config dependencies and comments are not right. update it depend to NUMA, and fix some comments Reported-by: Bartlomiej Zolnierkiewicz Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index bfe1245b1a3e..a74887b416cc 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2471,7 +2471,7 @@ static void irq_complete_move(struct irq_desc **descp) if (likely(!cfg->move_desc_pending)) return; - /* domain is not change, but affinity is changed */ + /* domain has not changed, but affinity did */ me = smp_processor_id(); if (cpu_isset(me, desc->affinity)) { *descp = desc = move_irq_desc(desc, me); -- cgit v1.2.3 From 34945ede31071ac7d72270cc6c1893323f392b3f Mon Sep 17 00:00:00 2001 From: Jaswinder Singh Date: Fri, 19 Dec 2008 22:33:52 +0530 Subject: x86: common.c boot_cpu_stack and boot_exception_stacks should be static Impact: cleanup, avoid sparse warnings, reduce kernel size a bit Fixes these sparse warnings: arch/x86/kernel/cpu/common.c:869:6: warning: symbol 'boot_cpu_stack' was not declared. Should it be static? arch/x86/kernel/cpu/common.c:910:6: warning: symbol 'boot_exception_stacks' was not declared. Should it be static? Signed-off-by: Jaswinder Singh Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index b9c9ea0217a9..aba49c782fd6 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -862,7 +862,7 @@ EXPORT_SYMBOL(_cpu_pda); struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table }; -char boot_cpu_stack[IRQSTACKSIZE] __page_aligned_bss; +static char boot_cpu_stack[IRQSTACKSIZE] __page_aligned_bss; void __cpuinit pda_init(int cpu) { @@ -903,8 +903,8 @@ void __cpuinit pda_init(int cpu) } } -char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + - DEBUG_STKSZ] __page_aligned_bss; +static char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + + DEBUG_STKSZ] __page_aligned_bss; extern asmlinkage void ignore_sysret(void); -- cgit v1.2.3 From bf53de907dfdaac178c92d774aae7370d7b97d20 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Fri, 19 Dec 2008 15:10:24 +0100 Subject: x86, bts: add fork and exit handling Impact: introduce new ptrace facility Add arch_ptrace_untrace() function that is called when the tracer detaches (either voluntarily or when the tracing task dies); ptrace_disable() is only called on a voluntary detach. Add ptrace_fork() and arch_ptrace_fork(). They are called when a traced task is forked. Clear DS and BTS related fields on fork. Release DS resources and reclaim memory in ptrace_untrace(). This releases resources already when the tracing task dies. We used to do that when the traced task dies. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ds.c | 11 ++++++++++ arch/x86/kernel/process_32.c | 20 ++++++++---------- arch/x86/kernel/process_64.c | 20 ++++++++---------- arch/x86/kernel/ptrace.c | 50 +++++++++++++++++++++++++++++++++++--------- 4 files changed, 69 insertions(+), 32 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 98d271e60e08..da91701a2348 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -1017,3 +1017,14 @@ void ds_switch_to(struct task_struct *prev, struct task_struct *next) update_debugctlmsr(next->thread.debugctlmsr); } + +void ds_copy_thread(struct task_struct *tsk, struct task_struct *father) +{ + clear_tsk_thread_flag(tsk, TIF_DS_AREA_MSR); + tsk->thread.ds_ctx = NULL; +} + +void ds_exit_thread(struct task_struct *tsk) +{ + WARN_ON(tsk->thread.ds_ctx); +} diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 605eff9a8ac0..3ba155d24884 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -60,6 +60,7 @@ #include #include #include +#include asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); @@ -251,17 +252,8 @@ void exit_thread(void) tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET; put_cpu(); } -#ifdef CONFIG_X86_DS - /* Free any BTS tracers that have not been properly released. */ - if (unlikely(current->bts)) { - ds_release_bts(current->bts); - current->bts = NULL; - - kfree(current->bts_buffer); - current->bts_buffer = NULL; - current->bts_size = 0; - } -#endif /* CONFIG_X86_DS */ + + ds_exit_thread(current); } void flush_thread(void) @@ -343,6 +335,12 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp, kfree(p->thread.io_bitmap_ptr); p->thread.io_bitmap_max = 0; } + + ds_copy_thread(p, current); + + clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); + p->thread.debugctlmsr = 0; + return err; } diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index 1cfd2a4bf853..416fb9282f4f 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -53,6 +53,7 @@ #include #include #include +#include asmlinkage extern void ret_from_fork(void); @@ -236,17 +237,8 @@ void exit_thread(void) t->io_bitmap_max = 0; put_cpu(); } -#ifdef CONFIG_X86_DS - /* Free any BTS tracers that have not been properly released. */ - if (unlikely(current->bts)) { - ds_release_bts(current->bts); - current->bts = NULL; - - kfree(current->bts_buffer); - current->bts_buffer = NULL; - current->bts_size = 0; - } -#endif /* CONFIG_X86_DS */ + + ds_exit_thread(current); } void flush_thread(void) @@ -376,6 +368,12 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp, if (err) goto out; } + + ds_copy_thread(p, me); + + clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); + p->thread.debugctlmsr = 0; + err = 0; out: if (err && p->thread.io_bitmap_ptr) { diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 45e9855da2d2..6ad2bb607650 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -769,8 +769,47 @@ static int ptrace_bts_size(struct task_struct *child) return (trace->ds.top - trace->ds.begin) / trace->ds.size; } + +static void ptrace_bts_fork(struct task_struct *tsk) +{ + tsk->bts = NULL; + tsk->bts_buffer = NULL; + tsk->bts_size = 0; + tsk->thread.bts_ovfl_signal = 0; +} + +static void ptrace_bts_untrace(struct task_struct *child) +{ + if (unlikely(child->bts)) { + ds_release_bts(child->bts); + child->bts = NULL; + + kfree(child->bts_buffer); + child->bts_buffer = NULL; + child->bts_size = 0; + } +} + +static void ptrace_bts_detach(struct task_struct *child) +{ + ptrace_bts_untrace(child); +} +#else +static inline void ptrace_bts_fork(struct task_struct *tsk) {} +static inline void ptrace_bts_detach(struct task_struct *child) {} +static inline void ptrace_bts_untrace(struct task_struct *child) {} #endif /* CONFIG_X86_PTRACE_BTS */ +void x86_ptrace_fork(struct task_struct *child, unsigned long clone_flags) +{ + ptrace_bts_fork(child); +} + +void x86_ptrace_untrace(struct task_struct *child) +{ + ptrace_bts_untrace(child); +} + /* * Called by kernel/ptrace.c when detaching.. * @@ -782,16 +821,7 @@ void ptrace_disable(struct task_struct *child) #ifdef TIF_SYSCALL_EMU clear_tsk_thread_flag(child, TIF_SYSCALL_EMU); #endif -#ifdef CONFIG_X86_PTRACE_BTS - if (child->bts) { - ds_release_bts(child->bts); - child->bts = NULL; - - kfree(child->bts_buffer); - child->bts_buffer = NULL; - child->bts_size = 0; - } -#endif /* CONFIG_X86_PTRACE_BTS */ + ptrace_bts_detach(child); } #if defined CONFIG_X86_32 || defined CONFIG_IA32_EMULATION -- cgit v1.2.3 From c5dee6177f4bd2095aab7d9be9f6ebdddd6deee9 Mon Sep 17 00:00:00 2001 From: Markus Metzger Date: Fri, 19 Dec 2008 15:17:02 +0100 Subject: x86, bts: memory accounting Impact: move the BTS buffer accounting to the mlock bucket Add alloc_locked_buffer() and free_locked_buffer() functions to mm/mlock.c to kalloc a buffer and account the locked memory to current. Account the memory for the BTS buffer to the tracer. Signed-off-by: Markus Metzger Signed-off-by: Ingo Molnar --- arch/x86/kernel/ptrace.c | 45 ++++++++++++++++++++++++++++++++++----------- 1 file changed, 34 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 6ad2bb607650..0a5df5f82fb9 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -650,6 +650,24 @@ static int ptrace_bts_drain(struct task_struct *child, return drained; } +static int ptrace_bts_allocate_buffer(struct task_struct *child, size_t size) +{ + child->bts_buffer = alloc_locked_buffer(size); + if (!child->bts_buffer) + return -ENOMEM; + + child->bts_size = size; + + return 0; +} + +static void ptrace_bts_free_buffer(struct task_struct *child) +{ + free_locked_buffer(child->bts_buffer, child->bts_size); + child->bts_buffer = NULL; + child->bts_size = 0; +} + static int ptrace_bts_config(struct task_struct *child, long cfg_size, const struct ptrace_bts_config __user *ucfg) @@ -679,14 +697,13 @@ static int ptrace_bts_config(struct task_struct *child, if ((cfg.flags & PTRACE_BTS_O_ALLOC) && (cfg.size != child->bts_size)) { - kfree(child->bts_buffer); + int error; - child->bts_size = cfg.size; - child->bts_buffer = kzalloc(cfg.size, GFP_KERNEL); - if (!child->bts_buffer) { - child->bts_size = 0; - return -ENOMEM; - } + ptrace_bts_free_buffer(child); + + error = ptrace_bts_allocate_buffer(child, cfg.size); + if (error < 0) + return error; } if (cfg.flags & PTRACE_BTS_O_TRACE) @@ -701,10 +718,8 @@ static int ptrace_bts_config(struct task_struct *child, if (IS_ERR(child->bts)) { int error = PTR_ERR(child->bts); - kfree(child->bts_buffer); + ptrace_bts_free_buffer(child); child->bts = NULL; - child->bts_buffer = NULL; - child->bts_size = 0; return error; } @@ -784,6 +799,9 @@ static void ptrace_bts_untrace(struct task_struct *child) ds_release_bts(child->bts); child->bts = NULL; + /* We cannot update total_vm and locked_vm since + child's mm is already gone. But we can reclaim the + memory. */ kfree(child->bts_buffer); child->bts_buffer = NULL; child->bts_size = 0; @@ -792,7 +810,12 @@ static void ptrace_bts_untrace(struct task_struct *child) static void ptrace_bts_detach(struct task_struct *child) { - ptrace_bts_untrace(child); + if (unlikely(child->bts)) { + ds_release_bts(child->bts); + child->bts = NULL; + + ptrace_bts_free_buffer(child); + } } #else static inline void ptrace_bts_fork(struct task_struct *tsk) {} -- cgit v1.2.3 From 280a9ca5d0663b185ddc4443052076c29652a328 Mon Sep 17 00:00:00 2001 From: Dmitry Adamushko Date: Sat, 20 Dec 2008 00:15:24 +0100 Subject: x86: fix resume (S2R) broken by Intel microcode module, on A110L Impact: fix deadlock This is in response to the following bug report: Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=12100 Subject : resume (S2R) broken by Intel microcode module, on A110L Submitter : Andreas Mohr Date : 2008-11-25 08:48 (19 days old) Handled-By : Dmitry Adamushko [ The deadlock scenario has been discovered by Andreas Mohr ] I think I might have a logical explanation why the system: (http://bugzilla.kernel.org/show_bug.cgi?id=12100) might hang upon resuming, OTOH it should have likely hanged each and every time. (1) possible deadlock in microcode_resume_cpu() if either 'if' section is taken; (2) now, I don't see it in spec. and can't experimentally verify it (newer ucodes don't seem to be available for my Core2duo)... but logically-wise, I'd think that when read upon resuming, the 'microcode revision' (MSR 0x8B) should be back to its original one (we need to reload ucode anyway so it doesn't seem logical if a cpu doesn't drop the version)... if so, the comparison with memcmp() for the full 'struct cpu_signature' is wrong... and that's how one of the aforementioned 'if' sections might have been triggered - leading to a deadlock. Obviously, in my tests I simulated loading/resuming with the ucode of the same version (just to see that the file is loaded/re-loaded upon resuming) so this issue has never popped up. I'd appreciate if someone with an appropriate system might give a try to the 2nd patch (titled "fix a comparison && deadlock..."). In any case, the deadlock situation is a must-have fix. Reported-by: Andreas Mohr Signed-off-by: Dmitry Adamushko Tested-by: Andreas Mohr Signed-off-by: Ingo Molnar Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/microcode_core.c | 19 ++++++++++++++----- arch/x86/kernel/microcode_intel.c | 6 ++++++ 2 files changed, 20 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c index 82fb2809ce32..c4b5b24e0217 100644 --- a/arch/x86/kernel/microcode_core.c +++ b/arch/x86/kernel/microcode_core.c @@ -272,13 +272,18 @@ static struct attribute_group mc_attr_group = { .name = "microcode", }; -static void microcode_fini_cpu(int cpu) +static void __microcode_fini_cpu(int cpu) { struct ucode_cpu_info *uci = ucode_cpu_info + cpu; - mutex_lock(µcode_mutex); microcode_ops->microcode_fini_cpu(cpu); uci->valid = 0; +} + +static void microcode_fini_cpu(int cpu) +{ + mutex_lock(µcode_mutex); + __microcode_fini_cpu(cpu); mutex_unlock(µcode_mutex); } @@ -306,12 +311,16 @@ static int microcode_resume_cpu(int cpu) * to this cpu (a bit of paranoia): */ if (microcode_ops->collect_cpu_info(cpu, &nsig)) { - microcode_fini_cpu(cpu); + __microcode_fini_cpu(cpu); + printk(KERN_ERR "failed to collect_cpu_info for resuming cpu #%d\n", + cpu); return -1; } - if (memcmp(&nsig, &uci->cpu_sig, sizeof(nsig))) { - microcode_fini_cpu(cpu); + if ((nsig.sig != uci->cpu_sig.sig) || (nsig.pf != uci->cpu_sig.pf)) { + __microcode_fini_cpu(cpu); + printk(KERN_ERR "cached ucode doesn't match the resuming cpu #%d\n", + cpu); /* Should we look for a new ucode here? */ return 1; } diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c index 622dc4a21784..a8e62792d171 100644 --- a/arch/x86/kernel/microcode_intel.c +++ b/arch/x86/kernel/microcode_intel.c @@ -155,6 +155,7 @@ static DEFINE_SPINLOCK(microcode_update_lock); static int collect_cpu_info(int cpu_num, struct cpu_signature *csig) { struct cpuinfo_x86 *c = &cpu_data(cpu_num); + unsigned long flags; unsigned int val[2]; memset(csig, 0, sizeof(*csig)); @@ -174,11 +175,16 @@ static int collect_cpu_info(int cpu_num, struct cpu_signature *csig) csig->pf = 1 << ((val[1] >> 18) & 7); } + /* serialize access to the physical write to MSR 0x79 */ + spin_lock_irqsave(µcode_update_lock, flags); + wrmsr(MSR_IA32_UCODE_REV, 0, 0); /* see notes above for revision 1.07. Apparent chip bug */ sync_core(); /* get the current revision from MSR 0x8B */ rdmsr(MSR_IA32_UCODE_REV, val[0], csig->rev); + spin_unlock_irqrestore(µcode_update_lock, flags); + pr_debug("microcode: collect_cpu_info : sig=0x%x, pf=0x%x, rev=0x%x\n", csig->sig, csig->pf, csig->rev); -- cgit v1.2.3 From adf77bac052bb5bf0722b2ce2af9fefc5b2d2a71 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 22 Dec 2008 17:56:05 -0800 Subject: x86: prioritize the FPU traps for the error code In the case of multiple FPU errors, prioritize the error codes, instead of returning __SI_FAULT, which ends up pushing a 0 as the error code to userspace, a POSIX violation. For i386, we will simply return if there are no errors at all; for x86-64 this is probably a "can't happen" (and the code should be unified), but for this patch, return __SI_FAULT|SI_KERNEL if this ever happens. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/traps.c | 34 +++++++++++++++------------------- 1 file changed, 15 insertions(+), 19 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index 04d242ab0161..c320c29255c2 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -664,7 +664,7 @@ void math_error(void __user *ip) { struct task_struct *task; siginfo_t info; - unsigned short cwd, swd; + unsigned short cwd, swd, err; /* * Save the info for the exception handler and clear the error. @@ -675,7 +675,6 @@ void math_error(void __user *ip) task->thread.error_code = 0; info.si_signo = SIGFPE; info.si_errno = 0; - info.si_code = __SI_FAULT; info.si_addr = ip; /* * (~cwd & swd) will mask out exceptions that are not set to unmasked @@ -689,34 +688,31 @@ void math_error(void __user *ip) */ cwd = get_fpu_cwd(task); swd = get_fpu_swd(task); - switch (swd & ~cwd & 0x3f) { - case 0x000: /* No unmasked exception */ -#ifdef CONFIG_X86_32 + + err = swd & ~cwd & 0x3f; + +#if CONFIG_X86_32 + if (!err) return; #endif - default: /* Multiple exceptions */ - break; - case 0x001: /* Invalid Op */ + + if (err & 0x001) { /* Invalid op */ /* * swd & 0x240 == 0x040: Stack Underflow * swd & 0x240 == 0x240: Stack Overflow * User must clear the SF bit (0x40) if set */ info.si_code = FPE_FLTINV; - break; - case 0x002: /* Denormalize */ - case 0x010: /* Underflow */ - info.si_code = FPE_FLTUND; - break; - case 0x004: /* Zero Divide */ + } else if (err & 0x004) { /* Divide by Zero */ info.si_code = FPE_FLTDIV; - break; - case 0x008: /* Overflow */ + } else if (err & 0x008) { /* Overflow */ info.si_code = FPE_FLTOVF; - break; - case 0x020: /* Precision */ + } else if (err & 0x012) { /* Denormal, Underflow */ + info.si_code = FPE_FLTUND; + } else if (err & 0x020) { /* Precision */ info.si_code = FPE_FLTRES; - break; + } else { + info.si_code = __SI_FAULT|SI_KERNEL; /* WTF? */ } force_sig_info(SIGFPE, &info, task); } -- cgit v1.2.3 From 0ca59dd948a51c95d5a366d35f897bc5ef9df55d Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Wed, 24 Dec 2008 23:30:02 +0100 Subject: tracing/ftrace: don't trace on early stage of a secondary cpu boot, v3 Impact: fix a crash/hard-reboot on certain configs while enabling cpu runtime On some archs, the boot of a secondary cpu can have an early fragile state. On x86-64, the pda is not initialized on the first stage of a cpu boot but it is needed to get the cpu number and the current task pointer. This data is needed during tracing. As they were dereferenced at this stage, we got a crash while tracing a cpu being enabled at runtime. Some other archs like ia64 can have such kind of issue too. Changes on v2: We dropped the previous solution of a per-arch called function to guess the current state of a cpu. That could slow down the tracing. This patch removes the -pg flag on arch/x86/kernel/cpu/common.c where the low level cpu boot functions exist, on start_secondary() and a helper function used at this stage. Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/Makefile | 5 +++++ arch/x86/kernel/smpboot.c | 2 +- 2 files changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index 82ec6075c057..4ae495a313f3 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -2,6 +2,11 @@ # Makefile for x86-compatible CPU details and quirks # +# Don't trace early stages of a secondary CPU boot +ifdef CONFIG_FUNCTION_TRACER +CFLAGS_REMOVE_common.o = -pg +endif + obj-y := intel_cacheinfo.o addon_cpuid_features.o obj-y += proc.o capflags.o powerflags.o common.o diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index f71f96fc9e62..f6174d229024 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -287,7 +287,7 @@ static int __cpuinitdata unsafe_smp; /* * Activate a secondary processor. */ -static void __cpuinit start_secondary(void *unused) +notrace static void __cpuinit start_secondary(void *unused) { /* * Don't put *anything* before cpu_init(), SMP booting is too -- cgit v1.2.3 From 1fcccb008be12ea823aaa392758e1e41fb82de9a Mon Sep 17 00:00:00 2001 From: Jaswinder Singh Date: Tue, 23 Dec 2008 21:50:11 +0530 Subject: x86: traps.c replace #if CONFIG_X86_32 with #ifdef CONFIG_X86_32 Impact: cleanup, avoid warning on X86_64 Fixes this warning on X86_64: CC arch/x86/kernel/traps.o arch/x86/kernel/traps.c:695:5: warning: "CONFIG_X86_32" is not defined Signed-off-by: Jaswinder Singh Signed-off-by: Ingo Molnar --- arch/x86/kernel/traps.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index c320c29255c2..f37cee75ab58 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -691,7 +691,7 @@ void math_error(void __user *ip) err = swd & ~cwd & 0x3f; -#if CONFIG_X86_32 +#ifdef CONFIG_X86_32 if (!err) return; #endif -- cgit v1.2.3