From b15ebe0b5d0b95aeb1d84cae3649df1e0e065e9b Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Thu, 17 Nov 2011 17:59:49 -0800 Subject: [SCSI] libsas: replace event locks with atomic bitops The locks only served to make sure the pending event bitmask was updated consistently. Signed-off-by: Dan Williams Signed-off-by: James Bottomley --- drivers/scsi/libsas/sas_event.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'drivers/scsi/libsas/sas_event.c') diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c index 9db30fb5caf2..9c084bc09bbd 100644 --- a/drivers/scsi/libsas/sas_event.c +++ b/drivers/scsi/libsas/sas_event.c @@ -30,7 +30,7 @@ static void notify_ha_event(struct sas_ha_struct *sas_ha, enum ha_event event) { BUG_ON(event >= HA_NUM_EVENTS); - sas_queue_event(event, &sas_ha->event_lock, &sas_ha->pending, + sas_queue_event(event, &sas_ha->pending, &sas_ha->ha_events[event].work, sas_ha); } @@ -40,7 +40,7 @@ static void notify_port_event(struct asd_sas_phy *phy, enum port_event event) BUG_ON(event >= PORT_NUM_EVENTS); - sas_queue_event(event, &ha->event_lock, &phy->port_events_pending, + sas_queue_event(event, &phy->port_events_pending, &phy->port_events[event].work, ha); } @@ -50,7 +50,7 @@ static void notify_phy_event(struct asd_sas_phy *phy, enum phy_event event) BUG_ON(event >= PHY_NUM_EVENTS); - sas_queue_event(event, &ha->event_lock, &phy->phy_events_pending, + sas_queue_event(event, &phy->phy_events_pending, &phy->phy_events[event].work, ha); } @@ -62,8 +62,6 @@ int sas_init_events(struct sas_ha_struct *sas_ha) int i; - spin_lock_init(&sas_ha->event_lock); - for (i = 0; i < HA_NUM_EVENTS; i++) { INIT_WORK(&sas_ha->ha_events[i].work, sas_ha_event_fns[i]); sas_ha->ha_events[i].ha = sas_ha; -- cgit v1.2.3 From b1124cd3ec97406c767b90bf7e93ecd2d2915592 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 19 Dec 2011 16:42:34 -0800 Subject: [SCSI] libsas: introduce sas_drain_work() When an lldd invokes ->notify_port_event() it can trigger a chain of libsas events to: 1/ form the port and find the direct attached device 2/ if the attached device is an expander perform domain discovery A call to flush_workqueue() will only flush the initial port formation work. Currently libsas users need to call scsi_flush_work() up to the max depth of chain (which will grow from 2 to 3 when ata discovery is moved to its own discovery event). Instead of open coding multiple calls switch to use drain_workqueue() to flush sas work. drain_workqueue() does not handle new work submitted during the drain so libsas needs a bit of infrastructure to hold off unchained work submissions while a drain is in flight. A lldd ->notify() event is considered 'unchained' while a sas_discover_event() is 'chained'. As Tejun notes: "For now, I think it would be best to add private wrapper in libsas to support deferring unchained work items while draining." Signed-off-by: Dan Williams Signed-off-by: James Bottomley --- drivers/scsi/libsas/sas_event.c | 55 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) (limited to 'drivers/scsi/libsas/sas_event.c') diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c index 9c084bc09bbd..e5035aa4c2a6 100644 --- a/drivers/scsi/libsas/sas_event.c +++ b/drivers/scsi/libsas/sas_event.c @@ -22,10 +22,65 @@ * */ +#include #include #include "sas_internal.h" #include "sas_dump.h" +static void sas_queue_work(struct sas_ha_struct *ha, struct work_struct *work) +{ + if (!test_bit(SAS_HA_REGISTERED, &ha->state)) + return; + + if (test_bit(SAS_HA_DRAINING, &ha->state)) + list_add(&work->entry, &ha->defer_q); + else + scsi_queue_work(ha->core.shost, work); +} + +static void sas_queue_event(int event, unsigned long *pending, + struct work_struct *work, + struct sas_ha_struct *ha) +{ + if (!test_and_set_bit(event, pending)) { + unsigned long flags; + + spin_lock_irqsave(&ha->state_lock, flags); + sas_queue_work(ha, work); + spin_unlock_irqrestore(&ha->state_lock, flags); + } +} + +int sas_drain_work(struct sas_ha_struct *ha) +{ + struct workqueue_struct *wq = ha->core.shost->work_q; + struct work_struct *w, *_w; + int err; + + err = mutex_lock_interruptible(&ha->drain_mutex); + if (err) + return err; + + set_bit(SAS_HA_DRAINING, &ha->state); + /* flush submitters */ + spin_lock_irq(&ha->state_lock); + spin_unlock_irq(&ha->state_lock); + + drain_workqueue(wq); + + spin_lock_irq(&ha->state_lock); + clear_bit(SAS_HA_DRAINING, &ha->state); + list_for_each_entry_safe(w, _w, &ha->defer_q, entry) { + list_del_init(&w->entry); + sas_queue_work(ha, w); + } + spin_unlock_irq(&ha->state_lock); + mutex_unlock(&ha->drain_mutex); + + return 0; +} +EXPORT_SYMBOL_GPL(sas_drain_work); + static void notify_ha_event(struct sas_ha_struct *sas_ha, enum ha_event event) { BUG_ON(event >= HA_NUM_EVENTS); -- cgit v1.2.3 From 87c8331fcf72e501c3a3c0cdc5c9391ec72f7cf2 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Thu, 17 Nov 2011 17:59:51 -0800 Subject: [SCSI] libsas: prevent domain rediscovery competing with ata error handling libata error handling provides for a timeout for link recovery. libsas must not rescan for previously known devices in this interval otherwise it may remove a device that is simply waiting for its link to recover. Let libata-eh make the determination of when the link is stable and prevent libsas (host workqueue) from taking action while this determination is pending. Using a mutex (ha->disco_mutex) to flush and disable revalidation while eh is running requires any discovery action that may block on eh be moved to its own context outside the lock. Probing ATA devices explicitly waits on ata-eh and the cache-flush-io issued during device removal may also pend awaiting eh completion. Essentially any rphy add/remove activity needs to run outside the lock. This adds two new cleanup states for sas_unregister_domain_devices() 'allocated-but-not-probed', and 'flagged-for-destruction'. In the 'allocated-but-not-probed' state dev->rphy points to a rphy that is known to have not been through a sas_rphy_add() event. At domain teardown check if this device is still pending probe and cleanup accordingly. Similarly if a device has already been queued for removal then sas_unregister_domain_devices has nothing to do. Signed-off-by: Dan Williams Signed-off-by: James Bottomley --- drivers/scsi/libsas/sas_event.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) (limited to 'drivers/scsi/libsas/sas_event.c') diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c index e5035aa4c2a6..933d757499b5 100644 --- a/drivers/scsi/libsas/sas_event.c +++ b/drivers/scsi/libsas/sas_event.c @@ -81,6 +81,32 @@ int sas_drain_work(struct sas_ha_struct *ha) } EXPORT_SYMBOL_GPL(sas_drain_work); +void sas_disable_revalidation(struct sas_ha_struct *ha) +{ + mutex_lock(&ha->disco_mutex); + set_bit(SAS_HA_ATA_EH_ACTIVE, &ha->state); + mutex_unlock(&ha->disco_mutex); +} + +void sas_enable_revalidation(struct sas_ha_struct *ha) +{ + int i; + + mutex_lock(&ha->disco_mutex); + clear_bit(SAS_HA_ATA_EH_ACTIVE, &ha->state); + for (i = 0; i < ha->num_phys; i++) { + struct asd_sas_port *port = ha->sas_port[i]; + const int ev = DISCE_REVALIDATE_DOMAIN; + struct sas_discovery *d = &port->disc; + + if (!test_and_clear_bit(ev, &d->pending)) + continue; + + sas_queue_event(ev, &d->pending, &d->disc_work[ev].work, ha); + } + mutex_unlock(&ha->disco_mutex); +} + static void notify_ha_event(struct sas_ha_struct *sas_ha, enum ha_event event) { BUG_ON(event >= HA_NUM_EVENTS); -- cgit v1.2.3 From 0b3e09da1350397f3f8b6fd839ab455b0b587451 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Tue, 20 Dec 2011 01:03:48 -0800 Subject: [SCSI] libsas: perform sas-transport resets in shost->workq context Extend the sas transport class to allow transport users to attach extra data to a sas_phy (->hostdata). Use this area in libsas to move resets to workq context in preparation for scheduling ata device resets through libata-eh. Signed-off-by: Dan Williams Signed-off-by: James Bottomley --- drivers/scsi/libsas/sas_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/scsi/libsas/sas_event.c') diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c index 933d757499b5..dbfaceeea0f7 100644 --- a/drivers/scsi/libsas/sas_event.c +++ b/drivers/scsi/libsas/sas_event.c @@ -27,7 +27,7 @@ #include "sas_internal.h" #include "sas_dump.h" -static void sas_queue_work(struct sas_ha_struct *ha, struct work_struct *work) +void sas_queue_work(struct sas_ha_struct *ha, struct work_struct *work) { if (!test_bit(SAS_HA_REGISTERED, &ha->state)) return; -- cgit v1.2.3 From 5d7f6d1071eadd020edb2cf366d358e0f6d0a0f9 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Thu, 12 Jan 2012 11:47:24 -0800 Subject: [SCSI] libsas: fix sas_unregister_ports vs sas_drain_work We need to hold drain_mutex across the unregistration as port down events queue device removal as chained events, so we need to make sure no other drainers are active. [ 1118.673968] WARNING: at kernel/workqueue.c:996 __queue_work+0x11a/0x326() [ 1118.681982] Hardware name: S2600CP [ 1118.686193] Modules linked in: isci(-) libsas scsi_transport_sas nls_utf8 ipv6 uinput sg iTCO_wdt iTCO_vendor_support i2c_i801 i2c_core ioatdma dca sd_mod sr_mod cdrom ahci libahci libata [last unloaded: scsi_transport_sas] [ 1118.709893] Pid: 6831, comm: rmmod Not tainted 3.2.0-isci+ #1 [ 1118.716727] Call Trace: [ 1118.719867] [] warn_slowpath_common+0x85/0x9d [ 1118.727000] [] warn_slowpath_null+0x1a/0x1c [ 1118.733942] [] __queue_work+0x11a/0x326 [ 1118.740481] [] queue_work_on+0x1b/0x22 [ 1118.746925] [] queue_work+0x37/0x3e [ 1118.753105] [] ? sas_discover_event+0x55/0x82 [libsas] [ 1118.761094] [] scsi_queue_work+0x42/0x44 [ 1118.767717] [] sas_discover_event+0x69/0x82 [libsas] [ 1118.775509] [] sas_unregister_dev+0xc3/0xcc [libsas] [ 1118.783319] [] sas_unregister_domain_devices+0x4a/0xc8 [libsas] [ 1118.792731] [] sas_deform_port+0x60/0x1a6 [libsas] [ 1118.800339] [] sas_unregister_ports+0x33/0x44 [libsas] [ 1118.808342] [] sas_unregister_ha+0x41/0x6b [libsas] [ 1118.816055] [] isci_unregister+0x22/0x4d [isci] [ 1118.823384] [] isci_pci_remove+0x2e/0x60 [isci] Reported-by: Jacek Danecki Signed-off-by: Dan Williams Signed-off-by: James Bottomley --- drivers/scsi/libsas/sas_event.c | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) (limited to 'drivers/scsi/libsas/sas_event.c') diff --git a/drivers/scsi/libsas/sas_event.c b/drivers/scsi/libsas/sas_event.c index dbfaceeea0f7..16639bbae629 100644 --- a/drivers/scsi/libsas/sas_event.c +++ b/drivers/scsi/libsas/sas_event.c @@ -51,15 +51,11 @@ static void sas_queue_event(int event, unsigned long *pending, } } -int sas_drain_work(struct sas_ha_struct *ha) + +void __sas_drain_work(struct sas_ha_struct *ha) { struct workqueue_struct *wq = ha->core.shost->work_q; struct work_struct *w, *_w; - int err; - - err = mutex_lock_interruptible(&ha->drain_mutex); - if (err) - return err; set_bit(SAS_HA_DRAINING, &ha->state); /* flush submitters */ @@ -75,6 +71,17 @@ int sas_drain_work(struct sas_ha_struct *ha) sas_queue_work(ha, w); } spin_unlock_irq(&ha->state_lock); +} + +int sas_drain_work(struct sas_ha_struct *ha) +{ + int err; + + err = mutex_lock_interruptible(&ha->drain_mutex); + if (err) + return err; + if (test_bit(SAS_HA_REGISTERED, &ha->state)) + __sas_drain_work(ha); mutex_unlock(&ha->drain_mutex); return 0; -- cgit v1.2.3