From 2c016dc2cbd160acf099bee8e4d5c056a8b691b1 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Wed, 26 Dec 2012 15:32:17 +0800 Subject: btrfs: fix comment typos Convert 'hepler' to 'helper'. Signed-off-by: Liu Bo Signed-off-by: Jiri Kosina --- fs/btrfs/extent-tree.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 06b2635073f3..604befdcc359 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6482,7 +6482,7 @@ reada: } /* - * hepler to process tree block while walking down the tree. + * helper to process tree block while walking down the tree. * * when wc->stage == UPDATE_BACKREF, this function updates * back refs for pointers in the block. @@ -6557,7 +6557,7 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans, } /* - * hepler to process tree block pointer. + * helper to process tree block pointer. * * when wc->stage == DROP_REFERENCE, this function checks * reference count of the block pointed to. if the block @@ -6695,7 +6695,7 @@ skip: } /* - * hepler to process tree block while walking up the tree. + * helper to process tree block while walking up the tree. * * when wc->stage == DROP_REFERENCE, this function drops * reference count on the block. -- cgit v1.2.3 From 10ee27a06cc8eb57f83342a8eabcb75deb872d52 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Jan 2013 13:47:57 +0800 Subject: vfs: re-implement writeback_inodes_sb(_nr)_if_idle() and rename them writeback_inodes_sb(_nr)_if_idle() is re-implemented by replacing down_read() with down_read_trylock() because - If ->s_umount is write locked, then the sb is not idle. That is writeback_inodes_sb(_nr)_if_idle() needn't wait for the lock. - writeback_inodes_sb(_nr)_if_idle() grabs s_umount lock when it want to start writeback, it may bring us deadlock problem when doing umount. In order to fix the problem, ext4 and btrfs implemented their own writeback functions instead of writeback_inodes_sb(_nr)_if_idle(), but it introduced the redundant code, it is better to implement a new writeback_inodes_sb(_nr)_if_idle(). The name of these two functions is cumbersome, so rename them to try_to_writeback_inodes_sb(_nr). This idea came from Christoph Hellwig. Some code is from the patch of Kamal Mostafa. Reviewed-by: Jan Kara Signed-off-by: Miao Xie Signed-off-by: Fengguang Wu --- fs/btrfs/extent-tree.c | 20 +++----------------- 1 file changed, 3 insertions(+), 17 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 521e9d4424f6..f31abb14e06f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3689,20 +3689,6 @@ static int can_overcommit(struct btrfs_root *root, return 0; } -static int writeback_inodes_sb_nr_if_idle_safe(struct super_block *sb, - unsigned long nr_pages, - enum wb_reason reason) -{ - if (!writeback_in_progress(sb->s_bdi) && - down_read_trylock(&sb->s_umount)) { - writeback_inodes_sb_nr(sb, nr_pages, reason); - up_read(&sb->s_umount); - return 1; - } - - return 0; -} - /* * shrink metadata reservation for delalloc */ @@ -3735,9 +3721,9 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, while (delalloc_bytes && loops < 3) { max_reclaim = min(delalloc_bytes, to_reclaim); nr_pages = max_reclaim >> PAGE_CACHE_SHIFT; - writeback_inodes_sb_nr_if_idle_safe(root->fs_info->sb, - nr_pages, - WB_REASON_FS_FREE_SPACE); + try_to_writeback_inodes_sb_nr(root->fs_info->sb, + nr_pages, + WB_REASON_FS_FREE_SPACE); /* * We need to wait for the async pages to actually start before -- cgit v1.2.3 From 53b381b3abeb86f12787a6c40fee9b2f71edc23b Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Tue, 29 Jan 2013 18:40:14 -0500 Subject: Btrfs: RAID5 and RAID6 This builds on David Woodhouse's original Btrfs raid5/6 implementation. The code has changed quite a bit, blame Chris Mason for any bugs. Read/modify/write is done after the higher levels of the filesystem have prepared a given bio. This means the higher layers are not responsible for building full stripes, and they don't need to query for the topology of the extents that may get allocated during delayed allocation runs. It also means different files can easily share the same stripe. But, it does expose us to incorrect parity if we crash or lose power while doing a read/modify/write cycle. This will be addressed in a later commit. Scrub is unable to repair crc errors on raid5/6 chunks. Discard does not work on raid5/6 (yet) The stripe size is fixed at 64KiB per disk. This will be tunable in a later commit. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 88 +++++++++++++++++++++++++++++++++----------------- 1 file changed, 59 insertions(+), 29 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d133edfcd449..3345f68fc64b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -31,6 +31,7 @@ #include "print-tree.h" #include "transaction.h" #include "volumes.h" +#include "raid56.h" #include "locking.h" #include "free-space-cache.h" #include "math.h" @@ -1852,6 +1853,8 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, *actual_bytes = discarded_bytes; + if (ret == -EOPNOTSUPP) + ret = 0; return ret; } @@ -3276,6 +3279,7 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) u64 num_devices = root->fs_info->fs_devices->rw_devices + root->fs_info->fs_devices->missing_devices; u64 target; + u64 tmp; /* * see if restripe for this chunk_type is in progress, if so @@ -3292,30 +3296,32 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) } spin_unlock(&root->fs_info->balance_lock); + /* First, mask out the RAID levels which aren't possible */ if (num_devices == 1) - flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0); + flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5); + if (num_devices < 3) + flags &= ~BTRFS_BLOCK_GROUP_RAID6; if (num_devices < 4) flags &= ~BTRFS_BLOCK_GROUP_RAID10; - if ((flags & BTRFS_BLOCK_GROUP_DUP) && - (flags & (BTRFS_BLOCK_GROUP_RAID1 | - BTRFS_BLOCK_GROUP_RAID10))) { - flags &= ~BTRFS_BLOCK_GROUP_DUP; - } - - if ((flags & BTRFS_BLOCK_GROUP_RAID1) && - (flags & BTRFS_BLOCK_GROUP_RAID10)) { - flags &= ~BTRFS_BLOCK_GROUP_RAID1; - } + tmp = flags & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID10); + flags &= ~tmp; - if ((flags & BTRFS_BLOCK_GROUP_RAID0) && - ((flags & BTRFS_BLOCK_GROUP_RAID1) | - (flags & BTRFS_BLOCK_GROUP_RAID10) | - (flags & BTRFS_BLOCK_GROUP_DUP))) { - flags &= ~BTRFS_BLOCK_GROUP_RAID0; - } + if (tmp & BTRFS_BLOCK_GROUP_RAID6) + tmp = BTRFS_BLOCK_GROUP_RAID6; + else if (tmp & BTRFS_BLOCK_GROUP_RAID5) + tmp = BTRFS_BLOCK_GROUP_RAID5; + else if (tmp & BTRFS_BLOCK_GROUP_RAID10) + tmp = BTRFS_BLOCK_GROUP_RAID10; + else if (tmp & BTRFS_BLOCK_GROUP_RAID1) + tmp = BTRFS_BLOCK_GROUP_RAID1; + else if (tmp & BTRFS_BLOCK_GROUP_RAID0) + tmp = BTRFS_BLOCK_GROUP_RAID0; - return extended_to_chunk(flags); + return extended_to_chunk(flags | tmp); } static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) @@ -3333,6 +3339,7 @@ static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) { u64 flags; + u64 ret; if (data) flags = BTRFS_BLOCK_GROUP_DATA; @@ -3341,7 +3348,8 @@ u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) else flags = BTRFS_BLOCK_GROUP_METADATA; - return get_alloc_profile(root, flags); + ret = get_alloc_profile(root, flags); + return ret; } /* @@ -3516,8 +3524,10 @@ static u64 get_system_chunk_thresh(struct btrfs_root *root, u64 type) { u64 num_dev; - if (type & BTRFS_BLOCK_GROUP_RAID10 || - type & BTRFS_BLOCK_GROUP_RAID0) + if (type & (BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) num_dev = root->fs_info->fs_devices->rw_devices; else if (type & BTRFS_BLOCK_GROUP_RAID1) num_dev = 2; @@ -3667,7 +3677,9 @@ static int can_overcommit(struct btrfs_root *root, /* * If we have dup, raid1 or raid10 then only half of the free - * space is actually useable. + * space is actually useable. For raid56, the space info used + * doesn't include the parity drive, so we don't have to + * change the math */ if (profile & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | @@ -5455,10 +5467,14 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, return ret; } -static u64 stripe_align(struct btrfs_root *root, u64 val) +static u64 stripe_align(struct btrfs_root *root, + struct btrfs_block_group_cache *cache, + u64 val, u64 num_bytes) { - u64 mask = ((u64)root->stripesize - 1); - u64 ret = (val + mask) & ~mask; + u64 mask; + u64 ret; + mask = ((u64)root->stripesize - 1); + ret = (val + mask) & ~mask; return ret; } @@ -5519,9 +5535,12 @@ int __get_raid_index(u64 flags) index = 2; else if (flags & BTRFS_BLOCK_GROUP_RAID0) index = 3; + else if (flags & BTRFS_BLOCK_GROUP_RAID5) + index = 5; + else if (flags & BTRFS_BLOCK_GROUP_RAID6) + index = 6; else - index = 4; - + index = 4; /* BTRFS_BLOCK_GROUP_SINGLE */ return index; } @@ -5665,6 +5684,8 @@ search: if (!block_group_bits(block_group, data)) { u64 extra = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID10; /* @@ -5835,7 +5856,8 @@ unclustered_alloc: goto loop; } checks: - search_start = stripe_align(root, offset); + search_start = stripe_align(root, used_block_group, + offset, num_bytes); /* move on to the next group */ if (search_start + num_bytes > @@ -7203,6 +7225,7 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags) root->fs_info->fs_devices->missing_devices; stripped = BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10; if (num_devices == 1) { @@ -7754,7 +7777,9 @@ int btrfs_read_block_groups(struct btrfs_root *root) btrfs_release_path(path); cache->flags = btrfs_block_group_flags(&cache->item); cache->sectorsize = root->sectorsize; - + cache->full_stripe_len = btrfs_full_stripe_len(root, + &root->fs_info->mapping_tree, + found_key.objectid); btrfs_init_free_space_ctl(cache); /* @@ -7808,6 +7833,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) if (!(get_alloc_profile(root, space_info->flags) & (BTRFS_BLOCK_GROUP_RAID10 | BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_DUP))) continue; /* @@ -7883,6 +7910,9 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY; cache->sectorsize = root->sectorsize; cache->fs_info = root->fs_info; + cache->full_stripe_len = btrfs_full_stripe_len(root, + &root->fs_info->mapping_tree, + chunk_offset); atomic_set(&cache->count, 1); spin_lock_init(&cache->lock); -- cgit v1.2.3 From 8de972b4faa4be9b2a3c53103e18d86092a5da45 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 4 Jan 2013 15:39:43 -0500 Subject: Btrfs: fix cluster alignment for mount -o ssd With the new raid56 code, we want to make sure we're properly aligning our allocation clusters with -o ssd Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 3345f68fc64b..f13402104c96 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5715,6 +5715,7 @@ have_block_group: * lets look there */ if (last_ptr) { + unsigned long aligned_cluster; /* * the refill lock keeps out other * people trying to start a new cluster @@ -5781,11 +5782,15 @@ refill_cluster: goto unclustered_alloc; } + aligned_cluster = max_t(unsigned long, + empty_cluster + empty_size, + block_group->full_stripe_len); + /* allocate a cluster in this block group */ ret = btrfs_find_space_cluster(trans, root, block_group, last_ptr, search_start, num_bytes, - empty_cluster + empty_size); + aligned_cluster); if (ret == 0) { /* * now pull our allocation out of this -- cgit v1.2.3 From bb721703aa551e98dc5c7fb259cf90343408baf2 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 29 Jan 2013 18:44:12 -0500 Subject: Btrfs: reduce CPU contention while waiting for delayed extent operations We batch up operations to the extent allocation tree, which allows us to deal with the recursive nature of using the extent allocation tree to allocate extents to the extent allocation tree. It also provides a mechanism to sort and collect extent operations, which makes it much more efficient to record extents that are close together. The delayed extent operations must all be finished before the running transaction commits, so we have code to make sure and run a few of the batched operations when closing our transaction handles. This creates a great deal of contention for the locks in the delayed extent operation tree, and also contention for the lock on the extent allocation tree itself. All the extra contention just slows down the operations and doesn't get things done any faster. This commit changes things to use a wait queue instead. As procs want to run the delayed operations, one of them races in and gets permission to hit the tree, and the others step back and wait for progress to be made. Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 60 ++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 56 insertions(+), 4 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index f13402104c96..87b0e856b6d0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2438,6 +2438,16 @@ int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans, return ret; } +static int refs_newer(struct btrfs_delayed_ref_root *delayed_refs, int seq, + int count) +{ + int val = atomic_read(&delayed_refs->ref_seq); + + if (val < seq || val >= seq + count) + return 1; + return 0; +} + /* * this starts processing the delayed reference count updates and * extent insertions we have queued up so far. count can be @@ -2472,6 +2482,44 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, delayed_refs = &trans->transaction->delayed_refs; INIT_LIST_HEAD(&cluster); + if (count == 0) { + count = delayed_refs->num_entries * 2; + run_most = 1; + } + + if (!run_all && !run_most) { + int old; + int seq = atomic_read(&delayed_refs->ref_seq); + +progress: + old = atomic_cmpxchg(&delayed_refs->procs_running_refs, 0, 1); + if (old) { + DEFINE_WAIT(__wait); + if (delayed_refs->num_entries < 16348) + return 0; + + prepare_to_wait(&delayed_refs->wait, &__wait, + TASK_UNINTERRUPTIBLE); + + old = atomic_cmpxchg(&delayed_refs->procs_running_refs, 0, 1); + if (old) { + schedule(); + finish_wait(&delayed_refs->wait, &__wait); + + if (!refs_newer(delayed_refs, seq, 256)) + goto progress; + else + return 0; + } else { + finish_wait(&delayed_refs->wait, &__wait); + goto again; + } + } + + } else { + atomic_inc(&delayed_refs->procs_running_refs); + } + again: loops = 0; spin_lock(&delayed_refs->lock); @@ -2480,10 +2528,6 @@ again: delayed_refs->run_delayed_start = find_middle(&delayed_refs->root); #endif - if (count == 0) { - count = delayed_refs->num_entries * 2; - run_most = 1; - } while (1) { if (!(run_all || run_most) && delayed_refs->num_heads_ready < 64) @@ -2505,9 +2549,12 @@ again: if (ret < 0) { spin_unlock(&delayed_refs->lock); btrfs_abort_transaction(trans, root, ret); + atomic_dec(&delayed_refs->procs_running_refs); return ret; } + atomic_add(ret, &delayed_refs->ref_seq); + count -= min_t(unsigned long, ret, count); if (count == 0) @@ -2576,6 +2623,11 @@ again: goto again; } out: + atomic_dec(&delayed_refs->procs_running_refs); + smp_mb(); + if (waitqueue_active(&delayed_refs->wait)) + wake_up(&delayed_refs->wait); + spin_unlock(&delayed_refs->lock); assert_qgroups_uptodate(trans); return 0; -- cgit v1.2.3 From eb6b88d92c6df083dd09a8c471011e3788dfd7c6 Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Sun, 27 Jan 2013 23:26:00 -0700 Subject: Btrfs: fix EDQUOT handling in btrfs_delalloc_reserve_metadata When btrfs_qgroup_reserve returned a failure, we were missing a counter operation for BTRFS_I(inode)->outstanding_extents++, leading to warning messages about outstanding extents and space_info->bytes_may_use != 0. Additionally, the error handling code didn't take into account that we dropped the inode lock which might require more cleanup. Luckily, all the cleanup code we need is already there and can be shared with reserve_metadata_bytes, which is exactly what this patch does. Reported-by: Lev Vainblat Signed-off-by: Jan Schmidt Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 85b8454d9608..61da9d0bb805 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4534,7 +4534,7 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) unsigned nr_extents = 0; int extra_reserve = 0; enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_ALL; - int ret; + int ret = 0; bool delalloc_lock = true; /* If we are a free space inode we need to not flush since we will be in @@ -4579,20 +4579,18 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) csum_bytes = BTRFS_I(inode)->csum_bytes; spin_unlock(&BTRFS_I(inode)->lock); - if (root->fs_info->quota_enabled) { + if (root->fs_info->quota_enabled) ret = btrfs_qgroup_reserve(root, num_bytes + nr_extents * root->leafsize); - if (ret) { - spin_lock(&BTRFS_I(inode)->lock); - calc_csum_metadata_size(inode, num_bytes, 0); - spin_unlock(&BTRFS_I(inode)->lock); - if (delalloc_lock) - mutex_unlock(&BTRFS_I(inode)->delalloc_mutex); - return ret; - } - } - ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush); + /* + * ret != 0 here means the qgroup reservation failed, we go straight to + * the shared error handling then. + */ + if (ret == 0) + ret = reserve_metadata_bytes(root, block_rsv, + to_reserve, flush); + if (ret) { u64 to_free = 0; unsigned dropped; -- cgit v1.2.3 From 78a6184a3ff9041280ee56273c01e5679a831b39 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 21 Nov 2012 02:21:28 +0000 Subject: Btrfs: use slabs for delayed reference allocation The delayed reference allocation is in the fast path of the IO, so use slabs to improve the speed of the allocation. And besides that, it can do check for leaked objects when the module is removed. Signed-off-by: Miao Xie --- fs/btrfs/extent-tree.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 61da9d0bb805..9bd87f0d73d0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2285,7 +2285,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, ref = &locked_ref->node; if (extent_op && must_insert_reserved) { - kfree(extent_op); + btrfs_free_delayed_extent_op(extent_op); extent_op = NULL; } @@ -2294,7 +2294,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, ret = run_delayed_extent_op(trans, root, ref, extent_op); - kfree(extent_op); + btrfs_free_delayed_extent_op(extent_op); if (ret) { list_del_init(&locked_ref->cluster); @@ -2338,7 +2338,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, must_insert_reserved); btrfs_put_delayed_ref(ref); - kfree(extent_op); + btrfs_free_delayed_extent_op(extent_op); count++; if (ret) { @@ -2586,7 +2586,7 @@ int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans, struct btrfs_delayed_extent_op *extent_op; int ret; - extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS); + extent_op = btrfs_alloc_delayed_extent_op(); if (!extent_op) return -ENOMEM; @@ -2598,7 +2598,7 @@ int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans, ret = btrfs_add_delayed_extent_op(root->fs_info, trans, bytenr, num_bytes, extent_op); if (ret) - kfree(extent_op); + btrfs_free_delayed_extent_op(extent_op); return ret; } @@ -5330,7 +5330,7 @@ static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans, if (head->extent_op) { if (!head->must_insert_reserved) goto out; - kfree(head->extent_op); + btrfs_free_delayed_extent_op(head->extent_op); head->extent_op = NULL; } @@ -6400,7 +6400,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans, if (root_objectid != BTRFS_TREE_LOG_OBJECTID) { struct btrfs_delayed_extent_op *extent_op; - extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS); + extent_op = btrfs_alloc_delayed_extent_op(); BUG_ON(!extent_op); /* -ENOMEM */ if (key) memcpy(&extent_op->key, key, sizeof(extent_op->key)); -- cgit v1.2.3 From 093486c453a55230ccdad4b48863b872fe68c46e Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 19 Dec 2012 08:10:10 +0000 Subject: Btrfs: make delayed ref lock logic more readable Locking and unlocking delayed ref mutex are in the different functions, and the name of lock functions is not uniform, so the readability is not so good, this patch optimizes the lock logic and makes it more readable. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 42 ++++++++++++++++++++++++------------------ 1 file changed, 24 insertions(+), 18 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 9bd87f0d73d0..b4cb8186035f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2143,7 +2143,6 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans, node->num_bytes); } } - mutex_unlock(&head->mutex); return ret; } @@ -2258,7 +2257,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, * process of being added. Don't run this ref yet. */ list_del_init(&locked_ref->cluster); - mutex_unlock(&locked_ref->mutex); + btrfs_delayed_ref_unlock(locked_ref); locked_ref = NULL; delayed_refs->num_heads_ready++; spin_unlock(&delayed_refs->lock); @@ -2297,25 +2296,22 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, btrfs_free_delayed_extent_op(extent_op); if (ret) { - list_del_init(&locked_ref->cluster); - mutex_unlock(&locked_ref->mutex); - - printk(KERN_DEBUG "btrfs: run_delayed_extent_op returned %d\n", ret); + printk(KERN_DEBUG + "btrfs: run_delayed_extent_op " + "returned %d\n", ret); spin_lock(&delayed_refs->lock); + btrfs_delayed_ref_unlock(locked_ref); return ret; } goto next; } - - list_del_init(&locked_ref->cluster); - locked_ref = NULL; } ref->in_tree = 0; rb_erase(&ref->rb_node, &delayed_refs->root); delayed_refs->num_entries--; - if (locked_ref) { + if (!btrfs_delayed_ref_is_head(ref)) { /* * when we play the delayed ref, also correct the * ref_mod on head @@ -2337,20 +2333,29 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, ret = run_one_delayed_ref(trans, root, ref, extent_op, must_insert_reserved); - btrfs_put_delayed_ref(ref); btrfs_free_delayed_extent_op(extent_op); - count++; - if (ret) { - if (locked_ref) { - list_del_init(&locked_ref->cluster); - mutex_unlock(&locked_ref->mutex); - } - printk(KERN_DEBUG "btrfs: run_one_delayed_ref returned %d\n", ret); + btrfs_delayed_ref_unlock(locked_ref); + btrfs_put_delayed_ref(ref); + printk(KERN_DEBUG + "btrfs: run_one_delayed_ref returned %d\n", ret); spin_lock(&delayed_refs->lock); return ret; } + /* + * If this node is a head, that means all the refs in this head + * have been dealt with, and we will pick the next head to deal + * with, so we must unlock the head and drop it from the cluster + * list before we release it. + */ + if (btrfs_delayed_ref_is_head(ref)) { + list_del_init(&locked_ref->cluster); + btrfs_delayed_ref_unlock(locked_ref); + locked_ref = NULL; + } + btrfs_put_delayed_ref(ref); + count++; next: cond_resched(); spin_lock(&delayed_refs->lock); @@ -2500,6 +2505,7 @@ again: ret = run_clustered_refs(trans, root, &cluster); if (ret < 0) { + btrfs_release_ref_cluster(&cluster); spin_unlock(&delayed_refs->lock); btrfs_abort_transaction(trans, root, ret); return ret; -- cgit v1.2.3 From da633a42170165cbf20a2d3886c7480ccc832ec3 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 20 Dec 2012 11:19:09 +0000 Subject: Btrfs: flush all dirty inodes if writeback can not start We may try to flush some dirty pages when there is no enough space to reserve. But it is possible that this operation fails, in order to get enough space to reserve successfully, we will sync all the delalloc file. This operation is safe, we needn't worry about the case that the filesystem goes from r/w to r/o. because the filesystem should guarantee all the dirty pages have been written into the disk after it becomes readonly, so the sync operation will do nothing if the filesystem is already readonly. Though it may waste lots of time, as a corner case, we needn't care. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 40 +++++++++++++++++++++++++++++++--------- 1 file changed, 31 insertions(+), 9 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b4cb8186035f..d5e60d25ca51 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3695,12 +3695,15 @@ static int can_overcommit(struct btrfs_root *root, return 0; } -static int writeback_inodes_sb_nr_if_idle_safe(struct super_block *sb, - unsigned long nr_pages, - enum wb_reason reason) +static inline int writeback_inodes_sb_nr_if_idle_safe(struct super_block *sb, + unsigned long nr_pages, + enum wb_reason reason) { - if (!writeback_in_progress(sb->s_bdi) && - down_read_trylock(&sb->s_umount)) { + /* the flusher is dealing with the dirty inodes now. */ + if (writeback_in_progress(sb->s_bdi)) + return 1; + + if (down_read_trylock(&sb->s_umount)) { writeback_inodes_sb_nr(sb, nr_pages, reason); up_read(&sb->s_umount); return 1; @@ -3709,6 +3712,28 @@ static int writeback_inodes_sb_nr_if_idle_safe(struct super_block *sb, return 0; } +void btrfs_writeback_inodes_sb_nr(struct btrfs_root *root, + unsigned long nr_pages) +{ + struct super_block *sb = root->fs_info->sb; + int started; + + /* If we can not start writeback, just sync all the delalloc file. */ + started = writeback_inodes_sb_nr_if_idle_safe(sb, nr_pages, + WB_REASON_FS_FREE_SPACE); + if (!started) { + /* + * We needn't worry the filesystem going from r/w to r/o though + * we don't acquire ->s_umount mutex, because the filesystem + * should guarantee the delalloc inodes list be empty after + * the filesystem is readonly(all dirty pages are written to + * the disk). + */ + btrfs_start_delalloc_inodes(root, 0); + btrfs_wait_ordered_extents(root, 0); + } +} + /* * shrink metadata reservation for delalloc */ @@ -3741,10 +3766,7 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, while (delalloc_bytes && loops < 3) { max_reclaim = min(delalloc_bytes, to_reclaim); nr_pages = max_reclaim >> PAGE_CACHE_SHIFT; - writeback_inodes_sb_nr_if_idle_safe(root->fs_info->sb, - nr_pages, - WB_REASON_FS_FREE_SPACE); - + btrfs_writeback_inodes_sb_nr(root, nr_pages); /* * We need to wait for the async pages to actually start before * we do anything. -- cgit v1.2.3 From c6b305a89b1903d63652691ad5eb9f05aa0326b8 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 18 Dec 2012 09:16:16 -0500 Subject: Btrfs: don't re-enter when allocating a chunk If we start running low on metadata space we will try to allocate a chunk, which could then try to allocate a chunk to add the device entry. The thing is we allocate a chunk before we try really hard to make the allocation, so we should be able to find space for the device entry. Add a flag to the trans handle so we know we're currently allocating a chunk so we can just bail out if we try to allocate another chunk. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d5e60d25ca51..c642fc2cfc2b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3570,6 +3570,10 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans, int wait_for_alloc = 0; int ret = 0; + /* Don't re-enter if we're already allocating a chunk */ + if (trans->allocating_chunk) + return -ENOSPC; + space_info = __find_space_info(extent_root->fs_info, flags); if (!space_info) { ret = update_space_info(extent_root->fs_info, flags, @@ -3612,6 +3616,8 @@ again: goto again; } + trans->allocating_chunk = true; + /* * If we have mixed data/metadata chunks we want to make sure we keep * allocating mixed chunks instead of individual chunks. @@ -3638,6 +3644,7 @@ again: check_system_chunk(trans, extent_root, flags); ret = btrfs_alloc_chunk(trans, extent_root, flags); + trans->allocating_chunk = false; if (ret < 0 && ret != -ENOSPC) goto out; -- cgit v1.2.3 From 17b85495cf87ac9b2d9e0878f62fa9f3ae630a6e Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:17 +0000 Subject: Btrfs: remove deprecated comments commit d53ba47484ed6245e640ee4bfe9d21e9bfc15765 (Btrfs: use commit root when loading free space cache) has remove the deadlock check, and the related comments can be removed as well. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 6 ------ 1 file changed, 6 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index c642fc2cfc2b..a4a062a0a79c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -527,12 +527,6 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, cache->cached = BTRFS_CACHE_FAST; spin_unlock(&cache->lock); - /* - * We can't do the read from on-disk cache during a commit since we need - * to have the normal tree locking. Also if we are currently trying to - * allocate blocks for the tree root we can't do the fast caching since - * we likely hold important locks. - */ if (fs_info->mount_opt & BTRFS_MOUNT_SPACE_CACHE) { ret = load_free_space_cache(fs_info, cache); -- cgit v1.2.3 From f6373bf3dc6b78cd2c8cfba0215104003ba4a1d0 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:18 +0000 Subject: Btrfs: kill unused arguments of cache_block_group Argument 'trans' and 'root' are not used any more. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a4a062a0a79c..a9dc12d5ac1b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -468,8 +468,6 @@ out: } static int cache_block_group(struct btrfs_block_group_cache *cache, - struct btrfs_trans_handle *trans, - struct btrfs_root *root, int load_cache_only) { DEFINE_WAIT(wait); @@ -4802,7 +4800,7 @@ static int update_block_group(struct btrfs_trans_handle *trans, * space back to the block group, otherwise we will leak space. */ if (!alloc && cache->cached == BTRFS_CACHE_NO) - cache_block_group(cache, trans, NULL, 1); + cache_block_group(cache, 1); byte_in_group = bytenr - cache->key.objectid; WARN_ON(byte_in_group > cache->key.offset); @@ -4917,7 +4915,7 @@ int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans, * to one because the slow code to read in the free extents does check * the pinned extents. */ - cache_block_group(cache, trans, root, 1); + cache_block_group(cache, 1); pin_down_extent(root, cache, bytenr, num_bytes, 0); @@ -5707,8 +5705,7 @@ have_block_group: cached = block_group_cache_done(block_group); if (unlikely(!cached)) { found_uncached_bg = true; - ret = cache_block_group(block_group, trans, - orig_root, 0); + ret = cache_block_group(block_group, 0); BUG_ON(ret < 0); ret = 0; } @@ -6244,7 +6241,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, u64 num_bytes = ins->offset; block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid); - cache_block_group(block_group, trans, NULL, 0); + cache_block_group(block_group, 0); caching_ctl = get_caching_control(block_group); if (!caching_ctl) { @@ -8187,7 +8184,7 @@ int btrfs_trim_fs(struct btrfs_root *root, struct fstrim_range *range) if (end - start >= range->minlen) { if (!block_group_cache_done(cache)) { - ret = cache_block_group(cache, NULL, root, 0); + ret = cache_block_group(cache, 0); if (!ret) wait_block_group_cache_done(cache); } -- cgit v1.2.3 From c53d613e5293ef248e42195bb0595502b646a5a6 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:19 +0000 Subject: Btrfs: kill unused argument of update_block_group Argument 'trans' is not used any more. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a9dc12d5ac1b..a162c7cb8fb5 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -72,8 +72,7 @@ enum { RESERVE_ALLOC_NO_ACCOUNT = 2, }; -static int update_block_group(struct btrfs_trans_handle *trans, - struct btrfs_root *root, +static int update_block_group(struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc); static int __btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, @@ -4762,8 +4761,7 @@ void btrfs_delalloc_release_space(struct inode *inode, u64 num_bytes) btrfs_free_reserved_data_space(inode, num_bytes); } -static int update_block_group(struct btrfs_trans_handle *trans, - struct btrfs_root *root, +static int update_block_group(struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc) { struct btrfs_block_group_cache *cache = NULL; @@ -5312,7 +5310,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, } } - ret = update_block_group(trans, root, bytenr, num_bytes, 0); + ret = update_block_group(root, bytenr, num_bytes, 0); if (ret) { btrfs_abort_transaction(trans, extent_root, ret); goto out; @@ -6134,7 +6132,7 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(path->nodes[0]); btrfs_free_path(path); - ret = update_block_group(trans, root, ins->objectid, ins->offset, 1); + ret = update_block_group(root, ins->objectid, ins->offset, 1); if (ret) { /* -ENOENT, logic error */ printk(KERN_ERR "btrfs update block group failed for %llu " "%llu\n", (unsigned long long)ins->objectid, @@ -6198,7 +6196,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); btrfs_free_path(path); - ret = update_block_group(trans, root, ins->objectid, ins->offset, 1); + ret = update_block_group(root, ins->objectid, ins->offset, 1); if (ret) { /* -ENOENT, logic error */ printk(KERN_ERR "btrfs update block group failed for %llu " "%llu\n", (unsigned long long)ins->objectid, -- cgit v1.2.3 From dcfac4156fa102c1bab0e4e31df37e47278292f6 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:20 +0000 Subject: Btrfs: kill unused argument of btrfs_pin_extent_for_log_replay Argument 'trans' is not used any more. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a162c7cb8fb5..825f23b13b58 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4898,8 +4898,7 @@ int btrfs_pin_extent(struct btrfs_root *root, /* * this function must be called within transaction */ -int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans, - struct btrfs_root *root, +int btrfs_pin_extent_for_log_replay(struct btrfs_root *root, u64 bytenr, u64 num_bytes) { struct btrfs_block_group_cache *cache; -- cgit v1.2.3 From a1897fddd28daf6b23d05a30dc2a18836f77f8e3 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:23 +0000 Subject: Btrfs: record first logical byte in memory This'd save us a rbtree search which may become expensive in large filesystem. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 825f23b13b58..82400b2b2517 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -161,6 +161,10 @@ static int btrfs_add_block_group_cache(struct btrfs_fs_info *info, rb_link_node(&block_group->cache_node, parent, p); rb_insert_color(&block_group->cache_node, &info->block_group_cache_tree); + + if (info->first_logical_byte > block_group->key.objectid) + info->first_logical_byte = block_group->key.objectid; + spin_unlock(&info->block_group_cache_lock); return 0; @@ -202,8 +206,11 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr, break; } } - if (ret) + if (ret) { btrfs_get_block_group(ret); + if (bytenr == 0 && info->first_logical_byte > ret->key.objectid) + info->first_logical_byte = ret->key.objectid; + } spin_unlock(&info->block_group_cache_lock); return ret; @@ -4848,6 +4855,13 @@ static u64 first_logical_byte(struct btrfs_root *root, u64 search_start) struct btrfs_block_group_cache *cache; u64 bytenr; + spin_lock(&root->fs_info->block_group_cache_lock); + bytenr = root->fs_info->first_logical_byte; + spin_unlock(&root->fs_info->block_group_cache_lock); + + if (bytenr < (u64)-1) + return bytenr; + cache = btrfs_lookup_first_block_group(root->fs_info, search_start); if (!cache) return 0; @@ -8059,6 +8073,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, spin_lock(&root->fs_info->block_group_cache_lock); rb_erase(&block_group->cache_node, &root->fs_info->block_group_cache_tree); + + if (root->fs_info->first_logical_byte == block_group->key.objectid) + root->fs_info->first_logical_byte = (u64)-1; spin_unlock(&root->fs_info->block_group_cache_lock); down_write(&block_group->space_info->groups_sem); -- cgit v1.2.3 From e6ec716f0ddbe51741ef261d0804f0c28038dda4 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 17 Jan 2013 05:38:51 +0000 Subject: Btrfs: make raid attr array more readable The current code of raid attr arry is hard to understand and it is easy to introduce some problem if we modify the array. So I changed it and made it more readable. Cc: Liu Bo Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 82400b2b2517..174c4d5c692c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5545,20 +5545,16 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache) int __get_raid_index(u64 flags) { - int index; - if (flags & BTRFS_BLOCK_GROUP_RAID10) - index = 0; + return BTRFS_RAID_RAID10; else if (flags & BTRFS_BLOCK_GROUP_RAID1) - index = 1; + return BTRFS_RAID_RAID1; else if (flags & BTRFS_BLOCK_GROUP_DUP) - index = 2; + return BTRFS_RAID_DUP; else if (flags & BTRFS_BLOCK_GROUP_RAID0) - index = 3; + return BTRFS_RAID_RAID0; else - index = 4; - - return index; + return BTRFS_RAID_SINGLE; } static int get_block_group_index(struct btrfs_block_group_cache *cache) @@ -7518,16 +7514,16 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr) index = get_block_group_index(block_group); } - if (index == 0) { + if (index == BTRFS_RAID_RAID10) { dev_min = 4; /* Divide by 2 */ min_free >>= 1; - } else if (index == 1) { + } else if (index == BTRFS_RAID_RAID1) { dev_min = 2; - } else if (index == 2) { + } else if (index == BTRFS_RAID_DUP) { /* Multiply by 2 */ min_free <<= 1; - } else if (index == 3) { + } else if (index == BTRFS_RAID_RAID0) { dev_min = fs_devices->rw_devices; do_div(min_free, dev_min); } -- cgit v1.2.3 From 963d678b0f7649300e3a67f2513ca9d830c6e303 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:10:51 +0000 Subject: Btrfs: use percpu counter for fs_info->delalloc_bytes fs_info->delalloc_bytes is accessed very frequently, so use percpu counter instead of the u64 variant for it to reduce the lock contention. This patch also fixed the problem that we access the variant without the lock protection.At worst, we would not flush the delalloc inodes, and just return ENOSPC error when we still have some free space in the fs. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 174c4d5c692c..115d1646bf50 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3760,7 +3760,8 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, space_info = block_rsv->space_info; smp_mb(); - delalloc_bytes = root->fs_info->delalloc_bytes; + delalloc_bytes = percpu_counter_sum_positive( + &root->fs_info->delalloc_bytes); if (delalloc_bytes == 0) { if (trans) return; @@ -3799,7 +3800,8 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, break; } smp_mb(); - delalloc_bytes = root->fs_info->delalloc_bytes; + delalloc_bytes = percpu_counter_sum_positive( + &root->fs_info->delalloc_bytes); } } -- cgit v1.2.3 From de98ced9e743656d108de41841797def0f5cb951 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:13:12 +0000 Subject: Btrfs: use seqlock to protect fs_info->avail_{data, metadata, system}_alloc_bits There is no lock to protect fs_info->avail_{data, metadata, system}_alloc_bits, it may introduce some problem, such as the wrong profile information, so we add a seqlock to protect them. Signed-off-by: Zhao Lei Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 115d1646bf50..faff98f720de 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3227,12 +3227,14 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) u64 extra_flags = chunk_to_extended(flags) & BTRFS_EXTENDED_PROFILE_MASK; + write_seqlock(&fs_info->profiles_lock); if (flags & BTRFS_BLOCK_GROUP_DATA) fs_info->avail_data_alloc_bits |= extra_flags; if (flags & BTRFS_BLOCK_GROUP_METADATA) fs_info->avail_metadata_alloc_bits |= extra_flags; if (flags & BTRFS_BLOCK_GROUP_SYSTEM) fs_info->avail_system_alloc_bits |= extra_flags; + write_sequnlock(&fs_info->profiles_lock); } /* @@ -3324,12 +3326,18 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) { - if (flags & BTRFS_BLOCK_GROUP_DATA) - flags |= root->fs_info->avail_data_alloc_bits; - else if (flags & BTRFS_BLOCK_GROUP_SYSTEM) - flags |= root->fs_info->avail_system_alloc_bits; - else if (flags & BTRFS_BLOCK_GROUP_METADATA) - flags |= root->fs_info->avail_metadata_alloc_bits; + unsigned seq; + + do { + seq = read_seqbegin(&root->fs_info->profiles_lock); + + if (flags & BTRFS_BLOCK_GROUP_DATA) + flags |= root->fs_info->avail_data_alloc_bits; + else if (flags & BTRFS_BLOCK_GROUP_SYSTEM) + flags |= root->fs_info->avail_system_alloc_bits; + else if (flags & BTRFS_BLOCK_GROUP_METADATA) + flags |= root->fs_info->avail_metadata_alloc_bits; + } while (read_seqretry(&root->fs_info->profiles_lock, seq)); return btrfs_reduce_alloc_profile(root, flags); } @@ -7967,12 +7975,14 @@ static void clear_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) u64 extra_flags = chunk_to_extended(flags) & BTRFS_EXTENDED_PROFILE_MASK; + write_seqlock(&fs_info->profiles_lock); if (flags & BTRFS_BLOCK_GROUP_DATA) fs_info->avail_data_alloc_bits &= ~extra_flags; if (flags & BTRFS_BLOCK_GROUP_METADATA) fs_info->avail_metadata_alloc_bits &= ~extra_flags; if (flags & BTRFS_BLOCK_GROUP_SYSTEM) fs_info->avail_system_alloc_bits &= ~extra_flags; + write_sequnlock(&fs_info->profiles_lock); } int btrfs_remove_block_group(struct btrfs_trans_handle *trans, -- cgit v1.2.3 From 96f1bb57771f71bf1d55d5031a1cf47908494330 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 30 Jan 2013 17:02:51 -0500 Subject: Btrfs: do not overcommit if we don't have enough space for global rsv Because of how little we allocate chunks now we can get really tight on metadata space before we will allocate a new chunk. This resulted in being unable to add device extents when allocating a new metadata chunk as we did not have enough space. This is because we were allowed to overcommit too much metadata without actually making sure we had enough space to make allocations. The idea behind overcommit is that we are allowed to say "sure you can have that reservation" when most of the free space is occupied by reservations, not actual allocations. But in this case where a majority of the total space is in use by actual allocations we can screw ourselves by not being able to make real allocations when it matters. So make sure we have enough real space for our global reserve, and if not then don't allow overcommitting. Thanks, Reported-and-tested-by: Jim Schutt Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index faff98f720de..e035731b3608 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3672,13 +3672,30 @@ static int can_overcommit(struct btrfs_root *root, struct btrfs_space_info *space_info, u64 bytes, enum btrfs_reserve_flush_enum flush) { + struct btrfs_block_rsv *global_rsv = &root->fs_info->global_block_rsv; u64 profile = btrfs_get_alloc_profile(root, 0); + u64 rsv_size = 0; u64 avail; u64 used; used = space_info->bytes_used + space_info->bytes_reserved + - space_info->bytes_pinned + space_info->bytes_readonly + - space_info->bytes_may_use; + space_info->bytes_pinned + space_info->bytes_readonly; + + spin_lock(&global_rsv->lock); + rsv_size = global_rsv->size; + spin_unlock(&global_rsv->lock); + + /* + * We only want to allow over committing if we have lots of actual space + * free, but if we don't have enough space to handle the global reserve + * space then we could end up having a real enospc problem when trying + * to allocate a chunk or some other such important allocation. + */ + rsv_size <<= 1; + if (used + rsv_size >= space_info->total_bytes) + return 0; + + used += space_info->bytes_may_use; spin_lock(&root->fs_info->free_chunk_lock); avail = root->fs_info->free_chunk_space; -- cgit v1.2.3 From 1971e917c8c99ff190264305803fbafcbbac1422 Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Thu, 31 Jan 2013 00:55:00 +0000 Subject: btrfs: remove unnecessary DEFINE_WAIT() declarations No point in DEFINE_WAIT(wait) if it's not used! Signed-off-by: Eric Sandeen Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index e035731b3608..3158817cd5a9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5541,7 +5541,6 @@ wait_block_group_cache_progress(struct btrfs_block_group_cache *cache, u64 num_bytes) { struct btrfs_caching_control *caching_ctl; - DEFINE_WAIT(wait); caching_ctl = get_caching_control(cache); if (!caching_ctl) @@ -5558,7 +5557,6 @@ static noinline int wait_block_group_cache_done(struct btrfs_block_group_cache *cache) { struct btrfs_caching_control *caching_ctl; - DEFINE_WAIT(wait); caching_ctl = get_caching_control(cache); if (!caching_ctl) -- cgit v1.2.3 From 70afa3998c9baed4186df38988246de1abdab56d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 6 Feb 2013 13:53:19 -0500 Subject: Btrfs: rework the overcommit logic to be based on the total size People have been complaining about random ENOSPC errors that will clear up after a umount or just a given amount of time. Chris was able to reproduce this with stress.sh and lots of processes and so was I. Basically the overcommit stuff would really let us get out of hand, in my tests I saw up to 30 gigs of outstanding reservations with only 2 gigs total of metadata space. This usually worked out fine but with so much outstanding reservation the flushing stuff short circuits to make sure we don't hang forever flushing when we really need ENOSPC. Plus we allocate chunks in order to alleviate the pressure, but this doesn't actually help us since we only use the non-allocated area in our over commit logic. So instead of basing overcommit on the amount of non-allocated space, instead just do it based on how much total space we have, and then limit it to the non-allocated space in case we are short on space to spill over into. This allows us to have the same performance as well as no longer giving random ENOSPC. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 3158817cd5a9..81aa7cf3ae86 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3677,6 +3677,7 @@ static int can_overcommit(struct btrfs_root *root, u64 rsv_size = 0; u64 avail; u64 used; + u64 to_add; used = space_info->bytes_used + space_info->bytes_reserved + space_info->bytes_pinned + space_info->bytes_readonly; @@ -3710,17 +3711,25 @@ static int can_overcommit(struct btrfs_root *root, BTRFS_BLOCK_GROUP_RAID10)) avail >>= 1; + to_add = space_info->total_bytes; + /* * If we aren't flushing all things, let us overcommit up to * 1/2th of the space. If we can flush, don't let us overcommit * too much, let it overcommit up to 1/8 of the space. */ if (flush == BTRFS_RESERVE_FLUSH_ALL) - avail >>= 3; + to_add >>= 3; else - avail >>= 1; + to_add >>= 1; + + /* + * Limit the overcommit to the amount of free space we could possibly + * allocate for chunks. + */ + to_add = min(avail, to_add); - if (used + bytes < space_info->total_bytes + avail) + if (used + bytes < space_info->total_bytes + to_add) return 1; return 0; } -- cgit v1.2.3 From 5d80366e9b5e56b3ffc1923b4995e83bbbf605e3 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 7 Feb 2013 16:06:02 -0500 Subject: Btrfs: steal from global reserve if we are cleaning up orphans Sometimes xfstest 83 will fail to remount the scratch device because we've gotten ourselves so full that we cannot cleanup the orphan items. In this case check to see if we're doing the orphan cleanup and if we are allow us to steal our reservation from the global block rsv. With this patch I've not been able to reproduce the failed mount problem. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 81aa7cf3ae86..1818dd90c27e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -102,6 +102,8 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes, int dump_block_groups); static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, u64 num_bytes, int reserve); +static int block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv, + u64 num_bytes); static noinline int block_group_cache_done(struct btrfs_block_group_cache *cache) @@ -4099,6 +4101,15 @@ again: goto again; out: + if (ret == -ENOSPC && + unlikely(root->orphan_cleanup_state == ORPHAN_CLEANUP_STARTED)) { + struct btrfs_block_rsv *global_rsv = + &root->fs_info->global_block_rsv; + + if (block_rsv != global_rsv && + !block_rsv_use_bytes(global_rsv, orig_bytes)) + ret = 0; + } if (flushing) { spin_lock(&space_info->lock); space_info->flush = 0; -- cgit v1.2.3 From 0934856d4697e63c14056375e26e3bd6e8ebd34b Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 7 Feb 2013 10:12:07 +0000 Subject: Btrfs: fix deadlock due to unsubmitted The deadlock problem happened when running fsstress(a test program in LTP). Steps to reproduce: # mkfs.btrfs -b 100M # mount # /fsstress -p 3 -n 10000000 -d The reason is: btrfs_direct_IO() |->do_direct_IO() |->get_page() |->get_blocks() | |->btrfs_delalloc_resereve_space() | |->btrfs_add_ordered_extent() ------- Add a new ordered extent |->dio_send_cur_page(page0) -------------- We didn't submit bio here |->get_page() |->get_blocks() |->btrfs_delalloc_resereve_space() |->flush_space() |->btrfs_start_ordered_extent() |->wait_event() ---------- Wait the completion of the ordered extent that is mentioned above But because we didn't submit the bio that is mentioned above, the ordered extent can not complete, we would wait for its completion forever. There are two methods which can fix this deadlock problem: 1. submit the bio before we invoke get_blocks() 2. reserve the space before we do dio Though the 1st is the simplest way, we need modify the code of VFS, and it is likely to break contiguous requests, and introduce performance regression for the other filesystems. So we have to choose the 2nd way. Signed-off-by: Miao Xie Cc: Josef Bacik Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 1818dd90c27e..51e1151a0a02 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4748,7 +4748,8 @@ void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes) spin_lock(&BTRFS_I(inode)->lock); dropped = drop_outstanding_extent(inode); - to_free = calc_csum_metadata_size(inode, num_bytes, 0); + if (num_bytes) + to_free = calc_csum_metadata_size(inode, num_bytes, 0); spin_unlock(&BTRFS_I(inode)->lock); if (dropped > 0) to_free += btrfs_calc_trans_metadata_size(root, dropped); -- cgit v1.2.3 From b069e0c3450ae388d7a9c94ded6d938a465de262 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 8 Feb 2013 21:28:17 +0000 Subject: btrfs: put some enospc messages under enospc_debug The warning in use_block_rsv is not useful for users and may fill the logs unnecessarily. Signed-off-by: David Sterba Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 51e1151a0a02..88831fab85d6 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6408,12 +6408,14 @@ use_block_rsv(struct btrfs_trans_handle *trans, if (!ret) return block_rsv; if (ret && !block_rsv->failfast) { - static DEFINE_RATELIMIT_STATE(_rs, - DEFAULT_RATELIMIT_INTERVAL, - /*DEFAULT_RATELIMIT_BURST*/ 2); - if (__ratelimit(&_rs)) - WARN(1, KERN_DEBUG "btrfs: block rsv returned %d\n", - ret); + if (btrfs_test_opt(root, ENOSPC_DEBUG)) { + static DEFINE_RATELIMIT_STATE(_rs, + DEFAULT_RATELIMIT_INTERVAL * 10, + /*DEFAULT_RATELIMIT_BURST*/ 1); + if (__ratelimit(&_rs)) + WARN(1, KERN_DEBUG + "btrfs: block rsv returned %d\n", ret); + } ret = reserve_metadata_bytes(root, block_rsv, blocksize, BTRFS_RESERVE_NO_FLUSH); if (!ret) { @@ -7730,11 +7732,13 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) space_info = list_entry(info->space_info.next, struct btrfs_space_info, list); - if (space_info->bytes_pinned > 0 || - space_info->bytes_reserved > 0 || - space_info->bytes_may_use > 0) { - WARN_ON(1); - dump_space_info(space_info, 0, 0); + if (btrfs_test_opt(info->tree_root, ENOSPC_DEBUG)) { + if (space_info->bytes_pinned > 0 || + space_info->bytes_reserved > 0 || + space_info->bytes_may_use > 0) { + WARN_ON(1); + dump_space_info(space_info, 0, 0); + } } list_del(&space_info->list); kfree(space_info); -- cgit v1.2.3 From 24542bf7ea5e4fdfdb5157ff544c093fa4dcb536 Mon Sep 17 00:00:00 2001 From: Zach Brown Date: Fri, 16 Nov 2012 00:04:43 +0000 Subject: btrfs: limit fallocate extent reservation to 256MB Very large fallocate requests are cpu bound and result in extents with a repeating pattern of ever decreasing size: $ time fallocate -l 1T file real 0m13.039s ( an excerpt of the extents from btrfs-debug-tree: ) prealloc data disk byte 1536292564992 nr 397312 prealloc data disk byte 1536292962304 nr 196608 prealloc data disk byte 1536293158912 nr 98304 prealloc data disk byte 1536293257216 nr 49152 prealloc data disk byte 1536293306368 nr 24576 prealloc data disk byte 1536293330944 nr 12288 prealloc data disk byte 1536293343232 nr 8192 prealloc data disk byte 1536293351424 nr 4096 prealloc data disk byte 1536293355520 nr 4096 prealloc data disk byte 1536293359616 nr 4096 The excessive cpu use comes from __btrfs_prealloc_file_range() trying to allocate the entire remaining size after each extent is allocated. btrfs_reserve_extent() repeatedly cuts this requested size in half until it gets down to the size that the allocators can return. We limit the problem for now by capping each reservation at 256 meg. The small extents come from a masking bug when decreasing the requested reservation size. The high 32bits are cleared and the remaining low bits might happen to reserve a small size. Fix this by using round_down() which properly casts the mask. After these fixes huge fallocate requests are fast and result in nice large extents: $ time fallocate -l 1T file real 0m0.082s prealloc data disk byte 1112425889792 nr 268435456 prealloc data disk byte 1112694325248 nr 268435456 prealloc data disk byte 1112962760704 nr 268435456 Reported-by: Eric Sandeen Signed-off-by: Zach Brown Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b3ecca447ddf..d2b3a5e9a621 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6143,7 +6143,7 @@ again: if (ret == -ENOSPC) { if (!final_tried) { num_bytes = num_bytes >> 1; - num_bytes = num_bytes & ~(root->sectorsize - 1); + num_bytes = round_down(num_bytes, root->sectorsize); num_bytes = max(num_bytes, min_alloc_size); if (num_bytes == min_alloc_size) final_tried = true; -- cgit v1.2.3 From a81cb9a2d939676dee6911993a37fa7e818ab72e Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Thu, 21 Feb 2013 21:15:14 +0000 Subject: clear chunk_alloc flag on retryable failure I've experienced filesystem freezes with permanent spikes in the active process count for quite a while, particularly on filesystems whose available raw space has already been fully allocated to chunks. While looking into this, I found a pretty obvious error in do_chunk_alloc: it sets space_info->chunk_alloc, but if btrfs_alloc_chunk returns an error other than ENOSPC, it returns leaving that flag set, which causes any other threads waiting for space_info->chunk_alloc to become zero to spin indefinitely. I haven't double-checked that this patch fixes the failure I've observed fully (it's not exactly trivial to trigger), but it surely is a bug and the fix is trivial, so... Please put it in :-) What I saw in that function also happens to explain why in some cases I see filesystems allocate a huge number of chunks that remain unused (leading to the scenario above, of not having more chunks to allocate). It happens for data and metadata, but not necessarily both. I'm guessing some thread sets the force_alloc flag on the corresponding space_info, and then several threads trying to get disk space end up attempting to allocate a new chunk concurrently. All of them will see the force_alloc flag and bump their local copy of force up to the level they see first, and they won't clear it even if another thread succeeds in allocating a chunk, thus clearing the force flag. Then each thread that observed the force flag will, on its turn, force the allocation of a new chunk. And any threads that come in while it does that will see the force flag still set and pick it up, and so on. This sounds like a problem to me, but... what should the correct behavior be? Clear force_flag once we copy it to a local force? Reset force to the incoming value on every loop? Set the flag to our incoming force if we have it at first, clear our local flag, and move it from the space_info when we determined that we are the thread that's going to perform the allocation? btrfs: clear chunk_alloc flag on retryable failure From: Alexandre Oliva If btrfs_alloc_chunk fails with e.g. ENOMEM, we exit do_chunk_alloc without clearing chunk_alloc in space_info. As a result, any further calls to do_chunk_alloc on that filesystem will start busy-waiting for chunk_alloc to be cleared, but it never will be. This patch adjusts do_chunk_alloc so that it clears this flag in case of an error. Signed-off-by: Alexandre Oliva Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d2b3a5e9a621..8520354f086e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3715,19 +3715,19 @@ again: ret = btrfs_alloc_chunk(trans, extent_root, flags); trans->allocating_chunk = false; - if (ret < 0 && ret != -ENOSPC) - goto out; spin_lock(&space_info->lock); + if (ret < 0 && ret != -ENOSPC) + goto out; if (ret) space_info->full = 1; else ret = 1; space_info->force_alloc = CHUNK_ALLOC_NO_FORCE; +out: space_info->chunk_alloc = 0; spin_unlock(&space_info->lock); -out: mutex_unlock(&fs_info->chunk_mutex); return ret; } -- cgit v1.2.3 From fda2832febb1928da0625b2c5d15559b29d7e740 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Tue, 26 Feb 2013 08:10:22 +0000 Subject: btrfs: cleanup for open-coded alignment Though most of the btrfs codes are using ALIGN macro for page alignment, there are still some codes using open-coded alignment like the following: ------ u64 mask = ((u64)root->stripesize - 1); u64 ret = (val + mask) & ~mask; ------ Or even hidden one: ------ num_bytes = (end - start + blocksize) & ~(blocksize - 1); ------ Sometimes these open-coded alignment is not so easy to understand for newbie like me. This commit changes the open-coded alignment to the ALIGN macro for a better readability. Also there is a previous patch from David Sterba with similar changes, but the patch is for 3.2 kernel and seems not merged. http://www.spinics.net/lists/linux-btrfs/msg12747.html Cc: David Sterba Signed-off-by: Qu Wenruo Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 8520354f086e..5681a91ed400 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3431,7 +3431,7 @@ int btrfs_check_data_free_space(struct inode *inode, u64 bytes) int ret = 0, committed = 0, alloc_chunk = 1; /* make sure bytes are sectorsize aligned */ - bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); + bytes = ALIGN(bytes, root->sectorsize); if (root == root->fs_info->tree_root || BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID) { @@ -3526,7 +3526,7 @@ void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes) struct btrfs_space_info *data_sinfo; /* make sure bytes are sectorsize aligned */ - bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); + bytes = ALIGN(bytes, root->sectorsize); data_sinfo = root->fs_info->data_sinfo; spin_lock(&data_sinfo->lock); @@ -5607,10 +5607,7 @@ static u64 stripe_align(struct btrfs_root *root, struct btrfs_block_group_cache *cache, u64 val, u64 num_bytes) { - u64 mask; - u64 ret; - mask = ((u64)root->stripesize - 1); - ret = (val + mask) & ~mask; + u64 ret = ALIGN(val, root->stripesize); return ret; } -- cgit v1.2.3 From d5c1207017cd8387b4d3224dd7ab6cf5cd7f1c9a Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 28 Feb 2013 10:04:33 +0000 Subject: Btrfs: fix wrong reserved space in qgroup during snap/subv creation There are two problems in the space reservation of the snapshot/ subvolume creation. - don't reserve the space for the root item insertion - the space which is reserved in the qgroup is different with the free space reservation. we need reserve free space for 7 items, but in qgroup reservation, we need reserve space only for 3 items. So we implement new metadata reservation functions for the snapshot/subvolume creation. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 65 ++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 53 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5681a91ed400..7cb9d734f6e5 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4560,19 +4560,60 @@ void btrfs_orphan_release_metadata(struct inode *inode) btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes); } -int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans, - struct btrfs_pending_snapshot *pending) +/* + * btrfs_subvolume_reserve_metadata() - reserve space for subvolume operation + * root: the root of the parent directory + * rsv: block reservation + * items: the number of items that we need do reservation + * qgroup_reserved: used to return the reserved size in qgroup + * + * This function is used to reserve the space for snapshot/subvolume + * creation and deletion. Those operations are different with the + * common file/directory operations, they change two fs/file trees + * and root tree, the number of items that the qgroup reserves is + * different with the free space reservation. So we can not use + * the space reseravtion mechanism in start_transaction(). + */ +int btrfs_subvolume_reserve_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + int items, + u64 *qgroup_reserved) { - struct btrfs_root *root = pending->root; - struct btrfs_block_rsv *src_rsv = get_block_rsv(trans, root); - struct btrfs_block_rsv *dst_rsv = &pending->block_rsv; - /* - * two for root back/forward refs, two for directory entries, - * one for root of the snapshot and one for parent inode. - */ - u64 num_bytes = btrfs_calc_trans_metadata_size(root, 6); - dst_rsv->space_info = src_rsv->space_info; - return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes); + u64 num_bytes; + int ret; + + if (root->fs_info->quota_enabled) { + /* One for parent inode, two for dir entries */ + num_bytes = 3 * root->leafsize; + ret = btrfs_qgroup_reserve(root, num_bytes); + if (ret) + return ret; + } else { + num_bytes = 0; + } + + *qgroup_reserved = num_bytes; + + num_bytes = btrfs_calc_trans_metadata_size(root, items); + rsv->space_info = __find_space_info(root->fs_info, + BTRFS_BLOCK_GROUP_METADATA); + ret = btrfs_block_rsv_add(root, rsv, num_bytes, + BTRFS_RESERVE_FLUSH_ALL); + if (ret) { + if (*qgroup_reserved) + btrfs_qgroup_free(root, *qgroup_reserved); + } + + return ret; +} + +void btrfs_subvolume_release_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + u64 qgroup_reserved) +{ + btrfs_block_rsv_release(root, rsv, (u64)-1); + if (qgroup_reserved) + btrfs_qgroup_free(root, qgroup_reserved); } /** -- cgit v1.2.3 From a9870c0e031527fbfa382019f30d2e9b98124a0d Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Fri, 1 Mar 2013 11:33:01 +0000 Subject: Btrfs: don't call btrfs_qgroup_free if just btrfs_qgroup_reserve fails commit eb6b88d92c6df083dd09a8c471011e3788dfd7c6 leads into another bug. If it is just because qgroup_reserve fails, the function btrfs_qgroup_free should not be called, otherwise, it will cause the wrong quota accounting. Signed-off-by: Wang Shilong Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 7cb9d734f6e5..075854e7a044 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4772,9 +4772,14 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) * ret != 0 here means the qgroup reservation failed, we go straight to * the shared error handling then. */ - if (ret == 0) + if (ret == 0) { ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush); + if (ret && root->fs_info->quota_enabled) { + btrfs_qgroup_free(root, num_bytes + + nr_extents * root->leafsize); + } + } if (ret) { u64 to_free = 0; @@ -4805,10 +4810,6 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) btrfs_ino(inode), to_free, 0); } - if (root->fs_info->quota_enabled) { - btrfs_qgroup_free(root, num_bytes + - nr_extents * root->leafsize); - } if (delalloc_lock) mutex_unlock(&BTRFS_I(inode)->delalloc_mutex); return ret; -- cgit v1.2.3 From 88e081bf82ffcb9be3ad33d04c829051d66da564 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Fri, 1 Mar 2013 11:36:01 +0000 Subject: Btrfs: cleanup to make the function btrfs_delalloc_reserve_metadata more logic The original code is a little confusing and not clear, The right way to deal with the kernel code like this: [...] if (ret) goto out; [...] So i move the common clean_up code to the place labeled with out_fail, this will be easier to maintain. Signed-off-by: Wang Shilong Signed-off-by: Josef Bacik --- fs/btrfs/extent-tree.c | 82 +++++++++++++++++++++++--------------------------- 1 file changed, 38 insertions(+), 44 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 075854e7a044..aaee2b7fee78 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4721,6 +4721,8 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_ALL; int ret = 0; bool delalloc_lock = true; + u64 to_free = 0; + unsigned dropped; /* If we are a free space inode we need to not flush since we will be in * the middle of a transaction commit. We also don't need the delalloc @@ -4764,55 +4766,19 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) csum_bytes = BTRFS_I(inode)->csum_bytes; spin_unlock(&BTRFS_I(inode)->lock); - if (root->fs_info->quota_enabled) + if (root->fs_info->quota_enabled) { ret = btrfs_qgroup_reserve(root, num_bytes + nr_extents * root->leafsize); + if (ret) + goto out_fail; + } - /* - * ret != 0 here means the qgroup reservation failed, we go straight to - * the shared error handling then. - */ - if (ret == 0) { - ret = reserve_metadata_bytes(root, block_rsv, - to_reserve, flush); - if (ret && root->fs_info->quota_enabled) { + ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush); + if (unlikely(ret)) { + if (root->fs_info->quota_enabled) btrfs_qgroup_free(root, num_bytes + nr_extents * root->leafsize); - } - } - - if (ret) { - u64 to_free = 0; - unsigned dropped; - - spin_lock(&BTRFS_I(inode)->lock); - dropped = drop_outstanding_extent(inode); - /* - * If the inodes csum_bytes is the same as the original - * csum_bytes then we know we haven't raced with any free()ers - * so we can just reduce our inodes csum bytes and carry on. - * Otherwise we have to do the normal free thing to account for - * the case that the free side didn't free up its reserve - * because of this outstanding reservation. - */ - if (BTRFS_I(inode)->csum_bytes == csum_bytes) - calc_csum_metadata_size(inode, num_bytes, 0); - else - to_free = calc_csum_metadata_size(inode, num_bytes, 0); - spin_unlock(&BTRFS_I(inode)->lock); - if (dropped) - to_free += btrfs_calc_trans_metadata_size(root, dropped); - - if (to_free) { - btrfs_block_rsv_release(root, block_rsv, to_free); - trace_btrfs_space_reservation(root->fs_info, - "delalloc", - btrfs_ino(inode), - to_free, 0); - } - if (delalloc_lock) - mutex_unlock(&BTRFS_I(inode)->delalloc_mutex); - return ret; + goto out_fail; } spin_lock(&BTRFS_I(inode)->lock); @@ -4833,6 +4799,34 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) block_rsv_add_bytes(block_rsv, to_reserve, 1); return 0; + +out_fail: + spin_lock(&BTRFS_I(inode)->lock); + dropped = drop_outstanding_extent(inode); + /* + * If the inodes csum_bytes is the same as the original + * csum_bytes then we know we haven't raced with any free()ers + * so we can just reduce our inodes csum bytes and carry on. + * Otherwise we have to do the normal free thing to account for + * the case that the free side didn't free up its reserve + * because of this outstanding reservation. + */ + if (BTRFS_I(inode)->csum_bytes == csum_bytes) + calc_csum_metadata_size(inode, num_bytes, 0); + else + to_free = calc_csum_metadata_size(inode, num_bytes, 0); + spin_unlock(&BTRFS_I(inode)->lock); + if (dropped) + to_free += btrfs_calc_trans_metadata_size(root, dropped); + + if (to_free) { + btrfs_block_rsv_release(root, block_rsv, to_free); + trace_btrfs_space_reservation(root->fs_info, "delalloc", + btrfs_ino(inode), to_free, 0); + } + if (delalloc_lock) + mutex_unlock(&BTRFS_I(inode)->delalloc_mutex); + return ret; } /** -- cgit v1.2.3 From 492104c866cb1b62a11393adccb477f5cd2c7768 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 8 Mar 2013 15:41:02 -0500 Subject: Btrfs: return EIO if we have extent tree corruption The callers of lookup_inline_extent_info all handle getting an error back properly, so return an error if we have corruption instead of being a jerk and panicing. Still WARN_ON() since this is kind of crucial and I've been seeing it a bit too much recently for my taste, I think we're doing something wrong somewhere. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index aaee2b7fee78..350b9b18140c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -1467,8 +1467,11 @@ int lookup_inline_extent_backref(struct btrfs_trans_handle *trans, if (ret && !insert) { err = -ENOENT; goto out; + } else if (ret) { + err = -EIO; + WARN_ON(1); + goto out; } - BUG_ON(ret); /* Corruption */ leaf = path->nodes[0]; item_size = btrfs_item_size_nr(leaf, path->slots[0]); -- cgit v1.2.3