From 8e62c2de6e23e5c1fee04f59de51b54cc2868ca5 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 12 Apr 2012 13:46:48 -0400 Subject: Revert "Btrfs: increase the global block reserve estimates" This reverts commit 5500cdbe14d7435e04f66ff3cfb8ecd8b8e44ebf. We've had a number of complaints of early enospc that bisect down to this patch. We'll hae to fix the reservations differently. CC: stable@kernel.org 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 a84420491c1..ace5e8cef03 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4205,7 +4205,7 @@ static u64 calc_global_metadata_size(struct btrfs_fs_info *fs_info) num_bytes += div64_u64(data_used + meta_used, 50); if (num_bytes * 3 > meta_used) - num_bytes = div64_u64(meta_used, 3) * 2; + num_bytes = div64_u64(meta_used, 3); return ALIGN(num_bytes, fs_info->extent_root->leafsize << 10); } -- cgit v1.2.3-70-g09d2 From c6664b42c4e567792abdb17c958fb01c5bcfcb3a Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 12 Apr 2012 16:03:56 -0400 Subject: Btrfs: remove lock assert from get_restripe_target() This fixes a regression introduced by fc67c450. spin_is_locked() always returns 0 on UP kernels, which caused assert in get_restripe_target() to be fired on every call from btrfs_reduce_alloc_profile() on UP systems. Remove it completely for now, it's not clear if it's going to be needed in future. Reported-by: Bobby Powers Reported-by: Mitch Harder Tested-by: Mitch Harder Signed-off-by: Ilya Dryomov Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 5 ++--- 1 file changed, 2 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 ace5e8cef03..a2134d8141c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3152,15 +3152,14 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) /* * returns target flags in extended format or 0 if restripe for this * chunk_type is not in progress + * + * should be called with either volume_mutex or balance_lock held */ static u64 get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags) { struct btrfs_balance_control *bctl = fs_info->balance_ctl; u64 target = 0; - BUG_ON(!mutex_is_locked(&fs_info->volume_mutex) && - !spin_is_locked(&fs_info->balance_lock)); - if (!bctl) return 0; -- cgit v1.2.3-70-g09d2 From d53ba47484ed6245e640ee4bfe9d21e9bfc15765 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 12 Apr 2012 16:03:57 -0400 Subject: Btrfs: use commit root when loading free space cache A user reported that booting his box up with btrfs root on 3.4 was way slower than on 3.3 because I removed the ideal caching code. It turns out that we don't load the free space cache if we're in a commit for deadlock reasons, but since we're reading the cache and it hasn't changed yet we are safe reading the inode and free space item from the commit root, so do that and remove all of the deadlock checks so we don't unnecessarily skip loading the free space cache. The user reported this fixed the slowness. Thanks, Tested-by: Calvin Walton Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 +--- fs/btrfs/free-space-cache.c | 9 ++------- 2 files changed, 3 insertions(+), 10 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a2134d8141c..2b35f8d14bb 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -529,9 +529,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, * allocate blocks for the tree root we can't do the fast caching since * we likely hold important locks. */ - if (trans && (!trans->transaction->in_commit) && - (root && root != root->fs_info->tree_root) && - btrfs_test_opt(root, SPACE_CACHE)) { + if (fs_info->mount_opt & BTRFS_MOUNT_SPACE_CACHE) { ret = load_free_space_cache(fs_info, cache); spin_lock(&cache->lock); diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 054707ed579..baaa518baaf 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -747,13 +747,6 @@ int load_free_space_cache(struct btrfs_fs_info *fs_info, bool matched; u64 used = btrfs_block_group_used(&block_group->item); - /* - * If we're unmounting then just return, since this does a search on the - * normal root and not the commit root and we could deadlock. - */ - if (btrfs_fs_closing(fs_info)) - return 0; - /* * If this block group has been marked to be cleared for one reason or * another then we can't trust the on disk cache, so just return. @@ -768,6 +761,8 @@ int load_free_space_cache(struct btrfs_fs_info *fs_info, path = btrfs_alloc_path(); if (!path) return 0; + path->search_commit_root = 1; + path->skip_locking = 1; inode = lookup_free_space_inode(root, block_group, path); if (IS_ERR(inode)) { -- cgit v1.2.3-70-g09d2 From 253beebd5a255e07d6a8b65515491f33664e82a2 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 18 Apr 2012 09:59:03 +0300 Subject: Btrfs: double unlock bug in error handling The caller expects this function to return with the lock held and releases it immediately on error. Signed-off-by: Dan Carpenter --- fs/btrfs/extent-tree.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 2b35f8d14bb..a0bb9dcd3c3 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2301,6 +2301,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, if (ret) { printk(KERN_DEBUG "btrfs: run_delayed_extent_op returned %d\n", ret); + spin_lock(&delayed_refs->lock); return ret; } @@ -2331,6 +2332,7 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans, if (ret) { printk(KERN_DEBUG "btrfs: run_one_delayed_ref returned %d\n", ret); + spin_lock(&delayed_refs->lock); return ret; } -- cgit v1.2.3-70-g09d2 From b9688bb8459b67e42327de6420edb405a9188775 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Wed, 18 Apr 2012 10:27:16 +0200 Subject: btrfs: don't return EINTR It is basically a good thing if we are interruptible when waiting for free space, but the generality in which it is implemented currently leads to system calls being interruptible that are not documented this way. For example git can't handle interrupted unlink(), leading to corrupt repos under space pressure. Instead we raise the bar to only be interruptible by SIGKILL. Thanks to David Sterba for suggesting this. Signed-off-by: Arne Jansen --- 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 a0bb9dcd3c3..84497f8eb04 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3771,13 +3771,10 @@ again: */ if (current->journal_info) return -EAGAIN; - ret = wait_event_interruptible(space_info->wait, - !space_info->flush); - /* Must have been interrupted, return */ - if (ret) { - printk(KERN_DEBUG "btrfs: %s returning -EINTR\n", __func__); + ret = wait_event_killable(space_info->wait, !space_info->flush); + /* Must have been killed, return */ + if (ret) return -EINTR; - } spin_lock(&space_info->lock); } -- cgit v1.2.3-70-g09d2 From 1f699d38b6556c393ac80f1c23c2053502a51631 Mon Sep 17 00:00:00 2001 From: Stefan Behrens Date: Fri, 27 Apr 2012 12:41:46 -0400 Subject: Btrfs: fix block_rsv and space_info lock ordering may_commit_transaction() calls spin_lock(&space_info->lock); spin_lock(&delayed_rsv->lock); and update_global_block_rsv() calls spin_lock(&block_rsv->lock); spin_lock(&sinfo->lock); Lockdep complains about this at run time. Everywhere except in update_global_block_rsv(), the space_info lock is the outer lock, therefore the locking order in update_global_block_rsv() is changed. Signed-off-by: Stefan Behrens Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 4 ++-- 1 file changed, 2 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 84497f8eb04..6fc2e6f5aab 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4214,8 +4214,8 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) num_bytes = calc_global_metadata_size(fs_info); - spin_lock(&block_rsv->lock); spin_lock(&sinfo->lock); + spin_lock(&block_rsv->lock); block_rsv->size = num_bytes; @@ -4241,8 +4241,8 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) block_rsv->full = 1; } - spin_unlock(&sinfo->lock); spin_unlock(&block_rsv->lock); + spin_unlock(&sinfo->lock); } static void init_global_block_rsv(struct btrfs_fs_info *fs_info) -- cgit v1.2.3-70-g09d2 From b9fab919b748c7b39c19ff236ed6c5682c266dde Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Sun, 6 May 2012 07:23:47 -0400 Subject: Btrfs: avoid sleeping in verify_parent_transid while atomic verify_parent_transid needs to lock the extent range to make sure no IO is underway, and so it can safely clear the uptodate bits if our checks fail. But, a few callers are using it with spinlocks held. Most of the time, the generation numbers are going to match, and we don't want to switch to a blocking lock just for the error case. This adds an atomic flag to verify_parent_transid, and changes it to return EAGAIN if it needs to block to properly verifiy things. Signed-off-by: Chris Mason --- fs/btrfs/ctree.c | 26 +++++++++++++++++--------- fs/btrfs/disk-io.c | 18 +++++++++++++----- fs/btrfs/disk-io.h | 3 ++- fs/btrfs/extent-tree.c | 2 +- fs/btrfs/tree-log.c | 2 +- 5 files changed, 34 insertions(+), 17 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 086303b9be6..4106264fbc6 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -725,7 +725,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, cur = btrfs_find_tree_block(root, blocknr, blocksize); if (cur) - uptodate = btrfs_buffer_uptodate(cur, gen); + uptodate = btrfs_buffer_uptodate(cur, gen, 0); else uptodate = 0; if (!cur || !uptodate) { @@ -1360,7 +1360,12 @@ static noinline int reada_for_balance(struct btrfs_root *root, block1 = btrfs_node_blockptr(parent, slot - 1); gen = btrfs_node_ptr_generation(parent, slot - 1); eb = btrfs_find_tree_block(root, block1, blocksize); - if (eb && btrfs_buffer_uptodate(eb, gen)) + /* + * if we get -eagain from btrfs_buffer_uptodate, we + * don't want to return eagain here. That will loop + * forever + */ + if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0) block1 = 0; free_extent_buffer(eb); } @@ -1368,7 +1373,7 @@ static noinline int reada_for_balance(struct btrfs_root *root, block2 = btrfs_node_blockptr(parent, slot + 1); gen = btrfs_node_ptr_generation(parent, slot + 1); eb = btrfs_find_tree_block(root, block2, blocksize); - if (eb && btrfs_buffer_uptodate(eb, gen)) + if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0) block2 = 0; free_extent_buffer(eb); } @@ -1506,8 +1511,9 @@ read_block_for_search(struct btrfs_trans_handle *trans, tmp = btrfs_find_tree_block(root, blocknr, blocksize); if (tmp) { - if (btrfs_buffer_uptodate(tmp, 0)) { - if (btrfs_buffer_uptodate(tmp, gen)) { + /* first we do an atomic uptodate check */ + if (btrfs_buffer_uptodate(tmp, 0, 1) > 0) { + if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) { /* * we found an up to date block without * sleeping, return @@ -1525,8 +1531,9 @@ read_block_for_search(struct btrfs_trans_handle *trans, free_extent_buffer(tmp); btrfs_set_path_blocking(p); + /* now we're allowed to do a blocking uptodate check */ tmp = read_tree_block(root, blocknr, blocksize, gen); - if (tmp && btrfs_buffer_uptodate(tmp, gen)) { + if (tmp && btrfs_buffer_uptodate(tmp, gen, 0) > 0) { *eb_ret = tmp; return 0; } @@ -1561,7 +1568,7 @@ read_block_for_search(struct btrfs_trans_handle *trans, * and give up so that our caller doesn't loop forever * on our EAGAINs. */ - if (!btrfs_buffer_uptodate(tmp, 0)) + if (!btrfs_buffer_uptodate(tmp, 0, 0)) ret = -EIO; free_extent_buffer(tmp); } @@ -4045,7 +4052,7 @@ again: tmp = btrfs_find_tree_block(root, blockptr, btrfs_level_size(root, level - 1)); - if (tmp && btrfs_buffer_uptodate(tmp, gen)) { + if (tmp && btrfs_buffer_uptodate(tmp, gen, 1) > 0) { free_extent_buffer(tmp); break; } @@ -4168,7 +4175,8 @@ next: struct extent_buffer *cur; cur = btrfs_find_tree_block(root, blockptr, btrfs_level_size(root, level - 1)); - if (!cur || !btrfs_buffer_uptodate(cur, gen)) { + if (!cur || + btrfs_buffer_uptodate(cur, gen, 1) <= 0) { slot++; if (cur) free_extent_buffer(cur); diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index d0c969beaad..a7ffc88a7db 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -323,7 +323,8 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf, * in the wrong place. */ static int verify_parent_transid(struct extent_io_tree *io_tree, - struct extent_buffer *eb, u64 parent_transid) + struct extent_buffer *eb, u64 parent_transid, + int atomic) { struct extent_state *cached_state = NULL; int ret; @@ -331,6 +332,9 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, if (!parent_transid || btrfs_header_generation(eb) == parent_transid) return 0; + if (atomic) + return -EAGAIN; + lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1, 0, &cached_state); if (extent_buffer_uptodate(eb) && @@ -372,7 +376,8 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, ret = read_extent_buffer_pages(io_tree, eb, start, WAIT_COMPLETE, btree_get_extent, mirror_num); - if (!ret && !verify_parent_transid(io_tree, eb, parent_transid)) + if (!ret && !verify_parent_transid(io_tree, eb, + parent_transid, 0)) break; /* @@ -1202,7 +1207,7 @@ static int __must_check find_and_setup_root(struct btrfs_root *tree_root, root->commit_root = NULL; root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item), blocksize, generation); - if (!root->node || !btrfs_buffer_uptodate(root->node, generation)) { + if (!root->node || !btrfs_buffer_uptodate(root->node, generation, 0)) { free_extent_buffer(root->node); root->node = NULL; return -EIO; @@ -3143,7 +3148,8 @@ int close_ctree(struct btrfs_root *root) return 0; } -int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) +int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid, + int atomic) { int ret; struct inode *btree_inode = buf->pages[0]->mapping->host; @@ -3153,7 +3159,9 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) return ret; ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf, - parent_transid); + parent_transid, atomic); + if (ret == -EAGAIN) + return ret; return !ret; } diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index a7ace1a2dd1..ab1830aaf0e 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -66,7 +66,8 @@ void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr); void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr); void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root); void btrfs_mark_buffer_dirty(struct extent_buffer *buf); -int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid); +int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid, + int atomic); int btrfs_set_buffer_uptodate(struct extent_buffer *buf); int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid); u32 btrfs_csum_data(struct btrfs_root *root, char *data, u32 seed, size_t len); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 6fc2e6f5aab..49fd7b66d57 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6568,7 +6568,7 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans, goto skip; } - if (!btrfs_buffer_uptodate(next, generation)) { + if (!btrfs_buffer_uptodate(next, generation, 0)) { btrfs_tree_unlock(next); free_extent_buffer(next); next = NULL; diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index d017283ae6f..eb1ae908582 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -279,7 +279,7 @@ static int process_one_buffer(struct btrfs_root *log, log->fs_info->extent_root, eb->start, eb->len); - if (btrfs_buffer_uptodate(eb, gen)) { + if (btrfs_buffer_uptodate(eb, gen, 0)) { if (wc->write) btrfs_write_tree_block(eb); if (wc->wait) -- cgit v1.2.3-70-g09d2