From a482039889b85c45fc9616e65d560db7a35d4f54 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Mon, 9 Sep 2013 13:19:42 +0800 Subject: Btrfs: allocate the free space by the existed max extent size when ENOSPC By the current code, if the requested size is very large, and all the extents in the free space cache are small, we will waste lots of the cpu time to cut the requested size in half and search the cache again and again until it gets down to the size the allocator can return. In fact, we can know the max extent size in the cache after the first search, so we needn't cut the size in half repeatedly, and just use the max extent size directly. This way can save lots of cpu time and make the performance grow up when there are only fragments in the free space cache. According to my test, if there are only 4KB free space extents in the fs, and the total size of those extents are 256MB, we can reduce the execute time of the following test from 5.4s to 1.4s. dd if=/dev/zero of= bs=1MB count=1 oflag=sync Changelog v2 -> v3: - fix the problem that we skip the block group with the space which is less than we need. Changelog v1 -> v2: - address the problem that we return a wrong start position when searching the free space in a bitmap. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 33 ++++++++++++++++++++++++--------- 1 file changed, 24 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 cfb3cf711b34..2f03181b777f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6117,10 +6117,13 @@ enum btrfs_loop_type { /* * walks the btree of allocated extents and find a hole of a given size. * The key ins is changed to record the hole: - * ins->objectid == block start + * ins->objectid == start position * ins->flags = BTRFS_EXTENT_ITEM_KEY - * ins->offset == number of blocks + * ins->offset == the size of the hole. * Any available blocks before search_start are skipped. + * + * If there is no suitable free space, we will record the max size of + * the free space extent currently. */ static noinline int find_free_extent(struct btrfs_root *orig_root, u64 num_bytes, u64 empty_size, @@ -6133,6 +6136,7 @@ static noinline int find_free_extent(struct btrfs_root *orig_root, struct btrfs_block_group_cache *block_group = NULL; struct btrfs_block_group_cache *used_block_group; u64 search_start = 0; + u64 max_extent_size = 0; int empty_cluster = 2 * 1024 * 1024; struct btrfs_space_info *space_info; int loop = 0; @@ -6292,7 +6296,10 @@ have_block_group: btrfs_get_block_group(used_block_group); offset = btrfs_alloc_from_cluster(used_block_group, - last_ptr, num_bytes, used_block_group->key.objectid); + last_ptr, + num_bytes, + used_block_group->key.objectid, + &max_extent_size); if (offset) { /* we have a block, we're done */ spin_unlock(&last_ptr->refill_lock); @@ -6355,8 +6362,10 @@ refill_cluster: * cluster */ offset = btrfs_alloc_from_cluster(block_group, - last_ptr, num_bytes, - search_start); + last_ptr, + num_bytes, + search_start, + &max_extent_size); if (offset) { /* we found one, proceed */ spin_unlock(&last_ptr->refill_lock); @@ -6391,13 +6400,18 @@ unclustered_alloc: if (cached && block_group->free_space_ctl->free_space < num_bytes + empty_cluster + empty_size) { + if (block_group->free_space_ctl->free_space > + max_extent_size) + max_extent_size = + block_group->free_space_ctl->free_space; spin_unlock(&block_group->free_space_ctl->tree_lock); goto loop; } spin_unlock(&block_group->free_space_ctl->tree_lock); offset = btrfs_find_space_for_alloc(block_group, search_start, - num_bytes, empty_size); + num_bytes, empty_size, + &max_extent_size); /* * If we didn't find a chunk, and we haven't failed on this * block group before, and this block group is in the middle of @@ -6515,7 +6529,8 @@ loop: ret = 0; } out: - + if (ret == -ENOSPC) + ins->offset = max_extent_size; return ret; } @@ -6573,8 +6588,8 @@ again: flags); if (ret == -ENOSPC) { - if (!final_tried) { - num_bytes = num_bytes >> 1; + if (!final_tried && ins->offset) { + num_bytes = min(num_bytes >> 1, ins->offset); num_bytes = round_down(num_bytes, root->sectorsize); num_bytes = max(num_bytes, min_alloc_size); if (num_bytes == min_alloc_size) -- cgit v1.2.3-59-g8ed1b From 14575aef4212d2cc02274bf1f5456f3e644f03f8 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 17 Sep 2013 10:48:00 -0400 Subject: Revert "Btrfs: rework the overcommit logic to be based on the total size" This reverts commit 70afa3998c9baed4186df38988246de1abdab56d. It is causing performance issues and wasn't actually correct. There were problems with the way we flushed delalloc and that was the real cause of the early enospc. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 15 +++------------ 1 file changed, 3 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 2f03181b777f..07f7f6942128 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3925,7 +3925,6 @@ static int can_overcommit(struct btrfs_root *root, u64 space_size; u64 avail; u64 used; - u64 to_add; used = space_info->bytes_used + space_info->bytes_reserved + space_info->bytes_pinned + space_info->bytes_readonly; @@ -3959,25 +3958,17 @@ 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) - to_add >>= 3; + avail >>= 3; else - to_add >>= 1; - - /* - * Limit the overcommit to the amount of free space we could possibly - * allocate for chunks. - */ - to_add = min(avail, to_add); + avail >>= 1; - if (used + bytes < space_info->total_bytes + to_add) + if (used + bytes < space_info->total_bytes + avail) return 1; return 0; } -- cgit v1.2.3-59-g8ed1b From f0de181c9b48a397c5a2fbe63dcdd2a26a872695 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 17 Sep 2013 10:55:51 -0400 Subject: Btrfs: kill delay_iput arg to the wait_ordered functions This is a left over of how we used to wait for ordered extents, which was to grab the inode and then run filemap flush on it. However if we have an ordered extent then we already are holding a ref on the inode, and we just use btrfs_start_ordered_extent anyway, so there is no reason to have an extra ref on the inode to start work on the ordered extent. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/dev-replace.c | 4 ++-- fs/btrfs/extent-tree.c | 6 +++--- fs/btrfs/ioctl.c | 2 +- fs/btrfs/ordered-data.c | 24 +++--------------------- fs/btrfs/ordered-data.h | 5 ++--- fs/btrfs/relocation.c | 2 +- fs/btrfs/super.c | 2 +- fs/btrfs/transaction.c | 2 +- 8 files changed, 14 insertions(+), 33 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c index af800ef677a0..5d844438b2d4 100644 --- a/fs/btrfs/dev-replace.c +++ b/fs/btrfs/dev-replace.c @@ -400,7 +400,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root, args->result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR; btrfs_dev_replace_unlock(dev_replace); - btrfs_wait_all_ordered_extents(root->fs_info, 0); + btrfs_wait_all_ordered_extents(root->fs_info); /* force writing the updated state information to disk */ trans = btrfs_start_transaction(root, 0); @@ -475,7 +475,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info, mutex_unlock(&dev_replace->lock_finishing_cancel_unmount); return ret; } - btrfs_wait_all_ordered_extents(root->fs_info, 0); + btrfs_wait_all_ordered_extents(root->fs_info); trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) { diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 07f7f6942128..88c6db611de3 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3991,7 +3991,7 @@ static void btrfs_writeback_inodes_sb_nr(struct btrfs_root *root, */ btrfs_start_all_delalloc_inodes(root->fs_info, 0); if (!current->journal_info) - btrfs_wait_all_ordered_extents(root->fs_info, 0); + btrfs_wait_all_ordered_extents(root->fs_info); } } @@ -4021,7 +4021,7 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, if (delalloc_bytes == 0) { if (trans) return; - btrfs_wait_all_ordered_extents(root->fs_info, 0); + btrfs_wait_all_ordered_extents(root->fs_info); return; } @@ -4049,7 +4049,7 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, loops++; if (wait_ordered && !trans) { - btrfs_wait_all_ordered_extents(root->fs_info, 0); + btrfs_wait_all_ordered_extents(root->fs_info); } else { time_left = schedule_timeout_killable(1); if (time_left) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 1a5b9462dd9a..fd6557e7118a 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -574,7 +574,7 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir, if (ret) return ret; - btrfs_wait_ordered_extents(root, 0); + btrfs_wait_ordered_extents(root); pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS); if (!pending_snapshot) diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index 966b413a33b8..c702cb62f78a 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -563,11 +563,10 @@ static void btrfs_run_ordered_extent_work(struct btrfs_work *work) * wait for all the ordered extents in a root. This is done when balancing * space between drives. */ -void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput) +void btrfs_wait_ordered_extents(struct btrfs_root *root) { struct list_head splice, works; struct btrfs_ordered_extent *ordered, *next; - struct inode *inode; INIT_LIST_HEAD(&splice); INIT_LIST_HEAD(&works); @@ -580,15 +579,6 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput) root_extent_list); list_move_tail(&ordered->root_extent_list, &root->ordered_extents); - /* - * the inode may be getting freed (in sys_unlink path). - */ - inode = igrab(ordered->inode); - if (!inode) { - cond_resched_lock(&root->ordered_extent_lock); - continue; - } - atomic_inc(&ordered->refs); spin_unlock(&root->ordered_extent_lock); @@ -605,21 +595,13 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput) list_for_each_entry_safe(ordered, next, &works, work_list) { list_del_init(&ordered->work_list); wait_for_completion(&ordered->completion); - - inode = ordered->inode; btrfs_put_ordered_extent(ordered); - if (delay_iput) - btrfs_add_delayed_iput(inode); - else - iput(inode); - cond_resched(); } mutex_unlock(&root->fs_info->ordered_operations_mutex); } -void btrfs_wait_all_ordered_extents(struct btrfs_fs_info *fs_info, - int delay_iput) +void btrfs_wait_all_ordered_extents(struct btrfs_fs_info *fs_info) { struct btrfs_root *root; struct list_head splice; @@ -637,7 +619,7 @@ void btrfs_wait_all_ordered_extents(struct btrfs_fs_info *fs_info, &fs_info->ordered_roots); spin_unlock(&fs_info->ordered_root_lock); - btrfs_wait_ordered_extents(root, delay_iput); + btrfs_wait_ordered_extents(root); btrfs_put_fs_root(root); spin_lock(&fs_info->ordered_root_lock); diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index d9a5aa097b4f..0c0b35612d7a 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -195,9 +195,8 @@ int btrfs_run_ordered_operations(struct btrfs_trans_handle *trans, void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode); -void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput); -void btrfs_wait_all_ordered_extents(struct btrfs_fs_info *fs_info, - int delay_iput); +void btrfs_wait_ordered_extents(struct btrfs_root *root); +void btrfs_wait_all_ordered_extents(struct btrfs_fs_info *fs_info); void btrfs_get_logged_extents(struct btrfs_root *log, struct inode *inode); void btrfs_wait_logged_extents(struct btrfs_root *log, u64 transid); void btrfs_free_logged_extents(struct btrfs_root *log, u64 transid); diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 09b2b9a0a491..a5a26320503f 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -4246,7 +4246,7 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start) err = ret; goto out; } - btrfs_wait_all_ordered_extents(fs_info, 0); + btrfs_wait_all_ordered_extents(fs_info); while (1) { mutex_lock(&fs_info->cleaner_mutex); diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 58cb83f23b1a..b341da3ee3ee 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -921,7 +921,7 @@ int btrfs_sync_fs(struct super_block *sb, int wait) return 0; } - btrfs_wait_all_ordered_extents(fs_info, 1); + btrfs_wait_all_ordered_extents(fs_info); trans = btrfs_attach_transaction_barrier(root); if (IS_ERR(trans)) { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index cac4a3f76323..e7a95356df83 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -1603,7 +1603,7 @@ static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info) static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info) { if (btrfs_test_opt(fs_info->tree_root, FLUSHONCOMMIT)) - btrfs_wait_all_ordered_extents(fs_info, 1); + btrfs_wait_all_ordered_extents(fs_info); } int btrfs_commit_transaction(struct btrfs_trans_handle *trans, -- cgit v1.2.3-59-g8ed1b From 363e4d354e51558629d13c5ae456f98c4b209576 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 17 Sep 2013 11:02:25 -0400 Subject: Btrfs: remove space_info->reservation_progress This isn't used for anything anymore, just remove it. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 9 --------- fs/btrfs/extent-tree.c | 3 --- 2 files changed, 12 deletions(-) (limited to 'fs/btrfs/extent-tree.c') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 36490b92ee33..0506f40ede83 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1118,15 +1118,6 @@ struct btrfs_space_info { */ struct percpu_counter total_bytes_pinned; - /* - * we bump reservation progress every time we decrement - * bytes_reserved. This way people waiting for reservations - * know something good has happened and they can check - * for progress. The number here isn't to be trusted, it - * just shows reclaim activity - */ - unsigned long reservation_progress; - unsigned int full:1; /* indicates that we cannot allocate any more chunks for this space */ unsigned int chunk_alloc:1; /* set if we are allocating a chunk */ diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 88c6db611de3..d58bef130a41 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4456,7 +4456,6 @@ static void block_rsv_release_bytes(struct btrfs_fs_info *fs_info, space_info->bytes_may_use -= num_bytes; trace_btrfs_space_reservation(fs_info, "space_info", space_info->flags, num_bytes, 0); - space_info->reservation_progress++; spin_unlock(&space_info->lock); } } @@ -4657,7 +4656,6 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info) sinfo->bytes_may_use -= num_bytes; trace_btrfs_space_reservation(fs_info, "space_info", sinfo->flags, num_bytes, 0); - sinfo->reservation_progress++; block_rsv->reserved = block_rsv->size; block_rsv->full = 1; } @@ -5437,7 +5435,6 @@ static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, space_info->bytes_readonly += num_bytes; cache->reserved -= num_bytes; space_info->bytes_reserved -= num_bytes; - space_info->reservation_progress++; } spin_unlock(&cache->lock); spin_unlock(&space_info->lock); -- cgit v1.2.3-59-g8ed1b