From 98cfee214394a3560bd4ce3209b55a71c4267783 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Sat, 1 Feb 2014 00:42:05 +0800 Subject: Btrfs: only add roots if necessary in find_parent_nodes() find_all_leafs() dosen't need add all roots actually, add roots only if we need, this can avoid unnecessary ulist dance. Signed-off-by: Wang Shilong Signed-off-by: Josef Bacik --- fs/btrfs/backref.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/backref.c') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index aded3ef3d3d4..903fe68e017b 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -965,7 +965,7 @@ again: while (!list_empty(&prefs)) { ref = list_first_entry(&prefs, struct __prelim_ref, list); WARN_ON(ref->count < 0); - if (ref->count && ref->root_id && ref->parent == 0) { + if (roots && ref->count && ref->root_id && ref->parent == 0) { /* no parent == root of tree */ ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS); if (ret < 0) @@ -1061,22 +1061,14 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans, u64 time_seq, struct ulist **leafs, const u64 *extent_item_pos) { - struct ulist *tmp; int ret; - tmp = ulist_alloc(GFP_NOFS); - if (!tmp) - return -ENOMEM; *leafs = ulist_alloc(GFP_NOFS); - if (!*leafs) { - ulist_free(tmp); + if (!*leafs) return -ENOMEM; - } ret = find_parent_nodes(trans, fs_info, bytenr, - time_seq, *leafs, tmp, extent_item_pos); - ulist_free(tmp); - + time_seq, *leafs, NULL, extent_item_pos); if (ret < 0 && ret != -ENOENT) { free_leaf_list(*leafs); return ret; -- cgit v1.2.3 From 850a8cdffe41abec9e3319d7801c49eced0778a1 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Thu, 6 Feb 2014 20:02:29 +0800 Subject: Btrfs: switch to btrfs_previous_extent_item() Since we have introduced btrfs_previous_extent_item() to search previous extent item, just switch into it. Signed-off-by: Wang Shilong Reviewed-by: Filipe Manana Signed-off-by: Josef Bacik --- fs/btrfs/backref.c | 37 ++++++------------------------------- 1 file changed, 6 insertions(+), 31 deletions(-) (limited to 'fs/btrfs/backref.c') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 903fe68e017b..a88da721dfc5 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -1325,38 +1325,13 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical, if (ret < 0) return ret; - while (1) { - u32 nritems; - if (path->slots[0] == 0) { - btrfs_set_path_blocking(path); - ret = btrfs_prev_leaf(fs_info->extent_root, path); - if (ret != 0) { - if (ret > 0) { - pr_debug("logical %llu is not within " - "any extent\n", logical); - ret = -ENOENT; - } - return ret; - } - } else { - path->slots[0]--; - } - nritems = btrfs_header_nritems(path->nodes[0]); - if (nritems == 0) { - pr_debug("logical %llu is not within any extent\n", - logical); - return -ENOENT; - } - if (path->slots[0] == nritems) - path->slots[0]--; - - btrfs_item_key_to_cpu(path->nodes[0], found_key, - path->slots[0]); - if (found_key->type == BTRFS_EXTENT_ITEM_KEY || - found_key->type == BTRFS_METADATA_ITEM_KEY) - break; + ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0); + if (ret) { + if (ret > 0) + ret = -ENOENT; + return ret; } - + btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]); if (found_key->type == BTRFS_METADATA_ITEM_KEY) size = fs_info->extent_root->leafsize; else if (found_key->type == BTRFS_EXTENT_ITEM_KEY) -- cgit v1.2.3 From e84752d434b5cca0869e906e7b94d0531b25c6d3 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Thu, 13 Feb 2014 11:19:47 +0800 Subject: Btrfs: skip locking when searching commit root We won't change commit root, skip locking dance with commit root when walking backrefs, this can speed up btrfs send operations. Signed-off-by: Wang Shilong Signed-off-by: Josef Bacik --- fs/btrfs/backref.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/backref.c') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index a88da721dfc5..860f4f22b9b0 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -873,8 +873,10 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans, path = btrfs_alloc_path(); if (!path) return -ENOMEM; - if (!trans) + if (!trans) { path->search_commit_root = 1; + path->skip_locking = 1; + } /* * grab both a lock on the path and a lock on the delayed ref head. -- cgit v1.2.3 From 4485386853454f184235c8a973b29fa7fa522eb1 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 19 Mar 2014 13:35:14 -0400 Subject: Btrfs: take into account total references when doing backref lookup I added an optimization for large files where we would stop searching for backrefs once we had looked at the number of references we currently had for this extent. This works great most of the time, but for snapshots that point to this extent and has changes in the original root this assumption falls on it face. So keep track of any delayed ref mods made and add in the actual ref count as reported by the extent item and use that to limit how far down an inode we'll search for extents. Thanks, Reportedy-by: Hugo Mills Signed-off-by: Josef Bacik Reported-by: Hugo Mills Tested-by: Hugo Mills Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/backref.c') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 860f4f22b9b0..aad7201ad11b 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -220,7 +220,8 @@ static int __add_prelim_ref(struct list_head *head, u64 root_id, static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path, struct ulist *parents, struct __prelim_ref *ref, - int level, u64 time_seq, const u64 *extent_item_pos) + int level, u64 time_seq, const u64 *extent_item_pos, + u64 total_refs) { int ret = 0; int slot; @@ -249,7 +250,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path, if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) ret = btrfs_next_old_leaf(root, path, time_seq); - while (!ret && count < ref->count) { + while (!ret && count < total_refs) { eb = path->nodes[0]; slot = path->slots[0]; @@ -306,7 +307,7 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info, struct btrfs_path *path, u64 time_seq, struct __prelim_ref *ref, struct ulist *parents, - const u64 *extent_item_pos) + const u64 *extent_item_pos, u64 total_refs) { struct btrfs_root *root; struct btrfs_key root_key; @@ -361,7 +362,7 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info, } ret = add_all_parents(root, path, parents, ref, level, time_seq, - extent_item_pos); + extent_item_pos, total_refs); out: path->lowest_level = 0; btrfs_release_path(path); @@ -374,7 +375,7 @@ out: static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info, struct btrfs_path *path, u64 time_seq, struct list_head *head, - const u64 *extent_item_pos) + const u64 *extent_item_pos, u64 total_refs) { int err; int ret = 0; @@ -400,7 +401,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info, if (ref->count == 0) continue; err = __resolve_indirect_ref(fs_info, path, time_seq, ref, - parents, extent_item_pos); + parents, extent_item_pos, + total_refs); /* * we can only tolerate ENOENT,otherwise,we should catch error * and return directly. @@ -557,7 +559,7 @@ static void __merge_refs(struct list_head *head, int mode) * smaller or equal that seq to the list */ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq, - struct list_head *prefs) + struct list_head *prefs, u64 *total_refs) { struct btrfs_delayed_extent_op *extent_op = head->extent_op; struct rb_node *n = &head->node.rb_node; @@ -593,6 +595,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq, default: BUG_ON(1); } + *total_refs += (node->ref_mod * sgn); switch (node->type) { case BTRFS_TREE_BLOCK_REF_KEY: { struct btrfs_delayed_tree_ref *ref; @@ -653,7 +656,8 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq, */ static int __add_inline_refs(struct btrfs_fs_info *fs_info, struct btrfs_path *path, u64 bytenr, - int *info_level, struct list_head *prefs) + int *info_level, struct list_head *prefs, + u64 *total_refs) { int ret = 0; int slot; @@ -677,6 +681,7 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info, ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item); flags = btrfs_extent_flags(leaf, ei); + *total_refs += btrfs_extent_refs(leaf, ei); btrfs_item_key_to_cpu(leaf, &found_key, slot); ptr = (unsigned long)(ei + 1); @@ -859,6 +864,7 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans, struct list_head prefs; struct __prelim_ref *ref; struct extent_inode_elem *eie = NULL; + u64 total_refs = 0; INIT_LIST_HEAD(&prefs); INIT_LIST_HEAD(&prefs_delayed); @@ -917,7 +923,7 @@ again: } spin_unlock(&delayed_refs->lock); ret = __add_delayed_refs(head, time_seq, - &prefs_delayed); + &prefs_delayed, &total_refs); mutex_unlock(&head->mutex); if (ret) goto out; @@ -938,7 +944,8 @@ again: (key.type == BTRFS_EXTENT_ITEM_KEY || key.type == BTRFS_METADATA_ITEM_KEY)) { ret = __add_inline_refs(fs_info, path, bytenr, - &info_level, &prefs); + &info_level, &prefs, + &total_refs); if (ret) goto out; ret = __add_keyed_refs(fs_info, path, bytenr, @@ -958,7 +965,7 @@ again: __merge_refs(&prefs, 1); ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs, - extent_item_pos); + extent_item_pos, total_refs); if (ret) goto out; -- cgit v1.2.3 From 9e351cc862b098d8ec8f8022d110932490794925 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 13 Mar 2014 15:42:13 -0400 Subject: Btrfs: remove transaction from send Lets try this again. We can deadlock the box if we send on a box and try to write onto the same fs with the app that is trying to listen to the send pipe. This is because the writer could get stuck waiting for a transaction commit which is being blocked by the send. So fix this by making sure looking at the commit roots is always going to be consistent. We do this by keeping track of which roots need to have their commit roots swapped during commit, and then taking the commit_root_sem and swapping them all at once. Then make sure we take a read lock on the commit_root_sem in cases where we search the commit root to make sure we're always looking at a consistent view of the commit roots. Previously we had problems with this because we would swap a fs tree commit root and then swap the extent tree commit root independently which would cause the backref walking code to screw up sometimes. With this patch we no longer deadlock and pass all the weird send/receive corner cases. Thanks, Reportedy-by: Hugo Mills Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 33 +++++++++++++++---- fs/btrfs/ctree.c | 88 -------------------------------------------------- fs/btrfs/ctree.h | 3 +- fs/btrfs/disk-io.c | 3 +- fs/btrfs/extent-tree.c | 20 ++++++------ fs/btrfs/inode-map.c | 14 ++++---- fs/btrfs/send.c | 57 ++------------------------------ fs/btrfs/transaction.c | 45 ++++++++++++++++---------- fs/btrfs/transaction.h | 1 + 9 files changed, 77 insertions(+), 187 deletions(-) (limited to 'fs/btrfs/backref.c') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index aad7201ad11b..10db21fa0926 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -330,7 +330,10 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info, goto out; } - root_level = btrfs_old_root_level(root, time_seq); + if (path->search_commit_root) + root_level = btrfs_header_level(root->commit_root); + else + root_level = btrfs_old_root_level(root, time_seq); if (root_level + 1 == level) { srcu_read_unlock(&fs_info->subvol_srcu, index); @@ -1099,9 +1102,9 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans, * * returns 0 on success, < 0 on error. */ -int btrfs_find_all_roots(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 bytenr, - u64 time_seq, struct ulist **roots) +static int __btrfs_find_all_roots(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 bytenr, + u64 time_seq, struct ulist **roots) { struct ulist *tmp; struct ulist_node *node = NULL; @@ -1137,6 +1140,20 @@ int btrfs_find_all_roots(struct btrfs_trans_handle *trans, return 0; } +int btrfs_find_all_roots(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 bytenr, + u64 time_seq, struct ulist **roots) +{ + int ret; + + if (!trans) + down_read(&fs_info->commit_root_sem); + ret = __btrfs_find_all_roots(trans, fs_info, bytenr, time_seq, roots); + if (!trans) + up_read(&fs_info->commit_root_sem); + return ret; +} + /* * this makes the path point to (inum INODE_ITEM ioff) */ @@ -1516,6 +1533,8 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info, if (IS_ERR(trans)) return PTR_ERR(trans); btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem); + } else { + down_read(&fs_info->commit_root_sem); } ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid, @@ -1526,8 +1545,8 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info, ULIST_ITER_INIT(&ref_uiter); while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) { - ret = btrfs_find_all_roots(trans, fs_info, ref_node->val, - tree_mod_seq_elem.seq, &roots); + ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val, + tree_mod_seq_elem.seq, &roots); if (ret) break; ULIST_ITER_INIT(&root_uiter); @@ -1549,6 +1568,8 @@ out: if (!search_commit_root) { btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem); btrfs_end_transaction(trans, fs_info->extent_root); + } else { + up_read(&fs_info->commit_root_sem); } return ret; diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 88d1b1eedc9c..9d89c1648e8e 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -5360,7 +5360,6 @@ int btrfs_compare_trees(struct btrfs_root *left_root, { int ret; int cmp; - struct btrfs_trans_handle *trans = NULL; struct btrfs_path *left_path = NULL; struct btrfs_path *right_path = NULL; struct btrfs_key left_key; @@ -5378,9 +5377,6 @@ int btrfs_compare_trees(struct btrfs_root *left_root, u64 right_blockptr; u64 left_gen; u64 right_gen; - u64 left_start_ctransid; - u64 right_start_ctransid; - u64 ctransid; left_path = btrfs_alloc_path(); if (!left_path) { @@ -5404,21 +5400,6 @@ int btrfs_compare_trees(struct btrfs_root *left_root, right_path->search_commit_root = 1; right_path->skip_locking = 1; - spin_lock(&left_root->root_item_lock); - left_start_ctransid = btrfs_root_ctransid(&left_root->root_item); - spin_unlock(&left_root->root_item_lock); - - spin_lock(&right_root->root_item_lock); - right_start_ctransid = btrfs_root_ctransid(&right_root->root_item); - spin_unlock(&right_root->root_item_lock); - - trans = btrfs_join_transaction(left_root); - if (IS_ERR(trans)) { - ret = PTR_ERR(trans); - trans = NULL; - goto out; - } - /* * Strategy: Go to the first items of both trees. Then do * @@ -5482,67 +5463,6 @@ int btrfs_compare_trees(struct btrfs_root *left_root, advance_left = advance_right = 0; while (1) { - /* - * We need to make sure the transaction does not get committed - * while we do anything on commit roots. This means, we need to - * join and leave transactions for every item that we process. - */ - if (trans && btrfs_should_end_transaction(trans, left_root)) { - btrfs_release_path(left_path); - btrfs_release_path(right_path); - - ret = btrfs_end_transaction(trans, left_root); - trans = NULL; - if (ret < 0) - goto out; - } - /* now rejoin the transaction */ - if (!trans) { - trans = btrfs_join_transaction(left_root); - if (IS_ERR(trans)) { - ret = PTR_ERR(trans); - trans = NULL; - goto out; - } - - spin_lock(&left_root->root_item_lock); - ctransid = btrfs_root_ctransid(&left_root->root_item); - spin_unlock(&left_root->root_item_lock); - if (ctransid != left_start_ctransid) - left_start_ctransid = 0; - - spin_lock(&right_root->root_item_lock); - ctransid = btrfs_root_ctransid(&right_root->root_item); - spin_unlock(&right_root->root_item_lock); - if (ctransid != right_start_ctransid) - right_start_ctransid = 0; - - if (!left_start_ctransid || !right_start_ctransid) { - WARN(1, KERN_WARNING - "BTRFS: btrfs_compare_tree detected " - "a change in one of the trees while " - "iterating. This is probably a " - "bug.\n"); - ret = -EIO; - goto out; - } - - /* - * the commit root may have changed, so start again - * where we stopped - */ - left_path->lowest_level = left_level; - right_path->lowest_level = right_level; - ret = btrfs_search_slot(NULL, left_root, - &left_key, left_path, 0, 0); - if (ret < 0) - goto out; - ret = btrfs_search_slot(NULL, right_root, - &right_key, right_path, 0, 0); - if (ret < 0) - goto out; - } - if (advance_left && !left_end_reached) { ret = tree_advance(left_root, left_path, &left_level, left_root_level, @@ -5672,14 +5592,6 @@ out: btrfs_free_path(left_path); btrfs_free_path(right_path); kfree(tmp_buf); - - if (trans) { - if (!ret) - ret = btrfs_end_transaction(trans, left_root); - else - btrfs_end_transaction(trans, left_root); - } - return ret; } diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 2a9d32e193a5..4253ab257c9c 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1440,7 +1440,7 @@ struct btrfs_fs_info { */ struct mutex ordered_extent_flush_mutex; - struct rw_semaphore extent_commit_sem; + struct rw_semaphore commit_root_sem; struct rw_semaphore cleanup_work_sem; @@ -1711,7 +1711,6 @@ struct btrfs_root { struct btrfs_block_rsv *block_rsv; /* free ino cache stuff */ - struct mutex fs_commit_mutex; struct btrfs_free_space_ctl *free_ino_ctl; enum btrfs_caching_type cached; spinlock_t cache_lock; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 98fe70193397..6d1ac7d46f81 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1567,7 +1567,6 @@ int btrfs_init_fs_root(struct btrfs_root *root) root->subv_writers = writers; btrfs_init_free_ino_ctl(root); - mutex_init(&root->fs_commit_mutex); spin_lock_init(&root->cache_lock); init_waitqueue_head(&root->cache_wait); @@ -2345,7 +2344,7 @@ int open_ctree(struct super_block *sb, mutex_init(&fs_info->transaction_kthread_mutex); mutex_init(&fs_info->cleaner_mutex); mutex_init(&fs_info->volume_mutex); - init_rwsem(&fs_info->extent_commit_sem); + init_rwsem(&fs_info->commit_root_sem); init_rwsem(&fs_info->cleanup_work_sem); init_rwsem(&fs_info->subvol_sem); sema_init(&fs_info->uuid_tree_rescan_sem, 1); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index e09db2c2f3b4..4d2508bbf6f0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -419,7 +419,7 @@ static noinline void caching_thread(struct btrfs_work *work) again: mutex_lock(&caching_ctl->mutex); /* need to make sure the commit_root doesn't disappear */ - down_read(&fs_info->extent_commit_sem); + down_read(&fs_info->commit_root_sem); next: ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0); @@ -443,10 +443,10 @@ next: break; if (need_resched() || - rwsem_is_contended(&fs_info->extent_commit_sem)) { + rwsem_is_contended(&fs_info->commit_root_sem)) { caching_ctl->progress = last; btrfs_release_path(path); - up_read(&fs_info->extent_commit_sem); + up_read(&fs_info->commit_root_sem); mutex_unlock(&caching_ctl->mutex); cond_resched(); goto again; @@ -513,7 +513,7 @@ next: err: btrfs_free_path(path); - up_read(&fs_info->extent_commit_sem); + up_read(&fs_info->commit_root_sem); free_excluded_extents(extent_root, block_group); @@ -633,10 +633,10 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, return 0; } - down_write(&fs_info->extent_commit_sem); + down_write(&fs_info->commit_root_sem); atomic_inc(&caching_ctl->count); list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups); - up_write(&fs_info->extent_commit_sem); + up_write(&fs_info->commit_root_sem); btrfs_get_block_group(cache); @@ -5471,7 +5471,7 @@ void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans, struct btrfs_block_group_cache *cache; struct btrfs_space_info *space_info; - down_write(&fs_info->extent_commit_sem); + down_write(&fs_info->commit_root_sem); list_for_each_entry_safe(caching_ctl, next, &fs_info->caching_block_groups, list) { @@ -5490,7 +5490,7 @@ void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans, else fs_info->pinned_extents = &fs_info->freed_extents[0]; - up_write(&fs_info->extent_commit_sem); + up_write(&fs_info->commit_root_sem); list_for_each_entry_rcu(space_info, &fs_info->space_info, list) percpu_counter_set(&space_info->total_bytes_pinned, 0); @@ -8256,14 +8256,14 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) struct btrfs_caching_control *caching_ctl; struct rb_node *n; - down_write(&info->extent_commit_sem); + down_write(&info->commit_root_sem); while (!list_empty(&info->caching_block_groups)) { caching_ctl = list_entry(info->caching_block_groups.next, struct btrfs_caching_control, list); list_del(&caching_ctl->list); put_caching_control(caching_ctl); } - up_write(&info->extent_commit_sem); + up_write(&info->commit_root_sem); spin_lock(&info->block_group_cache_lock); while ((n = rb_last(&info->block_group_cache_tree)) != NULL) { diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c index ab485e57b6fe..cc8ca193d830 100644 --- a/fs/btrfs/inode-map.c +++ b/fs/btrfs/inode-map.c @@ -55,7 +55,7 @@ static int caching_kthread(void *data) key.type = BTRFS_INODE_ITEM_KEY; again: /* need to make sure the commit_root doesn't disappear */ - mutex_lock(&root->fs_commit_mutex); + down_read(&fs_info->commit_root_sem); ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) @@ -88,7 +88,7 @@ again: btrfs_item_key_to_cpu(leaf, &key, 0); btrfs_release_path(path); root->cache_progress = last; - mutex_unlock(&root->fs_commit_mutex); + up_read(&fs_info->commit_root_sem); schedule_timeout(1); goto again; } else @@ -127,7 +127,7 @@ next: btrfs_unpin_free_ino(root); out: wake_up(&root->cache_wait); - mutex_unlock(&root->fs_commit_mutex); + up_read(&fs_info->commit_root_sem); btrfs_free_path(path); @@ -223,11 +223,11 @@ again: * or the caching work is done. */ - mutex_lock(&root->fs_commit_mutex); + down_write(&root->fs_info->commit_root_sem); spin_lock(&root->cache_lock); if (root->cached == BTRFS_CACHE_FINISHED) { spin_unlock(&root->cache_lock); - mutex_unlock(&root->fs_commit_mutex); + up_write(&root->fs_info->commit_root_sem); goto again; } spin_unlock(&root->cache_lock); @@ -240,7 +240,7 @@ again: else __btrfs_add_free_space(pinned, objectid, 1); - mutex_unlock(&root->fs_commit_mutex); + up_write(&root->fs_info->commit_root_sem); } } @@ -250,7 +250,7 @@ again: * and others will just be dropped, because the commit root we were * searching has changed. * - * Must be called with root->fs_commit_mutex held + * Must be called with root->fs_info->commit_root_sem held */ void btrfs_unpin_free_ino(struct btrfs_root *root) { diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index d00534b78382..6b5f13659317 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -1268,8 +1268,10 @@ static int find_extent_clone(struct send_ctx *sctx, } logical = disk_byte + btrfs_file_extent_offset(eb, fi); + down_read(&sctx->send_root->fs_info->commit_root_sem); ret = extent_from_logical(sctx->send_root->fs_info, disk_byte, tmp_path, &found_key, &flags); + up_read(&sctx->send_root->fs_info->commit_root_sem); btrfs_release_path(tmp_path); if (ret < 0) @@ -5367,57 +5369,21 @@ out: static int full_send_tree(struct send_ctx *sctx) { int ret; - struct btrfs_trans_handle *trans = NULL; struct btrfs_root *send_root = sctx->send_root; struct btrfs_key key; struct btrfs_key found_key; struct btrfs_path *path; struct extent_buffer *eb; int slot; - u64 start_ctransid; - u64 ctransid; path = alloc_path_for_send(); if (!path) return -ENOMEM; - spin_lock(&send_root->root_item_lock); - start_ctransid = btrfs_root_ctransid(&send_root->root_item); - spin_unlock(&send_root->root_item_lock); - key.objectid = BTRFS_FIRST_FREE_OBJECTID; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; -join_trans: - /* - * We need to make sure the transaction does not get committed - * while we do anything on commit roots. Join a transaction to prevent - * this. - */ - trans = btrfs_join_transaction(send_root); - if (IS_ERR(trans)) { - ret = PTR_ERR(trans); - trans = NULL; - goto out; - } - - /* - * Make sure the tree has not changed after re-joining. We detect this - * by comparing start_ctransid and ctransid. They should always match. - */ - spin_lock(&send_root->root_item_lock); - ctransid = btrfs_root_ctransid(&send_root->root_item); - spin_unlock(&send_root->root_item_lock); - - if (ctransid != start_ctransid) { - WARN(1, KERN_WARNING "BTRFS: the root that you're trying to " - "send was modified in between. This is " - "probably a bug.\n"); - ret = -EIO; - goto out; - } - ret = btrfs_search_slot_for_read(send_root, &key, path, 1, 0); if (ret < 0) goto out; @@ -5425,19 +5391,6 @@ join_trans: goto out_finish; while (1) { - /* - * When someone want to commit while we iterate, end the - * joined transaction and rejoin. - */ - if (btrfs_should_end_transaction(trans, send_root)) { - ret = btrfs_end_transaction(trans, send_root); - trans = NULL; - if (ret < 0) - goto out; - btrfs_release_path(path); - goto join_trans; - } - eb = path->nodes[0]; slot = path->slots[0]; btrfs_item_key_to_cpu(eb, &found_key, slot); @@ -5465,12 +5418,6 @@ out_finish: out: btrfs_free_path(path); - if (trans) { - if (!ret) - ret = btrfs_end_transaction(trans, send_root); - else - btrfs_end_transaction(trans, send_root); - } return ret; } diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 038177cfabbb..7579f6d0b854 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -75,10 +75,21 @@ void btrfs_put_transaction(struct btrfs_transaction *transaction) } } -static noinline void switch_commit_root(struct btrfs_root *root) +static noinline void switch_commit_roots(struct btrfs_transaction *trans, + struct btrfs_fs_info *fs_info) { - free_extent_buffer(root->commit_root); - root->commit_root = btrfs_root_node(root); + struct btrfs_root *root, *tmp; + + down_write(&fs_info->commit_root_sem); + list_for_each_entry_safe(root, tmp, &trans->switch_commits, + dirty_list) { + list_del_init(&root->dirty_list); + free_extent_buffer(root->commit_root); + root->commit_root = btrfs_root_node(root); + if (is_fstree(root->objectid)) + btrfs_unpin_free_ino(root); + } + up_write(&fs_info->commit_root_sem); } static inline void extwriter_counter_inc(struct btrfs_transaction *trans, @@ -208,6 +219,7 @@ loop: INIT_LIST_HEAD(&cur_trans->pending_snapshots); INIT_LIST_HEAD(&cur_trans->ordered_operations); INIT_LIST_HEAD(&cur_trans->pending_chunks); + INIT_LIST_HEAD(&cur_trans->switch_commits); list_add_tail(&cur_trans->list, &fs_info->trans_list); extent_io_tree_init(&cur_trans->dirty_pages, fs_info->btree_inode->i_mapping); @@ -920,9 +932,6 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans, return ret; } - if (root != root->fs_info->extent_root) - switch_commit_root(root); - return 0; } @@ -978,15 +987,16 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans, list_del_init(next); root = list_entry(next, struct btrfs_root, dirty_list); + if (root != fs_info->extent_root) + list_add_tail(&root->dirty_list, + &trans->transaction->switch_commits); ret = update_cowonly_root(trans, root); if (ret) return ret; } - down_write(&fs_info->extent_commit_sem); - switch_commit_root(fs_info->extent_root); - up_write(&fs_info->extent_commit_sem); - + list_add_tail(&fs_info->extent_root->dirty_list, + &trans->transaction->switch_commits); btrfs_after_dev_replace_commit(fs_info); return 0; @@ -1043,11 +1053,8 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans, smp_wmb(); if (root->commit_root != root->node) { - mutex_lock(&root->fs_commit_mutex); - switch_commit_root(root); - btrfs_unpin_free_ino(root); - mutex_unlock(&root->fs_commit_mutex); - + list_add_tail(&root->dirty_list, + &trans->transaction->switch_commits); btrfs_set_root_node(&root->root_item, root->node); } @@ -1858,11 +1865,15 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, btrfs_set_root_node(&root->fs_info->tree_root->root_item, root->fs_info->tree_root->node); - switch_commit_root(root->fs_info->tree_root); + list_add_tail(&root->fs_info->tree_root->dirty_list, + &cur_trans->switch_commits); btrfs_set_root_node(&root->fs_info->chunk_root->root_item, root->fs_info->chunk_root->node); - switch_commit_root(root->fs_info->chunk_root); + list_add_tail(&root->fs_info->chunk_root->dirty_list, + &cur_trans->switch_commits); + + switch_commit_roots(cur_trans, root->fs_info); assert_qgroups_uptodate(trans); update_super_roots(root); diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index 2bcba89d952e..b57b924e8e03 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -57,6 +57,7 @@ struct btrfs_transaction { struct list_head pending_snapshots; struct list_head ordered_operations; struct list_head pending_chunks; + struct list_head switch_commits; struct btrfs_delayed_ref_root delayed_refs; int aborted; }; -- cgit v1.2.3