Btrfs: make btrfs_search_forward return with nodes unlocked

None of the uses of btrfs_search_forward() need to have the path
nodes (level >= 1) read locked, only the leaf needs to be locked
while the caller processes it. Therefore make it return a path
with all nodes unlocked, except for the leaf.

This change is motivated by the observation that during a file
fsync we repeatdly call btrfs_search_forward() and process the
returned leaf while upper nodes of the returned path (level >= 1)
are read locked, which unnecessarily blocks other tasks that want
to write to the same fs/subvol btree.
Therefore instead of modifying the fsync code to unlock all nodes
with level >= 1 immediately after calling btrfs_search_forward(),
change btrfs_search_forward() to do it, so that it benefits all
callers.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>

5 files changed, 7 insertions, 15 deletions

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 783ea3bac7dc..39021bf2df9a 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -5144,8 +5144,9 @@ int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
 	u32 nritems;
 	int level;
 	int ret = 1;
+	int keep_locks = path->keep_locks;
 
-	WARN_ON(!path->keep_locks);
+	path->keep_locks = 1;
 again:
 	cur = btrfs_read_lock_root_node(root);
 	level = btrfs_header_level(cur);
@@ -5209,7 +5210,6 @@ find_next_key:
 		path->slots[level] = slot;
 		if (level == path->lowest_level) {
 			ret = 0;
-			unlock_up(path, level, 1, 0, NULL);
 			goto out;
 		}
 		btrfs_set_path_blocking(path);
@@ -5224,9 +5224,12 @@ find_next_key:
 		btrfs_clear_path_blocking(path, NULL, 0);
 	}
 out:
-	if (ret == 0)
+	path->keep_locks = keep_locks;
+	if (ret == 0) {
+		btrfs_unlock_up_safe(path, path->lowest_level + 1);
+		btrfs_set_path_blocking(path);
 		memcpy(min_key, &found_key, sizeof(found_key));
-	btrfs_set_path_blocking(path);
+	}
 	return ret;
 }
 
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 4492e017bdb2..85ca3cce8554 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -936,12 +936,9 @@ static int find_new_extents(struct btrfs_root *root,
 	min_key.offset = *off;
 
 	while (1) {
-		path->keep_locks = 1;
 		ret = btrfs_search_forward(root, &min_key, path, newer_than);
 		if (ret != 0)
 			goto none;
-		path->keep_locks = 0;
-		btrfs_unlock_up_safe(path, 1);
 process_slot:
 		if (min_key.objectid != ino)
 			goto none;
@@ -2088,8 +2085,6 @@ static noinline int search_ioctl(struct inode *inode,
 	key.type = sk->min_type;
 	key.offset = sk->min_offset;
 
-	path->keep_locks = 1;
-
 	while (1) {
 		ret = btrfs_search_forward(root, &key, path, sk->min_transid);
 		if (ret != 0) {
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 7b6d1428f033..82db14f5cf87 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -2983,8 +2983,6 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans,
 	min_key.type = key_type;
 	min_key.offset = min_offset;
 
-	path->keep_locks = 1;
-
 	ret = btrfs_search_forward(root, &min_key, path, trans->transid);
 
 	/*
@@ -3964,7 +3962,6 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		err = ret;
 		goto out_unlock;
 	}
-	path->keep_locks = 1;
 
 	while (1) {
 		ins_nr = 0;
diff --git a/fs/btrfs/uuid-tree.c b/fs/btrfs/uuid-tree.c
index f6a4c03ee7d8..778282944530 100644
--- a/fs/btrfs/uuid-tree.c
+++ b/fs/btrfs/uuid-tree.c
@@ -279,7 +279,6 @@ int btrfs_uuid_tree_iterate(struct btrfs_fs_info *fs_info,
 	key.offset = 0;
 
 again_search_slot:
-	path->keep_locks = 1;
 	ret = btrfs_search_forward(root, &key, path, 0);
 	if (ret) {
 		if (ret > 0)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 46cd29ad3156..ff8386eb3367 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3625,8 +3625,6 @@ static int btrfs_uuid_scan_kthread(void *data)
 	max_key.type = BTRFS_ROOT_ITEM_KEY;
 	max_key.offset = (u64)-1;
 
-	path->keep_locks = 1;
-
 	while (1) {
 		ret = btrfs_search_forward(root, &key, path, 0);
 		if (ret) {