In qgroup_rescan_leaf a copy is made of the target leaf by calling btrfs_clone_extent_buffer. The latter allocates a new buffer and attaches a new set of pages and copies the content of the source buffer. The new scratch buffer is only used to iterate it's items, it's not published anywhere and cannot be accessed by a third party. Hence, it's not necessary to perform any locking on it whatsoever. Furthermore, remove the extra extent_buffer_get call since the new buffer is always allocated with a reference count of 1 which is sufficient here. No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]> --- fs/btrfs/qgroup.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 4353bb69bb86..4d3efc8f391e 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -2689,9 +2689,6 @@ static int qgroup_rescan_leaf(struct btrfs_trans_handle *trans, mutex_unlock(&fs_info->qgroup_rescan_lock); goto out; } - extent_buffer_get(scratch_leaf); - btrfs_tree_read_lock(scratch_leaf); - btrfs_set_lock_blocking_rw(scratch_leaf, BTRFS_READ_LOCK); slot = path->slots[0]; btrfs_release_path(path); mutex_unlock(&fs_info->qgroup_rescan_lock); @@ -2717,10 +2714,8 @@ static int qgroup_rescan_leaf(struct btrfs_trans_handle *trans, goto out; } out: - if (scratch_leaf) { - btrfs_tree_read_unlock_blocking(scratch_leaf); + if (scratch_leaf) free_extent_buffer(scratch_leaf); - } if (done && !ret) { ret = 1; -- 2.7.4
