btrfs: fix deadlock when writing out space cache

[ Upstream commit b77000ed ]

If we fail to prepare our pages for whatever reason (out of memory in
our case) we need to make sure to drop the block_group->data_rwsem,
otherwise hilarity ensues.
Signed-off-by: Josef Bacik <jbacik@fb.com>
Reviewed-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add label and use existing unlocking code ]
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs/btrfs/free-space-cache.c

@@ -1258,7 +1258,7 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	/* Lock all pages first so we can lock the extent safely. */
 	ret = io_ctl_prepare_pages(io_ctl, inode, 0);
 	if (ret)
-		goto out;
+		goto out_unlock;
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
 			 0, &cached_state);
@@ -1351,6 +1351,7 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 out_nospc:
 	cleanup_write_cache_enospc(inode, io_ctl, &cached_state, &bitmap_list);
 
+out_unlock:
 	if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
 		up_write(&block_group->data_rwsem);