btrfs: drop root refs properly when orphan cleanup fails

When we mount the file system we do something like this:

	while (1) {
		lookup fs roots;

		for (i = 0; i < num_roots; i++) {
			ret = btrfs_orphan_cleanup(roots[i]);
			if (ret)
				break;
			btrfs_put_root(roots[i]);
		}
	}

	for (; i < num_roots; i++)
		btrfs_put_root(roots[i]);

As you can see if we break in that inner loop we just go back to the
outer loop and lose the fact that we have to drop references on the
remaining roots we looked up.  Fix this by making an out label and
jumping to that on error so we don't leak a reference to the roots we
looked up.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

fs/btrfs/disk-io.c

@@ -4404,12 +4404,12 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
 			root_objectid = gang[i]->root_key.objectid;
 			err = btrfs_orphan_cleanup(gang[i]);
 			if (err)
-				break;
+				goto out;
 			btrfs_put_root(gang[i]);
 		}
 		root_objectid++;
 	}
-
+out:
 	/* release the uncleaned roots due to error */
 	for (; i < ret; i++) {
 		if (gang[i])