Btrfs: fix error handling in btrfs_get_sb (0e78340f) · Commits · 戴 / test

If we failed to find the root subvol id, or the subvol=<name>, we would deactivate the locked super and close the devices. The problem is at this point we have gotten the SB all setup, which includes setting super_operations, so when we'd deactiveate the super, we'd do a close_ctree() which closes the devices, so we'd end up closing the devices twice. So if you do something like this mount /dev/sda1 /mnt/test1 mount /dev/sda1 /mnt/test2 -o subvol=xxx umount /mnt/test1 it would blow up (if subvol xxx doesn't exist). This patch fixes that problem. Thanks, Signed-off-by:

Josef Bacik <josef@redhat.com>

fs/btrfs/super.c

+3 −4

Original line number	Diff line number	Diff line
		@@ -629,7 +629,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
		if (IS_ERR(root)) {
		error = PTR_ERR(root);
		deactivate_locked_super(s);
		goto error;
		goto error_free_subvol_name;
		}
		/* if they gave us a subvolume name bind mount into that */
		if (strcmp(subvol_name, ".")) {
		@@ -643,14 +643,14 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
		deactivate_locked_super(s);
		error = PTR_ERR(new_root);
		dput(root);
		goto error_close_devices;
		goto error_free_subvol_name;
		}
		if (!new_root->d_inode) {
		dput(root);
		dput(new_root);
		deactivate_locked_super(s);
		error = -ENXIO;
		goto error_close_devices;
		goto error_free_subvol_name;
		}
		dput(root);
		root = new_root;
		@@ -668,7 +668,6 @@ error_close_devices:
		btrfs_close_devices(fs_devices);
		error_free_subvol_name:
		kfree(subvol_name);
		error:
		return error;
		}

Admin message