Re: [PATCH] vfs: return EINVAL instead of ENOENT when missing source

[Murphy Zhou <jencce.kernel@gmail.com>]

Hi AL, what do you think, ;)

On Mon, Mar 18, 2019 at 1:39 PM Murphy Zhou <jencce.kernel@gmail.com> wrote:
>
> mount(2) with a NULL source device would return ENOENT instead of EINVAL
> after this commit:
>
> commit f3a09c92018a91ad0981146a4ac59414f814d801
> Author: Al Viro <viro@zeniv.linux.org.uk>
> Date:   Sun Dec 23 18:55:56 2018 -0500
>
>     introduce fs_context methods
>
> Change the return value to be compatible with the old behaviour.
>
> This was caught by LTP mount02[1]. This testcase is calling mount(2) with a
> NULL device name and expecting EINVAL to PASS but now we are getting ENOENT.
>
> [1] https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/syscalls/mount/mount02.c
>
> Signed-off-by: Murphy Zhou <jencce.kernel@gmail.com>
> ---
>  fs/super.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/super.c b/fs/super.c
> index 583a0124bc39..48e51f13a4ba 100644
> --- a/fs/super.c
> +++ b/fs/super.c
> @@ -1469,7 +1469,7 @@ int vfs_get_tree(struct fs_context *fc)
>
>         if (fc->fs_type->fs_flags & FS_REQUIRES_DEV && !fc->source) {
>                 errorf(fc, "Filesystem requires source device");
> -               return -ENOENT;
> +               return -EINVAL;
>         }
>
>         if (fc->root)
> --
> 2.21.0
>