On Mon, Aug 19, 2024 at 10:54:59AM +0200, Christian Brauner wrote:
> On Sat, Aug 17, 2024 at 08:43:41PM GMT, Dan Carpenter wrote:
> > Hello Christian Brauner,
> >
> > Commit 71aacb4c8c3d ("fs: Allow statmount() in foreign mount
>
> Hey Dan,
>
> Thanks for the report.
>
> > Should copy_mnt_id_req() ensure that kreq->mnt_ns_id is non-zero the same as
> > ->mnt_id?
>
> No, mnt_ns_id can legitimately be zero.
>
> >
> > 5329 if (ret)
> > 5330 return ret;
> > 5331
> > 5332 ns = grab_requested_mnt_ns(&kreq);
> > 5333 if (!ns)
> > 5334 return -ENOENT;
> >
> > The grab_requested_mnt_ns() function returns a mix of error pointers and NULL.
>
> I'm not sure how you got that idea. If no mnt_ns_id is specified then
> current's mnt_ns will be returned which is always valid. And if
> mnt_ns_id is specified it'll return a valid mnt_ns or NULL.
Ah. You're right. When ->spare is zero it can't return error pointers. My
bad.
regards,
dan carpenter
