On Mon, Jun 28, 2021 at 6:42 AM Dave Chinner <david@xxxxxxxxxxxxx> wrote:
>
> On Fri, Jun 25, 2021 at 08:40:33AM -0400, Wang Shilong wrote:
> > From: Wang Shilong <wshilong@xxxxxxx>
> >
> > U32_MAX is reserved for special purpose,
> > qid_has_mapping() will return false if projid is
> > 4294967295, dqget() will return NULL for it.
> >
> > So U32_MAX is unsupported Project ID, fix to forbid
> > it.
>
> Actually, it's INVALID_PROJID, not U32_MAX, and we already have a
> check function for that:
>
> static inline bool projid_valid(kprojid_t projid)
> {
> return !projid_eq(projid, INVALID_PROJID);
> }
>
I was not aware of this, thanks for pointing it out.
> > Signed-off-by: Wang Shilong <wshilong@xxxxxxx>
> > ---
> > fs/ext4/ioctl.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
> > index 31627f7dc5cd..f3a8d962c291 100644
> > --- a/fs/ext4/ioctl.c
> > +++ b/fs/ext4/ioctl.c
> > @@ -744,6 +744,9 @@ int ext4_fileattr_set(struct user_namespace *mnt_userns,
> > u32 flags = fa->flags;
> > int err = -EOPNOTSUPP;
> >
> > + if (fa->fsx_projid >= U32_MAX)
> > + return -EINVAL;
> > +
>
> This should actually be calling qid_valid() or projid_valid(),
> and it should be in generic code because multiple filesystems
> support project quotas. i.e this should be checked in
> fileattr_set_prepare(), not in ext4 specific code.
I tried to fix ext4/f2fs, i am not sure about XFS, it looks to me XFS
implemented quota mostly by itself.
Anyway, let me fix this in generic code.
>
> Cheers,
>
> Dave.
> --
> Dave Chinner
> david@xxxxxxxxxxxxx
