Re: [PATCH v3] default_range glblub implementation

[Paul Moore <paul@paul-moore.com>]

On Fri, Aug 30, 2019 at 11:47 AM Joshua Brindle
<joshua.brindle@crunchydata.com> wrote:
> <snip>
> > > +
> > > +static inline int mls_context_glblub(struct context *dst, struct context *c1, struct context *c2)
> > > +{
> > > +     struct mls_range *dr = &dst->range, *r1 = &c1->range, *r2 = &c2->range;
> > > +     int rc = 0;
> > > +
> > > +     if (r1->level[1].sens < r2->level[0].sens || r2->level[1].sens < r1->level[0].sens)
> > > +             /* These ranges have no common sensitivities */
> > > +             return -1;
> >
> > I mentioned this on the prior patch, but perhaps you didn't read past
> > the first comment: I think you need to return -EINVAL or some other
> > -errno value here since this will get propagated up to the userspace
> > caller as an errno if I am not mistaken.
>
> Yes, you are right. Sorry about that. I'll fix this on the next spin
> if everyone is okay with everything else.  I think -EINVAL is
> appropriate unless anyone else objects.

Other than this, v3 seems reasonable to me, but it would be good to
see a thumbs-up on the latest userspace patches before we merge this.
Changes to the kernel/user interface are
<haunting_voice>FOREVER</haunting_voice>.

-- 
paul moore
www.paul-moore.com