[igt-dev] [PATCH i-g-t v7] i915/gem_mocs_settings: Add mocs table for icelake

[Kumar Valsan, Prathap]

On Mon, Mar 18, 2019 at 01:06:13PM +0000, Chris Wilson wrote:
> Quoting Kumar Valsan, Prathap (2019-03-18 13:10:43)
> > On Fri, Mar 15, 2019 at 03:15:55PM -0400, Prathap Kumar Valsan wrote:
> > > +static void isolation_test(int fd, unsigned engine)
> > > +{
> > > +     uint32_t ctx[2] = { gem_context_create(fd), gem_context_create(fd) };
> > > +
> > > +     /* Any writes by one normal client should not affect a second client */
> > > +     write_dirty_mocs(fd, engine, ctx[0], false);
> > > +     check_mocs_values(fd, engine, ctx[1], false);
> > > +
> > > +     for (int i = 0; i < ARRAY_SIZE(ctx); i++)
> > > +             gem_context_destroy(fd, ctx[i]);
> > > +}
> > > +
> > >  igt_main
> > >  {
> > >       const struct intel_execution_engine *e;
> > > @@ -490,6 +559,14 @@ igt_main
> > >                               }
> > >                       }
> > >               }
> > > +
> > > +             igt_subtest_f("mocs-isolation-%s",
> > > +                             e->name) {
> > > +                     gem_require_contexts(fd);
> > > +
> > > +                     isolation_test(fd, e->exec_id | e->flags);
> > > +             }
> > > +
> > >       }
> > Chris,
> > Added isolation_test as a subtest. Can i get a r-b if this looks good to
> > you.
> 
> Yup, that looks like what I expect we need to verify the ABI that mocs
> values are independent without overburdening ourselves with actual HW
> implementation.
> 
> Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
> -Chris

Thanks for the reviews.

-Prathap