[igt-dev] [PATCH i-g-t] i915/gem_exec_gttfill: Run basic test on simulation

[Chris Wilson]

Quoting Kumar Valsan, Prathap (2019-03-22 13:58:52)
> On Fri, Mar 22, 2019 at 07:19:01AM +0000, Chris Wilson wrote:
> > Quoting Prathap Kumar Valsan (2019-03-21 23:06:13)
> > > Basic test don't take long to complete, so allow basic test to
> > > run on simulation.
> > > ---
> > >  tests/i915/gem_exec_gttfill.c | 13 ++++++++-----
> > >  1 file changed, 8 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/tests/i915/gem_exec_gttfill.c b/tests/i915/gem_exec_gttfill.c
> > > index efd612bb..92c3cf62 100644
> > > --- a/tests/i915/gem_exec_gttfill.c
> > > +++ b/tests/i915/gem_exec_gttfill.c
> > > @@ -144,7 +144,8 @@ static void fillgtt(int fd, unsigned ring, int timeout)
> > >         count = size / BATCH_SIZE + 1;
> > >         igt_debug("Using %'d batches to fill %'llu aperture on %d engines\n",
> > >                   count, (long long)size, nengine);
> > > -       intel_require_memory(count, BATCH_SIZE, CHECK_RAM);
> > > +       if (!igt_run_in_simulation())
> > > +               intel_require_memory(count, BATCH_SIZE, CHECK_RAM);
> > 
> > No. That is too ugly. You are basically saying that memory limits don't
> > matter for simulation -- unless the reader is aware of the convoluted
> > story in the setup and that we've previously marked as anything that
> > requires to check memory is not suitable for the slow simulator.
> > 
> Chris,
> gem_exec_gtfill at basic has been marked as a test that we need to run on
> simulation. https://jira.devtools.intel.com/browse/VLK-1649
> This is the intention behind trying to enable this test.

This particular test is a eviction test; we run it against mock HW
inside selftests/ (and as live test). That alone should be enough to
indicate that we don't care about the nuances of HW here, it is pure
software (give or take the usual layering over HW and any serendipity
therein.)
-Chris