[igt-dev] [PATCH i-g-t v2] Follow kernel's resource streamer removal

[Lucas De Marchi]

On Thu, Jul 19, 2018 at 08:26:11PM +0100, Chris Wilson wrote:
> Quoting Lucas De Marchi (2018-07-19 20:16:14)
> > Resource streamer is being removed from all GENs, so just test it
> > returns EINVAL for all GENs when a batch is submitted with
> > I915_EXEC_RESOURCE_STREAMER keeping one test per ring.
> > 
> > v2: let one test per ring rather than just one test
> > 
> > Signed-off-by: Lucas De Marchi <lucas.demarchi at intel.com>
> > ---
> > 
> > This requires the patch for the kernel removing RS from all gens.
> > 
> >  tests/gem_exec_params.c | 6 +-----
> >  1 file changed, 1 insertion(+), 5 deletions(-)
> > 
> > diff --git a/tests/gem_exec_params.c b/tests/gem_exec_params.c
> > index 04c21c05..f822297f 100644
> > --- a/tests/gem_exec_params.c
> > +++ b/tests/gem_exec_params.c
> > @@ -375,25 +375,21 @@ igt_main
> >         }
> >  
> >         igt_subtest("rs-invalid-on-bsd-ring") {
> > -               igt_require(IS_HASWELL(devid) || intel_gen(devid) >= 8);
> 
> Now fails on old kernels...
> 
> igt_subtest("rs-invalid") {
> 	int has_rs = gem_param(HAS_RESOURCE_STREAMER);
> 	for_each_engine(fd, ring) {
> 		int expect = -EINVAL;
> 		if (has_rs && (ring == 0 || ring == I915_EXEC_RENDER))
> 			expect = 0;

if we are removing resource streamer from the kernel, shouldn't we
expect EINVAL on all rings now?

Lucas De Marchi

> 
> 		execbuf.flags = engine;
> 		igt_assert_eq(__gem_execbuf(fd, &execbuf), expect);
> 	}
> }
> 
> Now if we reinterpret the param as the uabi-class bitmask...
> -Chris
> _______________________________________________
> igt-dev mailing list
> igt-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/igt-dev