[igt-dev] [PATCH i-g-t v3 1/1] i915_pm_rpm: gem-execbuf-stress-extra-wait faster

[Caz Yokoyama]

Chris,

As I sai on Tue, 12 Mar 2019 09:11:14 -0700,
My intention is to remove gem-execbuf-stress-extra-wait subtest
because it does same as gem-execbuf-stress except for meaningless (10 x
5 sec) delay.

I believe the delay provides no extra coverage. It may be added as same
as other subtest such as modeset_subtest(). WAIT_EXTRA may provide
extra coverage for modeset_subtest() while it does not for gem-execbuf-
stress.
-caz

On Wed, 2019-03-13 at 21:12 +0000, Chris Wilson wrote:
> Quoting Caz Yokoyama (2019-03-14 00:46:08)
> > Replace (5sec * 10) delay by forcing to suspended state.
> > While this patch expects a corresponding patch in drm-tip,
> > the test runs successful without the corresponding patch.
> > Its runtime is 2-3 sec on my local machine, nuc.
> 
> WAIT_EXTRA is always being used in conjunction with WAIT_STATUS, so
> it
> simply is checking that things do not break with an extra sleep after
> suspending.
> 
> Removing that delay nerfs the test, you may as well simply not run
> the
> test.
> 
> Instead of removing the delay we need to be replacing the delay with
> something more likely to trigger a bug, or just remove the test as we
> believe it provides no extra coverage.
> -Chris