[PATCH] tests/intel/xe_exec_reset: Replace hardcoded sleep with GT C6 polling

[Cavitt, Jonathan]

-----Original Message-----
From: igt-dev <igt-dev-bounces at lists.freedesktop.org> On Behalf Of Sk Anirban
Sent: Tuesday, August 5, 2025 10:45 AM
To: igt-dev at lists.freedesktop.org
Cc: Gupta, Anshuman <anshuman.gupta at intel.com>; Tauro, Riana <riana.tauro at intel.com>; Nilawar, Badal <badal.nilawar at intel.com>; Brost, Matthew <matthew.brost at intel.com>; Anirban, Sk <sk.anirban at intel.com>
Subject: [PATCH] tests/intel/xe_exec_reset: Replace hardcoded sleep with GT C6 polling
> 
> Replace insufficient sleep with extended GT C6 polling to ensure proper
> power state transitions to RC6 and prevent subsequent RC6 test failures.
> 
> Signed-off-by: Sk Anirban <sk.anirban at intel.com>
> ---
>  tests/intel/xe_exec_reset.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/intel/xe_exec_reset.c b/tests/intel/xe_exec_reset.c
> index 7ae53c679..0c9503c7e 100644
> --- a/tests/intel/xe_exec_reset.c
> +++ b/tests/intel/xe_exec_reset.c
> @@ -421,9 +421,12 @@ test_compute_mode(int fd, struct drm_xe_engine_class_instance *eci,
>  			for (i = 0; i < n_exec_queues; i++)
>  				xe_exec_queue_destroy(fd, exec_queues[i]);
>  		}
> +
> +		igt_assert_f(igt_wait(xe_gt_is_in_c6(fd, eci->gt_id), 1000, 10),
> +			     "GT %d failed to enter C6\n", eci->gt_id);

I talked about this privately, but this igt_assert_f is causing a failure on CI testing.
I wonder if this is because of missing test coverage, or if this is an unexpected failure?

But beyond that, I think this change also needs to be copied over to test_balancer()
(once it's not failing on CI, that is).
-Jonathan Cavitt

> +
>  		drm_close_driver(fd);
> -		/* FIXME: wait for idle */
> -		usleep(150000);
> +
>  		return;
>  	}
>  
> -- 
> 2.43.0
> 
>