[igt-dev] [PATCH i-g-t] perf_pmu: Dump i915_runtime_pm_status

Dixit, Ashutosh ashutosh.dixit at intel.com
Thu Mar 2 23:09:48 UTC 2023


On Wed, 01 Mar 2023 04:12:07 -0800, Anshuman Gupta wrote:
>
> Dump i915_pm_runtime_status in case of i915 device failed to runtime
> suspend, it will dump the rpm wakeref tracking information which
> will useful to debug the runtime pm issues.
>
> Signed-off-by: Anshuman Gupta <anshuman.gupta at intel.com>
> ---
>  tests/i915/perf_pmu.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/tests/i915/perf_pmu.c b/tests/i915/perf_pmu.c
> index c1779fc9d0..567b2fa5cf 100644
> --- a/tests/i915/perf_pmu.c
> +++ b/tests/i915/perf_pmu.c
> @@ -1709,6 +1709,7 @@ test_rc6(int gem_fd, unsigned int flags)
>	int64_t duration_ns = 2e9;
>	uint64_t idle, busy, prev, ts[2];
>	unsigned long slept;
> +	bool suspended;
>	int fd, fw;
>
>	gem_quiescent_gpu(gem_fd);
> @@ -1727,7 +1728,12 @@ test_rc6(int gem_fd, unsigned int flags)
>		drmModeFreeResources(res);
>
>		igt_require(igt_setup_runtime_pm(gem_fd));
> -		igt_require(igt_wait_for_pm_status(IGT_RUNTIME_PM_STATUS_SUSPENDED));
> +		suspended = igt_wait_for_pm_status(IGT_RUNTIME_PM_STATUS_SUSPENDED);
> +
> +		if (!suspended)
> +			__igt_debugfs_dump(gem_fd, "i915_runtime_pm_status", IGT_LOG_INFO);
> +
> +		igt_require(suspended);

Can we use wait_for_suspended() here instead? OK otherwise too:

Reviewed-by: Ashutosh Dixit <ashutosh.dixit at intel.com>

>
>		/*
>		 * Sleep for a bit to see if once woken up estimated RC6 hasn't
> --
> 2.25.1
>


More information about the igt-dev mailing list