Re: [PATCH i-g-t] i915/drm_fdinfo: Quiesce GPU prior to running some tests

[Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>]

On Wed, Oct 02, 2024 at 04:19:31PM -0700, Dixit, Ashutosh wrote:
>On Wed, 02 Oct 2024 14:20:49 -0700, Umesh Nerlige Ramappa wrote:
>>
>> Sometimes render is still running a remnant of a prior workload when we
>> expect it to be idle. This throws some tests off. Ensure that these
>> tests wait for a quiescent state prior to checking if render is idle.
>>
>> Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/12179
>> Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
>
>Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>

Thanks

>
>Not related to this patch but little bit confusing, since I thought
>drm_fdinfo is driver/vendor agnostic. But this file is i915 specific?
>Should probably be named i915_fdinfo.c? Or if it's in intel/ directory
>should at least probably work for both i195 and xe? Anyway.

It was supposed to be driver agnostic, but interface changed for xe 
since we added 2 counters per class to indicate utilization (also units 
changed), so there is xe_drm_fdinfo.c in the tests/intel folder.

The driver/vendor-agnostic part of this feature is included as a tool 
- tools/gputop.c which can be used by other vendors too.

Regards,
Umesh