[Intel-gfx] [PATCH i-g-t] lib/igt_pm: Restore runtime pm state on test exit

Tvrtko Ursulin tvrtko.ursulin at linux.intel.com
Wed Feb 28 14:33:18 UTC 2018


On 28/02/2018 12:27, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2018-02-28 11:38:01)
>>
>> On 28/02/2018 11:12, Chris Wilson wrote:
>>> Quoting Tvrtko Ursulin (2018-02-28 11:08:29)
>>>> From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>>>
>>>> Some tests (the ones which call igt_setup_runtime_pm and
>>>> igt_pm_enable_audio_runtime_pm) change default system configuration and
>>>> never restore it.
>>>>
>>>> The configured runtime suspend is aggressive and may influence behaviour
>>>> of subsequent tests, so it is better to restore to previous values on test
>>>> exit.
>>>>
>>>> This way system behaviour, with regards to a random sequence of executed
>>>> tests, will be more consistent from one run to another.
>>>
>>> Otoh, if behaviour changes in subsequent tests, we likely have a lack of
>>> testing :(
>>
>> Yeah, and I am not saying it does - haven't spotted anything like that,
>> just that it leaves the auto-suspend with zero delay afterwards,
>> compared to otherwise default 10s.
>>
>> Maybe it is good for coverage, even with the downside of randomness
>> considering shard runs, or maybe it needs to be more explicit.
> 
> I actually thought (or at least picked up the idea) we enabled 0
> autosuspend delay throughout igt. And that it's only because most of time
> we have a display connected that prevents rpm madness.

Yes fair point, the effect of this patch is only visible on esoteric 
headless setups like mine. :)

Anyway, unless I am missing something, I think it is conceptually 
correct to restore, as long as the zero autosuspend delay is not a 
global IGT setup but a helper called by a few tests.

Regards,

Tvrtko


More information about the Intel-gfx mailing list