[igt-dev] [PATCH i-g-t] tests/kms_hdr: Fix failure to read output_bpc on AMD hardware

[Kazlauskas, Nicholas]

On 2021-10-07 10:56 a.m., Harry Wentland wrote:
> 
> 
> On 2021-10-07 10:54, Kazlauskas, Nicholas wrote:
>> [AMD Official Use Only]
>>
>>
>> I don't have the commits in front of me, but there was an issue with the port from amd_hdr to kms_hdr where the test would do a TEST_ONLY commit that change the output bpc to 10 but then the debugfs check /after /would try and verify that the output bpc was 10 - which doesn't work if the output bpc was previously 8 because the commit wasn't actually applied.
>>
> 
> https://patchwork.freedesktop.org/patch/456663/?series=95313&rev=1
> 
> Stylon, is that what is happening?
> 
> If so, please make sure the commit description reflects that. The existing
> description doesn't explain why switching from TEST_ONLY to ALLOW_MODESET
> fixes things.
> 
> Harry

Thanks for linking it - it looks like it's actually the reverse, where 
it tries to change bpc to 8 but the connector may not currently be in 8 
nor is there a requirement for it to be before the test starts.

Userspace might have already been running with an uncapped max bpc 
before the test starts.

Regards,
Nicholas Kazlauskas

> 
> 
>> Regards,
>> Nicholas Kazlauskas
>> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>> *From:* Simon Ser <contact at emersion.fr>
>> *Sent:* Thursday, October 7, 2021 10:49 AM
>> *To:* Siqueira, Rodrigo <Rodrigo.Siqueira at amd.com>
>> *Cc:* Saarinen, Jani <jani.saarinen at intel.com>; Vudum, Lakshminarayana <lakshminarayana.vudum at intel.com>; Latvala, Petri <petri.latvala at intel.com>; Wang, Chao-kai (Stylon) <Stylon.Wang at amd.com>; igt-dev at lists.freedesktop.org <igt-dev at lists.freedesktop.org>; Choi, Nicholas <Nicholas.Choi at amd.com>; Mark Yacoub <markyacoub at google.com>; Wentland, Harry <Harry.Wentland at amd.com>; Kazlauskas, Nicholas <Nicholas.Kazlauskas at amd.com>
>> *Subject:* Re: [PATCH i-g-t] tests/kms_hdr: Fix failure to read output_bpc on AMD hardware
>>   
>> I don't think removing TEST_ONLY is correct? bafd90661c26 ("tests/kms_hdr: Add
>> bpc switch subtests") reads:
>>
>>      v4: -Plane size back to CRTC size, regression observed in hsw again
>>           "Plane must cover entire CRTC" because of plane size restriction
>>           to 512x512. Petri/Ville suggested to use TEST_ONLY in try_commit()
>>           to probe whether we can use a smaller than fullscreen plane.
>>
>> cc Nicholas
>