[igt-dev] [i-g-t] Revert "syncobj_timeline: don't expect EINVAL for WAIT_UNSUBMITTED, | WAIT_AVAILABLE"

[Juha-Pekka Heikkila]

On 13.11.2023 12.30, Simon Ser wrote:
> On Monday, November 13th, 2023 at 11:06, Juha-Pekka Heikkila <juhapekka.heikkila at gmail.com> wrote:
> 
>> On 13.11.2023 12.02, Simon Ser wrote:
>>
>>> On Monday, November 13th, 2023 at 10:59, Juha-Pekka Heikkila juhapekka.heikkila at gmail.com wrote:
>>>
>>>> On 13.11.2023 11.55, Simon Ser wrote:
>>>>
>>>>> NACK. Please fix your docs/testplan/i915_tests.rst instead.
>>>>
>>>> Let's do fixes outside of master branch where it doesn't bother others.
>>>
>>> Sorry, but no. This commit fixes a bug already. The CI regression is
>>> completely unrelated, and Intel-specific, I really don't understand why
>>> it happens in the first place.
>>>
>>> Is it by design that actual bug fixes break Intel CI?
>>
>> As is this is blocking others totally unrelated to this. This patch
>> never passed ci in the first place hence never should've been merged.
>> Let's fix it somewhere else than on everyone's build machines independently.
> 
> If we want a quick fix, we can disable the broken Intel CI, instead of
> reverting a completely correct bugfix.

Simon, disabling CI doesn't universally fix the build for everyone. If 
you don't use testplan it's your story but it's not the story for all 
other igt users. Passing ci testing is the requirement for merging, 
that's why ci is running on patch sets. This reverted patch didn't even 
build in ci and after you merged it others started to see what ci 
already had reported about this patch.

/Juha-Pekka