Re: [igt-dev] Fwd: ✗ Fi.CI.IGT: failure for lib/igt_kms: Set pipe->plane_primary to driver-assigned primary plane when there are multiple possible primary planes (rev6)

["Dixit, Ashutosh" <ashutosh.dixit@intel.com>]

On Sat, 04 Jun 2022 23:19:51 -0700, Abhinav Kumar wrote:
>
> Hi Ashutosh
>
> On 6/4/2022 7:26 PM, Dixit, Ashutosh wrote:
> > On Thu, 19 May 2022 13:44:35 -0700, Jessica Zhang wrote:
> >>
> >>
> >>
> >> On 5/19/2022 1:34 PM, Jessica Zhang wrote:
> >>> Hi Lakshmi,
> >>>
> >>> I looked into this regression, but it doesn't seem to be related to my
> >>> patch. Can you help take a look at it?
> >
> > It seems the failures were related after all and CI correctly caught
> > them. They should be properly investigated by both the author an committer
> > before merging a patch. This has resulted in multiple people wasting a
> > fair amount of their time and multiple bugs being filed as a result :/
>
> Please correct if we are missing something.
>
> That time, CI had reported a different error. Please refer to the following
> CI report of the series:
>
> https://patchwork.freedesktop.org/series/103897/#rev5
>
> That we still think is unrelated to the change. Please comment if you think
> otherwise.
>
> The issues reported by all the following bugs in Jani's email have a very
> clear and consistent signature of the SEG fault:
>
> https://gitlab.freedesktop.org/drm/intel/issues/6133
> https://gitlab.freedesktop.org/drm/intel/-/issues/6134
> https://gitlab.freedesktop.org/drm/intel/-/issues/6135
> https://gitlab.freedesktop.org/drm/intel/-/issues/6136
> https://gitlab.freedesktop.org/drm/intel/-/issues/6137
>
>
> Stack trace:
>  #0 [fatal_sig_handler+0xd6]
>  #1 [killpg+0x40]
>  #2 [igt_display_require+0x260]
>  #3 [__igt_unique____real_main318+0x3c8]
>  #4 [main+0x34]
>  #5 [__libc_start_main+0xf3]
>  #6 [_start+0x2a]
>
> If this signature was seen in the CI runs earlier of our patch series, this
> could have been easily fixed by us as it was not really very hard to fix
> once we saw this signature. So i think the CI run on our change always gave
> an unrelated error which is different from the other issues seen here.

Hi Abhinav,

You appear to be correct. I was going by the clear PASS -> FAIL signatures
in CI but those failures indeed appear to be unrelated and it is the
igt_display_require() segfault above which is the signature for this
particular failure. That CI did not flag an obvious failure pre-merge is a
problem in itself, but in any case I cannot accuse you of not investigating
CI results so sorry about that.

Ashutosh