Re: [igt-dev] ✗ Fi.CI.IGT: failure for PMU: multi-tile support (rev4)

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

On Thu, 18 May 2023 23:44:03 -0700, Umesh Nerlige Ramappa wrote:
>

Hi Umesh,

>
> One of the SKIPs below is not expected,
>
> On Fri, May 19, 2023 at 12:13:01AM +0000, Patchwork wrote:
> >   Patch Details
> >
> >   Series:  PMU: multi-tile support (rev4)
> >   URL:     [1]https://patchwork.freedesktop.org/series/117406/
> >   State:   failure
> >   Details: [2]https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_8990/index.html
> >
> >         CI Bug Log - changes from CI_DRM_13160_full -> IGTPW_8990_full
> >
> > Summary
> >
> >   FAILURE
> >
> >   Serious unknown changes coming with IGTPW_8990_full absolutely need to be
> >   verified manually.
> >
> >   If you think the reported changes have nothing to do with the changes
> >   introduced in IGTPW_8990_full, please notify your bug team to allow them
> >   to document this new failure mode, which will reduce false positives in
> >   CI.
> >
> >   External URL:
> >   https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_8990/index.html
> >
> > Participating hosts (7 -> 7)
> >
> >   No changes in participating hosts
> >
> > Possible new issues
> >
> >   Here are the unknown changes that may have been introduced in
> >   IGTPW_8990_full:
> >
> >  IGT changes
> >
> >    Possible regressions
> >
> >     * igt@gem_ctx_isolation@preservation-s3@vcs0:
> >
> >          * shard-apl: [3]PASS -> [4]ABORT
> >
> >     * {igt@perf_pmu@frequency@gt0} (NEW):
> >
> >          * {shard-rkl}: NOTRUN -> [5]SKIP +1 similar issue
>

Yup, because gem_list_engines() had a bug. I have sent a patch:

https://patchwork.freedesktop.org/series/118001/

Failure can be seen here (for reviewers for the above patch):

https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_8990/shard-rkl-1/igt@perf_pmu@frequency@gt0.html

> With this IGT series - https://patchwork.freedesktop.org/series/117406/, we
> do not have igt@perf_pmu@frequency test anymore. It is replaced by
> igt@perf_pmu@frequency@gt0 for all single tile platforms. For non-MTL
> platforms, gem_engine_to_gt_map() should just return 0.
>
> >
> >     * {igt@perf_pmu@rc6-all-gts} (NEW):
> >
> >          * {shard-dg1}: NOTRUN -> [6]SKIP +1 similar issue
> >
> >          * {shard-tglu}: NOTRUN -> [7]SKIP +2 similar issues
>
> These are expected to skip.

Also, Umesh not sure if we need to run premerge CI on your latest rev:

https://patchwork.freedesktop.org/series/117913/

It seems to have got stuck and maybe you need to say 'test again'.

Thanks.
--
Ashutosh