Bug 110728

The CI Bug Log issue associated to this bug has been updated.

### New filters associated

* ICL: igt@perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull)
  - https://intel-gfx-ci.01.org/tree/drm-tip/Trybot_4275/shard-iclb2/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6087/shard-iclb3/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Trybot_4298/shard-iclb2/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/IGT_5000/shard-iclb7/igt@perf@polling.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/IGT_5000/shard-iclb8/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6112/shard-iclb7/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13055/shard-iclb8/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6114/shard-iclb4/igt@perf@polling.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6114/shard-iclb7/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13056/shard-iclb7/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13058/shard-iclb2/igt@perf@polling.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13058/shard-iclb3/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13059/shard-iclb2/igt@perf@blocking.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13059/shard-iclb7/igt@perf@polling.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13062/shard-iclb2/igt@perf@polling.html
  - https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_13064/shard-iclb6/igt@perf@blocking.html

A CI Bug Log filter associated to this bug has been updated:

{- ICL: igt@perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull) -}
{+ SKL ICL: igt@perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull) +}

New failures caught by the filter:

  * https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6176/shard-skl6/igt@perf@blocking.html
  * https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6179/shard-skl1/igt@perf@polling.html

In [1], 170ms (±10ms) are spent blocking in the kernel. This is more than 1% of the test duration (10s = 10000ms, time spent in kernel is 1.7%). In other logs, similar issues can be observed.

This means that we're burning too many CPU cycles in the kernel when reading from Observation Architecture.

User impact: some performance issues when measuring performance.

I suspect these performance issues only happen when actively using OA, so most users shouldn't be affected, but I'm not familiar with OA, so don't quote me on this.

[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6176/shard-skl6/igt@perf@blocking.html

Reproduction rate: 3%

Only ICL and SKL is affected. The bug is hit daily.

Posted a patch to improve logging: https://patchwork.freedesktop.org/patch/313695/

(In reply to emersion from comment #5)
> Posted a patch to improve logging:
> https://patchwork.freedesktop.org/patch/313695/

Erf, please disregard that, it's unrelated.

Issue is still hit daily (every 5 hours or so). However it seems like ICL is not hit anymore: last ICL hit was 1 month, 3 weeks ago. So now this seems to be SKL-specific.

A CI Bug Log filter associated to this bug has been updated:

{- SKL ICL: igt@perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull) -}
{+ SKL KBL ICL: igt@perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull) +}

New failures caught by the filter:

  * https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_354/fi-kbl-soraka/igt@perf@polling.html

The test itself runs for 10 seconds. In this time, I see these extra logs in the failed dmesg compared to the succeeding cases.

<7> [710.600658] [drm:intel_power_well_disable [i915]] disabling DC off
<7> [710.601387] [drm:skl_enable_dc6 [i915]] Enabling DC6
<7> [710.601896] [drm:gen9_set_dc_state [i915]] Setting DC state from 00 to 02

Issue is still hit daily. Last report was 1 week 3 days ago. Importance stays high.

Impact: Needs further root cause before impact can be assessed.

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/intel/issues/297.