[Intel-gfx] ✗ Fi.CI.BAT: failure for series starting with [v2,1/2] drm/i915: Don't include intel_de.h from intel_display_types.h (rev2)

[Ville Syrjälä]

On Mon, May 03, 2021 at 07:54:55PM -0000, Patchwork wrote:
> == Series Details ==
> 
> Series: series starting with [v2,1/2] drm/i915: Don't include intel_de.h from intel_display_types.h (rev2)
> URL   : https://patchwork.freedesktop.org/series/89698/
> State : failure
> 
> == Summary ==
> 
> CI Bug Log - changes from CI_DRM_10039 -> Patchwork_20053
> ====================================================
> 
> Summary
> -------
> 
>   **FAILURE**
> 
>   Serious unknown changes coming with Patchwork_20053 absolutely need to be
>   verified manually.
>   
>   If you think the reported changes have nothing to do with the changes
>   introduced in Patchwork_20053, 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/Patchwork_20053/index.html
> 
> Possible new issues
> -------------------
> 
>   Here are the unknown changes that may have been introduced in Patchwork_20053:
> 
> ### IGT changes ###
> 
> #### Possible regressions ####
> 
>   * igt at i915_selftest@live at blt:
>     - fi-bdw-5557u:       [PASS][1] -> [INCOMPLETE][2]
>    [1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10039/fi-bdw-5557u/igt@i915_selftest@live@blt.html
>    [2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20053/fi-bdw-5557u/igt@i915_selftest@live@blt.html

<4> [276.958769] WARNING: CPU: 0 PID: 358 at kernel/workqueue.c:1418 __queue_work+0x2d2/0x620
...
<4> [276.959009]  queue_work_on+0x68/0x80
<4> [276.959016]  node_free+0x43/0x60 [i915]
<4> [276.959140]  pool_free_older_than+0x132/0x170 [i915]
<4> [276.959270]  pool_free_work+0x12/0x40 [i915]
<4> [276.959395]  process_one_work+0x270/0x5c0
<4> [276.959407]  worker_thread+0x37/0x380

Which is
        /* if draining, only works from the same workqueue are allowed */
	if (unlikely(wq->flags & __WQ_DRAINING) &&
	    WARN_ON_ONCE(!is_chained_work(wq)))
	        return;

Based on a cursory glance the queue_work is probably coming from
i915_gem_free_object() -> queue_work(i915->wq, &i915->mm.free_work);

So looks like i915_gem_drain_workqueue() is broken.

Entirely unrelated to this patch series.

-- 
Ville Syrjälä
Intel