[Intel-gfx] ✗ Fi.CI.BAT: failure for series starting with [01/18] drm/i915: Always wrap the ring offset before resetting

[Chris Wilson]

Quoting Patchwork (2019-08-19 10:15:26)
> Possible new issues
> -------------------
> 
>   Here are the unknown changes that may have been introduced in Patchwork_14078:
> 
> ### IGT changes ###
> 
> #### Possible regressions ####
>   * igt at gem_exec_reloc@basic-cpu-gtt:
>     - fi-snb-2520m:       [PASS][3] -> [DMESG-WARN][4]
>    [3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6734/fi-snb-2520m/igt@gem_exec_reloc@basic-cpu-gtt.html
>    [4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_14078/fi-snb-2520m/igt@gem_exec_reloc@basic-cpu-gtt.html

That's the lack of lifting vma->ops->set_pages().
Partial/rotated/remapped pages require allocations.

>   * igt at gem_render_linear_blits@basic:
>     - fi-bwr-2160:        [PASS][15] -> [FAIL][16] +7 similar issues
>    [15]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6734/fi-bwr-2160/igt@gem_render_linear_blits@basic.html
>    [16]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_14078/fi-bwr-2160/igt@gem_render_linear_blits@basic.html

Not sure what's upsetting Broadwater. Hmm, we are rearranging the PTE
writes to be after the execbuf chipset/gtt-flush now. Could be.

>   * igt at kms_frontbuffer_tracking@basic:
>     - fi-ivb-3770:        [PASS][61] -> [DMESG-WARN][62]
>    [61]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6734/fi-ivb-3770/igt@kms_frontbuffer_tracking@basic.html
>    [62]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_14078/fi-ivb-3770/igt@kms_frontbuffer_tracking@basic.html

Sigh, I was trying to save the display changes for a later patch. Looks
like I missed a path to vm->mutex and so they need converting at the same time.
-Chris