[igt-dev] ✗ Fi.CI.IGT: failure for i915/gem_ctx_isolation: Sanitycheck nonpriv access

[Chris Wilson]

Quoting Patchwork (2019-02-22 23:10:31)
> == Series Details ==
> 
> Series: i915/gem_ctx_isolation: Sanitycheck nonpriv access
> URL   : https://patchwork.freedesktop.org/series/57091/
> State : failure
> 
> == Summary ==
> 
> CI Bug Log - changes from CI_DRM_5654_full -> IGTPW_2489_full
> ====================================================
> 
> Summary
> -------
> 
>   **FAILURE**
> 
>   Serious unknown changes coming with IGTPW_2489_full absolutely need to be
>   verified manually.
>   
>   If you think the reported changes have nothing to do with the changes
>   introduced in IGTPW_2489_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://patchwork.freedesktop.org/api/1.0/series/57091/revisions/1/mbox/
> 
> Possible new issues
> -------------------
> 
>   Here are the unknown changes that may have been introduced in IGTPW_2489_full:
> 
> ### IGT changes ###
> 
> #### Possible regressions ####
> 
>   * igt at gem_ctx_isolation@rcs0-dirty-switch:
>     - shard-glk:          PASS -> FAIL +1
> 
>   * igt at gem_ctx_isolation@rcs0-reset:
>     - shard-apl:          PASS -> FAIL
> 
>   * igt at gem_ctx_isolation@rcs0-s3:
>     - shard-kbl:          PASS -> FAIL
>     - shard-apl:          NOTRUN -> FAIL

So L3SQCREG4 is also not context saved but whitelisted. So anybody is
free to severely hinder performance of the rest of the system.

Sigh. I must push gem_ctx_isolation into a selftest so that we check the
whitelist before being merged.
-Chris