Re: ✗ i915.CI.BAT: failure for drm/i915/bw: Split bandwidth params into platform- and display-IP-specific structs (rev3)

[Gustavo Sousa <gustavo.sousa@intel.com>]

Patchwork <patchwork@emeril.freedesktop.org> writes:

> == Series Details ==
>
> Series: drm/i915/bw: Split bandwidth params into platform- and display-IP-specific structs (rev3)
> URL   : https://patchwork.freedesktop.org/series/164567/
> State : failure
>
> == Summary ==
>
> CI Bug Log - changes from CI_DRM_18491 -> Patchwork_164567v3
> ====================================================
>
> Summary
> -------
>
>   **FAILURE**
>
>   Serious unknown changes coming with Patchwork_164567v3 absolutely need to be
>   verified manually.
>   
>   If you think the reported changes have nothing to do with the changes
>   introduced in Patchwork_164567v3, please notify your bug team (I915-ci-infra@lists.freedesktop.org) 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_164567v3/index.html
>
> Participating hosts (42 -> 40)
> ------------------------------
>
>   Missing    (2): bat-dg2-13 fi-snb-2520m 
>
> Possible new issues
> -------------------
>
>   Here are the unknown changes that may have been introduced in Patchwork_164567v3:
>
> ### IGT changes ###
>
> #### Possible regressions ####
>
>   * igt@i915_module_load@load:
>     - bat-dg2-9:          [PASS][1] -> [ABORT][2]
>    [1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_18491/bat-dg2-9/igt@i915_module_load@load.html
>    [2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_164567v3/bat-dg2-9/igt@i915_module_load@load.html
>     - bat-dg2-14:         [PASS][3] -> [ABORT][4]
>    [3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_18491/bat-dg2-14/igt@i915_module_load@load.html
>    [4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_164567v3/bat-dg2-14/igt@i915_module_load@load.html
>     - bat-dg2-8:          [PASS][5] -> [ABORT][6]
>    [5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_18491/bat-dg2-8/igt@i915_module_load@load.html
>    [6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_164567v3/bat-dg2-8/igt@i915_module_load@load.html

Hm... This shows the warning because DG2 does not use bandwidth
parameters and get_soc_bw_params() rightfully does not cover DG2.

We could just get rid of the warnings in get_soc_bw_params() and
get_display_bw_params() all together.  The idea of the warning was to
serve as an aid to the developer, but I guess tracing the null pointer
dereference back to those functions shouldn't be too hard?

Another idea is to make sure that only the functions that use those
parameters make the call to get_{soc,display}_bw_params(). Jani, I know
you preferred the other way around, but maybe this is a compelling
reason for moving the call to the direct users?

--
Gustavo Sousa