[Intel-gfx] ✗ Fi.CI.BAT: failure for series starting with [1/2] drm/i915/guc: fix GuC suspend/resume

[Michal Wajdeczko]

On Tue, 16 Oct 2018 10:49:54 +0200, Chris Wilson  
<chris at chris-wilson.co.uk> wrote:

> Quoting Daniele Ceraolo Spurio (2018-10-16 00:43:59)
...
>> >
>> >      igt at drv_selftest@live_hangcheck:
>> >        fi-skl-gvtdvm:      PASS -> DMESG-FAIL
>> >
>>
>> <7> [464.966238] [drm:guc_fw_xfer [i915]] GuC status 0x20
>> <3> [464.966361] [drm:guc_fw_xfer [i915]] *ERROR* GuC firmware xfer
>> error -110
>>
>> This looks like GuC is stuck very early in the boot flow (even before
>> the RSA check). On SKL there are known issues that could cause this and
>> we should reset GuC and retry, but we aren't. Looks like we indirectly
>> stopped applying  WaEnableuKernelHeaderValidFix and
>> WaEnableGuCBootHashCheckNotSet by not returning -EAGAIN from
>> intel_guc_fw_upload in any case. Michal?
>
> That would be useful to fix, as it does fail regularly enough to be a
> nuisance.

Proposed fix is here [1]. Note that earlier patch was trying to avoid the
case when we were repeating loading of the GuC firmware that simply didn't
pass signature verification (not fixable that way ;)

Michal

[1] https://patchwork.freedesktop.org/patch/256913/