[Intel-gfx] ✓ Fi.CI.BAT: success for series starting with [CI,1/2] drm/i915: Mark the end of intel_ring_begin() and check in intel_ring_advance()

[Mika Kuoppala]

Chris Wilson <chris at chris-wilson.co.uk> writes:

> On Mon, Feb 06, 2017 at 07:52:15PM -0000, Patchwork wrote:
>> == Series Details ==
>> 
>> Series: series starting with [CI,1/2] drm/i915: Mark the end of intel_ring_begin() and check in intel_ring_advance()
>> URL   : https://patchwork.freedesktop.org/series/19169/
>> State : success
>> 
>> == Summary ==
>> 
>> Series 19169v1 Series without cover letter
>> https://patchwork.freedesktop.org/api/1.0/series/19169/revisions/1/mbox/
>> 
>> Test kms_force_connector_basic:
>>         Subgroup force-connector-state:
>>                 skip       -> PASS       (fi-snb-2520m)
>
> Ok, I pushed this as given Mika's feedback this may be responsible for
> skl (at least) sporadically dieing in CI. I'm sure the macros for the
> conditional code I added can and will be improved.

Observations with my kbl indicate that the retirement freed the request's
context, which then got referenced by the GEM_BUG_ON. Instead of neat
trace, full system hang ensued.

With this patch, my kbl seems very stable again.

I suspect that atleast some incomplete CI results we have seen,
on execlist machines, were due to this.

-Mika

> -Chris
>
> -- 
> Chris Wilson, Intel Open Source Technology Centre
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx