[Intel-gfx] [PATCH v2 01/11] drm/i915: Disable preemption and sleeping while using the punit sideband

[Mika Kuoppala]

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

> Quoting Mika Kuoppala (2018-01-15 12:04:40)
>> Chris Wilson <chris at chris-wilson.co.uk> writes:
>> 
>> > While we talk to the punit over its sideband, we need to prevent the cpu
>> > from sleeping in order to prevent a potential machine hang.
>> >
>> > Note that by itself, it appears that pm_qos_update_request (via
>> > intel_idle) doesn't provide a sufficient barrier to ensure that all core
>> > are indeed awake (out of Cstate) and that the package is awake. To do so,
>> > we need to supplement the pm_qos with a manual ping on_each_cpu.
>> >
>> > v2: Restrict the heavy-weight wakeup to just the ISOF_PORT_PUNIT, there
>> > is insufficient evidence to implicate a wider problem atm. Similarly,
>> > restrict the w/a to Valleyview, as Cherryview doesn't have an angry cadre
>> > of users.
>> >
>> 
>> One datapoint about the v1 of this patch with the cpu ping in it.
>> The j1900 byt did end up with system hang during weekend with
>> drm-tip + v1.
>
> Not much different (supposedly in v2, especially if we ignore the bits
> after the revert), and v2 ran most of Fri-Mon without incident. (Most
> because I tested a few other things as well, such as eliminating the
> waitboosting for glxgears in the reproducer.)
>
> 1 hang in 100 hours across 2 machines. Certainly the best we've achieved
> so far (while allowing RPS to frequently adjust gpufreq). Any fresh ideas?

Since you asked, I tried to express one idea with a fresh patch:
20180116152116.17900-1-mika.kuoppala at linux.intel.com

-Mika