Re: [PATCH] drm/i915/execlists: Clear STOP_RING bit on reset

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Quoting Mika Kuoppala (2019-09-10 10:31:05)
> Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> writes:
> 
> > During reset, we try to ensure no forward progress of the CS prior to
> > the reset by setting the STOP_RING bit in RING_MI_MODE. Since gen9, this
> > register is context saved and do we end up in the odd situation where we
> > save the STOP_RING bit and so try to stop the engine again immediately
> > upon resume. This is quite unexpected and causes us to complain about an
> > early CS completion event!
> 
> The completion event is a product of resume with a stop set?

A completion event is the product of STOP_RING. Whether it is the
completion event that we keep failing on...
 
> If my memory serves me well, we have fought this before.

Exactly. We've reduced the frequency of when we apply the STOP_RING, but
have not eliminated it.
 
> But I have still missing pieces. Why would we not want to
> set this for all contexts primed for execution? In gen8 too.

It's not in the gen8 context, afaict. I searched the context image for an
LRI with the RING_MI_MODE register:
https://patchwork.freedesktop.org/patch/329919/?series=66468&rev=1
 
> I mean, queuing context with a ring stopped just doesn't
> sound right on any gen.

We clear the STOP_RING in the register on resume just in case, and that
is being flagged on Icelake (which I put down to the reset not being very
thorough!). The remaining question was whether we were restoring it from
the context image.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx




[Index of Archives]     [AMD Graphics]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux