Re: [Intel-gfx] [PATCH] drm/i915/gt: Detect if we miss WaIdleLiteRestore

2019-12-10 Thread Tvrtko Ursulin
On 09/12/2019 02:32, Chris Wilson wrote: In order to avoid confusing the HW, we must never submit an empty ring during lite-restore, that is we should always advance the RING_TAIL before submitting to stay ahead of the RING_HEAD. Normally this is prevented by keeping a couple of spare NOPs in

Re: [Intel-gfx] [PATCH] drm/i915/gt: Detect if we miss WaIdleLiteRestore

2019-12-10 Thread Chris Wilson
Quoting Tvrtko Ursulin (2019-12-10 10:00:34) > > On 09/12/2019 02:32, Chris Wilson wrote: > > In order to avoid confusing the HW, we must never submit an empty ring > > during lite-restore, that is we should always advance the RING_TAIL > > before submitting to stay ahead of the RING_HEAD. > > >

[Intel-gfx] [PATCH] drm/i915/gt: Detect if we miss WaIdleLiteRestore

2019-12-08 Thread Chris Wilson
In order to avoid confusing the HW, we must never submit an empty ring during lite-restore, that is we should always advance the RING_TAIL before submitting to stay ahead of the RING_HEAD. Normally this is prevented by keeping a couple of spare NOPs in the request->wa_tail so that on resubmission

Re: [Intel-gfx] [PATCH] drm/i915/gt: Detect if we miss WaIdleLiteRestore

2019-12-08 Thread Chris Wilson
Quoting Chris Wilson (2019-12-09 02:01:45) > In order to avoid confusing the HW, we must never submit an empty ring > during lite-restore, that is we should always advance the RING_TAIL > before submitting to stay ahead of the RING_HEAD. > > Normally this is prevented by keeping a couple of spare

[Intel-gfx] [PATCH] drm/i915/gt: Detect if we miss WaIdleLiteRestore

2019-12-08 Thread Chris Wilson
In order to avoid confusing the HW, we must never submit an empty ring during lite-restore, that is we should always advance the RING_TAIL before submitting to stay ahead of the RING_HEAD. Normally this is prevented by keeping a couple of spare NOPs in the request->wa_tail so that on resubmission