[Intel-gfx] [PATCH 12/20] drm/i915: Add a delay between interrupt and inspecting the final seqno (ilk)

Chris Wilson chris at chris-wilson.co.uk
Fri Jul 1 14:35:26 UTC 2016


On Fri, Jul 01, 2016 at 03:27:40PM +0100, Tvrtko Ursulin wrote:
> 
> On 01/07/16 12:22, Chris Wilson wrote:
> >On Ironlake, there is no command nor register to ensure that the write
> >from a MI_STORE command is completed (and coherent on the CPU) before the
> >command parser continues. This means that the ordering between the seqno
> 
> Command *streamer* I think. (More instances below.)

Yeah, probably better to avoid confusing people into thinking about our
own command parser. I just had in mind that the execution is separate
from the streaming (and fetching is yet another phase).

> >@@ -3087,9 +3040,10 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev)
> >  	} else {
> >  		engine->mmio_base = BSD_RING_BASE;
> >  		engine->flush = bsd_ring_flush;
> >-		if (IS_GEN5(dev_priv))
> >+		if (IS_GEN5(dev_priv)) {
> >  			engine->irq_enable_mask = ILK_BSD_USER_INTERRUPT;
> >-		else
> >+			engine->irq_seqno_barrier = gen5_seqno_barrier;
> 
> This is already set in common setup AFAICS.

Yes, an oversight after sending the tidying patch earlier.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre


More information about the Intel-gfx mailing list