Re: [PATCH v2 05/14] drm/i915: Deconstruct execute fence

From: Chris Wilson <chris@chris-wilson.co.uk>
To: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH v2 05/14] drm/i915: Deconstruct execute fence
Date: Fri, 17 Feb 2017 14:41:05 +0000	[thread overview]
Message-ID: <20170217144105.GB4100@nuc-i3427.alporthouse.com> (raw)
In-Reply-To: <0ac032a6-5a8f-7e55-44dd-6bd3d286400b@linux.intel.com>

On Fri, Feb 17, 2017 at 02:26:04PM +0000, Tvrtko Ursulin wrote:
> 
> On 14/02/2017 09:54, Chris Wilson wrote:
> >@@ -1036,12 +1011,11 @@ long i915_wait_request(struct drm_i915_gem_request *req,
> > 	if (flags & I915_WAIT_LOCKED)
> > 		add_wait_queue(errq, &reset);
> >
> >-	if (!i915_sw_fence_done(&req->execute)) {
> >-		DEFINE_WAIT(exec);
> >-
> >+	reset_wait_queue(&req->execute, &exec);
> >+	if (!req->global_seqno) {
> > 		do {
> >-			prepare_to_wait(&req->execute.wait, &exec, state);
> 
> Somehow I missed the moment when reset_wait_queue was introduced.
> But why you can't just use prepare_to_wait here?

In a few patches, we starting looping around at this point, so reset is
required then (and choosing to do reset now just avoid a few lines
later on). Then in a few more patches, the reset_wait_queue is eliminated
entirely.
-Chris

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