[Intel-gfx] [PATCH v3] drm/i915/guc: Make wq_lock irq-safe

Chris Wilson chris at chris-wilson.co.uk
Tue Feb 28 10:16:05 UTC 2017


On Tue, Feb 28, 2017 at 11:07:58AM +0100, Michal Wajdeczko wrote:
> On Tue, Feb 28, 2017 at 09:28:10AM +0000, Chris Wilson wrote:
> > @@ -951,10 +958,12 @@ int i915_guc_submission_enable(struct drm_i915_private *dev_priv)
> >  		engine->schedule = NULL;
> >  
> >  		/* Replay the current set of previously submitted requests */
> > +		spin_lock_irq(&engine->timeline->lock);
> >  		list_for_each_entry(rq, &engine->timeline->requests, link) {
> > -			client->wq_rsvd += sizeof(struct guc_wq_item);
> > +			guc_wq_add_reserved(client, sizeof(struct guc_wq_item));
> 
> To be consistent with "unreserve" case, maybe we should define and then use wqi_size?
> 
> 	const int wqi_size = sizeof(struct guc_wq_item);

Fine, only because with the super long function name I need to shave a
few columns :-p
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre


More information about the Intel-gfx mailing list