[Intel-gfx] [PATCH 5/5] drm/i915: Implement an oom-notifier for last resort shrinking

Chris Wilson chris at chris-wilson.co.uk
Tue May 20 10:41:33 CEST 2014


On Tue, May 20, 2014 at 10:38:05AM +0200, Daniel Vetter wrote:
> On Tue, May 20, 2014 at 08:28:43AM +0100, Chris Wilson wrote:
> And one bikeshed below.

> > +	unbound = bound = pinned = 0;
> > +	list_for_each_entry(obj, &dev_priv->mm.unbound_list, global_list) {
> > +		if (!obj->base.filp) /* not backed by a freeable object */
> > +			continue;
> > +
> > +		if (obj->pages_pin_count)
> > +			pinned += obj->base.size;
> > +		else
> > +			unbound += obj->base.size;
> > +	}
> > +	list_for_each_entry(obj, &dev_priv->mm.bound_list, global_list) {
> > +		if (!obj->base.filp)
> > +			continue;
> > +
> > +		if (obj->pages_pin_count)
> > +			pinned += obj->base.size;
> > +		else
> > +			bound += obj->base.size;
> > +	}
> > +
> > +	if (unlock)
> > +		mutex_unlock(&dev->struct_mutex);
> > +
> > +	pr_info("Purging GPU memory, %lu bytes freed, %lu bytes still pinned.\n",
> > +		freed, pinned);
> > +	if (unbound | bound)
> 
> I prefer a boolean || here since we compare against non-zero for each and
> not bitfields.

Heh, and I prefer a union since we only care if either is non-zero and
don't want the code enlargement of short-circuiting. ;-)
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre



More information about the Intel-gfx mailing list