I've been seeing memory leaks on my system in the form of large
(300-400MB) GEM objects created by now-dead processes laying around
clogging up memory.  I usually notice when it gets to about 1.2GB of
them.  Hopefully this clears up the issue, but I just found this bug
by inspection.
Signed-off-by: Eric Anholt <eric@anholt.net>
Cc: stable@kernel.org
Signed-off-by: Keith Packard <keithp@keithp.com>
 
        if (obj->base.size > dev_priv->mm.gtt_mappable_end) {
                ret = -E2BIG;
-               goto unlock;
+               goto out;
        }
 
        if (obj->madv != I915_MADV_WILLNEED) {