drm/i915: fix scheduling while holding the new active list spinlock
[sfrench/cifs-2.6.git] / drivers / gpu / drm / i915 / i915_gem.c
index 6dca9fc7c1dbd40bb12bad95986d148501af1d49..4642115902d61a419ed112b39c5680810cbb6d35 100644 (file)
@@ -1596,8 +1596,19 @@ i915_gem_retire_request(struct drm_device *dev,
 
                if (obj->write_domain != 0)
                        i915_gem_object_move_to_flushing(obj);
-               else
+               else {
+                       /* Take a reference on the object so it won't be
+                        * freed while the spinlock is held.  The list
+                        * protection for this spinlock is safe when breaking
+                        * the lock like this since the next thing we do
+                        * is just get the head of the list again.
+                        */
+                       drm_gem_object_reference(obj);
                        i915_gem_object_move_to_inactive(obj);
+                       spin_unlock(&dev_priv->mm.active_list_lock);
+                       drm_gem_object_unreference(obj);
+                       spin_lock(&dev_priv->mm.active_list_lock);
+               }
        }
 out:
        spin_unlock(&dev_priv->mm.active_list_lock);