Removing a RenderLayer's CLM should trigger a repaint

Source/core/rendering/compositing/RenderLayerCompositor.cpp

 /*
  * Copyright (C) 2009, 2010 Apple Inc. All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
  * 1. Redistributions of source code must retain the above copyright
  *    notice, this list of conditions and the following disclaimer.
  * 2. Redistributions in binary form must reproduce the above copyright
  *    notice, this list of conditions and the following disclaimer in the
@@ skipping 414 matching lines @@
             if (layer->isReflection()) {
                 RenderLayer* sourceLayer = toRenderLayerModelObject(layer->renderer()->parent())->layer();
                 if (sourceLayer->hasCompositedLayerMapping()) {
                     ASSERT(sourceLayer->compositedLayerMapping()->mainGraphicsLayer()->replicaLayer() == layer->compositedLayerMapping()->mainGraphicsLayer());
                     sourceLayer->compositedLayerMapping()->mainGraphicsLayer()->setReplicatedByLayer(0);
                 }
             }
 
             layer->clearCompositedLayerMapping();
             compositedLayerMappingChanged = true;
+
+            // If we need to repaint, do so now that we've removed the compositedLayerMapping.
+            repaintOnCompositingChange(layer);

[Ian Vollick, 2014/06/12 20:17:10]

This is interesting. When compositedLayerMappingChanged is set to true, the
CompositingLayerAssign adds it to a list of layers to repaint later on. Is it
somehow invalid to be issuing deferred repaints?

[ajuma, 2014/06/12 20:33:50]

The call to allocateOrClearCompositedLayerMapping made in
applyUpdateLayerCompositingStateChickenEggHacks ignores the returned value (that
is, the value of compositedLayerMappingChanged). Issuing a repaint there when
the returned value is true fixes the bug too.

[Ian Vollick, 2014/06/12 20:37:07]

Ah, right! I think that's far preferable to issuing the repaint here. Not only
will that let us defer repaints correctly on the awesome-RLC-codepath, it'll
ensure that it's removed when the hacks go away.

[chrishtr, 2014/06/12 20:37:29]

Please do it there instead. That way it's contained to the chicken-egg hack
case.

[ajuma, 2014/06/12 21:08:00]

Done.

         }
 
         break;
     case RemoveFromSquashingLayer:
     case NoCompositingStateChange:
         // Do nothing.
         break;
     }
 
     if (layer->hasCompositedLayerMapping() && layer->compositedLayerMapping()->updateRequiresOwnBackingStoreForIntrinsicReasons())
@@ skipping 794 matching lines @@
     } else if (graphicsLayer == m_scrollLayer.get()) {
         name = "LocalFrame Scrolling Layer";
     } else {
         ASSERT_NOT_REACHED();
     }
 
     return name;
 }
 
 } // namespace WebCore