cc: Add checker-imaging support to TileManager.

cc/layers/picture_layer_impl.cc

Index: cc/layers/picture_layer_impl.cc
diff --git a/cc/layers/picture_layer_impl.cc b/cc/layers/picture_layer_impl.cc
index 4f6bfdb2981a21197bad72053e068f6b34ae6fd8..b6d37094a634689cfe6588a3145ef8405e0b0c06 100644
--- a/cc/layers/picture_layer_impl.cc
+++ b/cc/layers/picture_layer_impl.cc
@@ -1369,4 +1369,23 @@ bool PictureLayerImpl::HasValidTilePriorities() const {
          is_drawn_render_surface_layer_list_member();
 }
 
+void PictureLayerImpl::InvalidateRegionForImages(
+    const std::unordered_set<ImageId>& images_to_invalidate) {
+  InvalidationRegion image_invalidation;
+  for (auto image_id : images_to_invalidate) {
+    image_invalidation.Union(raster_source_->GetRectForImage(image_id));
+  }
+
+  if (image_invalidation.IsEmpty())

[vmpstr, 2017/02/10 19:25:42]

Should we just check if (images_to_invalidate.empty()) and return earlier?

I also want the trace to capture the loop when we union the regions, so if we do
that check and then move the trace above the for loop then you can use
TRACE_EVENT_BEGIN0/TRACE_EVENT_END1 to record the invalidation (you can see the
trace in TileManager::AssignGpuMemoryToTiles as an example)

[Khushal, 2017/02/10 22:09:19]

Good point. That check should be in LayerTreeImpl so we don't iterate through
the layers if we don't need to. I doubt that would be the case often anyway. The
empty invalidation here will most likely be from the fact that the image was
from a different layer, or may be it was removed in the next frame...
How is it now?

+    return;
+
+  Region invalidation;
+  image_invalidation.Swap(&invalidation);
+  TRACE_EVENT1("cc", "PictureLayerImpl::InvalidatingForImages", "invalidation",
+               invalidation.ToString());
+  invalidation_.Union(invalidation);
+  tilings_->UpdateTilingsForImplSideInvalidation(invalidation);
+  SetNeedsPushProperties();
+}
+
 }  // namespace cc