Mac ÜC: Enable back-pressure for the browser compositor

cc/output/output_surface.cc

Index: cc/output/output_surface.cc
diff --git a/cc/output/output_surface.cc b/cc/output/output_surface.cc
index 46096d2e3094febd271e31a47c8668fbc1158ea2..8d84cfb054a6f120172f9bcae8f80a2e56fb709f 100644
--- a/cc/output/output_surface.cc
+++ b/cc/output/output_surface.cc
@@ -342,10 +342,14 @@ void OutputSurface::UpdateAndMeasureGpuLatency() {
 void OutputSurface::PostSwapBuffersComplete() {
   base::MessageLoop::current()->PostTask(
       FROM_HERE,
-      base::Bind(&OutputSurface::OnSwapBuffersComplete,
+      base::Bind(&OutputSurface::DoOnSwapBuffersComplete,

[danakj, 2014/06/12 01:16:51]

Does virtual method in a posttask not do what you'd expect it to do, or why do
you need this indirection through DoOnSwapBuffersComplete, and not just bind
OnSwapBuffersComplete here directly?

[ccameron, 2014/06/12 01:28:25]

I was under the impression that a virtual method sent to posttask doesn't work
(well, it calls the precise method listed, not the derived class method).

But... I just tested this empirically, and it totally works! Huh! I wonder how
that's implemented...

                  weak_ptr_factory_.GetWeakPtr()));
 }
 
+void OutputSurface::DoOnSwapBuffersComplete() {
+  OnSwapBuffersComplete();
+}
+
 // We don't post tasks bound to the client directly since they might run
 // after the OutputSurface has been destroyed.
 void OutputSurface::OnSwapBuffersComplete() {