ThreadedTest that a non-null OutputSurface that fails initialization immediately doesn't crash

cc/thread_proxy.cc

 // Copyright 2011 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "cc/thread_proxy.h"
 
 #include "base/bind.h"
 #include "base/debug/trace_event.h"
 #include "cc/delay_based_time_source.h"
 #include "cc/draw_quad.h"
@@ skipping 994 matching lines @@
         DebugScopedSetMainThreadBlocked mainThreadBlocked(this);
         Proxy::implThread()->postTask(base::Bind(&ThreadProxy::commitPendingOnImplThreadForTesting, m_implThreadWeakPtr, &commitPendingRequest));
         commitPendingRequest.completion.wait();
     }
     return commitPendingRequest.commitPending;
 }
 
 void ThreadProxy::commitPendingOnImplThreadForTesting(CommitPendingRequest* request)
 {
     DCHECK(isImplThread());
-    request->commitPending = m_schedulerOnImplThread->commitPending();
+    if (m_layerTreeHostImpl->outputSurface())

[danakj, 2013/01/08 01:06:50]

Ya, I was thinking about something like this also. I'm fine with this, as it
shouldn't introduce races. You know that no commit is possibly sitting in a
message loop queue, so that sounds reasonable to me.

+        request->commitPending = m_schedulerOnImplThread->commitPending();
     request->completion.signal();
 }
 
 }  // namespace cc