Run VDA::Decode on GPU IO thread if VDA supports it.

content/common/gpu/gpu_channel.h

Index: content/common/gpu/gpu_channel.h
diff --git a/content/common/gpu/gpu_channel.h b/content/common/gpu/gpu_channel.h
index 0692b903b4fab7128f9c171a3c6a8dd24fcdc945..23c27e0a0a491333c1bee334dfa225e1fe31bcb8 100644
--- a/content/common/gpu/gpu_channel.h
+++ b/content/common/gpu/gpu_channel.h
@@ -86,6 +86,10 @@ class GpuChannel : public IPC::Listener,
 
   base::ProcessId renderer_pid() const { return channel_->peer_pid(); }
 
+  scoped_refptr<base::MessageLoopProxy> io_message_loop() const {

[Ami GONE FROM CHROMIUM, 2013/08/26 17:20:03]

Since this is a process-wide singleton, it would make more sense to grab this
from someplace like ChildProcess::current()->io_message_loop_proxy() in the impl
(instead of plumbing it through to all ctors).

Doing that highlights that the VDAs now depend on an IO thread running, which
makes me wonder whether vdatest satisfies that dependency.  
But this CL doesn't update vdatest!  You should probably do that, and verify
that everything still works :)

[wuchengli, 2013/08/27 12:32:24]

GpuChannelManager manages more than one GpuChannel. GpuChannelManager is a
process-wide singleton, but GpuChannel is not. Do you still think ctor
io_message_loop of GpuChannelManager and GpuChannel should be removed?
I'll test vdatest.

+    return io_message_loop_;
+  }
+
   // IPC::Listener implementation:
   virtual bool OnMessageReceived(const IPC::Message& msg) OVERRIDE;
   virtual void OnChannelError() OVERRIDE;
@@ -156,6 +160,9 @@ class GpuChannel : public IPC::Listener,
 
   void CacheShader(const std::string& key, const std::string& shader);
 
+  void AddFilter(IPC::ChannelProxy::MessageFilter* filter);
+  void RemoveFilter(IPC::ChannelProxy::MessageFilter* filter);
+
  protected:
   virtual ~GpuChannel();