Associate audio streams with their source/destination RenderView.

media/audio/audio_output_ipc.h

Index: media/audio/audio_output_ipc.h
diff --git a/media/audio/audio_output_ipc.h b/media/audio/audio_output_ipc.h
index 8543cdc170eb73139ece7ce5b004f3e9405f5e4e..45ef59a445ad25d09142a3c01cd95a1f307ca037 100644
--- a/media/audio/audio_output_ipc.h
+++ b/media/audio/audio_output_ipc.h
@@ -5,6 +5,7 @@
 #ifndef MEDIA_AUDIO_AUDIO_OUTPUT_IPC_H_
 #define MEDIA_AUDIO_AUDIO_OUTPUT_IPC_H_
 
+#include "base/memory/weak_ptr.h"
 #include "base/shared_memory.h"
 #include "base/sync_socket.h"
 #include "media/audio/audio_parameters.h"
@@ -55,7 +56,8 @@ class MEDIA_EXPORT AudioOutputIPCDelegate {
 // should asynchronously deliver the messages to an AudioOutputController object
 // (or create one in the case of CreateStream()), that may live in a separate
 // process.
-class MEDIA_EXPORT AudioOutputIPC {
+class MEDIA_EXPORT AudioOutputIPC
+    : public base::SupportsWeakPtr<AudioOutputIPC> {

[scherkus (not reviewing), 2012/11/27 22:55:26]

Ahh! What was the motivation for this change?

I don't approve of this sort of WeakPtr usage -- for details you can read
willchan@'s rant on chromium-dev:
https://groups.google.com/a/chromium.org/d/msg/chromium-dev/Oy3uuiTSwDk/5UDyR...

[miu, 2012/11/28 07:26:20]

Reverted the changes to this file.

I had meant to come back to this.  If you're curious, the original motivation
was due to the fact that I wasn't sure whether AudioOutputIPC would outlive
AudioOutputDevice, and I needed to be sure AudioOutputIPC::RemoveDelegate()
could be safely called from the destructor.  I resolved all of this.

  public:
   // Registers an AudioOutputIPCDelegate and returns a |stream_id| that must
   // be used with all other IPC functions in this interface.