Introduce AudioRendererSink::IsOptimizedForHardwareParameters

media/blink/webaudiosourceprovider_impl.cc

Index: media/blink/webaudiosourceprovider_impl.cc
diff --git a/media/blink/webaudiosourceprovider_impl.cc b/media/blink/webaudiosourceprovider_impl.cc
index 4cace4314b4203352d45914770eafa38930597a3..28a4057258cc3eb798a705d2342072d36e3112a6 100644
--- a/media/blink/webaudiosourceprovider_impl.cc
+++ b/media/blink/webaudiosourceprovider_impl.cc
@@ -258,6 +258,10 @@ OutputDeviceInfo WebAudioSourceProviderImpl::GetOutputDeviceInfo() {
                : OutputDeviceInfo(OUTPUT_DEVICE_STATUS_ERROR_NOT_FOUND);
 }
 
+bool WebAudioSourceProviderImpl::IsOptimizedForHardwareParameters() {
+  return client_ ? false : true;

[o1ka, 2017/05/16 10:31:19]

Access to |client_| is guarded everywhere by |sink_lock_|.
Is data race intentional here? Or is it not possible?
I would prefer to avoid locking here if we can, because we've had troubles with
that lock before (around l.119). But then we need a comment explaining what's
going on.

[flim-chromium, 2017/05/16 11:04:05]

Thanks for spotting this! I think it would actually be safer to add the lock; I
can't see a guarantee that there won't be a race with SetClient. If that also
makes sense to you, I'll upload the change.

+}
+
 bool WebAudioSourceProviderImpl::CurrentThreadIsRenderingThread() {
   NOTIMPLEMENTED();
   return false;