Fix VirtualAudioInputStream callback timing issues.

content/renderer/media/audio_renderer_mixer_manager.cc

Index: content/renderer/media/audio_renderer_mixer_manager.cc
diff --git a/content/renderer/media/audio_renderer_mixer_manager.cc b/content/renderer/media/audio_renderer_mixer_manager.cc
index 05782d231fb618d66e8e824388f045e8866185db..86d482960f9675a83bfd11d0cdcb73eaa302b5f8 100644
--- a/content/renderer/media/audio_renderer_mixer_manager.cc
+++ b/content/renderer/media/audio_renderer_mixer_manager.cc
@@ -52,12 +52,24 @@ media::AudioRendererMixer* AudioRendererMixerManager::GetMixer(
     return it->second.mixer;
   }
 
+  int buffer_size = hardware_buffer_size_;
+
+// TODO(justinlin): Enable this on other platforms when renderer side mixing
+// is enabled for them if we don't have renderer side device changes yet.
+#if defined(OS_WIN) || defined(OS_MACOSX)
+  // Force at least a 2048 buffer size until we have renderer-side device
+  // changes. Audio mirroring will use the input device's buffer size to
+  // drive mirroring, so we need the output device's buffer size to be big
+  // enough to avoid garbled audio.
+  buffer_size = std::max(hardware_buffer_size_, 2048);

[miu, 2013/01/16 22:56:57]

Why not make this a function of hardware_sample_rate_?  Meaning, assume we want
a 10ms buffer, then:

  buffer_size = std::max(hardware_buffer_size_, hardware_sample_rate_ / 100);

+#endif
+
   // Create output parameters based on the audio hardware configuration for
   // passing on to the output sink.  Force to 16-bit output for now since we
   // know that works well for WebAudio and WebRTC.
   media::AudioParameters output_params(
       media::AudioParameters::AUDIO_PCM_LOW_LATENCY, params.channel_layout(),
-      hardware_sample_rate_, 16, hardware_buffer_size_);
+      hardware_sample_rate_, 16, buffer_size);
 
   // If we've created invalid output parameters, simply pass on the input params
   // and let the browser side handle automatic fallback.