cc: Video layers without a frame should not occlude

content/renderer/media/android/webmediaplayer_android.cc

Index: content/renderer/media/android/webmediaplayer_android.cc
diff --git a/content/renderer/media/android/webmediaplayer_android.cc b/content/renderer/media/android/webmediaplayer_android.cc
index 763bdaaa652d7da069bce350e666e9a271c5f41e..d138c48ed05ce27bf22af7b725e2c167579c7518 100644
--- a/content/renderer/media/android/webmediaplayer_android.cc
+++ b/content/renderer/media/android/webmediaplayer_android.cc
@@ -1280,6 +1280,10 @@ bool WebMediaPlayerAndroid::UpdateCurrentFrame(base::TimeTicks deadline_min,
   return false;
 }
 
+bool WebMediaPlayerAndroid::HasCurrentFrame() const {
+  return current_frame_;

[danakj, 2015/05/18 18:15:45]

you need to lock current_frame_lock_ no?

[hendrikw, 2015/05/18 18:33:59]

Yeah, I was unsure about this.  We have a lock in the caller, and
VideoFrameCompositor doesn't lock on the GetCurrentFrame.  
Calling (bool)scoped_ptr is atomic.

I can play it safe and put the locks in here, but I'll lose constness.

Let me know that you think.

[danakj, 2015/05/18 18:44:57]

I /think/ in VFC the lock_ is locking access to the callback_ and not the
current_frame_. I guess it doesn't access the frame from other threads.
What makes you say it's atomic? It's implemented as:

return impl_.get() ? &scoped_ptr::impl_ : nullptr;
I think we need to for correctness.

+}
+
 scoped_refptr<media::VideoFrame> WebMediaPlayerAndroid::GetCurrentFrame() {
   scoped_refptr<VideoFrame> video_frame;
   {