Oilpan: Use transition types for HTMLMediaElement and make AudioSourceProviderClient a GC mixin.

Source/core/html/HTMLMediaElement.h

 /*
  * Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, 2013 Apple Inc. All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
  * 1. Redistributions of source code must retain the above copyright
  *    notice, this list of conditions and the following disclaimer.
  * 2. Redistributions in binary form must reproduce the above copyright
  *    notice, this list of conditions and the following disclaimer in the
@@ skipping 496 matching lines @@
     RefPtrWillBeMember<TextTrackList> m_textTracks;
     WillBeHeapVector<RefPtrWillBeMember<TextTrack> > m_textTracksWhenResourceSelectionBegan;
 
     CueIntervalTree m_cueTree;
 
     CueList m_currentlyActiveCues;
     int m_ignoreTrackDisplayUpdate;
 
 #if ENABLE(WEB_AUDIO)
     // This is a weak reference, since m_audioSourceNode holds a reference to us.
-    AudioSourceProviderClient* m_audioSourceNode;
+    RawPtrWillBeMember<AudioSourceProviderClient> m_audioSourceNode;

[haraken, 2014/05/21 16:35:58]

Shouldn't this be a weak member?

[zerny-chromium, 2014/05/22 05:54:32]

It appears to be raw simply to break what would otherwise be a ref cycle. I
can't see any apparent reason that the audioSourceNode must be reclaimed if only
the HTMLMediaElement is referencing it.

[haraken, 2014/05/22 07:33:36]

In the current implementation, it looks like that HTMLMediaElement can outlive
AudioSourceProviderClient. To keep the behavior, I guess this pointer needs to
be weak. (Also the comment in line 516 implies that this point is weak.)

[Mads Ager (chromium), 2014/05/22 08:02:56]

I think what that comment is saying is that this is a raw pointer because
m_audioSourceNode has a RefPtr to us and therefore it would introduce a cycle to
use a RefPtr here. The use of the work "weak" here just means that it is a "raw"
pointer.

I agree with Ian that these objects should just live and die together. As far as
I can tell this pointer is raw just to break a reference cycle between the
HTMLMediaElement and the AudioSourceNode that is providing audio for this
element.

 #endif
 
     friend class MediaController;
     RefPtr<MediaController> m_mediaController;
 
     friend class Internals;
     friend class TrackDisplayUpdateScope;
 
     static URLRegistry* s_mediaStreamRegistry;
 };
@@ skipping 25 matching lines @@
 inline bool isHTMLMediaElement(const HTMLElement& element)
 {
     return isHTMLAudioElement(element) || isHTMLVideoElement(element);
 }
 
 DEFINE_HTMLELEMENT_TYPE_CASTS_WITH_FUNCTION(HTMLMediaElement);
 
 } //namespace
 
 #endif