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

Source/core/html/HTMLMediaElement.cpp

Index: Source/core/html/HTMLMediaElement.cpp
diff --git a/Source/core/html/HTMLMediaElement.cpp b/Source/core/html/HTMLMediaElement.cpp
index 7aaf7d3a7165561a718cdaf70a6841012a38b1fb..9e387d720bc9d313683a41f986de8d9a3cf5d4cc 100644
--- a/Source/core/html/HTMLMediaElement.cpp
+++ b/Source/core/html/HTMLMediaElement.cpp
@@ -286,7 +286,7 @@ HTMLMediaElement::HTMLMediaElement(const QualifiedName& tagName, Document& docum
     , m_textTracks(nullptr)
     , m_ignoreTrackDisplayUpdate(0)
 #if ENABLE(WEB_AUDIO)
-    , m_audioSourceNode(0)
+    , m_audioSourceNode(nullptr)
 #endif
 {
     ASSERT(RuntimeEnabledFeatures::mediaEnabled());
@@ -363,7 +363,16 @@ HTMLMediaElement::~HTMLMediaElement()
     document().incrementLoadEventDelayCount();
 #endif
 
+#if ENABLE(OILPAN)
+#if ENABLE(WEB_AUDIO)
+    if (audioSourceProvider())
+        audioSourceProvider()->setClient(0);
+#endif
+    m_player.clear();
+#else
+    ASSERT(!m_audioSourceNode);

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

I'm curious why you added this ASSERT. If m_audioSourceNode is really guaranteed
to be 0 here, we don't need to have the following code in
clearMediaPlayerAndAudioSourceProviderClient.

if (m_audioSourceNode)
  m_audioSourceNode->lock();
...;
if (m_audioSourceNode)
  m_audioSourceNode->unlock();

(I added the lock/unlock to fix some threading issues a long time ago.)

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

Exactly. That is why I inlined the reduced code in the oilpan build and added
the assert to the non-oilpan build. It really must be the case since the audio
node has a ref-ptr to the element and upon destruction calls
setAudioSourceNode(0).

I could inline clearMediaPlayerAndAudioSourceProviderClient for both builds.
That would probably be more clear.

The reason why the locking is needed in
clearMediaPlayerAndAudioSourceProviderClient is because of its use outside of
the destructor.

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

Maybe we can split this into two methods:

clearMediaPlayerAndAudioSourceProviderClientWithoutLocking()
    - do the actual clearing

clearMediaPlayerAndAudioSourceProviderClient():
   - lock
   - clearMediaPlayerAndAudioSourceProviderClientWithoutLocking()
   - unlock

The call the one without locking here and have the assert that m_audioSourceNode
is NULL for the non-oilpan build.

     clearMediaPlayerAndAudioSourceProviderClient();
+#endif
 
 #if !ENABLE(OILPAN)
     document().decrementLoadEventDelayCount();
@@ -3652,6 +3661,7 @@ void HTMLMediaElement::trace(Visitor* visitor)
     visitor->trace(m_nextChildNodeToConsider);
     visitor->trace(m_textTracks);
     visitor->trace(m_textTracksWhenResourceSelectionBegan);
+    visitor->trace(m_audioSourceNode);
     WillBeHeapSupplementable<HTMLMediaElement>::trace(visitor);
     HTMLElement::trace(visitor);
 }