Web Audio: Unapply Oilpan to AudioSummingJunction.

Web Audio: Drop Oilpan from AudioSummingJunction.

Drop GarbageCollectedFinalized inheritance from AudioSummingJunction,
and add GarbageCollectedFinalized to AudioNodeInput and
AudioParamHandler, which are derived from AudioSummingJunction.

- AudioSummingJunction doesn't have Member<AudioContext>, has
  RefPtr<DeferredTaskHandler> instead.
- Code to trace m_renderingOutputs is moved to AudioNodeInput and
  AudioParamHandler.
- AudioContext::m_liveAudioSummingJunctions is unnecessary
  because we can unregister AudioSummingJunction from
  DeferredTaskHandler in ~AudioSummingJunction().

BUG=464617
R=haraken@chromium.org, rtoy@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=192574

[tkent, 2015-03-25 07:31:51]

tkent@chromium.org changed reviewers:
+ oilpan-reviews@chromium.org, rtoy@chromium.org

[tkent, 2015-03-25 07:31:52]

Please review this.

[haraken, 2015-03-25 12:34:48]

haraken@chromium.org changed reviewers:
+ haraken@chromium.org

[haraken, 2015-03-25 12:34:48]

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioNodeInput.cpp (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioNodeInput.cpp:35: inline
AudioNodeInput::AudioNodeInput(AudioNode& node)

Help me understand: Why is it OK to not have Member<AudioContext> in
AudioNodeInput? Conceptually, I guess AudioNodeInput should keep the
AudioContext alive. Also the previous code has a Member<AudioContext> in
AudioSummingJunction.

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioNodeInput.cpp:53: AudioContext::AutoLocker
locker(deferredTaskHandler());

Isn't it possible that a GC is triggered and the control reaches here before the
handler gets initialized?

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioSummingJunction.cpp (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioSummingJunction.cpp:44: m_didCallDispose = true;

I guess m_didCallDispose is no longer helpful.

[Raymond Toy, 2015-03-25 15:43:54]

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioSummingJunction.h (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioSummingJunction.h:45: DeferredTaskHandler&
deferredTaskHandler() { return *m_deferredTaskHandler; }
Is it possible for m_deferredTaskHandler to ever be null?

[tkent, 2015-03-25 22:28:08]

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioNodeInput.cpp (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioNodeInput.cpp:35: inline
AudioNodeInput::AudioNodeInput(AudioNode& node)
On 2015/03/25 12:34:48, haraken wrote:
> 
> Help me understand: Why is it OK to not have Member<AudioContext> in
> AudioNodeInput? Conceptually, I guess AudioNodeInput should keep the
> AudioContext alive. Also the previous code has a Member<AudioContext> in
> AudioSummingJunction.

AudioNodeInput doesn't need AudioContext.  It's one of the purpose to split
DeferredTaskHandler from AudioContext.

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioNodeInput.cpp:53: AudioContext::AutoLocker
locker(deferredTaskHandler());
On 2015/03/25 12:34:48, haraken wrote:
> 
> Isn't it possible that a GC is triggered and the control reaches here before
the
> handler gets initialized?

It's impossible.  We can't create AudioNodeInput before DeferredTaskHandler
initialization.

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioSummingJunction.cpp (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioSummingJunction.cpp:44: m_didCallDispose = true;
On 2015/03/25 12:34:48, haraken wrote:
> 
> I guess m_didCallDispose is no longer helpful.

Right.  Will remove it.

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
File Source/modules/webaudio/AudioSummingJunction.h (right):

https://codereview.chromium.org/1037683002/diff/1/Source/modules/webaudio/Aud...
Source/modules/webaudio/AudioSummingJunction.h:45: DeferredTaskHandler&
deferredTaskHandler() { return *m_deferredTaskHandler; }
On 2015/03/25 15:43:54, Raymond Toy wrote:
> Is it possible for m_deferredTaskHandler to ever be null?

It's never be null.  AudioSummingJunction constructor always take a valid
reference of DeferredTaskHandler, and AudioSummingJunction keeps a
RefPtr<DeferredTaskHandler>.

[Raymond Toy, 2015-03-25 23:08:17]

lgtm

[haraken, 2015-03-26 01:16:16]

LGTM

[tkent, 2015-03-26 02:34:24]

The CQ bit was checked by tkent@chromium.org

[tkent, 2015-03-26 02:34:25]

The patchset sent to the CQ was uploaded after l-g-t-m from rtoy@chromium.org
Link to the patchset: https://codereview.chromium.org/1037683002/#ps20001
(title: "Remove m_didCallDispose")

[commit-bot: I haz the power, 2015-03-26 02:34:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1037683002/20001

[tkent, 2015-03-26 03:51:40]

Committed patchset #2 (id:20001) manually as 192574 (presubmit successful).

[tkent, 2015-03-30 08:31:57]

It seems this change made audionode-disconnect-audioparam.html flaky. 
Investigating.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showExpec...

[tkent, 2015-04-01 06:23:34]

On 2015/03/30 08:31:57, tkent wrote:
> It seems this change made audionode-disconnect-audioparam.html flaky. 
> Investigating.
> 
>
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showExpec...

I confirmed this was not caused by my changes, and this flakiness was an test
issue and not a bug of Web Audio code.