MediaStream tests need to call WebHeap::collectGarbage

MediaStream tests need to call WebHeap::collectGarbage

Now that Blink-side objects about MediaStream are moved to oilpan's heap, we need to call WebHeap::collectGarbage in MediaStream tests to test that a particular object is gone. This CL adds necessary WebHeap::collectGarbage calls to content_unittests about MediaStreams.

Committed: https://crrev.com/106aa0ab544b4d0779bc975da79d4bf9b5d70552
Cr-Commit-Position: refs/heads/master@{#294099}

[haraken, 2014-09-09 04:37:44]

haraken@chromium.org changed reviewers:
+ oilpan-reviews@chromium.org

[haraken, 2014-09-09 04:51:16]

PTAL

https://codereview.chromium.org/549153002/diff/20001/content/renderer/media/w...
File content/renderer/media/webrtc_local_audio_track_unittest.cc (right):

https://codereview.chromium.org/549153002/diff/20001/content/renderer/media/w...
content/renderer/media/webrtc_local_audio_track_unittest.cc:195:
blink_source_.reset();

The reason we need to reset blink_source_ here is complicated.

The reset is needed to destruct capturer_source_ before destructing the
WebRtcLocalAudioTrackTest object. If we don't call the reset, the
capturer_source_ is destructed in a GC triggered after destructing the
WebRtcLocalAudioTrackTest object. This is too late because the testing system
complains that "the capturer_source_ was leaking when the
WebRtcLocalAudioTrackTest finished". Thus we need the reset.

The reason the capturer_source_ is not destructed soon is because the
WebRtcLocalAudioTrackTest object has a strong reference to the capturer_source_
as follows:

WebRtcLocalAudioTrackTest ==(part of object)==> WebMediaStreamSource
==(persistent handle)==> MediaStreamSource ==(OwnPtr)==> ExtraData ==(inherited
by)==> MediaStreamAudioSource ==(scoped_refptr)==> WebRtcAudioCapturer
==(scoped_refptr)==> MockCapturerSource.

Before oilpan, this long strong reference chain was broken promptly when the
WebRtcLocalAudioTrackTest gets destructed. On the other hand, after oilpan, the
chain is not broken until a next GC happens after the WebRtcLocalAudioTrackTest
gets destructed (if we don't call the explicit reset() in TearDown and
explicitly break the chain).

For the same reason, this CL adds a couple of reset() calls in TearDown(). (The
reset() calls in other tests are more complicated to explain why they are
needed. The chain is very long.)

[haraken, 2014-09-09 04:51:38]

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

[haraken, 2014-09-09 04:51:39]

tommi@: Would you take a look at this?

[haraken, 2014-09-10 01:49:01]

haraken@chromium.org changed reviewers:
+ dalecurtis@chromium.org, scherkus@chromium.org

[haraken, 2014-09-10 01:49:02]

dalecurtis, scherkus: Would you take a look at this?

[tommi (sloooow) - chröme, 2014-09-10 01:52:15]

lgtm.  sorry for the delay.

[haraken, 2014-09-10 02:07:55]

The CQ bit was checked by haraken@chromium.org

[commit-bot: I haz the power, 2014-09-10 02:10:11]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/haraken@chromium.org/549153002/40001

[commit-bot: I haz the power, 2014-09-10 04:12:34]

Committed patchset #3 (id:40001) as b248ac6ba0fad2eaeca43e26c7fd032b690e858e

[commit-bot: I haz the power, 2014-09-10 04:15:18]

Patchset 3 (id:??) landed as
https://crrev.com/106aa0ab544b4d0779bc975da79d4bf9b5d70552
Cr-Commit-Position: refs/heads/master@{#294099}