MediaStream content_unittests need to trigger a GC before tear down

MediaStream content_unittests need to trigger a GC before tear down

Given that MediaStream objects in the Blink side are allocated on Oilpan's heap, we need to trigger a GC to collect MediaStream objects in the Chromium side that depend on the lifetime of the MediaStream objects in the Blink side.

This CL adds Heap::collectAllGarbage to TearDown methods of MediaStream content_unittests. This guarantees that MediaStream objects used in the content_unittests are collected before the content_unittests finish. Otherwise, the LSan complains that the MediaStream objects leak in the content_unittests.

This is an issue that happens only in testing.

BUG=340522
Committed: https://crrev.com/02b15d5a415222f4ffc4a247e191fca76f1fddce
Cr-Commit-Position: refs/heads/master@{#295920}

[haraken, 2014-09-12 04:54:46]

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

[haraken, 2014-09-12 04:54:47]

PTAL

[tommi (sloooow) - chröme, 2014-09-12 18:04:26]

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

[tommi (sloooow) - chröme, 2014-09-12 18:04:27]

Per can you take a look?  I'll be happy to stamp afterwards (I'm at a summit
right now and completely out of time)

[perkj_chrome, 2014-09-15 07:17:53]

lgtm but with a small question. 

Is it necessary to destroy |dependency_factory_|? 

Thanks for doing this.

https://codereview.chromium.org/566793002/diff/1/content/renderer/media/webrt...
File content/renderer/media/webrtc/webrtc_media_stream_adapter_unittest.cc
(right):

https://codereview.chromium.org/566793002/diff/1/content/renderer/media/webrt...
content/renderer/media/webrtc/webrtc_media_stream_adapter_unittest.cc:34:
dependency_factory_.reset();
Why is it necessary to reset |dependency_factory_|? Is this hiding another
problem?

[haraken, 2014-09-22 03:09:14]

Thanks for review!

https://codereview.chromium.org/566793002/diff/1/content/renderer/media/webrt...
File content/renderer/media/webrtc/webrtc_media_stream_adapter_unittest.cc
(right):

https://codereview.chromium.org/566793002/diff/1/content/renderer/media/webrt...
content/renderer/media/webrtc/webrtc_media_stream_adapter_unittest.cc:34:
dependency_factory_.reset();
On 2014/09/15 07:17:53, perkj wrote:
> Why is it necessary to reset |dependency_factory_|? Is this hiding another
> problem?

Good point. I confirmed that resetting dependency_factory_ is not necessary.
Removed.

[haraken, 2014-09-22 03:09:17]

The CQ bit was checked by haraken@chromium.org

[commit-bot: I haz the power, 2014-09-22 03:11:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/566793002/20001

[commit-bot: I haz the power, 2014-09-22 03:51:34]

Committed patchset #2 (id:20001) as b767a289e6d7f516275c84f78aae533f7617f05a

[commit-bot: I haz the power, 2014-09-22 03:52:15]

Patchset 2 (id:??) landed as
https://crrev.com/02b15d5a415222f4ffc4a247e191fca76f1fddce
Cr-Commit-Position: refs/heads/master@{#295920}