Oilpan: HRTF dababase loader thread should be attached to Oilpan

Oilpan: HRTF dababase loader thread should be attached to Oilpan

When the main thread passes a pointer of GarbageCollected object to a
HRTFDatabase thread (via postTask), the HRTFDatabase thread creates
a CrossThreadPersistent handle for the pointer. However, this is not
allowed because the HRTFDatabase thread is not attached to Oilpan.

We should attach the HRTFDatabase thread to Oilpan.

This CL removes MutexLocker from HRTFDatabase's destructor since we don't need to clear m_htrfDatabase in the destructor.

BUG=427303
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=185024

[haraken, 2014-11-10 06:13:34]

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

[haraken, 2014-11-10 06:13:35]

PTAL

This is an alternative fix to https://codereview.chromium.org/670113002/. (This
CL doesn't address the issue in a WebAudio thread though.)

[tkent, 2014-11-10 06:27:55]

lgtm

https://codereview.chromium.org/685153005/diff/1/Source/platform/audio/HRTFDa...
File Source/platform/audio/HRTFDatabaseLoader.cpp (left):

https://codereview.chromium.org/685153005/diff/1/Source/platform/audio/HRTFDa...
Source/platform/audio/HRTFDatabaseLoader.cpp:75: MutexLocker locker(m_lock);
Please mention this change in the CL description.

[haraken, 2014-11-10 06:37:11]

https://codereview.chromium.org/685153005/diff/1/Source/platform/audio/HRTFDa...
File Source/platform/audio/HRTFDatabaseLoader.cpp (left):

https://codereview.chromium.org/685153005/diff/1/Source/platform/audio/HRTFDa...
Source/platform/audio/HRTFDatabaseLoader.cpp:75: MutexLocker locker(m_lock);
On 2014/11/10 06:27:54, tkent wrote:
> Please mention this change in the CL description.

Done.

[haraken, 2014-11-10 06:37:18]

The CQ bit was checked by haraken@chromium.org

[commit-bot: I haz the power, 2014-11-10 06:38:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/685153005/1

[commit-bot: I haz the power, 2014-11-10 07:13:11]

Committed patchset #1 (id:1) as 185024

[sof, 2014-11-11 07:27:03]

sigbjornf@opera.com changed reviewers:
+ sigbjornf@opera.com

[sof, 2014-11-11 07:27:03]

This leads to some webaudio/ crashes when finalizing a HRTFDataBaseLoader, as
finalizing its WebThreadSupportingGC is dependent on entering a safe point,
which we're not at when sweeping --

 
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Linux_Oil...

[sof, 2014-11-12 07:40:56]

On 2014/11/11 07:27:03, sof wrote:
> This leads to some webaudio/ crashes when finalizing a HRTFDataBaseLoader, as
> finalizing its WebThreadSupportingGC is dependent on entering a safe point,
> which we're not at when sweeping --
> 
>  
>
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Linux_Oil...

All well again after http://crrev.com/711323002 (r185177)