[Chromium] Use ThrottledTextureUploader with threaded compositing.

• Component: WebKit Misc.
• Status: RESOLVED FIXED

[David Reveman]

Can land as soon as http://code.google.com/p/chromium/issues/detail?id=123414 is fixed.

[David Reveman]

Created attachment 140638 [details]
Patch

[James Robinson]

Comment on attachment 140638 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=140638&action=review

> Source/WebCore/platform/graphics/chromium/cc/CCThreadProxy.cpp:808
> +    OwnPtr<TextureUploader> uploader = ThrottledTextureUploader::create(m_contextBeforeInitializationOnImplThread.get());

seems odd to create this uploader before we check if initialization is going to succeed or not. Is this ordering intentional - i.e. do we need an uploader to make it through initialization?

[David Reveman]

(In reply to comment #2)
> (From update of attachment 140638 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=140638&action=review
> 
> > Source/WebCore/platform/graphics/chromium/cc/CCThreadProxy.cpp:808
> > +    OwnPtr<TextureUploader> uploader = ThrottledTextureUploader::create(m_contextBeforeInitializationOnImplThread.get());
> 
> seems odd to create this uploader before we check if initialization is going to succeed or not. Is this ordering intentional - i.e. do we need an uploader to make it through initialization?

Good point. This is odd and need to be fixed. We don't need the uploader to make it through initialization but I also find it a bit odd to give the LRC an uploader instance post initialization that has certain requirements for when to be destroyed.

I think the best solution is to do land this change: https://bugs.webkit.org/show_bug.cgi?id=85893 which makes the change required to enable the throttled uploader:
s/layerRenderer()->unthrottledTextureUploader()/layerRenderer()->throttledTextureUploader()/ in CCThreadProxy::scheduledActionUpdateMoreResources().

[David Reveman]

Created attachment 140737 [details]
Patch

[David Reveman]

Created attachment 141478 [details]
Patch

Rebase

[Adrienne Walker]

Comment on attachment 141478 [details]
Patch

R=me, since the Chromium side fix is in now and the unit tests all passed.

[David Reveman]

Comment on attachment 141478 [details]
Patch

Finished the testing I wanted to do. I think this is ready to land.

[WebKit Review Bot]

Comment on attachment 141478 [details]
Patch

Clearing flags on attachment: 141478

Committed r117312: <http://trac.webkit.org/changeset/117312>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.