Reland of Fix slow repaint object registration

Reland of Fix slow repaint object registration (patchset #1 id:1 of https://codereview.chromium.org/1224083019/)

Reason for revert:
This patch was not the cause, relanding.

Original issue's description:
> Revert of Fix slow repaint object registration (patchset #2 id:20001 of https://codereview.chromium.org/1240443003/)
> 
> Reason for revert:
> Speculative revert: appears to have broken Linux MSAN bots.
> 
> http://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20MSAN/builds/4413
> 
> 
> Original issue's description:
> > Fix slow repaint object registration
> > 
> > The LayoutView now paints background, so the special condition to accelerate
> > frame scrolling with fixed root background should be updated as well.
> > 
> > As a drive-by fix, this CL also corrects slow repaint object accounting so
> > we no longer leak slow repaint object count.
> > 
> > BUG=506002
> > 
> > Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=198894
> 
> TBR=vollick@chromium.org,trchen@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=506002
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=198913

TBR=vollick@chromium.org,trchen@chromium.org,rtoy@chromium.org,yosin@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=506002
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=198920

[Noel Gordon, 2015-07-15 04:22:51]

Created Reland of Fix slow repaint object registration

[commit-bot: I haz the power, 2015-07-15 04:23:12]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1234153003/1

[commit-bot: I haz the power, 2015-07-15 04:23:55]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=198920

[trchen, 2015-07-15 06:16:16]

Really? The MSAN bots failure message looks exactly due to my patch, and I did
forget to initialize a member variable I added. Anyway let me know if the bots
failed again, I have a revised patch ready for submitting.

[Noel Gordon, 2015-07-15 07:09:35]

On 2015/07/15 06:16:16, trchen wrote:
> Really? The MSAN bots failure message looks exactly due to my patch, and I did
> forget to initialize a member variable I added. Anyway let me know if the bots
> failed again, I have a revised patch ready for submitting.

I'm still wating for the MSAN to come green.  But the bot did not go green after
I reverted your patch, so I relanded it.  Maybe I did not wait long enoigh.

Anyhow, you should fix your patch, of course, if you saw an unitialized and
submit it pronto.

I reverted another patch that I think might have caused the problem, namely,

https://codereview.chromium.org/1235163007

and still waiting for the bot to go green.

[Noel Gordon, 2015-07-15 07:23:29]

On 2015/07/15 07:09:35, noel gordon wrote:

> and still waiting for the bot to go green.

but webkit rolls are blocked, so it might take me some time to know for sure.

So ... will you be fixing you patch soon?  Should I just revert it?

[Noel Gordon, 2015-07-15 08:34:27]

Ah well, reverting is the safer option.

[Noel Gordon, 2015-07-15 08:35:44]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1234183003/ by noel@chromium.org.

The reason for reverting is: Speculative revert: appears to have broken Linux
MSAN bots. Author suggested there was in uninited varible use in their code..