ImageBufferBackingStoreCache should use DeferrableOneShotTimer

• Component: Platform
• Status: RESOLVED FIXED

[Tim Horton]

Right now, it uses plain old Timer, and reschedules it in every deallocate(), which can be very (very!) expensive (I'm seeing it at about 8% of main thread run time in my simple test case). That's why we have DeferrableOneShotTimer.

I have a patch.

[Tim Horton]

Created attachment 219905 [details]
patch

[Tim Horton]

Comment on attachment 219905 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=219905&action=review

> Source/WebCore/platform/graphics/cg/ImageBufferBackingStoreCache.cpp:78
> +    {
> +    }

unindented obviously

[Myles C. Maxfield]

Comment on attachment 219905 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=219905&action=review

> Source/WebCore/platform/graphics/cg/ImageBufferBackingStoreCache.cpp:32
> +static const double purgeInterval = 5;

shouldn't this be kPurgeInterval?

> Source/WebCore/platform/graphics/cg/ImageBufferBackingStoreCache.cpp:74
> +ImageBufferBackingStoreCache::ImageBufferBackingStoreCache()

Why did you move this into public?

[Tim Horton]

(In reply to comment #3)
> (From update of attachment 219905 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=219905&action=review
> 
> > Source/WebCore/platform/graphics/cg/ImageBufferBackingStoreCache.cpp:32
> > +static const double purgeInterval = 5;
> 
> shouldn't this be kPurgeInterval?

I think you'll find that that's generally not the case, nor do I see anything in the style guide.

> > Source/WebCore/platform/graphics/cg/ImageBufferBackingStoreCache.cpp:74
> > +ImageBufferBackingStoreCache::ImageBufferBackingStoreCache()
> 
> Why did you move this into public?

I didn't.

[Tim Horton]

http://trac.webkit.org/changeset/161239