Saving an extra backing store if the contents fill the entire border rect.

Saving an extra backing store if the contents fill the entire border rect.

Added an extra condition in CompositedLayerMapping::containsPaintedContent to check if the contents of the layer fill the entire border rect then return true so that the background colour can be painted into the same layer instead of allocating a backing store and painting in it.

BUG=

[patro, 2014-02-18 12:18:35]

Please Review

Thanks,
Shyam

[vivekg, 2014-02-18 12:30:39]

Not sure about the actual logic but a small code nit.

https://codereview.chromium.org/170423002/diff/1/Source/core/rendering/Compos...
File Source/core/rendering/CompositedLayerMapping.cpp (right):

https://codereview.chromium.org/170423002/diff/1/Source/core/rendering/Compos...
Source/core/rendering/CompositedLayerMapping.cpp:1618: 
How about reordering like here? 

if (renderer()->isVideo() || 
    renderer()->isImage() || 
    (renderer()->isCanvas() && isAcceleratedCanvas(renderer()))) {
    if (contentsBox() == toRenderBox(renderer())->borderBoxRect())
        return true;
}

[abarth-chromium, 2014-02-18 17:46:06]

Can you share some data about the magnitude of the benefit of this optimization.
 I'm learning about this code and am curious.  Thanks!

[patro, 2014-02-24 12:56:38]

Hi,
Can you please tell me what kind of data are you looking for.

Thanks

On 2014/02/18 17:46:06, abarth wrote:
> Can you share some data about the magnitude of the benefit of this
optimization.
>  I'm learning about this code and am curious.  Thanks!

[Ian Vollick, 2014-03-01 02:04:01]

On 2014/02/24 12:56:38, shyamp wrote:
> Hi,
> Can you please tell me what kind of data are you looking for.
> 
> Thanks
> 
> On 2014/02/18 17:46:06, abarth wrote:
> > Can you share some data about the magnitude of the benefit of this
> optimization.
> >  I'm learning about this code and am curious.  Thanks!

Do you have an example of a page that's faster with this optimization? As for
quantifying the performance differences, you might try looking at about:tracing
or telemetry (http://www.chromium.org/developers/telemetry).

[abarth-chromium, 2014-03-01 07:03:12]

not lgtm without data show that this change is a performance improvement.

This code is already quite complicated.  We should strive to make it simpler
rather than more complicated.  If there's a big performance win here, then it
might be worth making the code more complicated.