[RAL] Reduce SVG tree walks

[RAL] Reduce SVG tree walks

This is based on an earlier patch from <fs@opera.com>;.

The performance regression mostly came from excessive tree walks
on SVG_SvgCubics (with only a tiny part of it came also from
extra invalidations).

This change re-introduces what
SVGRenderSupport::checkForSVGRepaintingDuringLayout was doing to
prevent invalidation by checking if your parent had a transform
change, except that we use it to avoid entire subtree walks.

On my Z620, this changes the results of SVG_SvgCubics as follows:
- RAL without patch: 16 ms (std: 0.5)
- RAL with patch: 8.8 ms (std: 0.4)
- Pre-RAL: 9.2 ms (std: 0.3)

BUG=381138
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175983

[Julien - ping for review, 2014-06-10 23:28:13]

Performance fix. The test changes are invalidation removal (checked visually
that the changes are fine).

[leviw_travelin_and_unemployed, 2014-06-10 23:32:26]

lgtm

https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
File Source/core/rendering/svg/SVGRenderSupport.h (right):

https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
Source/core/rendering/svg/SVGRenderSupport.h:75: static bool
parentTransformDidChange(RenderObject*);
Why RenderObject and not RenderSVGModelObject? I guess it's consistent with the
rest of the code, it's just weird to me.

[pdr., 2014-06-11 03:09:24]

+cc fs@opera who may have further comments.

LGTM

https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
File Source/core/rendering/svg/SVGRenderSupport.h (right):

https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
Source/core/rendering/svg/SVGRenderSupport.h:75: static bool
parentTransformDidChange(RenderObject*);
On 2014/06/10 23:32:26, leviw wrote:
> Why RenderObject and not RenderSVGModelObject? I guess it's consistent with
the
> rest of the code, it's just weird to me.

Sometimes svg..
https://docs.google.com/a/chromium.org/drawings/d/1nayJX5XcclYj_-blbVuakVNjVW...

[leviw_travelin_and_unemployed, 2014-06-11 06:51:30]

On 2014/06/11 03:09:24, pdr wrote:
> +cc fs@opera who may have further comments.
> 
> LGTM
> 
>
https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
> File Source/core/rendering/svg/SVGRenderSupport.h (right):
> 
>
https://codereview.chromium.org/327093003/diff/1/Source/core/rendering/svg/SV...
> Source/core/rendering/svg/SVGRenderSupport.h:75: static bool
> parentTransformDidChange(RenderObject*);
> On 2014/06/10 23:32:26, leviw wrote:
> > Why RenderObject and not RenderSVGModelObject? I guess it's consistent with
> the
> > rest of the code, it's just weird to me.
> 
> Sometimes svg..
>
https://docs.google.com/a/chromium.org/drawings/d/1nayJX5XcclYj_-blbVuakVNjVW...

That is a terrifying graph. Thank you for clearing it up :(

[fs, 2014-06-11 10:32:20]

LGTM. Thanks for picking it up.

[Julien - ping for review, 2014-06-11 15:56:43]

The CQ bit was checked by jchaffraix@chromium.org

[commit-bot: I haz the power, 2014-06-11 15:57:40]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jchaffraix@chromium.org/327093003/1

[commit-bot: I haz the power, 2014-06-11 15:57:48]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-06-11 15:57:49]

Failed to apply patch for Source/core/rendering/svg/SVGRenderSupport.cpp:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file Source/core/rendering/svg/SVGRenderSupport.cpp
  Hunk #1 FAILED at 94.
  1 out of 1 hunk FAILED -- saving rejects to file
Source/core/rendering/svg/SVGRenderSupport.cpp.rej

Patch:       Source/core/rendering/svg/SVGRenderSupport.cpp
Index: Source/core/rendering/svg/SVGRenderSupport.cpp
diff --git a/Source/core/rendering/svg/SVGRenderSupport.cpp
b/Source/core/rendering/svg/SVGRenderSupport.cpp
index
722c7164e069689af4adc9f5fab26a3a1da5cd91..95a45f1ce22d205cdc781734f7c997c58ce20fa7
100644
--- a/Source/core/rendering/svg/SVGRenderSupport.cpp
+++ b/Source/core/rendering/svg/SVGRenderSupport.cpp
@@ -94,16 +94,22 @@ const RenderObject*
SVGRenderSupport::pushMappingToContainer(const RenderObject*
     return parent;
 }
 
-bool SVGRenderSupport::checkForSVGRepaintDuringLayout(RenderObject* object)
+bool SVGRenderSupport::parentTransformDidChange(RenderObject* object)
 {
-    if (!object->checkForRepaintDuringLayout())
-        return false;
     // When a parent container is transformed in SVG, all children will be
painted automatically
     // so we are able to skip redundant repaint checks.
     RenderObject* parent = object->parent();
     return !(parent && parent->isSVGContainer() &&
toRenderSVGContainer(parent)->didTransformToRootUpdate());
 }
 
+bool SVGRenderSupport::checkForSVGRepaintDuringLayout(RenderObject* object)
+{
+    if (!object->checkForRepaintDuringLayout())
+        return false;
+
+    return parentTransformDidChange(object);
+}
+
 bool SVGRenderSupport::paintInfoIntersectsRepaintRect(const FloatRect&
localRepaintRect, const AffineTransform& localTransform, const PaintInfo&
paintInfo)
 {
     return localTransform.mapRect(localRepaintRect).intersects(paintInfo.rect);

[Julien - ping for review, 2014-06-11 16:43:11]

The CQ bit was checked by jchaffraix@chromium.org

[commit-bot: I haz the power, 2014-06-11 16:43:24]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jchaffraix@chromium.org/327093003/10001

[commit-bot: I haz the power, 2014-06-11 18:28:59]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  mac_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/mac_blink_rel/builds/11105)
  win_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/win_blink_rel/builds/11588)

[commit-bot: I haz the power, 2014-06-11 19:11:30]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-06-11 19:11:31]

Try jobs failed on following builders:
  mac_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/mac_blink_rel/builds/11114)

[Julien - ping for review, 2014-06-11 19:18:44]

The CQ bit was checked by jchaffraix@chromium.org

[commit-bot: I haz the power, 2014-06-11 19:19:12]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jchaffraix@chromium.org/327093003/30001

[commit-bot: I haz the power, 2014-06-12 05:26:24]

Change committed as 175983