Invalidate SVG filter display items on subtree modification

Invalidate SVG filter display items on subtree modification

When an SVG filtered subtree is modified, the SVGFilter display item
needs to be invalidated. For example:
<g filter='url(#filter)'>
    <rect ...>
</g>

This example will generate a display item list of a single SVGFilter
display item for the <g> because the <rect> subtree will be drawn
as an SkPicture as part of the filter. If the <rect> is changed, we
need to invalidate the SVGFilter display item. Without this patch, we
would re-use the (stale) cached SVGFilter display item.

With this change, we should be able to remove the filter caching
entirely in a followup and rely on the display list.

BUG=451606
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=195645

[pdr., 2015-05-20 07:40:05]

pdr@chromium.org changed reviewers:
+ chrishtr@chromium.org, fs@opera.com

[fs, 2015-05-20 09:29:38]

https://codereview.chromium.org/1148763002/diff/20001/Source/core/layout/svg/...
File Source/core/layout/svg/LayoutSVGResourceFilter.cpp (right):

https://codereview.chromium.org/1148763002/diff/20001/Source/core/layout/svg/...
Source/core/layout/svg/LayoutSVGResourceFilter.cpp:98: invalidation =
PaintInvalidation;
So this means that boundaries are not marked as dirty? Seems fair when the
filter wasn't mutated, but can we discern that case here?

[pdr., 2015-05-20 21:09:27]

PTAL

https://codereview.chromium.org/1148763002/diff/20001/Source/core/layout/svg/...
File Source/core/layout/svg/LayoutSVGResourceFilter.cpp (right):

https://codereview.chromium.org/1148763002/diff/20001/Source/core/layout/svg/...
Source/core/layout/svg/LayoutSVGResourceFilter.cpp:98: invalidation =
PaintInvalidation;
On 2015/05/20 at 09:29:38, fs wrote:
> So this means that boundaries are not marked as dirty? Seems fair when the
filter wasn't mutated, but can we discern that case here?

Good catch (not sure why there weren't test failures). I've updated the patch so
the boundaries invalidation is preserved.

https://codereview.chromium.org/1148763002/diff/60001/LayoutTests/svg/filters...
File LayoutTests/svg/filters/filter-refresh.svg (right):

https://codereview.chromium.org/1148763002/diff/60001/LayoutTests/svg/filters...
LayoutTests/svg/filters/filter-refresh.svg:15: <set attributeName="x" to="0" />
Now that we are tracking invalidations, this flakiness caused by this animation
was exposed (this caused the debug failure). This change shouldn't actually
change this test.

[fs, 2015-05-20 21:57:08]

lgtm

https://codereview.chromium.org/1148763002/diff/60001/LayoutTests/svg/filters...
File LayoutTests/svg/filters/filter-refresh.svg (right):

https://codereview.chromium.org/1148763002/diff/60001/LayoutTests/svg/filters...
LayoutTests/svg/filters/filter-refresh.svg:15: <set attributeName="x" to="0" />
On 2015/05/20 21:09:27, pdr wrote:
> Now that we are tracking invalidations, this flakiness caused by this
animation
> was exposed (this caused the debug failure). This change shouldn't actually
> change this test.

Acknowledged.

[pdr., 2015-05-20 21:58:41]

The CQ bit was checked by pdr@chromium.org

[commit-bot: I haz the power, 2015-05-20 21:59:16]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1148763002/60001

[commit-bot: I haz the power, 2015-05-20 23:32:59]

Committed patchset #4 (id:60001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=195645