1202317 - Firefox slow with degenerate SVG filter

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: test.svg

https://information-node.net/illustration1.svg renders very slowly in Firefox but not in Chrome. The problem seems to be the SVG filter, which is very complex (55 primitives). The filter is applied to just one <path> element, which is fill:none and stroke:url(#pattern6816-3) ... but <pattern id="pattern6816-3"> is empty (though we can't just optimize away filter processing for empty elements).

The filter contains, among other things, 7 <feComposite operator="arithmetic"> with all-zero k values (i.e. transparent black), and 14 <feColorMatrix type="saturate"> with values="1" (i.e. noop).

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: MozReview Request: Bug 1202317. Bail out early if RequestRect is going to do redundant work. r=mstange

Bug 1202317. Bail out early if RequestRect is going to do redundant work. r=mstange

Comment 2

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: MozReview Request: Bug 1202317. Support PodEqual over fixed-length arrays. r=Waldo

Bug 1202317. Support PodEqual over fixed-length arrays. r=Waldo

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: MozReview Request: Bug 1202317. Optimize away identity ColorMatrix nodes. r=mstange

Bug 1202317. Optimize away identity ColorMatrix nodes. r=mstange

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: MozReview Request: Bug 1202317. Optimize feComposite type="arithmetic" with all-zero k values to transparent black. r=mstange

Bug 1202317. Optimize feComposite type="arithmetic" with all-zero k values to transparent black. r=mstange

Comment 5

[Robert O'Callahan (:roc) (email my personal email if necessary)]

The first patch is the real fix of course.

Comment 6

[Markus Stange [:mstange]]

Comment on attachment 8657673 [details]
MozReview Request: Bug 1202317. Bail out early if RequestRect is going to do redundant work. r=mstange

https://reviewboard.mozilla.org/r/18395/#review16567

Comment 7

[Markus Stange [:mstange]]

Comment on attachment 8657675 [details]
MozReview Request: Bug 1202317. Optimize away identity ColorMatrix nodes. r=mstange

https://reviewboard.mozilla.org/r/18399/#review16569

Comment 8

[Markus Stange [:mstange]]

Comment on attachment 8657676 [details]
MozReview Request: Bug 1202317. Optimize feComposite type="arithmetic" with all-zero k values to transparent black. r=mstange

https://reviewboard.mozilla.org/r/18401/#review16571

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b9c1705e4a6c

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Oooops, pushed without Waldo's review. Oh well. Let's assume he'll r+ it when he's ready :-).

Comment 11

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/e416ad732757
https://hg.mozilla.org/integration/mozilla-inbound/rev/e3a099db7378
https://hg.mozilla.org/integration/mozilla-inbound/rev/76240ee211ca
https://hg.mozilla.org/integration/mozilla-inbound/rev/f088e6dbb1ba

Comment 12

[Jeff Walden [:Waldo]]

Comment on attachment 8657674 [details]
MozReview Request: Bug 1202317. Support PodEqual over fixed-length arrays. r=Waldo

https://reviewboard.mozilla.org/r/18397/#review16679

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/e416ad732757
https://hg.mozilla.org/mozilla-central/rev/e3a099db7378
https://hg.mozilla.org/mozilla-central/rev/76240ee211ca
https://hg.mozilla.org/mozilla-central/rev/f088e6dbb1ba