Bug 240112

Summary:

REGRESSION (r291788): MotionMark Suits subtest is 9% regressed

Product:

WebKit

Reporter:

Antti Koivisto <koivisto>

Component:

CSS

Assignee:

Antti Koivisto <koivisto>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

cdumez, changseok, dino, esprehn+autocc, ews-watchlist, fmalita, fujii.hironori, glenn, gyuyoung.kim, kangil.han, kondapallykalyan, pdr, rbuis, sabouhallawa, schenney, sergio, zalan, zimmermann

Priority:

Keywords:

InRadar

Version:

WebKit Nightly Build

Hardware:

Unspecified

OS:

Unspecified

See Also:

https://bugs.webkit.org/show_bug.cgi?id=241489

Bug Depends on:

Bug Blocks:

230296

Attachments:

Description	Flags
Patch	none
Patch	none
Patch	none
Patch	none
Patch	none

Antti Koivisto

Reported 2022-05-05 01:08:02 PDT

This is https://browserbench.org/MotionMark1.2/ Individual subtests can be ran via https://browserbench.org/MotionMark1.2/developer.html

Attachments
Patch (10.28 KB, patch) 2022-05-11 01:58 PDT, Antti Koivisto	no flags	Details Formatted Diff Diff
Patch (52.44 KB, patch) 2022-05-11 04:54 PDT, Antti Koivisto	no flags	Details Formatted Diff Diff
Patch (52.72 KB, patch) 2022-05-11 05:21 PDT, Antti Koivisto	no flags	Details Formatted Diff Diff
Patch (52.58 KB, patch) 2022-05-11 05:32 PDT, Antti Koivisto	no flags	Details Formatted Diff Diff
Patch (53.46 KB, patch) 2022-05-11 07:50 PDT, Antti Koivisto	no flags	Details Formatted Diff Diff
Show Obsolete (4) View All Add attachment proposed patch, testcase, etc.

Antti Koivisto

Comment 1 2022-05-05 01:08:27 PDT

rdar://92409900

Nikolas Zimmermann

Comment 2 2022-05-05 13:30:17 PDT

Hmm, that's not a small one... You found this by chance Antti? Or did the patch make you suspicious? :-)

alan baradlay

Comment 3 2022-05-05 13:43:19 PDT

(In reply to Nikolas Zimmermann from comment #2) > Hmm, that's not a small one... You found this by chance Antti? Or did the > patch make you suspicious? :-) We run MM regularly to discover regressions (and progressions).

alan baradlay

Comment 4 2022-05-05 20:06:22 PDT

The Suits subtest is about calling svgElement.setAttribute('transform','translate...". in a repeated manner. r291788 went from "attribute changed + set needsLayout" to "attribute changed + invalidate style" which is fine as long as the attribute is translated to a RenderStyle property, however SVG transform is special and internal to SVGGraphicsElement and it does not require style invalidation. Will post a patch soon.

alan baradlay

Comment 5 2022-05-06 08:38:25 PDT

> Will post a patch soon. or Antti. Probably Antti.

Antti Koivisto

Comment 6 2022-05-11 01:58:40 PDT

Created attachment 459148 [details] Patch

Antti Koivisto

Comment 7 2022-05-11 04:54:23 PDT

Created attachment 459154 [details] Patch

Antti Koivisto

Comment 8 2022-05-11 05:21:31 PDT

Created attachment 459156 [details] Patch