1982914 - wpt reftest svg/text/reftests/text-context-fill.svg fatally asserts on new android 14 emulator image- 2 (" MOZ_ASSERT(!OuterSVGIsCallingReflowSVG(aFrame)) (Do not call under ISVGDisplayableFrame::ReflowSVG!) [@ mozilla::SVGUtils::ScheduleReflowSVG]")

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

we have 2 failures:

1. css/css-sizing/aspect-ratio/replaced-element-003.html (some letters have pixels, but hard to see with the naked eye): reftest-compare
2. /svg/text/reftests/text-context-fill.svg - crash on debug, specifically at this asset: MOZ_ASSERT(!OuterSVGIsCallingReflowSVG(aFrame)) (Do not call under ISVGDisplayableFrame::ReflowSVG!) [@ mozilla::SVGUtils::ScheduleReflowSVG]

Comment 1

[Boris Chiou [:boris]]

Hi Robert, it seems we hit an assertion on Android. The test file was added by you before. Could you please take a look at this crash? Thanks.

Comment 2

[Boris Chiou [:boris]]

aspect-ratio/replaced-element-003.html has been mentioned in Bug 1982907, so remove the dependency and perhaps we should focus on SVG part in this bug.

Comment 3

[Daniel Holbert [:dholbert]]

(In reply to Joel Maher ( :jmaher ) (UTC -8) from comment #0)

1. css/css-sizing/aspect-ratio/replaced-element-003.html (some letters have pixels, but hard to see with the naked eye): reftest-compare

(In reply to Boris Chiou [:boris] from comment #2)

aspect-ratio/replaced-element-003.html has been mentioned in Bug 1982907, so remove the dependency and perhaps we should focus on SVG part in this bug.

I think comment 0 just had a copypaste typo: jmaher really meant to paste svg/text/reftests/text-bidi-controls-anchors-2.svg for (1) there (he mentioned letters having fuzzy pixels, and that's the one with fuzzy letters shown in comment 0's reftest-analyzer link).

But nonetheless, it's still quite different from the text-context-fill.svg fatal-assert, so I spun off bug 1982937 to cover the fuzzy failure and we can leave this bug covering the fatal assert.

Comment 4

[Daniel Holbert [:dholbert]]

Looks like bug 1797417 and bug 1825459 are both about this same assertion-failure.

For now, simplest way forward would be to just skip this test on Android 14.

Comment 5

[Robert Longson [:longsonr]]

Attachment: Bug 1982914 - Change the conditions under which we skip ScheduleReflowSVG calls in SVGMarkerObserver::OnRenderingChange r=dholbert

Comment 6

[Robert Longson [:longsonr]]

https://treeherder.mozilla.org/jobs?repo=try&revision=7f8ecd300b0cc5f58a9e3bb7b736ad997d4199ef

Comment 7

[Joel Maher ( :jmaher ) (UTC -8)]

oh, thanks for this, I am pushing to try with android14...
https://treeherder.mozilla.org/jobs?repo=try&revision=e22e41c0ba2d6deda74b6894a2e35ed9f9e5fa9f

Comment 8

[Joel Maher ( :jmaher ) (UTC -8)]

this seems to be fixed :)

Comment 9

[Robert Longson [:longsonr]]

My inclination is to wait till the code freeze is over before landing this. Do you need it landed sooner Joel?

Comment 10

[Joel Maher ( :jmaher ) (UTC -8)]

next week is fine!

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/a3031bf4e67e
https://hg.mozilla.org/integration/autoland/rev/d367c1ff8d78
Change the conditions under which we skip ScheduleReflowSVG calls in SVGMarkerObserver::OnRenderingChange r=dholbert,firefox-svg-reviewers

Comment 12

[pstanciu]

https://hg.mozilla.org/mozilla-central/rev/d367c1ff8d78