2032262 - 291% regression on sunspider-regexpDna on 15Apr2026

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect, P2)

Description

[Mayank Bansal]

https://treeherder.mozilla.org/perfherder/graphs?highlightAlerts=1&highlightChangelogData=0&highlightCommonAlerts=0&highlightInitialDataPoints=0&replicates=0&selected=5256712,2420582097&series=autoland,5256712,1,13&timerange=5184000&zoom=1776107002197,1776328292410,0.6174860741110413,19.023936525432113

Chnagelong: https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=9de4b85dca4ed45ad46089e3c642fec80e0b9776&tochange=4b6887fd04a864c679a889de539f4a634bc47044

Suspect regressor: Bug 2022634

Comment 1

[Mayank Bansal]

Regression on sunspider-shell makes the regressor clear.
Changelog: https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=adf10b0109f1f953a131650c79f9a95666820bb3&tochange=9a64aadd3871e703101fbed232f65b6b7c5c80c3

Comment 2

[BugBot [:suhaib / :marco]]

Set release status flags based on info from the regressing bug 2022634

Comment 3

[Iain Ireland [:iain]]

Attachment: Bug 2032262: Initialize enable_regexp_unaligned_accesses flag correctly r=dminor

The code in irregexp that uses this was previously dead, but with the changes to compile from bytecode, now it's actually being checked.

When unaligned loads are supported, we can load four characters at once. Without that support, we are limited to one character at a time.

Comment 4

[Iain Ireland [:iain]]

Attachment: Bug 2032262: Fix --trace-regexp-assembler r=dminor

While investigating this, I realized that the update had accidentally broken some of the debug logging code.

I also realized that some of the newer stuff is not hooked up properly yet (eg --trace-regexp-compiler doesn't work), but that will take more work to fix. I'll open a separate bug.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/809f7d970263
https://hg.mozilla.org/integration/autoland/rev/4a216ead032c
Initialize enable_regexp_unaligned_accesses flag correctly r=dminor
https://github.com/mozilla-firefox/firefox/commit/8887c6f20a6e
https://hg.mozilla.org/integration/autoland/rev/afbf63e2c6cd
Fix --trace-regexp-assembler r=dminor

Comment 7

[Mayank Bansal]

The numbers are back to pre-regression.

Comment 8

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/4a216ead032c
https://hg.mozilla.org/mozilla-central/rev/afbf63e2c6cd

Comment 9

[Florin Bilt]

(In reply to Cosmin Sabou [:CosminS] from comment #8)

https://hg.mozilla.org/mozilla-central/rev/4a216ead032c
https://hg.mozilla.org/mozilla-central/rev/afbf63e2c6cd

Perfherder has detected a browsertime performance change from push afbf63e2c6cdd5738e4d803e61c0e011432fe0ba.

No action is required from the author; this comment is provided for informational purposes only.

Improvements	Test	Platform	Options	Absolute values [old vs new]
12%	jetstream3 Sunspider-Average (doc)	windows11-64-24h2-shippable	fission webrender	8.05 ms -> 7.07 ms
11%	jetstream3 Sunspider-Geometric (doc)	linux2404-64-shippable	fission webrender	320.58 score -> 355.19 score
11%	jetstream3 Sunspider-First (doc)	linux2404-64-shippable	fission webrender	17.54 ms -> 15.68 ms
11%	jetstream3 Sunspider-First (doc)	android-hw-a55-14-0-aarch64-shippable	webrender	21.35 ms -> 19.10 ms
10%	jetstream3 Sunspider-Geometric (doc)	android-hw-a55-14-0-aarch64-shippable	fission webrender	258.75 score -> 283.92 score
...	...	...	...	...
7%	jetstream3 UniPoker-Average (doc)	linux2404-64-shippable	fission webrender	10.62 ms -> 9.84 ms

Need Help or Information?

If you have any questions, please reach out to [e-mail]. Alternatively, you can find help on Slack by joining #perf-help, and on Matrix you can find help by joining #perftest.

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests.