462031 - CPU pegged, large memory use on page w/ big table and heavy JavaScript

Status: RESOLVED INCOMPLETE

(Firefox :: General, defect)

Description

[Ryan Tomayko]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b2pre) Gecko/20081025 Minefield/3.1b2pre
Build Identifier: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b2pre) Gecko/20081025 Minefield/3.1b2pre




[e-mail]

Reproducible: Always

Steps to Reproduce:
1. http://staging.translate.songbirdnest.com/languages/ca
2. Log in with user: [e-mail], pass: foobarbaz
3. Click within one of the table cells under the "Translation" header.
4. Enter some text in the edit box.
5. Hit <ENTER> to move to the next cell, repeat #4.
Actual Results:  
Pegs CPU and large memory fluctuation. Browser becomes responsive after 20-60 seconds. Process can be repeated.

Expected Results:  
Stay responsive.

This seems to be some kind of perf regression in Minefield. We're not able to reproduce under stable/official versions of Firefox 3.0. We have been able to reliably reproduce the issue with recent Minefield builds on Windows XP, Windows Vista, and MacOS 10.5 with and without the new jit features enabled.

Comment 1

[Ryan Tomayko]

(Oops. Sorry about the summary and description in the original report. Ignore the [e-mail] email address.)

We're seeing what looks like a perf regression in recent Minefield builds on the URL reported. The browser freezes with 100% CPU usage after editing a few translations. Scrolling the page up and down can also cause the browser to freeze.

Comment 2

[Wayne Mery (:wsmwk)]

Ryan, can you reproduce with http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/ ?

Comment 3

[Tyler Downer [He/Him]]

Ryan, any updates? Does it happen with 3.5.1 final or later?, In Safe Mode or a new profile?

Comment 4

[Wayne Mery (:wsmwk)]

no response and I can't reproduce, so => incomplete