595564 - [JaegerMonkey] Add-On LastPass hangs Fx4 with a script error.

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect)

Description

[Gary [:streetwolf52]]

User-Agent:       Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b6pre) Gecko/20100911 Firefox/4.0b6pre
Build Identifier: 20100911124454

If I set 'javascript.options.methodjit.chrome' to True starting up Fx4 causes a 'hang'(which might be why it defaults to False).  I'll eventually get a message that a script has stopped working.  I then stop the script and Fx4 comes up.  However, LastPass is not functional.  

LastPass @ https://addons.mozilla.org/en-US/firefox/addon/8542/

Reproducible: Always

Comment 1

[Gary [:streetwolf52]]

The problem appears to be caused by another Add-on; FEBE.  This is the message I get when FEBE is enabled.

A script on this page may be busy, or it may have stopped responding. You can stop the script now, or you can continue to see if the script will complete.

Script: XStringBundle:20

Comment 2

[Gary [:streetwolf52]]

While it appears FEBE causes the hang and script error, LastPass is still unusable even with FEBE disabled.

Comment 3

[Gary [:streetwolf52]]

Is there still work being done on whatever setting 'javascript.options.methodjit.chrome' to True does?  I would like to know if add-ons will need to be changed if and when this option is turned on by default.

Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b7pre) Gecko/20100919 Firefox/4.0b7pre - Build ID: 20100919042023

Comment 4

[Brendan Eich [:brendan]]

Gary, you don't need to ask -- the method JIT is new and bugs will be fixed. Any failure to follow JS standards (de-facto or de-jure) that the engine follows w/o the method JIT are bugs. Chrome being bit first does not mean these bugs won't bite content. So in general, they're serious bugs and they'll get fixed, without add-on authors having to change their code to work around.

Of course real software ships with bugs unfixed, whether discovered and diagnosed or not, so they may be some bugs to work around, but that would be something to avoid, and release-noted if known when shipped, else determined after the fact by add-on devs.

So while bugs happen in shipped products, this bug and others are all in beta software that we're trying to get ship-shape. The trick here, as in the D-Link router HTML/JS bug, is to help get reliable STR, and then reduced testcases. That is a good way to ensure a fix soon, before Firefox 4 is close to shipping.

/be

Comment 5

[Brendan Eich [:brendan]]

If we had some reason to believe our chrome JS exposed method JIT bugs that did not bite content on the web, say bugs peculiar to JS1.7 and up, then we might keep the method JIT turned off for chrome in Firefox 4. Then it would be worth filing bugs still, but you'd have to keep javascript.options.methodjit.chrome set to false, or else indeed develop workarounds for your add-ons.

But apart from JS1.7+ extensions popular in chrome scripts, we want to find and fix cross-browser methodjit bugs, so it would help to distinguish bugs biting chrome by whether the relevant JS seems to use specific Mozilla JS versions or features.

/be

Comment 6

[David Mandelin [:dmandelin]]

Gary, can you try again now? I just installed LastPass and FEBE, and LastPass seems to work fine. This was in a Tracemonkey nightly from Friday or so.

Comment 7

[Gary [:streetwolf52]]

Working Ok.

Comment 8

[David Mandelin [:dmandelin]]

(In reply to comment #7)
> Working Ok.

Thanks!