Firefox 10.0 crashes when visiting a website with a java applet

This is the output on the console when reproducing this bug from the command line and going directly to https://www.netbank.nordea.dk/netbank/index.jsp which makes it crash.

The interesting part is the last line:
Assertion failure: rt->onOwnerThread(), at /build/buildd/firefox-10.0+build1/build-tree/mozilla/js/src/jsapi.cpp:6316

Jarl

This might be related to https://bugzilla.mozilla.org/show_bug.cgi?id=704249

At least it is the same IcedTea version: 1.1.3-1ubuntu1.1

There is another bug report making Firefox crash at the same source code line:
http://code.google.com/p/fbug/issues/detail?id=5202

Here is a practical workaround (downgrade to Firefox 7):

sudo aptitude install firefox=7.0.1+build1+nobinonly-0ubuntu2 firefox-globalmenu=7.0.1+build1+nobinonly-0ubuntu2

Jarl

Status changed to 'Confirmed' because the bug affects multiple users.

There is the same problem with https://ebanking.danskebank.dk/html/index.html?site=DBNBEN&secsystem=DI
and ALL other danish netbanking sites since they all use nemid as a provider of security.

I can confirm the problem on mine and my girlfriends computers. (Running Firefox 10 and Ubuntu 11.10 both fully updated)

The bug is also present in Firefox 10 on Ubuntu 10.10.

This is not just with banking sites but also all government related websites. Switching to chromium also works as a workaround.

Not just banking and government, but all sites containing a java-applet which needs to be run.

Even the Java-tester at Oracle will crash FF:
http://java.com/en/download/testjava.jsp

I to can confirm that Chromium works with the exact same setup that crashes Firefox 10 and 10.0.1.

Using icedtea-7 solves this (please note that this is only possible in Pangolin)

On Fri, Feb 17, 2012 at 1:50 PM, Nicky <email address hidden> wrote:
> Not just banking and government, but all sites containing a java-applet
> which needs to be run.
>
> Even the Java-tester at Oracle will crash FF:
> http://java.com/en/download/testjava.jsp
>
> I to can confirm that Chromium works with the exact same setup that
> crashes Firefox 10 and 10.0.1.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/927065
>
> Title:
>  Firefox 10.0 crashes when visiting a website with a java applet
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/firefox/+bug/927065/+subscriptions

--
Hans Harhoff Andersen
PhD-student - Ion Trap Group
Department of Physics and Astronomy
University of Aarhus
Ny Munkegade 120, office 1525-224
8000 Aarhus - Denmark

T  +45 87155213
M +45 28990982

A forum post at http://ubuntudanmark.dk/forum/viewtopic.php?f=24&t=15645 has the following workaround, which solved the problem for me:

 - Type about:config in the adressline
 - Say yes to being cautious
 - Type dom.ipc.plugins.java.enabled into the filterline
 - Change the value to true, by doubleclicking the line
 - Close the tab
 - Restart firefox

Works for me too.

The problem is still present in the latest firefox package 10.0.2+build1-0ubuntu0.11.10.1

But the workaround described in comment #12 works for me.

Still present in 11.0+build1-0ubuntu0.11.10.1

The problem is with the icedtea plugin, so it isn't going to be fixed with a Firefox update

I disagree - yes there may be an error in the icedtea plugin but this is really poor errorhandling in firefox

Assertion failure: rt->onOwnerThread(), at /build/buildd/firefox-11.0+build1/build-tree/mozilla/js/src/jsapi.cpp:6317

Simply catching that error could avoid the browser crashing down alltogether. Perhaps even display some kind of sensible error message - "A plugin has made an error" or something.

And how can icedtea work in v7 of firefox if it is defunct?

This post explains what has changed: http://blog.mozilla.com/luke/2012/01/24/jsruntime-is-now-officially-single-threaded/. This change was reported to the icedtea developers in November btw, so it's not like it should have been a big surprise.

Without the assert, how do return an error through the calling function back to the plugin without changing the semantics of the API? Changing the semantics of the API might trigger other unexpected behaviour inside the icedtea plugin, which would result in a broken plugin and possibly expose other crashes inside plugin code in any case

firefox 11, bug is still there.

I strongly disagree with #16 - dumb assert() instead of normal error handling is firefox fault, not icedtea or whatever plugin to blame.

Max - Read my comments. It's a bug in the icedtea plugin.

MSU - So, returning an unexpected condition back to the plugin, and have it crash anyway (and maybe even in a way which is exploitable) is better than a safe, predictable runtime assert? Ok.......

This website is also affected:

http://www.wegenernet.org/mainpage.phtml (registration necessary...)

The Bug seems to be fixed. At least in Debian.
http://www.wegenernet.org/mainpage.phtml now works.

iceweasel 10.0.3esr-3
Icedtea-plugin 1.2-2
openjdk-6-jre 6b24-1.11.1-3

Juergen.

hello
Can You please check if Namid applets are running fine on Firefox 16.0.1 ?
I've tested some dk websites and they are failing using FF 16.0.1 (from other browsers they are fine).
Thanks.
Michal

I have had no problems for a very long time.