430699 - crash in jpinscp.dll (JavaScript calling an applet causes NS_ERROR_FAILURE)

Status: RESOLVED DUPLICATE of bug 405357

(Core :: JavaScript Engine, defect)

Description

[Jason Hunter]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.1.13) Gecko/20080311 Firefox/2.0.0.13
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.4; en-US; rv:1.9b5) Gecko/2008032619 Firefox/3.0b5

If I use FF3 to call from JavaScript into an applet I see this error:

[Exception... "Failure" nsresult: "0x80004005 (NS_ERROR_FAILURE)" location: "JS frame :: http://servlets.com/ff3/test.html ::  :: line 5" data: no] 

This doesn't happen with FF2.  I've isolated a simple reproducible test case in test.html.

Reproducible: Always

Steps to Reproduce:
1. Visit http://servlets.com/ff3/test.html
2. See the exception caught and written into the page.

Comment 1

[Ron]

Wow - just tried your test page under Windows XP and my results were even worse. In my case it CRASHED Firefox hard.

As soon as I entered the URL and hit return, Firefox crashed and displayed a message saying so, and asking if it could automatically report the bug, which I allowed it to do.  I tried this both with FF 3 beta 3 and FF 3 beta 5, both with the same exact result.

This is definitely a show stopper IMHO until its is resolved.  It will definitely cause MAJOR problems for our sites and I'm sure many others that use JavaScript to Java communication.

Thanks in advance for looking into this!

Comment 2

[Jason Hunter]

Updating Hardware, OS, and Severity based on Ron's comment.

I don't know if it's truly All but it's at least Mac + Windows.

Comment 3

[Bob Uliss]

Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.5; en-US; rv:1.9pre) Gecko/2008042904 Minefield/3.0pre


Works For Me with this build
Steps to Reproduce:

1. Visit http://servlets.com/ff3/test.html
Result: I get a white page
with 1 line:

Apple inc.

perhaps it is INTEL?

Comment 4

[Jason Hunter]

My original report was on PPC actually.  "Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.4; en-US; rv:1.9b5) Gecko/2008032619 Firefox/3.0b5".  I simply filed the bug from my laptop which is Intel.  I do notice you have a month later build than mine.

Comment 5

[Ron]

A few days ago I tried with the nightly build under Windows and it is still crashing. Sounds like progress has been made on the Mac front, so that is good. But certainly Windows remains a critical issue as we cannot have browsers crashing just by visiting a page. As mentioned previously this does not happen with any FF 2.x build, only in the 3.x branch. Please advise. Thank you!

Comment 6

[Mike Beltzner [:beltzner, not reading bugmail]]

--> Core::JS

Comment 7

[Mike Beltzner [:beltzner, not reading bugmail]]

Carrying over blocking nomination, cc'ing some likely suspects.

Comment 8

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

Do we have crash reporter URLs or a regression window for this?

Comment 9

[Mike Beltzner [:beltzner, not reading bugmail]]

FWIW, I can confirm the crash on Windows only, and only if the user has a JVM installed. The crash takes down Firefox without it sending a crash report.

Here's something interesting, though - sometimes if I start a new session or clear private data, I see:

java.lang.NullPointerException
	at sun.plugin.AppletViewer.loadJarFiles(Unknown Source)
	at sun.applet.AppletPanel.runLoader(Unknown Source)
	at sun.applet.AppletPanel.run(Unknown Source)
	at java.lang.Thread.run(Unknown Source)

in the JVM error console and:

Sun Microsystems Inc. 

in the content area of the page.

Not sure what would cause that.

Comment 10

[Mike Beltzner [:beltzner, not reading bugmail]]

It seems to be related, FWIW, to whether or not there are JVM icons left dangling in my system tray - that makes me believe that the crash is that the JVM not shutting down properly after catching the null pointer exception, and trying to start up a new JVM instance leading to a crash in Firefox.

Do we have an example of JS calling a servlet without a null pointer exception?

Comment 11

[Chris Pearce [:cpearce (Not reading bugmail)]]

I get the crash on Vista, in jpinscp.dll, which is a DLL of Sun's Java VM. I have jre1.6.0.02.

Comment 12

[Carsten Book [:Tomcat]]

i will search for the regression range.

Comment 13

[Mike Beltzner [:beltzner, not reading bugmail]]

tomcat: please renom if you get a regression range; thx!

Comment 14

[Brendan Eich [:brendan]]

(In reply to comment #13)
> tomcat: please renom if you get a regression range; thx!

Why is a blocker nomination approval dependent on finding a regression range? It would seem to depend on severity only, from first principles.

/be

Comment 15

[Carsten Book [:Tomcat]]

(In reply to comment #8)
> Do we have crash reporter URLs or a regression window for this?
> 

Crashreport is here : http://crash-stats.mozilla.com/report/index/fa49e4a3-1b9c-11dd-9c2c-0013211cbf8a?p=1

Could this be a dupe of Bug 405357 ?

Comment 16

[Carsten Book [:Tomcat]]

Duping to Bug 405357 because i did tests with this bug and Bug 405357 and got the identical crash stacks. 

Also the tryserver build from jst https://build.mozilla.org/tryserver-builds/2008-05-05_16:[e-mail]-plugin-fun/ fixes the crash for me :)

Comment 17

[Ron]

I just wanted to drop a quick line to let everyone know that I tested this fix with the May 07 nightly and it appears to have indeed resolved it!! Thank you to everyone for your help in resolving this important bug.