387850 - Firefox stops working when invoked java from javascript on Microsoft Windows Server 2008.

Status: RESOLVED INCOMPLETE

(Core Graveyard :: Java: OJI, defect)

Description

[Saujanya]

User-Agent:       Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322; InfoPath.1; .NET CLR 2.0.50727)
Build Identifier: Firefox version 2.0.0.4

Using Firefox if I call Java from a Javascript function (example java.lang.System.getProperty(...)) the browser crashes. OS presents "Firefox has stopped working" dialog. We have observed this behavior on the latest pre-release build of Microsoft Windows Server 2008 (previously called Longhorn).

Also, there is an error event logged under OS System Event Logs as Application Error with Error ID 1000 and Task Category 100.

We guess it might be an issue with liveconnect on this OS.


Reproducible: Always

Steps to Reproduce:
1. Save the following HTML page on your webserver (say test.html):
<HTML>
<HEAD>
<SCRIPT Language="Javascript">
	function testCrash() {
		alert ('Test Point 1');
		var osVer = java.lang.System.getProperty("os.version");
		alert ('Test Point 2 version=' + osVer);
	}
</SCRIPT>
</HEAD>
<BODY onload="testCrash()">
Hi
</BODY>
</HTML>

2. Try to browse to this page from Firefox
(e.g. https://hostname:portnumber/test.html )

3. Observe that only the first alert is displayed, then the browser crashes.
Actual Results:  
Per the reproducible steps provided only the first Alert is displayed, browser crashes after that. You can see the dialog "Firefox has stopped working". The second alert is never displayed.

Expected Results:  
The second alert should be shown with the OS version.

It should not have crashed, but displayed the second alert.
The issue was reproduced on multiple systems.

Additional Information:
OS: Windows Server 2008 x86 Enterprise Edition Build # 6001.16606 IDS3  
(This is a post Beta 3 version of the OS).

Release ID: R159869  (x86)
Browser: Latest Firefox downloaded, version 2.0.0.4
JRE: Sun Java version 6 and also 6 update 2 - it crashes with both.

Note that the same test.html provided in reproducible steps in the issue works fine on other operating system Windows Vista, with the same Firefox and JRE.

Also it is noteworthy that on the same installation on Windows Server 2008, where the above crash is seen, a simple HelloWorld Java applet run successfully without any problem.

Comment 1

[timeless]

please try:
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/firefox-3.0a7pre.en-US.win32.zip and report the breakpad incident id.

Comment 2

[Saujanya]

Attachment: Error Log General tab from System Event Logs.

Error Log General tab from System Event Logs.

Comment 3

[Saujanya]

Attachment: Error Log Details tab from System Event Logs.

Error Log Details tab from System Event Logs.

Comment 4

[Saujanya]

Hi timeless,
Thanks for looking into this.

I downloaded the zip you suggested. Should I just extract and overwrite all the files in my Firefox installation and reproduce the bug? Or does it involve more steps? (Sorry, I am new to this forum).

Meanwhile I have added 2 screenshots for the Error log that gets generated when the bug is seen (without your zip), in case it helps.

Comment 5

[Adam Guthrie]

If you unzip the file, a folder named "firefox" will be created. I would then suggest opening up cmd.exe and cd'ing to that directory and launching Firefox via "firefox -P". This will bring up the profile manager [0], which will allow you to create a new profile to test with.

When you crash, Breakpad should catch the crash. See http://kb.mozillazine.org/Breakpad for info. on how to get the stacktrace for the submitted crash.

[0] http://kb.mozillazine.org/Profile_Mangager

Comment 6

[timeless]

i'm assuming your pictures are useless. the only possible reason they might not be is if the problem matches the exception described in http://www.ddj.com/dept/windows/184416750 and http://www.ddj.com/article/printableArticle.jhtml?articleID=184406003&dept_url=/dept/windows/ which would be really weird. we want a breakpad id.

Comment 7

[Saujanya]

I tried the methods suggested by Adam Guthrie for almost 2 days, Firefox crashes every time. But breakpad is unable to catch the crash. Am I missing some step? I tried both with default profile and a new profile, it made no difference.

I repeated this several times, but could not get the breakpad id. I would expect the breakpad to either prompt with a dialog or store some logs etc. I tried to scan my system after several crashes this way and was able to locate 3 dumps from C:\Users\Administrator\AppData\Roaming\Mozilla\Firefox\Crash Reports\pending folder. I have zipped them and attaching with this defect. I cannot say for 100% sure whether the dumps are for exactly the same crash or some other reason. (I don't use the system for any other purpose, so hopefully they are related).

I even tried with http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/firefox-3.0a7pre.en-US.win32.installer.exe - the browser keeps crashing, but breakpad could not capture anything.

Comment 8

[Saujanya]

Attachment: Dumps as described in post dated 16 July 2007.

Comment 9

[Saujanya]

The bug appears to have got fixed in newer build of Microsoft Windows Server 2008 (Escrow build).
Rest of the configuration (Firefox version, Java version etc) remain the same, only the Windows Server 2008 build was a newer pre-release build.

Comment 10

[AndrewM]

Saujanya, can you still reproduce this on the release version of Windows Server 2008?

Comment 11

[Saujanya]

We are seeing the same bug with Windows x64 Vista operating system and Firefox 3 combination as well.
We tried putting both 32-bit and 64-bit versions of Sun Java Runtime Environment on the system. When the 32-bit JRE is used, it crashes exactly the same was as reported in this bug (with entry in system event log etc). When the 64-bit JRE is used, it crashes silently - the "Test Point 2 version=" alert never appears (from the source given for reproducible steps in the bug report).

Comment 12

[Robert Kaiser]

OJI has been discontinued and Java now runs out-of-process, so this stuff has probably changed a lot. This bug has no info about current software versions, please file a new bug for new issues or reopen this one with current info, including a crash signature if it still happens and move it to a component outside of graveyard.