237689 - Mozilla 1.6 final crashes when browsing to/using administrative functions on a Sun Cobalt Cube 3

Status: RESOLVED INVALID

(Core :: Layout, defect)

Description

[glawson]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113

Mozilla 1.6 final on both Red Hat 9.0 and Windows XP crashes when browsing
to/using administrative functions on a Sun Cobalt Cube 3.  Most functions seem
to work, except for: <Administration> Usage Information/Web.  The RH 9.0 machine
is using the Java plug-in 1.4.2_03-b02, the Windows computers also use java 1.4.
 On the windows machines, IE 6.x works fine with no crashes.

I'm not a programmer, but will be glad to provide additional information or help
with (non-programming) debugging.

--Greg
[e-mail]




Reproducible: Always
Steps to Reproduce:
1.  Browse to the administrative web page of the Cobalt Cube server--login.
2.  Select Administration/Usage Information/Web
3.  Select any statistical function on the pull-down.

Actual Results:  
Typically an immediate crash, sometimes a freeze then crash.

Expected Results:  
Display web usage statistical results from the Cube.

Comment 1

[Rene Pronk]

Since not many of us own a Cobalt Cube server, could you try to save a testcase
that crashes to your harddisk and attach it here in a zip?

Comment 2

[Olivier Cahagne]

can you attach a stacktrace by running 'mozilla -g' then attach the output of
"gdb> backtrace" via "create a new attachment" ?

Comment 3

[glawson]

Attachment: shell output from "mozilla -g"

Comment 4

[glawson]

(In reply to comment #3)
> Created an attachment (id=144138)
> shell output from "mozilla -g"
> 
----------addendum----------------

Sorry, not too familiar with bugzilla.  Above comment #3 should have included
the following text:

----------------------------------

>>>Tried running "mozilla -g" from a shell on Linux (RH 9.0).  This will run
gdb, but it doesn't start the browser.  Should it?  While gdb is open in the
shell, I can start mozilla from either a command line or icon, browse to the
admin functions on the Cube, and then crash the browser.  If I then run
"backtrace" from the debugger, there is no output--only "no stack".  I'm
guessing I'm not running something correctly.

I think the problem may be related to the Java plugin.  I ran an old version of
mozilla (1.2) without the link to the java plugin (libjavaplugin_oji.so), and it
ran fine--no crashes at all.  I then unlinked (remived) the java plugin in the
mozilla 1.6 directory, and it seemed to be a bit more stable--I checked several
web stat functions on the Cube before the browser crashed.  Again, no output
from "gdb backtrace".  I am attaching the shell output from "mozilla -g", if
it's of any help.

Tnx for both responses to this problem!!!

--Greg

Comment 5

[glawson]

Attachment: cube prob - gdb output

Ok, after getting some additional training on gdb :)  --tnx!

1.  believe the symlink to the java plugin is good, although I notice gdb has
references to java 1.3...

libjavaplugin_oji.so ->
/usr/local/j2re1.4.2_03/plugin/i386/ns610-gcc32/libjavaplugin_oji.so

$ gcc --version
gcc (GCC) 3.2.2 20030222 (Red Hat Linux 3.2.2-5)

2.  gdb - bt output attached

Comment 6

[Olivier Cahagne]

-> moving as it's crashing in layout
kw: +regression as 1.2 doesn't seem to crash.

Can you maybe produce a crashing reduced testcase from the pages you're crashing
on or eventually save the page on local drive, zip it and attach here if still
crashing when loading it from local drive ?
Is there any way that this appliance can be "seen" from internet so that
developers could also load pages to reproduce crash with a debug build ?

Comment 7

[Olivier Cahagne]

do you still crash with Mozilla 1.7rc2 ? If so, please post Talkback Incident ID
for this crash by running "mozilla/components/talkback/talkback"

Comment 8

[Bill Mason]

Reporter, if we cannot get further information from previous comments 6 and 7,
please resolve the bug.  Right now there is insufficient information to do
anything with the bug.

Comment 9

[Bill Mason]

Resolving, no reply from reporter.