Bug 81040 - IBM JDK core dump
Summary: IBM JDK core dump
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Linux
Classification: Retired
Component: kernel
Version: 9
Hardware: i386
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Arjan van de Ven
QA Contact: Brian Brock
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2003-01-03 17:48 UTC by Jon Willeke
Modified: 2008-01-17 17:49 UTC (History)
1 user (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2004-09-30 15:40:21 UTC
Embargoed:


Attachments (Terms of Use)
strace javac foo.java (1.3.1 on 8.0.92) (34.85 KB, text/plain)
2003-01-03 17:51 UTC, Jon Willeke
no flags Details
strace javac foo.java (1.3.1 on 8.0) (117.24 KB, text/plain)
2003-01-03 17:54 UTC, Jon Willeke
no flags Details
strace javac foo.java (1.4 on 8.0.92) (39.12 KB, text/plain)
2003-01-03 17:59 UTC, Jon Willeke
no flags Details
strace javac foo.java (1.4 on 8.0) (81.12 KB, text/plain)
2003-01-03 18:03 UTC, Jon Willeke
no flags Details

Description Jon Willeke 2003-01-03 17:48:08 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.2.1) Gecko/20021130

Description of problem:
The javac and java programs in the IBM Java Developer Kit core dump on trivial
operations.  I've found that 1.3.1-3 and 1.4-0 work on RH 8.0 (psyche), but not
on RH 8.0.92 (phoebe).  I suspect that this is related to the kernel.

Previous versions of the JDK mentioned a work around for early versions of the
2.4 kernel: setting the environment variable LD_ASSUME_KERNEL to 2.2.5.  This
work around is effective.  Still, since Red Hat 8.0 doesn't cause a problem,
this seems to be a regression.

I'll attach strace output of javac for a minimal Java program, foo.java.  1.3.1
appears to core dump calling mprotect(), whereas 1.4 hangs in futex(), shortly
after calling mprotect().

Version-Release number of selected component (if applicable):
2.4.20-2.2

How reproducible:
Always

Steps to Reproduce:
1. Install the IBM Java Developer Kit, version 1.3.1 SR 3 or 1.4.

2. Enter a trivial Java program, such as foo.java:

   public class foo {
     public static void main( String args[] ) {
     }
   }

3. Try to compile this program with javac:

  $ /opt/IBMJava2-131/bin/javac foo.java
  Segmentation fault (core dumped)

Comment 1 Jon Willeke 2003-01-03 17:51:37 UTC
Created attachment 89097 [details]
strace javac foo.java (1.3.1 on 8.0.92)

This is strace output showing a core dump after sigaltstack().	I will attach
strace output from RH 8.0, showing that the next call is mprotect().

Comment 2 Arjan van de Ven 2003-01-03 17:53:41 UTC
RHL 8.0  kernel didn't have futexes; the beta kernel does.
It looks like the IBM JDK has broken futex handling.....


Comment 3 Jon Willeke 2003-01-03 17:54:45 UTC
Created attachment 89098 [details]
strace javac foo.java (1.3.1 on 8.0)

This complements the 8.0.92 strace output, showing that the core dump may have
occurred in mprotect().

Comment 4 Jon Willeke 2003-01-03 18:00:00 UTC
Created attachment 89099 [details]
strace javac foo.java (1.4 on 8.0.92)

This strace output shows a hang in futex().  This may be more interesting than
the 1.3.1 output, as the phoebe kernel has extensive threading changes.

Comment 5 Jon Willeke 2003-01-03 18:03:43 UTC
Created attachment 89100 [details]
strace javac foo.java (1.4 on 8.0)

For completeness.  This complements the 1.4 strace output on phoebe, but the
calls after clone() don't look the same.

Comment 6 Arjan van de Ven 2003-01-21 11:30:13 UTC
have you reported this to IBM?
It doesn't look like it's our bug

Comment 7 Bugzilla owner 2004-09-30 15:40:21 UTC
Thanks for the bug report. However, Red Hat no longer maintains this version of
the product. Please upgrade to the latest version and open a new bug if the problem
persists.

The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, 
and if you believe this bug is interesting to them, please report the problem in
the bug tracker at: http://bugzilla.fedora.us/



Note You need to log in before you can comment on or make changes to this bug.