Is this a libc problem?

[Giuseppe Sacco]

Hi all,
one application on itanium2 system often crashes. The application is compiled for PA-RISC and call java via JNI methods.

When the application crashes, I get this information on a file, written by the JVM engine.

Is this a libc or java problem?

Thank you very much,
Giuseppe

An unexpected exception has been detected in native code outside the VM.
Unexpected Signal : 11 occurred at PC=0xC225DC30
Function=_doprnt
Library=/usr/lib/libc.2

Current Java thread:
"main" prio=7 tid=0x40339168 nid=1 lwp_id=361860 runnable [0x00000000..0x7b03caf0]

Dynamic libraries:
/appl/baan4/bse/bin/bshell6.1
text:0x00001000-0x00190394 data:0x40001000-0x4004d360
/opt/java1.4/jre/lib/PA_RISC/server/libjvm.sl
text:0xc6bf1000-0xc77c2000 data:0x7ae8b000-0x7aff4000
/usr/lib/librt.2
text:0xc444b000-0xc444f000 data:0x7b002000-0x7b003000
[....]
/usr/baan/shared/shlib/libBclmEx.sl
text:0xc6baf000-0xc6bf1000 data:0x7ae04000-0x7ae0f000

Local Time = Tue Jan 29 01:41:32 2008
Elapsed Time = 93
#
# The exception above was detected in native code outside the VM
#
# Java VM: Java HotSpot(TM) Server VM (1.4.1 1.4.1.03-030630-21:49-PA_RISC1.1 PA1.1 mixed mode)
#

[Peter Nikitka · Accepted solution]

Hi,

first: please tell us your HP-UX version.

Next: The java version 1.4.1.03 is really not the most current one - even in the 1.4 release tree.
Check, what the release notes of your 'baan4'-application for java references and check 'baan4' itself for known bugs and updates.

mfG Peter

[Giuseppe Sacco]

uname gives "B.11.23 U ia64 1161367452"

A few hours ago I found that HP released a new JVM (version 1.4.2.17) on September, so I will probably try this upgrade path, but I was wondering if anyone might give a different advice.

Thanks,
Giuseppe

[Hein van den Heuvel]

Ocam's razor thery says that this is unlikely to be a java probably and even less likely to be a libc problem.

Most likely an appliaction coding problel or application configuration problem (a string longer than allowed, a parameter not provided).

_doprt is an internal function that is used by all the printf style functions, which simply pass their format, arguments, and stream to this function.

For your application to find a bug in that is to big a bug. The world would fall apart.

$ grep 11 /usr/include/sys/signal.h
gives:
define SIGSEGV 11 /* Segmentation violation */

So some print function was called with a bad, or non-existing argument.

Look for warnings in any and all log files, notably startup logs.

Ask the application vendor for possible prior occurances.

And ask yourself why you? why now?
Why not your buddy yesteday?
What changed? Did it ever work?

Good luck!
Hein.

[Dennis Handly]

>Is this a libc problem?

Not likely. Either you have passed printf the wrong format/args or you have a thread stack overflow.

If you can get a core file or run it in gdb, you (really me ;-) could look at what's at PC=0xC225DC30 and figure out which of the 2 cases it is. A stack trace would be helpful.
If you could print the line under "Dynamic libraries:" for libc.2 and provide your libc patch version, I might be able to do that.

[Giuseppe Sacco]

Thanks for all you answers. We actually planned to update java to the latest 1.4 version available, but since the machine is part of the production environment we cannot do it this week.

So, I would leave this thread opened until we try this upgrade, then I will be back here with any result.

Thanks again,
Giuseppe

[Giuseppe Sacco]

Java upgrade fixed the problem, so I thank you very much for your help.

Bye,
Giuseppe

[Peter Nikitka]

Hi,

since you are new to this forum, please take the time an read this about its unique 'point system':
http://forums11.itrc.hp.com/service/forums/helptips.do?#33

mfG Peter