PRIH Core Dump in HP-UX 11.11 during running of application

[Sandip Mitra]

One of our application name is PRIH. It was executing successfully in Oracle 8i(8.1.7.4), but giving core dump some times in Oracle 10g(10.2.0.4).

This application picks up some files, applies logic and update the transaction value.

Please find the GDB output attached herewith for your reference.

Please help me out. It is a production issue.

[Dennis Handly]

Program terminated with signal 11, Segmentation fault.
#0 0xc00000000079ed78 in free+0x148 libc.2
#1 0x40000000000a2a88 in MpGetCgtab_sr+0x1358

You have heap corruption if you are aborting in free. (Or passed in a bad pointer.)
You need to download a newer gdb and then use the heap checking commands to track this down.

[Steven E. Protter]

Shalom,

I would look into if all Oracle environment and required patches are in place.

ORACLE_HOME and other variables if incorrect can cause this issue.

Also, you may want to try relinking Oracle 10.2.0.4 or updating to a more current Oracle patch set. See http://metalink.oracle.com for more information.

SEP

[Sandip Mitra]

Hi Denis & Steven,

Thanks for the reply.

Our server is applied with the latest Oracle Patch Set.

Denis, I have went through some of the forums in HP. I had found same kind of error with PERL. Can I do some testing with LD_PRELOAD environment variable in my user profile.

[Dennis Handly]

>I had found same kind of error with PERL.

Heap corruption is different each time.

Can I do some testing with LD_PRELOAD environment variable in my user profile?

No. You don't dare export LD_PRELOAD. You must only set it just before you need it:
LD_PRELOAD=xxxx command ...