Re: [CODE] [SYSTEM] Faulty Core Dumps and AIX

From: Chris Gilbert (
Date: 06/02/00

Robert Moon wrote:
> Since moving my MUD from a Linux box to an RS6000 with AIX nearly a month
> ago, I have not been able to generate a proper core dump.  Upon nohup
> boot-up, nohup.out produces:
> ./autorun[49]: 16626 IOT/Abort trap(coredump)
> ./autorun[49]: 22414 IOT/Abort trap(coredump)
> ./autorun[49]: 22510 Segmentation fault(coredump)
> ./autorun[49]: 22302 Segmentation fault(coredump)
> ./autorun[49]: 22350 Segmentation fault(coredump)
> Line 49 in autorun is what starts the mud (FLAGS = -q and PORT = 4000):
>   bin/circle $FLAGS $PORT >> syslog 2>&1
> When trying to run gdb on the core file after a crash, the following error
> message is generated:
> "/home/rmoon/mud/lib/core" is not a core dump: File format not recognized
> I *am* able to run the mud from within gdb using the RUN command and step
> through and set breaks in order to debug.  However, core dumps are
> definintely not being properly generated, and I can't maintain the MUD's
> operation in gdb forever.  The MUD-friendly sysadmin of the RS6000 machine
> and I have no clue why this is happening.  Any suggestions are most
> appreciated.

Which version of gdb?  and are you using c for AIX (IE ibm's compiler)
or gcc?

For gdb to work properly on AIX you need version 5.0 (it's the only one
that understands the 32-64 bit libraries (or maybe a post 4.18 cvs
version may as well. I'd recommend gcc 2.95.2 as well.  If you're using
the AIX compiler you may find a patch for it on IBM's web site somewhere
(not an easy thing to navigate to)

Other thought, you did remove your config.cache and rerun configure on
the AIX box, otherwise you're conf.h may be wrong...


     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     |  |

This archive was generated by hypermail 2b30 : 04/10/01 PDT