Re: [OMPI users] mpi with icc, icpc and ifort :: segfault (Jeff Squyres)

[George Bosilca]

There is another piece of information that can be useful in order to figure out what's wrong. If you can execute the ompi_info directly in gdb, run it until it segfault and then send us the output of "where" and "shared" (both are gdb commands). This will give us access to the exact location of the segfault, and the list of all shared libraries that get loaded by the application.

  Thanks,
    george.

On Jul 5, 2007, at 12:19 PM, Ricardo Reis wrote:

On Thu, 5 Jul 2007, Jeff Squyres wrote:

Yoinks -- that's not good.

I suspect that our included memory manager is borking things up
(Brian: can you comment?).  Can you try configuring OMPI --without-
memory-manager?

Yes. It compiles and links OK (execution of mpif90).

Trying to run (mpirun -np <p> <exec>) gives segmentation fault.



Added ompi_info log (without --all)
Added strace ompi_info --all log
Added strace mpirun log

 greets,

 Ricardo Reis

 'Non Serviam'

 PhD student @ Lasef
 Computational Fluid Dynamics, High Performance Computing, Turbulence
 <http://www.lasef.ist.utl.pt>

 &

 Cultural Instigator @ Rádio Zero

users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

Ompi_info gives output and then segfaults. ompi_info --all segfaults immediatly. http:// radio.ist.utl.pt<mpirun.log.bz2><ompi_info.log.bz2><ompi_info_all.log. bz2>_______________________________________________

smime.p7s
Description: S/MIME cryptographic signature