Re: [OMPI users] Run failure on Solaris Opteron with Sun Studio 11

[Pierre Valiron]

Thanks Jeff,

Things are better with 1.0.2a9. However open-mpi is not functional as yet.

- My new build script:
#! /bin/tcsh -v
setenv CC          "cc"
setenv CXX         "CC"
setenv F77         "f77"
setenv FC          "f95"
setenv CFLAGS      "-O -xtarget=opteron -xarch=amd64"
setenv CXXFLAGS    "-O -xtarget=opteron -xarch=amd64"
setenv FCFLAGS     "-O -xtarget=opteron -xarch=amd64"
setenv FFLAGS      "-O -xtarget=opteron -xarch=amd64"
setenv LD          "ld -64"
setenv LDFLAGS     "-xtarget=opteron -xarch=amd64"
setenv CXXLDFLAGS  "-xtarget=opteron -xarch=amd64"
./configure --prefix=/users/valiron/lib/openmpi-1.0.2a9
gmake
# gmake install

- Configure went through with no errors



- I could compile/link a small piece of code (rotate.f)
valiron@icare ~/config > mpif77 -xtarget=opteron -xarch=amd64 rotate.f

rotate.f:
MAIN rotate:

- However the resulting executable fails to launch:

Segmentation fault (core dumped)

- The problem seems buried into open-mpi:
valiron@icare ~/config > pstack core

fffffd7fff05dfe0 strlen () + 20
fffffd7fff0b6ab3 vsprintf () + 33
fffffd7fff2e4211 opal_vasprintf () + 41
fffffd7fff2e41c8 opal_asprintf () + 98
00000000004098a3 orterun () + 63
0000000000407214 main () + 34
000000000040708c ???????? ()

What additional info could help to track the problem ?
Pierre.



Jeff Squyres wrote:

- Make completed nicely, excepted compiling ompi/mpi/f90/mpi.f90 which took nearly half an hour to complete. I suspect the optimization flags in FFLAGS are not important for applications, and I could use -O0 or -O1 instead. NOTICE: Invoking /opt/Studio11/SUNWspro/bin/f90 -f77 -ftrap=%none -I/users/valiron/lib/openmpi-1.0.2a9/include -I/users/valiron/lib/openmpi-1.0.2a9/include/openmpi/ompi -xtarget=opteron -xarch=amd64 rotate.f -L/users/valiron/lib/openmpi-1.0.2a9/lib -lmpi -lorte -lopal -lnsl -lsocket -lthread -lm -lnsl -lsocket -lthread -ldl valiron@icare ~/config > mpirun --prefix /users/valiron/lib/openmpi-1.0.2a9 -np 2 a.out core 'core' of 27996: mpirun --prefix /users/valiron/lib/openmpi-1.0.2a9 -np 2 a.out

On Feb 24, 2006, at 12:17 PM, Pierre Valiron wrote:

*** Fortran 90/95 compiler
checking whether we are using the GNU Fortran compiler... no
checking whether f95 accepts -g... yes
checking for Fortran flag to compile .f files... none
checking for Fortran flag to compile .f90 files... none
checking whether f77 and f95 compilers are compatible... no

configure: WARNING: *** Disabling Fortran 90/95 bindings

configure: WARNING: *** Fortran 77 and Fortran 90 compilers are not link compatible

I see the problem here:

configure:32389: f95  -o conftest conftestf90.o conftestf77.o
ld: fatal: file conftestf77.o: wrong ELF class: ELFCLASS64





should be:

setenv FCFLAGS      "-O -xtarget=opteron -xarch=amd64"

(i.e., "FCFLAGS", not "FCLAGS").

We do a test in configure to check that both compilers are link- compatible to ensure that we don't run into any unexpected problems later. The problem here is that the f77 test file was compiled with the proper 64 bit flags, but the f90 test was not. This comes from a typo in your openmpi-build-icare.csh script. The following line: setenv FCLAGS "-O -xtarget=opteron -xarch=amd64"

Then the make fails in orte/mca/pls/base with message
"pls_base_context.c", line 48: undefined symbol: MAXHOSTNAMELEN

Oops. This is my fault from a recent commit. We fixed this, and after doing some sample builds with Sun Studio 11 on Solaris/SPARC (this is not one of the compilers/platforms we test with on a regular basis yet) we found another pair of minor issues that have now also been fixed. Specifically, we think that these were Solaris issues -- not SPARC issues. I just put 1.0.2a9 on the web site with all these fixes. Can you give it a whirl and let us know how it goes?

--
Soutenez le mouvement SAUVONS LA RECHERCHE :
http://recherche-en-danger.apinc.org/

      _/_/_/_/    _/       _/       Dr. Pierre VALIRON
     _/     _/   _/      _/   Laboratoire d'Astrophysique
    _/     _/   _/     _/    Observatoire de Grenoble / UJF
   _/_/_/_/    _/    _/    BP 53  F-38041 Grenoble Cedex 9 (France)
  _/          _/   _/    http://www-laog.obs.ujf-grenoble.fr/~valiron/
 _/          _/  _/     Mail: pierre.vali...@obs.ujf-grenoble.fr
_/          _/ _/      Phone: +33 4 7651 4787  Fax: +33 4 7644 8821

_/ _/_/

rotate.f.gz
Description: GNU Zip compressed data