[gmx-developers] Segfault/Core dump on Solaris -> reproduced on linux ! Debug output

[Marc Baaden]

Hi,

just reproduced my problem on a linux box, requesting 8 instead of
2 nodes. It gets weirder and weirder.

I now know that the bonds() routine is responsible for the problem
as shows eg the adb output from below. This same routine segfaults
for 1 (and presumably 2, 12 and 16) processors under Solaris and
for 8 processors on linux. Strangely the 2 processors on linux job
is running fine.

Any clues to this puzzle ??

Here is the copy of adb output


============================
oswell-1% adb `which mdrun` core
core file = core -- program ``mdrun'' on platform SUNW,Sun-Fire
SIGSEGV: Segmentation Fault
$c
bonds(0,9a4308,999038,305d18,10ab9d8,4a0390) + 2bc
calc_bonds(2b3ed0,12,24b558,2f9ae4,305d18,24b554) + 1fc
force(2b3ed0,0,4a0390,2b6118,2b5e20,ffbeeed0) + 644
do_force(2b3ed0,2b4890,24a000,4a0390,2f8ab8,ffbeee9c) + 8b4
do_md(2b3ed0,2b4890,0,2b5e20,3d2c30,10ab9d8) + 1094
mdrunner(2b4890,2b5f98,15,249474,49fb48,305d18) + 688
main(0,0,0,259c00,0,0) + 2a8
$C
bonds(0,9a4308,999038,305d18,10ab9d8,4a0390) + 2bc
        [savfp=0xffbee7a8,savpc=0xa893c]
calc_bonds(2b3ed0,12,24b558,2f9ae4,305d18,24b554) + 1fc
        [savfp=0xffbee860,savpc=0x524c4]
force(2b3ed0,0,4a0390,2b6118,2b5e20,ffbeeed0) + 644
        [savfp=0xffbee928,savpc=0x910d4]
do_force(2b3ed0,2b4890,24a000,4a0390,2f8ab8,ffbeee9c) + 8b4
        [savfp=0xffbeea00,savpc=0x39cb4]
do_md(2b3ed0,2b4890,0,2b5e20,3d2c30,10ab9d8) + 1094
        [savfp=0xffbef328,savpc=0x388c8]
mdrunner(2b4890,2b5f98,15,249474,49fb48,305d18) + 688
        [savfp=0xffbef460,savpc=0x3b848]
main(0,0,0,259c00,0,0) + 2a8
        [savfp=0x0,savpc=0x29728]
^D
============================


Cheers,
Marc


-- 
 Dr. Marc Baaden - Laboratory of Molecular Biophysics, Oxford University
 mailto:baaden at smplinux.de  - ICQ# 11466242 -  http://www.marc-baaden.de
 FAX/Voice +49 697912 39550  -  Tel: +44 1865 275380  or  +33 609 843217