[lammps-users] segmentation fault

Dear Friends

I want to run a simulation box containing a solution of water. but in the initial steps of the simulation, the running stops with “segmentation fault” message. I could not find this message in the manual. In the mailing list I saw that Cameron has encountered this problem and Axel answer him/her s the following:

if you get a segmentation fault, you can always force
a core dump and then generate a stack trace to find
out in which subroutine the executable craters.

But I don’t know this answer. would you please explain the answer?

I will be very thankful if I receive any suggestion for solving this problem.

Regards

Saly


But I don't know this answer. would you please explain the answer?

Dear Friends

I want to run a simulation box containing a solution of water. but in the
initial steps of the simulation, the running stops with "segmentation fault"
message. I could not find this message in the manual. In the mailing list I

that usually means, that something in your input is horribly wrong.
the segmentation fault is a "catch-all" error and does not point to
anything specific, so without additional information, nobody can help you.

saw that Cameron has encountered this problem and Axel answer him/her s the
following:

if you get a segmentation fault, you can always force
a core dump and then generate a stack trace to find
out in which subroutine the executable craters.

But I don't know this answer. would you please explain the answer?

this has very little to do with lammps and a lot with
general unix programming and debugging.

it would be too time consuming and rather difficult
to explain in detail. your best bet is to find somebody
with that kind of experience somewhere local (there
_have_ to be lots of those around).

what you have to to is to enable core dumps from
(e.g. ulimit -c unlimited) and then load the executable
and the core dump into gdb and ask for a stack trace
with "where". if you want more detail, you first have to
compile and link lammps with the -g flag added so
that your binary includes additional debug info.

if you don't understand this, you have to look for enlightenment
in some unix/programming beginners forum.

axel.