On 04/09/2015 08:22 AM, Patrick Begou wrote: > Andrew Haley wrote: >> On 09/04/15 08:06, Patrick Begou wrote: >>> What have I missed ? >> Running it in GDB? >> > The real application segfault when runing on 192 cores (mpi application) on the > cluster nodes..... > So it is a little bit difficult to investigate interactively. Did you call backtrace_symbols() ? To be honest with you: debugging a segfault without a debugger is nothing but a trail of tears. It is worth the effort to set this up. Andrew.
