[Valgrind-users] Problem with valgrind-3.17.0 and openmpi-4.0.5

[TESSER F. <fed...@po...>]

Good morning.

I have installed valgrind 3.17.0, having previously loaded 
the
module for openmpi 4.0.5, so it found the "MPI2-compliant 
mpicc
and mpi.h...".

However, trying to run just a simple program like this 
one:



#include <mpi.h>
#include <stdio.h>

int main(int argc, char** argv) {

MPI_Init(NULL, NULL);

int world_size;
int world_rank;
int name_len;
char processor_name[MPI_MAX_PROCESSOR_NAME];

MPI_Comm_size(MPI_COMM_WORLD, &world_size);
MPI_Comm_rank(MPI_COMM_WORLD, &world_rank);
MPI_Get_processor_name(processor_name, &name_len);

printf("Hello world from processor %s, rank %d out of %d 
processors\n",
	   processor_name, world_rank, world_size);

MPI_Finalize();

}



will produce the following errors:



==113228== Memcheck, a memory error detector
==113228== Copyright (C) 2002-2017, and GNU GPL'd, by 
Julian Seward et al.
==113228== Using Valgrind-3.17.0 and LibVEX; rerun with -h 
for copyright info
==113228== Command: ./pure_mpi_valgrind_try/a.out
==113228==
valgrind MPI wrappers 113228: Active for pid 113228
valgrind MPI wrappers 113228: Try MPIWRAP_DEBUG=help for 
possible options
vex amd64->IR: unhandled instruction bytes: 0x62 0xF2 0x7D 
0x8 0x7C 0xC5 0xC5 0xF9 0xD6 0x43
vex amd64->IR:   REX=0 REX.W=0 REX.R=0 REX.X=0 REX.B=0
vex amd64->IR:   VEX=0 VEX.L=0 VEX.nVVVV=0x0 ESC=NONE
vex amd64->IR:   PFX.66=0 PFX.F2=0 PFX.F3=0
==113228== valgrind: Unrecognised instruction at address 
0x5c79318.
==113228==    at 0x5C79318: opal_pointer_array_init (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5CA4BDB: mca_base_var_init (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5C82F11: opal_init_util (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5157FD9: ompi_mpi_init 
(ompi_mpi_init.c:428)
==113228==    by 0x50FB3A8: PMPI_Init (pinit.c:69)
==113228==    by 0x4E4BC26: PMPI_Init (libmpiwrap.c:2288)
==113228==    by 0x10893B: main (main.c:6)
==113228== Your program just tried to execute an 
instruction that Valgrind
==113228== did not recognise.  There are two possible 
reasons for this.
==113228== 1. Your program has a bug and erroneously 
jumped to a non-code
==113228==    location.  If you are running Memcheck and 
you just saw a
==113228==    warning about a bad jump, it's probably your 
program's fault.
==113228== 2. The instruction is legitimate but Valgrind 
doesn't handle it,
==113228==    i.e. it's Valgrind's fault.  If you think 
this is the case or
==113228==    you are not sure, please let us know and 
we'll try to fix it.
==113228== Either way, Valgrind will now raise a SIGILL 
signal which will
==113228== probably kill your program.
==113228==
==113228== Process terminating with default action of 
signal 4 (SIGILL): dumping core
==113228==  Illegal opcode at address 0x5C79318
==113228==    at 0x5C79318: opal_pointer_array_init (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5CA4BDB: mca_base_var_init (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5C82F11: opal_init_util (in 
/usr/local/openmpi-4.0.5/lib/libopen-pal.so.40.20.5)
==113228==    by 0x5157FD9: ompi_mpi_init 
(ompi_mpi_init.c:428)
==113228==    by 0x50FB3A8: PMPI_Init (pinit.c:69)
==113228==    by 0x4E4BC26: PMPI_Init (libmpiwrap.c:2288)
==113228==    by 0x10893B: main (main.c:6)
slurmstepd: error: *** JOB 159641 ON node01 CANCELLED AT 
2021-07-07T10:21:29 ***
srun: Job step aborted: Waiting up to 32 seconds for job 
step to finish.
srun: error: Timed out waiting for job step to complete
slurmstepd: error: *** STEP 159641.0 ON node01 CANCELLED 
AT 2021-07-07T10:22:48 ***



What am I doing wrong?

Regards,

Federico Tesser