1
1
openmpi/test/memchecker/irecv_uninit_check.c

65 строки
1.4 KiB
C
Исходник Обычный вид История

/*
* Copyright (c) 2004-2007 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include <stdio.h>
#include "mpi.h"
int main( int argc, char *argv[] )
{
int rank, size;
int send_value[10],recv_value[10];
int i;
MPI_Status status;
MPI_Request send_request;
MPI_Request recv_request;
MPI_Init( &argc, &argv );
MPI_Comm_rank( MPI_COMM_WORLD, &rank );
MPI_Comm_size( MPI_COMM_WORLD, &size );
for(i=0;i<10;i++)
{
send_value[i] = i;
printf("\n%d:%d",i,send_value[i]);
}
/*
* The last value of receive buffer is
* uninitialized.
*/
if (size != 2) {
fprintf (stderr, "Error: Need 2 processes\n");
MPI_Finalize ();
}
/*
* Now we only send 9 ints, but expect to receive 10.
* Perfectly valid, but the receiver (in strict mode?) should not
* depend on the data.
*/
MPI_Isend (&send_value, 9, MPI_INT,
(rank + 1) % size, 4711, MPI_COMM_WORLD, &send_request);
MPI_Irecv (&recv_value, 10, MPI_INT,
(rank + size - 1) % size, 4711, MPI_COMM_WORLD, &recv_request);
MPI_Wait (&send_request, &status);
MPI_Wait (&recv_request, &status);
/*
* This should return an error in any mode.
*/
printf("\nError: buf[9]:%d",recv_value[9]);
MPI_Finalize ();
return 0;
}