DeinoMPI

The Great and Terrible implementation of MPI-2

function index

MPI_Grequest_complete

Notify MPI that a user-defined request is complete
int MPI_Grequest_complete(
  MPI_Request request
);

Parameters

request
[in] Generalized request to mark as complete

Remarks

The call informs MPI that the operations represented by the generalized request request are complete. A call to MPI_WAIT(request, status) will return and a call to MPI_TEST(request, flag, status) will return flag=true only after a call to MPI_GREQUEST_COMPLETE has declared that these operations are complete.

MPI imposes no restrictions on the code executed by the callback functions. However, new nonblocking operations should be defined so that the general semantic rules about MPI calls such as MPI_TEST, MPI_REQUEST_FREE, or MPI_CANCEL still hold. For example, all these calls are supposed to be local and nonblocking. Therefore, the callback functions query_fn, free_fn, or cancel_fn should invoke blocking MPI communication calls only if the context is such that these calls are guaranteed to return in finite time. Once MPI_CANCEL is invoked, the cancelled operation should complete in finite time, irrespective of the state of other processes (the operation has acquired "local" semantics). It should either succeed, or fail without side-effects. The user should guarantee these same properties for newly defined operations.

Thread and Interrupt Safety

This routine is thread-safe. This means that this routine may be safely used by multiple threads without the need for any user-provided thread locks. However, the routine is not interrupt safe. Typically, this is due to the use of memory allocation routines such as malloc or other non-MPICH runtime routines that are themselves not interrupt-safe.

Notes for Fortran

All MPI routines in Fortran (except for MPI_WTIME and MPI_WTICK) have an additional argument ierr at the end of the argument list. ierr is an integer and has the same meaning as the return value of the routine in C. In Fortran, MPI routines are subroutines, and are invoked with the call statement.

All MPI objects (e.g., MPI_Datatype, MPI_Comm) are of type INTEGER in Fortran.

Errors

All MPI routines (except MPI_Wtime and MPI_Wtick) return an error value; C routines as the value of the function and Fortran routines in the last argument. Before the value is returned, the current MPI error handler is called. By default, this error handler aborts the MPI job. The error handler may be changed with MPI_Comm_set_errhandler (for communicators), MPI_File_set_errhandler (for files), and MPI_Win_set_errhandler (for RMA windows). The MPI-1 routine MPI_Errhandler_set may be used but its use is deprecated. The predefined error handler MPI_ERRORS_RETURN may be used to cause error values to be returned. Note that MPI does not guarentee that an MPI program can continue past an error; however, MPI implementations will attempt to continue whenever possible.

MPI_SUCCESS
No error; MPI routine completed successfully.

See Also

MPI_Grequest_start

Example Code

The following sample code illustrates MPI_Grequest_complete.

#include "mpi.h"
#include <stdio.h>
 
/* Simple test of generalized requests */
 
int query_fn( void *extra_state, MPI_Status *status )
{
   
/* Set a default status */
    status->MPI_SOURCE = MPI_UNDEFINED;
    status->MPI_TAG = MPI_UNDEFINED;
    MPI_Status_set_cancelled( status, 0 );
    MPI_Status_set_elements( status, MPI_BYTE, 0 );
    return 0;
}
int free_fn( void *extra_state )
{
    int *b = (int *)extra_state;
   
if (b) *b = *b - 1;
   
/* The value returned by the free function is the error code returned by the wait/test function */
   
return 0;
}
int cancel_fn( void *extra_state, int complete )
{
    return 0;
}
 
/*
* This is a very simple test of generalized requests. Normally, the
* MPI_Grequest_complete function would be called from another routine,
* often running in a separate thread. This simple code allows us to
* check that requests can be created, tested, and waited on in the
* case where the request is complete before the wait is called.
*
* Note that MPI did *not* define a routine that can be called within
* test or wait to advance the state of a generalized request.
* Most uses of generalized requests will need to use a separate thread.
*/
int main( int argc, char *argv[] )
{
   
int errs = 0;
   
int counter, flag;
    MPI_Status status;
    MPI_Request request;
 
    MPI_Init( &argc, &argv );
 
    MPI_Grequest_start( query_fn, free_fn, cancel_fn, NULL, &request );

    MPI_Test( &request, &flag, &status );
    if (flag) {
        errs++;
        fprintf( stderr, "Generalized request marked as complete\n" );
    }
 
    MPI_Grequest_complete( request );
 
    MPI_Wait( &request, &status );
 
    counter = 1;
    MPI_Grequest_start( query_fn, free_fn, cancel_fn, &counter, &request );
    MPI_Grequest_complete( request );
    MPI_Wait( &request, MPI_STATUS_IGNORE );

    if (counter) {
        errs++;
        fprintf( stderr, "Free routine not called, or not called with extra_data" );fflush(stderr);
    }
 
    MPI_Finalize();
   
return 0;
}