MPI_FINALIZE, MPI_Finalize

Purpose

Terminates all MPI processing.

C synopsis

#include <mpi.h>
int MPI_Finalize(void);

C++ synopsis

#include mpi.h
void MPI::Finalize();

FORTRAN synopsis

include 'mpif.h' or use mpi
MPI_FINALIZE(INTEGER IERROR)

Parameters

IERROR
     is the FORTRAN return code. It is always the last argument.

Description

Make sure this subroutine is the last MPI call. Any MPI calls made after
MPI_FINALIZE raise an error. You must be sure that all pending
communications involving a task have completed before the task calls
MPI_FINALIZE. You must also be sure that all files opened by the task have
been closed before the task calls MPI_FINALIZE.

Although MPI_FINALIZE terminates MPI processing, it does not terminate the
task. It is possible to continue with non-MPI processing after calling
MPI_FINALIZE, but no other MPI calls (including MPI_INIT) can be made.

In a threads environment, both MPI_INIT and MPI_FINALIZE must be called on
the same thread. MPI_FINALIZE closes the communication library and
terminates the service threads. It does not affect any threads you created,
other than returning an error if one subsequently makes an MPI call. If you
had registered a SIGIO handler, it is restored as a signal handler;
however, the SIGIO signal is blocked when MPI_FINALIZE returns. If you want
to catch SIGIO after MPI_FINALIZE has been called, you should unblock it.

At MPI_FINALIZE there is now an implicit MPI_COMM_FREE of MPI_COMM_SELF.
Because MPI_COMM_SELF cannot have been freed by user code and cannot be
used after MPI_FINALIZE, there is no direct effect of this change. The
value of this implicit free is that any attribute that a user may attach to
MPI_COMM_SELF will be deleted in MPI_FINALIZE and its attribute delete
function called. A library layered on MPI can take advantage of this to
force its own cleanup code to run whenever MPI_FINALIZE gets called. This
is done by packaging the cleanup logic as an attribute delete function and
attaching an attribute to MPI_COMM_SELF. It is legitimate to make MPI calls
in the attribute callbacks and a call to MPI_FINALIZED inside a delete
function will report that MPI is still active.

If an attribute delete function returns a non-zero return code, the code it
does return is passed to the error handler associated with MPI_COMM_WORLD.
The default handler, MPI_ERROR_ARE_FATAL, will embed the error code in the
message it prints. If there is a returning error handler on MPI_COMM_WORLD,
MPI_FINALIZE will return a code indicating that a delete callback failed.
MPI_FINALIZE does not return the error return code issued by the delete
function.

Notes

The MPI standard does not specify the state of MPI tasks after
MPI_FINALIZE, therefore, an assumption that all tasks continue may not be
portable. If MPI_BUFFER_ATTACH has been used and MPI_BUFFER_DETACH has been
not called, there will be an implicit MPI_BUFFER_DETACH within
MPI_FINALIZE. See MPI_BUFFER_DETACH.

Errors

MPI_COMM_SELF attribute delete function returned error

MPI already finalized

MPI not initialized

Related information

     MPI_ABORT
     MPI_BUFFER_DETACH
     MPI_INIT