| MPI_Grequest_start(3) | MPI | MPI_Grequest_start(3) |
MPI_Grequest_start - Create and return a user-defined request
int MPI_Grequest_start(MPI_Grequest_query_function * query_fn,
MPI_Grequest_free_function * free_fn,
MPI_Grequest_cancel_function * cancel_fn,
void *extra_state, MPI_Request * request)
The return values from the callback functions must be a valid MPI error code or class. This value may either be the return value from any MPI routine (with one exception noted below) or any of the MPI error classes. For portable programs, MPI_ERR_OTHER may be used; to provide more specific information, create a new MPI error class or code with MPI_Add_error_class or MPI_Add_error_code and return that value.
The MPI standard is not clear on the return values from the callback routines. However, there are notes in the standard that imply that these are MPI error codes. For example, pages 169 line 46 through page 170, line 1 require that the free_fn return an MPI error code that may be used in the MPI completion functions when they return MPI_ERR_IN_STATUS .
The one special case is the error value returned by MPI_Comm_dup when the attribute callback routine returns a failure. The MPI standard is not clear on what values may be used to indicate an error return. Further, the Intel MPI test suite made use of non-zero values to indicate failure, and expected these values to be returned by the MPI_Comm_dup when the attribute routines encountered an error. Such error values may not be valid MPI error codes or classes. Because of this, it is the user's responsibility to either use valid MPI error codes in return from the attribute callbacks, if those error codes are to be returned by a generalized request callback, or to detect and convert those error codes to valid MPI error codes (recall that MPI error classes are valid error codes).
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.
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.
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.
| 11/21/2018 |