1
1
openmpi/ompi/mpi/c/file_iwrite.c

89 строки
2.7 KiB
C
Исходник Обычный вид История

/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
/*
* Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2005 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2008 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
* Copyright (c) 2008 Sun Microsystems, Inc. All rights reserved.
* Copyright (c) 2013 Los Alamos National Security, LLC. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include "ompi/mpi/c/bindings.h"
#include "ompi/runtime/params.h"
#include "ompi/errhandler/errhandler.h"
- Split the datatype engine into two parts: an MPI specific part in OMPI and a language agnostic part in OPAL. The convertor is completely moved into OPAL. This offers several benefits as described in RFC http://www.open-mpi.org/community/lists/devel/2009/07/6387.php namely: - Fewer basic types (int* and float* types, boolean and wchar - Fixing naming scheme to ompi-nomenclature. - Usability outside of the ompi-layer. - Due to the fixed nature of simple opal types, their information is completely known at compile time and therefore constified - With fewer datatypes (22), the actual sizes of bit-field types may be reduced from 64 to 32 bits, allowing reorganizing the opal_datatype structure, eliminating holes and keeping data required in convertor (upon send/recv) in one cacheline... This has implications to the convertor-datastructure and other parts of the code. - Several performance tests have been run, the netpipe latency does not change with this patch on Linux/x86-64 on the smoky cluster. - Extensive tests have been done to verify correctness (no new regressions) using: 1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and ompi-ddt: a. running both trunk and ompi-ddt resulted in no differences (except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run correctly). b. with --enable-memchecker and running under valgrind (one buglet when run with static found in test-suite, commited) 2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt: all passed (except for the dynamic/ tests failed!! as trunk/MTT) 3. compilation and usage of HDF5 tests on Jaguar using PGI and PathScale compilers. 4. compilation and usage on Scicortex. - Please note, that for the heterogeneous case, (-m32 compiled binaries/ompi), neither ompi-trunk, nor ompi-ddt branch would successfully launch. This commit was SVN r21641.
2009-07-13 08:56:31 +04:00
#include "ompi/datatype/ompi_datatype.h"
#include "ompi/file/file.h"
#include "ompi/mca/io/io.h"
#include "ompi/mca/io/base/io_base_request.h"
#include "ompi/memchecker.h"
#if OPAL_HAVE_WEAK_SYMBOLS && OMPI_PROFILING_DEFINES
#pragma weak MPI_File_iwrite = PMPI_File_iwrite
#endif
#if OMPI_PROFILING_DEFINES
#include "ompi/mpi/c/profile/defines.h"
#endif
static const char FUNC_NAME[] = "MPI_File_iwrite";
int MPI_File_iwrite(MPI_File fh, const void *buf, int count, MPI_Datatype
datatype, MPI_Request *request)
{
int rc;
MEMCHECKER(
memchecker_datatype(datatype);
memchecker_call(&opal_memchecker_base_isdefined, buf, count, datatype);
);
if (MPI_PARAM_CHECK) {
rc = MPI_SUCCESS;
OMPI_ERR_INIT_FINALIZE(FUNC_NAME);
if (ompi_file_invalid(fh)) {
fh = MPI_FILE_NULL;
rc = MPI_ERR_FILE;
} else if (count < 0) {
rc = MPI_ERR_COUNT;
Clean up request handling in the I/O framework to be more consistent with other request-using frameworks. - Rather than having mpi/c/* functions allocate requests explicitly, pass the MPI_Request* down to the I/O component and have it perform the allocation. - While the I/O base provides a base request which can be used, it is not required and all request management occurs within the component. - Push progress management into the component, rather than having it happen in the base. Progress functions are now easily registered, and not all (ie, the one existing) components use progress functions in any rational way. ROMIO switched to generalized requests instead of MPIO_Requests many moons ago, and Open MPI now uses ROMIO's generalized requests, so there is no reason to wrap those requests (which are OMPI requests) in another level of request. Now the file function passes the MPI_Request* to the ROMIO component, which passes it to the underlying ROMIO function, which calls MPI_Grequest_start to create an OMPI request, which is what gets set as the request to the user. Much cleaner. This patch has two motivations. One, a whole heck of a lot of code just got removed, and request handling is now much cleaner for I/O components. Two, by adding support for Argonne's proposed generalized request extensions, we can allow ROMIO to provide async I/O through generalized requests, which we couldn't rationally do in the old setup due to the crazy request completion rules. This commit was SVN r22235.
2009-11-26 08:13:43 +03:00
} else if (NULL == request) {
rc = MPI_ERR_REQUEST;
} else {
OMPI_CHECK_DATATYPE_FOR_SEND(rc, datatype, count);
}
OMPI_ERRHANDLER_CHECK(rc, fh, rc, FUNC_NAME);
}
OPAL_CR_ENTER_LIBRARY();
/* Call the back-end io component function */
switch (fh->f_io_version) {
case MCA_IO_BASE_V_2_0_0:
/* XXX -- CONST -- do not cast away const -- update mca/io */
rc = fh->f_io_selected_module.v2_0_0.
io_module_file_iwrite(fh, (void *) buf, count, datatype, request);
break;
default:
rc = MPI_ERR_INTERN;
break;
}
/* All done */
OMPI_ERRHANDLER_RETURN(rc, fh, rc, FUNC_NAME);
}