2003-11-22 19:36:58 +03:00
# -*- shell-script -*-
#
2009-05-27 00:49:35 +04:00
# Copyright (c) 2004-2009 The Trustees of Indiana University and Indiana
2005-11-05 22:57:48 +03:00
# University Research and Technology
# Corporation. All rights reserved.
2011-01-13 09:08:54 +03:00
# Copyright (c) 2004-2010 The University of Tennessee and The University
2005-11-05 22:57:48 +03:00
# of Tennessee Research Foundation. All rights
# reserved.
2007-06-19 09:03:11 +04:00
# Copyright (c) 2004-2007 High Performance Computing Center Stuttgart,
2004-11-28 23:09:25 +03:00
# University of Stuttgart. All rights reserved.
2005-03-24 15:43:37 +03:00
# Copyright (c) 2004-2005 The Regents of the University of California.
# All rights reserved.
2013-01-12 00:20:02 +04:00
# Copyright (c) 2006-2013 Cisco Systems, Inc. All rights reserved.
2008-09-10 16:58:30 +04:00
# Copyright (c) 2006-2008 Sun Microsystems, Inc. All rights reserved.
2011-04-07 23:52:19 +04:00
# Copyright (c) 2006-2011 Los Alamos National Security, LLC. All rights
# reserved.
2012-06-27 05:28:28 +04:00
# Copyright (c) 2009 Oak Ridge National Labs. All rights reserved.
2013-07-18 02:12:29 +04:00
# Copyright (c) 2011-2013 NVIDIA Corporation. All rights reserved.
2012-10-04 18:39:51 +04:00
# Copyright (c) 2012 Oracle and/or its affiliates. All rights reserved.
2004-11-22 04:38:40 +03:00
# $COPYRIGHT$
#
# Additional copyrights may follow
#
2004-01-07 10:49:23 +03:00
# $HEADER$
2003-11-22 19:36:58 +03:00
#
############################################################################
# Initialization, version number, and other random setup/init stuff
############################################################################
2010-09-18 03:04:06 +04:00
# Load in everything found by autogen.pl
m4_include([config/autogen_found_items.m4])
# Load the version code. Because this is also used directly as a
# shell script, no ac_defun
m4_include([config/opal_get_version.m4])
2010-09-29 02:24:25 +04:00
AC_LANG([C])
Make the hwloc paffinity component available for everyone. hwloc
supports a wide variety of operating systems and platforms; see the
opal/mca/paffinity/hwloc/hwloc/README file for details.
This component includes an embedded copy of hwloc, currently based on
hwloc-1.0rc6. But note that hwloc is properly SVN imported into the
/vendor branch, so it will be easy to update when 1.0 GA is released.
Note that the hwloc tree embedded in opal/mca/paffinity/hwloc/hwloc is
identical to a hwloc distribution tarball, except that much of the
documentation was rm -rf'ed (because we don't need it for the embedded
case).
Since the paffinity framework currently does not understand hardware
threads, the hwloc component compensates for this by identifying cores
by the "first" hardware thread on that core. Hopefully we'll update
paffinity someday to understand hardware threads. :-)
configure grew a --with-hwloc option, analogous to what we do for many
other external libraries that OMPI supports. However, there's a new
feature: due to the request of several distros, OMPI can be configured
to build with its internal copy of hwloc or with an external copy of
hwloc (e.g., a system-installed hwloc).
1. If --with-hwloc is not specified, Open MPI will try to use its
internal copy (but silently fail/ignore hwloc if that fails).
1. If --with-hwloc=<dir> is supplied, Open MPI looks for hwloc
support in <dir> (and --with-hwloc-libdir=<dir>, if specified).
1. If --with-hwloc=external is supplied, Open MPI will look for hwloc
in a compiler/linker default external location.
1. If --with-hwloc=internal is supplied, Open MPI will use its
internal copy of hwloc.
Some of OMPI's main configury had to be slightly re-arranged in the
bootstrapping phase to accomodate hwloc's configry needs.
This commit was SVN r23125.
2010-05-14 03:56:05 +04:00
2003-11-22 19:36:58 +03:00
# Init autoconf
2005-09-07 09:54:53 +04:00
# We don't have the version number to put in here yet, and we can't
2010-09-18 03:04:06 +04:00
# call OPAL_GET_VERSION (etc.) before AC_INIT. So use the shell
2009-10-21 03:44:20 +04:00
# version. project_name_* comes from config/project_list.m4, which
# was set during autogen.sh.
2005-09-07 09:54:53 +04:00
2009-10-21 03:44:20 +04:00
AC_INIT([project_name_long],
2010-09-18 03:04:06 +04:00
[m4_normalize(esyscmd([config/opal_get_version.sh VERSION --base]))],
2009-10-21 03:44:20 +04:00
[http://www.open-mpi.org/community/help/], [project_name_short])
2007-05-15 08:23:48 +04:00
AC_PREREQ(2.60)
2006-05-07 23:19:37 +04:00
AC_CONFIG_AUX_DIR(config)
AC_CONFIG_MACRO_DIR(config)
2003-11-22 19:36:58 +03:00
2012-05-24 22:12:33 +04:00
# Get our platform support file. This has to be done very, very early
# because it twiddles random bits of autoconf
OMPI_LOAD_PLATFORM
Make the hwloc paffinity component available for everyone. hwloc
supports a wide variety of operating systems and platforms; see the
opal/mca/paffinity/hwloc/hwloc/README file for details.
This component includes an embedded copy of hwloc, currently based on
hwloc-1.0rc6. But note that hwloc is properly SVN imported into the
/vendor branch, so it will be easy to update when 1.0 GA is released.
Note that the hwloc tree embedded in opal/mca/paffinity/hwloc/hwloc is
identical to a hwloc distribution tarball, except that much of the
documentation was rm -rf'ed (because we don't need it for the embedded
case).
Since the paffinity framework currently does not understand hardware
threads, the hwloc component compensates for this by identifying cores
by the "first" hardware thread on that core. Hopefully we'll update
paffinity someday to understand hardware threads. :-)
configure grew a --with-hwloc option, analogous to what we do for many
other external libraries that OMPI supports. However, there's a new
feature: due to the request of several distros, OMPI can be configured
to build with its internal copy of hwloc or with an external copy of
hwloc (e.g., a system-installed hwloc).
1. If --with-hwloc is not specified, Open MPI will try to use its
internal copy (but silently fail/ignore hwloc if that fails).
1. If --with-hwloc=<dir> is supplied, Open MPI looks for hwloc
support in <dir> (and --with-hwloc-libdir=<dir>, if specified).
1. If --with-hwloc=external is supplied, Open MPI will look for hwloc
in a compiler/linker default external location.
1. If --with-hwloc=internal is supplied, Open MPI will use its
internal copy of hwloc.
Some of OMPI's main configury had to be slightly re-arranged in the
bootstrapping phase to accomodate hwloc's configry needs.
This commit was SVN r23125.
2010-05-14 03:56:05 +04:00
#
# Start it up
#
2010-11-13 02:22:11 +03:00
OPAL_CONFIGURE_SETUP
Make the hwloc paffinity component available for everyone. hwloc
supports a wide variety of operating systems and platforms; see the
opal/mca/paffinity/hwloc/hwloc/README file for details.
This component includes an embedded copy of hwloc, currently based on
hwloc-1.0rc6. But note that hwloc is properly SVN imported into the
/vendor branch, so it will be easy to update when 1.0 GA is released.
Note that the hwloc tree embedded in opal/mca/paffinity/hwloc/hwloc is
identical to a hwloc distribution tarball, except that much of the
documentation was rm -rf'ed (because we don't need it for the embedded
case).
Since the paffinity framework currently does not understand hardware
threads, the hwloc component compensates for this by identifying cores
by the "first" hardware thread on that core. Hopefully we'll update
paffinity someday to understand hardware threads. :-)
configure grew a --with-hwloc option, analogous to what we do for many
other external libraries that OMPI supports. However, there's a new
feature: due to the request of several distros, OMPI can be configured
to build with its internal copy of hwloc or with an external copy of
hwloc (e.g., a system-installed hwloc).
1. If --with-hwloc is not specified, Open MPI will try to use its
internal copy (but silently fail/ignore hwloc if that fails).
1. If --with-hwloc=<dir> is supplied, Open MPI looks for hwloc
support in <dir> (and --with-hwloc-libdir=<dir>, if specified).
1. If --with-hwloc=external is supplied, Open MPI will look for hwloc
in a compiler/linker default external location.
1. If --with-hwloc=internal is supplied, Open MPI will use its
internal copy of hwloc.
Some of OMPI's main configury had to be slightly re-arranged in the
bootstrapping phase to accomodate hwloc's configry needs.
This commit was SVN r23125.
2010-05-14 03:56:05 +04:00
ompi_show_title "Configuring project_name_long"
#
# Setup some things that must be done before AM-INIT-AUTOMAKE
#
ompi_show_subtitle "Startup tests"
AC_CANONICAL_HOST
AC_CANONICAL_TARGET
AC_DEFINE_UNQUOTED(OPAL_ARCH, "$target", [OMPI architecture string])
AS_IF([test "$host" != "$target"],
[AC_MSG_WARN([Cross-compile detected])
AC_MSG_WARN([Cross-compiling is only partially supported])
AC_MSG_WARN([Proceed at your own risk!])])
2010-05-20 04:25:23 +04:00
# AC_USE_SYSTEM_EXTENSIONS alters CFLAGS (e.g., adds -g -O2)
CFLAGS_save=$CFLAGS
Make the hwloc paffinity component available for everyone. hwloc
supports a wide variety of operating systems and platforms; see the
opal/mca/paffinity/hwloc/hwloc/README file for details.
This component includes an embedded copy of hwloc, currently based on
hwloc-1.0rc6. But note that hwloc is properly SVN imported into the
/vendor branch, so it will be easy to update when 1.0 GA is released.
Note that the hwloc tree embedded in opal/mca/paffinity/hwloc/hwloc is
identical to a hwloc distribution tarball, except that much of the
documentation was rm -rf'ed (because we don't need it for the embedded
case).
Since the paffinity framework currently does not understand hardware
threads, the hwloc component compensates for this by identifying cores
by the "first" hardware thread on that core. Hopefully we'll update
paffinity someday to understand hardware threads. :-)
configure grew a --with-hwloc option, analogous to what we do for many
other external libraries that OMPI supports. However, there's a new
feature: due to the request of several distros, OMPI can be configured
to build with its internal copy of hwloc or with an external copy of
hwloc (e.g., a system-installed hwloc).
1. If --with-hwloc is not specified, Open MPI will try to use its
internal copy (but silently fail/ignore hwloc if that fails).
1. If --with-hwloc=<dir> is supplied, Open MPI looks for hwloc
support in <dir> (and --with-hwloc-libdir=<dir>, if specified).
1. If --with-hwloc=external is supplied, Open MPI will look for hwloc
in a compiler/linker default external location.
1. If --with-hwloc=internal is supplied, Open MPI will use its
internal copy of hwloc.
Some of OMPI's main configury had to be slightly re-arranged in the
bootstrapping phase to accomodate hwloc's configry needs.
This commit was SVN r23125.
2010-05-14 03:56:05 +04:00
AC_USE_SYSTEM_EXTENSIONS
2010-05-20 04:25:23 +04:00
# AC_USE_SYSTEM_EXTENSIONS will modify CFLAGS if nothing was in there
# beforehand. We don't want that. So if there was nothing in
# CFLAGS, put nothing back in there.
2010-05-20 04:41:36 +04:00
AS_IF([test -z "$CFLAGS_save"], [CFLAGS=])
2010-05-20 04:25:23 +04:00
CFLAGS_save=
Make the hwloc paffinity component available for everyone. hwloc
supports a wide variety of operating systems and platforms; see the
opal/mca/paffinity/hwloc/hwloc/README file for details.
This component includes an embedded copy of hwloc, currently based on
hwloc-1.0rc6. But note that hwloc is properly SVN imported into the
/vendor branch, so it will be easy to update when 1.0 GA is released.
Note that the hwloc tree embedded in opal/mca/paffinity/hwloc/hwloc is
identical to a hwloc distribution tarball, except that much of the
documentation was rm -rf'ed (because we don't need it for the embedded
case).
Since the paffinity framework currently does not understand hardware
threads, the hwloc component compensates for this by identifying cores
by the "first" hardware thread on that core. Hopefully we'll update
paffinity someday to understand hardware threads. :-)
configure grew a --with-hwloc option, analogous to what we do for many
other external libraries that OMPI supports. However, there's a new
feature: due to the request of several distros, OMPI can be configured
to build with its internal copy of hwloc or with an external copy of
hwloc (e.g., a system-installed hwloc).
1. If --with-hwloc is not specified, Open MPI will try to use its
internal copy (but silently fail/ignore hwloc if that fails).
1. If --with-hwloc=<dir> is supplied, Open MPI looks for hwloc
support in <dir> (and --with-hwloc-libdir=<dir>, if specified).
1. If --with-hwloc=external is supplied, Open MPI will look for hwloc
in a compiler/linker default external location.
1. If --with-hwloc=internal is supplied, Open MPI will use its
internal copy of hwloc.
Some of OMPI's main configury had to be slightly re-arranged in the
bootstrapping phase to accomodate hwloc's configry needs.
This commit was SVN r23125.
2010-05-14 03:56:05 +04:00
2005-09-07 09:54:53 +04:00
#
2007-05-15 08:23:48 +04:00
# Init automake
2005-09-07 09:54:53 +04:00
#
2010-10-20 02:45:54 +04:00
AM_INIT_AUTOMAKE([foreign dist-bzip2 subdir-objects no-define 1.11 tar-ustar])
2005-09-07 09:54:53 +04:00
2010-10-20 02:45:54 +04:00
# SILENT_RULES is new in AM 1.11, but we require 1.11 or higher via
# autogen. Limited testing shows that calling SILENT_RULES directly
# works in more cases than adding "silent-rules" to INIT_AUTOMAKE
# (even though they're supposed to be identical). Shrug.
2010-09-25 02:34:53 +04:00
AM_SILENT_RULES([yes])
2009-11-04 05:07:02 +03:00
2005-09-16 06:49:31 +04:00
# Make configure depend on the VERSION file, since it's used in AC_INIT
2005-09-07 09:54:53 +04:00
AC_SUBST([CONFIGURE_DEPENDENCIES], ['$(top_srcdir)/VERSION'])
2009-10-21 03:44:20 +04:00
# Set up project specific AM_CONDITIONALs
2009-10-23 05:16:45 +04:00
AM_CONDITIONAL([PROJECT_OMPI], m4_ifdef([project_ompi], [true], [false]))
AM_CONDITIONAL([PROJECT_ORTE], m4_ifdef([project_orte], [true], [false]))
2009-10-21 03:44:20 +04:00
2006-01-22 01:53:16 +03:00
ompi_show_subtitle "Checking versions"
2004-06-07 19:33:53 +04:00
# Get the version of OMPI that we are installing
2009-10-21 03:44:20 +04:00
m4_ifdef([project_ompi],
2010-09-18 03:04:06 +04:00
[OPAL_SAVE_VERSION([OMPI], [Open MPI], [$srcdir/VERSION],
2009-10-21 03:44:20 +04:00
[ompi/include/ompi/version.h])])
2006-03-12 07:35:01 +03:00
2009-10-21 03:44:20 +04:00
m4_ifdef([project_orte],
2010-09-18 03:04:06 +04:00
[OPAL_SAVE_VERSION([ORTE], [Open MPI Run-Time Environment],
2009-10-21 03:44:20 +04:00
[$srcdir/VERSION],
[orte/include/orte/version.h])])
2006-03-12 07:35:01 +03:00
2010-09-18 03:04:06 +04:00
OPAL_SAVE_VERSION([OPAL], [Open Portable Access Layer], [$srcdir/VERSION],
2006-03-12 07:35:01 +03:00
[opal/include/opal/version.h])
2004-06-07 19:33:53 +04:00
2009-07-24 05:04:17 +04:00
# Get shared library version numbers
. $srcdir/VERSION
2009-10-21 03:44:20 +04:00
m4_ifdef([project_ompi],
[AC_SUBST(libmpi_so_version)
AC_SUBST(libmpi_cxx_so_version)
== Highlights ==
1. New mpifort wrapper compiler: you can utilize mpif.h, use mpi, and use mpi_f08 through this one wrapper compiler
1. mpif77 and mpif90 still exist, but are sym links to mpifort and may be removed in a future release
1. The mpi module has been re-implemented and is significantly "mo' bettah"
1. The mpi_f08 module offers many, many improvements over mpif.h and the mpi module
This stuff is coming from a VERY long-lived mercurial branch (3 years!); it'll almost certainly take a few SVN commits and a bunch of testing before I get it correctly committed to the SVN trunk.
== More details ==
Craig Rasmussen and I have been working with the MPI-3 Fortran WG and Fortran J3 committees for a long, long time to make a prototype MPI-3 Fortran bindings implementation. We think we're at a stable enough state to bring this stuff back to the trunk, with the goal of including it in OMPI v1.7.
Special thanks go out to everyone who has been incredibly patient and helpful to us in this journey:
* Rolf Rabenseifner/HLRS (mastermind/genius behind the entire MPI-3 Fortran effort)
* The Fortran J3 committee
* Tobias Burnus/gfortran
* Tony !Goetz/Absoft
* Terry !Donte/Oracle
* ...and probably others whom I'm forgetting :-(
There's still opportunities for optimization in the mpi_f08 implementation, but by and large, it is as far along as it can be until Fortran compilers start implementing the new F08 dimension(..) syntax.
Note that gfortran is currently unsupported for the mpi_f08 module and the new mpi module. gfortran users will a) fall back to the same mpi module implementation that is in OMPI v1.5.x, and b) not get the new mpi_f08 module. The gfortran maintainers are actively working hard to add the necessary features to support both the new mpi_f08 module and the new mpi module implementations. This will take some time.
As mentioned above, ompi/mpi/f77 and ompi/mpi/f90 no longer exist. All the fortran bindings implementations have been collated under ompi/mpi/fortran; each implementation has its own subdirectory:
{{{
ompi/mpi/fortran/
base/ - glue code
mpif-h/ - what used to be ompi/mpi/f77
use-mpi-tkr/ - what used to be ompi/mpi/f90
use-mpi-ignore-tkr/ - new mpi module implementation
use-mpi-f08/ - new mpi_f08 module implementation
}}}
There's also a prototype 6-function-MPI implementation under use-mpi-f08-desc that emulates the new F08 dimension(..) syntax that isn't fully available in Fortran compilers yet. We did that to prove it to ourselves that it could be done once the compilers fully support it. This directory/implementation will likely eventually replace the use-mpi-f08 version.
Other things that were done:
* ompi_info grew a few new output fields to describe what level of Fortran support is included
* Existing Fortran examples in examples/ were renamed; new mpi_f08 examples were added
* The old Fortran MPI libraries were renamed:
* libmpi_f77 -> libmpi_mpifh
* libmpi_f90 -> libmpi_usempi
* The configury for Fortran was consolidated and significantly slimmed down. Note that the F77 env variable is now IGNORED for configure; you should only use FC. Example:
{{{
shell$ ./configure CC=icc CXX=icpc FC=ifort ...
}}}
All of this work was done in a Mercurial branch off the SVN trunk, and hosted at Bitbucket. This branch has got to be one of OMPI's longest-running branches. Its first commit was Tue Apr 07 23:01:46 2009 -0400 -- it's over 3 years old! :-) We think we've pulled in all relevant changes from the OMPI trunk (e.g., Fortran implementations of the new MPI-3 MPROBE stuff for mpif.h, use mpi, and use mpi_f08, and the recent Fujitsu Fortran patches).
I anticipate some instability when we bring this stuff into the trunk, simply because it touches a LOT of code in the MPI layer in the OMPI code base. We'll try our best to make it as pain-free as possible, but please bear with us when it is committed.
This commit was SVN r26283.
2012-04-18 19:57:29 +04:00
AC_SUBST(libmpi_mpifh_so_version)
AC_SUBST(libmpi_usempi_tkr_so_version)
AC_SUBST(libmpi_usempi_ignore_tkr_so_version)
2012-02-21 02:12:43 +04:00
AC_SUBST(libmpi_java_so_version)
2009-10-27 23:58:34 +03:00
# It's icky that we have to hard-code the names of the
# common components here. :-( This could probably be done
# transparently by adding some intelligence in autogen.sh
# and/or ompi_mca.m4, but I don't have the cycles to do this
# right now.
AC_SUBST(libmca_common_sm_so_version)
AC_SUBST(libmca_common_mx_so_version)
2011-08-04 14:15:45 +04:00
AC_SUBST(libmca_common_cuda_so_version)
2009-10-27 23:58:34 +03:00
AC_SUBST(libmca_common_portals_so_version)])
2009-10-21 03:44:20 +04:00
m4_ifdef([project_orte],
[AC_SUBST(libopen_rte_so_version)])
2011-02-23 02:21:48 +03:00
AC_SUBST(libmca_opal_common_hwloc_so_version)
2009-07-24 05:04:17 +04:00
AC_SUBST(libopen_pal_so_version)
2012-02-17 06:55:06 +04:00
#
# Get the versions of the autotools that were used to bootstrap us
# (helpful for debugging reports)
#
AC_MSG_CHECKING([for bootstrap Autoconf version])
acversion=`grep "Generated by GNU Autoconf" $0 | head -n 1 | awk '{ print $6 }'`
AC_MSG_RESULT([$acversion])
AC_MSG_CHECKING([for bootstrap Automake version])
AC_MSG_RESULT([$am__api_version])
AC_MSG_CHECKING([for boostrap Libtool version])
ltversion=`grep VERSION= $srcdir/config/ltmain.sh | head -n 1 | cut -d= -f2`
AC_MSG_RESULT([$ltversion])
2009-07-24 05:04:17 +04:00
# List header files to generate
2013-01-12 00:20:02 +04:00
AC_CONFIG_HEADERS([opal/include/opal_config.h])
2009-10-21 03:44:20 +04:00
m4_ifdef([project_orte],
2013-01-12 00:20:02 +04:00
[AC_CONFIG_HEADERS([orte/include/orte_config.h])])
2009-10-21 03:44:20 +04:00
m4_ifdef([project_ompi],
2013-01-12 00:20:02 +04:00
[AC_CONFIG_HEADERS([ompi/include/ompi_config.h ompi/include/mpi.h])])
2007-06-06 02:07:30 +04:00
2005-09-07 09:54:53 +04:00
# override/fixup the version numbers set by AC_INIT, since on
# developer builds, there's no good way to know what the version is
# before running configure :(. We only use the base version number
# (ie, no svn r numbers) for the version set in AC_INIT. This will
# always match reality because we add the VERSION file (the only way
2005-09-27 06:06:05 +04:00
# to change the major.minor.release{greek}) into the configure
2006-03-12 07:35:01 +03:00
# dependencies. PACKAGE_VERSION the AC_DEFINE doesn't change once set
# the first time -- AC_INIT's input (so it doesn't have an r number in
# it). PACKAGE_VERSION the AC_SUBST can be rewritten along the way,
# and we'd like it to have the r number in it so that it shows up in
# the tarball name, so it is set to the full version here.
PACKAGE_VERSION="$OPAL_VERSION"
2005-09-07 09:54:53 +04:00
PACKAGE_STRING="${PACKAGE_NAME} ${PACKAGE_VERSION}"
VERSION="${PACKAGE_VERSION}"
2004-06-07 19:33:53 +04:00
ompi_show_subtitle "Initialization, setup"
2003-11-22 19:36:58 +03:00
2004-06-07 19:33:53 +04:00
OMPI_TOP_BUILDDIR="`pwd`"
AC_SUBST(OMPI_TOP_BUILDDIR)
2003-11-22 19:36:58 +03:00
cd "$srcdir"
2004-06-07 19:33:53 +04:00
OMPI_TOP_SRCDIR="`pwd`"
AC_SUBST(OMPI_TOP_SRCDIR)
cd "$OMPI_TOP_BUILDDIR"
2003-11-22 19:36:58 +03:00
2004-06-07 19:33:53 +04:00
AC_MSG_NOTICE([builddir: $OMPI_TOP_BUILDDIR])
AC_MSG_NOTICE([srcdir: $OMPI_TOP_SRCDIR])
if test "$OMPI_TOP_BUILDDIR" != "$OMPI_TOP_SRCDIR"; then
2003-11-22 19:36:58 +03:00
AC_MSG_NOTICE([Detected VPATH build])
fi
2006-02-12 04:33:29 +03:00
# Setup the top of the opal/include/opal_config.h file
2003-11-22 19:36:58 +03:00
AH_TOP([/* -*- c -*-
*
2004-11-22 04:38:40 +03:00
* Copyright (c) 2004-2005 The Trustees of Indiana University.
* All rights reserved.
* Copyright (c) 2004-2005 The Trustees of the University of Tennessee.
* All rights reserved.
2009-03-16 05:04:22 +03:00
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
2004-11-28 23:09:25 +03:00
* University of Stuttgart. All rights reserved.
2005-03-24 15:43:37 +03:00
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
2004-11-22 04:38:40 +03:00
* $COPYRIGHT$
2009-03-16 05:04:22 +03:00
*
2004-11-22 04:38:40 +03:00
* Additional copyrights may follow
2009-03-16 05:04:22 +03:00
*
2004-01-07 17:55:23 +03:00
* $HEADER$
2003-11-22 19:36:58 +03:00
*
2009-03-16 05:04:22 +03:00
* Function: - OS, CPU and compiler dependent configuration
2003-11-22 19:36:58 +03:00
*/
2006-02-12 04:33:29 +03:00
#ifndef OPAL_CONFIG_H
#define OPAL_CONFIG_H
2011-03-25 01:39:56 +03:00
#include "opal_config_top.h"
2003-12-22 19:29:21 +03:00
])
AH_BOTTOM([
2006-02-12 04:33:29 +03:00
#include "opal_config_bottom.h"
#endif /* OPAL_CONFIG_H */
2003-11-22 19:36:58 +03:00
])
2004-08-02 04:24:22 +04:00
# Other basic setup stuff (shared with components)
2003-11-22 19:36:58 +03:00
2010-11-13 02:22:11 +03:00
OPAL_BASIC_SETUP
2003-11-22 19:36:58 +03:00
2004-06-07 19:33:53 +04:00
top_ompi_srcdir="$OMPI_TOP_SRCDIR"
AC_SUBST(top_ompi_srcdir)
top_ompi_builddir="$OMPI_TOP_BUILDDIR"
AC_SUBST(top_ompi_builddir)
2003-11-22 19:36:58 +03:00
############################################################################
# Configuration options
############################################################################
2009-05-27 07:03:18 +04:00
OPAL_CONFIGURE_OPTIONS
2013-07-18 02:12:29 +04:00
OPAL_CHECK_CUDA
2009-10-21 03:44:20 +04:00
m4_ifdef([project_orte], [ORTE_CONFIGURE_OPTIONS])
2013-01-15 20:10:24 +04:00
m4_ifdef([project_ompi], [OMPI_CONFIGURE_OPTIONS])
2003-11-22 19:36:58 +03:00
2008-03-25 15:01:36 +03:00
if test "$enable_binaries" = "no" -a "$enable_dist" = "yes"; then
AC_MSG_WARN([--disable-binaries is incompatible with --enable dist])
AC_MSG_ERROR([Cannot continue])
fi
2003-11-22 19:36:58 +03:00
############################################################################
# Libtool: part one
# (before C compiler setup)
############################################################################
#
# Part one of libtool magic. Enable static so that we have the --with
# tests done up here and can check for OS. Save the values of
# $enable_static and $enable_shared before setting the defaults,
# because if the user specified --[en|dis]able-[static|shared] on the
# command line, they'll already be set. In this way, we can tell if
# the user requested something or if the default was set here.
#
2004-06-07 19:33:53 +04:00
ompi_enable_shared="$enable_shared"
ompi_enable_static="$enable_static"
2004-07-01 20:25:44 +04:00
AM_ENABLE_SHARED
AM_DISABLE_STATIC
2003-11-22 19:36:58 +03:00
2009-10-24 05:04:35 +04:00
OPAL_SETUP_WRAPPER_INIT
2005-04-18 20:38:27 +04:00
2009-01-28 04:06:53 +03:00
##################################
# Check for known incompatibility
##################################
# Do *not* print a message that we're checking the OS because this
# test is *not* meant to be an all-inclusive "if it passes this test,
# then configure must succeed" test. This test is *only* mean to
# screen out the versions of OS X where we know OMPI will cause kernel
# panics because of bad implementations of pty's. See
# https://svn.open-mpi.org/trac/ompi/ticket/1637 for details.
2012-02-21 19:56:19 +04:00
# OS X name OS X Version $host_os value
# OS X Tiger 10.4.x darwin8.x
# OS X Leopard 10.5.x darwin9.x
# OS X Snow Leopard 10.6.x darwin10.x
# OS X Lion 10.7.x darwin11.x
# We do not support OS X before version 10.5 (Leopard)
2009-01-28 04:06:53 +03:00
case $host_os in
2012-02-21 19:56:19 +04:00
# Corresponds to OS X 10.0 - 10.4 (additional [] quoting for m4)
darwin[[45678]]*)
AC_MSG_WARN([Open MPI does not support OS X prior to version 10.5 (Leopard)])
2009-01-28 04:06:53 +03:00
AC_MSG_ERROR([Cannot continue])
esac
2003-11-22 19:36:58 +03:00
############################################################################
# Check for compilers and preprocessors
############################################################################
2004-06-07 19:33:53 +04:00
ompi_show_title "Compiler and preprocessor tests"
2003-11-22 19:36:58 +03:00
##################################
# C compiler characteristics
##################################
2010-07-27 02:09:24 +04:00
OPAL_SETUP_CC
2003-11-22 19:36:58 +03:00
2005-12-18 01:29:12 +03:00
# If we build on a windows environment with the windows compiler and linker
# then we need some translation functions from the opal/win32 directory.
AM_CONDITIONAL(OMPI_NEED_WINDOWS_REPLACEMENTS,
test "$ompi_cv_c_compiler_vendor" = "microsoft" )
2008-03-27 02:20:33 +03:00
# Do all Interix detections if necessary
OMPI_INTERIX
2007-11-03 05:40:22 +03:00
# Does the compiler support "ident"-like constructs?
OMPI_CHECK_IDENT([CC], [CFLAGS], [c], [C])
2004-02-13 21:05:49 +03:00
#
# Check for some types
#
2004-01-09 04:17:00 +03:00
AC_CHECK_TYPES(int8_t)
AC_CHECK_TYPES(uint8_t)
AC_CHECK_TYPES(int16_t)
AC_CHECK_TYPES(uint16_t)
AC_CHECK_TYPES(int32_t)
AC_CHECK_TYPES(uint32_t)
AC_CHECK_TYPES(int64_t)
AC_CHECK_TYPES(uint64_t)
- 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
AC_CHECK_TYPES(int128_t)
AC_CHECK_TYPES(uint128_t)
AC_CHECK_TYPES(long long)
2011-08-26 01:56:05 +04:00
AC_CHECK_TYPES(__float128)
- 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
AC_CHECK_TYPES(long double)
2009-10-21 03:44:20 +04:00
# We only need these types if we're building the OMPI project, but
# OPAL currently doesn't protect for their lack of presence well.
== Highlights ==
1. New mpifort wrapper compiler: you can utilize mpif.h, use mpi, and use mpi_f08 through this one wrapper compiler
1. mpif77 and mpif90 still exist, but are sym links to mpifort and may be removed in a future release
1. The mpi module has been re-implemented and is significantly "mo' bettah"
1. The mpi_f08 module offers many, many improvements over mpif.h and the mpi module
This stuff is coming from a VERY long-lived mercurial branch (3 years!); it'll almost certainly take a few SVN commits and a bunch of testing before I get it correctly committed to the SVN trunk.
== More details ==
Craig Rasmussen and I have been working with the MPI-3 Fortran WG and Fortran J3 committees for a long, long time to make a prototype MPI-3 Fortran bindings implementation. We think we're at a stable enough state to bring this stuff back to the trunk, with the goal of including it in OMPI v1.7.
Special thanks go out to everyone who has been incredibly patient and helpful to us in this journey:
* Rolf Rabenseifner/HLRS (mastermind/genius behind the entire MPI-3 Fortran effort)
* The Fortran J3 committee
* Tobias Burnus/gfortran
* Tony !Goetz/Absoft
* Terry !Donte/Oracle
* ...and probably others whom I'm forgetting :-(
There's still opportunities for optimization in the mpi_f08 implementation, but by and large, it is as far along as it can be until Fortran compilers start implementing the new F08 dimension(..) syntax.
Note that gfortran is currently unsupported for the mpi_f08 module and the new mpi module. gfortran users will a) fall back to the same mpi module implementation that is in OMPI v1.5.x, and b) not get the new mpi_f08 module. The gfortran maintainers are actively working hard to add the necessary features to support both the new mpi_f08 module and the new mpi module implementations. This will take some time.
As mentioned above, ompi/mpi/f77 and ompi/mpi/f90 no longer exist. All the fortran bindings implementations have been collated under ompi/mpi/fortran; each implementation has its own subdirectory:
{{{
ompi/mpi/fortran/
base/ - glue code
mpif-h/ - what used to be ompi/mpi/f77
use-mpi-tkr/ - what used to be ompi/mpi/f90
use-mpi-ignore-tkr/ - new mpi module implementation
use-mpi-f08/ - new mpi_f08 module implementation
}}}
There's also a prototype 6-function-MPI implementation under use-mpi-f08-desc that emulates the new F08 dimension(..) syntax that isn't fully available in Fortran compilers yet. We did that to prove it to ourselves that it could be done once the compilers fully support it. This directory/implementation will likely eventually replace the use-mpi-f08 version.
Other things that were done:
* ompi_info grew a few new output fields to describe what level of Fortran support is included
* Existing Fortran examples in examples/ were renamed; new mpi_f08 examples were added
* The old Fortran MPI libraries were renamed:
* libmpi_f77 -> libmpi_mpifh
* libmpi_f90 -> libmpi_usempi
* The configury for Fortran was consolidated and significantly slimmed down. Note that the F77 env variable is now IGNORED for configure; you should only use FC. Example:
{{{
shell$ ./configure CC=icc CXX=icpc FC=ifort ...
}}}
All of this work was done in a Mercurial branch off the SVN trunk, and hosted at Bitbucket. This branch has got to be one of OMPI's longest-running branches. Its first commit was Tue Apr 07 23:01:46 2009 -0400 -- it's over 3 years old! :-) We think we've pulled in all relevant changes from the OMPI trunk (e.g., Fortran implementations of the new MPI-3 MPROBE stuff for mpif.h, use mpi, and use mpi_f08, and the recent Fujitsu Fortran patches).
I anticipate some instability when we bring this stuff into the trunk, simply because it touches a LOT of code in the MPI layer in the OMPI code base. We'll try our best to make it as pain-free as possible, but please bear with us when it is committed.
This commit was SVN r26283.
2012-04-18 19:57:29 +04:00
AC_CHECK_HEADERS(complex.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
AC_CHECK_TYPES(float _Complex)
AC_CHECK_TYPES(double _Complex)
AC_CHECK_TYPES(long double _Complex)
2004-01-09 04:17:00 +03:00
AC_CHECK_TYPES(intptr_t)
AC_CHECK_TYPES(uintptr_t)
2004-10-14 00:47:52 +04:00
AC_CHECK_TYPES(mode_t)
2006-10-20 07:24:59 +04:00
AC_CHECK_TYPES(ssize_t)
AC_CHECK_TYPES(ptrdiff_t)
2005-05-12 05:42:24 +04:00
2004-02-13 21:05:49 +03:00
#
# Check for type sizes
#
2004-01-09 04:17:00 +03:00
AC_CHECK_SIZEOF(char)
AC_CHECK_SIZEOF(short)
AC_CHECK_SIZEOF(int)
AC_CHECK_SIZEOF(long)
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type_long_long" = yes; then
2004-04-24 00:52:42 +04:00
AC_CHECK_SIZEOF(long long)
fi
- 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
AC_CHECK_SIZEOF(float)
AC_CHECK_SIZEOF(double)
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type_long_double" = yes; then
2004-04-24 00:52:42 +04:00
AC_CHECK_SIZEOF(long double)
fi
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type___float128" = yes; then
AC_CHECK_SIZEOF(__float128)
fi
2009-10-21 03:44:20 +04:00
# We only need these types if we're building the OMPI project, but
# OPAL currently doesn't protect for their lack of presence well.
- 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
if test "$ac_cv_type_float__Complex" = yes; then
AC_CHECK_SIZEOF(float _Complex)
fi
if test "$ac_cv_type_double__Complex" = yes; then
AC_CHECK_SIZEOF(double _Complex)
fi
if test "$ac_cv_type_long_double__Complex" = yes; then
AC_CHECK_SIZEOF(long double _Complex)
fi
2004-01-09 04:17:00 +03:00
AC_CHECK_SIZEOF(void *)
2005-04-05 06:32:36 +04:00
AC_CHECK_SIZEOF(size_t)
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type_ssize_t" = yes ; then
2006-10-20 07:24:59 +04:00
AC_CHECK_SIZEOF(ssize_t)
fi
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type_ptrdiff_t" = yes; then
2006-10-20 07:24:59 +04:00
AC_CHECK_SIZEOF(ptrdiff_t)
fi
- 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
AC_CHECK_SIZEOF(wchar_t)
2004-02-13 21:05:49 +03:00
#
# Check for type alignments
#
- 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
OMPI_C_GET_ALIGNMENT(_Bool, OPAL_ALIGNMENT_BOOL)
OMPI_C_GET_ALIGNMENT(int8_t, OPAL_ALIGNMENT_INT8)
OMPI_C_GET_ALIGNMENT(int16_t, OPAL_ALIGNMENT_INT16)
OMPI_C_GET_ALIGNMENT(int32_t, OPAL_ALIGNMENT_INT32)
OMPI_C_GET_ALIGNMENT(int64_t, OPAL_ALIGNMENT_INT64)
if test "$ac_cv_type_int128_t" = yes ; then
OMPI_C_GET_ALIGNMENT(int128_t, OPAL_ALIGNMENT_INT128)
fi
2009-05-07 00:11:28 +04:00
OMPI_C_GET_ALIGNMENT(char, OPAL_ALIGNMENT_CHAR)
OMPI_C_GET_ALIGNMENT(short, OPAL_ALIGNMENT_SHORT)
- 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
OMPI_C_GET_ALIGNMENT(wchar_t, OPAL_ALIGNMENT_WCHAR)
2009-05-07 00:11:28 +04:00
OMPI_C_GET_ALIGNMENT(int, OPAL_ALIGNMENT_INT)
OMPI_C_GET_ALIGNMENT(long, OPAL_ALIGNMENT_LONG)
- 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
if test "$ac_cv_type_long_long" = yes; then
2009-05-07 00:11:28 +04:00
OMPI_C_GET_ALIGNMENT(long long, OPAL_ALIGNMENT_LONG_LONG)
2004-04-24 00:52:42 +04:00
fi
- 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
OMPI_C_GET_ALIGNMENT(float, OPAL_ALIGNMENT_FLOAT)
OMPI_C_GET_ALIGNMENT(double, OPAL_ALIGNMENT_DOUBLE)
if test "$ac_cv_type_long_double" = yes; then
2009-05-07 00:11:28 +04:00
OMPI_C_GET_ALIGNMENT(long double, OPAL_ALIGNMENT_LONG_DOUBLE)
2004-04-24 00:52:42 +04:00
fi
2011-08-26 01:56:05 +04:00
if test "$ac_cv_type___float128" = yes; then
OMPI_C_GET_ALIGNMENT(__float128, OPAL_ALIGNMENT___FLOAT128)
fi
- 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
2009-10-21 03:44:20 +04:00
# We only need these types if we're building the OMPI project, but
# OPAL currently doesn't protect for their lack of presence well.
- 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
if test "$ac_cv_type_float__Complex" = yes; then
OMPI_C_GET_ALIGNMENT(float _Complex, OPAL_ALIGNMENT_FLOAT_COMPLEX)
fi
if test "$ac_cv_type_double__Complex" = yes; then
OMPI_C_GET_ALIGNMENT(double _Complex, OPAL_ALIGNMENT_DOUBLE_COMPLEX)
fi
if test "$ac_cv_type_long_double__Complex" = yes; then
OMPI_C_GET_ALIGNMENT(long double _Complex, OPAL_ALIGNMENT_LONG_DOUBLE_COMPLEX)
fi
OMPI_C_GET_ALIGNMENT(void *, OPAL_ALIGNMENT_VOID_P)
2012-10-04 18:39:51 +04:00
OMPI_C_GET_ALIGNMENT(size_t, OPAL_ALIGNMENT_SIZE_T)
2004-02-13 21:05:49 +03:00
2004-10-23 00:30:59 +04:00
#
# Does the C compiler native support "bool"? (i.e., without
# <stdbool.h> or any other help)
#
AC_MSG_CHECKING(for C bool type)
2010-09-24 02:37:52 +04:00
AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
AC_INCLUDES_DEFAULT],
[[bool bar, foo = true; bar = foo;]])],
2009-05-07 00:11:28 +04:00
[OPAL_NEED_C_BOOL=0 MSG=yes],[OPAL_NEED_C_BOOL=1 MSG=no])
AC_DEFINE_UNQUOTED(OPAL_NEED_C_BOOL, $OPAL_NEED_C_BOOL,
2004-10-23 00:30:59 +04:00
[Whether the C compiler supports "bool" without any other help (such as <stdbool.h>)])
AC_MSG_RESULT([$MSG])
2009-08-04 15:54:01 +04:00
AC_CHECK_SIZEOF(_Bool)
2004-10-23 00:30:59 +04:00
2004-02-13 21:05:49 +03:00
#
# Check for other compiler characteristics
#
2004-01-07 10:49:23 +03:00
2004-06-15 21:10:47 +04:00
if test "$GCC" = "yes"; then
# gcc 2.96 will emit oodles of warnings if you use "inline" with
# -pedantic (which we do in developer builds). However,
# "__inline__" is ok. So we have to force gcc to select the
# right one. If you use -pedantic, the AC_C_INLINE test will fail
# (because it names a function foo() -- without the (void)). So
# we turn off all the picky flags, turn on -ansi mode (which is
# implied by -pedantic), and set warnings to be errors. Hence,
# this does the following (for 2.96):
#
# - causes the check for "inline" to emit a warning, which then
# fails
# - checks for __inline__, which then emits no error, and works
#
# This also works nicely for gcc 3.x because "inline" will work on
# the first check, and all is fine. :-)
CFLAGS_save="$CFLAGS"
CFLAGS="$OMPI_CFLAGS_BEFORE_PICKY -Werror -ansi"
fi
2003-12-22 19:29:21 +03:00
AC_C_INLINE
2006-08-22 02:13:32 +04:00
# Microsoft compilers support 2 versions of restrict. One for functions, and
# one for variables. The problem is that they don't have an equivalent
# syntax, and the autoconf restrict detection is unable to detect them
# correctly. It detect the restrict keyword as __restrict which break the
# rules for function syntax which is declspec(restrict).
if test "x$ompi_cv_c_compiler_vendor" != "xmicrosoft"; then
2009-03-16 07:22:22 +03:00
AC_C_RESTRICT
2006-08-22 02:13:32 +04:00
fi
2004-06-07 19:33:53 +04:00
OMPI_C_WEAK_SYMBOLS
2004-06-15 21:10:47 +04:00
if test "$GCC" = "yes"; then
CFLAGS="$CFLAGS_save"
fi
2003-11-22 19:36:58 +03:00
2005-05-10 13:09:04 +04:00
if test "x$CC" = "xicc"; then
2009-03-16 07:22:22 +03:00
OMPI_CHECK_ICC_VARARGS
2005-05-10 13:09:04 +04:00
fi
2004-01-08 16:36:14 +03:00
# If we want the profiling layer:
# - If the C compiler has weak symbols, use those.
# - If not, then set to compile the code again with #define's in a
# separate directory.
if test "$WANT_WEAK_SYMBOLS" = "0"; then
2009-05-07 00:11:28 +04:00
OPAL_C_HAVE_WEAK_SYMBOLS=0
2004-01-08 16:36:14 +03:00
fi
if test "$WANT_MPI_PROFILING" = "1"; then
2009-05-07 00:11:28 +04:00
if test "$OPAL_C_HAVE_WEAK_SYMBOLS" = "1"; then
2009-03-16 07:22:22 +03:00
OMPI_PROFILING_COMPILE_SEPARATELY=0
2004-01-08 16:36:14 +03:00
else
2009-03-16 07:22:22 +03:00
OMPI_PROFILING_COMPILE_SEPARATELY=1
2004-01-08 16:36:14 +03:00
fi
else
2004-06-07 19:33:53 +04:00
OMPI_PROFILING_COMPILE_SEPARATELY=0
2004-01-08 16:36:14 +03:00
fi
2004-06-29 04:02:25 +04:00
2009-06-28 03:36:25 +04:00
# Check if we support the offsetof compiler directive
OPAL_CHECK_OFFSETOF
2009-10-21 05:13:21 +04:00
# Setup OMPI bindings (if we're building the OMPI project). Note that
# opal_wrapper.c has a hard-coded use of the OMPI_ENABLE_MPI_PROFILING
# macro, so we need to define it (to 0) even if we're not building the
# OMPI project.
m4_ifdef([project_ompi], [OMPI_SETUP_MPI_PROFILING],
[AC_DEFINE([OMPI_ENABLE_MPI_PROFILING], [0],
[We are not building OMPI, so no profiling])])
2004-01-08 16:36:14 +03:00
2004-01-07 10:49:23 +03:00
2003-11-22 19:36:58 +03:00
##################################
# C++ compiler characteristics
##################################
2010-06-12 07:15:47 +04:00
# We don't need C++ unless we're building Open MPI; ORTE and OPAL do
# not use C++ at all. The OPAL macro name appears to be a bit of a
# misnomer; I'm not sure why it was split into a second macro and put
# into OPAL...? All it does is setup the C++ compiler (the OMPI macro
# sets up the C++ MPI bindings, etc.). Perhaps it was moved to OPAL
# just on the rationale that all compiler setup should be done in
# OPAL...? Shrug.
m4_ifdef([project_ompi], [OPAL_SETUP_CXX
OMPI_SETUP_CXX])
2007-02-08 16:34:44 +03:00
##################################
# Only after setting up both
# C and C++ check compiler attributes.
##################################
2009-08-04 15:54:01 +04:00
ompi_show_subtitle "Compiler characteristics"
2010-06-12 07:15:47 +04:00
OPAL_CHECK_ATTRIBUTES
- As proposed in RFC and telcon, warn the user about deprecated
functionality (per MPI-2.1). This warning can be toggled using
--enable-mpi-interface-warning (default OFF), but can be
selectively turned on passing
mpicc -DOMPI_WANT_MPI_INTERFACE_WARNING
Using icc, gcc < 4.5, warnings (such as in mpi2basic_tests) show:
type_vector.c:83: warning: ‘MPI_Type_hvector’ is deprecated
(declared at /home/../usr/include/mpi.h:1379)
Using gcc-4.5 (gcc-svn) these show up as:
type_vector.c:83: warning: ‘MPI_Type_hvector’ is deprecated
(declared at /home/../usr/include/mpi.h:1379):
MPI_Type_hvector is superseded by MPI_Type_create_hvector in MPI-2.0
Jeff and I propose to turn such warnings on with Open MPI-1.7 by default.
- Detection of user-level compiler is handled using the preprocessor
checks of GASnet's other/portable_platform.h (thanks to Paul Hargrove
and Dan Bonachea) adapted into ompi/include/mpi_portable_platform.h
(see comments).
The OMPI-build time detection is output (Familyname and Version)
with ompi_info.
This functionality (actually any upcoming __attribute__) are turned
off, if a different compiler (and version) is being detected.
- Note, that any warnings regarding (user-compiler!=build-compiler)
as discussed in the RFC are _not_ included for now.
- Tested on Linux with --enable-mpi-interface-warning on
Linux, gcc-4.5 (deprecated w/ specific msg)
Linux, gcc-4.3 (deprecated w/o specific msg)
Linux, pathscale 3.1 (deprecated w/o specific msg)
Linux, icc-11.0 (deprecated w/o specific msg)
Linux, PGI-8.0.6 accepts __deprecated__ but does not issue a warning,
further investigation needed...
This commit was SVN r21262.
2009-05-22 08:39:43 +04:00
OPAL_CHECK_COMPILER_VERSION_ID
2004-01-07 10:49:23 +03:00
2009-10-21 03:44:20 +04:00
2012-02-21 02:12:43 +04:00
##################################
# Java compiler characteristics
##################################
2012-05-09 05:13:49 +04:00
# Needed for ORTE - OMPI will AC_REQUIRE where needed
m4_ifdef([project_orte], [ORTE_SETUP_JAVA])
2012-02-21 02:12:43 +04:00
2012-05-09 05:13:49 +04:00
##################################
# Java MPI Binding request
##################################
# Only needed for OMPI
2012-06-27 05:28:28 +04:00
m4_ifdef([project_ompi], [OMPI_SETUP_JAVA_BINDINGS])
2012-05-09 05:13:49 +04:00
2005-01-27 04:39:55 +03:00
##################################
# Assembler Configuration
##################################
2009-03-16 05:04:22 +03:00
ompi_show_subtitle "Assembler"
2005-01-27 04:39:55 +03:00
AM_PROG_AS
2010-07-27 08:51:50 +04:00
OPAL_CONFIG_ASM
2005-01-27 04:39:55 +03:00
2003-11-22 19:36:58 +03:00
##################################
# Fortran
##################################
2013-01-15 20:10:24 +04:00
m4_ifdef([project_ompi], [OMPI_SETUP_MPI_FORTRAN])
2006-04-11 07:33:38 +04:00
2005-07-09 22:52:53 +04:00
# checkpoint results
AC_CACHE_SAVE
2009-10-21 03:44:20 +04:00
2003-11-22 19:36:58 +03:00
##################################
# Header files
##################################
2004-06-07 19:33:53 +04:00
ompi_show_title "Header file tests"
2004-03-28 14:52:58 +04:00
2005-04-16 01:18:20 +04:00
AC_CHECK_HEADERS([alloca.h aio.h arpa/inet.h dirent.h \
2006-03-11 05:35:40 +03:00
dlfcn.h execinfo.h err.h fcntl.h grp.h inttypes.h libgen.h \
2009-03-05 05:40:25 +03:00
libutil.h memory.h netdb.h netinet/in.h netinet/tcp.h \
2010-07-20 22:45:48 +04:00
poll.h pthread.h pty.h pwd.h sched.h stdint.h stddef.h \
2011-04-07 23:52:19 +04:00
stdlib.h string.h strings.h stropts.h sys/fcntl.h sys/ipc.h sys/shm.h \
2012-12-22 20:32:26 +04:00
sys/ioctl.h sys/mman.h sys/param.h sys/queue.h \
2005-04-16 01:18:20 +04:00
sys/resource.h sys/select.h sys/socket.h sys/sockio.h \
2010-02-11 02:18:29 +03:00
stdarg.h sys/stat.h sys/statfs.h sys/statvfs.h sys/time.h sys/tree.h \
sys/types.h sys/uio.h net/uio.h sys/utsname.h sys/vfs.h sys/wait.h syslog.h \
2007-04-25 05:55:40 +04:00
time.h termios.h ulimit.h unistd.h util.h utmp.h malloc.h \
2012-12-22 20:32:26 +04:00
ifaddrs.h crt_externs.h regex.h signal.h \
2012-02-21 02:12:43 +04:00
ioLib.h sockLib.h hostLib.h shlwapi.h sys/synch.h limits.h db.h ndbm.h \
TargetConditionals.h])
2004-09-29 23:42:16 +04:00
2012-12-22 20:32:26 +04:00
AC_CHECK_HEADERS([sys/mount.h], [], [],
[AC_INCLUDES_DEFAULT
#if HAVE_SYS_PARAM_H
#include <sys/param.h>
#endif
])
AC_CHECK_HEADERS([sys/sysctl.h], [], [],
[AC_INCLUDES_DEFAULT
#if HAVE_SYS_PARAM_H
#include <sys/param.h>
#endif
])
2006-01-12 06:19:37 +03:00
# Needed to work around Darwin requiring sys/socket.h for
# net/if.h
AC_CHECK_HEADERS([net/if.h], [], [],
[#include <stdio.h>
#if STDC_HEADERS
# include <stdlib.h>
# include <stddef.h>
#else
# if HAVE_STDLIB_H
# include <stdlib.h>
# endif
#endif
#if HAVE_SYS_SOCKET_H
# include <sys/socket.h>
#endif
])
2004-08-05 15:12:25 +04:00
# Note that sometimes we have <stdbool.h>, but it doesn't work (e.g.,
# have both Portland and GNU installed; using pgcc will find GNU's
# <stdbool.h>, which all it does -- by standard -- is define "bool" to
# "_Bool" [see
# http://www.opengroup.org/onlinepubs/009695399/basedefs/stdbool.h.html],
# and Portland has no idea what to do with _Bool).
# So first figure out if we have <stdbool.h> (i.e., check the value of
# the macro HAVE_STDBOOL_H from the result of AC_CHECK_HEADERS,
# above). If we do have it, then check to see if it actually works.
2009-05-07 00:11:28 +04:00
# Define OPAL_USE_STDBOOL_H as approrpaite.
2004-09-29 23:42:16 +04:00
AC_CHECK_HEADERS([stdbool.h], [have_stdbool_h=1], [have_stdbool_h=0])
2004-08-05 15:12:25 +04:00
AC_MSG_CHECKING([if <stdbool.h> works])
if test "$have_stdbool_h" = "1"; then
2010-09-24 02:37:52 +04:00
AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
AC_INCLUDES_DEFAULT[
2004-08-05 15:12:25 +04:00
#if HAVE_STDBOOL_H
#include <stdbool.h>
2009-03-16 05:04:22 +03:00
#endif]],
2010-09-24 02:37:52 +04:00
[[bool bar, foo = true; bar = foo;]])],
2009-05-07 00:11:28 +04:00
[OPAL_USE_STDBOOL_H=1 MSG=yes],[OPAL_USE_STDBOOL_H=0 MSG=no])
2004-08-05 15:12:25 +04:00
else
2009-05-07 00:11:28 +04:00
OPAL_USE_STDBOOL_H=0
2004-08-05 15:12:25 +04:00
MSG="no (don't have <stdbool.h>)"
fi
2009-05-07 00:11:28 +04:00
AC_DEFINE_UNQUOTED(OPAL_USE_STDBOOL_H, $OPAL_USE_STDBOOL_H,
2004-08-05 15:12:25 +04:00
[Whether to use <stdbool.h> or not])
AC_MSG_RESULT([$MSG])
2003-11-22 19:36:58 +03:00
2005-07-09 22:52:53 +04:00
# checkpoint results
AC_CACHE_SAVE
2003-11-22 19:36:58 +03:00
2005-07-13 08:16:03 +04:00
##################################
# Types
##################################
ompi_show_title "Type tests"
# Size of pid_t
AC_CHECK_SIZEOF(pid_t)
2009-03-16 05:04:22 +03:00
AC_CHECK_TYPES([socklen_t, struct sockaddr_in, struct sockaddr_in6,
2007-07-20 05:34:02 +04:00
struct sockaddr_storage],
[], [], [AC_INCLUDES_DEFAULT
#if HAVE_SYS_SOCKET_H
2007-05-21 01:54:43 +04:00
#include <sys/socket.h>
2007-07-20 05:34:02 +04:00
#endif
2007-05-17 05:17:59 +04:00
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif])
2007-07-20 05:34:02 +04:00
2009-03-16 05:04:22 +03:00
AC_CHECK_DECLS([AF_UNSPEC, PF_UNSPEC, AF_INET6, PF_INET6],
2007-07-20 05:34:02 +04:00
[], [], [AC_INCLUDES_DEFAULT
#if HAVE_SYS_SOCKET_H
2007-05-27 07:52:56 +04:00
#include <sys/socket.h>
2007-07-20 05:34:02 +04:00
#endif
2007-05-17 05:17:59 +04:00
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif])
2006-09-08 04:10:40 +04:00
2005-07-13 08:16:03 +04:00
# SA_RESTART in signal.h
AC_MSG_CHECKING([if SA_RESTART defined in signal.h])
AC_EGREP_CPP(yes, [
#include <signal.h>
#ifdef SA_RESTART
yes
#endif ], [MSG=yes VALUE=1], [MSG=no VALUE=0])
2009-05-07 00:11:28 +04:00
AC_DEFINE_UNQUOTED(OPAL_HAVE_SA_RESTART, $VALUE,
2005-07-13 08:16:03 +04:00
[Whether we have SA_RESTART in <signal.h> or not])
AC_MSG_RESULT([$MSG])
2007-07-20 05:34:02 +04:00
AC_CHECK_MEMBERS([struct sockaddr.sa_len], [], [], [
#include <sys/types.h>
#if HAVE_SYS_SOCKET_H
#include <sys/socket.h>
#endif])
2005-07-13 08:16:03 +04:00
AC_CHECK_MEMBERS([struct dirent.d_type], [], [], [
#include <sys/types.h>
#include <dirent.h>])
AC_CHECK_MEMBERS([siginfo_t.si_fd],,,[#include <signal.h>])
2005-12-15 03:51:28 +03:00
AC_CHECK_MEMBERS([siginfo_t.si_band],,,[#include <signal.h>])
2005-07-13 08:16:03 +04:00
# checkpoint results
AC_CACHE_SAVE
2003-11-22 19:36:58 +03:00
##################################
# Libraries
##################################
2004-06-07 19:33:53 +04:00
ompi_show_title "Library and Function tests"
2004-03-28 14:02:38 +04:00
2008-08-14 00:56:06 +04:00
# Darwin doesn't need -lutil, as it's something other than this -lutil.
OMPI_CHECK_FUNC_LIB([openpty], [util])
2005-08-25 01:24:17 +04:00
2006-01-16 04:48:03 +03:00
AC_CHECK_LIB([nsl], [gethostbyname])
2005-08-25 01:24:17 +04:00
2006-01-16 04:48:03 +03:00
AC_CHECK_LIB([socket], [socket])
2009-08-12 17:07:04 +04:00
# Solaris has sched_yield in -lrt, usually in libc
2006-01-16 04:48:03 +03:00
OMPI_CHECK_FUNC_LIB([sched_yield], [rt])
# IRIX has dirname in -lgen, usually in libc
OMPI_CHECK_FUNC_LIB([dirname], [gen])
# Darwin doesn't need -lm, as it's a symlink to libSystem.dylib
OMPI_CHECK_FUNC_LIB([ceil], [m])
2005-08-25 01:24:17 +04:00
2011-02-02 13:05:57 +03:00
AC_CHECK_FUNCS([asprintf snprintf vasprintf vsnprintf openpty isatty getpwuid fork waitpid execve pipe ptsname setsid mmap tcgetpgrp posix_memalign strsignal sysconf syslog vsyslog regcmp regexec regfree _NSGetEnviron socketpair strncpy_s _strdup usleep mkfifo dbopen dbm_open])
2008-06-25 03:20:25 +04:00
# On some hosts, htonl is a define, so the AC_CHECK_FUNC will get
# confused. On others, it's in the standard library, but stubbed with
# the magic glibc foo as not implemented. and on other systems, it's
# just not there. This covers all cases.
AC_CACHE_CHECK([for htonl define],
[ompi_cv_htonl_define],
[AC_PREPROC_IFELSE([AC_LANG_PROGRAM([
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif
#ifdef HAVE_ARPA_INET_H
#include <arpa/inet.h>
#endif],[
#ifndef ntohl
#error "ntohl not defined"
#endif
])], [ompi_cv_htonl_define=yes], [ompi_cv_htonl_define=no])])
AC_CHECK_FUNC([htonl], [ompi_have_htonl=yes], [ompi_have_htonl=no])
AS_IF([test "$ompi_cv_htonl_define" = "yes" -o "$ompi_have_htonl" = "yes"],
2009-03-16 05:04:22 +03:00
[AC_DEFINE_UNQUOTED([HAVE_UNIX_BYTESWAP], [1],
2008-06-25 03:20:25 +04:00
[whether unix byteswap routines -- htonl, htons, nothl, ntohs -- are available])])
2004-08-11 02:41:17 +04:00
2004-01-14 09:39:55 +03:00
#
# Make sure we can copy va_lists (need check declared, not linkable)
#
2009-05-07 00:11:28 +04:00
AC_CHECK_DECL(va_copy, OPAL_HAVE_VA_COPY=1, OPAL_HAVE_VA_COPY=0,
2009-03-16 07:22:22 +03:00
[#include <stdarg.h>])
2009-05-07 00:11:28 +04:00
AC_DEFINE_UNQUOTED(OPAL_HAVE_VA_COPY, $OPAL_HAVE_VA_COPY,
2004-01-14 09:39:55 +03:00
[Whether we have va_copy or not])
2009-05-07 00:11:28 +04:00
AC_CHECK_DECL(__va_copy, OPAL_HAVE_UNDERSCORE_VA_COPY=1,
OPAL_HAVE_UNDERSCORE_VA_COPY=0, [#include <stdarg.h>])
AC_DEFINE_UNQUOTED(OPAL_HAVE_UNDERSCORE_VA_COPY, $OPAL_HAVE_UNDERSCORE_VA_COPY,
2004-01-14 09:39:55 +03:00
[Whether we have __va_copy or not])
2004-10-29 23:14:11 +04:00
AC_CHECK_DECLS(__func__)
2005-07-09 22:52:53 +04:00
# checkpoint results
AC_CACHE_SAVE
2003-11-22 19:36:58 +03:00
##################################
# System-specific tests
##################################
2004-06-07 19:33:53 +04:00
ompi_show_title "System-specific tests"
2004-06-29 04:02:25 +04:00
2004-01-19 04:33:03 +03:00
#
# Test to determine type of MPI_Offset. This is searched in the following order
# int64_t, long long, long, int. If none of these are 8 bytes, then we should
# search for int32_t, long long, long, int.
#
MPI_OFFSET_TYPE="not found"
2004-12-08 01:40:37 +03:00
MPI_OFFSET_DATATYPE="not found"
2004-01-19 04:33:03 +03:00
AC_MSG_CHECKING([checking for type of MPI_Offset])
2009-03-16 05:00:27 +03:00
if test $ac_cv_type_long_long = yes -a $ac_cv_sizeof_long_long = 8; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="long long"
MPI_OFFSET_DATATYPE=MPI_LONG_LONG
MPI_OFFSET_SIZE=8
2009-03-16 05:00:27 +03:00
elif test $ac_cv_type_long = yes -a $ac_cv_sizeof_long = 8; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="long"
MPI_OFFSET_DATATYPE=MPI_LONG
MPI_OFFSET_SIZE=8
2009-03-16 05:00:27 +03:00
elif test $ac_cv_sizeof_int = 8; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="int"
MPI_OFFSET_DATATYPE=MPI_INT
MPI_OFFSET_SIZE=8
2009-03-16 05:00:27 +03:00
elif test $ac_cv_type_long_long = yes -a $ac_cv_sizeof_long_long = 4; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="long long"
MPI_OFFSET_DATATYPE=MPI_LONG_LONG
MPI_OFFSET_SIZE=4
2009-03-16 05:00:27 +03:00
elif test $ac_cv_type_long = yes -a $ac_cv_sizeof_long = 4; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="long"
MPI_OFFSET_DATATYPE=MPI_LONG
MPI_OFFSET_SIZE=4
2009-03-16 05:00:27 +03:00
elif test $ac_cv_sizeof_int = 4; then
2009-03-16 07:22:22 +03:00
MPI_OFFSET_TYPE="int"
MPI_OFFSET_DATATYPE=MPI_INT
MPI_OFFSET_SIZE=4
2004-01-17 01:16:25 +03:00
fi
2004-01-19 04:33:03 +03:00
AC_MSG_RESULT([$MPI_OFFSET_TYPE])
2004-12-08 01:40:37 +03:00
if test "$MPI_OFFSET_TYPE" = "not found"; then
2009-03-16 07:22:22 +03:00
AC_MSG_WARN([*** Unable to find the right definition for MPI_Offset])
AC_MSG_ERROR([Cannot continue])
2004-01-17 01:16:25 +03:00
fi
2011-01-13 09:08:54 +03:00
AC_MSG_CHECKING([checking for an MPI datatype for MPI_Offset])
AC_MSG_RESULT([$MPI_OFFSET_DATATYPE])
# this should never happens
if test "$MPI_OFFSET_DATATYPE" = "not found"; then
AC_MSG_WARN([*** Unable to find an MPI datatype corresponding to MPI_Offset])
AC_MSG_ERROR([Cannot continue])
fi
2005-11-30 06:16:24 +03:00
AC_DEFINE_UNQUOTED(OMPI_MPI_OFFSET_TYPE, $MPI_OFFSET_TYPE, [Type of MPI_Offset -- has to be defined here and typedef'ed later because mpi.h does not get AC SUBST's])
2011-01-13 09:08:54 +03:00
AC_DEFINE_UNQUOTED(OMPI_MPI_OFFSET_SIZE, $MPI_OFFSET_SIZE, [Size of the MPI_Offset])
AC_DEFINE_UNQUOTED(OMPI_OFFSET_DATATYPE, $MPI_OFFSET_DATATYPE, [MPI datatype corresponding to MPI_Offset])
2004-01-19 04:33:03 +03:00
2006-10-20 07:24:59 +04:00
#
# Check for MPI_Aint type. Yes, there are platforms where
# sizeof(void*) != sizeof(long) (64 bit Windows, apparently).
#
if test $ac_cv_type_ptrdiff_t = yes ; then
2009-05-07 00:11:28 +04:00
opal_ptrdiff_t="ptrdiff_t"
2006-10-20 07:24:59 +04:00
elif test $ac_cv_sizeof_void_p -eq $ac_cv_sizeof_long ; then
2009-05-07 00:11:28 +04:00
opal_ptrdiff_t="long"
2006-10-20 07:24:59 +04:00
elif test $ac_cv_type_long_long = yes -a $ac_cv_sizeof_void_p -eq $ac_cv_sizeof_long_long ; then
2009-05-07 00:11:28 +04:00
opal_ptrdiff_t="long long"
2006-10-20 07:24:59 +04:00
else
AC_MSG_ERROR([Could not find datatype to emulate ptrdiff_t. Cannot continue])
fi
2009-05-07 00:11:28 +04:00
AC_DEFINE_UNQUOTED([OPAL_PTRDIFF_TYPE], [$opal_ptrdiff_t],
2006-10-20 07:24:59 +04:00
[type to use for ptrdiff_t])
2007-02-06 15:03:56 +03:00
# Do we have _SC_NPROCESSORS_ONLN? (only going to pass if we also have
# <unistd.h> and sysconf(), which is ok) OS X 10.4 has <unistd.h> and
# sysconf(), but does not have _SC_NPROCESSORS_ONLN. Doh!
2008-11-08 01:13:43 +03:00
AC_CACHE_CHECK([for _SC_NPROCESSORS_ONLN],
[ompi_cv_have__SC_NPROCESSORS_ONLN],
[AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
AC_INCLUDES_DEFAULT
#include <unistd.h>
],
[int i = _SC_NPROCESSORS_ONLN;])],
[ompi_cv_have__SC_NPROCESSORS_ONLN="yes"],
[ompi_cv_have__SC_NPROCESSORS_ONLN="no"])])
AS_IF([test "$ompi_cv_have__SC_NPROCESSORS_ONLN" = "yes"],
[result=1], [result=0])
AC_DEFINE_UNQUOTED([OPAL_HAVE__SC_NPROCESSORS_ONLN], [$result],
[Define to 1 ifyou have the declaration of _SC_NPROCESSORS_ONLN, and to 0 otherwise])
2007-02-06 15:03:56 +03:00
2003-11-22 19:36:58 +03:00
# all: endian
2004-10-14 16:55:34 +04:00
2006-03-04 16:59:14 +03:00
AC_C_BIGENDIAN
2004-10-14 16:55:34 +04:00
2005-06-27 03:11:37 +04:00
OMPI_CHECK_BROKEN_QSORT
2007-07-13 09:45:02 +04:00
AC_CACHE_CHECK([if word-sized integers must be word-size aligned],
[ompi_cv_c_word_size_align],
[AC_LANG_PUSH(C)
AC_RUN_IFELSE([AC_LANG_PROGRAM([dnl
#include <stdlib.h>], [[ long data[2] = {0, 0};
long *lp;
int *ip;
ip = (int*) data;
ip++;
lp = (long*) ip;
return lp[0]; ]])],
[ompi_cv_c_word_size_align=no],
[ompi_cv_c_word_size_align=yes],
[ompi_cv_c_word_size_align=yes])])
AS_IF([test $ompi_cv_c_word_size_align = yes], [results=1], [results=0])
2009-05-07 00:11:28 +04:00
AC_DEFINE_UNQUOTED([OPAL_ALIGN_WORD_SIZE_INTEGERS], [$results],
2007-07-13 09:45:02 +04:00
[set to 1 if word-size integers must be aligned to word-size padding to prevent bus errors])
2003-11-22 19:36:58 +03:00
# all: SYSV semaphores
# all: SYSV shared memory
# all: size of FD_SET
# all: sizeof struct stat members
# all: type of getsockopt optlen
# all: type of recvfrom optlen
2004-01-09 08:08:31 +03:00
2004-01-14 10:06:57 +03:00
#
# Check out what thread support we have
#
2010-03-17 02:10:50 +03:00
OPAL_CONFIG_THREADS
m4_ifdef([project_ompi], [OMPI_CONFIG_THREADS])
2004-01-14 10:06:57 +03:00
2004-03-17 06:04:44 +03:00
CFLAGS="$CFLAGS $THREAD_CFLAGS"
2010-02-03 07:28:32 +03:00
CPPFLAGS="$CPPFLAGS $THREAD_CPPFLAGS"
2013-01-29 04:00:43 +04:00
CXXFLAGS="$CXXFLAGS $THREAD_CXXFLAGS"
CXXCPPFLAGS="$CXXCPPFLAGS $THREAD_CXXCPPFLAGS"
2004-03-17 06:04:44 +03:00
LDFLAGS="$LDFLAGS $THREAD_LDFLAGS"
LIBS="$LIBS $THREAD_LIBS"
2013-01-29 04:00:43 +04:00
OPAL_WRAPPER_FLAGS_ADD([CFLAGS], [$THREAD_CFLAGS])
OPAL_WRAPPER_FLAGS_ADD([CXXFLAGS], [$THREAD_CXXFLAGS])
OPAL_WRAPPER_FLAGS_ADD([FCFLAGS], [$THREAD_FCFLAGS])
OPAL_WRAPPER_FLAGS_ADD([LDFLAGS], [$THREAD_LDFLAGS])
2006-01-16 04:48:03 +03:00
# no need to update WRAPPER_EXTRA_LIBS - we'll get it from LT later
2004-01-14 10:06:57 +03:00
2004-02-13 06:58:56 +03:00
#
# What is the local equivalent of "ln -s"
#
AC_PROG_LN_S
2007-06-01 06:31:15 +04:00
AC_PROG_GREP
AC_PROG_EGREP
2004-08-05 18:01:45 +04:00
#
2004-08-16 03:55:10 +04:00
# We need as and lex
2004-08-05 18:01:45 +04:00
#
2004-08-16 03:55:10 +04:00
AM_PROG_AS
2004-08-05 18:01:45 +04:00
AM_PROG_LEX
2004-09-05 21:45:59 +04:00
# If we don't have GNU Flex and we don't have a generated .c file
2004-08-26 20:17:20 +04:00
# (distribution tarballs will have the .c file included, but SVN
2004-09-05 21:45:59 +04:00
# checkouts will not), then error. Must have GNU Flex -- other
# versions of Lex are not workable (all things being equal, since this
# is *only* required for developers, we decided that it really was not
# worth it to be portable between different versions of lex ;-).
2007-06-01 06:31:15 +04:00
if test -z "$LEX" -o -n "`echo $LEX | $GREP missing`" -o \
2004-09-05 21:45:59 +04:00
"`basename $LEX`" != "flex"; then
2005-07-04 06:38:44 +04:00
if test ! -f "$srcdir/opal/util/show_help_lex.c"; then
2004-09-05 21:45:59 +04:00
AC_MSG_WARN([*** Could not find GNU Flex on your system.])
AC_MSG_WARN([*** GNU Flex required for developer builds of Open MPI.])
AC_MSG_WARN([*** Other versions of Lex are not supported.])
AC_MSG_WARN([*** YOU DO NOT NEED FLEX FOR DISTRIBUTION TARBALLS!])
AC_MSG_WARN([*** If you absolutely cannot install GNU Flex on this system])
AC_MSG_WARN([*** consider using a distribution tarball, or generate the])
AC_MSG_WARN([*** following files on another system (using Flex) and])
AC_MSG_WARN([*** copy them here:])
for lfile in `find . -name \*.l -print`; do
cfile="`echo $lfile | cut -d. -f-2`"
AC_MSG_WARN([*** $cfile.c])
done
2004-08-26 20:17:20 +04:00
AC_MSG_ERROR([Cannot continue])
fi
fi
2007-02-03 03:25:42 +03:00
#
# Look for ps command and arguments for orte-clean
#
2009-10-21 03:44:20 +04:00
m4_ifdef([project_orte], [OMPI_PS_FLAVOR_CHECK])
2007-02-03 03:25:42 +03:00
2004-01-09 11:25:18 +03:00
#
2004-01-09 08:08:31 +03:00
# File system case sensitivity
2004-01-09 11:25:18 +03:00
#
2004-01-09 08:08:31 +03:00
2012-12-19 03:51:38 +04:00
OPAL_CASE_SENSITIVE_FS_SETUP
2003-11-22 19:36:58 +03:00
# AIX: FIONBIO in sys/ioctl.h
# glibc: memcpy
2007-04-25 05:54:37 +04:00
#
# Do we have RLIMIT_NPROC in <sys/resources.h>? (e.g., Solaris does not)
#
AC_CHECK_DECLS([RLIMIT_NPROC], [], [], [
AC_INCLUDES_DEFAULT
#if HAVE_SYS_RESOURCE_H
#include <sys/resource.h>
2009-03-16 05:04:22 +03:00
#endif])
2007-04-25 05:54:37 +04:00
2008-11-25 06:13:09 +03:00
#
# Do we have RLIMIT_MEMLOCK in <sys/resources.h>? (e.g., Solaris does not)
#
AC_CHECK_DECLS([RLIMIT_MEMLOCK], [], [], [
AC_INCLUDES_DEFAULT
#if HAVE_SYS_RESOURCE_H
#include <sys/resource.h>
2009-03-16 05:04:22 +03:00
#endif])
2008-11-25 06:13:09 +03:00
2005-07-09 22:52:53 +04:00
# checkpoint results
AC_CACHE_SAVE
2013-01-29 22:30:41 +04:00
###########################################################
# Fault Tolerance
#
# The FT code in the OMPI trunk is currently broken. We don't
# have an active maintainer for it at this time, and it isn't
# clear if/when we will return to it. We have therefore removed
# the configure options supporting it until such time as it
# can be fixed.
#
# However, we recognize that there are researchers who use this
# option on their independent branches. In such cases, simply
# uncomment the line below to render the FT configure options
# visible again
#
###########################################################
#OPAL_SETUP_FT_OPTIONS
###########################################################
# The following line is always required as it contains the
# AC_DEFINE and AM_CONDITIONAL calls that set variables used
# throughout the build system. If the above line is commented
# out, then those variables will be set to "off". Otherwise,
# they are controlled by the options
OPAL_SETUP_FT
2003-11-22 19:36:58 +03:00
2004-01-12 00:35:37 +03:00
##################################
# MCA
##################################
2009-03-16 05:04:22 +03:00
ompi_show_title "Modular Component Architecture (MCA) setup"
2004-01-12 00:35:37 +03:00
AC_MSG_CHECKING([for subdir args])
2004-06-07 19:33:53 +04:00
OMPI_CONFIG_SUBDIR_ARGS([ompi_subdir_args])
AC_MSG_RESULT([$ompi_subdir_args])
2004-01-12 00:35:37 +03:00
2004-06-07 19:33:53 +04:00
OMPI_MCA
2004-01-12 00:35:37 +03:00
2005-07-09 22:52:53 +04:00
# checkpoint results
AC_CACHE_SAVE
2009-05-27 00:49:35 +04:00
##################################
# MPI Extended Interfaces
##################################
2009-10-21 03:44:20 +04:00
m4_ifdef([project_ompi], [OMPI_SETUP_MPI_EXT])
2009-05-27 00:49:35 +04:00
# checkpoint results
AC_CACHE_SAVE
2010-01-18 16:30:56 +03:00
##################################
# Contributed software
##################################
m4_ifdef([project_ompi], [OMPI_SETUP_CONTRIB])
# checkpoint results
AC_CACHE_SAVE
2008-03-07 01:15:10 +03:00
##################################
2009-03-16 05:00:27 +03:00
# Visibility
2008-03-07 01:15:10 +03:00
##################################
2009-03-16 05:00:27 +03:00
# Check the visibility declspec at the end to avoid problem with
# the previous tests that are not necessarily prepared for
2008-03-07 01:15:10 +03:00
# the visibility feature.
2009-05-27 00:49:35 +04:00
ompi_show_title "Symbol visibility feature"
2008-03-07 01:15:10 +03:00
2010-10-23 18:32:44 +04:00
OPAL_CHECK_VISIBILITY
2008-03-07 01:15:10 +03:00
2004-01-12 00:35:37 +03:00
2009-10-21 03:44:20 +04:00
2004-01-16 04:31:50 +03:00
############################################################################
2004-06-07 19:33:53 +04:00
# Final top-level OMPI configuration
2004-01-16 04:31:50 +03:00
############################################################################
2004-06-07 19:33:53 +04:00
ompi_show_title "Final top-level OMPI configuration"
2004-01-16 04:31:50 +03:00
2003-11-22 19:36:58 +03:00
############################################################################
# Libtool: part two
2008-03-07 01:17:40 +03:00
# (after C compiler setup = no compiler/linker tests after this)
2003-11-22 19:36:58 +03:00
############################################################################
2004-06-07 19:33:53 +04:00
ompi_show_subtitle "Libtool configuration"
2003-11-22 19:36:58 +03:00
2008-11-25 18:59:48 +03:00
# Use the undocumented solaris_use_stlport4 libtool variable to turn off any
# Cstd/stlport4 linkage. This allows Open MPI to be C++ STL agnostic.
if test "x$ompi_cv_c_compiler_vendor" = "xsun"; then
solaris_use_stlport4="yes"
fi
2013-02-12 19:24:40 +04:00
# Due to this thread:
# http://www.open-mpi.org/community/lists/users/2013/02/21356.php and
# a helpful tip from Dave Goodell, set the precious variables for
# compilers to "no" that we don't want. Libtool's m4 configry will
# interpret this as "I won't be using this language; don't bother
# setting it up."
AS_IF([test "$OMPI_WANT_FORTRAN_BINDINGS" != "1"],[F77=no FC=no])
AS_IF([test "$WANT_MPI_CXX_SUPPORT" != "1"],[CXX=no])
2007-08-17 21:11:36 +04:00
dnl Not all versions of LT set the PACKAGE_VERSION
m4_ifdef([LT_PACKAGE_VERSION], [], [m4_define([LT_PACKAGE_VERSION], [1.5.22])])
2007-08-17 08:08:23 +04:00
m4_if(m4_version_compare(m4_defn([LT_PACKAGE_VERSION]), 2.0), -1, [
2007-05-15 08:23:48 +04:00
AC_LIBLTDL_CONVENIENCE(opal/libltdl)
AC_LIBTOOL_DLOPEN
AC_PROG_LIBTOOL
2007-08-17 08:08:23 +04:00
], [
LT_CONFIG_LTDL_DIR([opal/libltdl], [subproject])
LTDL_CONVENIENCE
LT_INIT([dlopen win32-dll])
])
2007-08-17 21:11:36 +04:00
Put back the static-library-detection stuff from r27668, with some
additional functionality. Rationale (refs trac:3422):
* Normal MPI applications only ever use the MPI API. Hence, -lmpi is
sufficient (they'll never directly call ORTE or OPAL
functions). This is arguably the most common case.
* That being said, we do have some test programs (e.g., those in
orte/test/mpi) that call MPI functions but also call ORTE/OPAL
functions. I've also written the occasional MPI test program that
calls opal_output, for example (there even might be a few tests in
the IBM test suite that directly call ORTE/OPAL functions).
* Even though this is not a common case, these applications should
also compile/link with mpicc.
* So we should add a --openmpi:linkall option that will also link
in whatever is necessary to call ORTE/OPAL functions
* Yes, we could hard-code "-lopen-rte -lopen-pal" in Makefiles, but
we do reserve the right to change those library names and/or add
others someday, so it's better to abstract out the names and let
the wrapper supply whatever is necessary.
* ORTE programs, however, are different. They almost always call OPAL
functions (e.g., if they want to send a message, they must use the
OPAL DSS). As such, it seems like the ORTE programs should always
link in OPAL.
Therefore:
* Add undocumented --openmpi:linkall flag to the wrapper compilers.
See the comment in opal_wrapper.c for an explanation of what it
does. This flag is only intended for Open MPI developers -- not
end users. That's why it's undocumented.
* Update orte/test/mpi/Makefile.am to add --openmpi:linkall
* Make ortecc/ortec++'s wrapper data text files always explicitly
link in libopen-pal
This commit was SVN r27670.
The following SVN revision numbers were found above:
r27668 --> open-mpi/ompi@cf845897aa89407195aa13f56ddf4ec0d3c21a68
The following Trac tickets were found above:
Ticket 3422 --> https://svn.open-mpi.org/trac/ompi/ticket/3422
2012-12-14 02:31:37 +04:00
# What's the suffix of shared libraries? Inspired by generated
# Libtool code (even though we don't support several of these
# platforms, there didn't seem to be any harm in leaving in some of
# them, alhtough I did remove some that we have never/will never
# support, like OS/2).
2013-02-15 18:12:39 +04:00
OPAL_DYN_LIB_PREFIX=lib
Put back the static-library-detection stuff from r27668, with some
additional functionality. Rationale (refs trac:3422):
* Normal MPI applications only ever use the MPI API. Hence, -lmpi is
sufficient (they'll never directly call ORTE or OPAL
functions). This is arguably the most common case.
* That being said, we do have some test programs (e.g., those in
orte/test/mpi) that call MPI functions but also call ORTE/OPAL
functions. I've also written the occasional MPI test program that
calls opal_output, for example (there even might be a few tests in
the IBM test suite that directly call ORTE/OPAL functions).
* Even though this is not a common case, these applications should
also compile/link with mpicc.
* So we should add a --openmpi:linkall option that will also link
in whatever is necessary to call ORTE/OPAL functions
* Yes, we could hard-code "-lopen-rte -lopen-pal" in Makefiles, but
we do reserve the right to change those library names and/or add
others someday, so it's better to abstract out the names and let
the wrapper supply whatever is necessary.
* ORTE programs, however, are different. They almost always call OPAL
functions (e.g., if they want to send a message, they must use the
OPAL DSS). As such, it seems like the ORTE programs should always
link in OPAL.
Therefore:
* Add undocumented --openmpi:linkall flag to the wrapper compilers.
See the comment in opal_wrapper.c for an explanation of what it
does. This flag is only intended for Open MPI developers -- not
end users. That's why it's undocumented.
* Update orte/test/mpi/Makefile.am to add --openmpi:linkall
* Make ortecc/ortec++'s wrapper data text files always explicitly
link in libopen-pal
This commit was SVN r27670.
The following SVN revision numbers were found above:
r27668 --> open-mpi/ompi@cf845897aa89407195aa13f56ddf4ec0d3c21a68
The following Trac tickets were found above:
Ticket 3422 --> https://svn.open-mpi.org/trac/ompi/ticket/3422
2012-12-14 02:31:37 +04:00
case $host_os in
2013-02-15 18:12:39 +04:00
cygwin*)
OPAL_DYN_LIB_PREFIX=cyg
OPAL_DYN_LIB_SUFFIX=dll
;;
mingw* | pw32* | cegcc*)
Put back the static-library-detection stuff from r27668, with some
additional functionality. Rationale (refs trac:3422):
* Normal MPI applications only ever use the MPI API. Hence, -lmpi is
sufficient (they'll never directly call ORTE or OPAL
functions). This is arguably the most common case.
* That being said, we do have some test programs (e.g., those in
orte/test/mpi) that call MPI functions but also call ORTE/OPAL
functions. I've also written the occasional MPI test program that
calls opal_output, for example (there even might be a few tests in
the IBM test suite that directly call ORTE/OPAL functions).
* Even though this is not a common case, these applications should
also compile/link with mpicc.
* So we should add a --openmpi:linkall option that will also link
in whatever is necessary to call ORTE/OPAL functions
* Yes, we could hard-code "-lopen-rte -lopen-pal" in Makefiles, but
we do reserve the right to change those library names and/or add
others someday, so it's better to abstract out the names and let
the wrapper supply whatever is necessary.
* ORTE programs, however, are different. They almost always call OPAL
functions (e.g., if they want to send a message, they must use the
OPAL DSS). As such, it seems like the ORTE programs should always
link in OPAL.
Therefore:
* Add undocumented --openmpi:linkall flag to the wrapper compilers.
See the comment in opal_wrapper.c for an explanation of what it
does. This flag is only intended for Open MPI developers -- not
end users. That's why it's undocumented.
* Update orte/test/mpi/Makefile.am to add --openmpi:linkall
* Make ortecc/ortec++'s wrapper data text files always explicitly
link in libopen-pal
This commit was SVN r27670.
The following SVN revision numbers were found above:
r27668 --> open-mpi/ompi@cf845897aa89407195aa13f56ddf4ec0d3c21a68
The following Trac tickets were found above:
Ticket 3422 --> https://svn.open-mpi.org/trac/ompi/ticket/3422
2012-12-14 02:31:37 +04:00
OPAL_DYN_LIB_SUFFIX=dll
;;
darwin* | rhapsody*)
OPAL_DYN_LIB_SUFFIX=dylib
;;
hpux9* | hpux10* | hpux11*)
case $host_cpu in
ia64*)
OPAL_DYN_LIB_SUFFIX=so
;;
*)
OPAL_DYN_LIB_SUFFIX=sl
;;
esac
;;
*)
OPAL_DYN_LIB_SUFFIX=so
;;
esac
2013-02-15 18:12:39 +04:00
AC_SUBST(OPAL_DYN_LIB_PREFIX)
Put back the static-library-detection stuff from r27668, with some
additional functionality. Rationale (refs trac:3422):
* Normal MPI applications only ever use the MPI API. Hence, -lmpi is
sufficient (they'll never directly call ORTE or OPAL
functions). This is arguably the most common case.
* That being said, we do have some test programs (e.g., those in
orte/test/mpi) that call MPI functions but also call ORTE/OPAL
functions. I've also written the occasional MPI test program that
calls opal_output, for example (there even might be a few tests in
the IBM test suite that directly call ORTE/OPAL functions).
* Even though this is not a common case, these applications should
also compile/link with mpicc.
* So we should add a --openmpi:linkall option that will also link
in whatever is necessary to call ORTE/OPAL functions
* Yes, we could hard-code "-lopen-rte -lopen-pal" in Makefiles, but
we do reserve the right to change those library names and/or add
others someday, so it's better to abstract out the names and let
the wrapper supply whatever is necessary.
* ORTE programs, however, are different. They almost always call OPAL
functions (e.g., if they want to send a message, they must use the
OPAL DSS). As such, it seems like the ORTE programs should always
link in OPAL.
Therefore:
* Add undocumented --openmpi:linkall flag to the wrapper compilers.
See the comment in opal_wrapper.c for an explanation of what it
does. This flag is only intended for Open MPI developers -- not
end users. That's why it's undocumented.
* Update orte/test/mpi/Makefile.am to add --openmpi:linkall
* Make ortecc/ortec++'s wrapper data text files always explicitly
link in libopen-pal
This commit was SVN r27670.
The following SVN revision numbers were found above:
r27668 --> open-mpi/ompi@cf845897aa89407195aa13f56ddf4ec0d3c21a68
The following Trac tickets were found above:
Ticket 3422 --> https://svn.open-mpi.org/trac/ompi/ticket/3422
2012-12-14 02:31:37 +04:00
AC_SUBST(OPAL_DYN_LIB_SUFFIX)
2010-05-21 16:47:56 +04:00
OPAL_SETUP_LIBLTDL
2010-01-26 02:41:59 +03:00
2003-11-22 19:36:58 +03:00
############################################################################
2005-10-28 03:23:08 +04:00
# final compiler config
2003-11-22 19:36:58 +03:00
############################################################################
2009-10-24 05:04:35 +04:00
m4_ifdef([project_ompi], [ompi_show_subtitle "Compiler flags"],
[m4_ifdef([project_orte], [ompi_show_subtitle "Compiler flags"])])
2003-11-22 19:36:58 +03:00
#
# This is needed for VPATH builds, so that it will -I the appropriate
2005-09-01 16:16:36 +04:00
# include directory. We delayed doing it until now just so that
2003-11-22 19:36:58 +03:00
# '-I$(top_srcdir)' doesn't show up in any of the configure output --
# purely aesthetic.
#
2006-02-12 04:33:29 +03:00
# Because opal_config.h, orte_config.h, and ompi_config.h are all
# created by AC_CONFIG_HEADERS, we don't need to -I the builddir for
# <project>/include. If we VPATH building, we do need to include the
# source directories, however.
2005-09-01 16:16:36 +04:00
#
2006-02-12 04:33:29 +03:00
if test "$OMPI_TOP_BUILDDIR" != "$OMPI_TOP_SRCDIR"; then
2009-10-21 03:44:20 +04:00
# Note the embedded m4 directives here -- we must embed them
# rather than have successive assignments to these shell
# variables, lest the $(foo) names try to get evaluated here.
# Yuck!
2013-02-12 21:45:27 +04:00
CPPFLAGS='-I$(top_srcdir) -I$(top_builddir) -I$(top_srcdir)/opal/include m4_ifdef([project_orte], [-I$(top_srcdir)/orte/include]) m4_ifdef([project_ompi], [-I$(top_srcdir)/ompi/include])'" $CPPFLAGS"
2009-10-21 03:44:20 +04:00
# C++ is only relevant if we're building OMPI
2012-06-27 05:28:28 +04:00
m4_ifdef([project_ompi], [CXXCPPFLAGS='-I$(top_srcdir) -I$(top_builddir) -I$(top_srcdir)/opal/include -I$(top_srcdir)/orte/include -I$(top_srcdir)/ompi/include'" $CXXCPPFLAGS"])
2006-02-12 04:33:29 +03:00
else
2006-03-07 11:49:59 +03:00
CPPFLAGS='-I$(top_srcdir)'" $CPPFLAGS"
2009-10-21 03:44:20 +04:00
# C++ is only relevant if we're building OMPI
m4_ifdef([project_ompi], [CXXCPPFLAGS='-I$(top_srcdir)'" $CXXCPPFLAGS"])
2005-09-01 14:37:20 +04:00
fi
2009-10-23 17:00:59 +04:00
# OMPI/ORTE wants some additional processing of the flags (e.g., get
2009-10-21 03:44:20 +04:00
# versions without optimization for debugger modules).
2005-09-01 14:37:20 +04:00
2009-10-23 17:00:59 +04:00
m4_ifdef([project_orte], [ORTE_SETUP_DEBUGGER_FLAGS],
2012-06-27 05:28:28 +04:00
[m4_ifdef([project_ompi], [ORTE_SETUP_DEBUGGER_FLAGS])])
2005-02-08 08:06:15 +03:00
2004-01-19 04:41:40 +03:00
#
# Delayed the substitution of CFLAGS and CXXFLAGS until now because
# they may have been modified throughout the course of this script.
#
2003-11-22 19:36:58 +03:00
AC_SUBST(CFLAGS)
AC_SUBST(CPPFLAGS)
2009-10-24 05:04:35 +04:00
AC_SUBST(CXXFLAGS)
AC_SUBST(CXXCPPFLAGS)
m4_ifdef([project_ompi], [AC_SUBST(FFLAGS)
2009-10-21 03:44:20 +04:00
AC_SUBST(FCFLAGS)
2003-11-22 19:36:58 +03:00
2009-10-21 03:44:20 +04:00
AC_SUBST(OMPI_LIBMPI_EXTRA_LIBS)
AC_SUBST(OMPI_LIBMPI_EXTRA_LDFLAGS)])
2007-08-17 07:52:53 +04:00
2007-03-01 16:39:20 +03:00
#
# Aggregate MCA parameters directory
#
AC_SUBST([AMCA_PARAM_SETS_DIR], ['$(pkgdatadir)/amca-param-sets'])
2005-10-28 03:23:08 +04:00
############################################################################
# final wrapper compiler config
############################################################################
2010-10-20 02:45:54 +04:00
ompi_show_subtitle "Wrapper compiler final setup"
# The ORTE and OMPI wrapper scripts (i.e., not the C-compiled
# executables) need perl.
AC_PATH_PROG(PERL, perl, perl)
2009-10-24 05:04:35 +04:00
OPAL_SETUP_WRAPPER_FINAL
2003-11-22 19:36:58 +03:00
2008-02-26 04:45:32 +03:00
# Recreate some defines prefixed with OMPI_ so that there are no bare
# autoconf macro defines in mpi.h. Since AC sometimes changes whether
# things are defined as null tokens or an integer result, two projects
# with different versions of AC can cause problems.
if test $ac_cv_header_stdc = yes; then
2009-05-07 00:11:28 +04:00
AC_DEFINE(OPAL_STDC_HEADERS, 1,
2009-03-16 07:22:22 +03:00
[Do not use outside of mpi.h. Define to 1 if you have the ANSI C header files.])
2008-02-26 04:45:32 +03:00
fi
if test $ac_cv_header_sys_time_h = yes ; then
2009-05-07 00:11:28 +04:00
AC_DEFINE(OPAL_HAVE_SYS_TIME_H, 1,
2009-03-16 07:22:22 +03:00
[Do not use outside of mpi.h. Define to 1 if you have the <sys/time.h> header file.])
2008-02-26 04:45:32 +03:00
fi
if test $ac_cv_type_long_long = yes ; then
2009-05-07 00:11:28 +04:00
AC_DEFINE(OPAL_HAVE_LONG_LONG, 1,
2009-03-16 07:22:22 +03:00
[Do not use outside of mpi.h. Define to 1 if the system has the type `long long'.]) dnl `
2008-02-26 04:45:32 +03:00
fi
2008-09-10 16:58:30 +04:00
if test $ac_cv_header_sys_synch_h = yes ; then
2009-05-07 00:11:28 +04:00
AC_DEFINE(OPAL_HAVE_SYS_SYNCH_H, 1,
2009-03-16 07:22:22 +03:00
[Do not use outside of mpi.h. Define to 1 if you have the <sys/synch.h> header file.])
2008-09-10 16:58:30 +04:00
fi
2009-12-17 02:32:47 +03:00
# If there is a local hook for each project, call it. This allows 3rd
# parties to add configuration steps to OPAL, ORTE, and/or OMPI simply
# by placing a file in [opal|orte|ompi]/config/whatever.m4 that
# AC_DEFUN's the appropriate macro name -- no patching is necessary.
# If that macro is defined, we'll run it here.
#
# Unfortunately, aclocal is not smart enough to parse something like
# the following in ompi_mca.m4 (when we're already m4 looping over the
# project list):
#
# m4_foreach(mca_project, [mca_project_list],
# [m4_ifdef(mca_project[_CONFIG_LOCAL], mca_project[_CONFIG_LOCAL])])
#
# Meaning that aclocal doesn't see that, for example,
# "ompi_CONFIG_LOCAL" is actually invoked at the bottom and therefore
# go look for an .m4 file that contains it. Instead, we have to
# manually list the macros here. *Then* aclocal is smart enough to go
# look for an .m4 file containing each macro, and if found,
# automatically m4_include the corresponding in aclocal.m4. Bummer.
# :-\
m4_ifdef([opal_CONFIG_LOCAL], [opal_CONFIG_LOCAL])
2012-12-19 03:51:38 +04:00
m4_ifdef([project_orte],
[m4_ifdef([orte_CONFIG_LOCAL], [orte_CONFIG_LOCAL])])
m4_ifdef([project_ompi],
[m4_ifdef([ompi_CONFIG_LOCAL], [ompi_CONFIG_LOCAL])])
2009-12-17 02:32:47 +03:00
2003-11-22 19:36:58 +03:00
############################################################################
# Party on
############################################################################
2009-03-16 05:04:22 +03:00
ompi_show_subtitle "Final output"
2003-11-22 19:36:58 +03:00
AC_CONFIG_FILES([
Makefile
config/Makefile
2005-08-22 01:04:52 +04:00
contrib/Makefile
2004-08-30 14:44:30 +04:00
2005-03-22 07:25:01 +03:00
test/Makefile
2006-06-27 02:28:59 +04:00
test/event/Makefile
2005-03-22 07:25:01 +03:00
test/asm/Makefile
2009-10-29 04:54:11 +03:00
test/datatype/Makefile
2005-03-22 07:25:01 +03:00
test/class/Makefile
test/support/Makefile
test/threads/Makefile
2010-02-11 02:18:29 +03:00
test/util/Makefile
2003-11-22 19:36:58 +03:00
])
2009-10-21 03:44:20 +04:00
2009-10-24 05:04:35 +04:00
OPAL_CONFIG_FILES
m4_ifdef([project_orte], [ORTE_CONFIG_FILES])
m4_ifdef([project_ompi], [OMPI_CONFIG_FILES])
2009-10-21 03:44:20 +04:00
2003-11-22 19:36:58 +03:00
AC_OUTPUT