Use automake+libtool instead of simplemake+createshlib.  Parallel make
now works for real, and shared library creation should be more robust on
more platforms.  "make dist" now works, modulo building the documentation,
although for boring reasons we can't actually use it for the final
distribution tarball generation.

This is a massive change and without a doubt has some bugs in it.  We'll
work through them as we find them.

No reviewer.

[svn-r7239] Allowed autoconf-2.68 in maint/updatefiles.  Fixed incorrect uses of AC_*_IFELSE, i.e. without use of AC_LANG_SOURCE and friends in AC_*_IFELSE, that are detected by autoconf-2.68.  But there are still a lot of warning from use of AC_PROG_LIBTOOL(or AM_PROG_LIBTOOL) and LT_INIT.  Also warnings from hydra (hwloc and plpa) and openpa are not fixed here.

[svn-r7186] Fixes to allow job to continue with a node failure.  VCs associated with the failed process enter a failed state, and cannot be used for communication, but communication is still allowed with other processes.  Reviewed by goodell@.

as well.

Reviewed by goodell.

Uses a define added in r106 of OPA.  Now nemesis doesn't need to
use an AC_ARG_WITH matching OPA's argument.

Reviewed by buntinas@.

In particular, this seemed to be a problem on Solaris when "struct
ifconf" wasn't actually available.  The bad shell test resulted caused
it to always evaluate to true, even though the structure wasn't
available.

No reviewer.

Previously flags to AR were fixed by simplemake to be "cr".  With this
change this can be controlled by configure/make via the new AR_FLAGS
precious variable.  On darwin using "Scr" shaves ~10% off the normal
build time.

This is a minor, nibbling around the edges fix.  The real fix is to
rework all of the linking in MPICH2 so that parallel builds are possible
and only a single invocation of AR/RANLIB occurs for each library that
we produce.

Reviewed by balaji@.

are eventually dumped into CPPFLAGS.

variable. Since we moved to PAC_CONFIG_SUBDIR, it'll take care of
exporting the precious variables correctly. Fixes ticket #928.

flags need to go into CPPFLAGS, instead of CFLAGS.

WRAPPER_LIBS.

LIBS are all libraries detected and added by MPICH2. These are
maintained internally, so appropriate autoconf checks can use them
correctly. WRAPPER_LIBS are only used within mpicc and friends; they
are a combination of LIBS and other libraries such as opa and mpl
which are only generated at make time. These are maintained separately
so autoconf doesn't try to use them for link or run tests.

Brice.

into mpicc and friends.

and friends are automatically exported down to subconfigures. If a
sub-configure chooses not to use whatever is exported by the
upper-level configure, it can reset the flags to the user-specified
ones using a new macro.

Reviewed by buntinas.

[svn-r5116] updatefiles now matches top-level AC_PREREQ version and commented out AC_PREREQs are deleted.

This is a cleanup of r5041.

No reviewer.

[svn-r5041] commented out AC_PREREQ in all non-standalone configure.in, i.e. except top-level, test/mpi, romio, hydra, mpe2, slog2sdk, openpa.

Reviewed by buntinas@.

Reviewed by buntinas@.

Also add 'knem' as a valid way to specify the dma local LMT.

No reviewer.

This version supports the 0.5.0 interface to knem.

No reviewer.

[svn-r3945] Merging mx netmod branch into trunk.  This adds override functions to the vc for overriding mpid_ send functions, as well as hooks for posting receives.

[svn-r3853] Added an additional thread debugging option for enable-g; fix incorrect use of == in test

[svn-r3818] 1) Nemesis code now uses the shared mem util wrappers (2) Removing obsolete shared mem code in nemesis - Review@Darius

[svn-r3763] Moved the network module directory structure to remove the unnecessary "_module" tag. Made similar appropriate changes to the function and datatype names as well. The newgm module is left alone, as it's not in the configuration path, and will probably be deleted. The wintcp module is not changed either, as the naming convention for it is different from the rest of the modules.

queues as intended, rather than incorrectly defaulting to lock-ful queues.  A
performance/hang problem remains with the lock-ful queue implementation that
will be resolved in a later commit.

Reviewed by balaji@.

atomic operations.  These need some performance improvements still that are
outlined in a FIXME.

Also, fix a few format string warnings.

Reviewed by buntinas@.