Merging in latest from upstream (HDFFV/hdf5:refs/heads/develop)

* commit 'b30a1fe44aa804a67f2291005bf8135ac5c2dc99':
  Updated the parallel install docs.
  Eliminated the need for a separate script variable.
  Added a helpful message to the flush script.
  Added a shell script so we can run the parallel flush test on OpenMPI.

MANIFEST

@@ -1295,6 +1295,7 @@
 ./testpar/t_span_tree.c
 ./testpar/t_init_term.c
 ./testpar/testpar.h
+./testpar/testpflush.sh.in
 ./testpar/testphdf5.c
 ./testpar/testphdf5.h
 

config/conclude.am

@@ -249,7 +249,7 @@ build-check-p: $(LIB) $(PROGS) $(chk_TESTS)
 	    echo "**** Hint ****";                                            \
 	    echo "Parallel test files reside in the current directory"        \
 	         "by default.";                                               \
-	    echo "Set HDF5_PARAPREFIX to use another directory. E.g.,";       \
+	    echo "Set HDF5_PARAPREFIX to use another directory. e.g.,";       \
 	    echo "    HDF5_PARAPREFIX=/PFS/user/me";                          \
 	    echo "    export HDF5_PARAPREFIX";                                \
 	    echo "    make check";                                            \

configure.ac

@@ -3411,6 +3411,7 @@ AC_CONFIG_FILES([src/libhdf5.settings
                  test/test_usecases.sh
                  test/test_vol_plugin.sh
                  testpar/Makefile
+                 testpar/testpflush.sh
                  tools/Makefile
                  tools/lib/Makefile
                  tools/src/Makefile

release_docs/INSTALL_parallel

@@ -102,7 +102,7 @@ qsub -I -q debug -l mppwidth=8
 - configure HDF5:
   RUNSERIAL="aprun -q -n 1" RUNPARALLEL="aprun -q -n 6" FC=ftn CC=cc /path/to/source/configure --enable-fortran --enable-parallel --disable-shared
 
-  RUNSERIAL and RUNPARALLEL tells the library how it should launch programs that are part of the build procedure.
+  RUNSERIAL and RUNPARALLEL tell the library how it should launch programs that are part of the build procedure.
 
 - Compile HDF5: 
   gmake
@@ -155,12 +155,16 @@ to run a parallel application on one processor and on many processors. If the
 compiler is `mpicc' and the user hasn't specified values for RUNSERIAL and
 RUNPARALLEL then configure chooses `mpiexec' from the same directory as `mpicc':
 
-    RUNSERIAL:    /usr/local/mpi/bin/mpiexec -np 1
-    RUNPARALLEL:  /usr/local/mpi/bin/mpiexec -np $${NPROCS:=6}
+    RUNSERIAL:    mpiexec -n 1
+    RUNPARALLEL:  mpiexec -n $${NPROCS:=6}
 
 The `$${NPROCS:=6}' will be substituted with the value of the NPROCS
 environment variable at the time `make check' is run (or the value 6).
 
+Note that some MPI implementations (e.g. OpenMPI 4.0) disallow oversubscribing
+nodes by default so you'll have to either set NPROCS equal to the number of
+processors available (or fewer) or redefine RUNPARALLEL with appropriate
+flag(s) (--oversubscribe in OpenMPI).
 
 4. Parallel test suite
 ----------------------

test/h5test.c

@@ -613,7 +613,7 @@ h5_fixname_real(const char *base_name, hid_t fapl, const char *_suffix,
                 printf("*** Hint ***\n"
                        "You can use environment variable HDF5_PARAPREFIX to "
                        "run parallel test files in a\n"
-                       "different directory or to add file type prefix. E.g.,\n"
+                       "different directory or to add file type prefix. e.g.,\n"
                        "   HDF5_PARAPREFIX=pfs:/PFS/user/me\n"
                        "   export HDF5_PARAPREFIX\n"
                        "*** End of Hint ***\n");

testpar/Makefile.am

@@ -21,11 +21,19 @@ include $(top_srcdir)/config/commence.am
 
 AM_CPPFLAGS+=-I$(top_srcdir)/src -I$(top_srcdir)/test
 
+# Test scripts--
+#       testpflush.sh:
+TEST_SCRIPT_PARA = testpflush.sh
+SCRIPT_DEPEND = t_pflush1$(EXEEXT) t_pflush2$(EXEEXT) 
+
+check_SCRIPTS = $(TEST_SCRIPT_PARA)
+
 # Test programs.  These are our main targets.
 #
-TEST_PROG_PARA=t_mpi t_bigio testphdf5 t_cache t_cache_image t_pflush1 t_pflush2 t_pread t_pshutdown t_prestart t_init_term t_shapesame t_filters_parallel
+TEST_PROG_PARA=t_mpi t_bigio testphdf5 t_cache t_cache_image t_pread t_pshutdown t_prestart t_init_term t_shapesame t_filters_parallel
 
-check_PROGRAMS = $(TEST_PROG_PARA)
+# t_pflush1 and t_pflush2 are used by testpflush.sh
+check_PROGRAMS = $(TEST_PROG_PARA) t_pflush1 t_pflush2
 
 testphdf5_SOURCES=testphdf5.c t_dset.c t_file.c t_file_image.c t_mdset.c \
 	t_ph5basic.c t_coll_chunk.c t_span_tree.c t_chunk_alloc.c t_filter_read.c \

testpar/testpflush.sh.in

@@ -0,0 +1,64 @@
+#! /bin/sh
+#
+# Copyright by The HDF Group.
+# Copyright by the Board of Trustees of the University of Illinois.
+# All rights reserved.
+#
+# This file is part of HDF5.  The full HDF5 copyright notice, including
+# terms governing use, modification, and redistribution, is contained in
+# the COPYING file, which can be found at the root of the source code
+# distribution tree, or in https://support.hdfgroup.org/ftp/HDF5/releases.
+# If you do not have access to either file, you may request a copy from
+# help@hdfgroup.org.
+#
+#
+# Test script for the parallel flush test
+#
+# The parallel flush test uses two programs to test flush operations
+# in parallel HDF5. The first program purposely exits without calling
+# MPI_Finalize(), which is an error under the MPI standard and mpiexec
+# in some implementations will return an error code even though all
+# processes exit successfully. This script lets us swallow the error
+# from the first program.
+#
+# True errors in the first program will be detected as errors in the
+# second program, so watch out for that.
+#
+# Programmer:   Dana Robinson
+#               Fall 2018
+
+# The build (current) directory might be different than the source directory.
+if test -z "$srcdir"; then
+   srcdir=.
+fi
+
+# Turn the $$ we use to avoid Autotools munging into $
+#
+# Allowing $$ to substitute in both the RUNPARALLEL string and the
+# regexp is intentional. There doesn't seem to be a way around
+# this using quote shenanigans. The downside is that there is a remote
+# chance that the shell's pid will match a number in the RUNPARALLEL
+# variable, but that seems less likely to cause problems than expecting
+# library builders to specify two almost identical versions of the
+# RUNPARALLEL command, one for use in scripts and one via Makefiles.
+RUNPARALLELSCRIPT=`echo "@RUNPARALLEL@" | sed "s/$$/\$/g"`
+
+# ==========================================
+# Run the first parallel flush test program
+# (note that we ignore any errors here)
+# ==========================================
+echo "*** NOTE ***********************************************************"
+echo "You may see complaints from mpiexec et al. that not all processes"
+echo "called MPI_Finalize(). This is an intended characteristic of the"
+echo "test and should not be considered an  error."
+echo "********************************************************************"
+eval ${RUNPARALLELSCRIPT} ./t_pflush1
+
+
+# ===========================================
+# Run the second parallel flush test program
+# The return code of this call is the return
+# code of the script.
+# ===========================================
+eval ${RUNPARALLELSCRIPT} ./t_pflush2
+