Description of problem:
For parallel HDF5 (with openmpi-1.5-3.fc15.x86_64), file creation always fail.
The simplest example in the HDF documentation fails. As file creation is a
necessary step and fails, nothing can be done with parallel HDF5. 
The serial hdf5 package works well.


Version-Release number of selected component (if applicable):
hdf5-openmpi-1.8.5.patch1-9.fc15.x86_64 (and devel)

How reproducible: 100% reproducible (tested on a freshly installed Fedora 15).


Steps to Reproduce:
1. Install the packages openmpi-x86_64, hdf5 and hdf5-openmpi (with devel too)
2. Download the simplest test file for collective file creation from the HDF group site: http://www.hdfgroup.org/HDF5/Tutor/examples/parallel/File_create.c
3. Compile it with "mpicc File_create.c -lhdf5_hl -lhdf5" and run it: "a.out" 

  
Actual results:
$ a.out
HDF5-DIAG: Error detected in HDF5 (1.8.5-patch1) MPI-process 0:
  #000: ../../src/H5F.c line 1427 in H5Fcreate(): unable to create file
    major: File accessability
    minor: Unable to open file
  #001: ../../src/H5F.c line 1198 in H5F_open(): unable to open file: time = Fri Jun 10 10:14:16 2011
, name = 'SDS_row.h5', tent_flags = 13
    major: File accessability
    minor: Unable to open file
  #002: ../../src/H5FD.c line 1088 in H5FD_open(): open failed
    major: Virtual File Layer
    minor: Unable to initialize object
  #003: ../../src/H5FDmpio.c line 999 in H5FD_mpio_open(): MPI_File_open failed
    major: Internal error (too specific to document in detail)
    minor: Some MPI function failed
  #004: ../../src/H5FDmpio.c line 999 in H5FD_mpio_open(): MPI_ERR_OTHER: known error not in list
    major: Internal error (too specific to document in detail)
    minor: MPI Error String
HDF5-DIAG: Error detected in HDF5 (1.8.5-patch1) MPI-process 0:
  #000: ../../src/H5F.c line 1950 in H5Fclose(): not a file ID
    major: Invalid arguments to routine
    minor: Inappropriate type


Expected results: Smooth creation of the file and normal termination of the program


Additional info: 
1. Same bug in Fortran
2. Running "mpiexec -n xx a.out" produces the same error, once for each MPI process.
3. I think that the proper libraries are loaded, as the output of "ldd a.out" is:
	linux-vdso.so.1 =>  (0x00007fff372c7000)
	libhdf5_hl.so.6 => /usr/lib64/openmpi/lib/libhdf5_hl.so.6 (0x00007fa9e3b28000)
	libhdf5.so.6 => /usr/lib64/openmpi/lib/libhdf5.so.6 (0x00007fa9e3535000)
	libmpi.so.1 => /usr/lib64/openmpi/lib/libmpi.so.1 (0x0000003d54c00000)
	libdl.so.2 => /lib64/libdl.so.2 (0x0000003d53c00000)
	libnsl.so.1 => /lib64/libnsl.so.1 (0x0000003d62e00000)
	libutil.so.1 => /lib64/libutil.so.1 (0x0000003d66800000)
	libm.so.6 => /lib64/libm.so.6 (0x0000003d53400000)
	libpthread.so.0 => /lib64/libpthread.so.0 (0x0000003d53800000)
	libc.so.6 => /lib64/libc.so.6 (0x0000003d53000000)
	libz.so.1 => /lib64/libz.so.1 (0x0000003d54800000)
	/lib64/ld-linux-x86-64.so.2 (0x0000003d52c00000)