[MPICH] LoadLibrary("mpich2d.dll") fails, how can I link statically on Windows?
David Ashton
ashton at mcs.anl.gov
Fri Aug 19 02:30:05 CDT 2005
Robin Boerdijk,
Yes, I forgot about those changes. We moved the MPI_F_STATUS(ES)_IGNORE
variables from the Fortran library to the C library. Comment out the lines
in mpi.c that reference them. If you plan on using those variables, you
will have to wait for the next release.
-David Ashton
-----Original Message-----
From: owner-mpich-discuss at mcs.anl.gov
[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Robin Boerdijk
Sent: Friday, August 19, 2005 1:21 AM
To: David Ashton; mpich-discuss at mcs.anl.gov
Subject: RE: [MPICH] LoadLibrary("mpich2d.dll") fails, how can I link
statically on Windows?
Hi David,
> 1) This is a problem with the MPI wrapper dll. We've fixed this and
> it will
> be available in the next release. The changes are all in one file,
> mpich2/src/util/multichannel/mpi.c. You can grab the new version
> from CVS
> if you want to recompile the source.
> http://www-unix.mcs.anl.gov/mpi/mpich2/cvsweb.cgi
I recompiled MPICH2 with the latest mpi.c, but now my MPI program fails
to start with the following message displayed by Windows: "First-chance
exception at 0x1002885f (mpich2mpid.dll) in MyMPIProgram.exe:
0xC0000005: Access violation writing location 0x00000000.".
When I run my MPI program in the debugger it breaks at the following
line in mpi.c:
int MPI_Init( int *argc, char ***argv )
{
int result;
MPICH_CHECK_INIT(FCNAME);
result = fn.MPI_Init(argc, argv);
*fn.MPI_F_STATUS_IGNORE = MPI_F_STATUS_IGNORE; // <- crash
*fn.MPI_F_STATUSES_IGNORE = MPI_F_STATUSES_IGNORE;
return result;
}
Apparently, the
fn.MPI_F_STATUS_IGNORE = (MPI_Fint**)GetProcAddress(hPMPIModule,
"MPI_F_STATUS_IGNORE");
statement in LoadFunctions() has set fn.MPI_F_STATUS_IGNORE to NULL.
Do I need to update some more files to make it work?
Best regards,
Robin Boerdijk
____________________________________________________
Start your day with Yahoo! - make it your home page
http://www.yahoo.com/r/hs
More information about the mpich-discuss
mailing list