[mpich-discuss] Mac OS X and gfortran: undefined symbols

[Anthony Chan]

You don't need to do any of things that you described
if MPICH2's Fortran binding is built correctly.
Can you send us your configure and make outputs
as stated in the MPICH2's README ?

A.Chan

----- Original Message -----
> I am having trouble linking MPICH2 with routines compiled by gfortran
> on OS X.
> The computer is running OS 10.5.8 (but there are the same problems
> with
> another computer running 10.6.5)
> 
> With ifort, I can link to a version of libmpich.a that was build for
> Linux. There is a warning
> ld warning: in /c/nm72/mpi/mpi_lini/libmpich.a, file is not of
> required
> architecture
> but the run is successfully parallelized.
> 
> However, with gfortran, there are messages such as:
> Undefined symbols:
> "_mpi_recv_", referenced from:
> _pnm_mpi_recv_ in pnm_mpi.o
> 
> Are there any known problems with this platform?
> Is there a specific library that is needed instead of libmpich.a?
> ==========
> THe compiler is:
> gfortran -v
> Using built-in specs.
> Target: i386-apple-darwin8.10.1
> Configured with: /tmp/gfortran-20081211/ibin/../gcc/configure
> --prefix=/usr/local/gfortran --enable-languages=c,fortran
> --with-gmp=/tmp/gfortran-20081211/gfortran_libs --enable-bootstrap
> Thread model: posix
> gcc version 4.4.0 20081211 (experimental) [trunk revision 142675]
> (GCC)
> 
> 
> I have tried building MPICH2 (and libmpich.a)
> with many variations:
> 
> with and without
> --enable-sharedlibs=osx-gcc
> 
> The routines that call mpi_recv etc. have been compiled with and
> without gfortran options such as
> -fno-underscoring
> 
> and with and without Interface blocks such as
> SUBROUTINE
> MPI_RECV(IBUFF,N,IDATATYPE,IMODE1,IMODE2,ICOMM,ISTATUS,IERR)
> BIND(c, name='mpi_recv__')
> ! int MPI_Recv(void*, int, MPI_Datatype, int, int, MPI_Comm,
> MPI_Status
> *);
> INTEGER IBUFF(*),N,IDATATYPE,IMODE1,IMODE2,ICOMM,ISTATUS(*),IERR
> END SUBROUTINE MPI_RECV
> 
> --
> Alison Boeckmann
> alisonboeckmann at fastmail.fm
> 
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss