[petsc-users] PETSC ERROR: Can only handle MPIU_REAL or MPIU_COMPLEX with ML preconditioner

[Matthew Knepley]

On Wed, Oct 1, 2014 at 9:32 AM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
>   Recommend running under valgrind.
> http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind


Barry, it looks like this commit introduced it


https://bitbucket.org/petsc/petsc/commits/7be6b909d0172a4ed9ffbe3833ba966bb47c2681#Lsrc/ksp/pc/impls/ml/ml.cT422

and it should fail according to


https://bitbucket.org/petsc/petsc/src/643e5bfe5b452fa5bc89efaa4c603d713f2a6f29/src/sys/objects/pinit.c?at=master#cl-284

Why don't the nightly tests fail?

   Matt

On Oct 1, 2014, at 7:03 AM, Chris Richardson <chris at bpi.cam.ac.uk> wrote:
>
> > On 01/10/2014 11:33, Matthew Knepley wrote:
> >> On Wed, Oct 1, 2014 at 3:51 AM, Chris Richardson <chris at bpi.cam.ac.uk>
> >> wrote:
> >>> I have just run into this error after pulling the master branch of
> >>> petsc, when using the ML preconditioner in FEniCS.
> >>> It looks like a type error, but I guess FEniCS uses PETScInt, so
> >>> should be safe?
> >>> PETSC ERROR: Can only handle MPIU_REAL or MPIU_COMPLEX data types
> >> We need to see the entire error (with the stack).
> >
> > OK, this kind of thing:-
> >
> >
> > [1]PETSC ERROR: Can only handle MPIU_REAL or MPIU_COMPLEX data types
> > ^C
> > Program received signal SIGINT, Interrupt.
> > 0x00007ffff64d9653 in epoll_wait () at
> ../sysdeps/unix/syscall-template.S:81
> > 81      ../sysdeps/unix/syscall-template.S: No such file or directory.
> > (gdb) up
> > #1  0x00007ffff72c8591 in ?? () from /usr/lib/libmpi.so.1
> > (gdb) up
> > #2  0x00007ffff72ca787 in opal_event_base_loop () from
> /usr/lib/libmpi.so.1
> > (gdb) up
> > #3  0x00007ffff72eda3e in opal_progress () from /usr/lib/libmpi.so.1
> > (gdb) up
> > #4  0x00007fffb42884d5 in ?? ()
> >   from /usr/lib/openmpi/lib/openmpi/mca_grpcomm_bad.so
> > (gdb) up
> > #5  0x00007ffff723d02a in ompi_mpi_finalize () from /usr/lib/libmpi.so.1
> > (gdb) up
> > #6  0x00007ffff7766b12 in dolfin::SubSystemsManager::finalize_mpi ()
> >    at /opt/packages/src/dolfin/dolfin/common/SubSystemsManager.cpp:249
> > 249           MPI_Finalize();
> > (gdb) up
> > #7  0x00007ffff7766cde in dolfin::SubSystemsManager::finalize ()
> >    at /opt/packages/src/dolfin/dolfin/common/SubSystemsManager.cpp:219
> > 219       finalize_mpi();
> > (gdb) up
> > #8  0x00007ffff641a149 in __run_exit_handlers (status=1,
> >    listp=0x7ffff679d6c8 <__exit_funcs>,
> >    run_list_atexit=run_list_atexit at entry=true) at exit.c:82
> > 82      exit.c: No such file or directory.
> > (gdb) up
> > #9  0x00007ffff641a195 in __GI_exit (status=<optimised out>) at
> exit.c:104
> > 104     in exit.c
> > (gdb) up
> > #10 0x00007ffff729cc00 in orte_ess_base_app_abort () from
> /usr/lib/libmpi.so.1
> > (gdb) up
> > #11 0x00007ffff729c2a9 in orte_errmgr_base_error_abort ()
> >   from /usr/lib/libmpi.so.1
> > (gdb) up
> > #12 0x00007ffff723bb69 in ompi_mpi_abort () from /usr/lib/libmpi.so.1
> > (gdb) up
> > #13 0x00007ffff37b9ee3 in PetscSum_Local (in=0x230f4520,
> out=0x7fffffffc650,
> >    cnt=<optimised out>, datatype=0x7fffffffc428)
> >    at /opt/packages/src/petsc/src/sys/objects/pinit.c:302
> > 302         MPI_Abort(MPI_COMM_WORLD,1);
> > (gdb) up
> > #14 0x00007fffb10dd4dc in mca_coll_basic_scan_intra ()
> >   from /usr/lib/openmpi/lib/openmpi/mca_coll_basic.so
> > (gdb) up
> > #15 0x00007fffb08c7e19 in mca_coll_sync_scan ()
> >   from /usr/lib/openmpi/lib/openmpi/mca_coll_sync.so
> > (gdb) up
> > #16 0x00007ffff725835d in PMPI_Scan () from /usr/lib/libmpi.so.1
> > (gdb) up
> > #17 0x00007ffff413ec45 in MatWrapML_MPIAIJ (newmat=0x16a39228,
> >    reuse=MAT_INITIAL_MATRIX, mlmat=0x7ed6090)
> >    at /opt/packages/src/petsc/src/ksp/pc/impls/ml/ml.c:422
> > 422         ierr =
> MPI_Scan(&m,&rstart,1,MPIU_INT,MPIU_SUM,mlmat->comm->USR_com);CHKERRQ(ierr);
> > (gdb) list
> > 417         ierr =
> MatSetSizes(A,m,n,PETSC_DECIDE,PETSC_DECIDE);CHKERRQ(ierr);
> > 418         ierr = MatSetType(A,MATMPIAIJ);CHKERRQ(ierr);
> > 419         /* keep track of block size for A matrices */
> > 420         ierr = MatSetBlockSize (A,mlmat->num_PDEs);CHKERRQ(ierr);
> > 421         ierr = PetscMalloc3(m,&nnzA,m,&nnzB,m,&nnz);CHKERRQ(ierr);
> > 422         ierr =
> MPI_Scan(&m,&rstart,1,MPIU_INT,MPIU_SUM,mlmat->comm->USR_com);CHKERRQ(ierr);
> > 423         rstart -= m;
> > 424
> > 425         for (i=0; i<m; i++) {
> > 426           row = gordering[i] - rstart;
> > (gdb)
>
>


-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20141001/0d3cde89/attachment.html>