[petsc-users] MatMFFDSetBase

Wed Jun 20 14:11:57 CDT 2018

Hi Barry

When I changed it to CHKFORTRANNULLOBJECTDEREFERENCE and commented out the
duplicated stuff in the auto wrapper it worked just fine.

Thanks,
Gaetan

On Tue, Jun 19, 2018 at 2:55 PM Smith, Barry F. <bsmith at mcs.anl.gov> wrote:

>
>   Could you please try changing it to CHKFORTRANNULLOBJECTDEREFERENCE;
> recompile and run again
>
>    Barry
>
>
> > On Jun 19, 2018, at 3:28 PM, Gaetan Kenway <gaetank at gmail.com> wrote:
> >
> > I tried the patch. It initially didn't compile because of a conflict
> with the corresponding ftn-auto file. When i commented out the stuff there
> it compiled, but I got a segfault here:
> >
> >     at
> /nobackup/gkenway/packages/petsc-3.9.2/src/mat/impls/mffd/ftn-custom/zmffdf.c:52
> > 52     *ierr = MatMFFDSetBase(*mat,*u,*f);
> > (gdb) bt
> > #0  matmffdsetbase_ (mat=mat at entry=0x7fffb461ba10
> <__anksolver_MOD_drdw>, u=u at entry=0x7fffb461b9a0 <__anksolver_MOD_wvec>,
> f=0x0, f at entry=0x7fffffffb970, ierr=ierr at entry=0x7fffffffb938)
> >     at
> /nobackup/gkenway/packages/petsc-3.9.2/src/mat/impls/mffd/ftn-custom/zmffdf.c:52
> > #1  0x00007fffb3d7e871 in anksolver::ankstep (firstcall=<optimized out>)
> at ../NKSolver/NKSolvers.F90:2891
> > #2  0x00007fffb3ebbde9 in solvers::solvestate () at
> ../solver/solvers.F90:1114
> > #3  0x00007fffb3ebe4b5 in solvers::solver () at ../solver/solvers.F90:69
> > #4  0x00007fffb3cfcaa7 in f2py_rout_libadflow_solvers_solver
> (capi_self=<optimized out>, capi_args=<optimized out>,
> capi_keywds=<optimized out>, f2py_func=0x7fffb3ebe310 <solvers::solver>)
> >     at libadflowmodule.c:6710
> > #5  0x00007fffb3cf8643 in fortran_call (fp=<optimized out>,
> arg=<optimized out>, kw=<optimized out>) at
> /u/wk/gkenway/.local/lib/python2.7/site-packages/numpy/f2py/src/fortranobject.c:404
> > #6  0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > #7  0x00007ffff7aef236 in PyEval_EvalFrameEx () from
> /lib64/libpython2.7.so.1.0
> > #8  0x00007ffff7af603d in PyEval_EvalCodeEx () from
> /lib64/libpython2.7.so.1.0
> > #9  0x00007ffff7a7f978 in function_call () from
> /lib64/libpython2.7.so.1.0
> > #10 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > #11 0x00007ffff7a69a55 in instancemethod_call () from
> /lib64/libpython2.7.so.1.0
> > #12 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > #13 0x00007ffff7ab1e27 in slot_tp_call () from /lib64/libpython2.7.so.1.0
> > #14 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > #15 0x00007ffff7aef236 in PyEval_EvalFrameEx () from
> /lib64/libpython2.7.so.1.0
> > #16 0x00007ffff7af603d in PyEval_EvalCodeEx () from
> /lib64/libpython2.7.so.1.0
> > #17 0x00007ffff7af6142 in PyEval_EvalCode () from
> /lib64/libpython2.7.so.1.0
> > #18 0x00007ffff7b0f57f in run_mod () from /lib64/libpython2.7.so.1.0
> > #19 0x00007ffff7b1073e in PyRun_FileExFlags () from
> /lib64/libpython2.7.so.1.0
> > #20 0x00007ffff7b119c9 in PyRun_SimpleFileExFlags () from
> /lib64/libpython2.7.so.1.0
> > #21 0x00007ffff7b22b7f in Py_Main () from /lib64/libpython2.7.so.1.0
> > #22 0x00007ffff6d3f3d5 in __libc_start_main () from /lib64/libc.so.6
> > #23 0x000000000040066e in _start ()
> >
> > which was the new code that got added. So I never got the patch to work
> correctly.
> >
> > However, I circumvented the problem by making my own vector and calling
> the matfd-function, and then passing in the vector. This is essentially the
> same thing and I means I don't need an if-def for older versions.
> >
> > Gaetan
> >
> >
> >
> > On Tue, Jun 19, 2018 at 11:24 AM Smith, Barry F. <bsmith at mcs.anl.gov>
> wrote:
> >
> >     The code to handle the null vector got mistakenly removed. Please
> find attached a patch that should resolve the problem.
> >
> >
> >
> >     Please let us know if it works,
> >
> >      Thanks
> >
> >       Barry
> >
> >
> > > On Jun 19, 2018, at 12:36 PM, Gaetan Kenway <gaetank at gmail.com> wrote:
> > >
> > > Seems like the MatMFFDSetBase is now broken in petsc 3.9
> > >
> > > In 3.7 I used:
> > >
> > > call MatMFFDSetBase(dRdW, wVec, PETSC_NULL_OBJECT, ierr)
> > >
> > > and the most logical translation is now
> > >
> > > call MatMFFDSetBase(dRdW, wVec, PETSC_NULL_VEC, ierr)
> > >
> > > Unfortunately this just segfaults:
> > > #0  VecAXPY (y=y at entry=0x10a3750, alpha=alpha at entry=-1, x=x at entry=0xffffffffffffffff)
> at
> /nobackup/gkenway/packages/petsc-3.9.2/src/vec/vec/interface/rvector.c:601
> > > #1  0x00007fffb697faae in MatMult_MFFD (mat=0x291a150, a=0x10b25c0,
> y=0x10a3750) at
> /nobackup/gkenway/packages/petsc-3.9.2/src/mat/impls/mffd/mffd.c:368
> > > #2  0x00007fffb6847af2 in MatMult (mat=0x291a150, x=x at entry=0x10b25c0,
> y=y at entry=0x10a3750) at
> /nobackup/gkenway/packages/petsc-3.9.2/src/mat/interface/matrix.c:2305
> > > #3  0x00007fffb6e982e7 in PCApplyBAorAB (pc=0xe64d00, side=PC_RIGHT,
> x=0x10aaee0, y=y at entry=0x10a3750, work=0x10b25c0) at
> /nobackup/gkenway/packages/petsc-3.9.2/src/ksp/pc/interface/precon.c:680
> > > #4  0x00007fffb6f02b84 in KSP_PCApplyBAorAB (w=<optimized out>,
> y=0x10a3750, x=<optimized out>, ksp=0xf135e0) at
> /u/wk/gkenway/scratch/packages/petsc-3.9.2/include/petsc/private/kspimpl.h:304
> > > #5  KSPGMRESCycle (itcount=itcount at entry=0x7fffffffb73c, ksp=ksp at entry=0xf135e0)
> at
> /nobackup/gkenway/packages/petsc-3.9.2/src/ksp/ksp/impls/gmres/gmres.c:152
> > > #6  0x00007fffb6f035c2 in KSPSolve_GMRES (ksp=0xf135e0) at
> /nobackup/gkenway/packages/petsc-3.9.2/src/ksp/ksp/impls/gmres/gmres.c:234
> > > #7  0x00007fffb6f35050 in KSPSolve (ksp=0xf135e0, b=0xf7ced0,
> x=0xf77be0) at
> /nobackup/gkenway/packages/petsc-3.9.2/src/ksp/ksp/interface/itfunc.c:669
> > > #8  0x00007fffb6f482ee in kspsolve_ (ksp=ksp at entry=0x7fffb461bae8
> <__anksolver_MOD_ank_ksp>, b=<optimized out>, b at entry=0x7fffb461b9c8
> <__anksolver_MOD_rvec>, x=<optimized out>,
> > >     x at entry=0x7fffb461ba18 <__anksolver_MOD_deltaw>, ierr=ierr at entry=0x7fffffffb938)
> at
> /nobackup/gkenway/packages/petsc-3.9.2/src/ksp/ksp/interface/ftn-custom/zitfuncf.c:54
> > > #9  0x00007fffb3d7e697 in anksolver::ankstep (firstcall=<optimized
> out>) at ../NKSolver/NKSolvers.F90:2891
> > > #10 0x00007fffb3ebbbb9 in solvers::solvestate () at
> ../solver/solvers.F90:1114
> > > #11 0x00007fffb3ebe285 in solvers::solver () at
> ../solver/solvers.F90:69
> > > #12 0x00007fffb3cfca17 in f2py_rout_libadflow_solvers_solver
> (capi_self=<optimized out>, capi_args=<optimized out>,
> capi_keywds=<optimized out>, f2py_func=0x7fffb3ebe0e0 <solvers::solver>)
> > >     at libadflowmodule.c:6710
> > > #13 0x00007fffb3cf85b3 in fortran_call (fp=<optimized out>,
> arg=<optimized out>, kw=<optimized out>) at
> /u/wk/gkenway/.local/lib/python2.7/site-packages/numpy/f2py/src/fortranobject.c:404
> > > #14 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > > #15 0x00007ffff7aef236 in PyEval_EvalFrameEx () from
> /lib64/libpython2.7.so.1.0
> > > #16 0x00007ffff7af603d in PyEval_EvalCodeEx () from
> /lib64/libpython2.7.so.1.0
> > > #17 0x00007ffff7a7f978 in function_call () from
> /lib64/libpython2.7.so.1.0
> > > #18 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > > #19 0x00007ffff7a69a55 in instancemethod_call () from
> /lib64/libpython2.7.so.1.0
> > > #20 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > > #21 0x00007ffff7ab1e27 in slot_tp_call () from
> /lib64/libpython2.7.so.1.0
> > > #22 0x00007ffff7a5aa63 in PyObject_Call () from
> /lib64/libpython2.7.so.1.0
> > > #23 0x00007ffff7aef236 in PyEval_EvalFrameEx () from
> /lib64/libpython2.7.so.1.0
> > > #24 0x00007ffff7af603d in PyEval_EvalCodeEx () from
> /lib64/libpython2.7.so.1.0
> > > #25 0x00007ffff7af6142 in PyEval_EvalCode () from
> /lib64/libpython2.7.so.1.0
> > > #26 0x00007ffff7b0f57f in run_mod () from /lib64/libpython2.7.so.1.0
> > > #27 0x00007ffff7b1073e in PyRun_FileExFlags () from
> /lib64/libpython2.7.so.1.0
> > > #28 0x00007ffff7b119c9 in PyRun_SimpleFileExFlags () from
> /lib64/libpython2.7.so.1.0
> > > #29 0x00007ffff7b22b7f in Py_Main () from /lib64/libpython2.7.so.1.0
> > > #30 0x00007ffff6d3f3d5 in __libc_start_main () from /lib64/libc.so.6
> > > #31 0x000000000040066e in _start ()
> > >
> > > Thanks,
> > >
> > > Gaetan
> > >
> > >
> > >
> > >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20180620/534f0b2a/attachment.html>