Okay, I uninstalled the mpich packages, and the default compiler reverted to openmpi. Not sure how the mpich packages snuck in there. Anyway, I recompiled with openmpi, and everything seems to work dandy. I still get the warnings because of the swig issue, but using enableDebug=no allows the wrapper to compile, and all the demos work great. Thanks for your help!
Cheers, Tim On Tue, Aug 26, 2008 at 3:36 AM, Garth N. Wells <[EMAIL PROTECTED]> wrote: > > > Timothy Crone wrote: >> >> Well that is very strange. I am also using all Lenny packages >> (openmpi, petsc, etc.) except for FEniCS libraries. On my machine >> /usr/bin/mpicxx is a link pointing to a script called >> /usr/lib/mpich-shmem/bin/mpicxx, which appears to choose which >> compiler to use. > > I'm using openmpi. Try setting the environment variable CXX to point to the > openmpi compiler. mpich provided by one of the the Debian mpich packages > (which are MPI version 1.0) is unlikely to work. Easiest approach would be > to just remove the mpich packages. > > Garth > > Do you have an environment variable set to choose a >> >> non-default compiler? >> >> Thanks, >> Tim >> >> >> >> On Mon, Aug 25, 2008 at 5:51 PM, Garth N. Wells <[EMAIL PROTECTED]> wrote: >>> >>> It works perfectly for me under Lenny. Which MPI are you using? I used >>> Lenny >>> packages for all non-FEniCS libraries (petsc, openmpi, etc). >>> >>> Garth >>> >>> Timothy Crone wrote: >>>> >>>> Thanks for your reply Kent. I used enableDebug=no to get >>>> dolfin_wrap.cc to compile, because otherwise warnings are treated as >>>> errors. Not sure if this will affect the output from gdb. Anyway, here >>>> is the output from your suggestion: >>>> >>>> GNU gdb 6.8-debian >>>> Copyright (C) 2008 Free Software Foundation, Inc. >>>> License GPLv3+: GNU GPL version 3 or later >>>> <http://gnu.org/licenses/gpl.html> >>>> This is free software: you are free to change and redistribute it. >>>> There is NO WARRANTY, to the extent permitted by law. Type "show >>>> copying" >>>> and "show warranty" for details. >>>> This GDB was configured as "x86_64-linux-gnu"... >>>> (gdb) run demo.py >>>> Starting program: /usr/bin/python demo.py >>>> [Thread debugging using libthread_db enabled] >>>> [New Thread 0x7fe52a4266e0 (LWP 17418)] >>>> >>>> Program received signal SIGSEGV, Segmentation fault. >>>> [Switching to Thread 0x7fe52a4266e0 (LWP 17418)] >>>> 0x00007fe51fdadf94 in MPID_CH_InitMsgPass () from >>>> /home/tjc/computer/fenics/dolfin/dolfin/dolfin/libdolfin.so >>>> Current language: auto; currently asm >>>> (gdb) where >>>> #0 0x00007fe51fdadf94 in MPID_CH_InitMsgPass () from >>>> /home/tjc/computer/fenics/dolfin/dolfin/dolfin/libdolfin.so >>>> #1 0x00007fe51fdab053 in MPID_Init () from >>>> /home/tjc/computer/fenics/dolfin/dolfin/dolfin/libdolfin.so >>>> #2 0x00007fe51fd8faa0 in MPIR_Init () from >>>> /home/tjc/computer/fenics/dolfin/dolfin/dolfin/libdolfin.so >>>> #3 0x00007fe51fcf1271 in dolfin::SubSystemsManager::initMPI () at >>>> /usr/lib/openmpi/include/openmpi/ompi/mpi/cxx/functions_inln.h:104 >>>> #4 0x00007fe51fcf1069 in dolfin::MPI::numProcesses () at >>>> dolfin/main/MPI.cpp:29 >>>> #5 0x00007fe51fcf1099 in dolfin::MPI::receive () at >>>> dolfin/main/MPI.cpp:42 >>>> #6 0x00007fe51fd2afab in UnitSquare (this=0x2ba2390, nx=32, ny=32, >>>> type=dolfin::UnitSquare::right) at dolfin/mesh/UnitSquare.cpp:20 >>>> #7 0x00007fe522e031e4 in _wrap_new_UnitSquare (self=<value optimized >>>> out>, args=<value optimized out>) at dolfin/swig/dolfin_wrap.cc:90885 >>>> #8 0x00000000004187b3 in PyObject_Call (func=0x0, arg=0x0, >>>> kw=0x7fff324447a1) at ../Objects/abstract.c:1861 >>>> #9 0x0000000000490257 in PyEval_EvalFrameEx (f=0x2ceb3e0, >>>> throwflag=<value optimized out>) at ../Python/ceval.c:3855 >>>> #10 0x00000000004930fd in PyEval_EvalCodeEx (co=0x2b5f558, >>>> globals=<value optimized out>, locals=<value optimized out>, >>>> args=0x7fe52a3d3c98, >>>> argcount=3, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) >>>> at ../Python/ceval.c:2838 >>>> #11 0x00000000004dd692 in function_call (func=0x2c44b18, >>>> arg=0x7fe52a3d3c80, kw=0x0) at ../Objects/funcobject.c:517 >>>> #12 0x00000000004187b3 in PyObject_Call (func=0x0, arg=0x0, >>>> kw=0x7fff324447a1) at ../Objects/abstract.c:1861 >>>> #13 0x000000000041f658 in instancemethod_call (func=0x2c44b18, >>>> arg=0x7fe52a3d3c80, kw=0x0) at ../Objects/classobject.c:2519 >>>> #14 0x00000000004187b3 in PyObject_Call (func=0x0, arg=0x0, >>>> kw=0x7fff324447a1) at ../Objects/abstract.c:1861 >>>> #15 0x0000000000465ca6 in slot_tp_init (self=0x7fe52a3df6d0, >>>> args=0x7fe52a3e0ea8, kwds=0x0) at ../Objects/typeobject.c:4943 >>>> #16 0x000000000046df9b in type_call (type=0x2ceaa40, >>>> args=0x7fe52a3e0ea8, kwds=0x0) at ../Objects/typeobject.c:436 >>>> #17 0x00000000004187b3 in PyObject_Call (func=0x0, arg=0x0, >>>> kw=0x7fff324447a1) at ../Objects/abstract.c:1861 >>>> #18 0x000000000048feb2 in PyEval_EvalFrameEx (f=0x2245650, >>>> throwflag=<value optimized out>) at ../Python/ceval.c:3786 >>>> #19 0x00000000004930fd in PyEval_EvalCodeEx (co=0x7fe52a3d9288, >>>> globals=<value optimized out>, locals=<value optimized out>, args=0x0, >>>> argcount=0, >>>> kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at >>>> ../Python/ceval.c:2838 >>>> #20 0x00000000004932f2 in PyEval_EvalCode (co=0x0, globals=0x0, >>>> locals=0x7fff324447a1) at ../Python/ceval.c:494 >>>> #21 0x00000000004b2d08 in PyRun_FileExFlags (fp=0x222c010, >>>> filename=0x7fff324475a1 "demo.py", start=<value optimized out>, >>>> globals=0x224f580, >>>> locals=0x224f580, closeit=1, flags=0x7fff32445640) at >>>> ../Python/pythonrun.c:1273 >>>> #22 0x00000000004b2fab in PyRun_SimpleFileExFlags (fp=0x222c010, >>>> filename=0x7fff324475a1 "demo.py", closeit=1, flags=0x7fff32445640) >>>> at ../Python/pythonrun.c:879 >>>> ---Type <return> to continue, or q <return> to quit--- >>>> #23 0x0000000000414542 in Py_Main (argc=1, argv=<value optimized out>) >>>> at ../Modules/main.c:532 >>>> #24 0x00007fe5296511a6 in __libc_start_main () from /lib/libc.so.6 >>>> #25 0x0000000000413989 in _start () >>>> (gdb) >>>> >>>> By the way, the Lorenz Python demo seems to work okay, although it >>>> warns "Not solving the dual problem as requested" at the end. >>>> >>>> Thanks, >>>> Tim >>>> >>>> On Mon, Aug 25, 2008 at 4:02 PM, <[EMAIL PROTECTED]> wrote: >>>>> >>>>> Can you try with a debugger: >>>>> >>>>> gdb python >>>>> (gdb) run demo.py >>>>> .... >>>>> (gdb) where >>>>> >>>>> It is strange if these warnings have anything to do with the seg. >>>>> fault. >>>>> >>>>> Kent >>>>> >>>>> >>>>>> Hello, >>>>>> >>>>>> Does anyone know a workaround to the problem of compiling the Pydolfin >>>>>> file dolfin_wrap.cc with mpicxx version 4.3.1? I get several warnings >>>>>> during compile, and segmentation faults at runtime. I am using Debian >>>>>> Lenny, which uses 4.3.1 as the default compiler. SWIG is version >>>>>> 1.3.33. Any help in this matter would be greatly appreciated. >>>>>> >>>>>> Here are the errors when enableDebug=no: >>>>>> >>>>>> mpicxx -o dolfin/swig/dolfin_wrap.os -c -Wall -pipe -ansi -DNDEBUG -O3 >>>>>> -DHAS_MPI=1 -DMPICH_IGNORE_CXX_SEEK -DPACKAGE_VERSION=\"0.8.0\" >>>>>> -DHAS_PETSC=1 -DHAS_UMFPACK=1 -DHAS_GTS=1 -DHAS_CHOLMOD=1 >>>>>> -fno-strict-aliasing -fPIC -I. -I/usr/include/libxml2 >>>>>> -I/usr/lib/petscdir/2.3.3/bmake/linux-gnu-c-opt >>>>>> -I/usr/lib/petscdir/2.3.3/include -I/usr/lib/openmpi/include >>>>>> -I/usr/lib/openmpi/lib -Idolfin -I/usr/include/suitesparse >>>>>> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include >>>>>> -I/usr/include/suitesparse -I/usr/include/python2.5 >>>>>> -I/usr/include/python2.5 >>>>>> -I/usr/lib/python2.5/site-packages/numpy/core/include >>>>>> -I/usr/include/libxml2 -I/usr/lib/petscdir/2.3.3/bmake/linux-gnu-c-opt >>>>>> -I/usr/lib/petscdir/2.3.3/include -I/usr/lib/openmpi/include >>>>>> -I/usr/lib/openmpi/lib -Idolfin -I/usr/include/suitesparse >>>>>> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include >>>>>> -I/usr/include/suitesparse dolfin/swig/dolfin_wrap.cc >>>>>> dolfin/swig/dolfin_wrap.cc: In constructor >>>>>> 'swig::PyObject_ptr::PyObject_ptr(PyObject*, bool)': >>>>>> dolfin/swig/dolfin_wrap.cc:3286: warning: suggest explicit braces to >>>>>> avoid ambiguous 'else' >>>>>> dolfin/swig/dolfin_wrap.cc: In destructor >>>>>> 'swig::PySequence_Cont<T>::~PySequence_Cont()': >>>>>> dolfin/swig/dolfin_wrap.cc:4727: warning: suggest explicit braces to >>>>>> avoid ambiguous 'else' >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_summary(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:29893: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_ios_base_sync_with_stdio(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:17337: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_TimeDependent(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:28067: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_BoundaryMesh(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:90450: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_MeshTopology(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:74482: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_MeshConnectivity(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:76492: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_MeshGeometry(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:75192: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_Vector(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:50862: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_uBLASVector(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:41746: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_uBLASILUPreconditioner(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:50585: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_PETScVector(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:44047: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_Table(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:30960: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> dolfin/swig/dolfin_wrap.cc: In function 'PyObject* >>>>>> _wrap_new_Mesh(PyObject*, PyObject*)': >>>>>> dolfin/swig/dolfin_wrap.cc:78823: warning: 'argv[0]' may be used >>>>>> uninitialized in this function >>>>>> mpicxx -o dolfin/swig/_dolfin.so -pthread -Wl,--export-dynamic -shared >>>>>> dolfin/swig/dolfin_wrap.os -Ldolfin >>>>>> -L/usr/lib/petscdir/2.3.3/lib/linux-gnu-c-opt -L/usr/lib/atlas >>>>>> -L/usr/lib/atlas -lxml2 -lpetscts -lpetscsnes -lpetscksp -lpetscdm >>>>>> -lpetscmat -lpetscvec -lpetsc -lblas -lumfpack -lamd -lgts -lm >>>>>> -lgthread-2.0 -lrt -lgmodule-2.0 -ldl -lglib-2.0 -llapack -lblas >>>>>> -lcholmod -lamd -lcolamd -ldolfin >>>>>> scons: done building targets. >>>>>> >>>>>> Thanks, >>>>>> Tim >>>>>> _______________________________________________ >>>>>> DOLFIN-dev mailing list >>>>>> [email protected] >>>>>> http://www.fenics.org/mailman/listinfo/dolfin-dev >>>>>> >>>>> >>>> >>>> >>> >> >> >> > -- Timothy J. Crone Lamont-Doherty Earth Observatory Columbia University 61 Route 9W Palisades, NY 10964 Tel: 845-365-8687 http://dactyl.ldeo.columbia.edu _______________________________________________ DOLFIN-dev mailing list [email protected] http://www.fenics.org/mailman/listinfo/dolfin-dev
