24. mai 2014 kl. 00:09 skrev Imran Ali <[email protected]>:

> On 2014-05-21 15:23, Johannes Ring wrote:
>> On Wed, May 21, 2014 at 3:20 PM, Imran Ali
>> <[email protected]> wrote:
>>> On 2014-05-21 13:52, Johannes Ring wrote:
>>>> On Wed, May 21, 2014 at 1:35 PM, Imran Ali
>>>> <[email protected]> wrote:
>>>>> On 2014-05-20 15:16, Johannes Ring wrote:
>>>>>> On Tue, May 20, 2014 at 2:55 PM, Imran Ali
>>>>>> <[email protected]> wrote:
>>>>>>> I commented out the following in the cmake file :
>>>>>>> ## Try to figure out PETSC_DIR by finding petsc.h
>>>>>>> #find_path(PETSC_DIR include/petsc.h
>>>>>>> #  HINTS ${PETSC_DIR} $ENV{PETSC_DIR}
>>>>>>> #  PATHS ${petsc_dir_locations}
>>>>>>> #  DOC "PETSc directory")
>>>>>>> message("Skipped find petsc")
>>>>>>> # Report result of search for PETSC_DIR
>>>>>>> if (DEFINED PETSC_DIR)
>>>>>>>  message(STATUS "PETSC_DIR is ${PETSC_DIR}")
>>>>>>> else()
>>>>>>>  message(STATUS "PETSC_DIR is empty")
>>>>>>> endif()
>>>>>>> And the build went successfully, however without being able to find
>>>>>>> PETSc.
>>>>>>> Which I need.
>>>>>> I was thinking more of disabling some of the try_run calls in
>>>>>> FindPETSc.cmake. However, you can also try to turn off
>>>>>> DOLFIN_SKIP_BUILD_TESTS and then apply the following patch to
>>>>>> FindPETSc.cmake:
>>>>>>   # Build PETSc test program
>>>>>>  -if (DOLFIN_SKIP_BUILD_TESTS)
>>>>>>  +if (1)
>>>>>>     set(PETSC_TEST_RUNS TRUE)
>>>>>>  -  set(PETSC_VERSION "UNKNOWN")
>>>>>>  +  set(PETSC_VERSION "3.4.4")
>>>>>>     set(PETSC_VERSION_OK TRUE)
>>>>>>   elseif (FOUND_PETSC_CONF)
>>>>>> Change the version number according to your PETSc version.
>>>>>> Johannes
>>>>> I added this in the PETSc cmake file (along with skip build in dolfin
>>>>> script), however the build would crash with the following message
>>>>> CMake Error at CMakeLists.txt:335 (if):
>>>>>  if given arguments:
>>>>>    "NOT" "(" "3" "EQUAL" "AND" "4" "EQUAL" ")"
>>>>>  Unknown arguments specified
>>>> Haven't we already been through this?
>>>> http://fenicsproject.org/pipermail/fenics-support/2014-May/000552.html
>>>> Johannes
>>> Thanks! That did it. The dolfin build went successfully.
>> Good. However, the original problem was that you couldn't run the
>> poisson demo. Does that work now?
>> Johannes
> 
> Hi and sorry for the delayed answer. I have reinstalled every thing from 
> beginning after removing the Work folder, and decided to go for python2.6.

I forgot to mention that this was done after that the same error reoccured 
(after following the steps above).


> 
> I ran the cpp demo, and get the following segmentation fault :
> 
> $ ./demo_poisson
> [bell:14359] *** Process received signal ***
> [bell:14359] Signal: Segmentation fault (11)
> [bell:14359] Signal code: Address not mapped (1)
> [bell:14359] Failing at address: 0x1
> [bell:14359] [ 0] /lib64/libc.so.6[0x374fc329a0]
> [bell:14359] [ 1] /lib64/libc.so.6(_IO_vfprintf+0x3e5c)[0x374fc4812c]
> [bell:14359] [ 2] /lib64/libc.so.6(vasprintf+0xba)[0x374fc6f6da]
> [bell:14359] [ 3] /lib64/libc.so.6(asprintf+0x88)[0x374fc4f648]
> [bell:14359] [ 4] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libmpi.so.1(ompi_mpi_init+0x306)[0x7f853c87b9b6]
> [bell:14359] [ 5] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libmpi.so.1(MPI_Init_thread+0x184)[0x7f853c89f4e4]
> [bell:14359] [ 6] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin17SubSystemsManager8init_mpiEiPPci+0x9e)[0x7f85455bdaee]
> [bell:14359] [ 7] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin17SubSystemsManager8init_mpiEv+0x27)[0x7f85455bdcd7]
> [bell:14359] [ 8] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin3MPI13num_processesEv+0x9)[0x7f85455bc5d9]
> [bell:14359] [ 9] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin12MeshTopologyC1Ev+0xcf)[0x7f854573b41f]
> [bell:14359] [10] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin4MeshC2Ev+0xfa)[0x7f85456feeba]
> [bell:14359] [11] 
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/libdolfin.so.1.3(_ZN6dolfin13RectangleMeshC2EddddmmSs+0x40)[0x7f854537e650]
> [bell:14359] [12] ./demo_poisson(main+0x6c)[0x40921c]
> [bell:14359] [13] /lib64/libc.so.6(__libc_start_main+0xfd)[0x374fc1ed1d]
> [bell:14359] [14] ./demo_poisson[0x409059]
> [bell:14359] *** End of error message ***
> Segmentation fault (core dumped)
> 
> 
> While running the python demo, nothing happens...it simply hangs :
> 
> $ python demo_poisson.py
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/python2.6/site-packages/numpy/oldnumeric/__init__.py:11:
>  ModuleDeprecationWarning: The oldnumeric module will be dropped in Numpy 1.9
>  warnings.warn(_msg, ModuleDeprecationWarning)
> Calling FFC just-in-time (JIT) compiler, this may take some time.
> /mn/anatu/studenter-u1/imranal/Work/FEniCS/lib/python2.6/site-packages/numpy/linalg/linalg.py:520:
>  DeprecationWarning: Implicitly casting between incompatible kinds. In a 
> future numpy release, this will raise an error. Use casting="unsafe" if this 
> is intentional.
>  ainv = _umath_linalg.inv(a, signature=signature, extobj=extobj)
> Calling DOLFIN just-in-time (JIT) compiler, this may take some time.
> Calling DOLFIN just-in-time (JIT) compiler, this may take some time.
> 
> I tried some of the other cpp and python demos, and they all got stuck.
> 
> Imran
> 
>>> Imran
>>>>> Using the if test if (DOLFIN_SKIP_BUILD_TESTS) instead of if(1), would
>>>>> result in no error. However, then the dolfin build would not find PETSc :
>>>>> -- Found petscconf.h
>>>>> -- Performing test PETSC_TEST_RUNS - Failed
>>>>> -- Performing test PETSC_TEST_3RD_PARTY_LIBS_RUNS - Failed
>>>>> -- Performing Test PETSC_CUSP_FOUND
>>>>> -- Performing Test PETSC_CUSP_FOUND - Failed
>>>>> -- PETSc configured without Cusp support
>>>>> -- PETSc could not be found. Be sure to set PETSC_DIR and PETSC_ARCH.
>>>>> (missing:  PETSC_TEST_RUNS) (Required is at least version "3.2")
>>>>> And the build would continue without PETSc.
>>>>> Imran
> _______________________________________________
> fenics-support mailing list
> [email protected]
> http://fenicsproject.org/mailman/listinfo/fenics-support

_______________________________________________
fenics-support mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics-support

Reply via email to