[petsc-dev] DACreate3d should work for 1 node in Z

Jed Brown jed at 59A2.org
Tue Apr 19 09:13:22 CDT 2011


This would make it easier to solve 2D problems with a 3D code. It currently
crashes or produces NaN because the scatter is set up incorrectly:

==10402== Invalid read of size 2
==10402==    at 0x4C26C2C: memcpy (in
/usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==10402==    by 0x6D87D01: opal_convertor_pack (in
/usr/lib/openmpi/libmpi.so.1.0.1)
==10402==    by 0xF33D0C8: ??? (in /usr/lib/openmpi/openmpi/mca_btl_sm.so)
==10402==    by 0xEB1F1B7: ??? (in /usr/lib/openmpi/openmpi/mca_pml_ob1.so)
==10402==    by 0xEB0DC0C: ??? (in /usr/lib/openmpi/openmpi/mca_pml_ob1.so)
==10402==    by 0x6CD6650: PMPI_Isend (in /usr/lib/openmpi/libmpi.so.1.0.1)
==10402==    by 0x551F85A: PetscGatherMessageLengths (mpimesg.c:128)
==10402==    by 0x580D5AC: VecScatterCreate_PtoS (vpscat.c:1621)
==10402==    by 0x5823C9F: VecScatterCreate (vscat.c:1316)
==10402==    by 0x518A547: DACreate_3D (da3.c:957)
==10402==    by 0x51D1ACA: DASetType (dareg.c:48)
==10402==    by 0x5194FA7: DASetTypeFromOptions_Private (dacreate.c:65)
==10402==  Address 0xbc33562 is 498 bytes inside a block of size 516 alloc'd
==10402==    at 0x4C23918: memalign (in
/usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==10402==    by 0x53E3D3B: PetscMallocAlign (mal.c:30)
==10402==    by 0x5566432: PetscTrMallocDefault (mtr.c:192)
==10402==    by 0x580D2B4: VecScatterCreate_PtoS (vpscat.c:1601)
==10402==    by 0x5823C9F: VecScatterCreate (vscat.c:1316)
==10402==    by 0x518A547: DACreate_3D (da3.c:957)
==10402==    by 0x51D1ACA: DASetType (dareg.c:48)
==10402==    by 0x5194FA7: DASetTypeFromOptions_Private (dacreate.c:65)
==10402==    by 0x5195A81: DASetFromOptions (dacreate.c:131)
==10402==    by 0x518FFF5: DACreate3d (da3.c:1742)
==10402==    by 0x40669A: multi_init(int, char**) (multi_petsc.cpp:115)
==10402==    by 0x403AF5: main (mhd_main.cpp:351)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110419/c7651400/attachment.html>


More information about the petsc-dev mailing list