Segfault when computing gradient of a vector on polyhedral cells

ParaView Support
1

I am working on a grid that contains also polyhedral cells and my Paraview (5.10) crashes whenever I compute gradient of the velocity vector. I reduced my mesh to just 4 poly cells and can always reproduce the error. The cellst themselves look okay in terms of VTK format and I can compute gradients of a scalar and also gradient of a vector provided that I use up to 3 cells. With 4 connected polys I get the above mentioned segfault. I am wondering whether it is my mesh format problem or whether Paraview has a bug.

> Loguru caught a signal: SIGSEGV
> Stack trace:
> 13      0x7fa56673688d clone + 109
> 12      0x7fa55b130e65 /usr/lib64/libpthread.so.0(+0x7e65) [0x7fa55b130e65]
> 11      0x7fa55aee8196 /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libtbb.so.2(+0x1d196) [0x7fa55aee8196]
> 10      0x7fa55aee8206 /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libtbb.so.2(+0x1d206) [0x7fa55aee8206]
> 9       0x7fa55aeec56f /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libtbb.so.2(+0x2156f) [0x7fa55aeec56f]
> 8       0x7fa55aeee760 /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libtbb.so.2(+0x23760) [0x7fa55aeee760]
> 7       0x7fa55aef45ea /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libtbb.so.2(+0x295ea) [0x7fa55aef45ea]
> 6       0x7fa55f092fba /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libvtkFiltersGeneral-pv5.10.so.1(+0x24bfba) [0x7fa55f092fba]
> 5       0x7fa55f09286a /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libvtkFiltersGeneral-pv5.10.so.1(+0x24b86a) [0x7fa55f09286a]
> 4       0x7fa55dabe921 vtkPolyhedron::Derivatives(int, double const*, double const*, int, double*) + 641
> 3       0x7fa55da29c0c vtkMeanValueCoordinatesInterpolator::ComputeInterpolationWeights(double const*, vtkPoints*, vtkCellArray*, double*) + 236
> 2       0x7fa55da2783d vtkMeanValueCoordinatesInterpolator::ComputeInterpolationWeightsForPolygonMesh(double const*, vtkPoints*, vtkMVCPolyIterator&, double*) + 109
> 1       0x7fa55da20ffd /home/caldron-b/g478040/linux/Software/Paraview/v5.10.1/ParaView-5.10.1-MPI-Linux-Python3.9-x86_64/bin/../lib/libvtkCommonDataModel-pv5.10.so.1(+0x286ffd) [0x7fa55da20ffd]
> 0       0x7fa56666e3b0 /usr/lib64/libc.so.6(+0x363b0) [0x7fa56666e3b0]
> (  64.034s) [        813F4700]                       :0     FATL| Signal: SIGSEGV

poly.tar.zst (6.9 KB)

2

I reproduce, looks like a memory issue, you may want to open an issue.

3

I shamefully admit that I do not know how to open an issue. Either someone else opens an issue or shows me how to do it.

4

Issues are maintained using ParaView’s GitLab repo here: https://gitlab.kitware.com/paraview/paraview/-/issues. On that page in the upper right is a New issue button that you can use to submit a new issue.

1 Like
5

done
https://gitlab.kitware.com/paraview/paraview/-/issues/21549