ParaView 5.6.0 crash when loading raw binary file

Hi all,

I am using ParaView-5.6.0-MPI-Linux-64bit that I downloaded from paraview.org. I have a binary file of 16 x 16 x 16 floats in little endian that I try to load in ParaView using “Image Reader”. I then use:

  • Data Scalar Type: float
  • Data Byte Order: LittleEndian
  • Data Origin 0 0 0
  • Data Spacing 1 1 1
  • Number Of Scalar Components: 1
  • File Lower Left: true
  • Scalar Array Name: density
  • File DImensionality: 3
  • Data Extent:
    • 0 0
    • 0 0
    • 0 0

I then get the following crash message in the terminal:

Generic Warning: In /home/buildslave/dashboards/buildbot/paraview-pvbinsdash-linux-shared-release_superbuild/build/
superbuild/paraview/src/VTK/Parallel/MPI/vtkMPICommunicator.cxx, line 71
MPI had an error
------------------------------------------------
Invalid argument, error stack:
MPI_Type_create_subarray(344): MPI_Type_create_subarray(ndims=3, array_of_sizes=0x7fff1d1aeca0, array_of_subsizes=0
x7fff1d1aecb0, array_of_starts=0x7fff1d1aecc0, order=57, MPI_BYTE, newtype=0x7fff1d1aec98) failed
MPI_Type_create_subarray(127): Argument array_of_starts has value 4 but must be within [0,0]
------------------------------------------------

application called MPI_Abort(MPI_COMM_WORLD, 873079820) - process 0
[unset]: write_line error; fd=-1 buf=:cmd=abort exitcode=873079820
:
system msg for write_line failure : Bad file descriptor

I have included the raw binary file:
simpleData.raw (16 KB)

For comparison, I also wrote an XDMF file that describes the binary data. This works fine.
simpleData.xmf (778 Bytes)

Does this crash for other people as well? If so, I’ll create a bug report.
Or am I simply doing something wrong?

Thanks!

Hi Jarno,
It appears to me that you are not even getting as far as your data. Looks like you played around in the Settings file? Did you enable Auto MPI? If so, I would recommend turning it off. Generally, first thing to do if things break is go back to default. If you DIDN’T turn on MPI, then there is a different - and stranger - issue.

Alan

It is working fine here. Please reset your settings.

Hi, I reset my settings by deleting ~/.config/ParaView and did the same things as before.

jarno@workjarno:~/.config$ mv ParaView/ ParaView.bak
jarno@workjarno:~/.config$ paraview 
Generic Warning: In /home/buildslave/dashboards/buildbot/paraview-pvbinsdash-linux-shared-release_superbuild/build/superbuild/paraview/src/VTK/Parallel/MPI/vtkMPICommunicator.cxx, line 71
MPI had an error
------------------------------------------------
Invalid argument, error stack:
MPI_Type_create_subarray(344): MPI_Type_create_subarray(ndims=3, array_of_sizes=0x7ffe67da0160, array_of_subsizes=0x7ffe67da0170, array_of_starts=0x7ffe67da0180, order=57, MPI_BYTE, newtype=0x7ffe67da0158) failed
MPI_Type_create_subarray(127): Argument array_of_starts has value 4 but must be within [0,0]
------------------------------------------------

application called MPI_Abort(MPI_COMM_WORLD, 269100044) - process 0
[unset]: write_line error; fd=-1 buf=:cmd=abort exitcode=269100044
:
system msg for write_line failure : Bad file descriptor

I noticed I wrote some of the Image Reader setting wrong in my post. Data extent should be from 1 to 16, not 0 to 0 of course, but it looks like you spotted that already.

Here’s my settings file, though it doesn’t contain much. ParaView5.6.0.ini (1004 Bytes)
I still double-checked the Settings screen and “Enable Auto MPI” is off.

This is on a stock install of Kubuntu 18.04 by the way.

Sorry, I mistested this.

With your setup of the reading of the .raw file, I have the following error. It is different to your error though :

ERROR: In /home/buildslave/dashboards/buildbot/paraview-pvbinsdash-linux-shared-release_superbuild/build/superbuild/paraview/src/ParaViewCore/ClientServerCore/Rendering/vtkPVImageSliceMapper.cxx, line 241
vtkPVImageSliceMapper (0xa319900): Incorrect dimensionality.

If the raw file has dimension of 16 x 16 x 16, the data extent is set as below, isn’t it?

Data Extent:
0 15
0 15
0 15

Ugh, so it is… That worked. Thanks!
Been programming too much with arrays-start-at-one languages. :slight_smile:

Still somewhat strange error message I got and that it’s different from @mwestphal, but oh well.

Thanks again!