After a lot of debugging, it turns out the issue is with the SetCells
function of the vtk unstructured grid. Below are the bits of my
Programmable Filter code that are relevant to the seg fault:
output = self.GetOutputDataObject(0)
cellArray = vtk.vtkCellArray()
cellTypes = []
# Create a polygon to display the selected points
polygon = vtk.vtkPolygon()
polygon.GetPoints().DeepCopy(pointArray)
polygon.GetPointIds().SetNumberOfIds(len(coordinates))
for i in range(len(coordinates)):
polygon.GetPointIds().SetId(i, coordinates[i][0])
# Add the polygon to a list of Cells
cellArray.InsertNextCell(polygon)
cellTypes.append(vtk.VTK_POLYGON)
# Get the cell containing a given cell add it to the cellArray
selectedCell = input.GetCell(cellID)
cellArray.InsertNextCell(selectedCell)
cellTypes.append(vtk.VTK_TETRA)
# SetCells of the output with only the selected analyzed cells
output.SetCells(cellTypes, cellArray)
The last line above seg faults on one of my computers but not on the
other one, both using Linux Mint 15 OS, and similar versions of
ParaView. Any thoughts?
On 15-01-28 07:27 PM, Omid Mahabadi wrote:
Hi,
I've been experiencing a very weird seg faults in various versions of
ParaView (4.1.x, 4.2.0, 4.3.x) when running an animation (streaming
over time) of my Programmable Filter. I'm not even sure this is a
ParaView issue, but I'll still share my story in case someone has any
insights that they can kindly share.
My Programmable Filter code works fine for the Current time step,
regardless at which time step I run it (i.e., hit the Apply button of
the Programmable Filter). However, when I hit the Play button to loop
over time steps (which appends some information to a CSV file),
ParaView seg faults after reaching the /3rd time step/. The filter has
been working for some time and this started to happen all of a sudden,
without any system update.
It may be related to a weird combination of hardware, software, and
drivers. My script works fine on a desktop computer I have but it seg
faults on my main workstation, which has an AMD/ATI Radeon R9 280x GPU
running on Linux Mint 15, 64-bit. When I completely removed the AMD
drivers, it still seg faulted but after reaching the /6th time step/
after the current one. Then, I launched an Ubuntu 14.04 LTS 64-bit,
live on a USB flash drive. The script in Programmable Filter worked
perfectly fine. As a result, I installed the OS, but it acted up
exactly as the Linux Mint installation.
The input files are massive vtk unstructured grids (about 800 MB
each). I created smaller files, but unfortunately I couldn't reproduce
the error. But I'm fairly sure that the files are not corrupted. As
such, I can't send out the files that cause the issue. I also
monitored the memory usage using free -h command while Playing. Here's
the output for the -/+ buffers/cache line:
total used free
-/+ buffers/cache: 7.2G 8.4G (Opening the .vtu files)
-/+ buffers/cache: 7.2G 8.4G
-/+ buffers/cache: 7.4G 8.2G
-/+ buffers/cache: 7.6G 8.0G
-/+ buffers/cache: 7.6G 8.0G
-/+ buffers/cache: 7.6G 8.0G
-/+ buffers/cache: 7.9G 7.7G
-/+ buffers/cache: 8.3G 7.4G
-/+ buffers/cache: 8.5G 7.1G
-/+ buffers/cache: 8.6G 7.0G
-/+ buffers/cache: 8.5G 7.1G
-/+ buffers/cache: 8.5G 7.1G
-/+ buffers/cache: 8.5G 7.1G
-/+ buffers/cache: 8.5G 7.1G
-/+ buffers/cache: 8.6G 7.0G
-/+ buffers/cache: 8.9G 6.7G
-/+ buffers/cache: 9.1G 6.5G
-/+ buffers/cache: 9.3G 6.3G
-/+ buffers/cache: 9.4G 6.2G (Seg faults here)
-/+ buffers/cache: 7.0G 8.6G
As you can see, my system still had a lot of free memory when the seg
fault happened and I haven't done anything explicitly to limit memory
usage by a single process.
Looking at syslog, this is the error message for the seg fault:
kernel: [77995.911073] paraview[22483]: segfault at 7f3c8e0398ac
ip 7f3a47ebe169 sp 7fff9106ab50 error 4 in
fglrx_dri.so[7f3a479e9000+21e9000]
Am I missing something here? Is there any other debugging steps I
should take? Any insights or help will be hugely appreciated.
Thank you,
Omid
___
Powered by www.kitware.com
Visit other Kitware open-source projects at
http://www.kitware.com/opensource/opensource.html
Please keep messages on-topic and check the ParaView Wiki at:
http://paraview.org/Wiki/ParaView
Search the list archives at: http://markmail.org/search/?q=ParaView
Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview
--
Omid Mahabadi,
