[vtkusers] vtkTkRenderWidget:Linux:Python and VTK threaded:each render spawns zombie

[Donna]

Hi again Brent,

As I explained here, I don't want users to have to build my application:

http://public.kitware.com/pipermail/vtkusers/2003-March/016428.html

I know we'll need to upgrade our build host at some point, but right now 
building on this host produces binaries that are more interoperable than 
   those built with later Linux versions.

After posting the message below, I rebuilt VTK and python without 
threads, and this problem goes away (without any apparent performance 
hit).  My no-threads app runs fine on Linux RedHat 8.0 and 7.1 SMP.

I found the behavior (each render spawning a process) strange enough to 
report it.  Let's hope it is unique to my build environment.

Donna

Brent Goodrick wrote:
> Hi Donna,
> 
> You are building on RH 6.1 and executing on RH 8.0 which is bad news.
> GLIBC has changed, name mangling in the compiler has changed,
> etc. etc.  I would suggest compiling under the same system
> (compiler+libraries+OSrevision) in which you execute on.  
> 
> If you are bound and determined to do it, you may try the
> compatibility libraries that Red Hat has shipped in the past (RH 7.x
> anyhow).  I expect that they would discontinue those libraries at some
> point, so that would not be a good long term strategy.
> 
> Brent Goodrick
> PT Software
> 
> 
>>>>>>"donna" == donna  <donna at v1.wustl.edu> writes:
>>>>>
> 
> donna> Build environment:
> donna> 	Linux RedHat 6.1 (Cartman) 2.2.12-20 (uni-processor)
> donna> 	gcc 3.2.2
> donna> 	Python 2.1.3
> donna> 	VTK cvs co -D "10/5/2002"
> donna> 	both python and vtk configured with threads
> 
> donna> When I run the following simple python script on a dual processor host
> donna> running Linux RedHat 8.0 (Psyche) 2.4.18-26.8.0smp, each render spawns a
> donna> zombie python process:
> 
> donna> http://pulvinar.wustl.edu/donna/SureFit/Linux/volview.py
> 
> donna> (This script is a simplified version of
> donna> http://public.kitware.com/cgi-bin/cvsweb.cgi/VTK/Examples/Medical/Python/Medical3.py;
> donna> thanks to will and prabhu for making it available.)
> 
> donna> Keypress 'i' or 'd' in the image widget (increment or decrement the
> donna> slice displayed from VTK_DATA_ROOT + "/Data/headsq/quarter") triggers a
> donna> vtkTkRenderWidget render.  Running python in gdb, with several printf
> donna> messages inserted into VTK/Common/vtkDataObject.cxx and
> donna> VTK/Common/vtkImageData.cxx, produces this output:
> 
> donna> *** DISPLAYING slice 30
> donna> ...vtkImageData::UpdateData ...
> donna> ...vtkDataObject::UpdateData begin...
> donna> ...propagate the UpdateData to the source...
> ...Source-> UpdateData...
> donna> ...vtkImageData::UpdateData ...
> donna> ...vtkDataObject::UpdateData begin...
> donna> ...vtkDataObject::UpdateData end...
> donna> ...checking this->UpdateNumberOfPieces == 1 ...
> donna> ...confirmed this->UpdateNumberOfPieces == 1 ...
> donna> [New Thread 24580 (LWP 668)]
> ...this-> UpdatePiece...
> ...this-> UpdateNumberOfPieces...
> ...this-> UpdateGhostLevel...
> donna> ...vtkDataObject::UpdateData end...
> donna> ...checking this->UpdateNumberOfPieces == 1 ...
> donna> ...confirmed this->UpdateNumberOfPieces == 1 ...
> 
> donna> The modified versions of vtkDataObject.cxx and vtkImageData.cxx are here
> donna> -- just printf statements added to see where the thread is being
> donna> spawned:
> 
> donna> http://pulvinar.wustl.edu/donna/SureFit/Linux/vtkDataObject.cxx
> donna> http://pulvinar.wustl.edu/donna/SureFit/Linux/vtkImageData.cxx
> 
> donna> The thread is being spawned when vtkImageData::UpdateData returns to
> donna> vtkDataObject::UpdateData, because this->UpdateNumberOfPieces == 1. 
> donna> There seems to be some recursion here.  Can anyone help me understand
> donna> why python is spawning a new thread here?  These threads look like this
> donna> in ps -efl output:
> 
> donna> 044 Z donna     1590  1589  0  75   0    -     0 do_exi 10:59 pts/2   
> donna> 00:00:00 [python <defunct>]
> donna> 044 Z donna     1591  1589  0  80   0    -     0 do_exi 10:59 pts/2   
> donna> 00:00:00 [python <defunct>]
> donna> 044 Z donna     1592  1589  0  80   0    -     0 do_exi 10:59 pts/2   
> donna> 00:00:00 [python <defunct>]
> donna> 044 Z donna     1593  1589  0  76   0    -     0 do_exi 10:59 pts/2   
> donna> 00:00:00 [python <defunct>]
> 
> donna> They have caused crashes for one user, whose soft maxproc limit is
> donna> 1024.  My application has numerous render events, and in his case
> donna> several threads are being spawned per render.  When the number of
> donna> zombies approaches 1000, my application crashes.  This seems to happen
> donna> despite increasing the maxproc limit.  (Python seems to be crashing at
> donna> the machine's soft limit, regardless of the user's environment.)
> 
> donna> I can rebuild python and vtk without threads, and the user doesn't
> donna> crash.  But I'd like to understand why the renders are spawning these
> donna> threads.  Any ideas are appreciated.
> 
> donna> Thanks,
> 
> donna> Donna Hanlon
> donna> _______________________________________________
> donna> This is the private VTK discussion list. 
> donna> Please keep messages on-topic. Check the FAQ at: <http://public.kitware.com/cgi-bin/vtkfaq>
> donna> Follow this link to subscribe/unsubscribe:
> donna> http://public.kitware.com/mailman/listinfo/vtkusers
>