[Paraview-developers] Fw: Valgrind + Paraview

[pat marion]

Under certain configurations (release builds), what you think is the
paraview binary is really just a launcher that sets up the required
LD_LIBRARY_PATH and spawns a new process.  Look for another paraview binary,
maybe bin/../lib/paraview.  To run it the real binary, you may have to
manually set your LD_LIBRARY_PATH.

There could also be a way to make the launcher to play nice with valgrind
that I am not aware of...

Pat

On Tue, Feb 15, 2011 at 6:16 PM, Sohail Shafii <sohailshafii at yahoo.com>wrote:

> I have to re-send this message since I subscribed to the list after sending
> the original message. So it should go through this time around.
>
> Sohail
>
> --- On *Tue, 2/15/11, Sohail Shafii <sohailshafii at yahoo.com>* wrote:
>
>
> From: Sohail Shafii <sohailshafii at yahoo.com>
> Subject: Valgrind + Paraview
> To: paraview-developers at paraview.org
> Date: Tuesday, February 15, 2011, 3:04 PM
>
> Hi,
>
> I'm currently debugging some of my own filters so I thought I would run
> valgrind with paraview.  It appears that valgrind and paraview are entirely
> disconnected, even though my command line arguments appear to be correct and
> paraview has been built as a debug version.  Here is the terminal output:
> shodan at shodan:~/kitware/ParaView-3.8.0-bin-debug/bin$ valgrind
> --tool=memcheck --track-origins=yes --leak-check=full ./paraview
> ==13652== Memcheck, a memory error detector
> ==13652== Copyright (C) 2002-2010, and GNU GPL'd, by Julian Seward et al.
> ==13652== Using Valgrind-3.6.0.SVN-Debian and LibVEX; rerun with -h for
> copyright info
> ==13652== Command: ./paraview
> ==13652==
> Load file /home/shodan/Code/Flow/sample_timesteps/turbine/wtbl.2040
> Load file /home/shodan/Code/Flow/sample_timesteps/field/comp.out.2040
> Load file /home/shodan/Code/Flow/sample_timesteps/turbine/wtbl.2040
> Finished computing Lambda-2 and vorticity fields, it took: 0 second(s).
> Number of negative Lambda-2s: 291072 out of 896000 tuples.
> Values min/max: -17.3383/7.8664.
> Lower threshold: -0.01
> Num seeds for vortex cores: 1998.
>
> ....usually valgrind reports more information when the program exits.
> Also, paraview appears to be running way to fast, which tells me that
> valgrind isn't doing anything (usually the slowdowns are very noticeable).
> Has anyone used valgrind with Paraview?
>
> Thanks, Sohail
>
>
>
> _______________________________________________
> Paraview-developers mailing list
> Paraview-developers at paraview.org
> http://public.kitware.com/mailman/listinfo/paraview-developers
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20110215/5056dfa7/attachment.htm>