[Halld-offline] seg faults in mcsmear

[Richard Jones]

  Mark and all,

I am investigating a problem with segmentation faults in the current release that Jake discovered when trying to run simulations on my cluster.  To be sure it is not a local UConn problem, I reproduced it on the Jlab ifarm machines.  Here is how to reproduce it:

    * svn checkout --non-interactive https://halldsvn.jlab.org/repos/trunk/hdds
      cd hdds
      make
      cd ..
    * svn checkout --non-interactive https://halldsvn.jlab.org/repos/trunk/sim-recon/src
      cd src
      make
      cd ..
    * cd src/programs/Simulation/bggen
      make
      cd ../../../..

Now go to a work directory and run the new executables created in the above steps:

   1. bbgen
   2. hdgeant
   3. mcsmear hdgeant.hddm

For the input files for the above steps, see ~jonesrt/condortest/gluex directory.  About 2/3 of the times, mcsmear segmentation faults during step 3 above.  Wonder is, the other 1/3 of the time the command succeeds ok.  If you build the above with DEBUG=1 and run mcsmear_d instead of mcsmear, it never seg faults.  Weird.  Any ideas?

-Richard J.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20101010/b43f9b34/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4092 bytes
Desc: S/MIME Cryptographic Signature
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20101010/b43f9b34/attachment.p7s>