Problem 1590 - break from the application with optical processes in multi-threading mode
Summary: break from the application with optical processes in multi-threading mode
Status: RESOLVED FIXED
Alias: None
Product: Geant4
Classification: Unclassified
Component: track (show other problems)
Version: 10.0
Hardware: PC Linux
: P5 major
Assignee: asai
URL:
Depends on:
Blocks:
 
Reported: 2014-03-12 10:01 CET by Ivana.Hrivnacova
Modified: 2014-04-28 20:53 CEST (History)
1 user (show)

See Also:

Attachments
tar file with modified macro and gdb outputs (1.98 KB, application/gzip)
2014-03-12 10:01 CET, Ivana.Hrivnacova 		Details
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this problem.
Description Ivana.Hrivnacova 2014-03-12 10:01:56 CET 
Created attachment 257 [details]
tar file with modified macro and gdb outputs

We are getting break in Geant4 VMC test which defines the same setup as the extended OpNovice example when running in multi-threading mode.
The break can be reproduced with OpNovice example with modified test macro as attached and changing the verbosity level fVerboseLevel=0 in OpNovicePhysicsList.cc.

I attach the GDB back trace from OpNovice example (bt.txt) and also from Geant4 VMC example (bt_vmc.txt). They are a bit different, but both point at the updating the GROUPVEL material properties table in tracking.

The test was performed with Geant4 10.00.p01, run with 8 threads.
Comment 1 Vladimir.Ivantchenko 2014-03-15 19:53:24 CET 
Hello,

if bug happens in optical physics then it should be assigned to Peter Gumplinger to category processes/optical

VI
Comment 2 gum 2014-03-18 01:02:31 CET 
I suspect this has to do with:

G4Track::CalculateVelocityForOpticalPhoton()

which uses:

     groupvel = mat->GetMaterialPropertiesTable()->GetProperty("GROUPVEL");
     groupvel->Value(current_momentum);
Comment 3 asai 2014-04-28 20:53:59 CEST 
A race condition issue was identified in G4MaterialPropertyTable class and fixed.