Problem 1561 - Seg-fault due to missing re-initialization of physics in the workers, after change of geometry
Summary: Seg-fault due to missing re-initialization of physics in the workers, after c...
Status: RESOLVED FIXED
Alias: None
Product: Geant4
Classification: Unclassified
Component: run (show other problems)
Version: 10.0
Hardware: All All
: P5 normal
Assignee: asai
URL:
Depends on:
Blocks:
 
Reported: 2014-01-20 16:52 CET by Luciano Pandola
Modified: 2014-01-22 01:08 CET (History)
1 user (show)

See Also:

Attachments
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 Luciano Pandola 2014-01-20 16:52:59 CET 
* Only applies to the MT mode * 

After changes in the geometry, which trigger the re-initialization of the physics processes by the RunManager, the worker threads are not re-initialized. It happens in particular when the total number of materials is not changed, but different materials are selected for the geometry. The physics tables in the master thread are correctly re-initialized, but the re-initialization is not triggered for the workers.

It produces a segfault crash when using the Penelope EM builder. Also other EM builders are affected: no crashes, but physics results are not correct due to the incomplete re-initialization of the physics tables.
Comment 1 asai 2014-01-22 01:08:42 CET 
Fix is in run-V10-00-01 tag.
Makoto