Problem 1553 - Geometry Crash in Multithread v10.0
Summary: Geometry Crash in Multithread v10.0
Status: RESOLVED FIXED
Alias: None
Product: Examples/Basic and Novice
Classification: Unclassified
Component: B2 (show other problems)
Version: 10.0
Hardware: Apple Mac OS X
: P5 normal
Assignee: Gabriele Cosmo
URL:
Depends on:
Blocks:
 
Reported: 2013-12-29 20:14 CET by Matthew Lund
Modified: 2014-02-12 10:02 CET (History)
1 user (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this problem.
Description Matthew Lund 2013-12-29 20:14:54 CET
When compiling examples B2b, B5, and a multithread version of GDML application, Geant4-10.0 ceases every time on a beamOn command with the error message listed below and thread crash report underneath that. The same applications when compiled and ran in a single thread version of 10.0 with multithread OFF run just fine without a problem. 

G4WT0 > 
-------- EEEE ------- G4Exception-START -------- EEEE -------
*** G4Exception : GeomMgt0003
      issued by : G4VSolid::ComputeDimensions()
Illegal call to G4VSolid::ComputeDimensions()
Method not overloaded by derived class !
*** Fatal Exception *** core dump ***
-------- EEEE -------- G4Exception-END --------- EEEE -------

Thread 2 Crashed:
0 libsystem_kernel.dylib 0x00007fff90603866 __pthread_kill + 10
1 libsystem_pthread.dylib 0x00007fff8f11035c pthread_kill + 92
2 libsystem_c.dylib 0x00007fff8f2c3bba abort + 125
3 libG4global.dylib 0x000000010c671183 G4Exception(char const*, char const*, G4ExceptionSeverity, char const*) + 3027
4 libG4global.dylib 0x000000010c6713ab G4Exception(char const*, char const*, G4ExceptionSeverity, std::__1::basic_ostringstream<char, std::__1::char_traits<char>, std::__1::allocator<char> >&) + 91
5 libG4geometry.dylib 0x000000010c2c1e39 G4VSolid::ComputeDimensions(G4VPVParameterisation*, int, G4VPhysicalVolume const*) + 393
6 libG4geometry.dylib 0x000000010c2e9764 G4ParameterisedNavigation::LevelLocate(G4NavigationHistory&, G4VPhysicalVolume const*, int, CLHEP::Hep3Vector const&, CLHEP::Hep3Vector const*, bool, CLHEP::Hep3Vector&) + 708
7 libG4geometry.dylib 0x000000010c2dcfad G4Navigator::LocateGlobalPointAndSetup(CLHEP::Hep3Vector const&, CLHEP::Hep3Vector const*, bool, bool) + 4781
8 libG4event.dylib 0x000000010ad2d4e4 G4EventManager::DoProcessing(G4Event*) + 596
9 libG4run.dylib 0x000000010accb65e G4WorkerRunManager::ProcessOneEvent(int) + 46
10 libG4run.dylib 0x000000010accb5ee G4WorkerRunManager::DoEventLoop(int, char const*, int) + 238
11 libG4run.dylib 0x000000010acc1959 G4RunManager::BeamOn(int, char const*, int) + 89
12 libG4run.dylib 0x000000010acd2d4c G4MTRunManagerKernel::StartThread(void*) + 1084
13 libsystem_pthread.dylib 0x00007fff8f10f899 _pthread_body + 138
14 libsystem_pthread.dylib 0x00007fff8f10f72a _pthread_start + 137
15 libsystem_pthread.dylib 0x00007fff8f113fc9 thread_start + 13
Comment 1 Ivana Hrivnacova 2014-01-07 10:17:23 CET
Could you, please, give us more details about your platform and your Geant4 installation so that we can try to reproduce the problem?
What is your version of your Mac and which compiler do you use?
Also, please, give us the complete list of cmake options which you use for your Geant4 installation? Did you build the example with CMake as described at:

http://geant4.web.cern.ch/geant4/UserDocumentation/Doxygen/examples_doc/html/README_HowToRun.html

Thank you,
Comment 2 Andrea Dotti 2014-01-10 20:05:07 CET
Hello Matthew, 
triggered by your report, we have made a special run on Geant4 on Mac OS X Mavericks (10.9) and we did not see any failure for the two tests.

In addition to what Ivana asked let me ask also:
What are the conditions in which you have run the examples (how many threads, which macro files, etc). Also additional information on your system is welcome.

Andrea
Comment 3 Matthew Lund 2014-01-11 20:09:50 CET
I tested the built binary, everything worked fine. So I went back to my Geant4 build and rebuilt turning off options. After looking through the CMake, I realized I had accidentally turned on USE_USOLIDS. Once I rebuilt Geant4 without USOLIDS, every example works fine, thus the problem is in USOLIDS. Thanks!
Comment 4 Ivana Hrivnacova 2014-01-13 10:08:36 CET
Thank you for your investigations. I was now able to reproduce the exception with B2b example built against Geant4 installed with USOLIDS. We will let you know when a fix is available.

Best regards,
Comment 5 Gabriele Cosmo 2014-02-12 10:02:03 CET
The problem was reproduced and found affecting the cloning mechanism for the configuration of USolids with parameterised geometries in MT mode. The problem is now fixed in the development version and will be included in the next patch to release 10. Thanks for reporting.