Non-critical error: mutex lock failure in MacOS

Yanzhao_Wang · May 13, 2024, 2:06pm

Hi,

Issue

I’m using Geant4 in MacOS and got the following error message:

1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument
1: Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
1: If the RunManagerKernel has been deleted, it failed to delete an allocated resource
1: and this destructor is being called after the statics were destroyed.
1: Exception: [code: system:22] caught: mutex lock failed: Invalid argument

However, when I run exactly the same code in a linux system, I don’t have such the error.

Version info

Geant4 Version: 11.2.1
C++ standard : C++23

MacOS: 14.4.1
Apple clang: 15.0.0
CMake Version in MacOS: 3.29.3

Linux OS: RHEL 8.9
gcc: 13.1
CMake Version in Linux: 3.28.1

Thanks for your attention.

bmorgan · May 13, 2024, 2:19pm

Is this on an Intel or Apple chip? Does it occur when building against C++17? Are you able to share a minimum reproducible code?

Yanzhao_Wang · May 13, 2024, 3:09pm

Hi, @bmorgan. Thanks very much for your reply.

Is this on an Intel or Apple chip

It’s on Apple M1 chip.

Does it occur when building against C++17?

I’m afraid our project cannot be build with C++17.

Are you able to share a minimum reproducible code?

I’m not sure whether I could pinpoint the part which causes the error. But you could find the whole project here (please use dev branch).

Examples from Geant4 run pretty well. So maybe it’s caused by the bad ownership management somewhere in the code.

bmorgan · May 13, 2024, 3:27pm

Thanks for the link - it’s possible this is related to the problem reported here: Error related to multi-threading after upgraded to 11.2 so maybe see if any of the discussion there applies to your application first.

Yanzhao_Wang · May 13, 2024, 5:17pm

@bmorgan

Hi,

I managed to reproduce the errors from one of Geant4 examples. The cause of the error is the physics list “Shielding” in the MT mode.

Reproduction steps:

Go to the example B2a
Change to MT mode (line 63):

  auto runManager =
    G4RunManagerFactory::CreateRunManager(G4RunManagerType::MT);

Use the physics “shielding” (line 70):

  auto physListFactory = G4PhysListFactory{};
  G4VUserPhysicsList* physicsList = physListFactory.GetReferencePhysList("Shielding");
  runManager->SetUserInitialization(physicsList);

Compile and run with:

  ./exampleB2a run2.mac

And you have the following error messages:

Non-critical error: mutex lock failure in ~G4Cache<N26G4ParticleHPContAngularPar10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N26G4ParticleHPContAngularPar10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N26G4ParticleHPContAngularPar10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N26G4ParticleHPContAngularPar10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N26G4ParticleHPContAngularPar10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<P26G4ParticleHPContAngularPar>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N26G4VParticleHPEnergyAngular10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N19G4ParticleHPProduct10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<N28G4ParticleHPEnAngCorrelation10toBeCachedE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<PNSt3__16vectorIiNS_9allocatorIiEEEE>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument
Non-critical error: mutex lock failure in ~G4Cache<P15G4HadFinalState>.
If the RunManagerKernel has been deleted, it failed to delete an allocated resource
and this destructor is being called after the statics were destroyed.
Exception: [code: system:22] caught: mutex lock failed: Invalid argument

I digged a little bit deeper into the source file that generates these message. It’s the file “G4Cache.hh” at line 293~302. The error message will occur if the derived classes from G4HadFinalState are deleted manually by the Geant4. However, G4Cache requires those objects to be deleted before the static mutex object. Thus, this leads to a undefined behaviour as the system may delete the objects earlier or later than the static mutex object.

So I guess some memory leaking should be fixed on the “Shielding” physics list.

bmorgan · May 13, 2024, 6:07pm

Thanks! We’ll see if we can reproduce/(suggest)fix locally with that.

makotoasai · May 13, 2024, 10:21pm

Thank you for reporting this issue. I confirm it. The ParticleHP code seems not to be properly updated when we moved to version 11.0. We will fix it in our next patch release. Meanwhile, if you have no particular reason to stick to the original MT mode, please use task-based parallel mode, i.e.

  auto runManager =
    G4RunManagerFactory::CreateRunManager(G4RunManagerType::Default);

Kind regards,
Makoto

gcosmo · May 14, 2024, 7:32am

Can you please open a Bugzilla ticket with all the details, so we can properly track this? Thanks.

Yanzhao_Wang · May 14, 2024, 1:47pm

Thanks for your info. @makotoasai

I didn’t know Geant4 provides task based concurrency from the first glance of the recent Geant4 manuals. Just being curious, if I choose the default option (a.k.a Tasking), which framework am I using for the task management? Executor or sender/receiver? Does that change when I use the TBB option?

system · June 13, 2024, 1:47pm

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.