Mutex lock failure

Physics Lists
1

This was already mentioned (Non-critical error: mutex lock failure in MacOS) but I am opening a new topic since the linked topic was closed for new comments.

I see the exact same mutex lock failure error message as in Non-critical error: mutex lock failure in MacOS. I see it on Geant4 11.2.2 on intel macos.

In the other topic it was mentioned that this was a bug that would be fixed in the next patch release, however I could not find any related bugzilla issue. So I just wanted to confirm that a fix for this is underway for 11.2.3? And that the proposed solution (replace G4RunManager with G4MTRunManager with 1 thread) is actually still deemed the right way forward until 11.2.3 is out?

Cheers,
Thomas

2

Does anyone have any input or this, or can direct me to a relevant bugzilla bug report? We are held back on the 11.1.x branch due to this issue, and I am concerned that 11.2.3 will come out without a fix for this. In the linked forum thread @makotoasai confirmed the issue and mentioned that it would be fixed in 11.2.2, but apparently it slipped through the cracks.

Cheers,
Thomas

3

This issue is not yet resolved or understood.
To be clear, it is a warning message issued by G4Cache to indicate a mismatch in the deletion order of allocated objects through static containers, happening at job closure. According to our tests, it happens only in particular cases in recent releases, i.e. when using HP/shielding physics-lists with the old MT scheme.
Have you tried in latest releases to migrate your application to use the default tasking system for parallelism, as originally suggested by @makotoasai ?