Closed davidscn closed 4 years ago
Which preCICE version do you use? We recently changed quite a bit there in the destructor. The first logging setting here could help to find out whether preCICE or the adaper is the problem.
I'm working with the latest debian packages. Using a verbose logging doesn't give any additional information. Since the systemtest prints a stacktrace, which points (at first readable) to the data field in the adapter (Stress) my best guess would be, that the issue is in the adapter.
I inherited the Force calculation in my Stress data field, since they are related. This might be the problem here in case it is not a more general problem in the adapter.
The stack trace indeed indicates a problem in the adapter:
openfoam-adapter | ---[preciceAdapter] [DEBUG] Finalizing the preCICE solver interface...
openfoam-adapter | ---[preciceAdapter] [DEBUG] Destroying the preCICE solver interface...
openfoam-adapter | ---[preciceAdapter] [DEBUG] Deleting the interfaces...
openfoam-adapter | #0 Foam::error::printStack(Foam::Ostream&) at ??:?
openfoam-adapter | #1 Foam::sigSegv::sigHandler(int) at ??:?
openfoam-adapter | #2 ? in "/lib/x86_64-linux-gnu/libc.so.6"
openfoam-adapter | #3 preciceAdapter::FSI::Force::~Force() at ??:?
openfoam-adapter | #4 preciceAdapter::FSI::Stress::~Stress() at ??:?
openfoam-adapter | #5 preciceAdapter::Interface::~Interface() at ??:?
openfoam-adapter | #6 preciceAdapter::Adapter::teardown() at ??:?
openfoam-adapter | #7 preciceAdapter::Adapter::finalize() at ??:?
openfoam-adapter | #8 preciceAdapter::Adapter::execute() at ??:?
openfoam-adapter | #9 Foam::functionObjects::preciceAdapterFunctionObject::execute() at ??:?
openfoam-adapter | #10 Foam::functionObjectList::execute() at ??:?
openfoam-adapter | #11 Foam::Time::run() const at Time.C:?
openfoam-adapter | #12 ? at ??:?
openfoam-adapter | #13 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
openfoam-adapter | #14 ? at ??:?
openfoam-adapter | /bin/bash: line 1: 162 Segmentation fault (core dumped) pimpleDyMFoam -case /home/precice/Data/Input
Could you please:
-fsanitize=address
and run again (this may open Pandora's box, but this may be a good reason to).
Currently working with the
Stress
data field and OpenFOAM 7, I face the following error, when the simulation has finished:This error was reported by @Eder-K in the recent system tests and I can reproduce it locally.
I still need to investigate, if other data fields are affected as well, I guess there is something wrong in the destructors.