Thank you for involving yourself 
All inputs are welcome!
Yep, I agree that the stack is the suspect but I’ve already bumped the stack size up and down, with exactly the same assert. And I tried to remove this particular mutex (I don’t really need them in the unit tests) to get hit by an assert in another mutex release in a different unit. The mutex pointer is owned by the C++ class instance, so I believe that’s also in the heap. So I could have a problem with heap being overwritten, but then I’d not expect an assert from FreeRTOS to fail, but fatal access crashes in Linux when accessing the mutex pointer.