Thanks for your confirmation of the problem.
I found this very difficult to debug due to its random nature: it is unforeseeable and I don't know how to create a clean environment to test software/kernel versions when with one and the same kernel version and no change whatsoever on the software side the problem reappears fully randomly.