In the sub-test testExitedThreadRevocation, there is a check that the previousOwner field of an BiasedLockRevocationEvent is null if the previous owner was a thread that has terminated. Unfortunately, there is a slight possibility that a new thread is created with the exact same memory address, which would then implicitly cause it to be the new owner of the bias. This is valid behaviour from the point of view of the biased locking mechanism,

Have only seen this once, but the test should be updated to take this into account. Perhaps the check can be relaxed to only ensure that the previous owner isn't the terminated thread.