A good starting point is probably auditing usage of _syscallhandler_wasBlocked - some handlers assume that if they're running again after being blocked, then it must be the case that either the original blocking condition is somehow satisfied or a timeout expired. This is no longer true.