Dear all, I noticed that in kaffe lock implementation (in slowLockMutex() and slowUnlockMutex() ), the last thread suspending on the lock will always be resumed firstly, which doesn't embody equality policy. Right? Fang Wei Jian