Understanding Java's AbstractQueuedSynchronizer (AQS) and Concurrency Utilities

Java's concurrency utilities (JUC) were created by Doug Lea to improve performance beyond the heavyweight synchronized keyword, and the core of many utilities is the AbstractQueuedSynchronizer (AQS).

AQS implements a template‑method pattern where subclasses provide concrete implementations of methods such as tryAcquire , tryRelease , tryAcquireShared , and tryReleaseShared . The framework maintains a FIFO CLH queue of Node objects that represent waiting threads.

Key fields of Node include prev , next , thread , waitStatus (CANCELLED, SIGNAL, CONDITION, PROPAGATE, etc.), and nextWaiter for condition queues. The state variable is a volatile long that stores lock state; for ReentrantReadWriteLock the high 16 bits hold the read count and the low 16 bits hold the write count.

The main acquisition path is acquire(int arg) for exclusive locks and acquireShared(int arg) for shared locks. If tryAcquire fails, the thread is added to the queue via addWaiter , possibly spinning with enq , and then blocks with LockSupport.park after shouldParkAfterFailedAcquire returns true. Release operations invoke tryRelease and wake the successor with unparkSuccessor .

Condition objects maintain a separate wait queue (firstWaiter/lastWaiter). Methods await and signal / signalAll transfer nodes between the condition queue and the main AQS queue, using transferForSignal and transferAfterCancelledWait . This provides a more flexible and efficient alternative to Object's wait/notify .

Examples in the article illustrate custom lock implementations, usage of CountDownLatch , Semaphore , and ReentrantReadWriteLock , together with the underlying AQS code snippets.