Java并發(fā)編程源碼分析系列:
- 分析Java線程池的創(chuàng)建
- 分析Java線程池執(zhí)行原理
- 分析Java線程池Callable任務(wù)執(zhí)行原理
- 分析ReentrantLock的實(shí)現(xiàn)原理
上一篇通過(guò)研究ReentrantLock分析了AQS的獨(dú)占功能,本文將通過(guò)同樣是AQS子類的CountDownLatch分析AQS的共享功能。有了前文研究獨(dú)占功能的基礎(chǔ),再研究共享鎖就簡(jiǎn)單多了。
CountDownLatch的使用
CountDownLatch是同步工具類之一,可以指定一個(gè)計(jì)數(shù)值,在并發(fā)環(huán)境下由線程進(jìn)行減1操作,當(dāng)計(jì)數(shù)值變?yōu)?之后,被await方法阻塞的線程將會(huì)喚醒,實(shí)現(xiàn)線程間的同步。
public void startTestCountDownLatch() {
int threadNum = 10;
final CountDownLatch countDownLatch = new CountDownLatch(threadNum);
for (int i = 0; i < threadNum; i++) {
final int finalI = i + 1;
new Thread(() -> {
System.out.println("thread " + finalI + " start");
Random random = new Random();
try {
Thread.sleep(random.nextInt(10000) + 1000);
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.println("thread " + finalI + " finish");
countDownLatch.countDown();
}).start();
}
try {
countDownLatch.await();
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.println(threadNum + " thread finish");
}
主線程啟動(dòng)10個(gè)子線程后阻塞在await方法,需要等子線程都執(zhí)行完畢,主線程才能喚醒繼續(xù)執(zhí)行。
構(gòu)造器
CountDownLatch和ReentrantLock一樣,內(nèi)部使用Sync繼承AQS。構(gòu)造函數(shù)很簡(jiǎn)單地傳遞計(jì)數(shù)值給Sync,并且設(shè)置了state。
Sync(int count) {
setState(count);
}
上文已經(jīng)介紹過(guò)AQS的state,這是一個(gè)由子類決定含義的“狀態(tài)”。對(duì)于ReentrantLock來(lái)說(shuō),state是線程獲取鎖的次數(shù);對(duì)于CountDownLatch來(lái)說(shuō),則表示計(jì)數(shù)值的大小。
阻塞線程
接著來(lái)看await方法,直接調(diào)用了AQS的acquireSharedInterruptibly。
public void await() throws InterruptedException {
sync.acquireSharedInterruptibly(1);
}
public final void acquireSharedInterruptibly(int arg)
throws InterruptedException {
if (Thread.interrupted())
throw new InterruptedException();
if (tryAcquireShared(arg) < 0)
doAcquireSharedInterruptibly(arg);
}
首先嘗試獲取共享鎖,實(shí)現(xiàn)方式和獨(dú)占鎖類似,由CountDownLatch實(shí)現(xiàn)判斷邏輯。
protected int tryAcquireShared(int acquires) {
return (getState() == 0) ? 1 : -1;
}
返回1代表獲取成功,返回-1代表獲取失敗。如果獲取失敗,需要調(diào)用doAcquireSharedInterruptibly:
private void doAcquireSharedInterruptibly(int arg)
throws InterruptedException {
final Node node = addWaiter(Node.SHARED);
boolean failed = true;
try {
for (;;) {
final Node p = node.predecessor();
if (p == head) {
int r = tryAcquireShared(arg);
if (r >= 0) {
setHeadAndPropagate(node, r);
p.next = null; // help GC
failed = false;
return;
}
}
if (shouldParkAfterFailedAcquire(p, node) &&
parkAndCheckInterrupt())
throw new InterruptedException();
}
} finally {
if (failed)
cancelAcquire(node);
}
}
doAcquireSharedInterruptibly的邏輯和獨(dú)占功能的acquireQueued基本相同,阻塞線程的過(guò)程是一樣的。不同之處:
- 創(chuàng)建的Node是定義成共享的(Node.SHARED);
- 被喚醒后重新嘗試獲取鎖,不只設(shè)置自己為head,還需要通知其他等待的線程。(重點(diǎn)看后文釋放操作里的setHeadAndPropagate)
釋放操作
public void countDown() {
sync.releaseShared(1);
}
countDown操作實(shí)際就是釋放鎖的操作,每調(diào)用一次,計(jì)數(shù)值減少1:
public final boolean releaseShared(int arg) {
if (tryReleaseShared(arg)) {
doReleaseShared();
return true;
}
return false;
}
同樣是首先嘗試釋放鎖,具體實(shí)現(xiàn)在CountDownLatch中:
protected boolean tryReleaseShared(int releases) {
// Decrement count; signal when transition to zero
for (;;) {
int c = getState();
if (c == 0)
return false;
int nextc = c-1;
if (compareAndSetState(c, nextc))
return nextc == 0;
}
}
死循環(huán)加上cas的方式保證state的減1操作,當(dāng)計(jì)數(shù)值等于0,代表所有子線程都執(zhí)行完畢,被await阻塞的線程可以喚醒了,下一步調(diào)用doReleaseShared:
private void doReleaseShared() {
for (;;) {
Node h = head;
if (h != null && h != tail) {
int ws = h.waitStatus;
if (ws == Node.SIGNAL) {
//1
if (!compareAndSetWaitStatus(h, Node.SIGNAL, 0))
continue; // loop to recheck cases
unparkSuccessor(h);
}
//2
else if (ws == 0 &&
!compareAndSetWaitStatus(h, 0, Node.PROPAGATE))
continue; // loop on failed CAS
}
if (h == head) // loop if head changed
break;
}
}
標(biāo)記1里,頭節(jié)點(diǎn)狀態(tài)如果SIGNAL,則狀態(tài)重置為0,并調(diào)用unparkSuccessor喚醒下個(gè)節(jié)點(diǎn)。
標(biāo)記2里,被喚醒的節(jié)點(diǎn)狀態(tài)會(huì)重置成0,在下一次循環(huán)中被設(shè)置成PROPAGATE狀態(tài),代表狀態(tài)要向后傳播。
private void unparkSuccessor(Node node) {
int ws = node.waitStatus;
if (ws < 0)
compareAndSetWaitStatus(node, ws, 0);
Node s = node.next;
if (s == null || s.waitStatus > 0) {
s = null;
for (Node t = tail; t != null && t != node; t = t.prev)
if (t.waitStatus <= 0)
s = t;
}
if (s != null)
LockSupport.unpark(s.thread);
}
在喚醒線程的操作里,分成三步:
- 處理當(dāng)前節(jié)點(diǎn):非CANCELLED狀態(tài)重置為0;
- 尋找下個(gè)節(jié)點(diǎn):如果是CANCELLED狀態(tài),說(shuō)明節(jié)點(diǎn)中途溜了,從隊(duì)列尾開(kāi)始尋找排在最前還在等著的節(jié)點(diǎn)
- 喚醒:利用LockSupport.unpark喚醒下個(gè)節(jié)點(diǎn)里的線程。
線程是在doAcquireSharedInterruptibly里被阻塞的,喚醒后調(diào)用到setHeadAndPropagate。
private void setHeadAndPropagate(Node node, int propagate) {
Node h = head;
setHead(node);
if (propagate > 0 || h == null || h.waitStatus < 0 ||
(h = head) == null || h.waitStatus < 0) {
Node s = node.next;
if (s == null || s.isShared())
doReleaseShared();
}
}
setHead設(shè)置頭節(jié)點(diǎn)后,再判斷一堆條件,取出下一個(gè)節(jié)點(diǎn),如果也是共享類型,進(jìn)行doReleaseShared釋放操作。下個(gè)節(jié)點(diǎn)被喚醒后,重復(fù)上面的步驟,達(dá)到共享狀態(tài)向后傳播。
要注意,await操作看著好像是獨(dú)占操作,但它可以在多個(gè)線程中調(diào)用。當(dāng)計(jì)數(shù)值等于0的時(shí)候,調(diào)用await的線程都需要知道,所以使用共享鎖。
限定時(shí)間的await
CountDownLatch的await方法還有個(gè)限定阻塞時(shí)間的版本.
public boolean await(long timeout, TimeUnit unit)
throws InterruptedException {
return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
}
跟蹤代碼,最后來(lái)看doAcquireSharedNanos方法,和上文介紹的doAcquireShared邏輯基本一樣,不同之處是加了time字眼的處理。
private boolean doAcquireSharedNanos(int arg, long nanosTimeout)
throws InterruptedException {
if (nanosTimeout <= 0L)
return false;
final long deadline = System.nanoTime() + nanosTimeout;
final Node node = addWaiter(Node.SHARED);
boolean failed = true;
try {
for (;;) {
final Node p = node.predecessor();
if (p == head) {
int r = tryAcquireShared(arg);
if (r >= 0) {
setHeadAndPropagate(node, r);
p.next = null; // help GC
failed = false;
return true;
}
}
nanosTimeout = deadline - System.nanoTime();
if (nanosTimeout <= 0L)
return false;
if (shouldParkAfterFailedAcquire(p, node) &&
nanosTimeout > spinForTimeoutThreshold)
LockSupport.parkNanos(this, nanosTimeout);
if (Thread.interrupted())
throw new InterruptedException();
}
} finally {
if (failed)
cancelAcquire(node);
}
}
進(jìn)入方法時(shí),算出能夠執(zhí)行多久的deadline,然后在循環(huán)中判斷時(shí)間。注意到代碼中間有句:
nanosTimeout > spinForTimeoutThreshold
static final long spinForTimeoutThreshold = 1000L;
spinForTimeoutThreshold寫(xiě)死了1000ns,這就是所謂的自旋操作。當(dāng)超時(shí)在1000ns內(nèi),讓線程在循環(huán)中自旋,否則阻塞線程。
總結(jié)
兩篇文章分別以ReentrantLock和CountDownLatch為例研究了AQS的獨(dú)占功能和共享功能。AQS里的主要方法研究得七七八八了,趁熱打鐵下一篇將會(huì)研究其他同步工具類的實(shí)現(xiàn)。