标签：启用 tac 方法 void volatile final 非公平锁 condition lock
启动N个线程, 这N个线程要不间断按顺序打印数字1-N. 将问题简化为3个线程无限循环打印1到3
三个线程无序竞争同步锁, 如果遇上的是自己的数字, 就打印. 这种方式会浪费大量的循环
public class TestSequential1 {
private volatile int pos = 1;
private volatile int count = 0;
public void one(int i) {
synchronized (this) {
if (pos == i) {
System.out.println("T-" + i + " " + count);
pos = i % 3 + 1;
count = 0;
} else {
count++;
}
}
}
public static void main(String[] args) {
TestSequential1 demo = new TestSequential1();
for (int i = 1; i <=3; i++) {
int j = i;
new Thread(()->{
while(true) {
demo.one(j);
}
}).start();
}
}
}
输出
T-1 0 T-2 5793 T-3 5285 T-1 2616 T-2 33 T-3 28 T-1 22 T-2 44 T-3 6 T-1 881 T-2 118358 T-3 247380 T-1 30803 T-2 29627 T-3 52044 ...
竞争同步锁时使用wait()和notifyAll(), 可以避免浪费循环
public class TestSequential4 {
private volatile int pos = 1;
private volatile int count = 0;
private final Object obj = new Object();
public void one(int i) {
System.out.println(i + " try");
synchronized (obj) {
System.out.println(i + " in");
try {
while (pos != i) {
count++;
System.out.println(i + " wait");
obj.wait();
}
System.out.println("T-" + i + " " + count);
pos = i % 3 + 1;
count = 0;
obj.notifyAll();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public static void main(String[] args) {
TestSequential4 demo = new TestSequential4();
for (int i = 3; i >=1; i--) {
int j = i;
new Thread(()->{
while(true) {
demo.one(j);
}
}).start();
}
}
}
输出
3 try 3 in 3 wait 2 try 2 in 2 wait 1 try 1 in T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 1 try 1 in 1 wait T-2 1 2 try 2 in 2 wait T-3 1 3 try 3 in 3 wait 2 wait T-1 2 ...
.
用Lock做, 非公平锁, 三个线程竞争, 如果遇上的是自己的数字, 就打印. 这种方式也会浪费大量的循环
public class TestSequential2 {
private final Lock lock = new ReentrantLock();
private volatile int pos = 1;
private volatile int count = 0;
public void one(int i) {
lock.lock();
if (pos == i) {
System.out.println("T-" + i + " " + count);
pos = i % 3 + 1;
count = 0;
} else {
count++;
}
lock.unlock();
}
public static void main(String[] args) {
TestSequential2 demo = new TestSequential2();
for (int i = 1; i <=3; i++) {
int j = i;
new Thread(()->{
while(true) {
demo.one(j);
}
}).start();
}
}
}
输出
T-1 0 T-2 0 T-3 323 T-1 54 T-2 68964 T-3 97642 T-1 6504 T-2 100603 T-3 6989 T-1 1313 T-2 0 T-3 183741 T-1 233 T-2 5081 T-3 164367 ..
.
和3一样, 但是使用公平锁, 这种情况下基本上可以做到顺序执行, 偶尔会产生多一次循环
private final Lock lock = new ReentrantLock(true);
输出
T-1 0 T-2 0 T-3 0 T-1 0 T-2 0 T-3 0 T-1 0 T-2 0 T-3 0 T-1 0 T-2 0 T-3 1 T-1 1 T-2 1 T-3 1 ...
.
这种情况可以在非公平锁的环境下, 实现和4接近的效果. 每个线程在拿到lock后如果看到不是自己的计数, 就通过await把lock交出去, 如果是自己的计数, 就完成打印动作, 再通知所有其他线程去抢lock, 自己在下一个循环后, 即使又拿到lock, 也会因为计数已经变了而await把lock交出去.
public class TestSequential5 {
private final Lock lock = new ReentrantLock();
private final Condition condition = lock.newCondition();
private volatile int pos = 1;
private volatile int count = 0;
public void one(int i) {
while (true) {
System.out.println(i + " try lock");
lock.lock();
System.out.println(i + " has lock");
try {
while (pos != i) {
System.out.println(i + " await");
count++;
condition.await();
}
System.out.println("T-" + i + " " + count);
pos = pos % 3 + 1;
condition.signalAll();
count = 0;
} catch (InterruptedException e) {
e.printStackTrace();
} finally {
System.out.println(i + " unlock");
lock.unlock();
}
}
}
public static void main(String[] args) {
TestSequential5 demo = new TestSequential5();
for (int i = 1; i <=3; i++) {
int j = i;
new Thread(()->{
while(true) {
demo.one(j);
}
}).start();
}
}
}
输出, 可以看到具体的线程序列, 每个线程是如何得到lock, 再进入await的
2 try lock 3 try lock 2 has lock 2 await 1 try lock 1 has lock T-1 1 1 unlock 1 try lock 1 has lock 1 await 3 has lock 3 await T-2 2 2 unlock 2 try lock 2 has lock 2 await 1 await T-3 2 3 unlock 3 try lock 2 await T-1 1 1 unlock 1 try lock 1 has lock 1 await 3 has lock 3 await T-2 2 2 unlock 2 try lock 2 has lock 2 await 1 await T-3 2 3 unlock 3 try lock 3 has lock 3 await 2 await T-1 2 1 unlock 1 try lock 1 has lock 1 await 3 await T-2 2 2 unlock 2 try lock 1 await T-3 1 3 unlock 3 try lock 2 has lock 2 await T-1 1 1 unlock 1 try lock 3 has lock 3 await T-2 1 2 unlock 2 try lock 2 has lock 2 await 1 has lock 1 await T-3 2 ...
.
给每个线程不同的condition. 这个和4的区别是, 会精确地通知对应的线程(在对应的condition上await的线程, 可能是多个)去抢lock.
public class TestSequential3 {
private final Lock lock = new ReentrantLock();
private final Condition[] conditions = {lock.newCondition(), lock.newCondition(), lock.newCondition()};
private volatile int pos = 1;
private volatile int count = 0;
public void one(int i) {
while (true) {
System.out.println(i + " try lock");
lock.lock();
System.out.println(i + " has lock");
try {
while (pos != i) {
System.out.println(i + " await");
count++;
conditions[i - 1].await();
}
System.out.println("T-" + i + " " + count);
pos = pos % 3 + 1;
conditions[pos - 1].signal();
count = 0;
} catch (InterruptedException e) {
e.printStackTrace();
} finally {
System.out.println(i + " unlock");
lock.unlock();
}
}
}
public static void main(String[] args) {
TestSequential3 demo = new TestSequential3();
for (int i = 1; i <=3; i++) {
int j = i;
new Thread(()->{
while(true) {
demo.one(j);
}
}).start();
}
}
}
输出
2 try lock 1 try lock 3 try lock 2 has lock 2 await 1 has lock T-1 1 1 unlock 1 try lock 3 has lock 3 await T-2 1 2 unlock 2 try lock 1 has lock 1 await T-3 1 3 unlock 3 try lock 3 has lock 3 await 2 has lock 2 await T-1 2 1 unlock 1 try lock T-2 0 2 unlock 2 try lock 1 has lock 1 await T-3 1 3 unlock 3 try lock 3 has lock 3 await 2 has lock 2 await T-1 2 1 unlock 1 try lock T-2 0 2 unlock 2 try lock 1 has lock 1 await T-3 1 ...
.
标签：启用 tac 方法 void volatile final 非公平锁 condition lock
原文地址：https://www.cnblogs.com/milton/p/11311148.html
