Reentrantlock 在 Java 中工作正常，但在 Scala 中导致 IllegalMonitorException

Question

我想迁移 Java 函数

protected static final Lock LOCK = new ReentrantLock();
public double calculate(...){
    try {
        LOCK.tryLock(20, TimeUnit.SECONDS);
        ...
    }finally{
        LOCK.unlock()
    }
}

Scala 中的相同函数：

protected final def LOCK = new ReentrantLock
def calculate(...): double = {
    try{
        LOCK.tryLock(20, TimeUnit.Seconds)
        ...
    }finally{
        LOCK.unlock()
    }
}

LOCK.unlock() 总是导致 IllegalMonitorStateException。我不明白为什么会发生这种情况。

谁能告诉我问题出在哪里？

Answer 1

你绝对应该让 LOCK 变成 val 而不是 def。就目前而言，您每次都在重新创建 ReetrantLock 的新实例。实际上你正在做的是：

try {
    // Useless as we are creating a new lock
    (new ReentrantLock).tryLock(20, TimeUnit.SECONDS).tryLock(20, TimeUnit.SECONDS).tryLock(20, TimeUnit.SECONDS);
    ...
}finally{
    // Useless too, and will actually throw because we unlock a fresh (thus unlocked) lock
    (new ReentrantLock).unlock()
}

显然是注定要失败的

你应该这样做：

object MyClass {
  private val LOCK = new ReentrantLock
}
class MyClass {
  def calculate(...): double = {
      try{
          LOCK.tryLock(20, TimeUnit.Seconds)
          ...
      }finally{
          LOCK.unlock()
      }
  }
}

你的原始 java 代码直接翻译成 scala。

最后，Jon Skeet 在他（现已删除）的回答中正确地建议：

You should only unlock the lock if you managed to acquire it - and the conventional pattern is to put the lock/tryLock call before the try. (It doesn't matter with tryLock(), but it does matter for lock(), so we might as well be consistent.)

给出：

object MyClass {
  private val LOCK = new ReentrantLock
}
class MyClass {
  def calculate(...): double = {
      val gotLock = LOCK.tryLock(20, TimeUnit.Seconds)
      try {
          ...
      } finally {
        if (gotLock) {
          LOCK.unlock()
        }
      }
  }
}

Reentrantlock 在 Java 中工作正常，但在 Scala 中导致 IllegalMonitorException

Reentrantlock works fine in Java but causes IllegalMonitorException in Scala

java

concurrency

multithreading

scala

reentrantlock