Previous | Next | Trail Map | Writing Java Programs | Threads of Control


Synchronization

So far, this lesson has contained examples with independent, asynchronous threads. That is, each thread contained all of the data and methods required for its execution and didn't require any outside resources or methods. In addition, the threads in those examples ran at their own pace without concern over the state or activities of any other concurrently running threads.

However, there are many interesting situations where separate concurrently running threads do share data and must consider the state and activities of those other threads. One such set of programming situations are known as Producer/Consumer scenarios where the Producer generates a stream of data which then is consumed by a Consumer.

For example, you can imagine a Java application where one thread (the producer) writes data to a file while a second thread (the consumer) reads data from the same file. Or, as you type characters on the keyboard, the producer thread places key events in an event queue and the consumer thread reads the events from the same queue. Both of these examples use concurrent threads that share a common resource: a file, an event queue. And because the threads share a common resource, they must be synchronized in some way.

This lesson teaches you about Java thread synchronization through a simple Producer/Consumer example.

Producer/Consumer Example

The Producer generates integers ranging from 0 to 9, stores it in a "CubbyHole" object, prints the generated number, and (just to make the synchronization problem more interesting) the Producer sleeps for a random amount of time between 0 and 100 milliseconds.
class Producer extends Thread {
    private CubbyHole cubbyhole;
    private int number;

    public Producer(CubbyHole c, int number) {
	cubbyhole = c;
	this.number = number;
    }

    public void run() {
	for (int i = 0; i < 10; i++) {
	    cubbyhole.put(i);
	    System.out.println("Producer #" + this.number + " put: " + i);
	    try {
	        sleep((int)(Math.random() * 100));
	    } catch (InterruptedException e) {
	    }
	}
    }
}
The Consumer, being ravenous, consumes all integers from the CubbyHole (the exact same object into which the Producer put the integers in the first place) as quickly as they become available.
class Consumer extends Thread {
    private CubbyHole cubbyhole;
    private int number;

    public Consumer(CubbyHole c, int number) {
	cubbyhole = c;
	this.number = number;
    }

    public void run() {
	int value = 0;
	for (int i = 0; i < 10; i++) {
	    value = cubbyhole.get();
	    System.out.println("Consumer #" + this.number + " got: " + value);
	}
    }
}

The Producer and Consumer in this example share data through a common CubbyHole object. And you will note that neither the Producer nor the Consumer make any effort whatsoever to ensure that the Consumer is getting each value produced once and only once. The synchronization between these two threads actually occurs at a lower level, within the get() and put() methods of the CubbyHole object. However, let's assume for a moment that no synchronization between these two threads and talk about the potential problems that might arise in that situation.

One problem that might arise if the Producer and Consumer threads make no arrangements to run synchronously occurs when the Producer is quicker than the Consumer and generates two numbers before the Consumer has a chance to consume the first one. Thus the Consumer would skip a number. Part of the output might look like this:

    . . .

Consumer #1 got: 3
Producer #1 put: 4
Producer #1 put: 5
Consumer #1 got: 5

    . . .
Another problem that might arise is when the Consumer is quicker than the Producer and consumes the same value twice. In this situation, the Consumer would print the same value twice and might produce output that looked like this:
    . . .

Producer #1 put: 4
Consumer #1 got: 4
Consumer #1 got: 4
Producer #1 put: 5

    . . .
Either way, the result is wrong. You want the Consumer to get each integer produced by the Producer exactly once. Problems, such as those just described, that arise from multiple, asynchronously executing threads trying to a single object at the same time and getting the wrong result, are called race conditions.

To prevent race conditions in our Producer/Consumer example, the storage of a new integer into the CubbyHole by the Producer must be synchronized with the retrieval of an integer from the CubbyHole by the Consumer. The Consumer must consume each integer exactly once. Objects such as the CubbyHole, which are shared between two threads and whose accesses must be synchronized, are called condition variables. The Java language allows you to synchronize threads around a condition variable through the use of monitors.

Monitors

The Java language and runtime system support thread synchronization through the use of monitors which were first outlined in C. A. R. Hoare's article Communicating Sequential Processes (Communications of the ACM, Vol. 21, No. 8, August 1978, pp. 666-677). In general, a monitor is associated with a specific data item (the condition variable) and functions as a lock on that data. When a thread holds the monitor for some data item, other threads are locked out and cannot inspect or modify the data.

The code segments within a program that make it possible for separate, concurrent threads to access the same data items are known as critical sections. In the Java language, you identify critical sections in your program with the synchronized keyword.
Note: Generally, critical sections in Java programs are methods. You can mark smaller code segments as synchronized. However, this violates object-oriented paradigms and leads to confusing code that is difficult to debug and maintain. For the majority of your Java programming purposes, it's best to use synchronized only at the method level.

In the Java language, a unique monitor is associated with every object that has a synchronized method. The CubbyHole class for the Producer/Consumer example introduced above has two synchronized methods: the put() method used to change the value in the CubbyHole and the get() method used to retrieve the current value.

class CubbyHole {
    private int seq;
    private boolean available = false;

    public synchronized int get() {
        while (available == false) {
	    try {
		wait();
	    } catch (InterruptedException e) {
	    }
	}
        available = false;
	notify();
	return seq;
    }

    public synchronized void put(int value) {
	while (available == true) {
	    try {
		wait();
	    } catch (InterruptedException e) {
	    }
	}
	seq = value;
        available = true;
        notify();
    }
}
In addition, the CubbyHole has two private variables: seq which is the current contents of the CubbyHole, and the boolean variable available which indicates whether the CubbyHole contents can be retrieved. When available is true the Producer has just put a new value in the CubbyHole and the Consumer has not yet consumed it. The Consumer can only consume the value in the CubbyHole when available is true.

Because CubbyHole has synchronized methods, the Java language provides a unique monitor for each instance of CubbyHole, including the one shared by the Producer and the Consumer. Whenever the Producer calls the CubbyHole's put() method, the Producer acquires the monitor for the CubbyHole thereby preventing the Consumer from calling the CubbyHole'sget() method.

public synchronized void put(int value) {
	// monitor has been acquired by the Producer
    while (available == true) {
        try {
            wait();
        } catch (InterruptedException e) {
        }
    }
    seq = value;
    available = true;
    notify();
	// monitor is released by the Producer
}
When the put() method returns, the Producer releases the monitor thereby unlocking the CubbyHole.

Conversely, whenever the Consumer calls the CubbyHole's get() method, the Consumer acquires the monitor for the CubbyHole thereby preventing the Producer from calling the put() method.

public synchronized int get() {
	// monitor has been acquired by the Consumer
    while (available == false) {
	try {
	    wait();
	} catch (InterruptedException e) {
	}
    }
    available = false;
    notify();
    return seq;
	// monitor is released by the Consumer
}
The acquisition and release of a monitor is done atomically by the Java runtime system. This ensures that race conditions cannot occur in the underlying implementation of the threads and ensures data integrity.

You might have noticed a potential problem in CubbyHole's put() and get() methods. At the beginning of the get() method, if the value in the CubbyHole is not available (that is, the Producer has not generated a new number since the last time the Consumer consumed it) then the Consumer waits for the Producer to put a new value into the CubbyHole. So, the question arises--how can the Producer put a new value into the CubbyHole, if the Consumer holds the monitor (the Consumer holds the CubbyHole's monitor because it's within the synchronized method get())?

Similarly, at the beginning of the put() method, if the value in the CubbyHole has not yet been consumed then the Producer waits for the Consumer to consume the value in the CubbyHole. And again the question arises--how can the Consumer consume the value in the CubbyHole, if the Producer holds the monitor (the Producer holds the CubbyHole's monitor because it's within the synchronized method put())?

Well, the designers of the Java language thought of this too. When the thread enters the wait() method, which happens at the beginning of both the put() and get methods, the monitor is released atomically, and when the thread exits the wait() method, the monitor is acquired again. This gives the waiting object the opportunity to acquire the monitor and, depending on who's waiting, consume the value in the CubbyHole or produce a new value for the CubbyHole.

The Main Program

Here's a small stand-alone Java application that creates a CubbyHole object, one Producer, one Consumer, and then starts both the Producer and the Consumer.
class ProducerConsumerTest {
    public static void main(String args[]) {
	CubbyHole c = new CubbyHole();
	Producer p1 = new Producer(c, 1);
	Consumer c1 = new Consumer(c, 1);

        p1.start();
        c1.start();
    }
}

The Output

Here's the output of ProducerConsumerTest.
Producer #1 put: 0
Consumer #1 got: 0
Producer #1 put: 1
Consumer #1 got: 1
Producer #1 put: 2
Consumer #1 got: 2
Producer #1 put: 3
Consumer #1 got: 3
Producer #1 put: 4
Consumer #1 got: 4
Producer #1 put: 5
Consumer #1 got: 5
Producer #1 put: 6
Consumer #1 got: 6
Producer #1 put: 7
Consumer #1 got: 7
Producer #1 put: 8
Consumer #1 got: 8
Producer #1 put: 9
Consumer #1 got: 9

Try this: Remove the lines that are shown in bold in the listing of the CubbyHole class shown above. Recompile the program and run it again. What happened? Because no explicit effort has been made to synchronize the Producer and Consumer threads, the Consumer consumes with reckless abandon and gets a whole bunch of zeros instead of getting each integer between 0 and 9 exactly once.


Previous | Next | Trail Map | Writing Java Programs | Threads of Control