Reading 22: Queues and Message-Passing

In our introduction to concurrency, we saw two models for concurrent programming: shared memory and message passing.

• 

  In the shared memory model, concurrent modules interact by reading and writing shared mutable objects in memory. Creating multiple threads inside a single Java process is our primary example of shared-memory concurrency.

• In the message passing model, concurrent modules interact by sending immutable messages to one another over a communication channel. That communication channel might connect different computers over a network, as in some of our initial examples: web browsing, instant messaging, etc.

  

The message passing model has several advantages over the shared memory model, which boil down to greater safety from bugs. In message-passing, concurrent modules interact explicitly, by passing messages through the communication channel, rather than implicitly through mutation of shared data. The implicit interaction of shared memory can too easily lead to inadvertent interaction, sharing and manipulating data in parts of the program that don’t know they’re concurrent and aren’t cooperating properly in the thread safety strategy. Message passing also shares only immutable objects (the messages) between modules, whereas shared memory requires sharing mutable objects, which we have already seen can be a source of bugs.

We’ll discuss in this reading how to implement message passing within a single process. We’ll use blocking queues (an existing threadsafe type) to implement message passing between threads within a process.

In the next concurrency reading we’ll see how to implement message passing between client/server processes over the network.

We saw in Locks and Synchronization that a thread blocks trying to acquire a lock until the lock has been released by its current owner. Blocking means that a thread waits (without doing further work) until an event occurs. We can use this term to describe methods and method calls: if a method is a blocking method, then a call to that method can block, waiting until some event occurs before it returns to the caller.

We can use a queue with blocking operations for message passing between threads — and the buffered network communication channel in client/server message passing will work the same way. Java provides the BlockingQueue interface for queues with blocking operations.

In an ordinary Queue:

• add(e) adds element e to the end of the queue.
• remove() removes and returns the element at the head of the queue, or throws an exception if the queue is empty.

A BlockingQueue extends this interface:

additionally supports operations that wait for the queue to become non-empty when retrieving an element, and wait for space to become available in the queue when storing an element.

• put(e) blocks until it can add element e to the end of the queue (if the queue does not have a size bound, put will not block).
• take() blocks until it can remove and return the element at the head of the queue, waiting until the queue is non-empty.

When you are using a BlockingQueue for message passing between threads, make sure to use the put() and take() operations, not add() and remove().

We will implement the producer-consumer design pattern for message passing between threads. Producer threads and consumer threads share a synchronized queue. Producers put data or requests onto the queue, and consumers remove and process them. One or more producers and one or more consumers might all be adding and removing items from the same queue. This queue must be safe for concurrency.

Java provides two implementations of BlockingQueue:

• ArrayBlockingQueue is a fixed-size queue that uses an array representation. putting a new item on the queue will block if the queue is full.
• LinkedBlockingQueue is a growable queue using a linked-list representation. If no maximum capacity is specified, the queue will never fill up, so put will never block.

Like other collections classes in Java, these synchronized queues can hold objects of an arbitrary type. We must choose or design a type for messages in the queue: we will choose an immutable type because our goal in using message passing is to avoid the problems of shared memory. Producers and consumers will only communicate only by sending and receiving messages, and there will be no opportunity for (mis)communication by mutating an aliased message object.

And just as we designed the operations on a threadsafe ADT to prevent race conditions and enable clients to perform the atomic operations they need, we will design our design our message objects with those same requirements.

Bank account example

Our first example of message passing was the bank account example.

Each cash machine and each account is its own module, and modules interact by sending messages to one another. Incoming messages arrive on a queue.

We designed messages for get-balance and withdraw, and said that each cash machine checks the account balance before withdrawing to prevent overdrafts:

get-balance
if balance >= 1 then withdraw 1

But it is still possible to interleave messages from two cash machines so they are both fooled into thinking they can safely withdraw the last dollar from an account with only $1 in it.

We need to choose a better atomic operation: withdraw-if-sufficient-funds would be a better operation than just withdraw.

You can see all the code for this example on GitHub: squarer example. All the relevant parts are excerpted below.

Here’s a message passing module for squaring integers:

/** Squares integers. */
public class Squarer {

    private final BlockingQueue<Integer> in;
    private final BlockingQueue<SquareResult> out;
    // Rep invariant: in, out != null

    /** Make a new squarer.
     *  @param requests queue to receive requests from
     *  @param replies queue to send replies to */
    public Squarer(BlockingQueue<Integer> requests,
                   BlockingQueue<SquareResult> replies) {
        this.in = requests;
        this.out = replies;
    }

    /** Start handling squaring requests. */
    public void start() {
        new Thread(new Runnable() {
            public void run() {
                while (true) {
                    // TODO: we may want a way to stop the thread
                    try {
                        // block until a request arrives
                        int x = in.take();
                        // compute the answer and send it back
                        int y = x * x;
                        out.put(new SquareResult(x, y));
                    } catch (InterruptedException ie) {
                        ie.printStackTrace();
                    }
                }
            }
        }).start();
    }
}

Incoming messages to the Squarer are integers; the squarer knows that its job is to square those numbers, so no further details are required.

Outgoing messages are instances of SquareResult:

/** An immutable squaring result message. */
public class SquareResult {
    private final int input;
    private final int output;

    /** Make a new result message.
     *  @param input input number
     *  @param output square of input */
    public SquareResult(int input, int output) {
        this.input = input;
        this.output = output;
    }

    @Override public String toString() {
        return input + "^2 = " + output;
    }
}

We would probably add additional observers to SquareResult so clients can retrieve the input number and output result.

Finally, here’s a main method that uses the squarer:

public static void main(String[] args) {

    BlockingQueue<Integer> requests = new LinkedBlockingQueue<>();
    BlockingQueue<SquareResult> replies = new LinkedBlockingQueue<>();

    Squarer squarer = new Squarer(requests, replies);
    squarer.start();

    try {
        // make a request
        requests.put(42);
        // ... maybe do something concurrently ...
        // read the reply
        System.out.println(replies.take());
    } catch (InterruptedException ie) {
        ie.printStackTrace();
    }
}

private void checkRep() {
  assert REP_INVARIANT;
}

What if we want to shut down the Squarer so it is no longer waiting for new inputs? One strategy is a poison pill: a special message on the queue that signals the consumer of that message to end its work.

To shut down the squarer, since its input messages are merely integers, we would have to choose a magic poison integer (everyone knows the square of 0 is 0 right? surely they’ll never ask for the square of 0? don’t use magic numbers) or use null (don’t use null). Instead, we might change the type of elements on the requests queue to an ADT:

SquareRequest = IntegerRequest + StopRequest

with operations:

input : SquareRequest → int
shouldStop : SquareRequest → boolean

and when we want to stop the squarer, we enqueue a SquareRequest where shouldStop returns true.

For example, in Squarer.start():

public void run() {
    while (true) {
        try {
            // block until a request arrives
            SquareRequest req = in.take();
            // see if we should stop
            if (▶▶A◀◀) { ▶▶B◀◀; }
            // compute the answer and send it back
            int x = ▶▶C◀◀;
            int y = x * x;
            out.put(new SquareResult(x, y));
        } catch (InterruptedException ie) {
            ie.printStackTrace();
        }
    }
}

SquareRequest = IntegerRequest + StopRequest

interface SquareRequest { ... }
class IntegerRequest implements SquareRequest { ... }
class StopRequest implements SquareRequest { ... }

class SquareRequest { ... }
class IntegerRequest { ... }
class StopRequest { ... }

class SquareRequest {
  private final String requestType;
  public static final String INTEGER_REQUEST = "integer";
  public static final String STOP_REQUEST    = "stop";
  ...
}

It is also possible to interrupt a thread by calling its interrupt() method. If the thread is blocked waiting, the method it’s blocked in will throw an Interrupted­Exception (that’s why we have to try-catch that exception almost any time we call a blocking method). If the thread was not blocked, an interrupted flag will be set. The thread must both handle any Interrupted­Exceptions and check for the interrupted flag to see whether it should stop working. For example:

public void run() {
    // handle requests until we are interrupted
    while ( ! Thread.interrupted()) {
        try {
            // block until a request arrives
            int x = in.take();
            // compute the answer and send it back
            int y = x * x;
            out.put(new SquareResult(x, y));
        } catch (InterruptedException ie) {
            // stop
            break;
        }
    }
}

A thread safety argument with message passing might rely on:

• Existing threadsafe data types for the synchronized queue. This queue is definitely shared and definitely mutable, so we must ensure it is safe for concurrency.

• Immutability of messages or data that might be accessible to multiple threads at the same time.

• Confinement of data to individual producer/consumer threads. Local variables used by one producer or consumer are not visible to other threads, which only communicate with one another using messages in the queue.

• Confinement of mutable messages or data that are sent over the queue but will only be accessible to one thread at a time. This argument must be carefully articulated and implemented. Suppose one thread has some mutable data to send to another thread. If the first thread drops all references to the data like a hot potato as soon as it puts them onto a queue for delivery to the other thread, then only one thread will have access to those data at a time, precluding concurrent access.

In comparison to synchronization, message passing can make it easier for each module in a concurrent system to maintain its own thread safety invariants. We don’t have to reason about multiple threads accessing shared data if the data are instead transferred between modules using a threadsafe communication channel.

public interface Trade {
    public int numShares();
    public String stockName();
}

public class TradeWorker implements Runnable {
    private final Queue<Trade> tradesQueue;

    public TradeWorker(Queue<Trade> tradesQueue) {
        this.tradesQueue = tradesQueue;
    }

    public void run() {
        while (true) {
            Trade trade = tradesQueue.poll();
            TradeProcessor.handleTrade(trade.numShares(), trade.stockName());
        }
    }
}

public class TradeProcessor {
    public static void handleTrade(int numShares, String stockName) {
        /* ... process the trade ... takes a while ... */
    }
}

• Rather than synchronize with locks, message passing systems synchronize on a shared communication channel, e.g. a stream or a queue.

• Threads communicating with blocking queues is a useful pattern for message passing within a single process.