Acknowledgment: OOPSLA 2007 tutorial by Joe Bowbeer and David Holmes

http://www.oopsla.org/oopsla2007/index.php?page=sub/&id=69

Java Concurrency Utilities in Practice

Joe Bowbeer
Java ME Specialist Mobile App Consulting
[email protected]

David Holmes
Senior Java Technologist Sun Microsystems Australia
[email protected]

Contributing authors: Doug Lea

State University of New York, Oswego [email protected]

Tim Peierls
BoxPop.biz [email protected]

Brian Goetz
Sun Microsystem Inc. [email protected]

About these slides

Java is a trademark of Sun Microsystems, Inc. Material presented is based on latest information available for JavaTM Platform Standard Edition, as implemented in JDK 6.0 Code fragments elide
Exception handling for simplicity Access modifiers unless relevant

More extensive coverage of most topics can be found in the book

Java Concurrency in Practice, by Brian Goetz et al, Addison-Wesley (JCiP)

See also
Concurrent Programming in Java, by Doug Lea, Addison-Wesley (CPJ)
4

Review: Java Threading Model

The Java virtual machine (JVM)

Creates the initial thread which executes the main method of the class passed to the JVM Creates internal JVM helper threads
Garbage collection, finalization, signal dispatching

The code executed by the main thread can create other threads
Either explicitly; or Implicitly via libraries:
AWT/Swing, Applets Servlets, web services RMI image loading 5

Review: Java Thread Creation

Concurrency is introduced through objects of the class Thread
Provides a handle to an underlying thread of control

There is always a current thread running: a

Static method Thread.currentThread()

The start() method

Creates a new thread of control to execute the Thread objects run() method

Two ways to provide a run() method:

Subclass Thread and override run() Define a class that implements the Runnable interface and get the Thread object to run it
new Thread(aRunnable).start();

Runnable defines the abstraction of work Thread defines the abstraction of a worker
6

Review: Thread Interaction

void start()
Creates a new thread of control to execute the run() method of the Thread object Can only be invoked once per Thread object

void join()
Waits for a thread to terminate

t1.join(); // blocks current thread until t1 terminates

static void sleep(long ms)throws InterruptedException

Blocks current thread for approximately at least the specified time

static void yield()

Allows the scheduler to select another thread to run

Review: Java Synchronization

Every Java object has an associated lock acquired via:

synchronized statements
} synchronized( foo ){ // execute code while holding foos lock

synchronized methods

public synchronized void op1(){ // execute op1 while holding this lock }

Only one thread can hold a lock at a time

If the lock is unavailable the thread is blocked Locks are granted per-thread: reentrant or recursive locks

Locking and unlocking are automatic

Cant forget to release a lock Locks are released when a block goes out of scope

By normal means or when an exception is thrown

8

Review: Use of wait/notify

Waiting for a condition to hold:

synchronized (obj) { // obj protects the mutable state while (!condition) { try { obj.wait(); } catch (InterruptedException ex) { ... } } // make use of condition while obj still locked }

Changing a condition:
synchronized (obj) { // obj protects the mutable state condition = true; obj.notifyAll(); // or obj.notify() }

Golden rule: Always test a condition in a loop

Change of state may not be what you need Condition may have changed again

No built-in protection from barging

Spurious wakeups are permitted and can occur 9

java.util.concurrent

General purpose toolkit for developing concurrent applications
No more reinventing the wheel!

Goals: Something for Everyone!

Make some problems trivial to solve by everyone
Develop thread-safe classes, such as servlets, built on concurrent building blocks like ConcurrentHashMap

Make some problems easier to solve by concurrent programmers

Develop concurrent applications using thread pools, barriers, latches, and blocking queues

Make some problems possible to solve by concurrency experts

Develop custom locking classes, lock-free algorithms

10

Overview of j.u.c

Executors Executor ExecutorService ScheduledExecutorService Callable Future ScheduledFuture Delayed CompletionService ThreadPoolExecutor ScheduledThreadPoolExecutor AbstractExecutorService Executors FutureTask ExecutorCompletionService

Concurrent Collections ConcurrentMap ConcurrentHashMap CopyOnWriteArray{List,Set}

Synchronizers CountDownLatch Semaphore Exchanger CyclicBarrier

Locks: java.util.concurrent.locks Lock Condition ReadWriteLock AbstractQueuedSynchronizer LockSupport ReentrantLock ReentrantReadWriteLock

Queues BlockingQueue ConcurrentLinkedQueue LinkedBlockingQueue ArrayBlockingQueue SynchronousQueue PriorityBlockingQueue DelayQueue

Atomics: java.util.concurrent.atomic Atomic[Type] Atomic[Type]Array Atomic[Type]FieldUpdater Atomic{Markable,Stampable}Reference

11

Key Functional Groups in j.u.c.

Executors, Thread pools and Futures
Execution frameworks for asynchronous tasking

Concurrent Collections:
Queues, blocking queues, concurrent hash map, Data structures designed for concurrent environments

Locks and Conditions

More flexible synchronization control Read/write locks

Synchronizers: Semaphore, Latch, Barrier, Exchanger

Ready made tools for thread coordination

Atomic variables
The key to writing lock-free algorithms
12

The Executor Framework

Framework for asynchronous task execution Standardize asynchronous invocation
Framework to execute Runnable and Callable tasks
Runnable: void run() Callable<V>: V call() throws Exception

Separate submission from execution policy

Use anExecutor.execute(aRunnable) Not new Thread(aRunnable).start()

Cancellation and shutdown support Usually created via Executors factory class
Configures flexible ThreadPoolExecutor Customize shutdown methods, before/after hooks, saturation policies, queuing
13

Creating Executors

Sample ExecutorService implementations from Executors

newSingleThreadExecutor
A pool of one, working from an unbounded queue

newFixedThreadPool(int N)
A fixed pool of N, working from an unbounded queue

newCachedThreadPool
A variable size pool that grows as needed and shrinks when idle

newScheduledThreadPool(int N)
Pool for executing tasks after a given delay, or periodically

14

Thread Pools

Use a collection of worker threads, not just one

Can limit maximum number and priorities of threads Dynamic worker thread management

Sophisticated policy controls

Often faster than thread-per-message for I/O bound actions
15

ThreadPoolExecutor

Sophisticated ExecutorService implementation with numerous tuning parameters

Core and maximum pool size
Thread created on task submission until core size reached Additional tasks queued until queue is full Thread created if queue full until maximum size reached Note: unbounded queue means the pool wont grow above core size

Keep-alive time
Threads above the core size terminate if idle for more than the keep-alive time In JDK 6 core threads can also terminate if idle

Pre-starting of core threads, or else on demand

16

Working with ThreadPoolExecutor

ThreadFactory used to create new threads
Default: Executors.defaultThreadFactory

Queuing strategies: must be a BlockingQueue<Runnable>

Direct hand-off via SynchronousQueue: zero capacity; hands-off to waiting thread, else creates new one if allowed, else task rejected Bounded queue: enforces resource constraints, when full permits pool to grow to maximum, then tasks rejected Unbounded queue: potential for resource exhaustion but otherwise never rejects tasks

Queue is used internally

Use remove or purge to clear out cancelled tasks You should not directly place tasks in the queue
Might work, but you need to rely on internal details

Subclass customization hooks: beforeExecute and afterExecute

17

Futures

Encapsulates waiting for the result of an asynchronous computation launched in another thread
The callback is encapsulated by the Future object

Usage pattern
Client initiates asynchronous computation via oneway message Client receives a handle to the result: a Future Client performs additional tasks prior to using result Client requests result from Future, blocking if necessary until result is available Client uses result

Assumes truly concurrent execution between client and task

Otherwise no point performing an asynchronous computation

Assumes client doesnt need result immediately

Otherwise it may as well perform the task directly
18

Future<V> Interface

V get()
Retrieves the result held in this Future object, blocking if necessary until the result is available Timed version throws TimeoutException If cancelled then CancelledException thrown If computation fails throws ExecutionException

boolean isDone()
Queries if the computation has completedwhether successful, cancelled or threw an exception

boolean isCancelled()
Returns true if the computation was cancelled before it completed

19

Simple Future Example

Asynchronous rendering in a graphics application interface Pic { byte[] getImage(); } interface Renderer { Pic render(byte[] raw); } class App { // sample usage void app(final byte[] raw) throws ... { final Renderer r = ; FutureTask<Pic> p = new FutureTask<Pic>( new Callable<Pic>() { Pic call() { return r.render(raw); } }); new Thread(p).start(); doSomethingElse(); display(p.get()); // wait if not yet ready } // ... }
20

Key Functional Groups in j.u.c.

Executors, Thread pools and Futures
Execution frameworks for asynchronous tasking

Concurrent Collections:
Queues, blocking queues, concurrent hash map, Data structures designed for concurrent environments

Locks and Conditions

More flexible synchronization control Read/write locks

Synchronizers: Semaphore, Latch, Barrier, Exchanger

Ready made tools for thread coordination

Atomic variables
The key to writing lock-free algorithms
21

Concurrent Collections

22

Concurrent Collections

23

Iteration Semantics

24

Concurrent Collection Performance

25

ConcurrentMap

Atomic get-and-maybe-set methods for maps interface ConcurrentMap<K,V> extends Map<K,V> { V putIfAbsent(K key, V value); V replace(K key, V value); boolean replace(K key, V oldValue, V newValue); boolean remove(K key, V value); }

26

Key Functional Groups in j.u.c.

Executors, Thread pools and Futures
Execution frameworks for asynchronous tasking

Concurrent Collections:
Queues, blocking queues, concurrent hash map, Data structures designed for concurrent environments

Locks and Conditions

More flexible synchronization control Read/write locks

Synchronizers: Semaphore, Latch, Barrier, Exchanger

Ready made tools for thread coordination

Atomic variables
The key to writing lock-free algorithms
27

Locks

28

Lock / ReentrantLock
interface Lock { void lock(); void lockInterruptibly() throws InterruptedException; boolean tryLock(); boolean tryLock(long timeout, TimeUnit unit) throws InterruptedException; void unlock(); Condition newCondition(); }

Additional flexibility
Interruptible, try-lock, not block-structured, multiple conditions Advanced uses: e.g. Hand-over-hand or chained locking

ReentrantLock: mutual-exclusion Lock implementation

Same basic semantics as synchronized
Reentrant, must hold lock before using condition,

Supports fair and non-fair behavior

Fair lock granted to waiting threads ahead of new requests

High performance under contention

29

Simple lock example

30

Key Functional Groups in j.u.c.

Executors, Thread pools and Futures
Execution frameworks for asynchronous tasking

Concurrent Collections:
Queues, blocking queues, concurrent hash map, Data structures designed for concurrent environments

Locks and Conditions

More flexible synchronization control Read/write locks

Synchronizers: Semaphore, Latch, Barrier, Exchanger

Ready made tools for thread coordination

Atomic variables
The key to writing lock-free algorithms
31

Synchronizers

32

CountDownLatch

A counter that releases waiting threads when it reaches zero

Allows one or more threads to wait for one or more events Initial value of 1 gives a simple gate or latch
CountDownLatch(int initialValue)

await: wait (if needed) until the counter is zero

Timeout version returns false on timeout Interruptible

countDown: decrement the counter if > 0 Query: getCount() Very simple but widely useful:
Replaces error-prone constructions ensuring that a group of threads all wait for a common signal 33

Single-use tool: no reset

Semaphores

Conceptually serve as permit holders

Construct with an initial number of permits acquire: waits for permit to be available, then takes one release: returns a permit But no actual permits change hands
The semaphore just maintains the current count No need to acquire a permit before you release it

fair variant hands out permits in FIFO order Supports balking and timed versions of acquire Applications:
Resource controllers Designs that otherwise encounter missed signals
Semaphores remember how often they were signalled 34

Bounded Blocking Concurrent List

Concurrent list with fixed capacity
Insertion blocks until space is available

Tracking free space, or available items, can be done using a Semaphore Demonstrates composition of data structures with library synchronizers
Much, much easier than modifying implementation of concurrent list directly

35

Bounded Blocking Concurrent List

public class BoundedBlockingList { final int capacity; final ConcurrentLinkedList list = new ConcurrentLinkedList(); final Semaphore sem; public BoundedBlockingList(int capacity) { this.capacity = capacity; sem = new Semaphore(capacity); } public void addFirst(Object x) throws InterruptedException { sem.acquire(); try { list.addFirst(x); } catch (Throwable t){ sem.release(); rethrow(t); } } public boolean remove(Object x) { if (list.remove(x)) { sem.release(); return true; } return false; } }

36

Key Functional Groups in j.u.c.

Executors, Thread pools and Futures
Execution frameworks for asynchronous tasking

Concurrent Collections:
Queues, blocking queues, concurrent hash map, Data structures designed for concurrent environments

Locks and Conditions

More flexible synchronization control Read/write locks

Synchronizers: Semaphore, Latch, Barrier, Exchanger

Ready made tools for thread coordination

Atomic variables
The key to writing lock-free algorithms
37

Atomic Variables

Holder classes for scalars, references and fields
java.util.concurrent.atomic

Support atomic operations

Compare-and-set (CAS)
boolean compareAndSet(T expected, T update) Atomically sets value to update if currently expected Returns true on successful update

Get, set and arithmetic operations (where applicable)

Increment, decrement operations

Nine main classes:

{ int, long, reference } X { value, field, array } E.g. AtomicInteger useful for counters, sequence numbers, statistics gathering
38

AtomicInteger Example

39

Case Study: Memoizer

Implement a class for memorizing function results Memo Function:
A function that memorizes its previous results
Optimization for recursive functions, etc.

Invented by Prof. Donald Michie, Univ. of Edinburgh

Goal: Implement Memoizer

Function wrapper Provide concurrent access Compute each result at most once

Tools:
ConcurrentHashMap FutureTask
40

Memoizer: Generic Computation

Generic computation
interface Computable<A, V> { V compute(A arg) throws Exception; }

Representative example
class ComplexFunction implements Computable<String, BigInteger> { public BigInteger compute(String arg) { // after deep thought... return new BigInteger("2"); } }
41

Memoizer: Usage

Current use of ComplexFunction requires local caching of result (or expensive re-compute) Computable<String, BigInteger> f = new ComplexFunction(); BigInteger result = f.compute("1+1"); // cache result for future use

Memoizer encapsulates its own caching Computable<String, BigInteger> f = new ComplexFunction(); f = new Memoizer<String, BigInteger>(f); BigInteger result = f.compute("1+1"); // call f.compute whenever we need to
42

Synchronized Memoizer

Safe but not concurrent class SyncMemoizer<A,V> implements Computable<A,V> { final Map<A, V> cache = new HashMap<A, V>(); final Computable<A, V> func; SyncMemoizer(Computable<A, V> func) { this.func = func; } public synchronized V compute(A arg)throws Exception{ if (!cache.containsKey(arg)) cache.put(arg, func.compute(arg)); return cache.get(arg); } }
43

Non-atomic Concurrent Memoizer

Safe, concurrent (no sync) but computes may overlap class NonAtomicMemoizer<A,V> implements Computable<A,V> { final Map<A, V> cache = new ConcurrentHashMap<A, V>(); final Computable<A, V> func; NonAtomicMemoizer(Computable<A, V> func) { this.func = func; } public V compute(A arg) throws Exception { if (!cache.containsKey(arg)) cache.put(arg, func.compute(arg)); return cache.get(arg); }

44

Concurrent Memoizer Using Future

Safe, concurrent and exactly one compute per argument class ConcurrentMemoizer<A, V> implements Computable<A, V> { final ConcurrentMap<A, Future<V>> cache = new ConcurrentHashMap<A, Future<V>>(); final Computable<A, V> func; ConcurrentMemoizer(Computable<A, V> func) { this.func = func; } ...
45

Concurrent Memoizer Using Future (2)

public V compute(final A arg) throws Exception{ Future<V> f = cache.get(arg); if (f == null) { Callable<V> eval = new Callable<V>() { public V call() throws Exception { return func.compute(arg); } }; FutureTask<V> ft = new FutureTask<V>(eval); f = cache.putIfAbsent(arg, ft); if (f == null) { f = ft; ft.run(); } } return f.get(); }

46

Case Study: Concurrent Linked List

Goal: Implement a concurrent linked-list
Demonstrate chained-locking

Tools:
ReentrantLock

Goal: Implement a blocking bounded list

Demonstrate composition: data structure + synchronizer

Tools:
Semaphore

47

Concurrent Linked List Locking Strategy

Design goal: fine-grained concurrent access Solution: lock-per-node Basic principle: all accesses traverse from the head in-order
To access a node it must be locked To add a new node the node before must be locked To remove a node both the node and the node before must be locked

Hand-over-hand Locking:
Lock n1, lock n2, unlock n1, lock n3, unlock n2, lock n4, Order in which threads acquire the first lock is maintained
No overtaking once traversal starts

Full version would implement java.util.List

48

Concurrent Linked List #1

public class ConcurrentLinkedList { /** * Holds one item in a singly-linked list. * It's convenient here to subclass ReentrantLock * rather than add one as a field. */ private static class Node extends ReentrantLock { Object item; Node next; Node(Object item, Node next) { this.item = item; this.next = next; } } /** * Sentinel node. This node's next field points to * the first node in the list. */ private final Node sentinel = new Node(null, null);49

Concurrent Linked List #2

public void addFirst(Object x) { Node p = sentinel; p.lock(); // acquire first lock try { p.next = new Node(x, p.next); // Attach new node } finally { p.unlock(); } }

Locking considerations
What needs to be unlocked in the normal case? What needs to be unlocked if an exception occurs?
Will the list still be in a consistent state? Note: cant protect against asynchronous exceptions

Simple in this case: only one lock held, only one failure mode

Note: Lock.lock() could throw exception e.g. OutOfMemoryError

50

Concurrent Linked List #3

public void addLast(Object x) { Node p = sentinel; p.lock(); // Acquire first lock try { // Find tail, using hand-over-hand locking while (p.next != null) { // p is always locked here Node prevp = p; p.next.lock(); // Acquire next lock p = p.next; prevp.unlock(); // Release previous lock } // only p is still locked here p.next = new Node(x, null); // Attach new node } finally { p.unlock(); // Release final lock } }

Again exception handling is easy to do but harder to reason about! Note: NullPointerException and IllegalMonitorStateException only possible if list code is broken 51

Concurrent Linked List #4

public boolean contains(Object x) { Node p = sentinel; p.lock(); // Acquire first lock try { // Find item, using hand-over-hand locking while (p.next != null) { // p is always locked here Node prevp = p; p.next.lock(); // Acquire next lock p = p.next; prevp.unlock(); // Release previous lock // found it? if (x == p.item || x != null && x.equals(p.item)) return true; } // only p is still locked now return false; } finally { p.unlock(); // Release final lock } } 52

Concurrent Linked List #5

public boolean remove(Object x) { Node p = sentinel; p.lock(); // Acquire first lock try { // Find item, using hand-over-hand locking while (p.next != null) { Node prevp = p; p.next.lock(); // Acquire next lock p = p.next; // cant unlock prevp yet as removal of p // requires update of prevp.next try { if (x==p.item || x!=null && x.equals(p.item)) { prevp.next = p.next; // remove node p return true; } } finally { prevp.unlock(); // Release previous lock } } return false; } finally { p.unlock(); // Release final lock } }

53