1 / 29

Effective Java: Concurrency

Effective Java: Concurrency. Last Updated: Spring 2010. Agenda. Material From Joshua Bloch Effective Java: Programming Language Guide Cover Items 66-73 “Concurrency” Chapter Bottom Line: Primitive Java concurrency is complex.

kellerm
Download Presentation

Effective Java: Concurrency

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Effective Java:Concurrency Last Updated: Spring 2010

  2. Agenda • Material From Joshua Bloch • Effective Java: Programming Language Guide • Cover Items 66-73 • “Concurrency” Chapter • Bottom Line: • Primitive Java concurrency is complex

  3. Some Background (from Java Concurrency in Practice, B. Goetz, T. Peierls, J. Bloch, J. Bowbeer, D. Holmes, and D. Lea) • Thread: lightweight processes; execute simultaneously and asynchronously with each other; share same memory address space of owning process, all thread within a process have access to the same variables and allocate objects from the same heap • Thread Safety • Managing the state: with respect to shared and mutable state. • Shared: accessed by multiple threads. • Mutable: value could change during lifecycle.

  4. Some Background (from Java Concurrency in Practice, B. Goetz, T. Peierls, J. Bloch, J. Bowbeer, D. Holmes, and D. Lea) • Thread Safety???: • Correctness – invariants; postconditions • “A class is thread-safe if it behaves correctly when accessed from multiple threads, regardless of the scheduling or interleaving of the execution of the threads by the runtime environment, and with no additoinal synchronization or other coordination on the part of the calling code.” • “Thread-safe encapsulate any needed synchronization so that clients need not provide their own.”

  5. Java Concurrency Problems: Examples • Race conditions – threads try to update the same data structure at the same time • Deadlock – two threads need variables A and B to perform a calculation; one thread locks A and the other thread locks B; • Starvation – a thread is unable to obtain CPU time due a higher priority threads

  6. Thread Interference class Counter { private int c = 0; public void increment() { c++; } public void decrement() { c--; } public int value() { return c; } } • Interference happens when two operations, running in different threads, but acting on the same data, interleave. This means that the two operations consist of multiple steps, and the sequences of steps overlap.

  7. Thread Interference (concluded) • c++ can be decomposed into three steps • Retrieve the current value of c • Increment the retrieved value by 1 • Store the incremented value back in c • Suppose Thread A invokes increment at about the same time Thread B invokes decrement. If the initial value of c is 0, their interleaved actions might follow this sequence: • Thread A: Retrieve c. • Thread B: Retrieve c. • Thread A: Increment retrieved value; result is 1. • Thread B: Decrement retrieved value; result is -1. • Thread A: Store result in c; c is now 1. • Thread B: Store result in c; c is now -1. • N.B., Thread A's result is lost, overwritten by Thread B. This particular interleaving is only one possibility. Under different circumstances it might be Thread B's result that gets lost, or there could be no error at all. Because they are unpredictable, thread interference bugs can be difficult to detect and fix.

  8. Memory Consistency Errors - Java Memory Model (from Java Concurrency in Practice, B. Goetz, T. Peierls, J. Bloch, J. Bowbeer, D. Holmes, and D. Lea) • Partial ordering on Java program actions (happens-before) • Read/write • Lock/unlock • Start/join threads • If action X happens-before Y, then X’s results are visible to Y. • Within a thread the order is the program order. • Between threads, if synchronized or volatile is not used, there are no visibility guarantees (i.e., there is no guarantee that thread A will see them in the order that thread be executes them).

  9. Item 66: Synchronize Access to Shared Mutable Data • Method synchronization yields atomic transitions: • public synchronized boolean doStuff() {…} • Fairly well understood… • Method synchronization also ensures that other threads “see” earlier threads • Not synchronizing on shared “atomic” data produces wildly counterintuitive results • Not well understood

  10. Java Synchronization • Communication between Java Threads • Java synchronization is implemented using monitors • Each object in Java is associated with a monitor, which a thread can lock and unlock • Only one thread at a time may hold a lock on a monitor. Any other threads attempting to lock that monitor are blocked until they can obtain a lock on that monitor.

  11. Basic tools of Java Thread Synchronization • Synchronized keyword – an unlock happens-before every subsequent lock on the same monitor • Volatile keyword – a write to a volatile variable happens-before subsequent reads of that variable; compiler and runtime are notified that the variable should not be reordered with memory operations; variables are not cached in registers or in caches where they are hidden from other processors, so a read of volatile variable always returns the most recent write by any thread (from Java Concurrency in Practice, B. Goetz, T. Peierls, J. Bloch, J. Bowbeer, D. Holmes, and D. Lea) • Static initialization – done by the class loader, so the JVM guarantees thread safety

  12. Atomic Example (from Java Concurrency in Practice, B. Goetz, T. Peierls, J. Bloch, J. Bowbeer, D. Holmes, and D. Lea) @NotThreadSafe Public class UnsafeCountingFactorizer implements Servlet { private long count; public long getCount() { return count; } public void service(ServletRequest req, ServletResponse resp) { BigInteger I = extractFromRequest(req); BigInteger[] factors = factor(i); ++count; encodeIntoResponse(resp, factors); Why is this unsafe? Atomic – execute as a single, indivisible operation.

  13. Item 66: Unsafe Example // Broken! How long do you expect this program to run? public class StopThread { private static boolean stopRequested; public static void main (String[] args) throws InterruptedException { Thread backgroundThread = new Thread(new Runnable() { public void run() { // May run forever! Liveness failure int i=o; while (! stopRequested) i++; // See below }}); backgroundThread.start(); TimeUnit.SECONDS.sleep(1); stopRequested = true; } } // Hoisting transform: // while (!loopTest) {i++;}  if (!loopTest) while(true) {i++;} // Also note anonymous class

  14. Item 66: Fixing the Example // As before, but with synchronized calls public class StopThread { private static boolean stopReq; public static synchronized void setStop() {stopReq = true;} public static synchronized void getStop() {return stopReq;} public static void main (String[] args) throws InterruptedException { Thread backgroundThread = new Thread(new Runnable() { public void run() { // Now “sees” main thread int i=o; while (! getStop() ) i++; }}); backgroundThread.start(); TimeUnit.SECONDS.sleep(1); setStop(); } } // Note that both setStop() and getStop() are synchronized

  15. Item 66: A volatile Fix for the Example // A fix with volatile public class StopThread { // Pretty subtle stuff, using the volatile keyword private static volatile boolean stopRequested; public static void main (String[] args) throws InterruptedException { Thread backgroundThread = new Thread(new Runnable() { public void run() { int i=o; while (! stopRequested) i++; }}); backgroundThread.start(); TimeUnit.SECONDS.sleep(1); stopRequested = true; } }

  16. Item 66: volatile Does Not Guarantee Mutual Exclusion // Broken! Requires Synchronization! private static volatile int nextSerialNumber = 0; public static int generateSerialNumber() { return nextSerialNumber++; } Problem is that the “++” operator is not atomic // Even better! (See Item 47) private static final AtomicLong nextSerial = new AtomicLong(); public static long generateSerialNumber() { return nextSerial.getAndIncrement(); }

  17. Item 66: Advice on Sharing Data Between Threads • Confine mutable data to a single Thread • May modify, then share (no further changes) • Called “Effectively Immutable” • Allows for “Safe Publication” • Mechanisms for safe publication • In static field at class initialization • volatile field • final field • field accessed with locking (ie synchronization) • Store in concurrent collection (Item 69)

  18. Item 67: Avoid Excessive Synchronization // Broken! Invokes alien method from sychronized block public interface SetOb<E> { void added(ObservableSet<E> set, E el);} public class ObservableSet<E> extends ForwardingSet<E> { // Bloch 16 public ObservableSet(Set<E> set) { super(set); } private final List<SetOb<E>> obs = new ArrayList<SetOb<E>>(); public void addObserver (SetObs<E> ob ) { synchronized (obs) { obs.add(ob); } } public boolean removeObserver (SetOb<E> ob ) { synchronized (obs) { return obs.remove(ob); } } private void notifyElementAdded (E el) { synchronized(obs) {for (SetOb<E> ob:obs) // Exceptions? ob.added(this, el);} @Override public boolean add(E el) { // from Set interface boolean added = super.add(el); if (added) notifyElementAdded (el); return added; }}

  19. More Item 67: What’s the Problem? public static void main (String[] args) { ObservableSet<Integer> set = new ObservableSet<Integer> (new HashSet<Integer>); set.addObserver (new SetOb<Integer>() { public void added (ObservableSet<Integer> s, Integer e) { System.out.println(e); if (e.equals(23)) s.removeObserver(this); // Oops! CME // See Bloch for a variant that deadlocks instead of CME } }); for (int i=0; i < 100; i++) set.add(i); }

  20. More Item 67: Turning the Alien Call into an Open Call // Alien method moved outside of synchronized block – open call private void notifyElementAdded(E el) { List<SetOb<E>> snapshot = null; synchronized (observers) { snapshot = new ArrayList<SetOb<E>>(obs); } for (SetObserver<E> observer : snapshot) observer.added(this, el) // No more CME }} Open Calls increase concurrency and prevent failures Rule: Do as little work inside synch block as possible When designing a new class: Do NOT internally synchronize absent strong motivation Example: StringBuffer vs. StringBuilder

  21. Item 67: Alternate Fix Using CopyOnWriteArray public interface SetOb<E> { void added(ObservableSet<E> set, E el);} public class ObservableSet<E> extends ForwardingSet<E> { // Bloch 16 public ObservableSet(Set<E> set) { super(set); } private final List<SetOb<E>> obs = new CopyOnWriteArrayList<SetOb<E>>(); public void addObserver (SetObs<E> ob ) { synchronized (obs) { obs.add(ob); } } public boolean removeObserver (SetOb<E> ob ) { synchronized (obs) { return obs.remove(ob); } } private void notifyElementAdded (E el) { {for (SetOb<E> ob:obs) // Iterate on copy – No Synch! ob.added(this, el);} @Override public boolean add(E el) { // from Set interface boolean added = super.add(el); if (added) notifyElementAdded (el); return added; }}

  22. Item 68: Prefer Executors and Tasks to Threads • Old key abstraction: Thread • Unit of work and • Mechanism for execution • New key abstractions: • Task (Unit of work) • Runnable and Callable • Mechanism for execution • Executor Service • Start tasks, wait on particular tasks, etc. • See Bloch for references

  23. Item 69: Prefer Concurrency Utilities to wait and notify • wait() and notify() are complex • Java concurrency facilities much better • Legacy code still requires understanding low level primitives • Three mechanisms • Executor Framework (Item 68) • Concurrent collections • Internally synchronized versions of Collection classes • Extensions for blocking, Example: BlockingQueue • Synchronizers • Objects that allow Threads to wait for one another

  24. More Item 69: Timing Example // Simple framework for timing concurrent execution public static long time (Executor executor, int concurrency, final Runnable action) throws InterrruptedExecution { final CountDownLatch ready = new CountDownLatch(concurrency); final CountDownLatch start = new CountDownLatch(1); final CountDownLatch done = new CountDownLatch(concurrency); for (int i=0; i< concurrency; i++) { executor.execute (new Runnable() { public void run() { ready.countDown(); // Tell Timer we’re ready try { start.await(); action.run(); // Wait till peers are ready } catch (...){ ...} } finally { done.countDown(); }} // Tell Timer we’re done });} ready.await(); // Wait for all workers to be ready long startNanos = System.nanoTime(); start.countDown(); // And they’re off! done.await() // Wait for all workers to finish return System.nanoTime() – startNanos; }

  25. Item 70: Document Thread Safety • Levels of Thread safety • Immutable: • Instances of class appear constant • Example: String • Unconditionally thread-safe • Instances of class are mutable, but is internally synchronized • Example: ConcurrentHashMap • Conditionally thread-safe • Some methods require external synchronization • Example: Collections.synchronized wrappers • Not thread-safe • Client responsible for synchronization • Examples: Collection classes • Thread hostile: Not to be emulated!

  26. Item 71: Use Lazy Initialization Judiciously • Under most circumstances, normal initialization is preferred // Normal initialization of an instance field private final FieldType field = computeFieldValue(); // Lazy initialization of instance field – synchronized accessor private FieldType field; synchronized FieldType getField() { if (field == null) field = computeFieldValue(); return field; }

  27. More Item 71: Double Check Lazy Initialization // Double-check idiom for lazy initialization of instance fields private volatile FieldType field; // volatile key – see Item 66 FieldType getField() { FieldType result = field; if (result == null) { // check with no locking synchronized (this) { result = field; if (result == null) // Second check with a lock field = result = computeFieldValue(); } } return result; }

  28. Item 72: Don’t Depend on the Thread Scheduler • Any program that relies on the thread scheduler is likely to be unportable • Threads should not busy-wait • Use concurrency facilities instead (Item 69) • Don’t “Fix” slow code with Thread.yield calls • Restructure instead • Avoid Thread priorities

  29. Item 73: Avoid Thread Groups • Thread groups originally envisioned as a mechanism for isolating Applets for security purposes • Unfortunately, doesn’t really work

More Related