Listings xii Preface xvii Chapter 1: Introduction 1 1.1 A (very) brief history of concurrency 1 1.2 Benefits of threads 3 1.3 Risks of threads 5 1.4 Threads are everywhere 9 Part I: Fundamentals 13 Chapter 2: Thread Safety 15 2.1 What is thread safety? 17 2.2 Atomicity 19 2.3 Locking 23 2.4 Guarding state with locks 27 2.5 Liveness and performance 29 Chapter 3: Sharing Objects 33 3.1 Visibility 33 3.2 Publication and escape 39 3.3 Thread confinement 42 3.4 Immutability 46 3.5 Safepublication 49 Chapter 4: Composing Objects 55 4.1 Designing a thread-safe class 55 4.2 Instance confinement 58 4.3 Delegating thread safety 62 4.4 Adding functionality to existing thread-safe classes 71 4.5 Documenting synchronization policies 74 Chapter 5: Building Blocks 79 5.1 Synchronized collections 79 5.2 Concurrent collections 84 5.3 Blocking queues and the producer-consumer pattern 87 5.4 Blocking and interruptible methods 92 5.5 Synchronizers 94 5.6 Building an efficient, scalable result cache 101 Part II: Structuring Concurrent Applications 111 Chapter 6: Task Execution 113 6.1 Executing tasks in threads 113 6.2 The Executor framework 117 6.3 Finding exploitable parallelism 123 Chapter 7: Cancellation and Shutdown 135 7.1 Task cancellation 135 7.2 Stopping a thread-based service 150 7.3 Handling abnormal thread termination 161 7.4 JVM shutdown 164 Chapter 8: Applying Thread Pools 167 8.1 Implicit couplings between tasks and execution policies 167 8.2 Sizing thread pools 170 8.3 Configuring ThreadPoolExecutor 171 8.4 Extending ThreadPoolExecutor 179 8.5 Parallelizing recursive algorithms 181 Chapter 9: GUI Applications 189 9.1 Why are GUIs single-threaded? 189 9.2 Short-running GUI tasks 192 9.3 Long-running GUI tasks 195 9.4 Shared data models 198 9.5 Other forms of single-threaded subsystems 202 Part III: Liveness, Performance, and Testing 203 Chapter 10: Avoiding Liveness Hazards 205 10.1 Deadlock 205 10.2 Avoiding and diagnosing deadlocks 215 10.3 Other liveness hazards 218 Chapter 11: Performance and Scalability 221 11.1 Thinking about performance 221 11.2 Amdahl's law 225 11.3 Costs introduced by threads 229 11.4 Reducing lock contention 232 11.5 Example: Comparing Map performance 242 11.6 Reducing context switch overhead 243 Chapter 12: Testing Concurrent Programs 247 12.1 Testing for correctness 248 12.2 Testing for performance 260 12.3 Avoiding performance testing pitfalls 266 12.4 Complementary testing approaches 270 Part IV: Advanced Topics 275 Chapter 13: Explicit Locks 277 13.1 Lock and ReentrantLock 277 13.2 Performance considerations 282 13.3 Fairness 283 13.4 Choosing between synchronized and ReentrantLock 285 13.5 Read-write locks 286 Chapter 14: Building Custom Synchronizers 291 14.1 Managing state dependence 291 14.2 Using condition queues 298 14.3 Explicit condition objects 306 14.4 Anatomy of a synchronizer 308 14.5 AbstractQueuedSynchronizer 311 14.6 AQS in java.util.concurrent synchronizer classes 314 Chapter15: Atomic Variables and Nonblocking Synchronization 319 15.1 Disadvantages of locking 319 15.2 Hardware support for concurrency 321 15.3 Atomic variable classes 324 15.4 Nonblocking algorithms 329 Chapter 16: The Java Memory Model 337 16.1 What is a memory model, and why would I want one? 337 16.2 Publication 344 16.3 Initialization safety 349 Appendix A: Annotations for Concurrency 353 A.1 Class annotations 353 A.2 Field andmethod annotations 353 Bibliography 355 Index 359
As processors become faster and multiprocessor systems become cheaper, the need to take advantage of multithreading in order to achieve full hardware resource utilization only increases the importance of being able to incorporate concurrency in a wide variety of application categories. For many developers, concurrency remains a mystery. Developing, testing, and debugging multithreaded programs is extremely difficult because concurrency hazards do not manifest themselves uniformly or reliably. This book is intended to be neither an introduction to concurrency (any threading chapter in an "intro" book does that) nor is it an encyclopedic reference of All Things Concurrency (that would be Doug Lea's Concurrent Programming in Java). Instead, this title is a combination of concepts, guidelines, and examples intended to assist developers in the difficult process of understanding concurrency and its new tools in J2SE 5.0. Filled with contributions from Java gurus such as Josh Bloch, David Holmes, and Doug Lea, this book provides any Java programmers with the basic building blocks they need to gain a basic understanding of concurrency and its benefits.
Brian Goetz is a software consultant with twenty years industry experience, with over 75 articles on Java development. He is one of the primary members of the Java Community Process JSR 166 Expert Group (Concurrency Utilities), and has served on numerous other JCP Expert Groups. Tim Peierls is the very model of a modern multiprocessor, with BoxPop.biz, recording arts, and goings on theatrical. He is one of the primary members of the Java Community Process JSR 166 Expert Group (Concurrency Utilities), and has served on numerous other JCP Expert Groups. Joshua Bloch is a principal engineer at Google and a Jolt Award-winner. He was previously a distinguished engineer at Sun Microsystems and a senior systems designer at Transarc. Josh led the design and implementation of numerous Java platform features, including JDK 5.0 language enhancements and the award-winning Java Collections Framework. He holds a Ph.D. in computer science from Carnegie Mellon University. Joseph Bowbeer is a software architect at Vizrea Corporation where he specializes in mobile application development for the Java ME platform, but his fascination with concurrent programming began in his days at Apollo Computer. He served on the JCP Expert Group for JSR-166 (Concurrency Utilities). David Holmes is director of DLTeCH Pty Ltd, located in Brisbane, Australia. He specializes in synchronization and concurrency and was a member of the JSR-166 expert group that developed the new concurrency utilities. He is also a contributor to the update of the Real-Time Specification for Java, and has spent the past few years working on an implementation of that specification. Doug Lea is one of the foremost experts on object-oriented technology and software reuse. He has been doing collaborative research with Sun Labs for more than five years. Lea is Professor of Computer Science at SUNY Oswego, Co-director of the Software Engineering Lab at the New York Center for Advanced Technology in Computer Applications, and Adjunct Professor of Electrical and Computer Engineering at Syracuse University. In addition, he co-authored the book, Object-Oriented System Development (Addison-Wesley, 1993). He received his B.A., M.A., and Ph.D. from the University of New Hampshire.