p Compilers Principles, Techniques, <§ Tools Second Edition Second Edition Compilers Principles, Techniques, & Tools Second Edition Alfred V. Aho Columbia University Monica S. Lam Stanford University Ravi Sethi Avaya Jeffrey D. Ullman Stanford University Boston San Francisco New York London Toronto Sydney Tokyo Singapore Madrid Mexico City Munich Paris Cape Town Hong Kong Montreal Publisher Greg Tobin Executive Editor Michael Hirsch Acquisitions Editor Matt Goldstein Project Editor Katherine Harutunian Associate Managing Editor Jeffrey Holcomb Cover Designer Joyce Cosentino Wells Digital Assets Manager Marianne Groth Media Producer Bethany Tidd Senior Marketing Manager Michelle Brown Marketing Assistant Sarah Milmore Senior Author Support/ Technology Specialist Joe Vetere Senior Manufacturing Buyer Carol Melville Cover Image Scott Ullman of Strange Tonic Productions (www. strangetonic.com) Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. Where those designations appear in this book, and Addison-Wesley was aware of a trademark claim, the designations have been printed in initial caps or all caps. This interior of this book was composed in LATX. E Library of Congress Cataloging-in-Publication Data Compilers : principles, techniques, and tools / Alfred V. Aho ... [et al.]. ~ 2nd ed. p. cm. Rev. ed. of: Compilers, principles, techniques, and tools / Alfred V. Aho, Ravi Sethi, Jeffrey D. Ullman. 1986. ISBN 0-321-48681-1 (alk. paper) 1. Compilers (Computer programs) I. Aho, Alfred V. II. Aho, Alfred V. Compilers, principles, techniques, and tools. QA76.76.C65A37 2007 005.4'53~dc22 2006024333 Copyright © 2007 Pearson Education, Inc. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Printed in the United States of America. For information on obtaining permission for use of material in this work, please submit a written request to Pearson Education, Inc., Rights and Contracts Department, 75 Arlington Street, Suite 300, Boston, MA 02116, fax your request to 617-848-7047, or e-mail at http://www.pearsoned.com/legal/permissions.htm. 23456789 10—CW—10 09 08 07 06 Preface In the time since the 1986 edition of this book, the world of compiler design has changed significantly. Programming languages have evolved to present new compilation problems. Computer architectures offer a variety of resources of which the compiler designer must take advantage. Perhaps most interestingly, the venerable technology of code optimization has found use outside compilers. It is now used in tools that find bugs in software, and most importantly, find security holes in existing code. And much of the "front-end" technology — grammars, regular expressions, parsers, and syntax-directed translators — are still in wide use. Thus, our philosophy from previous versions of the book has not changed. We recognize that few readers will build, or even maintain, a compiler for a major programming language. Yet the models, theory, and algorithms associ ated with a compiler can be applied to a wide range of problems in software design and software development. We therefore emphasize problems that are most commonly encountered in designing a language processor, regardless of the source language or target machine. Use of the Book It takes at least two quarters or even two semesters to cover all or most of the material in this book. It is common to cover the first half in an undergraduate course and the second half of the book — stressing code optimization — in a second course at the graduate or mezzanine level. Here is an outline of the chapters: Chapter 1 contains motivational material and also presents some background issues in computer architecture and programming-language principles. Chapter 2 develops a miniature compiler and introduces many of the impor tant concepts, which are then developed in later chapters. The compiler itself appears in the appendix. Chapter 3 covers lexical analysis, regular expressions, finite-state machines, and scanner-generator tools. This material is fundamental to text-processing of all sorts. v vi PREFACE Chapter 4 covers the major parsing methods, top-down (recursive-descent, LL) and bottom-up (LR and its variants). Chapter 5 introduces the principal ideas in syntax-directed definitions and syntax-directed translations. Chapter 6 takes the theory of Chapter 5 and shows how to use it to generate intermediate code for a typical programming language. Chapter 7 covers run-time environments, especially management of the run-time stack and garbage collection. Chapter 8 is on object-code generation. It covers construction of basic blocks, generation of code from expressions and basic blocks, and register-allocation techniques. Chapter 9 introduces the technology of code optimization, including flow graphs, data-flow frameworks, and iterative algorithms for solving these frameworks. Chapter 10 covers instruction-level optimization. The emphasis is on the ex traction of parallelism from small sequences of instructions and scheduling them on single processors that can do more than one thing at once. Chapter 11 talks about larger-scale parallelism detection and exploitation. Here, the emphasis is on numeric codes that have many tight loops that range over multidimensional arrays. Chapter 12 is on interprocedural analysis. It covers pointer analysis, aliasing, and data-flow analysis that takes into account the sequence of procedure calls that reach a given point in the code. Courses from material in this book have been taught at Columbia, Harvard, and Stanford. At Columbia, a senior/first-year graduate course on program ming languages and translators has been regularly offered using material from the first eight chapters. A highlight of this course is a semester-long project in which students work in small teams to create and implement a little lan guage of their own design. The student-created languages have covered diverse application domains including quantum computation, music synthesis, com puter graphics, gaming, matrix operations and many other areas. Students use compiler-component generators such as ANTLR, Lex, and Yacc and the syntax- directed translation techniques discussed in chapters two and five to build their compilers. A follow-on graduate course has focused on material in Chapters 9 through 12, emphasizing code generation and optimization for contemporary machines including network processors and multiprocessor architectures. At Stanford, a one-quarter introductory course covers roughly the mate rial in Chapters 1 through 8, although there is an introduction to global code optimization from Chapter 9. The second compiler course covers Chapters 9 through 12, plus the more advanced material on garbage collection from Chap ter 7. Students use a locally developed, Java-based system called Joeq for implementing data-flow analysis algorithms. PREFACE vii Prerequisites The reader should possess some "computer-science sophistication," including at least a second course on programming, and courses in data structures and discrete mathematics. Knowledge of several different programming languages is useful. Exercises The book contains extensive exercises, with some for almost every section. We indicate harder exercises or parts of exercises with an exclamation point. The hardest exercises have a double exclamation point. Gradiance On-Line Homeworks A feature of the new edition is that there is an accompanying set of on-line homeworks using a technology developed by Gradiance Corp. Instructors may assign these homeworks to their class, or students not enrolled in a class may enroll in an "omnibus class" that allows them to do the homeworks as a tutorial (without an instructor-created class). Gradiance questions look like ordinary questions, but your solutions are sampled. If you make an incorrect choice you are given specific advice or feedback to help you correct your solution. If your instructor permits, you are allowed to try again, until you get a perfect score. A subscription to the Gradiance service is offered with all new copies of this text sold in North America. For more information, visit the Addison-Wesley web site www.aw.com/gradiance or send email to [email protected]. Support on the World Wide Web The book's home page is dragonbook.Stanford.edu Here, you will find errata as we learn of them, and backup materials. We hope to make available the notes for each offering of compiler-related courses as we teach them, including homeworks, solutions, and exams. We also plan to post descriptions of important compilers written by their implementers. Acknowledgements Cover art is by S. D. Ullman of Strange Tonic Productions. Jon Bentley gave us extensive comments on a number of chapters of an earlier draft of this book. Helpful comments and errata were received from: viii PREFACE Domenico Bianculli, Peter Bosch, Marcio Buss, Marc Eaddy, Stephen Edwards, Vibhav Garg, Kim Hazelwood, Gaurav Kc, Wei Li, Mike Smith, Art Stamness, Krysta Svore, Olivier Tardieu, and Jia Zeng. The help of all these people is gratefully acknowledged. Remaining errors are ours, of course. In addition, Monica would like to thank her colleagues on the SUIF com piler team for an 18-year lesson on compiling: Gerald Aigner, Dzintars Avots, Saman Amarasinghe, Jennifer Anderson, Michael Carbin, Gerald Cheong, Amer Diwan, Robert French, Anwar Ghuloum, Mary Hall, John Hennessy, David Heine, Shih-Wei Liao, Amy Lim, Benjamin Livshits, Michael Martin, Dror Maydan, Todd Mowry, Brian Murphy, Jeffrey Oplinger, Karen Pieper, Mar tin Rinard, Olatunji Ruwase, Constantine Sapuntzakis, Patrick Sathyanathan, Michael Smith, Steven Tjiang, Chau-Wen Tseng, Christopher Unkel, John Whaley, Robert Wilson, Christopher Wilson, and Michael Wolf. A. V. A., Chatham NJ M. S. L., Menlo Park CA R. S., Far Hills NJ J. D. U., Stanford CA June, 2006 Table of Contents 1 Introduction 1 1.1 Language Processors 1 1.1.1 Exercises for Section 1.1 3 1.2 The Structure of a Compiler 4 1.2.1 Lexical Analysis 5 1.2.2 Syntax Analysis 8 1.2.3 Semantic Analysis 8 1.2.4 Intermediate Code Generation 9 1.2.5 Code Optimization 10 1.2.6 Code Generation 10 1.2.7 Symbol-Table Management 11 1.2.8 The Grouping of Phases into Passes 11 1.2.9 Compiler-Construction Tools 12 1.3 The Evolution of Programming Languages 12 1.3.1 The Move to Higher-level Languages 13 1.3.2 Impacts on Compilers 14 1.3.3 Exercises for Section 1.3 14 1.4 The Science of Building a Compiler 15 1.4.1 Modeling in Compiler Design and Implementation . . .. 15 1.4.2 The Science of Code Optimization 15 1.5 Applications of Compiler Technology 17 1.5.1 Implementation of High-Level Programming Languages . 17 1.5.2 Optimizations for Computer Architectures 19 1.5.3 Design of New Computer Architectures 21 1.5.4 Program Translations 22 1.5.5 Software Productivity Tools 23 1.6 Programming Language Basics 25 1.6.1 The Static/Dynamic Distinction 25 1.6.2 Environments and States 26 1.6.3 Static Scope and Block Structure 28 1.6.4 Explicit Access Control 31 1.6.5 Dynamic Scope 31 1.6.6 Parameter Passing Mechanisms 33 ix TABLE OF CONTENTS 1.6.7 Aliasing 35 1.6.8 Exercises for Section 1.6 35 1.7 Summary of Chapter 1 36 1.8 References for Chapter 1 38 A Simple Syntax-Directed Translator 39 2.1 Introduction 40 2.2 Syntax Definition 42 2.2.1 Definition of Grammars 42 2.2.2 Derivations 44 2.2.3 Parse Trees 45 2.2.4 Ambiguity 47 2.2.5 Associativity of Operators 48 2.2.6 Precedence of Operators 48 2.2.7 Exercises for Section 2.2 51 2.3 Syntax-Directed Translation . 52 2.3.1 Postfix Notation . 53 2.3.2 Synthesized Attributes 54 2.3.3 Simple Syntax-Directed Definitions 56 2.3.4 Tree Traversals 56 2.3.5 Translation Schemes 57 2.3.6 Exercises for Section 2.3 60 2.4 Parsing 60 2.4.1 Top-Down Parsing 61 2.4.2 Predictive Parsing 64 2.4.3 When to Use e-Productions 65 2.4.4 Designing a Predictive Parser 66 2.4.5 Left Recursion 67 2.4.6 Exercises for Section 2.4 68 2.5 A Translator for Simple Expressions 68 2.5.1 Abstract and Concrete Syntax 69 2.5.2 Adapting the Translation Scheme 70 2.5.3 Procedures for the Nonterminals 72 2.5.4 Simplifying the Translator 73 2.5.5 The Complete Program 74 2.6 Lexical Analysis 76 2.6.1 Removal of White Space and Comments 77 2.6.2 Reading Ahead 78 2.6.3 Constants 78 2.6.4 Recognizing Keywords and Identifiers 79 2.6.5 A Lexical Analyzer 81 2.6.6 Exercises for Section 2.6 84 2.7 Symbol Tables 85 2.7.1 Symbol Table Per Scope 86 2.7.2 The Use of Symbol Tables 89 TABLE OF CONTENTS xi 2.8 Intermediate Code Generation 91 2.8.1 Two Kinds of Intermediate Representations 91 2.8.2 Construction of Syntax Trees 92 2.8.3 Static Checking 97 2.8.4 Three-Address Code 99 2.8.5 Exercises for Section 2.8 105 2.9 Summary of Chapter 2 105 3 Lexical Analysis 3.1 The Role of the Lexical Analyzer 109 3.1.1 Lexical Analysis Versus Parsing 110 3.1.2 Tokens, Patterns, and Lexemes Ill 3.1.3 Attributes for Tokens 112 3.1.4 Lexical Errors 113 3.1.5 Exercises for Section 3.1 114 3.2 Input Buffering 115 3.2.1 Buffer Pairs 115 3.2.2 Sentinels 116 3.3 Specification of Tokens 116 3.3.1 Strings and Languages 117 3.3.2 Operations on Languages 119 3.3.3 Regular Expressions 120 3.3.4 Regular Definitions 123 3.3.5 Extensions of Regular Expressions 124 3.3.6 Exercises for Section 3.3 125 3.4 Recognition of Tokens 128 3.4.1 Transition Diagrams 130 3.4.2 Recognition of Reserved Words and Identifiers 132 3.4.3 Completion of the Running Example 133 3.4.4 Architecture of a Transition-Diagram-Based Lexical An alyzer 134 3.4.5 Exercises for Section 3.4 136 3.5 The Lexical-Analyzer Generator Lex 140 3.5.1 Use of Lex 140 3.5.2 Structure of Lex Programs 141 3.5.3 Conflict Resolution in Lex 144 3.5.4 The Lookahead Operator 144 3.5.5 Exercises for Section 3.5 146 3.6 Finite Automata 147 3.6.1 Nondeterministic Finite Automata 147 3.6.2 Transition Tables 148 3.6.3 Acceptance of Input Strings by Automata 149 3.6.4 Deterministic Finite Automata 149 3.6.5 Exercises for Section 3.6 151 3.7 From Regular Expressions to Automata 152