DOCSLIB.ORG

Partial Escape Analysis and Scalar Replacement for Java

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Partial Escape Analysis and Scalar Replacement for Java Partial Escape Analysis and Scalar Replacement for Java Lukas Stadler Thomas Würthinger Hanspeter Mössenböck Johannes Kepler University Oracle Labs Johannes Kepler University Linz, Austria thomas.wuerthinger Linz, Austria [email protected] @oracle.com [email protected] ABSTRACT 1. INTRODUCTION Escape Analysis allows a compiler to determine whether an State-of-the-art Virtual Machines employ techniques such object is accessible outside the allocating method or thread. as advanced garbage collection, alias analysis and biased This information is used to perform optimizations such as locking to make working with dynamically allocated objects Scalar Replacement, Stack Allocation and Lock Elision, al- as efficient as possible. But even if allocation is cheap, it lowing modern dynamic compilers to remove some of the still incurs some overhead. Even if alias analysis can remove abstractions introduced by advanced programming models. most object accesses, some of them cannot be removed. And The all-or-nothing approach taken by most Escape Anal- although acquiring a biased lock is simple, it is still more ysis algorithms prevents all these optimizations as soon as complex than not acquiring a lock at all. there is one branch where the object escapes, no matter how Escape Analysis can be used to determine whether an ob- unlikely this branch is at runtime. ject needs to be allocated at all, and whether its lock can This paper presents a new, practical algorithm that per- ever be contended. This can help the compiler to get rid of forms control flow sensitive Partial Escape Analysis in a dy- the object's allocation, using Scalar Replacement to replace namic Java compiler. It allows Escape Analysis, Scalar Re- the object's fields with local variables. placement and Lock Elision to be performed on individual Escape Analysis checks whether an object escapes its al- branches. We implemented the algorithm on top of Graal, locating method, i.e., whether it is accessible outside this an open-source Java just-in-time compiler, and it performs method. An object escapes, for example, if it is assigned well on a diverse set of benchmarks. to a static field, if it is passed as an argument to another In this paper, we evaluate the effect of Partial Escape method, or if it is returned by a method. In these cases the Analysis on the DaCapo, ScalaDaCapo and SpecJBB2005 object needs to exist on the heap, because it will be accessed benchmarks, in terms of run-time, number and size of al- as an object in some other context. locations and number of monitor operations. It performs In many cases, however, an object escapes just in a single particularly well in situations with additional levels of ab- unlikely branch. Nevertheless, this prevents optimizations. straction, such as code generated by the Scala compiler. It Therefore, we suggest a flow-sensitive Escape Analysis which reduces the amount of allocated memory by up to 58.5%, we call Partial Escape Analysis. and improves performance by up to 33%. The idea behind Partial Escape Analysis is to perform op- timizations such as Scalar Replacement in branches where Categories and Subject Descriptors the object does not escape, and make sure that the object exists in the heap in branches where it does escape. Addi- D.3.4 [Programming Languages]: Processors|Compil- tionally, our Partial Escape Analysis works not on bytecodes ers, Optimization but on the compiler's intermediate representation, so that it can be applied, possibly multiple times, at any point during General Terms compilation. This paper contributes the following novel aspects: Algorithms, Languages, Performance • A control-flow-sensitive Partial Escape Analysis algo- Keywords rithm that checks the escapability of objects for indi- escape analysis, java, virtual machine, just-in-time compila- vidual branches. tion, intermediate representation, speculative optimization • The integration of our Partial Escape Analysis in a Permission to make digital or hard copies of all or part of this work for Java compiler based on SSA form, speculative opti- personal or classroom use is granted without fee provided that copies are not mization, and deoptimization. made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components • An evaluation of this algorithm on a set of current of this work owned by others than ACM must be honored. Abstracting with benchmarks, in terms of run-time, number and size of credit is permitted. To copy otherwise, or republish, to post on servers or to allocations and number of monitor operations, show- redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]. ing that our algorithm performs well on a variety of CGO ’14 February 15 - 19 2014, Orlando, FL, USA benchmarks. Copyright 2014 ACM 978-1-4503-2670-4/14/02 ...$15.00. Graal 1 c l a s s Key f Compiler Java 2 i n t i d x ; 3 Object r e f ; C++ Graal API Client Server 4 Key ( i n t idx, Object ref) f Compiler Compiler 5 t h i s . i d x = i d x ; 6 t h i s . r e f = r e f ; Interpreter GC 7 g 8 synchronized boolean e q u a l s ( Key Class Loading . o t h e r ) f HotSpot VM 9 r e t u r n idx == other.idx && 10 ref == other.ref; 11 g Figure 1: Overview of HotSpot and Graal. 12 g 13 s t a t i c CacheKey cacheKey; 14 s t a t i c Object cacheValue; 15 16 Object getValue( i n t idx, Object ref) f 2. SYSTEM OVERVIEW 17 Key key = new Key(idx, ref); 18 i f (key. equals(cacheKey)) f We implemented our analysis for Graal, which is a Java 19 r e t u r n cacheValue; just-in-time compiler written in Java that runs on top of 20 g e l s e f the HotSpot VM (Figure 1). While it can completely re- 21 r e t u r n createValue (...) ; place the client and server compilers, it reuses all other VM 22 g components, such as the interpreter, the garbage collection 23 g subsystem and class loading. It is open-source and available via the OpenJDK Graal project [11]. Listing 1: Simple example. Graal translates Java bytecode into a high-level intermedi- ate representation called Graal IR [5], on which it performs all optimizations. This SSA-based intermediate represen- have side effects; they can be reexecuted and will lead to the tation models both control flow and data flow dependencies same result. between nodes. While it explicitly expresses the control flow For example, a field store is associated with a frame state of the code, many operations are not fixed at specific loca- that describes the state after the store. A machine code tions. Rather, the position at which these operations are instruction that causes or can cause a deoptimization will executed is determined solely by their data flow dependen- be associated with the after state of the last side-effecting cies. There are usually many positions where an operation instruction. All instructions that were executed in-between could be placed, and it is up to the so-called Scheduler to do not have side effects and can therefore be reexecuted in determine a good one. the interpreter. Graal is an aggressive and optimistic compiler that often makes assumptions about the state and behavior of the run- ning application. This includes assumptions such as some 3. ESCAPE ANALYSIS classes not having subclasses, and some branches never being Escape Analysis checks whether an allocated object es- taken. When one of these assumptions is invalidated, e.g., capes (i.e., can be used outside) the allocating method or by loading a new class or by entering an unexpected branch, thread. This happens, for example, if it is assigned to a the execution needs to be transferred from compiled code global variable or heap object, or if it is passed as a param- back to the interpreter (which does not make any assump- eter to some other method. Compilers use Escape Analysis tions and can execute all code). This switch back to the to determine the dynamic scope and the lifetime of allo- interpreter is called Deoptimization [7] and requires a trans- cated objects. The result of this analysis allows the com- lation from the machine state (native stack frames) back to piler to perform numerous optimizations on operations such the Java VM state, which is used by the interpreter. as object allocations, synchronization primitives and field The Graal IR contains a mapping to Java VM state for accesses. all positions that can cause a deoptimization to occur. This Listing 1 shows a small piece of code that will serve as an mapping is expressed as FrameState nodes and consists of example to show the benefits of Escape Analysis: The get- the current position (bytecode index and method), the local Value method creates a new Key object and checks whether variables, the contents of the expression stack and the locked it is in the cache. If so, the method returns the cached value. objects. Otherwise, it creates and returns a new value (the method After inlining, one position can map to multiple Java VM createValue is not discussed here). stack frames. A frame state thus contains a reference to an When getValue is compiled, the compiler will most likely outer frame state, which is the caller's state. This reference perform some inlining, which might cause the actually com- is used to create chains of FrameState nodes that describe piled code to look like Listing 2.
Load more

Recommended publications
  • Incrementalized Pointer and Escape Analysis Frédéric Vivien, Martin Rinard
    Incrementalized Pointer and Escape Analysis Frédéric Vivien, Martin Rinard To cite this version: Frédéric Vivien, Martin Rinard. Incrementalized Pointer and Escape Analysis. PLDI ’01 Proceedings of the ACM SIGPLAN 2001 conference on Programming language design and implementation, Jun 2001, Snowbird, United States. pp.35–46, 10.1145/381694.378804. hal-00808284 HAL Id: hal-00808284 https://hal.inria.fr/hal-00808284 Submitted on 14 Oct 2018 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Incrementalized Pointer and Escape Analysis∗ Fred´ eric´ Vivien Martin Rinard ICPS/LSIIT Laboratory for Computer Science Universite´ Louis Pasteur Massachusetts Institute of Technology Strasbourg, France Cambridge, MA 02139 [email protected]­strasbg.fr [email protected] ABSTRACT to those parts of the program that offer the most attrac- We present a new pointer and escape analysis. Instead of tive return (in terms of optimization opportunities) on the analyzing the whole program, the algorithm incrementally invested resources. Our experimental results indicate that analyzes only those parts of the program that may deliver this approach usually delivers almost all of the benefit of the useful results. An analysis policy monitors the analysis re- whole-program analysis, but at a fraction of the cost.
    [Show full text]
  • Generalized Escape Analysis and Lightweight Continuations
    Generalized Escape Analysis and Lightweight Continuations Abstract Might and Shivers have published two distinct analyses for rea- ∆CFA and ΓCFA are two analysis techniques for higher-order soning about programs written in higher-order languages that pro- functional languages (Might and Shivers 2006b,a). The former uses vide first-class continuations, ∆CFA and ΓCFA (Might and Shivers an enrichment of Harrison’s procedure strings (Harrison 1989) to 2006a,b). ∆CFA is an abstract interpretation based on a variation reason about control-, environment- and data-flow in a program of procedure strings as developed by Sharir and Pnueli (Sharir and represented in CPS form; the latter tracks the degree of approxi- Pnueli 1981), and, most proximately, Harrison (Harrison 1989). mation inflicted by a flow analysis on environment structure, per- Might and Shivers have applied ∆CFA to problems that require mitting the analysis to exploit extra precision implicit in the flow reasoning about the environment structure relating two code points lattice, which improves not only the precision of the analysis, but in a program. ΓCFA is a general tool for “sharpening” the precision often its run-time as well. of an abstract interpretation by tracking the amount of abstraction In this paper, we integrate these two mechanisms, showing how the analysis has introduced for the specific abstract interpretation ΓCFA’s abstract counting and collection yields extra precision in being performed. This permits the analysis to opportunistically ex- ∆CFA’s frame-string model, exploiting the group-theoretic struc- ploit cases where the abstraction has introduced no true approxi- ture of frame strings. mation.
    [Show full text]
  • Escape from Escape Analysis of Golang
    Escape from Escape Analysis of Golang Cong Wang Mingrui Zhang Yu Jiang∗ KLISS, BNRist, School of Software, KLISS, BNRist, School of Software, KLISS, BNRist, School of Software, Tsinghua University Tsinghua University Tsinghua University Beijing, China Beijing, China Beijing, China Huafeng Zhang Zhenchang Xing Ming Gu Compiler and Programming College of Engineering and Computer KLISS, BNRist, School of Software, Language Lab, Huawei Technologies Science, ANU Tsinghua University Hangzhou, China Canberra, Australia Beijing, China ABSTRACT KEYWORDS Escape analysis is widely used to determine the scope of variables, escape analysis, memory optimization, code generation, go pro- and is an effective way to optimize memory usage. However, the gramming language escape analysis algorithm can hardly reach 100% accurate, mistakes ACM Reference Format: of which can lead to a waste of heap memory. It is challenging to Cong Wang, Mingrui Zhang, Yu Jiang, Huafeng Zhang, Zhenchang Xing, ensure the correctness of programs for memory optimization. and Ming Gu. 2020. Escape from Escape Analysis of Golang. In Software In this paper, we propose an escape analysis optimization ap- Engineering in Practice (ICSE-SEIP ’20), May 23–29, 2020, Seoul, Republic of proach for Go programming language (Golang), aiming to save Korea. ACM, New York, NY, USA, 10 pages. https://doi.org/10.1145/3377813. heap memory usage of programs. First, we compile the source code 3381368 to capture information of escaped variables. Then, we change the code so that some of these variables can bypass Golang’s escape 1 INTRODUCTION analysis mechanism, thereby saving heap memory usage and reduc- Memory management is very important for software engineering.
    [Show full text]
  • Escape Analysis for Java
    Escape Analysis for Java Jong-Deok Choi Manish Gupta Mauricio Serrano Vugranam C. Sreedhar Sam Midkiff IBM T. J. Watson Research Center P. O. Box 218, Yorktown Heights, NY 10598 jdchoi, mgupta, mserrano, vugranam, smidkiff ¡ @us.ibm.com ¢¤£¦¥¨§ © ¦ § § © § This paper presents a simple and efficient data flow algorithm Java continues to gain importance as a language for general- for escape analysis of objects in Java programs to determine purpose computing and for server applications. Performance (i) if an object can be allocated on the stack; (ii) if an object is an important issue in these application environments. In is accessed only by a single thread during its lifetime, so that Java, each object is allocated on the heap and can be deallo- synchronization operations on that object can be removed. We cated only by garbage collection. Each object has a lock asso- introduce a new program abstraction for escape analysis, the ciated with it, which is used to ensure mutual exclusion when connection graph, that is used to establish reachability rela- a synchronized method or statement is invoked on the object. tionships between objects and object references. We show that Both heap allocation and synchronization on locks incur per- the connection graph can be summarized for each method such formance overhead. In this paper, we present escape analy- that the same summary information may be used effectively in sis in the context of Java for determining whether an object (1) different calling contexts. We present an interprocedural al- may escape the method (i.e., is not local to the method) that gorithm that uses the above property to efficiently compute created the object, and (2) may escape the thread that created the connection graph and identify the non-escaping objects for the object (i.e., other threads may access the object).
    [Show full text]
  • Making Collection Operations Optimal with Aggressive JIT Compilation
    Making Collection Operations Optimal with Aggressive JIT Compilation Aleksandar Prokopec David Leopoldseder Oracle Labs Johannes Kepler University, Linz Switzerland Austria [email protected] [email protected] Gilles Duboscq Thomas Würthinger Oracle Labs Oracle Labs Switzerland Switzerland [email protected] [email protected] Abstract 1 Introduction Functional collection combinators are a neat and widely ac- Compilers for most high-level languages, such as Scala, trans- cepted data processing abstraction. However, their generic form the source program into an intermediate representa- nature results in high abstraction overheads – Scala collec- tion. This intermediate program representation is executed tions are known to be notoriously slow for typical tasks. We by the runtime system. Typically, the runtime contains a show that proper optimizations in a JIT compiler can widely just-in-time (JIT) compiler, which compiles the intermediate eliminate overheads imposed by these abstractions. Using representation into machine code. Although JIT compilers the open-source Graal JIT compiler, we achieve speedups of focus on optimizations, there are generally no guarantees up to 20× on collection workloads compared to the standard about the program patterns that result in fast machine code HotSpot C2 compiler. Consequently, a sufficiently aggressive – most optimizations emerge organically, as a response to JIT compiler allows the language compiler, such as Scalac, a particular programming paradigm in the high-level lan- to focus on other concerns. guage. Scala’s functional-style collection combinators are a In this paper, we show how optimizations, such as inlin- paradigm that the JVM runtime was unprepared for. ing, polymorphic inlining, and partial escape analysis, are Historically, Scala collections [Odersky and Moors 2009; combined in Graal to produce collections code that is optimal Prokopec et al.
    [Show full text]
  • Escape Analysis on Lists∗
    Escape Analysis on Lists∗ Young Gil Parkyand Benjamin Goldberg Department of Computer Science Courant Institute of Mathematical Sciences New York Universityz Abstract overhead, and other optimizations that are possible when the lifetimes of objects can be computed stati- Higher order functional programs constantly allocate cally. objects dynamically. These objects are typically cons Our approach is to define a high-level non-standard cells, closures, and records and are generally allocated semantics that, in many ways, is similar to the stan- in the heap and reclaimed later by some garbage col- dard semantics and captures the escape behavior caused lection process. This paper describes a compile time by the constructs in a functional language. The ad- analysis, called escape analysis, for determining the vantage of our analysis lies in its conceptual simplicity lifetime of dynamically created objects in higher or- and portability (i.e. no assumption is made about an der functional programs, and describes optimizations underlying abstract machine). that can be performed, based on the analysis, to im- prove storage allocation and reclamation of such ob- jects. In particular, our analysis can be applied to 1 Introduction programs manipulating lists, in which case optimiza- tions can be performed to allow whole cons cells in Higher order functional programs constantly allocate spines of lists to be either reclaimed at once or reused objects dynamically. These objects are typically cons without incurring any garbage collection overhead. In cells, closures, and records and are generally allocated a previous paper on escape analysis [10], we had left in the heap and reclaimed later by some garbage col- open the problem of performing escape analysis on lection process.
    [Show full text]
  • Compiling Haskell by Program Transformation
    Compiling Haskell by program transformation a rep ort from the trenches Simon L Peyton Jones Department of Computing Science University of Glasgow G QQ Email simonpjdcsglaacuk WWW httpwwwdcsglaacuksimonpj January Abstract Many compilers do some of their work by means of correctnesspre se rving and hop e fully p erformanceimproving program transformations The Glasgow Haskell Compiler GHC takes this idea of compilation by transformation as its warcry trying to express as much as p ossible of the compilation pro cess in the form of program transformations This pap er rep orts on our practical exp erience of the transformational approach to compilation in the context of a substantial compiler The paper appears in the Proceedings of the European Symposium on Programming Linkoping April Introduction Using correctnesspreserving transformations as a compiler optimisation is a wellestablished technique Aho Sethi Ullman Bacon Graham Sharp In the functional programming area esp ecially the idea of compilation by transformation has received quite a bit of attention App el Fradet Metayer Kelsey Kelsey Hudak Kranz Steele A transformational approach to compiler construction is attractive for two reasons Each transformation can b e implemented veried and tested separately This leads to a more mo dular compiler design in contrast to compilers that consist of a few huge passes each of which accomplishes a great deal In any framework transformational or otherwise each optimisation often exp oses new opp ortunities for other optimisations
    [Show full text]
  • Current State of EA and Its Uses in The
    Jfokus 2020 Current state of EA and Charlie Gracie Java Engineering Group at Microsoft its uses in the JVM Overview • Escape Analysis and Consuming Optimizations • Current State of Escape Analysis in JVM JITs • Escape Analysis and Allocation Elimination in Practice • Stack allocation 2 Escape Analysis and Consuming Optimizations 3 What is Escape Analysis? • Escape Analysis is a method for determining the dynamic scope of objects -- where in the program an object can be accessed. • Escape Analysis determines all the places where an object can be stored and whether the lifetime of the object can be proven to be restricted only to the current method and/or thread. 4 https://en.wikipedia.org/wiki/Escape_analysis Partial Escape Analysis • A variant of Escape Analysis which tracks object lifetime along different control flow paths of a method. • An object can be marked as not escaping along one path even though it escapes along a different path. 5 https://en.wikipedia.org/wiki/Escape_analysis EA Consuming Optimizations 1. Monitor elision • If an object does not escape the current method or thread, then operations can be performed on this object without synchronization 2. Stack allocation • If an object does not escape the current method, it may be allocated in stack memory instead of heap memory 3. Scalar replacement • Improvement to (2) by breaking an object up into its scalar parts which are just stored as locals 6 Current State of Escape Analysis in JVM JITs 7 HotSpot C2 EA and optimizations • Flow-insensitive1 implementation based on the
    [Show full text]
  • Evaluating the Impact of Thread Escape Analysis on a Memory Consistency Model-Aware Compiler
    Evaluating the Impact of Thread Escape Analysis on a Memory Consistency Model-aware Compiler Chi-Leung Wong1 Zehra Sura2 Xing Fang3 Kyungwoo Lee3 Samuel P. Midki®3 Jaejin Lee4 David Padua1 Dept. of Computer Science, University of Illinois at Urbana-Champaign, USA fcwong1,[email protected] IBM Thomas J. Watson Research Center, Yorktown Heights, NY, USA [email protected] Purdue University, West Lafayette, IN, USA fxfang,kwlee,[email protected] Seoul National University, Seoul, Korea [email protected] Abstract. The widespread popularity of languages allowing explicitly parallel, multi-threaded programming, e.g. Java and C#, have focused attention on the issue of memory model design. The Pensieve Project is building a compiler that will enable both language designers to pro- totype di®erent memory models, and optimizing compilers to adapt to di®erent memory models. Among the key analyses required to implement this system are thread escape analysis, i.e. detecting when a referenced object is accessible by more than one thread, delay set analysis, and synchronization analysis. In this paper, we evaluate the impact of di®erent escape analysis algo- rithms on the e®ectiveness of the Pensieve system when both delay set analysis and synchronization analysis are used. Since both analyses make use of results of escape analyses, their precison and cost is dependent on the precision of the escape analysis algorithm. It is the goal of this paper to provide a quantitative evalution of this impact. 1 Introduction In shared memory parallel programs, di®erent threads of the program commu- nicate with each other by reading from and writing to shared memory locations.
    [Show full text]
  • Hotspot VM Graal VM
    Polyglot on the JVM with Graal Thomas Wuerthinger Senior Research Director, Oracle Labs @thomaswue Java User Group Zurich, 15 of December 2016 Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | Safe Harbor Statement The following is intended to provide some insight into a line of research in Oracle Labs. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described in connection with any Oracle product or service remains at the sole discretion of Oracle. Any views expressed in this presentation are my own and do not necessarily reflect the views of Oracle. Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 3 One language to rule them all? Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | One Language to Rule Them All? Let’s ask Stack Overflow… Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | The World Is Polyglot better is Lower 3 Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 6 Graal Overview A new compiler for HotSpot written in Java and with a focus on speculative optimiZations. JVMCI and Graal includedin JDK9, modified version of JDK8 available via OTN. HotSpot VM Graal VM Client Server Client Graal Compiler Compiler Interface JVMCI Java Interface HotSpot HotSpot C++ Copyright © 2016, Oracle
    [Show full text]
  • Pointer Analysis for Source-To-Source Transformations
    Pointer Analysis for Source-to-Source Transformations Marcio Buss∗ Stephen A. Edwards† Bin Yao Daniel Waddington Department of Computer Science Network Platforms Research Group Columbia University Bell Laboratories, Lucent Technologies New York, NY 10027 Holmdel, NJ 07733 {marcio,sedwards}@cs.columbia.edu {byao,dwaddington}@lucent.com Abstract p=&x; p=&y; q=&z; p=q; x=&a; y=&b; z=&c; We present a pointer analysis algorithm designed for Andersen [1] Steensgaard [15] Das [5] Heintze [8] source-to-source transformations. Existing techniques for p q p q p q p q pointer analysis apply a collection of inference rules to a dismantled intermediate form of the source program, mak- x y z x,y,z x,y z x y z ing them difficult to apply to source-to-source tools that generally work on abstract syntax trees to preserve details a b c a,b,c a,b,c a b c of the source program. Our pointer analysis algorithm operates directly on the Figure 1. Results of various flow-insensitive abstract syntax tree of a C program and uses a form of stan- pointer analysis algorithms. dard dataflow analysis to compute the desired points-to in- formation. We have implemented our algorithm in a source- to-source translation framework and experimental results dismantling the program into increasingly lower-level rep- show that it is practical on real-world examples. resentations that deliberately discard most of the original structure of the source code to simplify its analysis. 1 Introduction By contrast, source-to-source techniques strive to pre- serve everything about the structure of the original source The role of pointer analysis in understanding C programs so that only minimal, necessary changes are made.
    [Show full text]
  • Compiler Construction
    Compiler construction PDF generated using the open source mwlib toolkit. See http://code.pediapress.com/ for more information. PDF generated at: Sat, 10 Dec 2011 02:23:02 UTC Contents Articles Introduction 1 Compiler construction 1 Compiler 2 Interpreter 10 History of compiler writing 14 Lexical analysis 22 Lexical analysis 22 Regular expression 26 Regular expression examples 37 Finite-state machine 41 Preprocessor 51 Syntactic analysis 54 Parsing 54 Lookahead 58 Symbol table 61 Abstract syntax 63 Abstract syntax tree 64 Context-free grammar 65 Terminal and nonterminal symbols 77 Left recursion 79 Backus–Naur Form 83 Extended Backus–Naur Form 86 TBNF 91 Top-down parsing 91 Recursive descent parser 93 Tail recursive parser 98 Parsing expression grammar 100 LL parser 106 LR parser 114 Parsing table 123 Simple LR parser 125 Canonical LR parser 127 GLR parser 129 LALR parser 130 Recursive ascent parser 133 Parser combinator 140 Bottom-up parsing 143 Chomsky normal form 148 CYK algorithm 150 Simple precedence grammar 153 Simple precedence parser 154 Operator-precedence grammar 156 Operator-precedence parser 159 Shunting-yard algorithm 163 Chart parser 173 Earley parser 174 The lexer hack 178 Scannerless parsing 180 Semantic analysis 182 Attribute grammar 182 L-attributed grammar 184 LR-attributed grammar 185 S-attributed grammar 185 ECLR-attributed grammar 186 Intermediate language 186 Control flow graph 188 Basic block 190 Call graph 192 Data-flow analysis 195 Use-define chain 201 Live variable analysis 204 Reaching definition 206 Three address
    [Show full text]