Difference between revisions of "Compiler Construction With Embedded Xinu"
(→Potential Course Structure: added a section about monitors) |
(→Potential Course Structure: added link to grammar) |
||
Line 23: | Line 23: | ||
== Potential Course Structure == | == Potential Course Structure == | ||
− | The course outlined below describes a compiler construction course focusing on a semester long project in which students build most of the pieces of a complete working compiler. For this example course we take the compiler project from Appel and Palsberg's [http://www.cambridge.org/us/catalogue/catalogue.asp?isbn=052182060x Modern Compiler Implementation in Java] [[Compiler Construction With Embedded Xinu#References|[2]]] and modify it to target a MIPS platform running the Xinu operating system. We take advantage of the fact that our compiler targets a runtime with an operating system and add high-level I/O and concurrency features to Appel and Palsberg's [http://www.cambridge.org/resources/052182060X/ MiniJava] language, creating our own [ | + | The course outlined below describes a compiler construction course focusing on a semester long project in which students build most of the pieces of a complete working compiler. For this example course we take the compiler project from Appel and Palsberg's [http://www.cambridge.org/us/catalogue/catalogue.asp?isbn=052182060x Modern Compiler Implementation in Java] [[Compiler Construction With Embedded Xinu#References|[2]]] and modify it to target a MIPS platform running the Xinu operating system. We take advantage of the fact that our compiler targets a runtime with an operating system and add high-level I/O and concurrency features to Appel and Palsberg's [http://www.cambridge.org/resources/052182060X/ MiniJava] language, creating our own [http://www.mscs.mu.edu/~brylow/cosc4400/Spring2011/ConcurrentMiniJava.html Concurrent MiniJava Grammar] language. |
We allow Java-like threading and synchronization with our added support for class declarations inheriting the built in ''Thread'' class and with added support for Java's ''synchronized'' keyword. We also add external operating system calls for I/O operations and for operations to create and manipulate multiple threads of execution. Specifically, we add the ability to print strings with ''Xinu.print(String s)'', the ability to print a new line with ''Xinu.println()'', the ability to print an integer with ''Xinu.printint(int x)'', the ability to read in an integer input with ''Xinu.readint()'', the ability to create a thread of execution with ''Xinu.threadCreate(Thread t)'', the ability for a thread to yield control of the processor with ''Xinu.yield()'', and the ability for a thread to sleep for a given number of milliseconds with ''Xinu.sleep(int time)''. | We allow Java-like threading and synchronization with our added support for class declarations inheriting the built in ''Thread'' class and with added support for Java's ''synchronized'' keyword. We also add external operating system calls for I/O operations and for operations to create and manipulate multiple threads of execution. Specifically, we add the ability to print strings with ''Xinu.print(String s)'', the ability to print a new line with ''Xinu.println()'', the ability to print an integer with ''Xinu.printint(int x)'', the ability to read in an integer input with ''Xinu.readint()'', the ability to create a thread of execution with ''Xinu.threadCreate(Thread t)'', the ability for a thread to yield control of the processor with ''Xinu.yield()'', and the ability for a thread to sleep for a given number of milliseconds with ''Xinu.sleep(int time)''. | ||
Line 29: | Line 29: | ||
The links in the outline below describe the changes necessary in each assignment to add these high-level I/O and concurrency features to the language, including the modifications for targeting a Xinu backend instead of the book's intended MIPS simulator. In addition to these compiler changes, modifications must also be made to Xinu to offer the runtime support required by the ''synchronized'' keyword. Since Java's ''synchronized'' feature depends on the JVM monitor system, which has subtly different semantics from standard O/S semaphores, [[Adding Monitors To Xinu|monitor constucts]] must be added to Xinu. | The links in the outline below describe the changes necessary in each assignment to add these high-level I/O and concurrency features to the language, including the modifications for targeting a Xinu backend instead of the book's intended MIPS simulator. In addition to these compiler changes, modifications must also be made to Xinu to offer the runtime support required by the ''synchronized'' keyword. Since Java's ''synchronized'' feature depends on the JVM monitor system, which has subtly different semantics from standard O/S semaphores, [[Adding Monitors To Xinu|monitor constucts]] must be added to Xinu. | ||
− | Appel and Palsberg's [http://www.cambridge.org/resources/052182060X/ MiniJava] language is a subset of the standard Java language, and this means test cases written in MiniJava can be compiled and run using standard Java compilers. To use standard Java compilers to compile programs written in our [ | + | Appel and Palsberg's [http://www.cambridge.org/resources/052182060X/ MiniJava] language is a subset of the standard Java language, and this means test cases written in MiniJava can be compiled and run using standard Java compilers. To use standard Java compilers to compile programs written in our [http://www.mscs.mu.edu/~brylow/cosc4400/Spring2011/ConcurrentMiniJava.html Concurrent MiniJava Grammar] language one needs our [[Xinu Helper Class|Xinu.java]] helper class. |
===== Course Outline ===== | ===== Course Outline ===== |
Revision as of 19:17, 27 August 2010
Contents
Overview
Having students construct a compiler which targets a runtime that uses their own, or a provided, Xinu operating system is one of the potential tracks for a professor that is Teaching With Xinu.
Including Embedded Xinu in a compiler construction course allows students to explore the compilation of high level language constructs that rely on interacting with the underlying runtime. Many traditional compilers courses simply target a processor or simulator, but by targeting a platform (a processor and operating system combination) one can extend the source language to include more advanced language features such as I/O operations and thread creation, manipulation, and concurrency. This also allows students to run their test cases on real hardware and see these programs actually interacting with a real runtime. In modern programming these high level language features are vital, and it is important for students to see what the processor and runtime are doing when they use these features in their own programs.
Course Outcomes
Course development can parallel learning objectives and topics associated with many Programming Language Translation or Compiler Construction courses. [1] However, by targeting a platform with an operating system students can also focus on learning how compilers interact with the runtime to achieve thread concurrency and synchronization; topics which many traditional compilers courses avoid. [2, 3, 4]
Topics
- Lexical Analysis
- Syntax Analysis
- Semantic Analysis
- IR Translation
- Instruction Selection
- Register Allocation
Learning Objectives
- Recognize various classes of grammars, languages, and automata, and employ these to solve common software problems.
- Explain the major steps involved in compiling a high-level programming language down to a low-level target machine language.
- Construct and use the major components of a modern compiler.
- Work together effectively in teams on a substantial software implementation project.
Potential Course Structure
The course outlined below describes a compiler construction course focusing on a semester long project in which students build most of the pieces of a complete working compiler. For this example course we take the compiler project from Appel and Palsberg's Modern Compiler Implementation in Java [2] and modify it to target a MIPS platform running the Xinu operating system. We take advantage of the fact that our compiler targets a runtime with an operating system and add high-level I/O and concurrency features to Appel and Palsberg's MiniJava language, creating our own Concurrent MiniJava Grammar language.
We allow Java-like threading and synchronization with our added support for class declarations inheriting the built in Thread class and with added support for Java's synchronized keyword. We also add external operating system calls for I/O operations and for operations to create and manipulate multiple threads of execution. Specifically, we add the ability to print strings with Xinu.print(String s), the ability to print a new line with Xinu.println(), the ability to print an integer with Xinu.printint(int x), the ability to read in an integer input with Xinu.readint(), the ability to create a thread of execution with Xinu.threadCreate(Thread t), the ability for a thread to yield control of the processor with Xinu.yield(), and the ability for a thread to sleep for a given number of milliseconds with Xinu.sleep(int time).
The links in the outline below describe the changes necessary in each assignment to add these high-level I/O and concurrency features to the language, including the modifications for targeting a Xinu backend instead of the book's intended MIPS simulator. In addition to these compiler changes, modifications must also be made to Xinu to offer the runtime support required by the synchronized keyword. Since Java's synchronized feature depends on the JVM monitor system, which has subtly different semantics from standard O/S semaphores, monitor constucts must be added to Xinu.
Appel and Palsberg's MiniJava language is a subset of the standard Java language, and this means test cases written in MiniJava can be compiled and run using standard Java compilers. To use standard Java compilers to compile programs written in our Concurrent MiniJava Grammar language one needs our Xinu.java helper class.
Course Outline
Week | Topics | Assignments | |||||
01 | Introduction | Project 1: Interpreter | |||||
02 | Lexical Analysis, Automata | Project 2: Scanner | |||||
03 | Syntax Analysis, Grammars | Homework 1: Automata and Grammars | |||||
04 | Parser Generators | Project 3: Parser | |||||
05 | Abstract Syntax Trees | ||||||
06 | Semantic Analysis | Project 4: Semantic Analysis | |||||
07 | Activation Records | ||||||
08 | IR Translation | ||||||
09 | Basic Blocks | Project 5: Translation | |||||
10 | Instruction Selection | Homework 2: Activation Records | |||||
11 | Liveness Analysis | ||||||
12 | Register Allocation | ||||||
13 | Register Allocation | Project 6: Instruction Selection | |||||
14 | Advanced Topics | ||||||
15 | Advanced Topics | Homework 3: Register Allocation |
Books
References
[1] Course topics and learning objectives have been adapted from the ACM's Computing Curricula 2001 Computer Science.
[3] A. V. Aho, M. Lam, R. Sethi, and J. D. Ullman. Compilers: Principles, Techniques and Tools. Pearson, 2nd edition, 1985.
[4] S. Muchnick. Advanced Compiler Design and Implementation. Morgan Kaufmann, 1997.
This work funded in part by NSF grant DUE-CCLI-0737476.