Comp Const Week1(Lec1 & 2)

8/2/2019 Comp Const Week1(Lec1 & 2)

1/88

1

Compiler ConstructionIntroduction


2/88

Background You Should Have

1. Programming Good C++ programmer (essential)

2. Basics of computer organization Architecture, pipelining, registers, assembly code

3. Basic theory of computation Finite automata, regular expressions, context-free

grammars


3/88

Textbook and Other Classroom

Material

Class textbook

Compilers: Principles, Techniques, and Tools, Aho, Sethi,

Ullman (aka red dragon book)

Other useful books

Lex & Yacc, Levine, Mason and Brown

Advanced Compiler Design & Implementation, StevenMuchnick

Building an Optimizing Compiler, Robert Morgan

Modern Compiler Implementation in Java, Andrew Appel


4/88

The Five Segments

Lexical and Syntax Analysis

Semantic Analysis

Code Generation Data-flow Optimizations

Instruction Optimizations


5/88

Compiler

A program written in one language istranslated into an equivalent program inanother language the target language.

Compiler Target ProgramSource Program

Error Messages


6/88

Why Compilers?

Compiler

A program that translates from 1 language toanother

It must preserve semantics of the source

It should create an efficient version of the targetlanguage

It reports to its user the presence of errors in thesource program


7/88

The Phases of a Compiler

Conceptually, a compiler operates in phases,each of which transforms the source programfrom one representation to another.

Generically speaking there are two parts tocompilation:

Analysis Synthesis


8/88

Many SW tools that manipulate source programs

first perform some kind of analysis like

Structure Editors

Pretty Printers

Static Checkers

Interpreters


9/88

Structure Editors

Takes as input a sequence of commands tobuild a source program.

performs the text creation and modification

Also analyzes the program text putting anappropriate hierarchal structure on the sourceprogram.

Examples Check that the input is correctly formed

Can supply keywords automatically


10/88

Pretty Printers

Analyzes a program and prints it in such away that the structure of the programbecomes clearly visible.

For eg: Comments may appear in a special font

Statement may appear with an amount of

indentation proportional to the depth of theirnesting in the hierarchical organization of thestatements


11/88

Static Checkers

Reads a program, analyzes it and attempts todiscover potential bugs without running theprogram

For e.g. It may detects the parts of the source program

that can never be executed, or that a certainvariable might no be used. Can catch logicalerrors such as trying to use a real variable as apointer.


12/88

Interpreters

Instead of producing a target program astranslation, an interpreter performs theoperations implied by the source program.

For e.g. An interpreter might build a tree and then carry

out the operations at the node as it walks the tree

Frequently used to execute commandlanguages.


13/88

Seemingly Unrelated places where

compiler technology is regularly used

Text formatter

Takes input that is a stream of characters, which

may include commands to indicate paragraphs,figures or mathematical structures likesuperscripts and subscripts


14/88

Silicon Compilers

Has a source language similar to a conventional

programming language. However the variables ofthe language represent, not location in memorybut logical signals (0 or 1) or group of signals in aswitching circuit. Output is a circuit design in an

appropriate language.




15/88

Query Interpreters

Translates a predicate containing relational and

Boolean operators into commands to search adatabase for records satisfying the predicate.




16/88

The Context of a Compiler

In addition to a compiler, several otherprograms may be required to create anexecutable target program.

A source program may be divided intomodules stored in separate files. The task ofcollecting the source program is entrusted to

a distinct program called a preprocessor.


17/88

A Language Processing SystemSkeletal source program

preprocessor

source program

compiler

assembler

Loader/linker - editor

Target assembly program

Relocatable machine code

Absolute machine code

Library, relocatable object files


18/88

Optimized Intermediate

Representation

Assembly code

Intermediate Code

generator

Lexical

Analyzer

Token Stream

Parse Tree

Program

(character stream)

Syntax

Analyzer

Semantic

Analyzer

Intermediate

Code Generator

CodeOptimizer

Code

Generator

Symbol TableManager

Error Handler

Source Program

Target Program


19/88

Token Stream

Program (character stream)

Lexical Analyzer (Scanner)



20/88

Lexical Analyzer (Scanner)/ Linear

Analyzer

In which the stream of characters making upthe source program is read from the left-to-right and grouped into tokens that are

sequences of characters having a collectivemeaning.

For example:

Position := initial + rate * 60



21/88


18..23 + val#ue

Num(234) mul_op lpar_op Num(11) add_op rpar_op

2 3 4 * ( 1 1 + - 2 2 )

Num(-22)


22/88


18..23 + val#ue

Num(234) mul_op lpar_op Num(11) add_op rpar_op

2 3 4 * ( 1 1 + - 2 2 )

Num(-22)

Not a number

Variable names cannot have # character


23/88

Token Stream




Syntax Analyzer (Parser)Parse Tree


24/88

Syntax Analyzer (Parser)

60

identifier

:=

position

+


Assignment Statement

identifier

initial

*

identifier

ratenumber


25/88


num num+

( )*num

num * ( num + num )


26/88


int * foo(i, j, k))

int i;

int j;{

for(i=0; i j) {

fi(i>j)

return j;}

Extra parentheses

Missing increment

Not an expression

Not a keyword


27/88

Token Stream

Program (character stream)Lexical Analyzer (Scanner)


Syntax Analyzer (Parser)Parse Tree

Semantic Analyzer

Intermediate Representation


28/88

Semantic Analyzer

60

:=

position+

initial *

rate intoreal


29/88

Semantic Analyzer

int * foo(i, j, k)

int i;

int j;{

int x;

x = x + j + N;

return j;}

Type not declared

Mismatched return type

Uninitialized variable used

Undeclared variable


30/88

Token Stream





Parse Tree

Semantic Analyzer

Intermediate RepresentationIntermediate Code Generator

Intermediate Code Representation


31/88

Intermediate Code Generator

Some compilers generate an explicitintermediate representation of the sourceprogram.

Can have variety of forms.. Three-address code (like the assembly language

for a machine in which every memory location canact like a register.. It consists of a sequence ofinstructions, each of which has at most threeoperands.


32/88

Intermediate Code Generator

temp1 := inttoreal(60)

temp2 := id3 + temp1

temp3 := id2 + temp2

id1 := temp3


33/88

Another Example (Input Program)

int expr(int n)

{

int d;

d = 4 * n * n * (n + 1) * (n

+ 1);

return d;

}


34/88

Example (Output assembly code)

lda $30,-32($30)stq $26,0($30)

stq $15,8($30)

bis $30,$30,$15

bis $16,$16,$1

stl $1,16($15)

lds $f1,16($15)

sts $f1,24($15)ldl $5,24($15)

bis $5,$5,$2

s4addq $2,0,$3

ldl $4,16($15)

mull $4,$3,$2

ldl $3,16($15)

addq $3,1,$4

mull $2,$4,$2ldl $3,16($15)addq $3,1,$4mull $2,$4,$2stl $2,20($15)ldl $0,20($15)

br $31,$33$33:

bis $15,$15,$30ldq $26,0($30)ldq $15,8($30)addq $30,32,$30

ret $31,($26),1


35/88

Token Stream





Parse Tree

Semantic Analyzer

Intermediate Representation

Intermediate Code GeneratorIntermediate Code Representation

Code Optimizer

Optimized Intermediate Representation


36/88

Code Optimizer

Attempts to improve the intermediatecode, so that faster-running machine codebe generated

Temp1 := id3 + 60.0

Id1 := id2 + temp1


37/88

Optimization

How to make the code go faster Classical optimizations

Dead code elimination remove useless code

Common sub expression eliminationrecomputing the same thing multiple times

Machine independent (classical)

Useful for almost all architectures

Machine dependent Depends on processor architecture

Memory system, branches, dependences


38/88

Example

int sumcalc(int a, int b, int N){

int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


39/88

test:

subu $fp, 16

sw zero, 0($fp) # x = 0

sw zero, 4($fp) # y = 0

sw zero, 8($fp) # i = 0

lab1: # for(i=0;i


40/88

Lets Optimize...


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


41/88

Constant Propagation


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


42/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


43/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


44/88

Algebraic Simplification


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


45/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


46/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


47/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


48/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


49/88



int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


50/88

Copy Propagation


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


51/88

Copy Propagation


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


52/88

Common Subexpression Elimination


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


53/88


int i;

int x, y;

x = 0;

y = 0;

for(i = 0; i


54/88


int i;

int x, y, t;

x = 0;

y = 0;

for(i = 0; i


55/88


int i;

int x, y, t;

x = 0;

y = 0;

for(i = 0; i


56/88

Dead Code Elimination


int i;

int x, y, t;

x = 0;

y = 0;

for(i = 0; i


57/88



int i;

int x, y, t;

x = 0;

y = 0;

for(i = 0; i


58/88



int i;

int x, y, t;

x = 0;

for(i = 0; i


59/88



int i;

int x, t;

x = 0;

for(i = 0; i


60/88

Loop Invariant Removal


int i;

int x, t;

x = 0;

for(i = 0; i


61/88



int i;

int x, t;

x = 0;

for(i = 0; i


62/88



int i;

int x, t, u;

x = 0;

u = (4*a/b);

for(i = 0; i


63/88

Strength Reduction


int i;

int x, t, u;

x = 0;

u = (4*a/b);

for(i = 0; i


64/88

Strength Reduction


int i;

int x, t, u;

x = 0;

u = (4*a/b);for(i = 0; i


65/88

Strength Reduction


int i;

int x, t, u, v;

x = 0;

u = (4*a/b);v = 0;

for(i = 0; i


66/88

Strength Reduction


int i;

int x, t, u, v;

x = 0;

u = (4*a/b);v = 0;

for(i = 0; i


67/88

Strength Reduction


int i;

int x, t, u, v;

x = 0;

u = (4*a/b);v = 0;

for(i = 0; i


68/88

Optimized Example


int i;

int x, t, u, v;

x = 0;

u = (4*a/b);v = 0;

for(i = 0; i


69/88

subu $fp, 16

add $t9, zero, zero # x = 0

sll $t0, $a0, 2 # a


70/88

test:subu $fp, 16add $t9, zero, zerosll $t0, $a0, 2div $t7, $t0, $a1add $t6, zero, zeroadd $t5, zero, zero

lab1:addui $t8, $t5, 1

mul $t0, $t8, $t8addu $t1, $t0, $t6addu $t9, t9, $t1

addu $6, $6, $7

addui $t5, $t5, 1

ble $t5, $a3, lab1

addu $v0, $t9, zeroaddu $fp, 16 b $ra

test:subu $fp, 16sw zero, 0($fp)sw zero, 4($fp)sw zero, 8($fp)

lab1: mul $t0, $a0, 4

div $t1, $t0, $a1lw $t2, 8($fp) mul $t3, $t1, $t2lw $t4, 8($fp)addui $t4, $t4, 1lw $t5, 8($fp)addui $t5, $t5, 1 mul $t6, $t4, $t5addu $t7, $t3, $t6lw $t8, 0($fp)add $t8, $t7, $t8sw $t8, 0($fp)lw $t0, 4($fp) mul $t1, $t0, a1lw $t2, 0($fp)add $t2, $t2, $t1sw $t2, 0($fp)lw $t0, 8($fp)addui $t0, $t0, 1

sw $t0, 8($fp) ble $t0, $a3, lab1

lw $v0, 0($fp)addu $fp, 16 b $ra

Execution time = 17 secExecution time = 43 sec

C mpil r Optimiz Pr r m f r


71/88

Compilers Optimize Programs for

Performance/Speed

Code Size

Power Consumption

Fast/Efficient Compilation

Security/Reliability

Debugging

C d G n r t r


72/88

Code Generator

MOVF id3, R2

MULF #60.0,R2

MOVF id2, R1

ADDF R2, R1

MOVF R1, id1

Code Generator


73/88

Code Generator

test:subu $fp, 16add $t9, zero, zerosll $t0, $a0, 2div $t7, $t0, $a1add $t6, zero, zeroadd $t5, zero, zero

lab1:addui $t8, $t5, 1

mul $t0, $t8, $t8addu $t1, $t0, $t6addu $t9, t9, $t1

addu $6, $6, $7

addui $t5, $t5, 1 ble $t5, $a3, lab1

addu $v0, $t9, zeroaddu $fp, 16 b $ra

int sumcalc(int a, int b, int N)

{

int i;

int x, t, u, v;

x = 0;

u = ((a


74/88

Symbol Table Manager


75/88

Error Detection and Reporting

Dataflow and Control Flow Analysis


76/88

Dataflow and Control Flow Analysis

Dataflow analysis

Provide the necessary information about variableusage and execution behavior to determine when

a transformation is legal/illegal Control flow analysis

Execution behavior caused by control statements

Ifs, for/while loops, gotos

Control flow graph

Cousins of the Compiler (Preprocessors)


77/88

Cousins of the Compiler (Preprocessors)

Produces input to the compilers

1. Macro processing

To define macros that are shorthand for longer

constructs.2. File Inclusion

To include header files into the program text

Cousins of the Compiler


78/88

Cousins of the Compiler

3. Rational Preprocessor Augment older languages with modern flow of

control and data structuring facilities.

4. Language extensions

These processors attempt to add capabilities tothe language by what amounts to built in macros.

For example, such the language Equel is adatabase query language embedded in C.(Statements Beginning with ##)

Cousins of the Compiler (Assemblers)


79/88


Production of relocatable machine code thatcan be passed directly to the loader/linker-editor.

Assembly code (menmonic version of machinecode)

Example:

MOV a, R1

ADD #2, R1

MOV R1, b



80/88


Two-Pass Assembly (Two passes over theinput, Reading an input file once)

First phase: Identifiers r found & stored in a

symbol table Identifier Address

a 0

b 4Assume that a word, consisting of four bytes is setaside for each identifier and the addresses areassigned stating from byte 0.



81/88


Two-Pass Assembly Second phase:

Assembler scans the input again.

It translates each operation code into thesequence of bits representing that operation inmachine language.

Translates each identifier into the address given

for that in symbol table. Output is relocatable machine code.



82/88


A hypothetical machine codeInstruction R1 tag address/value

0001 01 00 00000000

0011 01 10 00000010 0010 01 00 00000100

0001 LOAD 0011 ADD 0010 STORE

Cousins of the Compiler (Loader and Link


83/88


- Editors)

2 Functions Loading

Linking

Loading Taking relocatable machine code,

altering the relocatable address

Placing the altered instruction and data in memory atproper locations



84/88


- Editors)

Represent the Relocation bit i.e. associatedwith each operand in relocatable machinecode.

Suppose address space containing thedata is to be located at location L

L must be added to address of the

instruction



85/88


- Editors)

If L = 00001111 i.e. 15, then a and b would beat 15 and 19

0001 01 00 00000000

0011 01 10 000000100010 01 00 00000100



86/88


- Editors)

Linker Editor allows to make a single programfrom several files of relocatable machinecode.

Files may be the result of several differentcompilations

One or more may be the library files of

routine

General Structure of a Modern Compiler


87/88

General Structure of a Modern Compiler

Lexical Analysis

Syntax Analysis

Semantic Analysis

Controlflow/Dataflow

Optimization

Code Generation

Source

Program

AssemblyCode

Scanner

Parser

High-level IR to low-level IR conversion

Build high-level IRContext

Symbol Table

CFG

Machine independent asm to machine dependent

Front end

Back end

Front and Back Ends


88/88

Front and Back Ends

Front End Phases or parts of phases depend primarily on

the source language

Back End Phases or parts of phases depend primarily on

the target machine

Generally do not depend on source language

Comp Const Week1(Lec1 & 2)

Documents

Transcript of Comp Const Week1(Lec1 & 2)