Programmer's Guide

Programmer's%20Guide

User Manual:

Open the PDF directly: View PDF .
Page Count: 520

Download
Open PDF In Browser	View PDF

Programmer’s Guide

Microsoft MASM
®

Assembly-Language Development System
Version 6.1
For MS-DOS ® and Windows ™ Operating Systems

Microsoft Corporation

Filename: LMAPGTTL.DOC Project:
Template: FRONTA1.DOT Author: Bart Simpson, Who the Hell Are You?
Revision #: 16 Page: 1 of 1 Printed: 10/02/00 04:19 PM

Last Saved By: Mike Eddy

Information in this document is subject to change without notice. Companies, names, and data used in
examples herein are fictitious unless otherwise noted. No part of this document maybe reproduced or
transmitted in any form or by any means, electronic or mechanical, for any purpose, without the
express written permission of Microsoft Corporation.
©1992 Microsoft Corporation. All rights reserved.

Microsoft, MS, MS-DOS, XENIX, CodeView, and QuickC are registered trademarks and Microsoft
QuickBasic, QuickPascal, Windows and Windows NT are trademarks of Microsoft Corporation in
the USA and other countries.
U.S. Patent No. 4,955,066
Hercules is a registered trademark of Hercules Computer Technology.
IBM, PS/2, and OS/2 are registered trademarks of International Business Machines Corporation.
Intel is a registered trademark of Intel Corporation.
NEC and V25 are registered trademarks and V35 is a trademark of NEC Corporation.

Document No. DB35747-1292
Printed in the United States of America.

Filename: LMAPGCPY.DOC Project:
Template: FRONTA1.DOT Author: Ruth L Silverio Last Saved By: Mike Eddy
Revision #: 6 Page: 2 of 1 Printed: 10/02/00 04:21 PM

iii

Contents
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
New and Extended Features in MASM 6.1 . . . . . . . . . . . . . . . . . . . . . . . . xiii
MASM Features New Since Version 5.1 . . . . . . . . . . . . . . . . . . . . . . . . xiv
MASM Features New Since Version 6.0 . . . . . . . . . . . . . . . . . . . . . . . . xv
ML and MASM Command Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
Compatibility with Earlier Versions of MASM . . . . . . . . . . . . . . . . . . . . xvi
A Word About Instruction Timings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Books for Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii
Document Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
Getting Assistance and Reporting Problems . . . . . . . . . . . . . . . . . . . . . . . . xx
Chapter 1 Understanding Global Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
The Processing Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
8086-Based Processors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Operating Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Segmented Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Segment Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Segmented Addressing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Segment Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Language Components of MASM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Reserved Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Identifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Predefined Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Integer Constants and Constant Expressions . . . . . . . . . . . . . . . . . . . . . 11
Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Data Types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
The Assembly Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Generating and Running Executable Programs . . . . . . . . . . . . . . . . . . . . 23
Using the OPTION Directive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Conditional Directives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Chapter 2 Organizing Segments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Physical Memory Segments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Logical Segments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Using Simplified Segment Directives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 3 of 1 Printed: 10/02/00 04:19 PM

Contents

Defining Basic Attributes with .MODEL . . . . . . . . . . .
Specifying a Processor and Coprocessor . . . . . . . . . . .
Creating a Stack. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating Data Segments . . . . . . . . . . . . . . . . . . . . . . .
Creating Code Segments . . . . . . . . . . . . . . . . . . . . . . .
Starting and Ending Code with .STARTUP and .EXIT .
Using Full Segment Definitions . . . . . . . . . . . . . . . . . . . .
Defining Segments with the SEGMENT Directive. . . . .
Controlling the Segment Order . . . . . . . . . . . . . . . . . .
Setting the ASSUME Directive for Segment Registers . .
Defining Segment Groups . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

34
38
38
39
40
41
44
44
47
49
51

Chapter 3 Using Addresses and Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Programming Segmented Addresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Initializing Default Segment Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Near and Far Addresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Operands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Register Operands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Immediate Operands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Direct Memory Operands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Indirect Memory Operands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
The Program Stack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Saving Operands on the Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Saving Flags on the Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Saving Registers on the Stack (80186–80486 Only). . . . . . . . . . . . . . . . . 74
Accessing Data with Pointers and Addresses . . . . . . . . . . . . . . . . . . . . . . . . 74
Defining Pointer Types with TYPEDEF . . . . . . . . . . . . . . . . . . . . . . . . 75
Defining Register Types with ASSUME . . . . . . . . . . . . . . . . . . . . . . . . . 77
Basic Pointer and Address Operations . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Chapter 4 Defining and Using Simple Data Types . . . . . . . . . . . . . . . . . . . . . . . 85
Declaring Integer Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Allocating Memory for Integer Variables . . . . . . . . . . . . . . . . . . . . . . . . 85
Data Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Working with Simple Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Copying Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Adding and Subtracting Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Multiplying and Dividing Integers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Manipulating Numbers at the Bit Level. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Logical Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Shifting and Rotating Bits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 4 of 2 Printed: 10/02/00 04:19 PM

Contents

Multiplying and Dividing with Shift Instructions . . . . . . . . . . . . . . . . . . 102
Chapter 5 Defining and Using Complex Data Types. . . . . . . . . . . . . . . . . . . . . 105
Arrays and Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Declaring and Referencing Arrays. . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Declaring and Initializing Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Processing Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Structures and Unions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Declaring Structure and Union Types . . . . . . . . . . . . . . . . . . . . . . . . . 118
Defining Structure and Union Variables. . . . . . . . . . . . . . . . . . . . . . . . 121
Referencing Structures, Unions, and Fields . . . . . . . . . . . . . . . . . . . . . 126
Nested Structures and Unions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Declaring Record Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Defining Record Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Record Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers . . . . . . . 135
Using Floating-Point Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Declaring Floating-Point Variables and Constants . . . . . . . . . . . . . . . . . 136
Storing Numbers in Floating-Point Format. . . . . . . . . . . . . . . . . . . . . . 138
Using a Math Coprocessor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Coprocessor Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Instruction and Operand Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Coordinating Memory Access. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Using Coprocessor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Using An Emulator Library. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Using Binary Coded Decimal Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Defining BCD Constants and Variables . . . . . . . . . . . . . . . . . . . . . . . . 157
BCD Calculations on a Coprocessor . . . . . . . . . . . . . . . . . . . . . . . . . . 157
BCD Calculations on the Main Processor . . . . . . . . . . . . . . . . . . . . . . 158
Chapter 7 Controlling Program Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Unconditional Jumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Conditional Jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Loop-Generating Directives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Writing Loop Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Defining Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 5 of 3 Printed: 10/02/00 04:19 PM

Contents

Passing Arguments on the Stack . . . . . . . . . . .
Declaring Parameters with the PROC Directive
Using Local Variables. . . . . . . . . . . . . . . . . . .
Creating Local Variables Automatically . . . . . .
Declaring Procedure Prototypes . . . . . . . . . . .
Calling Procedures with INVOKE . . . . . . . . . .
Generating Prologue and Epilogue Code. . . . . .
MS-DOS Interrupts . . . . . . . . . . . . . . . . . . . . . .
Calling MS-DOS and ROM-BIOS Interrupts . .
Replacing an Interrupt Routine . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.

182
184
188
190
193
194
198
204
204
206

Chapter 8 Sharing Data and Procedures Among Modules and Libraries . . . . . 211
Selecting Data-Sharing Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
Sharing Symbols with Include Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Organizing Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Declaring Symbols Public and External . . . . . . . . . . . . . . . . . . . . . . . . 214
Positioning External Declarations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Using Alternatives to Include Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
PUBLIC and EXTERN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Other Alternatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Developing Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Associating Libraries with Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Using EXTERN with Library Routines . . . . . . . . . . . . . . . . . . . . . . . . 223
Chapter 9 Using Macros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Text Macros. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Macro Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Creating Macro Procedures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Passing Arguments to Macros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Specifying Required and Default Parameters . . . . . . . . . . . . . . . . . . . . 229
Defining Local Symbols in Macros . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Assembly-Time Variables and Macro Operators . . . . . . . . . . . . . . . . . . . . 233
Text Delimiters and the Literal-Character Operator . . . . . . . . . . . . . . . . 234
Expansion Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Substitution Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Defining Repeat Blocks with Loop Directives . . . . . . . . . . . . . . . . . . . . . . 239
REPEAT Loops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
WHILE Loops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
FOR Loops and Variable-Length Parameters . . . . . . . . . . . . . . . . . . . . 242
FORC Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
String Directives and Predefined Functions . . . . . . . . . . . . . . . . . . . . . . . . 245

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 6 of 4 Printed: 10/02/00 04:19 PM

Contents

Returning Values with Macro Functions. . . . . . . . . . . . . . . . . . .
Returning Values with EXITM. . . . . . . . . . . . . . . . . . . . . . .
Using Macro Functions with Variable-Length Parameter Lists .
Expansion Operator in Macro Functions . . . . . . . . . . . . . . . .
Advanced Macro Techniques . . . . . . . . . . . . . . . . . . . . . . . . . .
Defining Macros within Macros . . . . . . . . . . . . . . . . . . . . . .
Testing for Argument Type and Environment . . . . . . . . . . . .
Using Recursive Macros . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.

vii

.
.
.
.
.
.
.
.

248
248
249
251
251
251
252
255

Chapter 10 Writing a Dynamic-Link Library For Windows . . . . . . . . . . . . . . . . 257
Overview of DLLs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Loading a DLL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Building a DLL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
DLL Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
DLL Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
DLL Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
DLL Extension Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Example of a DLL: SYSINFO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Entry Routine for SYSINFO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
Expanding SYSINFO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
Chapter 11 Writing Memory-Resident Software . . . . . . . . . . . . . . . . . . . . . . . . 273
Terminate-and-Stay-Resident Programs. . . . . . . . . . . . . . . . . . . . . . . . . . 273
Structure of a TSR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Passive TSRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Active TSRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Interrupt Handlers in Active TSRs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Auditing Hardware Events for TSR Requests . . . . . . . . . . . . . . . . . . . 275
Monitoring System Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Determining Whether to Invoke the TSR . . . . . . . . . . . . . . . . . . . . . . 279
Example of a Simple TSR: ALARM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Using MS-DOS in Active TSRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Understanding MS-DOS Stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Determining MS-DOS Activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Interrupting MS-DOS Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Monitoring the Critical Error Flag . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Preventing Interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Trapping Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Preserving an Existing Condition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Preserving Existing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 7 of 5 Printed: 10/02/00 04:19 PM

viii

Contents

Communicating Through the Multiplex Interrupt . . . . . . . . . .
The Multiplex Handler . . . . . . . . . . . . . . . . . . . . . . . . . .
Using the Multiplex Interrupt Under MS-DOS Version 2.x.
Deinstalling a TSR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example of an Advanced TSR: SNAP . . . . . . . . . . . . . . . . .
Building SNAP.EXE . . . . . . . . . . . . . . . . . . . . . . . . . . .
Outline of SNAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.

290
291
292
292
293
294
295

Chapter 12 Mixed-Language Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Naming and Calling Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
Naming Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
The C Calling Convention. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
The Pascal Calling Convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
The STDCALL and SYSCALL Calling Conventions. . . . . . . . . . . . . . . 311
Writing an Assembly Procedure For a Mixed-Language Program . . . . . . . . 312
The MASM/High-Level–Language Interface . . . . . . . . . . . . . . . . . . . . . . . 313
The C/MASM Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
The C++/MASM Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
The FORTRAN/MASM Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
The Basic/MASM Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
Chapter 13 Writing 32-Bit Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
32-Bit Memory Addressing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
MASM Directives for 32-Bit Programming . . . . . . . . . . . . . . . . . . . . . . . . 336
Sample Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337

Appendixes
Appendix A Differences Between MASM 6.1 and 5.1. . . . . . . . . . . . . . . . . . . . 341
New Features of Version 6.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
The Assembler, Environment, and Utilities . . . . . . . . . . . . . . . . . . . . . . 342
Segment Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
Procedures, Loops, and Jumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Simplifying Multiple-Module Projects . . . . . . . . . . . . . . . . . . . . . . . . . 348
Expanded State Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
New Processor Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
Renamed Directives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
Macro Enhancements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
MASM 6.1 Programming Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
Compatibility Between MASM 5.1 and 6.1. . . . . . . . . . . . . . . . . . . . . . . . 352

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 8 of 6 Printed: 10/02/00 04:19 PM

Contents

Rewriting Code for Compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Using the OPTION Directive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Changes to Instruction Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Appendix B BNF Grammar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
Appendix Generating and Reading Assembly Listings. . . . . . . . . . . . . . . . . . 397
Generating Listing Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
Precedence of Command-Line Options and Listing Directives. . . . . . . . 399
Reading the Listing File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
Generated Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
Error Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
Symbols and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
Reading Tables in a Listing File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
Appendix D MASM Reserved Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
Operands and Symbols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
Special Operands for the 80386/486 . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Predefined Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Operators and Directives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410
Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
8086/8088 Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
80186 Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
80286 Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
80286 and 80386 Privileged-Mode Instructions . . . . . . . . . . . . . . . . . . 413
80386 Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
80486 Processor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Instruction Prefixes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Coprocessor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
8087 Coprocessor Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
80287 Privileged-Mode Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
80387 Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
Appendix E Default Segment Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 9 of 7 Printed: 10/02/00 04:19 PM

Contents

Figures and Tables
Figures
1.1 Segment Allocation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Calculating Physical Addresses . . . . . . . . . . . . . . . . . . . . . .
1.3 Registers for 8088-80286 Processors . . . . . . . . . . . . . . . . . .
1.4 Extended Registers for the 80386/486 Processors . . . . . . . . .
1.5 Flags for 8088-80486 Processors. . . . . . . . . . . . . . . . . . . . .
3.1 Stack Status Before and After Pushes and Pops . . . . . . . . . .
4.1 Integer Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Shifts and Rotates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1 Encoding for Real Numbers in IEEE Format . . . . . . . . . . . .
6.2 Coprocessor Data Registers. . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Status of the Register Stack. . . . . . . . . . . . . . . . . . . . . . . . .
6.4 Status of the Register Stack and Memory Locations . . . . . . .
6.5 Status of the Previously Initialized Register Stack . . . . . . . . .
6.6 Status of the Already Initialized Register Stack . . . . . . . . . . .
6.7 Status of the Register Stack: Main Memory and Coprocessor.
6.8 Coprocessor Control Registers. . . . . . . . . . . . . . . . . . . . . . .
6.9 Coprocessor and Processor Control Flags . . . . . . . . . . . . . . .
7.1 Program Arguments on the Stack . . . . . . . . . . . . . . . . . . . .
7.2 Local Variables on the Stack . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Operation of Interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1 Using EXTERNDEF for Variables. . . . . . . . . . . . . . . . . . . .
8.2 Using PROTO and INVOKE . . . . . . . . . . . . . . . . . . . . . . .
8.3 Using PUBLIC and EXTERN. . . . . . . . . . . . . . . . . . . . . . .
11.1 Time Line of Interaction Between Interrupt Handlers
for a Typical TSR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2 Flowchart for SNAP.EXE: Installation Phase . . . . . . . . . . .
11.3 Flowchart for SNAP.EXE Resident Phase . . . . . . . . . . . . .
11.4 Flowchart for SNAP.EXE Deinstallation Phase. . . . . . . . . .
12.1 C String Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.2 C Stack Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.3 FORTRAN String Frame . . . . . . . . . . . . . . . . . . . . . . . . .
12.4 FORTRAN Stack Frame. . . . . . . . . . . . . . . . . . . . . . . . . .
12.5 Basic String Descriptor Format . . . . . . . . . . . . . . . . . . . . .
12.6 Basic Stack Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
B.1 BNF Definition of the TYPEDEF Directive . . . . . . . . . . . . .

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 10 of 8 Printed: 10/02/00 04:19 PM

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. 6
. 8
. 17
. 18
. 20
. 72
. 87
101
138
140
142
143
144
144
148
154
155
183
190
206
215
217
221

.
.
.
.
.
.
.
.
.
.
.

278
296
297
298
316
320
324
327
330
333
380

Contents

Tables
1.1 8086 Family of Processors. . . . . . . . . . . . . . . . . . . . . . . .
1.2 The MS-DOS and Windows Operating Systems Compared
1.3 Operator Precedence . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1 Attributes of Memory Models . . . . . . . . . . . . . . . . . . . . .
3.1 Indirect Addressing with 16-Bit Registers . . . . . . . . . . . . .
4.1 Division Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1 Requirements for String Instructions . . . . . . . . . . . . . . . . .
6.1 Ranges of Floating-Point Variables . . . . . . . . . . . . . . . . . .
6.2 Coprocessor Operand Formats . . . . . . . . . . . . . . . . . . . . .
6.3 Control-Flag Settings After Comparison or Test . . . . . . . . .
7.1 Conditional Jumps Based on Comparisons of Two Values .
9.1 MASM Macro Operators . . . . . . . . . . . . . . . . . . . . . . . . .
11.1 MS-DOS Internal Stacks . . . . . . . . . . . . . . . . . . . . . . . .
12.1 Naming and Calling Conventions . . . . . . . . . . . . . . . . . .
12.2 Register Conventions for Simple Return Values . . . . . . . .
A.1 Requirements for String Instructions. . . . . . . . . . . . . . . . .
C.1 Options for Generating or Modifying Listing Files . . . . . . .
C.2 Symbols and Abbreviations in Listings . . . . . . . . . . . . . . .
C.3 Symbols in Timing Column . . . . . . . . . . . . . . . . . . . . . . .

Filename: LMAPGTOC.DOC Project:
Template: FRONTA1.DOT Author: Don Hayward Last Saved By: Ruth L Silverio
Revision #: 18 Page: 11 of 9 Printed: 10/02/00 04:19 PM

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.

. 2
. 4
14
35
68
97
112
136
141
151
167
234
286
309
317
353
398
400
401

xiii

Introduction
The Microsoft® Macro Assembler Programmer’s Guide provides the
information you need to write and debug assembly-language programs with the
Microsoft Macro Assembler (MASM), version 6.1. This book documents
enhanced features of the language and the programming environment for
MASM 6.1.
This Programmer’s Guide is written for experienced programmers who know
assembly language and are familiar with an assembler. The book does not teach
the basics of assembly language; it does explain Microsoft-specific features. If
you want to learn or review the basics of assembly language, refer to “Books for
Further Reading” in this introduction.
This book teaches you how to write efficient code with the new and advanced
features of MASM. Getting Started explains how to set up MASM 6.1.
Environment and Tools introduces the integrated development environment
called the Programmer’s WorkBench (PWB). It also includes a detailed
reference to Microsoft tools and utilities such as Microsoft ® CodeView ®,
LINK, and NMAKE. The Microsoft Macro Assembler Reference provides a full
listing of all MASM instructions, directives, statements, and operators, and it
serves as a quick reference to utility commands.
For more information on these same topics, see the online Microsoft Advisor,
which is a complete reference to Macro Assembler language topics, to the
utilities, and to PWB. You should be able to find most of the information you
need in the Microsoft Advisor.

New and Extended Features in MASM 6.1
MASM 6.1 continues the break with tradition established by version 6.0. It
incorporates conveniences of high-level languages while offering all the
traditional advantages of assembly-language programming.
For example, MASM 6.1 includes the Programmer’s WorkBench, which
provides the same integrated software development environment enjoyed by
programmers of Microsoft high-level languages such as C and Basic. From
within PWB you can edit, build, debug, or run a program. You can perform
most of these operations with either menu selections or keyboard commands.
You can also customize PWB to suit your individual programming and editing
requirements and preferences.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 13 of 1 Printed: 10/02/00 04:20 PM

xiv

Programmer’s Guide

MASM Features New Since Version 5.1
MASM 6.1 includes several features designed to make programming more
efficient and productive. The following list briefly describes how MASM 6.1
improves on the language features of the popular version 5.1.
u

MASM 6.1 has many enhancements related to types. You can now use the
same type specifiers in initializations as in other contexts (BYTE instead of
DB). You can also define your own types, including pointer types, with the
new TYPEDEF directive. See Chapter 3, “Using Addresses and Pointers,”
and Chapter 4, “Defining and Using Simple Data Types.”

The syntax for defining and using structures and records has been enhanced
since version 5.1. You can also define unions with the new UNION
directive. See Chapter 5, “Defining and Using Complex Data Types.”
MASM now generates complete CodeView information for all types. See
Chapter 3, “Using Addresses and Pointers,” and Chapter 4, “Defining and
Using Simple Data Types.”
New control-flow directives let you use high-level – language constructs such
as loops and if-then-else blocks defined with .REPEAT and .UNTIL (or
.UNTILCXZ); .WHILE and .ENDW; and .IF, .ELSE, and .ELSEIF. The
assembler generates the appropriate code to implement the control structure.
See Chapter 7, “Controlling Program Flow.”
MASM now has more powerful features for defining and calling procedures.
The extended PROC syntax for generating stack frames has been enhanced
since version 5.1. You can also use the PROTO directive to prototype a
procedure, which you can then call with the INVOKE directive. INVOKE
automatically generates code to pass arguments (converting them to a related
type, if appropriate) and makes the call according to the specified calling
convention. See Chapter 7, “Controlling Program Flow.”
MASM optimizes jumps by automatically determining the most efficient
coding for a jump and then generating the appropriate code. See Chapter 7,
“Controlling Program Flow.”
Maintaining multiple-module programs is easier in MASM 6.1 than in version
5.1. The EXTERNDEF and PROTO directives make it easy to maintain all
global definitions in include files shared by all the source modules of a
project. See Chapter 8, “Sharing Data and Procedures Among Modules and
Libraries.”

The assembler has many new macro features that make complex macros clearer
and easier to write:
u

You can specify default values for macro arguments or mark arguments as
required. And with the VARARG keyword, one parameter can accept a
variable number of arguments.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 14 of 2 Printed: 10/02/00 04:20 PM

Introduction

You can implement loops inside of macros in various ways. For example, the
new WHILE directive expands the statements in a macro body while an
expression is not zero.

You can define macro functions, which return text macros. Several
predefined text macros are also provided for processing strings. Macro
operators and other features related to processing text macros and macro
arguments have been enhanced. For more information on all these macro
features, see Chapter 9, “Using Macros.”

MASM 6.1 has other improved capabilities, such as:
u

The .STARTUP and .EXIT directives automatically generate appropriate
startup and exit code for your assembly-language programs. See Chapter 2,
“Organizing Segments.”
MASM 6.1 supports flat memory model, available with the new Microsoft ®
Windows NT ™ operating system. Flat model allows segments as large as 4
gigabytes instead of 64K (kilobytes). Offsets are 32 bits instead of 16 bits.
See Chapter 2, “Organizing Segments.”
The program H2INC.EXE converts C include files to MASM include files
and translates data structures and declarations. See Chapter 20 in
Environment and Tools.
MASM 6.1 provides a library of assembly routines that let you create a
terminate-and-stay-resident program (TSR) in a high-level language.

MASM 6.1 includes many other minor new features as well as extensive support
for features of earlier versions of MASM. For a complete list of enhancements,
refer to Appendix A, “Differences between MASM 6.1 and 5.1.” The crossreferences in Appendix A guide you to the chapters where the new features are
described in detail.

MASM Features New Since Version 6.0
MASM 6.1 offers several new features:
u

ML now runs in 32-bit protected mode under MS-DOS, giving it direct
access to extended memory for assembling very large source files.
A collection of tools lets you write a dynamic-link library (DLL) for the
Microsoft ® Windows ™ operating system without the Windows Software
Development Kit. The LIBW.LIB library provides access to all functions in
the Windows application programming interface (API), so your DLL can
display menus, dialog boxes, and scroll bars. Chapter 10, “Writing a
Dynamic-Link Library for Windows,” shows you how.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 15 of 3 Printed: 10/02/00 04:20 PM

xvi

Programmer’s Guide
u

Program listings now show instruction timings. The number of required
processor cycles appears adjacent to each instruction in the listing, based on
the selected processor. For an example listing and instructions on how to use
this feature, see Appendix C, “Generating and Reading Assembly Listings.”

All utilities have been updated for version 6.1. Documentation is clearer and
better arranged, with a new Environment and Tools reference book.

Version 6.1 generates debugging information for CodeView version 4.0 and
later.

MASM 6.1 provides even greater compatibility with version 5.1 than does
MASM 6.0. Many programs written with version 5.1 will assemble
unchanged under MASM 6.1.

ML and MASM Command Lines
MASM 6.1 provides an updated version of the command-line driver, ML,
introduced in version 6.0. ML is more powerful and flexible than the MASM
driver of version 5.1. ML assembles and links with one command. It recognizes
all the old MASM driver command syntax, however, to support existing batch
files and makefiles that use MASM command lines.
Note The name MASM has traditionally referred to the Microsoft Macro
Assembler. It is used in that context throughout this book. However, MASM
also refers to MASM.EXE, which has been replaced by ML.EXE. In MASM
6.1, MASM.EXE is a small utility that translates command-line options to those
accepted by ML.EXE, and then calls ML.EXE. The distinction between
ML.EXE and MASM.EXE is made whenever necessary. Otherwise, MASM
refers to the assembler and its features.

Compatibility with Earlier Versions of MASM
MASM 6.1 is fully compatible with version 6.0 and, in many cases, with version
5.1. Code written for MASM 5.1 will often assemble correctly without
modification under MASM 6.1. However, MASM 6.1 provides the OPTION
directive to let you selectively modify the assembly process. In particular, you
can use the M510 argument with OPTION or the /Zm command-line option to
set most features to be compatible with version 5.1 code.
For information about obsolete features that will not assemble correctly under
MASM 6.1, see Appendix A, “Differences Between MASM 6.1 and 5.1.” The
appendix also explains how to update code to use the new features.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 16 of 4 Printed: 10/02/00 04:20 PM

Introduction

xvii

A Word About Instruction Timings
As an assembly-language programmer, whether novice or expert, you are
probably interested in producing lightning-fast code. After all, one of the main
reasons to program in assembly is to take advantage of its ability to streamline
execution speeds to the limit of the processor. This book will help you write
efficient and fast programs.
When discussing the speed of individual instructions, the chapters in this book
often speak of “timing,” which is the number of processor cycles required to
carry out an instruction. The Reference lists instruction timings for processors in
the 8086 family. It is tempting to use timing as the only criterion when judging
an instruction’s actual execution speed, but the world within the processor is not
so simple.
The clock for instruction timing does not begin ticking until the processor has
read and begins to execute an instruction. When you read about instruction
timings (in this book or any other), keep in mind that other factors also influence
the real speed of an instruction: the instruction’s size, whether it resides in cache
memory, whether it accesses memory, its position in the processor’s prefetch
queue, and the processor type. These factors make it impossible to say precisely
how fast an instruction executes. Accept the references to timing in this book as
guidelines, but use these simple rules to write fast code:
u

u
u

Whenever possible, use registers rather than constant values, and constant
values rather than memory.
Minimize changes in program flow.
Smaller is often better. For example, the instructions
dec
sub

bx
bx, 1

accomplish the same thing and have the same timings on 80386/486
processors. But the first instruction is 3 bytes smaller than the second, and so
may reach the processor faster.
• When possible, use the string instructions described in Chapter 5, “Defining
and Using Complex Data Types.”

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 17 of 5 Printed: 10/02/00 04:20 PM

xviii

Programmer’s Guide

Books for Further Reading
The following books may help you learn to program in assembly language or
write specialized programs. These books are listed only for your convenience.
Microsoft makes no specific recommendations concerning any of these books.

Books About Programming in Assembly Language
Abrash, Michael. Zen of Assembly Language. Glenview, IL: Scott, Foresman
and Co., 1990. Out of print.
Duntemann, Jeff. Assembly Language from Square One: For the PC AT and
Compatibles. Glenview, IL: Scott, Foresman and Co., 1990. Out of print.
Fernandez, Judi N., and Ruth Ashley. Assembly Language Programming for
the 80386. New York: McGraw-Hill, 1990.
Miller, Alan R. DOS Assembly Language Programming. San Francisco:
SYBEX, 1988. Out of print.
Scanlon, Leo J. 80286 Assembly Language Programming on MS-DOS
Computers. New York: Brady Communications, 1986. Out of print.
Turley, James L. Advanced 80386 Programming Techniques. Berkeley, CA:
Osborne McGraw-Hill, 1988.

Books About MS-DOS and BIOS
“Terminate-and-Stay-Resident Utilities.” MS-DOS Encyclopedia. Redmond,
WA: Microsoft Press, 1989.
Duncan, Ray. Advanced MS-DOS Programming: The Microsoft Guide for
Assembly Language and C Programmers. 2d ed. Redmond, WA: Microsoft
Press, 1988.
Duncan, Ray. Extending DOS: Programmer’s Guide to Protected-Mode DOS.
Redding, MA: Addison-Wesley. 1991.
Jourdain, Robert. Programmer’s Problem Solver for the IBM PC, XT and AT.
New York: Brady Communications, 1985. Out of print.
Microsoft MS-DOS Programmer’s Reference. Redmond, WA: Microsoft Press,
1991.
Norton, Peter and Richard Wilton. The New Peter Norton Programmer’s Guide
to the IBM PC and PS/2. Redmond, WA: Microsoft Press, 1988.
Wilton, Richard. Programmer’s Guide to PC & PS/2 Video Systems: Maximum
Video Performance from the EGA, VGA, HGC, and MCGA. Redmond, WA:
Microsoft Press, 1987. Out of print.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 18 of 6 Printed: 10/02/00 04:20 PM

Introduction

xix

Books and Articles About Windows
Kauler, Barry. Windows Assembly Language & Systems Programming: ObjectOriented & Systems Programming in Assembly Language for Windows 3.0
and 3.1. New York, NY: Prentice Hall, 1993.
Klein, Mike. Windows Programmer’s Guide to DLLs & Memory Management.
Carmel, IN: Sams, 1992.
Petzold, Charles. Programming Windows. 3d ed. Redmond, WA: Microsoft
Press, 1992.
Petzold, Charles. “Environments.” PC Magazine. New York, NY: Ziff-Davis
Publishing Company, June 1990–1992.
Programmer’s Reference. 4 vols. Microsoft Windows Software Development
Kit (SDK). Redmond, WA: Microsoft Press, 1992.

Books About Other Topics
Nelson, Ross P. The 80386/80486 Programming Guide. 2d ed. Redmond, WA:
Microsoft Press, 1991.
Startz, Richard. 8087/80287/80387 for the IBM PC and Compatibles:
Applications and Programming with Intel’s Math Coprocessors. Bowie, MD:
Robert J. Brady Co., 1988. Out of print.

Document Conventions
The following document conventions are used throughout this manual:
Example of
Convention

Description

SAMPLE2.ASM

Uppercase letters indicate filenames, segment names, registers, and
terms used at the command level.

.MODEL

Boldface type indicates assembly-language directives, instructions,
type specifiers, and predefined macros, as well as keywords in other
programming languages.

placeholder

Italic letters indicate placeholders for information you must supply,
such as a filename. Italics are used occasionally for emphasis in the
text.

target

This font is used to indicate example programs, user input, and screen
output.

;

A semicolon in the first column of an example signals illegal code. A
semicolon also marks a comment.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 19 of 7 Printed: 10/02/00 04:20 PM

Programmer’s Guide
SHIFT

Small capital letters signify names of keys on the keyboard. Notice
that a plus (+) indicates a combination of keys. For example, CTRL+E
means to hold down the CTRL key while pressing the E key.

[[argument]]

Items inside double square brackets are optional.

{register|memory}

Braces and a vertical bar indicate a choice between two or more
items. You must choose one of the items unless double square
brackets surround the braces.

Repeating elements...

A horizontal ellipsis (...) following an item indicates that more items
having the same form may appear.

Program
.
.
.
Fragment

A vertical ellipsis tells you that part of a program has been intentionally
omitted.

Getting Assistance and Reporting Problems
If you need help or think you have discovered a problem in the software, please
provide the following information to help us locate the source of the problem:
u
u

The version of MS-DOS or Windows you run.
Your system configuration: the type of machine you use, its total memory,
and its total free memory at assembler execution time, as well as any other
information you think might be useful.
The command line you used for the assembler, linker, or other MASM tool
that was running when the problem occurred.
Any object files or libraries you linked with if the problem occurred at link
time.

If your program is very large, reduce it to the smallest possible program that still
produces the problem.
Note the circumstances of the error and notify Microsoft Corporation by
following the instructions in the section “Microsoft Support Services” in the
introduction to Environment and Tools. If you have comments or suggestions
regarding any of the books accompanying this product, please indicate them on
the Document Feedback page at the back of this book and send it to Microsoft.
If you have not yet registered your copy of the Macro Assembler, you should
fill out and return the Registration Card. This enables Microsoft to keep you
informed of updates and other information about the assembler.

Filename: LMAPGINT.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 33 Page: 20 of 8 Printed: 10/02/00 04:20 PM

C H A P T E R

Understanding Global Concepts

With the development of the Microsoft Macro Assembler (MASM) version 6.1,
you now have more options available to you for approaching a programming
task. This chapter explains the general concepts of programming in assembly
language, beginning with the environment and a review of the components you
need to work in the assembler environment. Even if you are familiar with
previous versions of MASM, you should examine this chapter for information
on new terms and features.
The first section of this chapter reviews available processors and operating
systems and how they work together. The section also discusses segmented
architecture and how it affects a protected-mode operating environment such as
Windows.
The second section describes some of the language components of MASM that
are common to most programs, such as reserved words, constant expressions,
operators, and registers. The remainder of this book was written with the
assumption that you understand the information presented in this section.
The last section summarizes the assembly process, from assembling a program
through running it. You can affect this process by the way you develop your
code. Finally, this section explores how you can change the assembly process
with the OPTION directive and conditional assembly.

The Processing Environment
The processing environment for MASM 6.1 includes the processor on which
your programs run, the operating system your programs use, and the aspects of
the segmented architecture that influence the choice of programming models.
This section summarizes these elements of the environment and how they affect
your programming choices.

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 1 of 1

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Programmer’s Guide

8086-Based Processors
The 8086 “family” of processors uses segments to control data and code. The
later 8086-based processors have larger instruction sets and more memory
capacity, but they still support the same segmented architecture. Knowing the
differences between the various 8086-based processors can help you select the
appropriate target processor for your programs.
The instruction set of the 8086 processor is upwardly compatible with its
successors. To write code that runs on the widest number of machines, select
the 8086 instruction set. By using the instruction set of a more advanced
processor, you increase the capabilities and efficiency of your program, but you
also reduce the number of systems on which the program can run.
Table 1.1 lists modes, memory, and segment size of processors on which your
application may need to run. Each processor is discussed in more detail
following.
Table 1.1

8086 Family of Processors

Processor

Available
Modes

Addressable
Memory

Segment
Size

8086/8088

Real

1 megabyte

16 bits

80186/80188

Real

1 megabyte

16 bits

80286

Real and Protected

16 megabytes

16 bits

80386

Real and Protected

4 gigabytes

16 or 32 bits

80486

Real and Protected

4 gigabytes

16 or 32 bits

Processor Modes
Real mode allows only one process to run at a time. The mode gets its name
from the fact that addresses in real mode always correspond to real locations in
memory. The MS-DOS operating system runs in real mode.
Windows 3.1 operates only in protected mode, but runs MS-DOS programs in
real mode or in a simulation of real mode called virtual-86 mode. In protected
mode, more than one process can be active at any one time. The operating
system protects memory belonging to one process from access by another
process; hence the name protected mode.
Protected-mode addresses do not correspond directly to physical memory.
Under protected-mode operating systems, the processor allocates and manages
memory dynamically. Additional privileged instructions initialize protected mode
and control multiple processes. For more information, see “Operating Systems,”
following.

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 2 of 2

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

8086 and 8088
The 8086 is faster than the 8088 because of its 16-bit data bus; the 8088 has
only an 8-bit data bus. The 16-bit data bus allows you to use EVEN and
ALIGN on an 8086 processor to word-align data and thus improve datahandling efficiency. Memory addresses on the 8086 and 8088 refer to actual
physical addresses.

80186 and 80188
These two processors are identical to the 8086 and 8088 except that new
instructions have been added and several old instructions have been optimized.
These processors run significantly faster than the 8086.

80286
The 80286 processor adds some instructions to control protected mode, and it
runs faster. It also provides protected mode services, allowing the operating
system to run multiple processes at the same time. The 80286 is the minimum
for running Windows 3.1 and 16-bit versions of OS/2 ®.

80386
Unlike its predecessors, the 80386 processor can handle both 16-bit and 32-bit
data. It supports the entire instruction set of the 80286, and adds several new
instructions as well. Software written for the 80286 runs unchanged on the
80386, but is faster because the chip operates at higher speeds.
The 80386 implements many new hardware-level features, including paged
memory, multiple virtual 8086 processes, addressing of up to 4 gigabytes of
memory, and specialized debugging registers. Thirty-two–bit operating systems
such as Windows NT and OS/2 2.0 can run only on an 80386 or higher
processor.

80486
The 80486 processor is an enhanced version of the 80386, with instruction
“pipelining” that executes many instructions two to three times faster. The chip
incorporates both a math coprocessor and an 8K (kilobyte) memory cache. (The
math coprocessor is disabled on a variation of the chip called the 80486SX.)
The 80486 includes new instructions and is fully compatible with 80386
software.

8087, 80287, and 80387
These math coprocessors work concurrently with the 8086 family of processors.
Performing floating-point calculations with math coprocessors is up to 100 times
faster than emulating the calculations with integer instructions. Although there
are technical and performance differences among the three coprocessors, the
main difference to the applications programmer is that the 80287 and 80387 can

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 3 of 3

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Programmer’s Guide

operate in protected mode. The 80387 also has several new instructions. The
80486 does not use any of these coprocessors; its floating-point processor is
built in and is functionally equivalent to the 80387.

Operating Systems
With MASM, you can create programs that run under MS-DOS, Windows, or
Windows NT — or all three, in some cases. For example, ML.EXE can produce
executable files that run in any of the target environments, regardless of the
programmer’s environment. For information on building programs for different
environments, see “Building and Running Programs” in Help for PWB.
MS-DOS and Windows 3.1 provide different processing modes. MS-DOS runs
in the single-process real mode. Windows 3.1 operates in protected mode,
allowing multiple processes to run simultaneously.
Although Windows requires another operating system for loading and file
services, it provides many functions normally associated with an operating
system. When an application requests an MS-DOS service, Windows often
provides the service without invoking MS-DOS. For consistency, this book
refers to Windows as an operating system.
MS-DOS and Windows (in protected mode) differ primarily in system access
methods, size of addressable memory, and segment selection. Table 1.2
summarizes these differences.
Table 1.2

The MS-DOS and Windows Operating Systems Compared
Available
Active
Processes

Addressable
Memory

Direct to
hardware
and OS call

One

1 megabyte

Actual
address

16 bits

Windows
virtual-86 mode

Operating
system call

Multiple

1 megabyte

Segment
selectors

16 bits

Windows
protected mode

Operating
system call

Multiple

16 megabytes

Segment
selectors

16 bits

Windows NT

Operating
system call

Multiple

512
megabytes

Segment
selectors

32 bits

Operating
System

System
Access

MS-DOS and
Windows real
mode

Contents of
Segment
Register

Word
Length

MS-DOS
In real-mode programming, you can access system functions by calling MSDOS, calling the basic input/output system (BIOS), or directly addressing
hardware. Access is through MS-DOS Interrupt 21h.

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 4 of 4

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

Windows
As you can see in Table 1.2, protected mode allows for much larger data
structures than real mode, since addressable memory extends to 16 megabytes.
In protected mode, segment registers contain selector values rather than actual
segment addresses. These selectors cannot be calculated by the program; they
must be obtained by calling the operating system. Programs that attempt to
calculate segment values or to address memory directly do not work in protected
mode.
Protected mode uses privilege levels to maintain system integrity and security.
Programs cannot access data or code that is in a higher privilege level. Some
instructions that directly access ports or affect interrupts (such as CLI, STI, IN,
and OUT) are available at privilege levels normally used only by systems
programmers.
Windows protected mode provides each application with up to 16 megabytes of
“virtual memory,” even on computers that have less physical memory. The term
virtual memory refers to the operating system’s ability to use a swap area on the
hard disk as an extension of real memory. When a Windows application requires
more memory than is available, Windows writes sections of occupied memory
to the swap area, thus freeing those sections for other use. It then provides the
memory to the application that made the memory request. When the owner of
the swapped data regains control, Windows restores the data from disk to
memory, swapping out other memory if required.

Windows NT
Windows NT uses the so-called “flat model” of 80386/486 processors. This
model places the processor’s entire address space within one 32-bit segment.
The section “Defining Basic Attributes with .MODEL” in Chapter 2 explains
how to use the flat model. In flat model, your program can (in theory) access up
to 4 gigabytes of virtual memory. Since code, data, and stack reside in the same
segment, each segment register can hold the same value, which need never
change.

Segmented Architecture
The 8086 family of processors employs a segmented architecture — that is, each
address is represented as a segment and an offset. Segmented addresses affect
many aspects of assembly-language programming, especially addresses and
pointers.
Segmented architecture was originally designed to enable a 16-bit processor to
access an address space larger than 64K. (The section “Segmented Addressing,”
later in this chapter, explains how the processor uses both the segment and
offset to create addresses larger than 64K.) MS-DOS is an example of an
operating system that uses segmented architecture on a 16-bit processor.

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 5 of 5

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Programmer’s Guide

With the advent of protected-mode processors such as the 80286, segmented
architecture gained a second purpose. Segments can separate different blocks of
code and data to protect them from undesirable interactions. Windows takes
advantage of the protection features of the 16-bit segments on the 80286.
Segmented architecture went through another significant change with the release
of 32-bit processors, starting with the 80386. These processors are compatible
with the older 16-bit processors, but allow flat model 32-bit offset values up to 4
gigabytes. Offset values of this magnitude remove the memory limitations of
segmented architecture. The Windows NT operating system uses 32-bit
addressing.

Segment Protection
Segmented architecture is an important part of the Windows memory-protection
scheme. In a “multitasking” operating system in which numerous programs can
run simultaneously, programs cannot access the code and data of another
process without permission.
In MS-DOS, the data and code segments are usually allocated adjacent to each
other, as shown in Figure 1.1. In Windows, the data and code segments can be
anywhere in memory. The programmer knows nothing about, and has no
control over, their location. The operating system can even move the segments
to a new memory location or to disk while the program is running.

Figure 1.1

Segment Allocation

Segment protection makes software development easier and more reliable in
Windows than in MS-DOS, because Windows immediately detects illegal

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 6 of 6

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

memory accesses. The operating system intercepts illegal memory accesses,
terminates the program, and displays a message. This makes it easier for you to
track down and fix the bug.
Because it runs in real mode, MS-DOS contains no mechanism for detecting an
improper memory access. A program that overwrites data not belonging to it
may continue to run and even terminate correctly. The error may not surface
until later, when MS-DOS or another program reads the corrupted memory.

Segmented Addressing
Segmented addressing refers to the internal mechanism that combines a segment
value and an offset value to form a complete memory address. The two parts of
an address are represented as
segment:offset
The segment portion always consists of a 16-bit value. The offset portion is a
16-bit value in 16-bit mode or a 32-bit value in 32-bit mode.
In real mode, the segment value is a physical address that has an arithmetic
relationship to the offset value. The segment and offset together create a 20-bit
physical address (explained in the next section). Although 20-bit addresses can
access up to 1 megabyte of memory, the BIOS and operating system on
International Standard Architecture (IBM PC/AT and compatible) computers
use part of this memory, leaving the remainder available for programs.

Segment Arithmetic
Manipulating segment and offset addresses directly in real-mode programming is
called “segment arithmetic.” Programs that perform segment arithmetic are not
portable to protected-mode operating systems, in which addresses do not
correspond to a known segment and offset.
To perform segment arithmetic successfully, it helps to understand how the
processor combines a 16-bit segment and a 16-bit offset to form a 20-bit linear
address. In effect, the segment selects a 64K region of memory, and the offset
selects the byte within that region. Here’s how it works:
1. The processor shifts the segment address to the left by four binary places,
producing a 20-bit address ending in four zeros. This operation has the effect
of multiplying the segment address by 16.
2. The processor adds this 20-bit segment address to the 16-bit offset address.
The offset address is not shifted.
3. The processor uses the resulting 20-bit address, called the “physical
address,” to access an actual location in the 1-megabyte address space.

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 7 of 7

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Programmer’s Guide

Figure 1.2 illustrates this process.

Figure 1.2

Calculating Physical Addresses

A 20-bit physical address may actually be specified by 4,096 equivalent
segment:offset addresses. For example, the addresses 0000:F800, 0F00:0800,
and 0F80:0000 all refer to the same physical address 0F800.

Language Components of MASM
Programming with MASM requires that you understand the MASM concepts of
reserved words, identifiers, predefined symbols, constants, expressions,
operators, data types, registers, and statements. This section defines important
terms and provides lists that summarize these topics. For detailed information,
see Help or the Reference.

Reserved Words
A reserved word has a special meaning fixed by the language. You can use it
only under certain conditions. Reserved words in MASM include:
u
u
u
u

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 8 of 8

Instructions, which correspond to operations the processor can execute.
Directives, which give commands to the assembler.
Attributes, which provide a value for a field, such as segment alignment.
Operators, which are used in expressions.

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts
u

Predefined symbols, which return information to your program.

MASM reserved words are not case sensitive except for predefined symbols
(see “Predefined Symbols,” later in this chapter).
The assembler generates an error if you use a reserved word as a variable, code
label, or other identifier within your source code. However, if you need to use a
reserved word for another purpose, the OPTION NOKEYWORD directive can
selectively disable a word’s status as a reserved word.
For example, to remove the STR instruction, the MASK operator, and the
NAME directive from the set of words MASM recognizes as reserved, use this
statement in the code segment of your program before the first reference to
STR, MASK, or NAME:
OPTION NOKEYWORD:

The section “Using the OPTION Directive,” later in this chapter, discusses the
OPTION directive. Appendix D provides a complete list of MASM reserved
words.
With the /Zm command-line option or OPTION M510 in effect, MASM does
not reserve any operators or instructions that do not apply to the current CPU
mode. For example, you can use the symbol ENTER when assembling under the
default CPU mode but not under .286 mode, since the 80186/486 processors
recognize ENTER as an instruction. The USE32, FLAT, FAR32, and NEAR32
segment types and the 80386/486 register names are not keywords with
processors other than the 80386/486.

Identifiers
An identifier is a name that you invent and attach to a definition. Identifiers can
be symbols representing variables, constants, procedure names, code labels,
segment names, and user-defined data types such as structures, unions, records,
and types defined with TYPEDEF. Identifiers longer than 247 characters
generate an error.
Certain restrictions limit the names you can use for identifiers. Follow these
rules to define a name for an identifier:
u

The first character of the identifier can be an alphabetic character (A–Z) or
any of these four characters: @ _ $ ?
The other characters in the identifier can be any of the characters listed
above or a decimal digit (0–9).

Avoid starting an identifier with the at sign (@), because MASM 6.1 predefines
some special symbols starting with @ (see “Predefined Symbols,” following).

Filename: LMAPGC01.DOC
Template: MSGRIDA1.DOT
Revision #: 57 Page: 9 of 9

Project:
Author: Terri Sharkey Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:24 PM

Programmer’s Guide

Beginning an identifier with @ may also cause conflicts with future versions of
the Macro Assembler.
The symbol — and thus the identifier — is visible as long as it remains within
scope. (For more information about visibility and scope, see “Sharing Symbols
with Include Files” in Chapter 8.)

Predefined Symbols
The assembler includes a number of predefined symbols (also called predefined
equates). You can use these symbol names at any point in your code to
represent the equate value. For example, the predefined equate @FileName
represents the base name of the current file. If the current source file is
TASK.ASM, the value of @FileName is TASK. The MASM predefined
symbols are listed according to the kinds of information they provide. Case is
important only if the /Cp option is used. (For additional details, see Help on ML
command-line options.)
The predefined symbols for segment information include:
Symbol

Description

@code

Returns the name of the code segment.

@CodeSize

Returns an integer representing the default code distance.

@CurSeg

Returns the name of the current segment.

@data

Expands to DGROUP.

@DataSize

Returns an integer representing the default data distance.

@fardata

Returns the name of the segment defined by the .FARDATA directive.

@fardata?

Returns the name of the segment defined by the .FARDATA? directive.

@Model

Returns the selected memory model.

@stack

Expands to DGROUP for near stacks or STACK for far stacks. (See
“Creating a Stack” in Chapter 2.)

@WordSize

Provides the size attribute of the current segment.

The predefined symbols for environment information include:
Symbol

Description

@Cpu

Contains a bit mask specifying the processor mode.

@Environ

Returns values of environment variables during assembly.

@Interface

Contains information about the language parameters.

@Version

Represents the text equivalent of the MASM version number. In MASM 6.1,
this expands to 610.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 10 of 10 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

The predefined symbols for date and time information include:
Symbol

Description

@Date

Supplies the current system date during assembly.

@Time

Supplies the current system time during assembly.

The predefined symbols for file information include:
Symbol

Description

@FileCur

Names the current file (base and suffix).

@FileName

Names the base name of the main file being assembled as it appears on the
command line.

@Line

Gives the source line number in the current file.

The predefined symbols for macro string manipulation include:
Symbol

Description

@CatStr

Returns concatenation of two strings.

@InStr

Returns the starting position of a string within another string.

@SizeStr

Returns the length of a given string.

@SubStr

Returns substring from a given string.

Integer Constants and Constant Expressions
An integer constant is a series of one or more numerals followed by an optional
radix specifier. For example, in these statements
mov
mov

ax, 25
bx, 0B3h

the numbers 25 and 0B3h are integer constants. The h appended to 0B3 is a
radix specifier. The specifiers are:
u

y for binary (or b if the default radix is not hexadecimal)

o or q for octal

t for decimal (or d if the default radix is not hexadecimal)

h for hexadecimal

Radix specifiers can be either uppercase or lowercase letters; sample code in this
book is in lowercase. If you do not specify a radix, the assembler interprets the
integer according to the current radix. The default radix is decimal, but you can
change the default with the .RADIX directive.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 11 of 11 Printed: 10/02/00 04:24 PM

Programmer’s Guide

Hexadecimal numbers must always start with a decimal digit (0–9). If necessary,
add a leading zero to distinguish between symbols and hexadecimal numbers
that start with a letter. For example, MASM interprets ABCh as an identifier.
The hexadecimal digits A through F can be either uppercase or lowercase letters.
Sample code in this book is in uppercase letters.
Constant expressions contain integer constants and (optionally) operators such
as shift, logical, and arithmetic operators. The assembler evaluates constant
expressions at assembly time. (In addition to constants, expressions can contain
labels, types, registers, and their attributes.) Constant expressions do not change
value during program execution.

Symbolic Integer Constants
You can define symbolic integer constants with either of the data assignment
directives, EQU or the equal sign (=). These directives assign values to symbols
during assembly, not during program execution. Symbolic constants are used to
assign names to constant values. You can use a symbol with an assigned value
in place of an immediate operand. For example, instead of referring in your code
to keyboard scan codes with numbers such as 30 or 48, you can create more
recognizable symbols:
SCAN_A
SCAN_B

EQU
EQU

30
48

then use the appropriate symbol in your program rather than the number. Using
symbolic constants instead of undescriptive numbers makes your code more
readable and easier to maintain. The assembler does not allocate data storage
when you use either EQU or =. It simply replaces each occurrence of the
symbol with the value of the expression.
The directives EQU and = have slightly different purposes. Integers defined
with the = directive can be redefined with another value in your source code,
but those defined with EQU cannot. Once you’ve defined a symbolic constant
with the EQU directive, attempting to redefine it generates an error. The syntax
is:
symbol EQU expression
The symbol is a unique name of your choice, except for words reserved by
MASM. The expression can be an integer, a constant expression, a one- or twocharacter string constant (four-character on the 80386/486), or an expression
that evaluates to an address. Symbolic constants let you change a constant value
used throughout your source code by merely altering expression in the
definition. This removes the potential for error and saves you the inconvenience
of having to find and replace each occurrence of the constant in your program.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 12 of 12 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

The following example shows the correct use of EQU to define symbolic
integers.
column
row
screen
line

EQU
EQU
EQU
EQU

80
25
column * row
row

;
;
;
;

Constant 80
Constant 25
Constant - 2000
Constant 25

.DATA
.CODE
.
.
.
mov
mov

cx, column
bx, line

The value of a symbol defined with the = directive can be different at different
places in the source code. However, a constant value is assigned during
assembly for each use, and that value does not change at run time.
The syntax for the = directive is:
symbol = expression

Size of Constants
The default word size for MASM 6.1 expressions is 32 bits. This behavior can
be modified using OPTION EXPR16 or OPTION M510. Both of these options
set the expression word size to 16 bits, but OPTION M510 affects other
assembler behavior as well (see Appendix A).
It is illegal to change the expression word size once it has been set with
OPTION M510, OPTION EXPR16, or OPTION EXPR32. However, you
can repeat the same directive in your source code as often as you wish. You can
place the same directive in every include file, for example.

Operators
Operators are used in expressions. The value of the expression is determined at
assembly time and does not change when the program runs.
Operators should not be confused with processor instructions. The reserved
word ADD is an instruction; the plus sign (+) is an operator. For example,
Amount+2 illustrates a valid use of the plus operator (+). It tells the assembler
to add 2 to the constant value Amount, which might be a value or an address.
Contrast this operation, which occurs at assembly time, with the processor’s
ADD instruction. ADD tells the processor at run time to add two numbers and
store the result.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 13 of 13 Printed: 10/02/00 04:24 PM

Programmer’s Guide

The assembler evaluates expressions that contain more than one operator
according to the following rules:
u

Operations in parentheses are performed before adjacent operations.

Binary operations of highest precedence are performed first.

Operations of equal precedence are performed from left to right.

Unary operations of equal precedence are performed right to left.

Table 1.3 lists the order of precedence for all operators. Operators on the same
line have equal precedence.
Table 1.3

Operator Precedence

Precedence

Operators

( ), [ ]

LENGTH, SIZE, WIDTH, MASK, LENGTHOF, SIZEOF

. (structure-field-name operator)

: (segment-override operator), PTR

LROFFSET, OFFSET, SEG, THIS , TYPE

HIGH, HIGHWORD, LOW, LOWWORD

+ ,– (unary)

*, /, MOD, SHL, SHR

+, – (binary)

EQ, NE, LT, LE, GT, GE

NOT

AND

OR, XOR

OPATTR, SHORT, .TYPE

Data Types
A “data type” describes a set of values. A variable of a given type can have any
of a set of values within the range specified for that type.
The intrinsic types for MASM 6.1 are BYTE, SBYTE, WORD, SWORD,
DWORD, SDWORD, FWORD, QWORD, and TBYTE. These types define
integers and binary coded decimals (BCDs), as discussed in Chapter 6. The
signed data types SBYTE, SWORD, and SDWORD work in conjunction with
directives such as INVOKE (for calling procedures) and .IF (introduced in
Chapter 7). The REAL4, REAL8, and REAL10 directives define floating-point
types. (See Chapter 6.)

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 14 of 14 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

Versions of MASM prior to 6.0 had separate directives for types and initializers.
For example, BYTE is a type and DB is the corresponding initializer. The
distinction does not apply in MASM 6.1. You can use any type (intrinsic or
user-defined) as an initializer.
MASM does not have specific types for arrays and strings. However, you can
treat a sequence of data units as arrays, and character or byte sequences as
strings. (See “Arrays and Strings” in Chapter 5.)
Types can also have attributes such as langtype and distance (NEAR and
FAR). For information on these attributes, see “Declaring Parameters with the
PROC Directive” in Chapter 7.
You can also define your own types with STRUCT, UNION, and RECORD.
The types have fields that contain string or numeric data, or records that contain
bits. These data types are similar to the user-defined data types in high-level
languages such as C, Pascal, and FORTRAN. (See Chapter 5, “Defining and
Using Complex Data Types.”)
You can define new types, including pointer types, with the TYPEDEF
directive. TYPEDEF assigns a qualifiedtype (explained in the following) to a
typename of your choice. This lets you build new types with descriptive names
of your choosing, making your programs more readable. For example, the
following statement makes the symbol CHAR a synonym for the intrinsic type
BYTE:
CHAR

TYPEDEF BYTE

The qualifiedtype is any type or pointer to a type of the form:
[[distance]] PTR [[qualifiedtype]]
where distance is NEAR, FAR, or any distance modifier. (For more
information on distance, see “Declaring Parameters with the PROC Directive”
in Chapter 7.)
The qualifiedtype can also be any type previously defined with TYPEDEF. For
example, if you use TYPEDEF to create an alias for BYTE — say, CHAR as in
the preceding example — you can use CHAR as a qualifiedtype when defining the
pointer type PCHAR, like this:
CHAR
PCHAR

TYPEDEF BYTE
TYPEDEF PTR CHAR

The typename CHAR in the first line becomes a qualifiedtype in the second line.
Use of the TYPEDEF directive to define pointers is explained in “Accessing
Data with Pointers and Addresses” in Chapter 3.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 15 of 15 Printed: 10/02/00 04:24 PM

Programmer’s Guide

Since distance and qualifiedtype are optional syntax elements, you can use
variables of type PTR or FAR PTR. You can also define procedure prototypes
with qualifiedtype. For more information about procedure prototypes, see
“Declaring Procedure Prototypes” in Chapter 7.
These rules govern the use of qualifiedtype:
u

The only component of a qualifiedtype definition that can be forwardreferenced is a structure or union type identifier.

If you do not specify distance, the assembler assumes a distance that
corresponds to the memory model. The assumed distance is NEAR for tiny,
small, and medium models, and FAR for other models.
If you do not specify a memory model with .MODEL, the assembler
assumes SMALL model (and therefore NEAR pointers).

You can use a qualifiedtype in seven places:
Use

Example

In procedure arguments

proc1 PROC pMsg:PTR BYTE

In prototype arguments

proc2 PROTO pMsg:FAR PTR WORD

With local variables declared inside
procedures

LOCAL pMsg:PTR

With the LABEL directive

TempMsg LABEL PTR WORD

With the EXTERN and EXTERNDEF
directives

EXTERN pMsg:FAR PTR BYTE
EXTERNDEF MyProc:PROTO

With the COMM directive

COMM var1:WORD:3

With the TYPEDEF directive

PBYTE TYPEDEF PTR BYTE
PFUNC TYPEDEF PROTO MyProc

“Defining Pointer Types with TYPEDEF” in Chapter 3 shows ways to write a
TYPEDEF type for a qualifiedtype. Attributes such as NEAR and FAR can
also apply to a qualifiedtype.
You can determine an accurate definition for TYPEDEF and qualifiedtype from
the BNF grammar definitions given in Appendix B. The BNF grammar defines
each component of the syntax for any directive, showing the recursive
properties of components such as qualifiedtype.

Registers
The 8086 family of processors have the same base set of 16-bit registers. Each
processor can treat certain registers as two separate 8-bit registers. The
80386/486 processors have extended 32-bit registers. To maintain compatibility

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 16 of 16 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

with their predecessors, 80386/486 processors can access their registers as 16bit or, where appropriate, as 8-bit values.
Figure 1.3 shows the registers common to all the 8086-based processors. Each
register has its own special uses and limitations.

Figure 1.3

Registers for 8088 – 80286 Processors

80386/486 Only
The 80386/486 processors use the same 8-bit and 16-bit registers used by the
rest of the 8086 family. All of these registers can be further extended to 32 bits,
except segment registers, which always occupy 16 bits. The extended register
names begin with the letter “E.” For example, the 32-bit extension of AX is
EAX. The 80386/486 processors have two additional segment registers, FS and
GS. Figure 1.4 shows the extended registers of the 80386/486.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 17 of 17 Printed: 10/02/00 04:24 PM

Programmer’s Guide

Figure 1.4

Extended Registers for the 80386/486 Processors

Segment Registers
At run time, all addresses are relative to one of four segment registers: CS, DS,
SS, or ES. (The 80386/486 processors add two more: FS and GS.) These
registers, their segments, and their purposes include:
Register and Segment

Purpose

CS (Code Segment)

Contains processor instructions and their immediate operands.

DS (Data Segment)

Normally contains data allocated by the program.

SS (Stack Segment)

Contains the program stack for use by PUSH , POP, CALL,
and RET.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 18 of 18 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts
Register and Segment

Purpose

ES (Extra Segment)

References secondary data segment. Used by string
instructions.

FS, GS

Provides extra segments on the 80386/486.

General-Purpose Registers
The AX, DX, CX, BX, BP, DI, and SI registers are 16-bit general-purpose
registers, used for temporary data storage. Since the processor accesses registers
more quickly than it accesses memory, you can make your programs run faster
by keeping the most-frequently used data in registers.
The 8086-based processors do not perform memory-to-memory operations. For
example, the processor cannot directly copy a variable from one location in
memory to another. You must first copy from memory to a register, then from
the register to the new memory location. Similarly, to add two variables in
memory, you must first copy one variable to a register, then add the contents of
the register to the other variable in memory.
The processor can access four of the general registers — AX, DX, CX, and BX
— either as two 8-bit registers or as a single 16-bit register. The AH, DH, CH,
and BH registers represent the high-order 8 bits of the corresponding registers.
Similarly, AL, DL, CL, and BL represent the low-order 8 bits of the registers.
The 80386/486 processors can extend all the general registers to 32 bits, though
as Figure 1.4 shows, you cannot treat the upper 16 bits as a separate register as
you can the lower 16 bits. To use EAX as an example, you can directly
reference the low byte as AL, the next lowest byte as AH, and the low word as
AX. To access the high word of EAX, however, you must first shift the upper
16 bits into the lower 16 bits.

Special-Purpose Registers
The 8086 family of processors has two additional registers, SP and IP, whose
values are changed automatically by the processor.

SP (Stack Pointer)
The SP register points to the current location within the stack segment. Pushing
a value onto the stack decreases the value of SP by two; popping from the stack
increases the value of SP by two. Thirty-two–bit operands on 80386/486
processors increase or decrease SP by four instead of two. The CALL and INT
instructions store the return address on the stack and reduce SP accordingly.
Return instructions retrieve the stored address from the stack and reset SP to its
value before the call. SP can also be adjusted with instructions such as ADD.
The program stack is described in detail in Chapter 3.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 19 of 19 Printed: 10/02/00 04:24 PM

Programmer’s Guide

IP (Instruction Pointer)
The IP register always contains the address of the next instruction to be
executed. You cannot directly access or change the instruction pointer.
However, instructions that control program flow (such as calls, jumps, loops,
and interrupts) automatically change the instruction pointer.

Flags Register
The 16 bits in the flags register control the execution of certain instructions and
reflect the current status of the processor. In 80386/486 processors, the flags
register is extended to 32 bits. Some bits are undefined, so there are actually 9
flags for real mode, 11 flags (including a 2-bit flag) for 80286 protected mode,
13 for the 80386, and 14 for the 80486. The extended flags register of the
80386/486 is sometimes called “Eflags.”
Figure 1.5 shows the bits of the 32-bit flags register for the 80386/486. Earlier
8086-family processors use only the lower word. The unmarked bits are
reserved for processor use, and should not be modified.

Figure 1.5 Flags for 8088-80486 Processors

In the following descriptions and throughout this book, “set” means a bit value
of 1, and “cleared” means the bit value is 0. The nine flags common to all 8086family processors, starting with the low-order flags, include:

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 20 of 20 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

Flag

Description

Carry

Set if an operation generates a carry to or a borrow from a destination
operand.

Parity

Set if the low-order bits of the result of an operation contain an even
number of set bits.

Auxiliary Carry

Set if an operation generates a carry to or a borrow from the low-order 4
bits of an operand. This flag is used for binary coded decimal (BCD)
arithmetic.

Zero

Set if the result of an operation is 0.

Sign

Equal to the high-order bit of the result of an operation (0 is positive, 1 is
negative).

Trap

If set, the processor generates a single-step interrupt after each
instruction. A debugging program can use this feature to execute a
program one instruction at a time.

Interrupt Enable

If set, interrupts are recognized and acted on as they are received. The bit
can be cleared to turn off interrupt processing temporarily.

Direction

If set, string operations process down from high addresses to low
addresses. If cleared, string operations process up from low addresses to
high addresses.

Overflow

Set if the result of an operation is too large or small to fit in the destination
operand.

Although all flags serve a purpose, most programs require only the carry, zero,
sign, and direction flags.

Statements
Statements are the line-by-line components of source files. Each MASM
statement specifies an instruction or directive for the assembler. Statements have
up to four fields, as shown here:
[[name:]] [[operation]] [[operands]] [[;comment]]
The following list explains each field:
Field

Purpose

name

Labels the statement, so that instructions elsewhere in the program can refer to
the statement by name. The name field can label a variable, type, segment, or
code location.

operation

Defines the action of the statement. This field contains either an instruction or an
assembler directive.

operands

Lists one or more items on which the instruction or directive operates.

comment

Provides a comment for the programmer. Comments are for documentation
only; they are ignored by the assembler.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 21 of 21 Printed: 10/02/00 04:24 PM

Programmer’s Guide

The following line contains all four fields:
mainlp: mov

ax, 7

; Load AX with the value 7

Here, mainlp is the label, mov is the operation, and ax and 7 are the operands,
separated by a comma. The comment follows the semicolon.
All fields are optional, although certain directives and instructions require an
entry in the name or operand field. Some instructions and directives place
restrictions on the choice of operands. By default, MASM is not case sensitive.
Each field (except the comment field) must be separated from other fields by
white-space characters (spaces or tabs). MASM also requires code labels to be
followed by a colon, operands to be separated by commas, and comments to be
preceded by a semicolon.
A logical line can contain up to 512 characters and occupy one or more physical
lines. To extend a logical line into two or more physical lines, put the backslash
character (\) as the last non-whitespace character before the comment or end of
the line. You can place a comment after the backslash as shown in this example:
.IF
&&
&&
mov
.ENDIF

(x > 0)
(ax > x)
(cx == 0)
dx, 20h

\
\

; X must be positive
; Result from function must be > x
; Check loop counter, too

Multiline comments can also be specified with the COMMENT directive. The
assembler ignores all text and code between the delimiters or on the same line as
the delimiters. This example illustrates the use of COMMENT.
COMMENT ^
^

mov

ax, 1

The assembler
ignores this text
and this code

The Assembly Process
Creating and running an executable file involves four steps:
1. Assembling the source code into an object file
2. Linking the object file with other modules or libraries into an executable
program
3. Loading the program into memory
4. Running the program

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 22 of 22 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

Once you have written your assembly-language program, MASM provides
several options for assembling it. The OPTION directive has several different
arguments that let you control the way MASM assembles your programs.
Conditional assembly allows you to create one source file that can generate a
variety of programs, depending on the status of various conditional-assembly
statements.

Generating and Running Executable Programs
This section briefly lists all the actions that take place during each of the
assembly steps. You can change the behavior of some of these actions in
various ways, such as using macros instead of procedures, or using the
OPTION directive or conditional assembly. The other chapters in this book
include specific programming methods; this section simply gives you an
overview.

Assembling
The ML.EXE program does two things to create an executable program. First, it
assembles the source code into an intermediate object file. Second, it calls the
linker, LINK.EXE, which links the object files and libraries into an executable
program.
At assembly time, the assembler:
u

u
u

u
u
u
u
u

Evaluates conditional-assembly directives, assembling if the conditions are
true.
Expands macros and macro functions.
Evaluates constant expressions such as MYFLAG AND 80H, substituting the
calculated value for the expression.
Encodes instructions and nonaddress operands. For example, mov cx, 13
can be encoded at assembly time because the instruction does not access
memory.
Saves memory offsets as offsets from their segments.
Places segments and segment attributes in the object file.
Saves placeholders for offsets and segments (relocatable addresses).
Outputs a listing if requested.
Passes messages (such as INCLUDELIB and .DOSSEG) directly to the
linker.

For information about conditional assembly, see “Conditional Directives” in this
chapter; for macros, see Chapter 9. Further details about segments and offsets
are included in Chapters 2 and 3. Assembly listings are explained in Appendix C.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 23 of 23 Printed: 10/02/00 04:24 PM

Programmer’s Guide

Linking
Once your source code is assembled, the resulting object file is passed to the
linker. At this point, the linker may combine several object files into an
executable program. The linker:
u

Combines segments according to the instructions in the object files,
rearranging the positions of segments that share the same class or group.

Fills in placeholders for offsets (relocatable addresses).

Writes relocations for segments into the header of .EXE files (but not .COM
files).

Writes the result as an executable program file.

Classes and groups are defined in “Defining Segment Groups” in Chapter 2.
Segments and offsets are explained in Chapter 3, “Using Addresses and
Pointers.”

Loading
After loading the executable file into memory, the operating system:
u
u
u
u
u

Creates the program segment prefix (PSP) header in memory.
Allocates memory for the program, based on the values in the PSP.
Loads the program.
Calculates the correct values for absolute addresses from the relocation table.
Loads the segment registers SS, CS, DS, and ES with values that point to the
proper areas of memory.

For information about segment registers, the instruction pointer (IP), and the
stack pointer (SP), see “Registers” earlier in this chapter. For more information
on the PSP see Help or an MS-DOS reference.

Running
To run your program, MS-DOS jumps to the program’s first instruction. Some
program operations, such as resolving indirect memory operands, cannot be
handled until the program runs. For a description of indirect references, see
“Indirect Operands” in Chapter 7.

Using the OPTION Directive
The OPTION directive lets you modify global aspects of the assembly process.
With OPTION, you can change command-line options and default arguments.
These changes affect only statements that follow the OPTION keyword.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 24 of 24 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

For example, you may have MASM code in which the first character of a
variable, macro, structure, or field name is a dot (.). Since a leading dot causes
MASM 6.1 to generate an error, you can use this statement in your program:
OPTION DOTNAME

This enables the use of the dot for the first character.
Changes made with OPTION override any corresponding command-line option.
For example, suppose you compile a module with this command line (which
enables M510 compatibility):
ML /Zm TEST.ASM

The assembler disables M510 compatibility options for all code following this
statement:
OPTION NOM510

The following lists explain each of the arguments for the OPTION directive.
Where appropriate, an underline identifies the default argument. If you wish to
place more than one OPTION statement on a line, separate them by commas.
Options for M510 compatibility include:
Argument

Description

CASEMAP: maptype

CASEMAP:NONE (or /Cx) causes internal

symbol recognition to be case sensitive and causes
the case of identifiers in the .OBJ file to be the same
as specified in the EXTERNDEF, PUBLIC, or
COMM statement. The default is
CASEMAP:NOTPUBLIC (or /Cp). It specifies
case insensitivity for internal symbol recognition and
the same behavior as CASEMAP:NONE for case
of identifiers in .OBJ files. CASEMAP:ALL (/Cu)
specifies case insensitivity for identifiers and
converts all identifier names to uppercase.
DOTNAME | NODOTNAME

Enables the use of the dot (.) as the leading

character in variable, macro, structure, union, and
member names.
M510 | NOM510

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 25 of 25 Printed: 10/02/00 04:24 PM

Sets all features to be compatible with MASM
version 5.1, disabling the SCOPED argument and
enabling OLDMACROS, DOTNAME, and,
OLDSTRUCTS. OPTION M510 conditionally
sets other arguments for the OPTION directive.
For more information on using OPTION M510,
see Appendix A.

Programmer’s Guide
Argument

Description

OLDMACROS | NOOLDMACROS

Enables the version 5.1 treatment of macros.
MASM 6.1 treats macros differently.

OLDSTRUCTS | NOOLDSTRUCTS

Enables compatibility with MASM 5.1 for treatment
of structure members. See Chapter 5 for
information on structures.

SCOPED | NOSCOPED

Guarantees that all labels inside procedures are
local to the procedure when SCOPED (the default)
is enabled.

SETIF2: TRUE | FALSE

If TRUE, .ERR2 statements and IF2 and
ELSEIF2 conditional blocks are evaluated on
every pass. If FALSE, they are not evaluated. If
SETIF2 is not specified (or implied), .ERR2, IF2 ,
and ELSEIF2 expressions cause an error. Both the
/Zm command-line argument and OPTION M510
imply SETIF2:TRUE.

Options for procedure use include:
Argument

Description

LANGUAGE : langtype

Specifies the default language type (C, PASCAL,
FORTRAN, BASIC, SYSCALL, or STDCALL)
to be used with PROC, EXTERN, and PUBLIC.
This use of the OPTION directive overrides the
.MODEL directive but is normally used when
.MODEL is not given.

EPILOGUE: macroname

Instructs the assembler to call the macroname to
generate a user-defined epilogue instead of the
standard epilogue code when a RET instruction is
encountered. See Chapter 7.

PROLOGUE: macroname

Instructs the assembler to call macroname to
generate a user-defined prologue instead of
generating the standard prologue code. See
Chapter 7.

PROC: visibility

Lets you explicitly set the default visibility as
PUBLIC, EXPORT, or PRIVATE.

Other options include:
Argument

Description

EXPR16 | EXPR32

Sets the expression word size to 16 or 32 bits. The
default is 32 bits. The M510 argument to the
OPTION directive sets the word size to 16 bits.
Once set with the OPTION directive, the
expression word size cannot be changed.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 26 of 26 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

Argument

Description

EMULATOR | NOEMULATOR

Controls the generation of floating-point
instructions.The NOEMULATOR option
generates the coprocessor instructions directly. The
EMULATOR option generates instructions with
special fixup records for the linker so that the
Microsoft floating-point emulator, supplied with
other Microsoft languages, can be used. It produces
the same result as setting the /Fpi command-line
option. You can set this option only once per
module.

LJMP | NOLJMP

Enables automatic conditional-jump lengthening.
For information about conditional-jump lengthening,
see Chapter 7.

NOKEYWORD:

Disables the specified reserved words. For an
example of the syntax for this argument, see
“Reserved Words” in this chapter.

NOSIGNEXTEND

Overrides the default sign-extended opcodes for the
AND, OR, and XOR instructions and generates the
larger non-sign-extended forms of these
instructions. Provided for compatibility with NEC
V25 and NEC V35 controllers.

OFFSET: offsettype

Determines the result of OFFSET operator fixups.
SEGMENT sets the defaults for fixups to be
segment-relative (compatible with MASM 5.1).
GROUP, the default, generates fixups relative to the
group (if the label is in a group). FLAT causes
fixups to be relative to a flat frame. (The .386 mode
must be enabled to use FLAT.) See Appendix A.

READONLY | NOREADONLY

Enables checking for instructions that modify code
segments, thereby guaranteeing that read-only code
segments are not modified. Same as the /p
command-line option of MASM 5.1, except that it
affects only segments with at least one assembly
instruction, not all segments. The argument is useful
for protected mode programs, where code
segments must remain read-only.

SEGMENT: segSize

Allows global default segment size to be set. Also
determines the default address size for external
symbols defined outside any segment. The segSize
can be USE16, USE32, or FLAT.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 27 of 27 Printed: 10/02/00 04:24 PM

Programmer’s Guide

Conditional Directives
MASM 6.1 provides conditional-assembly directives and conditional-error
directives. Conditional-assembly directives let you test for a specified condition
and assemble a block of statements if the condition is true. Conditional-error
directives allow you to test for a specified condition and generate an assembly
error if the condition is true.
Both kinds of conditional directives test assembly-time conditions, not run-time
conditions. You can test only expressions that evaluate to constants during
assembly. For a list of the predefined symbols often used in conditional
assembly, see “Predefined Symbols,” earlier in this chapter.

Conditional-Assembly Directives
The IF and ENDIF directives enclose the conditional statements. The optional
ELSEIF and ELSE blocks follow the IF directive. There are many forms of the
IF and ELSE directives. Help provides a complete list.
The following statements show the syntax for the IF directives. The syntax for
other condition-assembly directives follow the same form.
IF expression1
ifstatements
[[ELSEIF expression2
elseifstatements]]
[[ELSE
elsestatements]]
ENDIF
The statements within an IF block can be any valid instructions, including other
conditional blocks, which in turn can contain any number of ELSEIF blocks.
ENDIF ends the block.
MASM assembles the statements following the IF directive only if the
corresponding condition is true. If the condition is not true and the block
contains an ELSEIF directive, the assembler checks to see if the corresponding
condition is true. If so, it assembles the statements following the ELSEIF
directive. If no IF or ELSEIF conditions are satisfied, the assembler processes
only the statements following the ELSE directive.
For example, you may want to assemble a line of code only if your program
defines a particular variable. In this example,
IFDEF
buff
ENDIF

buffer
BYTE

buffer DUP(?)

the assembler allocates buff only if buffer has been previously defined.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 28 of 28 Printed: 10/02/00 04:24 PM

Chapter 1 Understanding Global Concepts

MASM 6.1 provides the directives IF1, IF2, ELSEIF1, and ELSIF2 to grant
assembly only on pass one or pass two. To use these directives, you must either
enable 5.1 compatibility (with the /Zm command-line switch or OPTION
M510) or set OPTION SETIF2:TRUE, as described in the previous section.
The following list summarizes the conditional-assembly directives:
The Directive

Grants Assembly If

IF expression

expression is true (nonzero)

IFE expression

expression is false (zero)

IFDEF name

name has been previously defined

IFNDEF name

name has not been previously defined

IFB argument*

argument is blank

IFNB argument*

argument is not blank

IFIDN[I] arg1, arg2*

arg1 equals arg2

IFDIF[I] arg1, arg2*

arg1 does not equal arg2
The optional I suffix (IFIDNI and IFDIFI) makes comparisons
insensitive to differences in case.

* Used only in macros.

Conditional-Error Directives
You can use conditional-error directives to debug programs and check for
assembly-time errors. By inserting a conditional-error directive at a key point in
your code, you can test assembly-time conditions at that point. You can also use
conditional-error directives to test for boundary conditions in macros.
Like other severe errors, those generated by conditional-error directives cause
the assembler to return a nonzero exit code. If MASM encounters a severe error
during assembly, it does not generate the object module.
For example, the .ERRNDEF directive produces an error if the program has not
defined a given label. In the following example, .ERRNDEF makes sure a label
called publevel actually exists.
.ERRNDEF
IF
PUBLIC
ELSE
PUBLIC
ENDIF

publevel
publevel LE 2
var1, var2
var1, var2, var3

The conditional-error directives use the syntax given in the previous section.
The following list summarizes the conditional-error directives. Note their close
correspondence with the previous list of conditional-assembly directives.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 29 of 29 Printed: 10/02/00 04:24 PM

Programmer’s Guide
The Directive

Generates an Error

.ERR

Unconditionally where it occurs in the source file. Usually placed
within a conditional-assembly block.

.ERRE expression

If expression is false (zero).

.ERRNZ expression

If expression is true (nonzero).

.ERRDEF name

If name has been defined.

.ERRNDEF name

If name has not been defined.

.ERRB argument*

If argument is blank.

.ERRNB argument*

If argument is not blank.

.ERRIDN[I] arg1, arg2*

If arg1 equals arg2.

.ERRDIF[I] arg1, arg2*

If arg1 does not equal arg2.
The optional I suffix (.ERRIDNI and .ERRDIFI) makes
comparisons insensitive to case.

* Used only in macros

Two special conditional-error directives, .ERR1 and .ERR2, generate an error
only on pass one or pass two. To use these directives, you must either enable
5.1 compatibility (with the /Zm command-line switch or OPTION M510) or set
OPTION SETIF2:TRUE, as described in the previous section.

Filename: LMAPGC01.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 57 Page: 30 of 30 Printed: 10/02/00 04:24 PM

C H A P T E R

Organizing Segments

Understanding segments is an essential part of programming in assembly
language. In the family of 8086-based processors, the term segment has two
meanings:
u

A block of memory of discrete size, called a “physical segment.” The
number of bytes in a physical memory segment is 64K for 16-bit processors
or 4 gigabytes for 32-bit processors.
A variable-sized block of memory, called a “logical segment,” occupied by a
program’s code or data.

As you read this chapter, the distinction between the two definitions will become
clear. The adjectives “physical” and “logical” are not often used when speaking
of segments. The beginning programmer is left to infer from context which
definition applies. Fortunately, this is not difficult, and a distinction is often not
required.
This chapter begins with a close look at physical memory segments. This lays
the foundation for understanding logical segments, which form the subject of
most of the following sections.
The section “Using Simplified Segment Directives” explains how to begin, end,
and organize segments. It also explains how to access far data and code with
simplified segment directives.
The next section, “Using Full Segment Definitions,” describes how to order,
combine, and divide segments, and how to use the SEGMENT directive to
define full segments. It also explains how to create a segment group so that you
can use one segment address to access all the data.
Most of the information in this chapter also applies to writing modules to be
called from other programs. Exceptions are noted when they apply. For more
information about multiple-module programming, see Chapter 8, “Sharing Data
and Procedures Among Modules and Libraries.”

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 31 of 1 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Physical Memory Segments
As explained in Chapter 1, a physical segment can begin only at memory
locations evenly divisible by 16, including address 0. Intel calls such locations
“paragraphs.” You can easily recognize a paragraph location because its
hexadecimal address always ends with 0, as in 10000h or 2EA70h. The
8086/286 processors allow segments 64K in size, the largest number 16 bits can
represent. The 80386/486 processors still adhere to the 64K limit when running
in real mode. In protected mode, however, they use 32-bit registers that can
hold addresses up to 4 gigabytes.
Segmented architecture presents certain hurdles for the assembly-language
programmer. For small programs, the limitations lose importance. Code and data
each occupy less than 64K and reside in individual segments. A simple offset
locates each variable or instruction within a segment.
Larger programs, however, must contend with problems of segmented memory
areas. If data occupies two or more segments, the program must specify both
segment and offset to access a variable. When the data forms a continuous
stream across segments — such as the text in a word processor’s workspace —
the problems become more acute. Whenever it adds or deletes text in the first
segment, the word processor must seamlessly move data back and forth over
the boundaries of each following segment.
The problem of segment boundaries disappears in the so-called flat address
space of 32-bit protected mode. Although segments still exist, they easily hold all
the code and data of the largest programs. Even a very large program becomes
in effect a small application, able to reach all code and data with a single offset
address.

Logical Segments
Logical segments contain the three components of a program: code, data, and
stack. MASM organizes the three parts for you so they occupy physical
segments of memory. The segment registers CS, DS, and SS contain the
addresses of the physical memory segments where the logical segments reside.
You can define segments in two ways: with simplified segment directives and
with full segment definitions. You can also use both kinds of segment definitions
in the same program.
Simplified segment directives hide many of the details of segment definition and
assume the same conventions used by Microsoft high-level languages. (See the
following section, “Using Simplified Segment Directives.”) The simplified
segment directives generate necessary code, specify segment attributes, and
arrange segment order.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 32 of 2 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

Full segment definitions require more complex syntax but provide more
complete control over how the assembler generates segments. (See “Using Full
Segment Definitions” later in this chapter.) If you use full segment definitions,
you must write code to handle all the tasks performed automatically by the
simplified segment directives.

Using Simplified Segment Directives
Structuring a MASM program using simplified segments requires use of several
directives to assign standard names, alignment, and attributes to the segments in
your program. These directives define the segments in such a way that linking
with Microsoft high-level languages is easy.
The simplified segment directives are .MODEL, .CODE, .CONST, .DATA,
.DATA?, .FARDATA, .FARDATA?, .STACK, .STARTUP, and .EXIT. The
following sections discuss these directives and the arguments they take.
MASM programs consist of modules made up of segments. Every program
written only in MASM has one main module, where program execution begins.
This main module can contain code, data, or stack segments defined with all of
the simplified segment directives. Any additional modules should contain only
code and data segments. Every module that uses simplified segments must,
however, begin with the .MODEL directive.
The following example shows the structure of a main module using simplified
segment directives. It uses the default processor (8086) and the default stack
distance (NEARSTACK). Additional modules linked to this main program
would use only the .MODEL, .CODE, and .DATA directives and the END
statement.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 33 of 3 Printed: 10/02/00 04:23 PM

Programmer’s Guide
; This is the structure of a main module
; using simplified segment directives
.MODEL small, c ; This statement is required before you
;
can use other simplified segment directives
.STACK

; Use default 1-kilobyte stack

.DATA

; Begin data segment
; Place data declarations here

.CODE
.STARTUP

; Begin code segment
; Generate start-up code
; Place instructions here

.EXIT
END

; Generate exit code

The .DATA and .CODE statements do not require any separate statements to
define the end of a segment. They close the preceding segment and then open a
new segment. The .STACK directive opens and closes the stack segment but
does not close the current segment. The END statement closes the last segment
and marks the end of the source code. It must be at the end of every module.

Defining Basic Attributes with .MODEL
The .MODEL directive defines the attributes that affect the entire module:
memory model, default calling and naming conventions, operating system, and
stack type. This directive enables use of simplified segments and controls the
name of the code segment and the default distance for procedures.
You must place .MODEL in your source file before any other simplified
segment directive. The syntax is:
.MODEL memorymodel [[, modeloptions ]]
The memorymodel field is required and must appear immediately after the
.MODEL directive. The use of modeloptions, which define the other attributes,
is optional. The modeloptions must be separated by commas. You can also use
equates passed from the ML command line to define the modeloptions.
The following list summarizes the memorymodel field and the modeloptions
fields, which specify language and stack distance:

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 34 of 4 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments
Field

Description

Memory model

TINY, SMALL, COMPACT, MEDIUM , LARGE , HUGE , or
FLAT. Determines size of code and data pointers. This field is

required.
Language

C, BASIC, FORTRAN, PASCAL, SYSCALL, or STDCALL. Sets

calling and naming conventions for procedures and public symbols.
Stack distance

NEARSTACK or FARSTACK. Specifying NEARSTACK groups

the stack segment into a single physical segment (DGROUP) along with
data. SS is assumed to equal DS. FARSTACK does not group the
stack with DGROUP; thus SS does not equal DS.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 35 of 5 Printed: 10/02/00 04:23 PM

Programmer’s Guide

You can use no more than one reserved word from each field. The following
examples show how you can combine various fields:
.MODEL
.MODEL

small
large, c, farstack

.MODEL

medium, pascal

; Small memory model
; Large memory model,
;
C conventions,
;
separate stack
; Medium memory model,
;
Pascal conventions,
;
near stack (default)

The next four sections give more detail on each field.

Defining the Memory Model
MASM supports the standard memory models used by Microsoft high-level
languages — tiny, small, medium, compact, large, huge, and flat. You specify the
memory model with attributes of the same name placed after the .MODEL
directive. With the exception of the flat model, which requires instructions
specific to the 80386/486, your choice of a memory model does not limit the
kind of instructions you can write. The memory model does, however, control
segment defaults and determine whether data and code are near or far by
default, as indicated in the following table.
Table 2.1 Attributes of Memory Models
Memory
Model

Default
Code

Default
Data

Operating
System

Data and Code
Combined

Tiny

Near

MS-DOS

Yes

Small

Near

MS-DOS, Windows

Medium

Far

Near

MS-DOS, Windows

Compact

Near

Far

MS-DOS, Windows

Large

Far

MS-DOS, Windows

Huge

Far

MS-DOS, Windows

Flat

Near

Windows NT

Yes

When writing assembler modules for a high-level language, you should use the
same memory model as the calling language. Choose the smallest memory
model available that can contain your data and code, since near references
operate more efficiently than far references.
The predefined symbol @Model returns the memory model, encoding memory
models as integers 1 through 7. For more information on predefined symbols,
see “Predefined Symbols” in Chapter 1. For an example of how to use them,
see Help.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 36 of 6 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

The seven memory models supported by MASM 6.1 fall into three groups,
described in the following paragraphs.

Small, Medium, Compact, Large, and Huge Models
The traditional memory models recognized by many languages are small,
medium, compact, large, and huge. Small model supports one data segment and
one code segment. All data and code are near by default. Large model supports
multiple code and multiple data segments. All data and code are far by default.
Medium and compact models are in-between. Medium model supports multiple
code and single data segments; compact model supports multiple data segments
and a single code segment.
Huge model implies individual data items larger than a single segment, but the
implementation of huge data items must be coded by the programmer. Since the
assembler provides no direct support for this feature, huge model is essentially
the same as large model.
In each of these models, you can override the default. For example, you can
make large data items far in small model, or internal procedures near in large
model.

Tiny Model
Tiny-model programs run only under MS-DOS. Tiny model places all data and
code in a single segment. Therefore, the total program file size can occupy no
more than 64K. The default is near for code and static data items; you cannot
override this default. However, you can allocate far data dynamically at run time
using MS-DOS memory allocation services.
Tiny model produces MS-DOS .COM files. Specifying .MODEL tiny
automatically sends the /TINY argument to the linker. Therefore, the /AT
argument is not necessary with .MODEL tiny. However, /AT does not insert a
.MODEL directive. It only verifies that there are no base or pointer fixups, and
sends /TINY to the linker.

Flat Model
The flat memory model is a nonsegmented configuration available in 32-bit
operating systems. It is similar to tiny model in that all code and data go in a
single 32-bit segment.
To write a flat model program, specify the .386 or .486 directive before
.MODEL FLAT. All data and code (including system resources) are in a single
32-bit segment. The operating system automatically initializes segment registers
at load time; you need to modify them only when mixing 16-bit and 32-bit
segments in a single application. CS, DS, ES, and SS all occupy the supergroup
FLAT. Addresses and pointers passed to system services are always 32-bit near
addresses and pointers.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 37 of 7 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Choosing the Language Convention
The language option facilitates compatibility with high-level languages by
determining the internal encoding for external and public symbol names, the
code generated for procedure initialization and cleanup, and the order that
arguments are passed to a procedure with INVOKE. It also facilitates
compatibility with high-level – language modules. The PASCAL, BASIC, and
FORTRAN conventions are identical. C and SYSCALL have the same calling
convention but different naming conventions. Functions in the Windows API
use the Pascal calling convention.
Procedure definitions (PROC) and high-level procedure calls (INVOKE)
automatically generate code consistent with the calling convention of the
specified language. The PROC, INVOKE, PUBLIC, and EXTERN directives
all use the naming convention of the language. These directives follow the
default language conventions from the .MODEL directive unless you
specifically override the default. Use of these directives is explained in
“Controlling Program Flow,” Chapter 7. You can also use the OPTION
directive to set the language type. (See “Using the OPTION Directive” in
Chapter 1.) Not specifying a language type in either the .MODEL, OPTION,
EXTERN, PROC, INVOKE, or PROTO statement causes the assembler to
generate an error.
The predefined symbol @Interface provides information about the language
parameters. For a description of the bit flags, see Help.
For more information on calling and naming conventions, see Chapter 12,
“Mixed-Language Programming.” For information about writing procedures and
prototypes, see Chapter 7, “Controlling Program Flow.” For information on
multiple-module programming, refer to Chapter 8, “Sharing Data and
Procedures Among Modules and Libraries.”

Setting the Stack Distance
The NEARSTACK keyword places the stack segment in the group DGROUP
along with the data segment. The .STARTUP directive then generates code to
adjust SS:SP so that SS (Stack Segment register) holds the same address as DS
(Data Segment register). If you do not use .STARTUP, you must make this
adjustment or your program may fail to run. (For information about startup
code, see “Starting and Ending Code with .STARTUP and .EXIT,” later in this
chapter.) In this case, you can use DS to access stack items (including
parameters and local variables) and SS to access near data. Furthermore, since
stack items share the same segment address as near data, you can reliably pass
near pointers to stack items.
The FARSTACK setting gives the stack a segment of its own. That is, SS does
not equal DS. The default stack type, NEARSTACK, is a convenient setting for

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 38 of 8 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

most programs. Use FARSTACK for special cases such as memory-resident
programs

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 39 of 9 Printed: 10/02/00 04:23 PM

Programmer’s Guide

and dynamic-link libraries (discussed in Chapters 10 and 11) when you cannot
assume that the caller’s stack is near. You can use the predefined symbol
@Stack to determine if the stack location is DGROUP (for near stacks) or
STACK (for far stacks).

Specifying a Processor and Coprocessor
MASM supports a set of directives for selecting processors and coprocessors.
Once you select a processor, you must use only the instruction set for that
processor. The default is the 8086 processor. If you always want your code to
run on this processor, you do not need to add any processor directives.
To enable a different processor mode and the additional instructions available on
that processor, use the directives .186, .286, .386, and .486. The instruction
timings on a listing (see Appendix C, “Generating and Reading Assembly
Listings”) correspond to whichever processor directive you select.
The .286P, .386P, and .486P directives enable the instructions available only at
higher privilege levels in addition to the normal instruction set for the given
processor. Generally, you don’t need privileged instructions unless you are
writing operating-systems code or device drivers.
In addition to enabling different instruction sets, the processor directives also
affect the behavior of extended language features. For example, the INVOKE
directive pushes arguments onto the stack. If the .286 directive is in effect,
INVOKE takes advantage of operations possible only on 80286 and later
processors.
Use the directives .8087 (the default), .287, .387, and .NO87 to select a math
coprocessor instruction set. The .NO87 directive turns off assembly of all
coprocessor instructions. Note that .486 also enables assembly of all
coprocessor instructions because the 80486 processor has a complete set of
coprocessor registers and instructions built into the chip. The processor
instructions imply the corresponding coprocessor directive. The coprocessor
directives are provided to override the defaults.

Creating a Stack
The stack is the section of memory used for pushing or popping registers and
storing the return address when a subroutine is called. The stack often holds
temporary and local variables.
If your main module is written in a high-level language, that language handles
the details of creating a stack. Use the .STACK directive only when you write a
main module in assembly language.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 40 of 10 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

The .STACK directive creates a stack segment. By default, the assembler
allocates 1K of memory for the stack. This size is sufficient for most small
programs.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 41 of 11 Printed: 10/02/00 04:23 PM

Programmer’s Guide

To create a stack of a size other than the default size, give .STACK a single
numeric argument indicating stack size in bytes:
.STACK

2048

; Use 2K stack

For a description of how stack memory is used with procedure calls and local
variables, see Chapter 7, “Controlling Program Flow.”

Creating Data Segments
Programs can contain both near and far data. In general, you should place
important and frequently used data in the near data area, where data access is
faster. This area can get crowded, however, because in 16-bit operating systems
the total amount of all near data in all modules cannot exceed 64K. Therefore,
you may want to place infrequently used or particularly large data items in a far
data segment.
The .DATA, .DATA?, .CONST, .FARDATA, and .FARDATA? directives
create data segments. You can access the various segments within DGROUP
without reloading segment registers (see “Defining Segment Groups,” later in
this chapter). These five directives also prevent instructions from appearing in
data segments by assuming CS to ERROR.

Near Data Segments
The .DATA directive creates a near data segment. This segment contains the
frequently used data for your program. It can occupy up to 64K in MS-DOS or
512 megabytes under flat model in Windows NT. It is placed in a special group
identified as DGROUP, which is also limited to 64K.
When you use .MODEL, the assembler automatically defines DGROUP for
your near data segment. The segments in DGROUP form near data, which can
normally be accessed directly through DS or SS.
You can also define the .DATA? and .CONST segments that go into DGROUP
unless you are using flat model. Although all of these segments (along with the
stack) are eventually grouped together and handled as data segments, .DATA?
and .CONST enhance compatibility with Microsoft high-level languages. In
Microsoft languages, .CONST is used to define constant data such as strings
and floating-point numbers that must be stored in memory. The .DATA?
segment is used for storing uninitialized variables. You can follow this
convention if you want. If you use C startup code, .DATA? is initialized to 0.
You can use @data to determine the group of the data segment and
@DataSize to determine the size of the memory model set by the .MODEL
directive. The predefined symbols @WordSize and @CurSeg return the size
attribute and name of the current segment, respectively. See “Predefined
Symbols” in Chapter 1.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 42 of 12 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

Far Data Segments
The compact, large, and huge memory models use far data addresses by default.
With these memory models, however, you can still construct data segments
using .DATA, .DATA?, and .CONST. The effect of these directives does not
change from one memory model to the next. They always contribute segments
to the default data area, DGROUP, which has a total limit of 64K.
When you use .FARDATA or .FARDATA? in the small and medium memory
models, the assembler creates far data segments FAR_DATA and FAR_BSS,
respectively. You can access variables with:
mov
mov

ax, SEG farvar2
ds, ax

For more information on far data, see “Near and Far Addresses” in Chapter 3.

Creating Code Segments
Whether you are writing a main module or a module to be called from another
module, you can have both near and far code segments. This section explains
how to use near and far code segments and how to use the directives and
predefined equates that relate to code segments.

Near Code Segments
The small memory model is often the best choice for assembly programs that
are not linked to modules in other languages, especially if you do not need more
than 64K of code. This memory model defaults to near (two-byte) addresses for
code and data, which makes the program run faster and use less memory.
When you use .MODEL and simplified segment directives, the .CODE
directive in your program instructs the assembler to start a code segment. The
next segment directive closes the previous segment; the END directive at the
end of your program closes remaining segments. The example at the beginning
of “Using Simplified Segment Directives,” earlier in this chapter, shows how to
do this.
You can use the predefined symbol @CodeSize to determine whether code
pointers default to NEAR or FAR.

Far Code Segments
When you need more than 64K of code, use the medium, large, or huge
memory model to create far segments.
The medium, large, and huge memory models use far code addresses by
default. In the larger memory models, the assembler creates a different code
segment for each module. If you use multiple code segments in the small,

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 43 of 13 Printed: 10/02/00 04:23 PM

Programmer’s Guide

compact, or tiny model, the linker combines the .CODE segments for all
modules into one segment.
For far code segments, the assembler names each code segment
MODNAME_TEXT, in which MODNAME is the name of the module. With
near code, the assembler names every code segment _TEXT, causing the linker
to concatenate these segments into one. You can override the default name by
providing an argument after .CODE. (For a complete list of segment names
generated by MASM, see Appendix E, “Default Segment Names.”)
With far code, a single module can contain multiple code segments. The .CODE
directive takes an optional text argument that names the segment. For instance,
the following example creates two distinct code segments, FIRST_TEXT and
SECOND_TEXT.
.CODE
.
.
.
.CODE
.
.
.

FIRST
; First set of instructions here
SECOND
; Second set of instructions here

Whenever the processor executes a far call or jump, it loads CS with the new
segment address. No special action is necessary other than making sure that you
use far calls and jumps. See “Near and Far Addresses” in Chapter 3.
Note The assembler always assumes that the CS register contains the address
of the current code segment or group.

Starting and Ending Code with .STARTUP and .EXIT
The easiest way to begin and end an MS-DOS program is to use the
.STARTUP and .EXIT directives in the main module. The main module
contains the starting point and usually the termination point. You do not need
these directives in a module called by another module.
These directives make MS-DOS programs easy to maintain. They automatically
generate code appropriate to the stack distance specified with .MODEL.
However, they do not apply to flat-model programs written for 32-bit operating
systems. Thus, you should not use .STARTUP or .EXIT in programs written
for Windows NT.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 44 of 14 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

To start a program, place the .STARTUP directive where you want execution to
begin. Usually, this location immediately follows the .CODE directive:
.CODE
.STARTUP
.
.
.
.EXIT
END

; Place executable code here

Note that .EXIT generates executable code, while END does not. The END
directive informs the assembler that it has reached the end of the module. All
modules must end with the END directive whether you use simplified or full
segments.
If you do not use .STARTUP, you must give the starting address as an
argument to the END directive. For example, the following fragment shows how
to identify a program’s starting instruction with the label start:
.CODE
start:
.
.
.
END

; Place executable code here
start

Only the END directive for the module with the starting instruction should have
an argument. When .STARTUP is present, the assembler ignores any argument
to END.
For the default NEARSTACK attribute, .STARTUP points DS to DGROUP
and sets SS:SP relative to DGROUP, generating the following code:

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 45 of 15 Printed: 10/02/00 04:23 PM

Programmer’s Guide
@Startup:
mov
mov
mov
sub
shl
shl
shl
shl
cli
mov
add
sti
.
.
.
END

dx,
ds,
bx,
bx,
bx,
bx,
bx,
bx,

DGROUP
dx
ss
dx
1
1
1
1

; If .286 or higher, this is
;
shortened to shl bx, 4

; Not necessary in .286 or higher
ss, dx
sp, bx
; Not necessary in .286 or higher

@Startup

An MS-DOS program with the FARSTACK attribute does not need to adjust
SS:SP, so .STARTUP just initializes DS, like this:
@Startup:
mov
mov
.
.
.
END

dx, DGROUP
ds, dx

@Startup

When the program terminates, you can return an exit code to the operating
system. Applications that check exit codes usually assume that an exit code of 0
means no problem occurred, and that an exit code of 1 means an error
terminated the program. The .EXIT directive accepts a 1-byte exit code as its
optional argument:
.EXIT

; Return exit code 1

.EXIT generates the following code that returns control to MS-DOS, thus
terminating the program. The return value, which can be a constant, memory
reference, or 1-byte register, goes into AL:
mov
mov
int

al, value
ah, 04Ch
21h

If your program does not specify a return value, .EXIT returns whatever value
happens to be in AL.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 46 of 16 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

Using Full Segment Definitions
If you need complete control over segments, you can fully define the segments
in your program. This section explains segment definitions, including how to
order segments and how to define the segment types.
If you write a program under MS-DOS without .MODEL and .STARTUP, you
must initialize registers yourself and use the END directive to indicate the
starting address. The Windows operating system does not require you to
initialize registers, as described in Chapter 3. For a description of typical startup
code, see “Controlling the Segment Order,” later in this chapter.

Defining Segments with the SEGMENT Directive
A defined segment begins with the SEGMENT directive and ends with the
ENDS directive:
name SEGMENT [[align]] [[READONLY]] [[combine]] [[use]] [[’class’]]
statements
name ENDS
The name defines the name of the segment. Within a module, all segment
definitions with the same name are treated as though they reference the same
segment. The linker also combines identically named segments from different
modules unless the combine type is PRIVATE. In addition, segments can be
nested.
The optional types that follow the SEGMENT directive give the linker and the
assembler instructions on how to set up and combine segments. The optional
types, which are explained in detail in the following sections, include:
Type

Description

align

Defines the memory boundary on which a new segment begins.

READONLY

Tells the assembler to report an error if it detects an instruction
modifying any item in a READONLY segment.

combine

Determines how the linker combines segments from different modules
when building executable files.

use (80386/486 only)

Determines the size of a segment. USE16 indicates that offsets in the
segment are 16 bits wide. USE32 indicates 32-bit offsets.

class

Provides a class name for the segment. The linker automatically
groups segments of the same class in memory.

Types can be specified in any order. You can specify only one attribute from
each of these fields; for example, you cannot have two different align types.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 47 of 17 Printed: 10/02/00 04:23 PM

Programmer’s Guide

You can close a segment and reopen it later with another SEGMENT directive.
When you reopen a segment, you need only give the segment name. You
cannot change the attributes of a segment once you have defined it.
Note The PAGE align type and the PUBLIC combine type are distinct from
the PAGE and PUBLIC directives. The assembler distinguishes them by means
of context.

Aligning Segments
The optional align type in the SEGMENT directive defines the range of
memory addresses from which a starting address for the segment can be
selected. The align type can be any of the following:
Align Type

Starting Address

BYTE

Next available byte address.

WORD

Next available word address.

DWORD

Next available doubleword address.

PARA

Next available paragraph address (16 bytes per paragraph). Default.

PAGE

Next available page address (256 bytes per page).

The linker uses the alignment information to determine the relative starting
address for each segment. The operating system calculates the actual starting
address when the program is loaded.

Making Segments Read-Only
The optional READONLY attribute is helpful when creating read-only code
segments for protected mode, or when writing code to be placed in read-only
memory (ROM). It protects against illegal self-modifying code.
The READONLY attribute causes the assembler to check for instructions that
modify the segment and to generate an error if it finds any. The assembler
generates an error if you attempt to write directly to a read-only segment.

Combining Segments
The optional combine type in the SEGMENT directive defines how the linker
combines segments having the same name but appearing in different modules.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 48 of 18 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

The combine type controls linker behavior, not assembler behavior. The
combine types, which are described in full detail in Help, include:
Combine Type

Linker Action

PRIVATE

Does not combine the segment with segments from other modules,
even if they have the same name. Default.

PUBLIC

Concatenates all segments having the same name to form a single,
contiguous segment.

STACK

Concatenates all segments having the same name and causes the
operating system to set SS:00 to the bottom and SS:SP to the top
of the resulting segment. Data initialization is unreliable, as
discussed following.

COMMON

Overlaps segments. The length of the resulting area is the length of
the largest of the combined segments. Data initialization is
unreliable, as discussed following.

MEMORY

Used as a synonym for the PUBLIC combine type.

AT address

Assumes address as the segment location. An AT segment cannot
contain any code or initialized data, but is useful for defining
structures or variables that correspond to specific far memory
locations, such as a screen buffer or low memory.
You cannot use the AT combine type in protected-mode
programs.

Do not place initialized data in STACK or COMMON segments. With these
combine types, the linker overlays initialized data for each module at the
beginning of the segment. The last module containing initialized data writes over
any data from other modules.
Note Normally, you should provide at least one stack segment (having STACK
combine type) in a program. If no stack segment is declared, LINK displays a
warning message. You can ignore this message if you have a specific reason for
not declaring a stack segment. For example, you would not have a separate
stack segment in a MS-DOS tiny model (.COM) program, nor would you need
a separate stack in a DLL that uses the caller’s stack.

Setting Segment Word Sizes (80386/486 Only)
The use type in the SEGMENT directive specifies the segment word size on the
80386/486 processors. Segment word size determines the default operand and
address size of all items in a segment.
The size attribute can be USE16, USE32, or FLAT. If you specify the .386 or
.486 directive before the .MODEL directive, USE32 is the default. This
attribute specifies that items in the segment are addressed with a 32-bit offset
rather than a

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 49 of 19 Printed: 10/02/00 04:23 PM

Programmer’s Guide

16-bit offset. If .MODEL precedes the .386 or .486 directive, USE16 is the
default. To make USE32 the default, put .386 or .486 before .MODEL. You
can override the USE32 default with the USE16 attribute, or vice versa.
Note Programs written for MS-DOS must not specify USE32. Mixing 16-bit
and 32-bit segments in the same program is possible but usually applies only to
systems programming.

Setting Segment Order with Class Type
The optional class type in the SEGMENT directive helps control segment
ordering. Two segments with the same name are not combined if their class is
different. The linker arranges segments so that all segments identified with a
given class type are next to each other in the executable file. However, within a
particular class, the linker arranges segments in the order encountered. The
.ALPHA, .SEQ, or .DOSSEG directive determines this order in each .OBJ file.
The most common method for specifying a class type is to place all code
segments first in the executable file.

Controlling the Segment Order
The assembler normally positions segments in the object file in the order in
which they appear in source code. The linker, in turn, processes object files in
the order in which they appear on the command line. Within each object file, the
linker outputs segments in the order they appear, subject to any group, class,
and .DOSSEG requirements.
You can usually ignore segment ordering. However, it is important whenever
you want certain segments to appear at the beginning or end of a program or
when you make assumptions about which segments are next to each other in
memory. For tiny model (.COM) programs, code segments must appear first in
the executable file, because execution must start at the address 100h.

Segment Order Directives
You can control the order in which segments appear in the executable program
with three directives. The default, .SEQ, arranges segments in the order in
which you declare them.
The .ALPHA directive specifies alphabetical segment ordering within a module.
.ALPHA is provided for compatibility with early versions of the IBM
assembler. If you have trouble running code from older books on assembly
language, try using .ALPHA.
The .DOSSEG directive specifies the MS-DOS segment-ordering convention. It
places segments in the standard order required by Microsoft languages. Do not
use .DOSSEG in a module to be called from another module.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 50 of 20 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

The .DOSSEG directive orders segments as follows:
1. Code segments
2. Data segments, in this order:
a. Segments not in class BSS or STACK
b. Class BSS segments
c. Class STACK segments
When you declare two or more segments to be in the same class, the linker
automatically makes them contiguous. This rule overrides the segment-ordering
directives. (For more about segment classes, see “Setting Segment Order with
Class Type” in the previous section.)

Linker Control
Most of the segment-ordering techniques (class names, .ALPHA, and .SEQ)
control the order in which the assembler outputs segments. Usually, you are
more interested in the order in which segments appear in the executable file. The
linker controls this order.
The linker processes object files in the order in which they appear on the
command line. Within each module, it then outputs segments in the order given
in the object file. If the first module defines segments DSEG and STACK and
the second module defines CSEG, then CSEG is output last. If you want to
place CSEG first, there are two ways to do so.
The simpler method is to use .DOSSEG. This directive is output as a special
record to the object file linker, and it tells the linker to use the Microsoft
segment-ordering convention. This convention overrides command-line order of
object files, and it places all segments of class 'CODE' first. (See “Defining
Segments with the SEGMENT Directive,” previous.)
The other method is to define all the segments as early as possible (in an include
file, for example, or in the first module). These definitions can be “dummy
segments” — that is, segments with no content. The linker observes the segment
ordering given, then later combines the empty segments with segments in other
modules that have the same name.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 51 of 21 Printed: 10/02/00 04:23 PM

Programmer’s Guide

For example, you might include the following at the start of the first module of
your program or in an include file:
_TEXT
_TEXT
_DATA
_DATA
CONST
CONST
STACK
STACK

SEGMENT
ENDS
SEGMENT
ENDS
SEGMENT
ENDS
SEGMENT
ENDS

WORD PUBLIC 'CODE'
WORD PUBLIC 'DATA'
WORD PUBLIC 'CONST'
PARA STACK 'STACK'

Later in the program, the order in which you write _TEXT, _DATA, or other
segments does not matter because the ultimate order is controlled by the
segment order defined in the include file.

Setting the ASSUME Directive for Segment Registers
Many of the assembler instructions assume a default segment. For example,
JMP assumes the segment associated with the CS register, PUSH and POP
assume the segment associated with the SS register, and MOV instructions
assume the segment associated with the DS register.
When the assembler needs to reference an address, it must know what segment
contains the address. It finds this by using the default segment or group
addresses assigned with the ASSUME directive. The syntax is:
ASSUME
ASSUME
ASSUME
ASSUME
ASSUME

segregister : seglocation [, segregister : seglocation] ]
dataregister : qualifiedtype [, dataregister : qualifiedtype]
register : ERROR [, register : ERROR]
[register :] NOTHING [, register : NOTHING]
register : FLAT [, register : FLAT]

The seglocation must be the name of the segment or group that is to be
associated with segregister. Subsequent instructions that assume a default
register for referencing labels or variables automatically assume that if the
default segment is segregister, the label or variable is in the seglocation. MASM
6.1 automatically gives CS the address of the current code segment. Therefore,
you do not need to include
ASSUME CS : MY_CODE

at the beginning of your program if you want the current segment associated
with CS.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 52 of 22 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

Note Using the ASSUME directive to tell the assembler which segment to
associate with a segment register is not the same as telling the processor. The
ASSUME directive affects only assembly-time assumptions. You may need to
use instructions to change run-time conditions. Initializing segment registers at
run time is discussed in “Informing the Assembler About Segment Values,”
Chapter 3.
The ASSUME directive can define a segment for each of the segment registers.
The segregister can be CS, DS, ES, or SS (and FS and GS on the 80386/486).
The seglocation must be one of the following:
u

u
u
u
u

The name of a segment defined in the source file with the SEGMENT
directive.
The name of a group defined in the source file with the GROUP directive.
The keyword NOTHING, ERROR, or FLAT.
A SEG expression (see “Immediate Operands” in Chapter 3).
A string equate (text macro) that evaluates to a segment or group name (but
not a string equate that evaluates to a SEG expression).

It is legal to combine assumes to FLAT with assumes to specific segments.
Combinations might be necessary in operating-system code that handles both
16- and 32-bit segments.
The keyword NOTHING cancels the current segment assumptions. For
example, the statement ASSUME NOTHING cancels all register assumptions
made by previous ASSUME statements.
Usually, a single ASSUME statement defines all four segment registers at the
start of the source file. However, you can use the ASSUME directive at any
point to change segment assumptions.
Using the ASSUME directive to change segment assumptions is often equivalent
to changing assumptions with the segment-override operator (:). See “Direct
Memory Operands” in Chapter 3. The segment-override operator is more
convenient for one-time overrides. The ASSUME directive may be more
convenient if previous assumptions must be overridden for a sequence of
instructions.
However, in either case, your program must explicitly load a segment register
with a segment address before accessing data within the segment. ASSUME
only tells the assembler to assume that the register is correctly initialized; it does
not by itself generate any code to load the register.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 53 of 23 Printed: 10/02/00 04:23 PM

Programmer’s Guide

You can also prevent the use of a register with:
ASSUME

SegRegister : ERROR

The assembler generates an ASSUME CS:ERROR when you use simplified
directives to create data segments, effectively preventing instructions or code
labels from appearing in a data segment.
For more information about ASSUME, refer to “Defining Register Types with
ASSUME” in Chapter 3.

Defining Segment Groups
A group is a collection of segments totalling not more than 64K in 16-bit mode.
A program addresses a code or data item in the group relative to the beginning
of the group.
A group lets you develop separate logical segments for different kinds of data
and then combine these into one segment (a group) for all the data. Using a
group can save you from having to continually reload segment registers to
access different segments. As a result, the program uses fewer instructions and
runs faster.
The most common example of a group is the specially named group for near
data, DGROUP. In the Microsoft segment model, several segments (_DATA,
_BSS, CONST, and STACK) are combined into a single group called
DGROUP. Microsoft high-level languages place all near data segments in this
group. (By default, the stack is placed here, too.) The .MODEL directive
automatically defines DGROUP. The DS register normally points to the
beginning of the group, giving you relatively fast access to all data in DGROUP.
The syntax of the group directive is:
name GROUP segment [[, segment]]...
The name labels the group. It can refer to a group that was previously defined.
This feature lets you add segments to a group one at a time. For example, if
MYGROUP was previously defined to include ASEG and BSEG, then the
statement
MYGROUP GROUP

CSEG

is perfectly legal. It simply adds CSEG to the group MYGROUP; ASEG and BSEG
are not removed.
Each segment can be any valid segment name (including a segment defined later
in source code), with one restriction: a segment cannot belong to more than one
group.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 54 of 24 Printed: 10/02/00 04:23 PM

Chapter 2 Organizing Segments

The GROUP directive does not affect the order in which segments of a group
are loaded. You can place any number of 16-bit segments in a group as long as
the total size does not exceed 65,536 bytes. If the processor is in 32-bit mode,
the maximum size is 4 gigabytes. You need to make sure that non-grouped
segments do not get placed between grouped segments in such a way that the
size of the group exceeds 64K or 4 gigabytes. Neither can you place a 16-bit and
a 32-bit segment in the same group.

Filename: LMAPGC02.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 38 Page: 55 of 25 Printed: 10/02/00 04:23 PM

C H A P T E R

Using Addresses and Pointers

MASM applications running in real mode require segmented addresses to access
code and data. The address of the code or data in a segment is relative to a
segment address in a segment register. You can also use pointers to access data
in assembly language programs. (A pointer is a variable that contains an address
as its value.)
The first section of this chapter describes how to initialize default segment
registers to access near and far addresses. The next section describes how to
access code and data. It also describes related operators, syntax, and
displacements. The discussion of memory operands lays the foundation for the
third section, which describes the stack.
The fourth section of this chapter explains how to use the TYPEDEF directive
to declare pointers and the ASSUME directive to give the assembler information
about registers containing pointers. This section also shows you how to do
typical pointer operations and how to write code that works for pointer variables
in any memory model.

Programming Segmented Addresses
Before you use segmented addresses in your programs, you need to initialize the
segment registers. The initialization process depends on the registers used and
on your choice of simplified segment directives or full segment definitions. The
simplified segment directives (introduced in Chapter 2) handle most of the
initialization process for you. This section explains how to inform the assembler
and the processor of segment addresses, and how to access the near and far
code and data in those segments.

Initializing Default Segment Registers
The segmented architecture of the 8086-family of processors does not require
that you specify two addresses every time you access memory. As explained in
Chapter 2, “Organizing Segments,” the 8086 family of processors uses a system

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 53 of 1 Printed: 10/02/00 04:23 PM

Programmer’s Guide

of default segment registers to simplify access to the most commonly used data
and code.
The segment registers DS, SS, and CS are normally initialized to default
segments at the beginning of a program. If you write the main module in a highlevel language, the compiler initializes the segment registers. If you write the
main module in assembly language, you must initialize the segment registers
yourself. Follow these steps to initialize segments:
1. Tell the assembler which segment is associated with a register. The
assembler must know the default segments at assembly time.
2. Tell the processor which segment is associated with a register by writing the
necessary code to load the correct segment value into the segment register on
the processor.
These steps are discussed separately in the following sections.

Informing the Assembler About Segment Values
The first step in initializing segments is to tell the assembler which segment to
associate with a register. You do this with the ASSUME directive. If you use
simplified segment directives, the assembler automatically generates the
appropriate ASSUME statements. If you use full segment definitions, you must
code the ASSUME statements for registers other than CS yourself. (ASSUME
can also be used on general-purpose registers, as explained in “Defining Register
Types with ASSUME” later in this chapter.)
The .STARTUP directive generates startup code that sets DS equal to SS
(unless you specify FARSTACK), allowing default data to be accessed through
either SS or DS. This can improve efficiency in the code generated by
compilers. The “DS equals SS” convention may not work with certain
applications, such as memory-resident programs in MS-DOS and Windows
dynamic-link libraries (see Chapter 10). The code generated for .STARTUP is
shown in “Starting and Ending Code with .STARTUP and .EXIT” in Chapter 2.
You can use similar code to set DS equal to SS in programs using full segment
definitions.
Here is an example of ASSUME using full segment definitions:
ASSUME cs:_TEXT, ds:DGROUP, ss:DGROUP

This example is equivalent to the ASSUME statement generated with simplified
segment directives in small model with NEARSTACK. Note that DS and SS
are part of the same segment group. It is also possible to have different
segments for data and code, and to use ASSUME to set ES, as shown here:

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 54 of 2 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

ASSUME cs:MYCODE, ds:MYDATA, ss:MYSTACK, es:OTHER

Correct use of the ASSUME statement can help find addressing errors. With
.CODE, the assembler assumes CS is the current segment. When you use the
simplified segment directives .DATA, .DATA?, .CONST, .FARDATA, or
.FARDATA?, the assembler automatically assumes CS is the ERROR
segment. This prevents instructions from appearing in these segments. If you
use full segment definitions, you can accomplish the same by placing ASSUME
CS:ERROR in a data segment.
With simple or full segments, you can cancel the control of an ASSUME
statement by assuming NOTHING. You can cancel the previous assumption for
ES with the following statement:
ASSUME es:NOTHING

Prior to the .MODEL statement (or in its absence), the assembler sets the
ASSUME statement for DS, ES, and SS to the current segment.

Informing the Processor About Segment Values
The second and final step in initializing segments is to inform the processor of
segment values at run time. How segment values are initialized at run time
differs for each segment register and depends on the operating system and on
your use of simplified segment directives or full segment definitions.

Specifying a Starting Address
A program’s starting address determines where execution begins. After the
operating system loads a program, it simply jumps to the starting address, giving
processor control to the program. The true starting address is known only to the
loader; the linker determines only the offset of the address within an
undetermined code segment. That’s why a normal application is often referred
to as “relocatable code,” because it runs regardless of where the loader places it
in memory.
The offset of the starting address depends on the program type. Programs with
an .EXE extension contain a header from which the loader reads the offset and
combines it with a segment to form the starting address. Programs with a .COM
extension (tiny model) have no such header, so by convention the loader jumps
to the first byte of the program.
In either case, the .STARTUP directive identifies where execution begins,
provided you use simplified segment directives. For an .EXE program, place
.STARTUP immediately before the instruction where you want execution to
start. In a .COM program, place .STARTUP before the first assembly
instruction in your source code.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 55 of 3 Printed: 10/02/00 04:23 PM

Programmer’s Guide

If you use full segment directives or prefer not to use .STARTUP, you must
identify the starting instruction in two steps:
1. Label the starting instruction.
2. Provide the same label in the END directive.
These steps tell the linker where execution begins in the program. The following
example illustrates the two steps for a tiny model program:
_TEXT
start:

_TEXT

SEGMENT WORD PUBLIC 'CODE'
ORG
100h
; Use this declaration for .COM files only
.
; First instruction here
.
.
ENDS
END
start
; Name of starting label

Notice the ORG statement in this example. This statement is mandatory in a
tiny model program without the .STARTUP directive. It places the first
instruction at offset 100h in the code segment to create space for a 256-byte
(100h) data area called the Program Segment Prefix (PSP). The operating
system takes care of initializing the PSP, so you need only make sure the area
exists. (For a description of what data resides in the PSP, refer to the “Tables”
chapter in the Reference.)

Initializing DS
The DS register is automatically initialized to the correct value (DGROUP) if
you use .STARTUP or if you are writing a program for Windows. If you do not
use .STARTUP with MS-DOS, you must initialize DS using the following
instructions:
mov
mov

ax, DGROUP
ds, ax

The initialization requires two instructions because the segment name is a
constant and the assembler does not allow a constant to be loaded directly to a
segment register. The previous example loads DGROUP, but you can load any
valid segment or group.

Initializing SS and SP
The SS and SP registers are initialized automatically if you use the .STACK
directive with simplified segments or if you define a segment that has the
STACK combine type with full segment definitions. Using the STACK directive
initializes SS to the stack segment. If you want SS to be equal to DS, use
.STARTUP or its equivalent. (See “Combining Segments,” page 45.) For an
.EXE file, the stack address is encoded into the executable header and resolved

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 56 of 4 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

at load time. For a .COM file, the loader sets SS equal to CS and initializes SP
to 0FFFEh.
If your program does not access far data, you do not need to initialize the ES
register. If you choose to initialize, use the same technique as for the DS
register. You can initialize SS to a far stack in the same way.

Near and Far Addresses
Addresses that have an implied segment name or segment registers associated
with them are called “near addresses.” Addresses that have an explicit segment
associated with them are called “far addresses.” The assembler handles near and
far code automatically, as described in the following sections. You must specify
how to handle far data.
The Microsoft segment model puts all near data and the stack in a group called
DGROUP. Near code is put in a segment called _TEXT. Each module’s far
code or far data is placed in a separate segment. This convention is described in
“Controlling the Segment Order” in Chapter 2.
The assembler cannot determine the address for some program components;
these are said to be relocatable. The assembler generates a fixup record and the
linker provides the address once it has determined the location of all segments.
Usually a relocatable operand references a label, but there are exceptions.
Examples in the next two sections include information about relocating near and
far data.

Near Code
Control transfers within near code do not require changes to segment registers.
The processor automatically handles changes to the offset in the IP register
when control-flow instructions such as JMP, CALL, and RET are used. The
statement
call

nearproc

; Change code offset

changes the IP register to the new address but leaves the segment unchanged.
When the procedure returns, the processor resets IP to the offset of the next
instruction after the CALL instruction.

Far Code
The processor automatically handles segment register changes when dealing with
far code. The statement
call

farproc

; Change code segment and offset

automatically moves the segment and offset of the farproc procedure to the
CS and IP registers. When the procedure returns, the processor sets CS to the

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 57 of 5 Printed: 10/02/00 04:23 PM

Programmer’s Guide

original code segment and sets IP to the offset of the next instruction after the
call.

Near Data
A program can access near data directly, because a segment register already
holds the correct segment for the data item. The term “near data” is often used
to refer to the data in the DGROUP group.
After the first initialization of the DS and SS registers, these registers normally
point into DGROUP. If you modify the contents of either of these registers
during the execution of the program, you must reload the register with
DGROUP’s address before referencing any DGROUP data.
The processor assumes all memory references are relative to the segment in the
DS register, with the exception of references using BP or SP. The processor
associates these registers with the SS register. (You can override these
assumptions with the segment override operator, described in “Direct Memory
Operands,” on page 62.)
The following lines illustrate how the processor accesses either the DS or SS
segments, depending on whether the pointer operand contains BP or SP. Note
the distinction loses significance when DS and SS are equal.
nearvar WORD
.
.
.
mov
mov
mov
mov
mov

ax, nearvar
di, [bx]
[di], cx
[bp+6], ax
bx, [bp]

;
;
;
;
;

Reads from
Reads from
Writes to
Writes to
Reads from

DS:[nearvar]
DS:[bx]
DS:[di]
SS:[bp+6]
SS:[bp]

Far Data
To read or modify a far address, a segment register must point to the segment of
the data. This requires two steps. First load the segment (normally either ES or
DS) with the correct value, and then (optionally) set an assume of the segment
register to the segment of the address.
Note Flat model does not require far addresses. By default, all addressing is
relative to the initial values of the segment registers. Therefore, this section on
far addressing does not apply to flat model programs.
One method commonly used to access far data is to initialize the ES segment
register. This example shows two ways to do this:

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 58 of 6 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers
; First method
mov
mov
mov

ax, SEG farvar
es, ax
ax, es:farvar

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 59 of 7 Printed: 10/02/00 04:23 PM

; Load segment of the
,
far address into ES
; Provide an explicit segment
;
override on the addressing

Programmer’s Guide
; Second method
mov
ax, SEG farvar2 ; Load the segment of the
mov
es, ax
;
far address into ES
ASSUME ES:SEG farvar2 ; Tell the assembler that ES points
;
to the segment containing farvar2
mov
ax, farvar2
; The assembler provides the ES
;
override since it knows that
;
the label is addressable

After loading the segment of the address into the ES segment register, you can
explicitly override the segment register so that the addressing is correct (method
1) or allow the assembler to insert the override for you (method 2). The
assembler uses ASSUME statements to determine which segment register can
be used to address a segment of memory. To use the segment override operator,
the left operand must be a segment register, not a segment name. (For more
information on segment overrides, see “Direct Memory Operands” on page 62.)
If an instruction needs a segment override, the resulting code is slightly larger
and slower, since the override must be encoded into the instruction. However,
the resulting code may still be smaller than the code for multiple loads of the
default segment register for the instruction.
The DS, SS, FS, and GS segment registers (FS and GS are available only on the
80386/486 processors) may also be used for addressing through other segments.
If a program uses ES to access far data, it need not restore ES when finished
(unless the program uses flat model). However, some compilers require that you
restore ES before returning to a module written in a high-level language.
To access far data, first set DS to the far segment and then restore the original
DS when finished. Use the ASSUME directive to let the assembler know that
DS no longer points to the default data segment, as shown here:
push
mov
mov
ASSUME
mov
mov
.
.
.
pop
ASSUME

ds
ax, SEG fararray
ds, ax
ds:SEG fararray
ax, fararray[0]
dx, fararray[2]

;
;
;
;
;
;

ds
ds:@DATA

; Restore segment
;
and default assumption

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 60 of 8 Printed: 10/02/00 04:23 PM

Save original segment
Move segment into data register
Initialize segment register
Tell assembler where data is
Set DX:AX = dword variable
fararray

Chapter 3 Using Addresses and Pointers

“Direct Memory Operands,”on page 62, describes an alternative method for
accessing far data. The technique of resetting DS as shown in the previous
example is best for a lengthy series of far data references. The segment override
method described in “Direct Memory Operands” serves best when accessing
only one or two far variables.
If your program changes DS to access far data, it should restore DS when
finished. This allows procedures to assume that DS is the segment for near data.
Many compilers, including Microsoft compilers, use this convention.

Operands
With few exceptions, assembly language instructions work on sources of data
called operands. In a listing of assembly code (such as the examples in this
book), operands appear in the operand field immediately to the right of the
instructions.
This section describes the four kinds of instruction operands: register,
immediate, direct memory, and indirect memory. Some instructions, such as
POPF and STI, have implied operands which do not appear in the operand
field. Otherwise, an implied operand is just as real as one stated explicitly.
Certain other instructions such as NOP and WAIT deserve special mention.
These instructions affect only processor control and do not require an operand.
The following four types of operands are described in the rest of this section:
Operand Type

Addressing Mode

An 8-bit or 16-bit register on the 8086–80486; can also be 32-bit on the
80386/486.

Immediate

A constant value contained in the instruction itself.

Direct memory

A fixed location in memory.

Indirect memory

A memory location determined at run time by using the address stored in
one or two registers.

Instructions that take two or more operands always work right to left. The right
operand is the source operand. It specifies data that will be read, but not
changed, in the operation. The left operand is the destination operand. It
specifies the data that will be acted on and possibly changed by the instruction.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 61 of 9 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Register Operands
Register operands refer to data stored in registers. The following examples show
typical register operands:
mov
add
jmp

bx, 10
ax, bx
di

; Load constant to BX
; Add BX to AX
; Jump to the address in DI

An offset stored in a base or index register often serves as a pointer into
memory. You can store an offset in one of the base or index registers, then use
the register as an indirect memory operand. (See “Indirect Memory Operands,”
following.) For example:
mov
inc
mov

[bx], dl ; Store DL in indirect memory operand
bx
; Increment register operand
[bx], dl ; Store DL in new indirect memory operand

This example moves the value in DL to 2 consecutive bytes of a memory
location pointed to by BX. Any instruction that changes the register value also
changes the data item pointed to by the register.

Immediate Operands
An immediate operand is a constant or the result of a constant expression. The
assembler encodes immediate values into the instruction at assembly time. Here
are some typical examples showing immediate operands:
mov
add
sub

cx, 20
var, 1Fh
bx, 25 * 80

; Load constant to register
; Add hex constant to variable
; Subtract constant expression

Immediate data is never permitted in the destination operand. If the source
operand is immediate, the destination operand must be either a register or direct
memory to provide a place to store the result of the operation.
Immediate expressions often involve the useful OFFSET and SEG operators,
described in the following paragraphs.

The OFFSET Operator
An address constant is a special type of immediate operand that consists of an
offset or segment value. The OFFSET operator returns the offset of a memory
location, as shown here:
mov

bx, OFFSET var

; Load offset address

For information on differences between MASM 5.1 behavior and MASM 6.1
behavior related to OFFSET, see Appendix A.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 62 of 10 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

Since data in different modules may belong to a single segment, the assembler
cannot know for each module the true offsets within a segment. Thus, the offset
for var, although an immediate value, is not determined until link time.

The SEG Operator
The SEG operator returns the segment of a memory location:
mov
mov

ax, SEG farvar
es, ax

; Load segment address

The actual value of a particular segment is not known until the program is
loaded into memory. For .EXE programs, the linker makes a list in the
program’s header of all locations in which the SEG operator appears. The
loader reads this list and fills in the required segment address at each location.
Since .COM programs have no header, the assembler does not allow relocatable
segment expressions in tiny model programs.
The SEG operator returns a variable’s “frame” if it appears in the instruction.
The frame is the value of the segment, group, or segment override of a
nonexternal variable. For example, the instruction
mov

ax, SEG DGROUP:var

places in AX the value of DGROUP, where var is located. If you do not
include a frame, SEG returns the value of the variable’s group if one exists. If
the variable is not defined in a group, SEG returns the variable’s segment
address.
This behavior can be changed with the /Zm command-line option or with the
OPTION OFFSET:SEGMENT statement. (See Appendix A, “Differences
between MASM 6.1 and 5.1.”) “Using the OPTION Directive” in Chapter 1
introduces the OPTION directive.

Direct Memory Operands
A direct memory operand specifies the data at a given address. The instruction
acts on the contents of the address, not the address itself. Except when size is
implied by another operand, you must specify the size of a direct memory
operand so the instruction accesses the correct amount of memory. The
following example shows how to explicitly specify data size with the BYTE
directive:

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 63 of 11 Printed: 10/02/00 04:23 PM

Programmer’s Guide

var

.DATA?
BYTE
?
.CODE
.
.
.
mov
var, al

; Segment for uninitialized data
; Reserve one byte, labeled "var"

; Copy AL to byte at var

Any location in memory can be a direct memory operand as long as a size is
specified (or implied) and the location is fixed. The data at the address can
change, but the address cannot. By default, instructions that use direct memory
addressing use the DS register. You can create an expression that points to a
memory location using any of the following operators:
Operator Name

Symbol

Plus

Minus

–

Index

[]

Structure member

Segment override

These operators are discussed in more detail in the following section.

Plus, Minus, and Index
The plus and index operators perform in exactly the same way when applied to
direct memory operands. For example, both the following statements move the
second word value from an array into the AX register:
mov
mov

ax, array[2]
ax, array+2

The index operator can contain any direct memory operand. The following
statements are equivalent:
mov
mov

ax, var
ax, [var]

Some programmers prefer to enclose the operand in brackets to show that the
contents, not the address, are used.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 64 of 12 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

The minus operator behaves as you would expect. Both the following
instructions retrieve the value located at the word preceding array:
mov
mov

ax, array[-2]
ax, array-2

Structure Field
The structure operator (.) references a particular element of a structure or
“field,” to use C terminology:
mov

bx, structvar.field1

The address of the structure operand is the sum of the offsets of structvar
and field1. For more information about structures, see “Structures and
Unions” in Chapter 5.

Segment Override
The segment override operator (:) specifies a segment portion of the address
that is different from the default segment. When used with instructions, this
operator can apply to segment registers or segment names:
mov

ax, es:farvar

; Use segment override

The assembler will not generate a segment override if the default segment is
explicitly provided. Thus, the following two statements assemble in exactly the
same way:
mov
mov

[bx], ax
ds:[bx], ax

A segment name override or the segment override operator identifies the
operand as an address expression.

;

mov
mov
mov
mov

WORD
WORD
WORD
WORD

PTR
PTR
PTR
PTR

FARSEG:0, ax
es:100h, ax
es:[100h], ax
[100h], ax

; Segment name override
; Legal and equivalent
;
expressions
; Illegal, not an address

As the example shows, a constant expression cannot be an address expression
unless it has a segment override.

Indirect Memory Operands
Like direct memory operands, indirect memory operands specify the contents of
a given address. However, the processor calculates the address at run time by
referring to the contents of registers. Since values in the registers can change at
run time, indirect memory operands provide dynamic access to memory.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 65 of 13 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Indirect memory operands make possible run-time operations such as pointer
indirection and dynamic indexing of array elements, including indexing of
multidimensional arrays.
Strict rules govern which registers you can use for indirect memory operands
under 16-bit versions of the 8086-based processors. The rules change
significantly for 32-bit processors starting with the 80386. However, the new
rules apply only to code that does not need to be compatible with earlier
processors.
This section covers features of indirect operands in either mode. The specific
16-bit rules and 32-bit rules are then explained separately.

Indirect Operands with 16- and 32-Bit Registers
Some rules and options for indirect memory operands always apply, regardless
of the size of the register. For example, you must always specify the register and
operand size for indirect memory operands. But you can use various syntaxes to
indicate an indirect memory operand. This section describes the rules that apply
to both 16-bit and 32-bit register modes.

Specifying Indirect Memory Operands
The index operator specifies the register or registers for indirect operands. The
processor uses the data pointed to by the register. For example, the following
instruction moves into AX the word value at the address in DS:BX.
mov

ax, WORD PTR [bx]

When you specify more than one register, the processor adds the contents of the
two addresses together to determine the effective address (the address of the
data to operate on):
mov

ax, [bx+si]

Specifying Displacements
You can specify an address displacement, which is a constant value added to the
effective address. A direct memory specifier is the most common displacement:
mov

ax, table[si]

In this relocatable expression, the displacement table is the base address of an
array; SI holds an index to an array element. The SI value is calculated at run
time, often in a loop. The element loaded into AX depends on the value of SI at
the time the instruction executes.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 66 of 14 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

Each displacement can be an address or numeric constant. If there is more than
one displacement, the assembler totals them at assembly time and encodes the
total displacement. For example, in the statement
table

WORD
.
.
.
mov

100 DUP (0)

ax, table[bx][di]+6

both table and 6 are displacements. The assembler adds the value of 6 to
table to get the total displacement. However, the statement
mov ax, mem1[si] + mem2

is not legal, because it attempts to use a single command to join the contents of
two different addresses.

Specifying Operand Size
You must give the size of an indirect memory operand in one of three ways:
u
u
u

By the variable’s declared size
With the PTR operator
Implied by the size of the other operand

The following lines illustrate all three methods. Assume the size of the table
array is WORD, as declared earlier.
mov
mov
mov

table[bx], 0
; 2 bytes - from size of table
BYTE PTR table, 0 ; 1 byte - specified by BYTE
ax, [bx]
; 2 bytes - implied by AX

Syntax Options
The assembler allows a variety of syntaxes for indirect memory operands.
However, all registers must be inside brackets. You can enclose each register in
its own pair of brackets, or you can place the registers in the same pair of
brackets separated by a plus operator (+). All the following variations are legal
and assemble the same way:
mov
mov
mov
mov
mov

ax,
ax,
ax,
ax,
ax,

table[bx][di]
table[di][bx]
table[bx+di]
[table+bx+di]
[bx][di]+table

All of these statements move the value in table indexed by BX+DI into AX.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 67 of 15 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Scaling Indexes
The value of index registers pointing into arrays must often be adjusted for zerobased arrays and scaled according to the size of the array items. For a word
array, the item number must be multiplied by two (shifted left by one place).
When using 16-bit registers, you must scale with separate instructions, as shown
here:
mov
shl
inc

bx, 5
bx, 1
wtable[bx]

; Get sixth element (adjust for 0)
; Scale by two (word size)
; Increment sixth element in table

When using 32-bit registers on the 80386/486 processor, you can include scaling
in the operand, as described in “Indirect Memory Operands with 32-Bit
Registers,” following.

Accessing Structure Elements
The structure member operator can be used in indirect memory operands to
access structure elements. In this example, the structure member operator loads
the year field of the fourth element of the students array into AL:
STUDENT
grade
name
year
STUDENT

STRUCT
WORD
BYTE
BYTE
ENDS

students

?
20 DUP (?)
?

STUDENT
.
.
mov
mov
mov
mul
mov
mov

< >

; Assume array is initialized
; Point to array of students
; Get fourth element
; Get size of STUDENT
; Multiply size times
ax
;
elements to point DI
;
to current element
al, (STUDENT PTR[bx+di]).year
bx,
ax,
di,
di
di,

OFFSET students
4
SIZE STUDENT

For more information on MASM structures, see “Structures and Unions” in
Chapter 5.

Indirect Memory Operands with 16-Bit Registers
For 8086-based computers and MS-DOS, you must follow the strict indexing
rules established for the 8086 processor. Only four registers are allowed — BP,
BX, SI, and DI — those only in certain combinations.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 68 of 16 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

BP and BX are base registers. SI and DI are index registers. You can use either
a base or an index register by itself. But if you combine two registers, one must
be a base and one an index. Here are legal and illegal forms:

;
;

mov
mov
mov
mov
mov
mov

ax,
ax,
ax,
ax,
ax,
ax,

[bx+di]
[bx+si]
[bp+di]
[bp+si]
[bx+bp]
[di+si]

;
;
;
;
;
;

Legal
Legal
Legal
Legal
Illegal - two base registers
Illegal - two index registers

Table 3.1 shows the register modes in which you can specify indirect memory
operands.
Table 3.1

Indirect Addressing with 16-Bit Registers

Mode

Syntax

Effective Address

[BX]
[BP]
[DI]
[SI]

Contents of register

Base or index

displacement[BX]
displacement[BP]
displacement[DI]
displacement[SI]

Contents of register plus
displacement

Base plus index

[BX][DI]
[BP][DI]
[BX][SI]
[BP][SI]

Contents of base register plus
contents of index register

Base plus index with
displacement

displacement[BX][DI]
displacement[BP][DI]
displacement[BX][SI]
displacement[BP][SI]

Sum of base register, index
register, and displacement

Different combinations of registers and displacements have different timings, as
shown in Reference.

Indirect Memory Operands with 32-Bit Registers
You can write instructions for the 80386/486 processor using either 16-bit or
32-bit segments. Indirect memory operands are different in each case.
In 16-bit real mode, the 80386/486 operates the same way as earlier 8086-based
processors, with one difference: you can use 32-bit registers. If the 80386/486
processor is enabled (with the .386 or .486 directive), 32-bit general-purpose
registers are available with either 16-bit or 32-bit segments. Thirty-two–bit

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 69 of 17 Printed: 10/02/00 04:23 PM

Programmer’s Guide

registers eliminate many of the limitations of 16-bit indirect memory operands.
You can use 80386/486 features to make your MS-DOS programs run faster
and more efficiently if you are willing to sacrifice compatibility with earlier
processors.
In 32-bit mode, an offset address can be up to 4 gigabytes. (Segments are still
represented in 16 bits.) This effectively eliminates size restrictions on each
segment, since few programs need 4 gigabytes of memory. Windows NT uses
32-bit mode and flat model, which spans all segments. XENIX 386 uses 32-bit
mode with multiple segments.

80386/486 Enhancements
On the 80386/486, the processor allows you to use any general-purpose 32-bit
register as a base or index register, except ESP, which can be a base but not an
index. However, you cannot combine 16-bit and 32-bit registers. Several
examples are shown here:
add
mov
dec
cmp
jmp

edx, [eax]
dl, [esp+10]
WORD PTR [edx][eax]
ax, array[ebx][ecx]
FWORD PTR table[ecx]

;
;
;
;
;

Add double
Copy byte from stack
Decrement word
Compare word from array
Jump into pointer table

Scaling Factors
With 80386/486 registers, the index register can have a scaling factor of 1, 2, 4,
or 8. Any register except ESP can be the index register and can have a scaling
factor. To specify the scaling factor, use the multiplication operator (*) adjacent
to the register.
You can use scaling to index into arrays with different sizes of elements. For
example, the scaling factor is 1 for byte arrays (no scaling needed), 2 for word
arrays, 4 for doubleword arrays, and 8 for quadword arrays. There is no
performance penalty for using a scaling factor. Scaling is illustrated in the
following examples:
mov
mov
mov

eax, darray[edx*4]
; Load double of double array
eax, [esi*8][edi]
; Load double of quad array
ax, wtbl[ecx+2][edx*2] ; Load word of word array

Scaling is also necessary on earlier processors, but it must be done with separate
instructions before the indirect memory operand is used, as described in
“Indirect Memory Operands with 16-Bit Registers,” previous.
The default segment register is SS if the base register is EBP or ESP. However,
if EBP is scaled, the processor treats it as an index register with a value relative
to DS, not SS.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 70 of 18 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

All other base registers are relative to DS. If two registers are used, only one can
have a scaling factor. The register with the scaling factor is defined as the index
register. The other register is defined as the base. If scaling is not used, the first
register is the base. If only one register is used, it is considered the base for
deciding the default segment unless it is scaled. The following examples illustrate
how to determine the base register:
mov
mov
mov
mov
mov
mov

eax,
eax,
eax,
eax,
eax,
eax,

[edx][ebp*4]
[edx*1][ebp]
[edx][ebp]
[ebp][edx]
[ebp]
[ebp*2]

;
;
;
;
;
;

EDX base (not scaled - seg DS)
EBP base (not scaled - seg SS)
EDX base (first - seg DS)
EBP base (first - seg SS)
EBP base (only - seg SS)
EBP*2 index (seg DS)

Mixing 16-Bit and 32-Bit Registers
Assembly statements can mix 16-bit and 32-bit registers. For example, the
following statement is legal for 16-bit and 32-bit segments:
mov

eax, [bx]

This statement moves the 32-bit value pointed to by BX into the EAX register.
Although BX is a 16-bit pointer, it can still point into a 32-bit segment.
However, the following statement is never legal, since you cannot use the CX
register as a 16-bit pointer:
;

mov

eax, [cx]

; illegal

Operands that mix 16-bit and 32-bit registers are also illegal:
;

mov

eax, [ebx+si]

; illegal

The following statement is legal in either 16-bit or 32-bit mode:
mov

bx, [eax]

This statement moves the 16-bit value pointed to by EAX into the BX register.
This works in 32-bit mode. However, in 16-bit mode, moving a 32-bit pointer
into a 16-bit segment is illegal. If EAX contains a 16-bit value (the top half of
the 32-bit register is 0), the statement works. However, if the top half of the
EAX register is not 0, the operand points into a part of the segment that doesn’t
exist, generating an error. If you use 32-bit registers as indexes in 16-bit mode,
you must make sure that the index registers contain valid 16-bit addresses.

The Program Stack
The preceding discussion on memory operands lays the groundwork for
understanding the important data area known as the “stack.”

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 71 of 19 Printed: 10/02/00 04:23 PM

Programmer’s Guide

A stack is an area of memory for storing data temporarily. Unlike other
segments that store data starting from low memory, the stack stores data starting
from high memory. Data is always pushed onto, or “popped” from the top of
the stack.
The stack gets its name from its similarity to the spring-loaded plate holders in
cafeterias. You add and remove plates from only the top of the stack. To
retrieve the third plate, you must remove — that is, “pop” — the first two plates.
Stacks are often referred to as LIFO buffers, from their last-in-first-out
operation.
A stack is an essential part of any nontrivial program. A program continually
uses its stack to temporarily store return addresses, procedure arguments,
memory data, flags, or registers.
The SP register serves as an indirect memory operand to the top of the stack. At
first, the stack is an uninitialized segment of a finite size. As your program adds
data to the stack, the stack grows downward from high memory to low
memory. When you remove items from the stack, it shrinks upward from low to
high memory.

Saving Operands on the Stack
The PUSH instruction stores a 2-byte operand on the stack. The POP
instruction retrieves the most recent pushed value. When a value is pushed onto
the stack, the assembler decreases the SP (Stack Pointer) register by 2. On
8086-based processors, the SP register always points to the top of the stack.
The PUSH and POP instructions use the SP register to keep track of the
current position.
When a value is popped off the stack, the assembler increases the SP register by
2. Since the stack always contains word values, the SP register changes in
multiples of two. When a PUSH or POP instruction executes in a 32-bit code
segment (one with USE32 use type), the assembler transfers a 4-byte value, and
ESP changes in multiples of four.
Note The 8086 and 8088 processors differ from later Intel processors in how
they push and pop the SP register. If you give the statement push sp with the
8086 or 8088, the word pushed is the word in SP after the push operation.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 72 of 20 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

Figure 3.1 illustrates how pushes and pops change the SP register.

Figure 3.1

Stack Status Before and After Pushes and Pops

On the 8086, PUSH and POP take only registers or memory expressions as
their operands. The other processors allow an immediate value to be an operand
for PUSH. For example, the following statement is legal on the 80186–80486
processors:
push

; 3 clocks on 80286

That statement is faster than these equivalent statements, which are required on
the 8088 or 8086:
mov
push

ax, 7
ax

; 2 clocks plus
; 3 clocks on 80286

Words are popped off the stack in reverse order: the last item pushed is the first
popped. To return the stack to its original status, you do the same number of

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 73 of 21 Printed: 10/02/00 04:23 PM

Programmer’s Guide

pops as pushes. You can subtract the correct number of words from the SP
register if you want to restore the stack without using the values on it.
To reference operands on the stack, remember that the values pointed to by the
BP (Base Pointer) and SP registers are relative to the SS (Stack Segment)
register. The BP register is often used to point to the base of a frame of
reference (a stack frame) within the stack. This example shows how you can
access values on the stack using indirect memory operands with BP as the base
register.
push
mov
push
push
push
.
.
.
mov
mov
mov
.
.
.
add

bp
bp, sp
ax
bx
cx

;
;
;
;
;

Save current value of BP
Set stack frame
Push first; SP = BP - 2
Push second; SP = BP - 4
Push third; SP = BP - 6

ax, [bp-6]
bx, [bp-4]
cx, [bp-2]

; Put third word in AX
; Put second word in BX
; Put first word in CX

sp, 6

pop

; Restore stack pointer
;
(two bytes per push)
; Restore BP

If you often use these stack values in your program, you may want to give them
labels. For example, you can use TEXTEQU to create a label such as count
TEXTEQU <[bp-6]>. Now you can replace the mov ax, [bp - 6]
statement in the previous example with mov ax, count. For more information
about the TEXTEQU directive, see “Text Macros” in Chapter 9.

Saving Flags on the Stack
Your program can push and pop flags onto the stack with the PUSHF and
POPF instructions. These instructions save and then restore the status of the
flags. You can also use them within a procedure to save and restore the flag
status of the caller. The 32-bit versions of these instructions are PUSHFD and
POPFD.
This example saves the flags register before calling the systask procedure:
pushf
call
popf

systask

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 74 of 22 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

If you do not need to store the entire flags register, you can use the LAHF
instruction to manually load and store the status of the lower byte of the flag
register in the AH register. SAHF restores the value.

Saving Registers on the Stack (80186–80486 Only)
Starting with the 80186 processor, the PUSHA and POPA instructions push or
pop all the general-purpose registers with only one instruction. These
instructions save the status of all registers before a procedure call and restore
them after the return. Using PUSHA and POPA is significantly faster and takes
fewer bytes of code than pushing and popping each register individually.
The processor pushes the registers in the following order: AX, CX, DX, BX,
SP, BP, SI, and DI. The SP word pushed is the value before the first register is
pushed.
The processor pops the registers in the opposite order. The 32-bit versions of
these instructions are PUSHAD and POPAD.

Accessing Data with Pointers and Addresses
A pointer is simply a variable that contains an address of some other variable.
The address in the pointer “points” to the other object. Pointers are useful when
transferring a large data object (such as an array) to a procedure. The caller
places only the pointer on the stack, which the called procedure uses to locate
the array. This eliminates the impractical step of having to pass the entire array
back and forth through the stack.
There is a difference between a far address and a far pointer. A “far address” is
the address of a variable located in a far data segment. A “far pointer” is a
variable that contains the segment address and offset of some other data. Like
any other variable, a pointer can be located in either the default (near) data
segment or in a far segment.
Previous versions of MASM allow pointer variables but provide little support for
them. In previous versions, any address loaded into a variable can be considered
a pointer, as in the following statements:
Var
npVar
fpVar

BYTE
WORD
DWORD

0
Var
Var

; Variable
; Near pointer to variable
; Far pointer to variable

If a variable is initialized with the name of another variable, the initialized
variable is a pointer, as shown in this example. However, in previous versions of
MASM, the CodeView debugger recognizes npVar and fpVar as word and
doubleword variables. CodeView does not treat them as pointers, nor does it
recognize the type of data they point to (bytes, in the example).

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 75 of 23 Printed: 10/02/00 04:23 PM

Programmer’s Guide

The TYPEDEF directive and enhanced capabilities of ASSUME (introduced in
MASM 6.0) make it easier to manage pointers in registers and variables. The
rest of this chapter describes these directives and how they apply to basic
pointer operations.

Defining Pointer Types with TYPEDEF
The TYPEDEF directive can define types for pointer variables. A type so
defined is considered the same as the intrinsic types provided by the assembler
and can be used in the same contexts. When used to define pointers, the syntax
for TYPEDEF is:
typename TYPEDEF [[distance]] PTR qualifiedtype
The typename is the name assigned to the new type. The distance can be
NEAR, FAR, or any distance modifier. The qualifiedtype can be any
previously intrinsic or defined MASM type, or a type previously defined with
TYPEDEF. (For a full definition of qualifiedtype, see “Data Types” in Chapter
1.)
Here are some examples of user-defined types:
PBYTE
NPBYTE
FPBYTE
PWORD
NPWORD
FPWORD

TYPEDEF
TYPEDEF
TYPEDEF
TYPEDEF
TYPEDEF
TYPEDEF

PTR BYTE
NEAR PTR BYTE
FAR PTR BYTE
PTR WORD
NEAR PTR WORD
FAR PTR WORD

;
;
;
;
;
;

Pointer to bytes
Near pointer to bytes
Far pointer to bytes
Pointer to words
Near pointer to words
Far pointer to words

PPBYTE

TYPEDEF

PTR PBYTE

PVOID

TYPEDEF

PTR

PERSON
name
num
PERSON
PPERSON

STRUCT
; Structure type
BYTE
20 DUP (?)
WORD
?
ENDS
TYPEDEF
PTR PERSON ; Pointer to structure type

; Pointer to pointer to bytes
;
(in C, an array of strings)
; Pointer to any type of data

The distance of a pointer can be set specifically or determined automatically by
the memory model (set by .MODEL) and the segment size (16 or 32 bits). If
you don’t use .MODEL, near pointers are the default.
In 16-bit mode, a near pointer is 2 bytes that contain the offset of the object
pointed to. A far pointer requires 4 bytes, and contains both the segment and
offset. In 32-bit mode, a near pointer is 4 bytes and a far pointer is 6 bytes,
since segments are

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 76 of 24 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

still word values in 32-bit mode. If you specify the distance with NEAR or
FAR, the processor uses the default distance of the current segment size. You
can use NEAR16, NEAR32, FAR16, and FAR32 to override the defaults set
by the current segment size. In flat model, NEAR is the default.
You can declare pointer variables with a pointer type created with TYPEDEF.
Here are some examples using these pointer types.
; Type declarations
Array
WORD
25 DUP (0)
Msg
BYTE
"This is a string", 0
pMsg
PBYTE
Msg
; Pointer to string
pArray PWORD
Array
; Pointer to word array
npMsg
NPBYTE Msg
; Near pointer to string
npArray NPWORD Array
; Near pointer to word array
fpArray FPWORD Array
; Far pointer to word array
fpMsg
FPBYTE Msg
; Far pointer to string
S1
S2
S3
pS123
ppS123

BYTE
BYTE
BYTE
PBYTE
PPBYTE

"first", 0
"second", 0
"third", 0
S1, S2, S3, 0
pS123

Andy
pAndy

PERSON <>
PPERSON Andy

; Some strings

; Array of pointers to strings
; A pointer to pointers to strings
; Structure variable
; Pointer to structure variable
; Procedure prototype

EXTERN
Sort

ptrArray:PBYTE
PROTO
pArray:PBYTE

; Parameter for procedure
Sort
PROC
pArray:PBYTE
LOCAL
pTmp:PBYTE
.
.
.
ret
Sort
ENDP

; External variable
; Parameter for prototype

; Local variable

Once defined, pointer types can be used in any context where intrinsic types are
allowed.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 77 of 25 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Defining Register Types with ASSUME
You can use the ASSUME directive with general-purpose registers to specify
that a register is a pointer to a certain size of object. For example:
ASSUME
inc
add
mov
.
.
.
ASSUME

bx:PTR WORD
[bx]
bx, 2
[bx], 0

;
;
;
;

Assume BX is now a word pointer
Increment word pointed to by BX
Point to next word
Word pointed to by BX = 0

; Other pointer operations with BX
bx:NOTHING

; Cancel assumption

In this example, BX is specified as a pointer to a word. After a sequence of
using BX as a pointer, the assumption is canceled by assuming NOTHING.
Without the assumption to PTR WORD, many instructions need a size
specifier. The INC and MOV statements from the previous examples would
have to be written like this to specify the sizes of the memory operands:
inc
mov

WORD PTR [bx]
WORD PTR [bx], 0

When you have used ASSUME, attempts to use the register for other purposes
generate assembly errors. In this example, while the PTR WORD assumption is
in effect, any use of BX inconsistent with its ASSUME declaration generates an
error. For example,
;

mov

al, [bx]

; Can't move word to byte register

You can also use the PTR operator to override defaults:
mov

al, BYTE PTR [bx]

; Legal

Similarly, you can use ASSUME to prevent the use of a register as a pointer, or
even to disable a register:
;
;

ASSUME
mov
mov

bx:WORD, dx:ERROR
al, [bx] ; Error - BX is an integer, not a pointer
ax, dx
; Error - DX disabled

For information on using ASSUME with segment registers, refer to “Setting the
ASSUME Directive for Segment Registers” in Chapter 2.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 78 of 26 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

Basic Pointer and Address Operations
A program can perform the following basic operations with pointers and
addresses:
u

Initialize a pointer variable by storing an address in it.

Load an address into registers, directly or from a pointer.

The sections in the rest of this chapter describe variations of these tasks with
pointers and addresses. The examples are used with the assumption that you
have previously defined the following pointer types with the TYPEDEF
directive:
PBYTE
NPBYTE
FPBYTE

TYPEDEF
PTR BYTE
TYPEDEF NEAR PTR BYTE
TYPEDEF FAR PTR BYTE

; Pointer to bytes
; Near pointer to bytes
; Far pointer to bytes

Initializing Pointer Variables
If the value of a pointer is known at assembly time, the assembler can initialize it
automatically so that no processing time is wasted on the task at run time. The
following example shows how to do this, placing the address of msg in the
pointer pmsg.
Msg
pMsg

BYTE
PBYTE

"String", 0
Msg

If a pointer variable can be conditionally defined to one of several constant
addresses, initialization must be delayed until run time. The technique is
different for near pointers than for far pointers, as shown here:
Msg1
Msg2
npMsg
fpMsg

BYTE
BYTE
NPBYTE
FPBYTE
.
.
.
mov

"String1"
"String2"
?
?

npMsg, OFFSET Msg1

; Load near pointer

mov
mov

WORD PTR fpMsg[0], OFFSET Msg2
WORD PTR fpMsg[2], SEG Msg2

; Load far offset
; Load far segment

If you know that the segment for a far pointer is in a register, you can load it
directly:
mov

WORD PTR fpMsg[2], ds

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 79 of 27 Printed: 10/02/00 04:23 PM

; Load segment of
;
far pointer

Programmer’s Guide

Dynamic Addresses
Often a pointer must point to a dynamic address, meaning the address depends
on a run-time condition. Typical situations include memory allocated by MSDOS (see “Interrupt 21h Function 48h” in Help) and addresses found by the
SCAS or CMPS instructions (see “Processing Strings” in Chapter 5). The
following illustrates the technique for saving dynamic addresses:
; Dynamically allocated buffer
fpBuf
FPBYTE 0
; Initialize so offset will be zero
.
.
.
mov
ah, 48h
; Allocate memory
mov
bx, 10h
; Request 16 paragraphs
int
21h
; Call DOS
jc
error
; Return segment in AX
mov
WORD PTR fpBuf[2], ax
; Load segment
.
;
(offset is already 0)
.
.
error:
; Handle error

Copying Pointers
Sometimes one pointer variable must be initialized by copying from another.
Here are two ways to copy a far pointer:
fpBuf1
fpBuf2

FPBYTE ?
FPBYTE ?
.
.
.
; Copy through registers is faster, but requires a spare register
mov
ax, WORD PTR fpBuf1[0]
mov
WORD PTR fpBuf2[0], ax
mov
ax, WORD PTR fpBuf1[2]
mov
WORD PTR fpBuf2[2], ax
; Copy through stack is slower, but does not use a register
push
WORD PTR fpBuf1[0]
push
WORD PTR fpBuf1[2]
pop
WORD PTR fpBuf2[2]
pop
WORD PTR fpBuf2[0]

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 80 of 28 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

Pointers as Arguments
Most high-level-language procedures and library functions accept arguments
passed on the stack. “Passing Arguments on the Stack” in Chapter 7 covers this
subject in detail. A pointer is passed in the same way as any other variable, as
this fragment shows:
; Push a far pointer (segment always pushed first)
push
WORD PTR fpMsg[2]
; Push segment
push
WORD PTR fpMsg[0]
; Push offset

Pushing an address has the same result as pushing a pointer to the address:
; Push a far address as a far pointer
mov
ax, SEG fVar
; Load and push segment
push
ax
mov
ax, OFFSET fVar ; Load and push offset
push
ax

On the 80186 and later processors, you can push a constant in one step:
push
push

SEG fVar
OFFSET fVar

; Push segment
; Push offset

Loading Addresses into Registers
Loading a near address into a register (or a far address into a pair of registers) is
a common task in assembly-language programming. To reference data pointed
to by a pointer, your program must first place the pointer into a register or pair
of registers.
Load far addresses as segment:offset pairs. The following pairs have specific
uses:
Segment:Offset Pair

Standard Use

DS:SI

Source for string operations

ES:DI

Destination for string operations

DS:DX

Input for certain DOS functions

ES:BX

Output from certain DOS functions

Addresses from Data Segments
For near addresses, you need only load the offset; the segment is assumed as SS
for stack-based data and as DS for other data. You must load both segment and
offset for far pointers.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 81 of 29 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Here is an example of loading an address into DS:BX from a near data segment:
Msg

.DATA
BYTE
.
.
.
mov

"String"

bx, OFFSET Msg

; Load address to BX
;
(DS already loaded)

Far data can be loaded like this:
.FARDATA
Msg
BYTE
.
.
.
mov
mov
mov

"String"

ax, SEG Msg
es, ax
bx, OFFSET Msg

; Load address to ES:BX

You can also read a far address from a pointer in one step, using the LES and
LDS instructions described next.

Far Pointers
The LES and LDS instructions load a far pointer into a segment pair. The
instructions copy the pointer’s low word into either ES or DS, and the high
word into a given register. The following example shows how to load a far
pointer into ES:DI:
OutBuf

BYTE

20 DUP (0)

fpOut

FPBYTE
.
.
.
les

OutBuf

di, fpOut

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 82 of 30 Printed: 10/02/00 04:23 PM

; Load far pointer into ES:DI

Chapter 3 Using Addresses and Pointers

Stack Variables
The technique for loading the address of a stack variable is significantly different
from the technique for loading near addresses. You may need to put the correct
segment value into ES for string operations. The following example illustrates
how to load the address of a local (stack) variable to ES:DI:
Task

PROC
LOCAL

Arg[4]:BYTE

push
pop
lea

ss
; Since it's stack-based, segment is SS
es
; Copy SS to ES
di, Arg ; Load offset to DI

The local variable in this case actually evaluates to SS:[BP-4]. This is an offset
from the stack frame (described in “Passing Arguments on the Stack,” Chapter
7). Since you cannot use the OFFSET operator to get the offset of an indirect
memory operand, you must use the LEA (Load Effective Address) instruction.

Direct Memory Operands
To get the address of a direct memory operand, use either the LEA instruction
or the MOV instruction with OFFSET. Though both methods have the same
effect, the MOV instruction produces smaller and faster code, as shown in this
example:
lea
mov

si, Msg
; Four byte instruction
si, OFFSET Msg ; Three byte equivalent

Copying Between Segment Pairs
Copying from one register pair to another is complicated by the fact that you
cannot copy one segment register directly to another. Two copying methods are
shown here. Timings are for the 8088 processor.
; Copy DS:SI to ES:DI, generating smaller code
push
ds
; 1 byte, 14 clocks
pop
es
; 1 byte, 12 clocks
mov
di, si
; 2 bytes, 2 clocks
; Copy DS:SI to ES:DI, generating faster code
mov
di, ds
; 2 bytes, 2 clocks
mov
es, di
; 2 bytes, 2 clocks
mov
di, si
; 2 bytes, 2 clocks

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 83 of 31 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Model-Independent Techniques
Often you may want to write code that is memory-model independent. If you
are writing libraries that must be available for different memory models, you can
use conditional assembly to handle different sizes of pointers. You can use the
predefined symbols @DataSize and @Model to test the current assumptions.
You can use conditional assembly to write code that works with pointer
variables that have no specified distance. The predefined symbol @DataSize
tests the pointer size for the current memory model:
Msg1
pMsg

BYTE
PBYTE
.
.
.
IF
mov
mov
ELSE
mov
ENDIF

"String1"
?

@DataSize
WORD PTR pMsg[0], OFFSET Msg1
WORD PTR pMsg[2], SEG Msg1
pMsg, OFFSET Msg1

;
;
;
;
;

@DataSize > 0 for far
Load far offset
Load far segment
@DataSize = 0 for near
Load near pointer

In the following example, a procedure receives as an argument a pointer to a
word variable. The code inside the procedure uses @DataSize to determine
whether the current memory model supports far or near data. It loads and
processes the data accordingly:
; Procedure that receives an argument by reference
mul8
PROC
arg:PTR WORD

mul8

IF
les
mov
ELSE
mov
mov
ENDIF
shl
shl
shl
ret
ENDP

@DataSize
bx, arg
; Load far pointer to ES:BX
ax, es:[bx] ; Load the data pointed to
bx, arg
ax, [bx]

; Load near pointer to BX (assume DS)
; Load the data pointed to

ax, 1
ax, 1
ax, 1

; Multiply by 8

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 84 of 32 Printed: 10/02/00 04:23 PM

Chapter 3 Using Addresses and Pointers

If you have many routines, writing the conditionals for each case can be tedious.
The following conditional statements automatically generate the proper
instructions and segment overrides.
; Equates for conditional handling of pointers
IF @DataSize
lesIF
TEXTEQU

ldsIF
TEXTEQU

esIF
TEXTEQU

ELSE
lesIF
TEXTEQU

ldsIF
TEXTEQU

esIF
TEXTEQU
<>
ENDIF

Once you define these conditionals, you can use them to simplify code that must
handle several types of pointers. This next example rewrites the above mul8
procedure to use conditional code.
mul8

PROC

arg:PTR WORD
bx,
ax,
ax,
ax,
ax,

mul8

lesIF
mov
shl
shl
shl
ret
ENDP

arg
esIF [bx]
1
1
1

; Load pointer to BX or ES:BX
; Load the data from [BX] or ES:[BX]
; Multiply by 8

The conditional statements from these examples can be defined once in an
include file and used whenever you need to handle pointers.

Filename: LMAPGC03.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 49 Page: 85 of 33 Printed: 10/02/00 04:23 PM

C H A P T E R

Defining and Using Simple
Data Types

This chapter covers the concepts essential for working with simple data types in
assembly-language programs. The first section shows how to declare integer
variables. The second section describes basic operations including moving,
loading, and sign-extending numbers, as well as calculating. The last section
describes how to do various operations with numbers at the bit level, such as
using bitwise logical instructions and shifting and rotating bits.
The complex data types introduced in the next chapter — arrays, strings,
structures, unions, and records — use many of the operations illustrated in this
chapter. Floating-point operations require a different set of instructions and
techniques. These are covered in Chapter 6, “Using Floating-Point and Binary
Coded Decimal
Numbers.”

Declaring Integer Variables
An integer is a whole number, such as 4 or 4,444. Integers have no fractional
part, as do the real numbers discussed in Chapter 6. You can initialize integer
variables in several ways with the data allocation directives. This section
explains how to use the SIZEOF and TYPE operators to provide information to
the assembler about the types in your program. For information on symbolic
integer constants, see “Integer Constants and Constant Expressions” in Chapter
1.

Allocating Memory for Integer Variables
When you declare an integer variable by assigning a label to a data allocation
directive, the assembler allocates memory space for the integer. The variable’s
name becomes a label for the memory space. The syntax is:
[[name]] directive initializer

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 85 of 1

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Programmer’s Guide

The following directives indicate the integer’s size and value range:
Directive

Description of Initializers

BYTE, DB (byte)

Allocates unsigned numbers from 0 to 255.

SBYTE (signed byte)

Allocates signed numbers from –128 to +127.

WORD, DW (word = 2 bytes)

Allocates unsigned numbers from
0 to 65,535 (64K).

SWORD (signed word)

Allocates signed numbers from
–32,768 to +32,767.

DWORD, DD (doubleword = 4

Allocates unsigned numbers from
0 to 4,294,967,295 (4 megabytes).

bytes),
SDWORD (signed doubleword)

Allocates signed numbers from
–2,147,483,648 to +2,147,483,647.

FWORD, DF (farword = 6 bytes)

Allocates 6-byte (48-bit) integers. These values are
normally used only as pointer variables on the
80386/486 processors.

QWORD, DQ (quadword = 8 bytes)

Allocates 8-byte integers used with 8087-family
coprocessor instructions.

TBYTE, DT (10 bytes),

Allocates 10-byte (80-bit) integers if the initializer has
a radix specifying the base of the number.

See Chapter 6 for information on the REAL4, REAL8, and REAL10 directives
that allocate real numbers.
The SIZEOF and TYPE operators, when applied to a type, return the size of an
integer of that type. The size attribute associated with each data type is:
Data Type

Bytes

BYTE, SBYTE

WORD, SWORD

DWORD, SDWORD

FWORD

QWORD

TBYTE

The data types SBYTE, SWORD, and SDWORD tell the assembler to treat the
initializers as signed data. It is important to use these signed types with high-level
constructs such as .IF, .WHILE, and .REPEAT, and with PROTO and
INVOKE directives. For descriptions of these directives, see the sections
“Loop-Generating Directives,” “Declaring Procedure Prototypes,” and “Calling
Procedures with INVOKE” in Chapter 7.

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 86 of 2

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

The assembler stores integers with the least significant bytes lowest in memory.
Note that assembler listings and most debuggers show the bytes of a word in the
opposite order — high byte first.
Figure 4.1 illustrates the integer formats.

Figure 4.1

Integer Formats

Although the TYPEDEF directive’s primary purpose is to define pointer
variables (see “Defining Pointer Types with TYPEDEF” in Chapter 3), you can
also use TYPEDEF to create an alias for any integer type. For example, these
declarations
char
long
float
double

TYPEDEF
TYPEDEF
TYPEDEF
TYPEDEF

SBYTE
DWORD
REAL4
REAL8

allow you to use char, long, float, or double in your programs if you
prefer the C data labels.

Data Initialization
You can initialize variables when you declare them with constants or expressions
that evaluate to constants. The assembler generates an error if you specify an
initial value too large for the variable type.
A ? in place of an initializer indicates you do not require the assembler to
initialize the variable. The assembler allocates the space but does not write in it.
Use ? for buffer areas or variables your program will initialize at run time.

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 87 of 3

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Programmer’s Guide

You can declare and initialize variables in one step with the data directives, as
these examples show.
integer
negint
expression
signedexp
empty
long

BYTE
SBYTE
WORD
SWORD
QWORD
BYTE
DWORD

longnum

SDWORD

TBYTE

16
-16
4*3
4*3
?
1,2,3,4,5,6
4294967295

;
;
;
;
;
;
;
;
-2147433648 ;
;
2345t
;

Initialize byte to 16
Initialize signed byte to -16
Initialize word to 12
Initialize signed word to 12
Allocate uninitialized long int
Initialize six unnamed bytes
Initialize doubleword to
4,294,967,295
Initialize signed doubleword
to -2,147,433,648
Initialize 10-byte binary number

For information on arrays and on using the DUP operator to allocate initializer
lists, see “Arrays and Strings” in Chapter 5.

Working with Simple Variables
Once you have declared integer variables in your program, you can use them to
copy, move, and sign-extend integer variables in your MASM code. This section
shows how to do these operations as well as how to add, subtract, multiply, and
divide numbers and do bit-level manipulations with logical, shift, and rotate
instructions.
Since MASM instructions require operands to be the same size, you may need
to operate on data in a size other than that originally declared. You can do this
with the PTR operator. For example, you can use the PTR operator to access
the high-order word of a DWORD-size variable. The syntax for the PTR
operator is
type PTR expression
where the PTR operator forces expression to be treated as having the type
specified. An example of this use is
num

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 88 of 4

.DATA
DWORD
.CODE

mov
mov

ax, WORD PTR num[0] ; Loads a word-size value from
dx, WORD PTR num[2] ;
a doubleword variable

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

Copying Data
The primary instructions for moving data from operand to operand and loading
them into registers are MOV (Move), XCHG (Exchange), CWD (Convert
Word to Double), and CBW (Convert Byte to Word).

Moving Data
The most common method of moving data, the MOV instruction, is essentially
a copy instruction, since it always copies the source operand to the destination
operand without affecting the source. After a MOV instruction, the source and
destination operands contain the same value.
The following example illustrates the MOV instruction. As explained in
“General-Purpose Registers,” Chapter 1, you cannot move a value from one
location in memory to another in a single operation.
; Immediate value moves
mov
ax, 7
mov
mem, 7
mov
mem[bx], 7

; Immediate to register
; Immediate to memory direct
; Immediate to memory indirect

; Register moves
mov
mem, ax
; Register to memory direct
mov
mem[bx], ax ; Register to memory indirect
mov
ax, bx
; Register to register
mov
ds, ax
; General register to segment register
; Direct memory moves
mov
ax, mem
mov
ds, mem

; Memory direct to register
; Memory to segment register

; Indirect memory moves
mov
ax, mem[bx] ; Memory indirect to register
mov
ds, mem[bx] ; Memory indirect to segment register
; Segment register moves
mov
mem, ds
; Segment register to memory
mov
mem[bx], ds ; Segment register to memory indirect
mov
ax, ds
; Segment register to general register

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 89 of 5

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Programmer’s Guide

The following example shows several common types of moves that require two
instructions.
; Move immediate to segment register
mov
ax, DGROUP ; Load AX with immediate value
mov
ds, ax
; Copy AX to segment register
; Move memory to memory
mov
ax, mem1
mov
mem2, ax

; Load AX with memory value
; Copy AX to other memory

; Move segment register to segment register
mov
ax, ds
; Load AX with segment register
mov
es, ax
; Copy AX to segment register

The MOVSX and MOVZX instructions for the 80386/486 processors extend
and copy values in one step. See “Extending Signed and Unsigned Integers,”
following.

Exchanging Integers
The XCHG (Exchange) instruction exchanges the data in the source and
destination operands. You can exchange data between registers or between
registers and memory, but not from memory to memory:

;

xchg
xchg
xchg

ax, bx
memory, ax
mem1, mem2

; Put AX in BX and BX in AX
; Put "memory" in AX and AX in "memory"
; Illegal- can't exchange memory locations

Extending Signed and Unsigned Integers
Since moving data between registers of different sizes is illegal, you must “signextend” integers to convert signed data to a larger size. Sign-extending means
copying the sign bit of the unextended operand to all bits of the operand’s next
larger size. This widens the operand while maintaining its sign and value.
8086-based processors provide four instructions specifically for sign-extending.
The four instructions act only on the accumulator register (AL, AX, or EAX), as
shown in the following list.
Instruction

Sign-extend

CBW (convert byte to word)

AL to AX

CWD (convert word to doubleword)

AX to DX:AX

CWDE (convert word to doubleword extended)*

AX to EAX

CDQ (convert doubleword to quadword)*

EAX to EDX:EAX

*Requires an extended register and applies only to 80386/486 processors.

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 90 of 6

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

On the 80386/486 processors, the CWDE instruction converts a signed 16-bit
value in AX to a signed 32-bit value in EAX. The CDQ instruction converts a
signed 32-bit value in EAX to a signed 64-bit value in the EDX:EAX register
pair.
This example converts signed integers using CBW, CWD, CWDE, and CDQ.
mem8
mem16
mem32

.DATA
SBYTE
SWORD
SDWORD
.CODE
.
.
.
mov
cbw
mov
cwd
mov
cwde
mov
cdq

-5
+5
-5

al, mem8
ax, mem16
ax, mem16
eax, mem32

;
;
;
;
;
;
;
;
;

Load 8-bit -5 (FBh)
Convert to 16-bit -5 (FFFBh) in AX
Load 16-bit +5
Convert to 32-bit +5 (0000:0005h) in DX:AX
Load 16-bit +5
Convert to 32-bit +5 (00000005h) in EAX
Load 32-bit -5 (FFFFFFFBh)
Convert to 64-bit -5
(FFFFFFFF:FFFFFFFBh) in EDX:EAX

These four instructions efficiently convert unsigned values as well, provided the
sign bit is zero. This example, for instance, correctly widens mem16 whether
you treat the variable as signed or unsigned.
The processor does not differentiate between signed and unsigned values. For
instance, the value of mem8 in the previous example is literally 251 (0FBh) to
the processor. It ignores the human convention of treating the highest bit as an
indicator of sign. The processor can ignore the distinction between signed and
unsigned numbers because binary arithmetic works the same in either case.
If you add 7 to mem8, for example, the result is 258 (102h), a value too large to
fit into a single byte. The byte-sized mem8 can accommodate only the leastsignificant digits of the result (02h), and so receives the value of 2. The result is
the same whether we treat mem8 as a signed value (-5) or unsigned value (251).
This overview illustrates how the programmer, not the processor, must keep
track of which values are signed or unsigned, and treat them accordingly. If
AL=127 (01111111y), the instruction CBW sets AX=127 because the sign bit is
zero. If AL=128 (10000000y), however, the sign bit is 1. CBW thus sets
AX=65,280

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 91 of 7

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Programmer’s Guide

(FF00h), which may not be what you had in mind if you assumed AL originally
held an unsigned value.To widen unsigned values, explicitly set the higher
register to zero, as shown in the following example:
mem8
mem16

.DATA
BYTE
WORD
.CODE
.
.
.
mov
sub

251
251

al, mem8
ah, ah

; Load 251 (FBh) from 8-bit memory
; Zero upper half (AH)

mov
sub

ax, mem16 ; Load 251 (FBh) from 16-bit memory
dx, dx
; Zero upper half (DX)

sub
mov

eax, eax ; Zero entire extended register (EAX)
ax, mem16 ; Load 251 (FBh) from 16-bit memory

The 80386/486 processors provide instructions that move and extend a value to
a larger data size in a single step. MOVSX moves a signed value into a register
and sign-extends it. MOVZX moves an unsigned value into a register and zeroextends it.
; 80386/486 instructions
movzx
dx, bl

; Load unsigned 8-bit value into
;
16-bit register and zero-extend

These special 80386/486 instructions usually execute much faster than the
equivalent 8086/286 instructions.

Adding and Subtracting Integers
You can use the ADD, ADC, INC, SUB, SBB, and DEC instructions for
adding, incrementing, subtracting, and decrementing values in single registers.
You can also combine them to handle larger values that require two registers for
storage.

Adding and Subtracting Integers Directly
The ADD, INC (Increment), SUB, and DEC (Decrement) instructions operate
on 8- and 16-bit values on the 8086–80286 processors, and on 8-, 16-, and 32bit values on the 80386/486 processors. They can be combined with the ADC
and SBB instructions to work on 32-bit values on the 8086 and 64-bit values on
the 80386/486 processors. (See “Adding and Subtracting in Multiple Registers,”
following.)

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 92 of 8

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

These instructions have two requirements:
1. If there are two operands, only one operand can be a memory operand.
2. If there are two operands, both must be the same size.
To meet the second requirement, you can use the PTR operator to force an
operand to the size required. (See “Working with Simple Variables,” previous.)
For example, if Buffer is an array of bytes and BX points to an element of the
array, you can add a word from Buffer with
add

ax, WORD PTR Buffer[bx] ; Add word from byte array

The next example shows 8-bit signed and unsigned addition and subtraction.
mem8

.DATA
BYTE
.CODE

; Addition

mov

;
;
;
;
;
;
al, mem8 ;
;
ah, al
;

add

al, ah

mov
inc
add

add

al, 26
al
al, 76

Start with register
Increment
Add immediate

Add memory
Copy to AH

; Add register
;
;

signed
unsigned
26
26
1
1
76
+ 76
------103
103
39
+ 39
-------114
142
+overflow
142
---28+carry

; Subtraction
mov
dec
sub

sub

mov
sub

Filename: LMAPGC04.DOC
Template: MSGRIDA1.DOT
Revision #: 2 Page: 93 of 9

;
;
;
;
;
;
al, mem8 ;
;
;

al, 95
al
al, 23

ah, 119
al, ah

Project:
Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Printed: 10/02/00 04:23 PM

Load register
Decrement
Subtract immediate

Subtract memory

; Load register
;
and subtract
;
;

signed
95
-1
-23
---71
-122
----51

unsigned
95
-1
-23
---71
-122
---205+sign

119
-51
---86+overflow

Programmer’s Guide

The INC and DEC instructions treat integers as unsigned values and do not
update the carry flag for signed carries and borrows.
When the sum of 8-bit signed operands exceeds 127, the processor sets the
overflow flag. (The overflow flag is also set if both operands are negative and
the sum is less than or equal to -128.) Placing a JO (Jump on Overflow) or
INTO (Interrupt on Overflow) instruction in your program at this point can
transfer control to error-recovery statements. When the sum exceeds 255, the
processor sets the carry flag. A JC (Jump on Carry) instruction at this point can
transfer control to error-recovery statements.
In the previous subtraction example, the processor sets the sign flag if the result
goes below 0. At this point, you can use a JS (Jump on Sign) instruction to
transfer control to error-recovery statements. Jump instructions are described in
the “Jumps” section in Chapter 7.

Adding and Subtracting in Multiple Registers
You can add and subtract numbers larger than the register size on your
processor with the ADC (Add with Carry) and SBB (Subtract with Borrow)
instructions. If the operations prior to an ADC or SBB instruction do not set the
carry flag, these instructions are identical to ADD and SUB. When you operate
on large values in more than one register, use ADD and SUB for the least
significant part of the number and ADC or SBB for the most significant part.
The following example illustrates multiple-register addition and subtraction. You
can also use this technique with 64-bit operands on the 80386/486 processors.
.DATA
DWORD
DWORD
DWORD
.CODE
.
.
.
; Addition
mov
sub
add
adc

mem32
mem32a
mem32b

; Subtraction
mov
mov
sub
sbb

316423
316423
156739

ax,
dx,
ax,
dx,

43981
dx
WORD PTR mem32[0]
WORD PTR mem32[2]

; Load immediate
43981
;
into DX:AX
; Add to both
+ 316423
;
memory words
-----; Result in DX:AX
360404

ax,
dx,
ax,
dx,

WORD
WORD
WORD
WORD

; Load mem32
316423
;
into DX:AX
; Subtract low
- 156739
;
then high
-----; Result in DX:AX
159684

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 94 of 10 Printed: 10/02/00 04:23 PM

PTR
PTR
PTR
PTR

mem32a[0]
mem32a[2]
mem32b[0]
mem32b[2]

Chapter 4 Defining and Using Simple Data Types

For 32-bit registers on the 80386/486 processors, only two steps are necessary.
If your program needs to be assembled for more than one processor, you can
assemble the statements conditionally, as shown in this example:
.DATA
DWORD
DWORD
DWORD
TEXTEQU
.CODE
.
.
.
; Addition
IF
mov
add
ELSE
.
.
.
ENDIF
mem32
mem32a
mem32b
p386

; Subtraction
IF
mov
sub
ELSE
.
.
.
ENDIF

316423
316423
156739
(@Cpu AND 08h)

p386
eax, 43981
eax, mem32

; Load immediate
; Result in EAX

; do steps in previous example

p386
eax, mem32a ; Load memory
eax, mem32b ; Result in EAX

; do steps in previous example

Since the status of the carry flag affects the results of calculations with ADC
and SBB, be sure to turn off the carry flag with the CLC (Clear Carry Flag)
instruction or use ADD or SUB for the first calculation, when appropriate.

Multiplying and Dividing Integers
The 8086 family of processors uses different multiplication and division
instructions for signed and unsigned integers. Multiplication and division
instructions also have special requirements depending on the size of the
operands and the processor the code runs on.

Using Multiplication Instructions
The MUL instruction multiplies unsigned numbers. IMUL multiplies signed
numbers. For both instructions, one factor must be in the accumulator register
(AL for 8-bit numbers, AX for 16-bit numbers, EAX for 32-bit numbers). The

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 95 of 11 Printed: 10/02/00 04:23 PM

Programmer’s Guide

other factor can be in any single register or memory operand. The result
overwrites the contents of the accumulator register.
Multiplying two 8-bit numbers produces a 16-bit result returned in AX.
Multiplying two 16-bit operands yields a 32-bit result in DX:AX. The 80386/486
processor handles 64-bit products in the same way in the EDX:EAX pair.
This example illustrates multiplication of signed 16- and 32-bit integers.
.DATA
SWORD
-30000
.CODE
.
.
.
; 8-bit unsigned multiply
mov
al, 23
mov
bl, 24
mul
bl

mem16

; 16-bit signed multiply
mov
ax, 50
imul

mem16

;
;
;
;
;

Load AL
23
Load BL
* 24
Multiply BL
----Product in AX
552
overflow and carry set

; Load AX
50
;
-30000
; Multiply memory
----; Product in DX:AX
-1500000
;
overflow and carry set

A nonzero number in the upper half of the result (AH for byte, DX or EDX for
word) sets the overflow and carry flags.
On the 80186–80486 processors, the IMUL instruction supports three
additional operand combinations. The first syntax option allows for 16-bit
multipliers producing a 16-bit product or 32-bit multipliers for 32-bit products on
the 80386/486. The result overwrites the destination. The syntax for this
operation is:
IMUL register16, immediate
The second syntax option specifies three operands for IMUL. The first operand
must be a 16-bit register operand, the second a 16-bit memory (or register)
operand, and the third a 16-bit immediate operand. IMUL multiplies the
memory (or register) and immediate operands and stores the product in the
register operand with this syntax:
IMUL register16,{ memory16 | register16}, immediate

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 96 of 12 Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

For the 80386/486 only, a third option for IMUL allows an additional operand
for multiplication of a register value by a register or memory value. The syntax
is:
IMUL register,{register | memory}
The destination can be any 16-bit or 32-bit register. The source must be the
same size as the destination.
In all of these options, products too large to fit in 16 or 32 bits set the overflow
and carry flags. The following examples show these three options for IMUL.
imul
imul

dx, 456
ax, [bx],6

; Multiply DX times 456 on 80186-80486
; Multiply the value pointed to by BX
;
by 6 and put the result in AX

imul
imul

dx, ax
ax, [bx]

; Multiply DX times AX on 80386
; Multiply AX by the value pointed to
;
by BX on 80386

The IMUL instruction with multiple operands can be used for either signed or
unsigned multiplication, since the 16-bit product is the same in either case. To
get a 32-bit result, you must use the single-operand version of MUL or IMUL.

Using Division Instructions
The DIV instruction divides unsigned numbers, and IDIV divides signed
numbers. Both return a quotient and a remainder.
Table 4.1 summarizes the division operations. The dividend is the number to be
divided, and the divisor is the number to divide by. The quotient is the result.
The divisor can be in any register or memory location except the registers where
the quotient and remainder are returned.
Table 4.1 Division Operations
Size of
Operand

Dividend
Register

Size of
Divisor

Quotient

Remainder

16 bits

8 bits

32 bits

DX:AX

16 bits

64 bits (80386
and 80486)

EDX:EAX

32 bits

EAX

EDX

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 97 of 13 Printed: 10/02/00 04:23 PM

Programmer’s Guide

Unsigned division does not require careful attention to flags. The following
examples illustrate signed division, which can be more complex.
.DATA
SWORD
SDWORD
.CODE
.
.
.
; Divide 16-bit
mov
mov
div
mem16
mem32

; Divide 32-bit
mov
mov
idiv

; Divide 16-bit
mov
cwd
mov
idiv

-2000
500000

unsigned by 8-bit
ax, 700
bl, 36
bl

;
;
;
;
;

signed by 16-bit
ax, WORD PTR mem32[0] ;
dx, WORD PTR mem32[2] ;
mem16
;
;
;
;
signed by 16-bit
ax, WORD PTR mem16
bx,-421
bx

;
;
;
;
;
;

Load dividend
700
Load divisor DIV
36
Divide BL
-----Quotient in AL
19
Remainder in AH

Load into DX:AX
500000
DIV -2000
Divide memory
-----Quotient in AX
-250
Remainder in DX

Load into AX
-2000
Extend to DX:AX
DIV -421
Divide by BX
----Quotient in AX
4
Remainder in DX
-316

If the dividend and divisor are the same size, sign-extend or zero-extend the
dividend so that it is the length expected by the division instruction. See
“Extending Signed and Unsigned Integers,” earlier in this chapter.

Manipulating Numbers at the Bit Level
The instructions introduced so far in this chapter access numbers at the byte or
word level. The logical, shift, and rotate instructions described in this section
access individual bits in a number. You can use logical instructions to evaluate
characters and do other text and screen operations. The shift and rotate
instructions do similar tasks by shifting and rotating bits through registers. This
section reviews some applications of these bit-level operations.

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 98 of 14 Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

Logical Instructions
The logical instructions AND, OR, and XOR compare bits in two operands.
Based on the results of the comparisons, the instructions alter bits in the first
(destination) operand. The logical instruction NOT also changes bits, but
operates on a single operand.
The following list summarizes these four logical instructions. The list makes
reference to the “destination bit,” meaning the bit in the destination operand.
The terms “both bits” and “either bit” refer to the corresponding bits in the
source and destination operands. These instructions include:
Instruction

Sets Destination Bit If

Clears Destination Bit If

AND

Both bits set

Either or both bits clear

Either or both bits set

Both bits clear

XOR

Either bit (but not both) set

Both bits set or both clear

NOT

Destination bit clear

Destination bit set

Note Do not confuse logical instructions with the logical operators, which
perform these operations at assembly time, not run time. Although the names
are the same, the assembler recognizes the difference.
The following example shows the result of the AND, OR, XOR, and NOT
instructions operating on a value in the AX register and in a mask. A mask is
any number with a pattern of bits set for an intended operation.
mov
and

ax, 035h
ax, 0FBh

ax, 016h

xor

ax, 0ADh

not

;
;
;
;
;
;
;
;
;
;
;

Load value
Clear bit 2
Value is now 31h
Set bits 4,2,1
Value is now 37h
Toggle bits 7,5,3,2,0
Value is now 9Ah
Value is now 65h

00110101
AND 11111011
-------00110001
OR 00010110
-------00110111
XOR 10101101
-------10011010
01100101

The AND instruction clears unmasked bits — that is, bits not protected by 1 in
the mask. To mask off certain bits in an operand and clear the others, use an
appropriate masking value in the source operand. The bits of the mask should
be 0 for any bit positions you want to clear and 1 for any bit positions you want
to remain unchanged.

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 99 of 15 Printed: 10/02/00 04:23 PM

100

Programmer’s Guide

The OR instruction forces specific bits to 1 regardless of their current settings.
The bits of the mask should be 1 for any bit positions you want to set and 0 for
any bit positions you want to remain unchanged.
The XOR instruction toggles the value of specific bits on and off — that is,
reverses them from their current settings. This instruction sets a bit to 1 if the
corresponding bits are different or to 0 if they are the same. The bits of the
mask should be 1 for any bit positions you want to toggle and 0 for any bit
positions you want to remain unchanged.
The following examples show an application for each of these instructions. The
code illustrating the AND instruction converts a “y” or “n” read from the
keyboard to uppercase, since bit 5 is always clear in uppercase letters. In the
example for OR, the first statement is faster and uses fewer bytes than cmp
bx, 0. When the operands for XOR are identical, each bit cancels itself,
producing 0.
;AND example - converts characters to uppercase
mov
ah, 7
; Get character without echo
int
21h
and
al, 11011111y
; Convert to uppercase by clearing bit 5
cmp
al, 'Y'
; Is it Y?
je
yes
; If so, do Yes actions
.
; Else do No actions
.
yes:
.
;OR example - compares operand to 0
or
bx, bx
; Compare to 0
jg
positive
; BX is positive
jl
negative
; BX is negative
; else BX is zero
;XOR example - sets a register to 0
xor
cx, cx
; 2 bytes, 3 clocks on 8088
sub
cx, cx
; 2 bytes, 3 clocks on 8088
mov
cx, 0
; 3 bytes, 4 clocks on 8088

On the 80386/486 processors, the BSF (Bit Scan Forward) and the BSR (Bit
Scan Reverse) instructions perform operations like those of the logical
instructions. They scan the contents of a register to find the first-set or last-set
bit. You can use BSF or BSR to find the position of a set bit in a mask or to
check if a register value is 0.

Shifting and Rotating Bits
The 8086-based processors provide a complete set of instructions for shifting
and rotating bits. Shift instructions move bits a specified number of places to the

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 100 of 16 Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

101

right or left. The last bit in the direction of the shift goes into the carry flag, and
the first bit is filled with 0 or with the previous value of the first bit.
Rotate instructions also move bits a specified number of places to the right or
left. For each bit rotated, the last bit in the direction of the rotate operation
moves into the first bit position at the other end of the operand. With some
variations, the carry bit is used as an additional bit of the operand. Figure 4.2
illustrates the eight variations of shift and rotate instructions for 8-bit operands.
Notice that SHL and SAL are identical.

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 101 of 17 Printed: 10/02/00 04:23 PM

102

Programmer’s Guide
Figure 4.2

Shifts and Rotates

All shift instructions use the same format. Before the instruction executes, the
destination operand contains the value to be shifted; after the instruction
executes, it contains the shifted operand. The source operand contains the
number of bits to shift or rotate. It can be the immediate value 1 or the CL
register. The 8088 and 8086 processors do not accept any other values or
registers with these instructions.
Starting with the 80186 processor, you can use 8-bit immediate values larger
than 1 as the source operand for shift or rotate instructions, as shown here:
shr

bx, 4

; 9 clocks, 3 bytes on 80286

The following statements are equivalent if the program must run on the 8088 or
8086 processor:
mov
shr

cl, 4
bx, cl

; 2 clocks, 3 bytes on 80286
; 9 clocks, 2 bytes on 80286
; 11 clocks, 5 bytes total

Masks for logical instructions can be shifted to new bit positions. For example,
an operand that masks off a bit or group of bits can be shifted to move the mask
to a different position, allowing you to mask off a different bit each time the
mask is used. This technique, illustrated in the following example, is useful only
if the mask value is unknown until run time.
masker

.DATA
BYTE
.CODE
.
.
.
mov
mov
rol
or
rol
or

00000010y

; Mask that may change at run time

cl, 2
bl, 57h
masker, cl
bl, masker

;
;
;
;
;
;
;
;

masker, cl
bl, masker

Rotate two at a time
Load value to be changed
Rotate two to left
Turn on masked values
New value is 05Fh
Rotate two more
Turn on masked values
New value is 07Fh

01010111y
00001000y
--------01011111y
00100000y
--------01111111y

Multiplying and Dividing with Shift Instructions
You can use the shift and rotate instructions (SHR, SHL, SAR, and SAL) for
multiplication and division. Shifting a value right by one bit has the effect of
dividing by two; shifting left by 1 bit has the effect of multiplying by two. You
can take advantage of shifts to do fast multiplication and division by powers of

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 102 of 18 Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

103

two. For example, shifting left twice multiplies by four, shifting left three times
multiplies by eight, and so on.
Use SHR (Shift Right) to divide unsigned numbers. You can use SAR (Shift
Arithmetic Right) to divide signed numbers, but SAR rounds negative numbers
down — IDIV always rounds negative numbers up (toward 0). Division using
SAR must adjust for this difference. Multiplication by shifting is the same for
signed and unsigned numbers, so you can use either SAL or SHL.
Multiply and divide instructions are relatively slow, particularly on the 8088 and
8086 processors. When multiplying or dividing by a power of two, use shifts to
speed operations by a factor of 10 or more. For example, these statements take
only four clocks on an 8088 or 8086 processor:
sub
shl

ah, ah
ax, 1

; Clear AH
; Multiply byte in AL by 2

The following statements produce the same results, but take between 74 and 81
clocks on the 8088 or 8086 processors. The same statements take 15 clocks on
the 80286 and between 11 and 16 clocks on the 80386. (For a discussion about
instruction timings, see “A Word on Instruction Timings” in the Introduction.)
mov
mul

bl, 2
bl

; Multiply byte in AL by 2

As the following macro shows, it’s possible to multiply by any number — in this
case, 10 — without resorting to the MUL instruction. However, such a
procedure is no more than an interesting arithmetic exercise, since the additional
code almost certainly takes more time to execute than a single MUL. You
should consider using shifts in your program only when multiplying or dividing
by a power of two.
mul_10

MACRO
mov
shl
mov
shl
shl
add
ENDM

factor
ax, factor
ax, 1
bx, ax
ax, 1
ax, 1
ax, bx

;
;
;
;
;
;
;
;

Factor must be unsigned
Load into AX
AX = factor * 2
Save copy in BX
AX = factor * 4
AX = factor * 8
AX = (factor * 8) + (factor * 2)
AX = factor * 10

Here’s another macro that divides by 512. In contrast to the previous example,
this macro uses little code and operates faster than an equivalent DIV
instruction.

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 103 of 19 Printed: 10/02/00 04:23 PM

104

Programmer’s Guide
div_512 MACRO
mov
shr
xchg

dividend
ax, dividend
ax, 1
al, ah

cbw
ENDM

;
;
;
;
;
;
;

Dividend must be unsigned
Load into AX
AX = dividend / 2 (unsigned)
XCHG is like rotate right 8
AL = (dividend / 2) / 256
Clear upper byte
AX = (dividend / 512)

If you need to shift a value that is too large to fit in one register, you can shift
each part separately. The RCR (Register Carry Right) and RCL (Register Carry
Left) instructions carry values from the first register to the second by passing the
leftmost or rightmost bit through the carry flag.
This example shifts a multiword value.
mem32

.DATA
DWORD
.CODE

; Divide 32-bit
mov
again: shr
rcr
loop

500000

unsigned by 16
cx, 4
; Shift right 4
WORD PTR mem32[2], 1 ; Shift into carry
WORD PTR mem32[0], 1 ; Rotate carry in
again
;

DIV

500000
16
-----31250

Since the carry flag is treated as part of the operand (it’s like using a 9-bit or 17bit operand), the flag value before the operation is crucial. The carry flag can be
adjusted by a previous instruction, but you can also set or clear the flag directly
with the CLC (Clear Carry Flag), CMC (Complement Carry Flag), and STC
(Set Carry Flag) instructions.
On the 80386 and 80486 processors, an alternate method for multiplying quickly
by constants takes advantage of the LEA (Load Effective Address) instruction
and the scaling of indirect memory operands. By using a 32-bit value as both the
index and the base register in an indirect memory operand, you can multiply by
the constants 2, 3, 4, 5, 8, and 9 more quickly than you can by using the MUL
instruction. LEA calculates the offset of the source operand and stores it into
the destination register, EBX, as this example shows:
lea
lea
lea
lea
lea
lea

ebx,
ebx,
ebx,
ebx,
ebx,
ebx,

[eax*2]
[eax*2+eax]
[eax*4]
[eax*4+eax]
[eax*8]
[eax*8+eax]

;
;
;
;
;
;

EBX
EBX
EBX
EBX
EBX
EBX

=
=
=
=
=
=

2
3
4
5
8
9

*
*
*
*
*
*

EAX
EAX
EAX
EAX
EAX
EAX

Scaling of 80386 indirect memory operands is reviewed in “Indirect Memory
Operands with 32-Bit Registers” in Chapter 3. LEA is introduced in “Loading
Addresses into Registers” in Chapter 3.

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 104 of 20 Printed: 10/02/00 04:23 PM

Chapter 4 Defining and Using Simple Data Types

105

The next chapter deals with more complex data types — arrays, strings,
structures, unions, and records. Many of the operations presented in this chapter
can also be applied to the data structures covered in Chapter 5, “Defining and
Using Complex Data Types.”

Filename: LMAPGC04.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 105 of 21 Printed: 10/02/00 04:23 PM

105

C H A P T E R

Defining and Using Complex
Data Types

With the complex data types available in MASM 6.1 — arrays, strings, records,
structures, and unions — you can access data as a unit or as individual elements
that make up a unit. The individual elements of complex data types are often the
integer types discussed in Chapter 4, “Defining and Using Simple Data Types.”
“Arrays and Strings” reviews how to declare, reference, and initialize arrays and
strings. This section summarizes the general steps needed to process arrays and
strings and describes the MASM instructions for moving, comparing, searching,
loading, and storing.
“Structures and Unions” covers similar information for structures and unions:
how to declare structure and union types, how to define structure and union
variables, and how to reference structures and unions and their fields.
“Records” explains how to declare record types, define record variables, and use
record operators.

Arrays and Strings
An array is a sequential collection of variables, all of the same size and type,
called “elements.” A string is an array of characters. For example, in the string
“ABC,” each letter is an element. You can access the elements in an array or
string relative to the first element. This section explains how to handle arrays
and strings in your programs.

Declaring and Referencing Arrays
Array elements occupy memory contiguously, so a program references each
element relative to the start of the array. To declare an array, supply a label
name, the element type, and a series of initializing values or ? placeholders. The
following examples declare the arrays warray and xarray:
warray
xarray

WORD
DWORD

1, 2, 3, 4
0FFFFFFFFh, 789ABCDEh

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 105 of 1 Printed: 10/02/00 04:23 PM

106

Programmer’s Guide

Initializer lists of array declarations can span multiple lines. The first initializer
must appear on the same line as the data type, all entries must be initialized,
and, if you want the array to continue to the new line, the line must end with a
comma. These examples show legal multiple-line array declarations:
big

BYTE

21, 22, 23, 24, 25,
26, 27, 28

somelist

WORD

10,
20,
30

If you do not use the LENGTHOF and SIZEOF operators discussed later in
this section, an array may span more than one logical line, although a separate
type declaration is needed on each logical line:
var1

BYTE
BYTE
BYTE

10, 20, 30
40, 50, 60
70, 80, 90

The DUP Operator
You can also declare an array with the DUP operator. This operator works with
any of the data allocation directives described in “Allocating Memory for Integer
Variables” in Chapter 4. In the syntax
count DUP (initialvalue [[, initialvalue]]...)
the count value sets the number of times to repeat all values within the
parentheses. The initialvalue can be an integer, character constant, or another
DUP operator, and must always appear within parentheses. For example, the
statement
barray

BYTE

5 DUP (1)

allocates the integer 1 five times for a total of 5 bytes.
The following examples show various ways to allocate data elements with the
DUP operator:
array

DWORD

10 DUP (1)

buffer

BYTE

256 DUP (?)

masks

BYTE

20 DUP (040h, 020h, 04h, 02h) ; 80-byte buffer
;
with bit masks
5 DUP (5 DUP (5 DUP (0)))
; 125 doublewords
;
initialized to 0

three_d DWORD

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 106 of 2 Printed: 10/02/00 04:23 PM

; 10 doublewords
;
initialized to 1
; 256-byte buffer

Chapter 5 Defining and Using Complex Data Types

107

Referencing Arrays
Each element in an array is referenced with an index number, beginning with
zero. The array index appears in brackets after the array name, as in
array[9]

Assembly-language indexes differ from indexes in high-level languages, where
the index number always corresponds to the element’s position. In C, for
example, array[9] references the array’s tenth element, regardless of whether
each element is 1 byte or 8 bytes in size.
In assembly language, an element’s index refers to the number of bytes between
the element and the start of the array. This distinction can be ignored for arrays
of byte-sized elements, since an element’s position number matches its index.
For example, defining the array
prime

BYTE 1, 3, 5, 7, 11, 13, 17

gives a value of 1 to prime[0], a value of 3 to prime[1], and so forth.
However, in arrays with elements larger than 1 byte, index numbers (except
zero) do not correspond to an element’s position. You must multiply an
element’s position by its size to determine the element’s index. Thus, for the
array
wprime

WORD 1, 3, 5, 7, 11, 13, 17

wprime[4] represents the third element (5), which is 4 bytes from the
beginning of the array. Similarly, the expression wprime[6] represents the
fourth element (7) and wprime[10] represents the sixth element (13).

The following example determines an index at run time. It multiplies the position
by two (the size of a word element) by shifting it left:
mov
shl
mov

si, cx
si, 1
ax, wprime[si]

; CX holds position number
; Scale for word referencing
; Move element into AX

The offset required to access an array element can be calculated with the
following formula:
nth element of array = array[(n-1) * size of element]
Referencing an array element by distance rather than position is not difficult to
master, and is actually very consistent with how assembly language works.
Recall that a variable name is a symbol that represents the contents of a
particular address in memory. Thus, if the array wprime begins at address
DS:2400h, the reference wprime[6] means to the processor “the word value
contained in the DS segment at offset 2400h-plus-6-bytes.”

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 107 of 3 Printed: 10/02/00 04:23 PM

108

Programmer’s Guide

As described in “Direct Memory Operands,” Chapter 3, you can substitute the
plus operator (+) for brackets, as in:
wprime[9]
wprime+9

Since brackets simply add a number to an address, you don’t need them when
referencing the first element. Thus, wprime and wprime[0] both refer to the
first element of the array wprime.
If your program runs only on an 80186 processor or higher, you can use the
BOUND instruction to verify that an index value is within the bounds of an
array. For a description of BOUND, see the Reference.

LENGTHOF, SIZEOF, and TYPE for Arrays
When applied to arrays, the LENGTHOF, SIZEOF, and TYPE operators
return information about the length and size of the array and about the type of
the
initializers.
The LENGTHOF operator returns the number of elements in the array. The
SIZEOF operator returns the number of bytes used by the initializers in the
array definition. TYPE returns the size of the elements of the array. The
following examples illustrate these operators:
array

WORD

40 DUP (5)

larray
sarray
tarray

EQU
EQU
EQU

LENGTHOF array
SIZEOF
array
TYPE
array

num

DWORD

4, 5, 6, 7, 8, 9, 10, 11

lnum
snum
tnum

EQU
EQU
EQU

LENGTHOF num
SIZEOF
num
TYPE
num

warray

WORD

40 DUP (40 DUP (5))

len
siz
typ

EQU
EQU
EQU

LENGTHOF warray
SIZEOF
warray
TYPE
warray

; 40 elements
; 80 bytes
; 2 bytes per element

; 8 elements
; 32 bytes
; 4 bytes per element

; 1600 elements
; 3200 bytes
;
2 bytes per element

Declaring and Initializing Strings
A string is an array of characters. Initializing a string like "Hello, there"
allocates and initializes 1 byte for each character in the string. An initialized
string can be no longer than 255 characters.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 108 of 4 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

109

For data directives other than BYTE, a string may initialize only the first
element. The initializer value must fit into the specified size and conform to the
expression word size in effect (see “Integer Constants and Constant
Expressions” in Chapter 1), as shown in these examples:
wstr
dstr

WORD
DWORD

"OK"
"DATA"

; Legal under EXPR32 only

As with arrays, string initializers can span multiple lines. The line must end with
a comma if you want the string to continue to the next line.
str1

BYTE

"This is a long string that does not ",
"fit on one line."

You can also have an array of pointers to strings.
PBYTE
msg1
msg2
msg3
pmsg

TYPEDEF
.DATA
BYTE
BYTE
BYTE
PBYTE
PBBYTE
PBYTE

PTR BYTE
"Operation completed successfully."
"Unknown command"
"File not found"
msg1
; pmsg is an array
msg2
;
of pointers to
msg3
;
above messages

Strings must be enclosed in single (') or double (") quotation marks. To put a
single quotation mark inside a string enclosed by single quotation marks, use two
single quotation marks. Likewise, if you need quotation marks inside a string
enclosed by double quotation marks, use two sets. These examples show the
various uses of quotation marks:
char
message
warn
string

BYTE
BYTE
BYTE
BYTE

'a'
"That's the message."
; That's the message.
'Can''t find file.'
; Can't find file.
"This ""value"" not found." ; This "value" not found.

You can always use single quotation marks inside a string enclosed by double
quotation marks, as the initialization for message shows, and vice versa.

The ? Initializer
You do not have to initialize an array. The ? operator lets you allocate space for
the array without placing specific values in it. Object files contain records for
initialized data. Unspecified space left in the object file means that no records
contain initialized data for that address. The actual values stored in arrays
allocated with ? depend on certain conditions. The ? initializer is treated as a
zero in a DUP statement that contains initializers in addition to the ? initializer.
If the ? initializer does not appear in a DUP statement, or if the DUP statement
contains only ? initializers, the assembler leaves the allocated space unspecified.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 109 of 5 Printed: 10/02/00 04:23 PM

110

Programmer’s Guide

LENGTHOF, SIZEOF, and TYPE for Strings
Because strings are simply arrays of byte elements, the LENGTHOF, SIZEOF,
and TYPE operators behave as you would expect, as illustrated in this example:
msg

BYTE

"This string extends ",
"over three ",
"lines."

lmsg
smsg
tmsg

EQU
EQU
EQU

LENGTHOF msg
SIZEOF
msg
TYPE
msg

; 37 elements
; 37 bytes
; 1 byte per element

Processing Strings
The 8086-family instruction set has seven string instructions for fast and
efficient processing of entire strings and arrays. The term “string” in “string
instructions” refers to a sequence of elements, not just character strings. These
instructions work directly only on arrays of bytes and words on the 8086–80486
processors, and on arrays of bytes, words, and doublewords on the 80386/486
processors. Processing larger elements must be done indirectly with loops.
The following list gives capsule descriptions of the five instructions discussed in
this section.
Instruction

Description

MOVS

Copies a string from one location to another
Stores contents of the accumulator register to a string
Compares one string with another
Loads values from a string to the accumulator register
Scans a string for a specified value

STOS
CMPS
LODS
SCAS

All of these instructions use registers in a similar way and have a similar syntax.
Most are used with the repeat instruction prefixes REP, REPE (or REPZ), and
REPNE (or REPNZ). REPZ is a synonym for REPE (Repeat While Equal) and
REPNZ is a synonym for REPNE (Repeat While Not Equal).
This section first explains the general procedures for using all string instructions.
It then illustrates each instruction with an example.

Overview of String Instructions
The string instructions have specific requirements for the location of strings and
the use of registers. To operate on any string, follow these three steps:
1. Set the direction flag to indicate the direction in which you want to process
the string. The STD instruction sets the flag, while CLD clears it.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 110 of 6 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

111

If the direction flag is clear, the string is processed upward (from low
addresses to high addresses, which is from left to right through the string). If
the direction flag is set, the string is processed downward (from high
addresses to low addresses, or from right to left). Under MS-DOS, the
direction flag is normally clear if your program has not changed it.
2. Load the number of iterations for the string instruction into the CX register.
If you want to process 100 elements in a string, move 100 into CX. If you
wish the string instruction to terminate conditionally (for example, during a
search when a match is found), load the maximum number of iterations that
can be performed without an error.
3. Load the starting offset address of the source string into DS:SI and the
starting address of the destination string into ES:DI. Some string instructions
take only a destination or source, not both (see Table 5.1).
Normally, the segment address of the source string should be DS, but you
can use a segment override to specify a different segment for the source
operand. You cannot override the segment address for the destination string.
Therefore, you may need to change the value of ES. For information on
changing segment registers, see “Programming Segmented Addresses” in
Chapter 3.
Note Although you can use a segment override on the source operand, a
segment override combined with a repeat prefix can cause problems in certain
situations on all processors except the 80386/486. If an interrupt occurs during
the string operation, the segment override is lost and the rest of the string
operation processes incorrectly. Segment overrides can be used safely when
interrupts are turned off or with the 80386/486 processors.
You can adapt these steps to the requirements of any particular string operation.
The syntax for the string instructions is:
[[prefix]] CMPS [[segmentregister:]] source, [[ES:]] destination
LODS [[segmentregister:]] source
[[prefix]] MOVS [[ES:]] destination, [[segmentregister:]] source
[[prefix]] SCAS [[ES:]] destination
[[prefix]] STOS [[ES:]] destination
Some instructions have special forms for byte, word, or doubleword operands.
If you use the form of the instruction that ends in B (BYTE), W (WORD), or D
(DWORD) with LODS, SCAS, and STOS, the assembler knows whether the
element is in the AL, AX, or EAX register. Therefore, these instruction forms
do not require operands.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 111 of 7 Printed: 10/02/00 04:23 PM

112

Programmer’s Guide

Table 5.1 lists each string instruction with the type of repeat prefix it uses and
indicates whether the instruction works on a source, a destination, or both.
Table 5.1 Requirements for String Instructions
Instruction

Repeat Prefix

Source/Destination

MOVS

REP

Both

DS:SI, ES:DI

SCAS

REPE/REPNE

Destination

ES:DI

CMPS

REPE/REPNE

Both

DS:SI, ES:DI

LODS

None

Source

DS:SI

STOS

REP

Destination

ES:DI

INS

REP

Destination

ES:DI

OUTS

REP

Source

DS:SI

The repeat prefix causes the instruction that follows it to repeat for the number
of times specified in the count register or until a condition becomes true. After
each iteration, the instruction increments or decrements SI and DI so that it
points to the next array element. The direction flag determines whether SI and
DI are incremented (flag clear) or decremented (flag set). The size of the
instruction determines whether SI and DI are altered by 1, 2, or 4 bytes each
time.
Each prefix governs the number of repetitions as follows:
Prefix

Description

REP

Repeats instruction CX times

REPE, REPZ

Repeats instruction maximum CX times while values are equal

REPNE, REPNZ

Repeats instruction maximum CX times while values are not
equal

The prefixes apply to only one string instruction at a time. To repeat a block of
instructions, use a loop construction. (See “Loops” in Chapter 7.)
At run time, if a string instruction is preceded by a repeat sequence, the
processor:
1. Checks the CX register and exits if CX is 0.
2. Performs the string operation once.
3. Increases SI and/or DI if the direction flag is clear. Decreases SI and/or DI if
the direction flag is set. The amount of increase or decrease is 1 for byte
operations, 2 for word operations, and 4 for doubleword operations.
4. Decrements CX without modifying the flags.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 112 of 8 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

113

5. Checks the zero flag (for SCAS or CMPS) if the REPE or REPNE prefix is
used. If the repeat condition holds, loops back to step 1. Otherwise, the loop
ends and execution proceeds to the next instruction.
When the repeat loop ends, SI (or DI) points to the position following a match
(when using SCAS or CMPS), so you need to decrement or increment DI or SI
to point to the element where the last match occurred.
Although string instructions (except LODS) are used most often with repeat
prefixes, they can also be used by themselves. In these cases, the SI and/or DI
registers are adjusted as specified by the direction flag and the size of operands.

Using String Instructions
To use the 8086-family string instructions, follow the steps outlined in the
previous section. Examples in this section illustrate each instruction.
You can also use the techniques in this section with structures and unions, since
arrays and strings can be fields in structures and unions. (See the section
“Structures and Unions,” following.)

Moving Array Data
The MOVS instruction copies data from one area of memory to another. To
move data, first load the count, source and destination addresses into the
appropriate registers. Then use REP with the MOVS instruction.

source
destin

.MODEL
.DATA
BYTE
BYTE
.CODE
mov
mov
mov
.
.
.
cld
mov
mov
mov
rep

small
10 DUP ('0123456789')
100 DUP (?)
ax, @data
ds, ax
es, ax

; Load same segment
;
to both DS
;
and ES

cx, LENGTHOF source
si, OFFSET source
di, OFFSET destin
movsb

;
;
;
;
;

Work upward
Set iteration count to 100
Load address of source
Load address of destination
Move 100 bytes

Filling Arrays
The STOS instruction stores a specified value in each position of a string. The
string is the destination, so it must be pointed to by ES:DI. The value to store
must be in the accumulator.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 113 of 9 Printed: 10/02/00 04:23 PM

114

Programmer’s Guide

The next example stores the character 'a' in each byte of a 100-byte string,
filling the entire string with “aaaa....” Notice how the code stores 50 words
rather than

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 114 of 10 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

115

100 bytes. This makes the fill operation faster by reducing the number of
iterations. To fill an odd number of bytes, you need to adjust for the last byte.
.MODEL
.DATA
destin BYTE
ldestin EQU
.CODE
.
.
.
cld
mov
mov
mov
rep

small, C
100 DUP (?)
(LENGTHOF destin) / 2
; Assume ES = DS

ax, 'aa'
cx, ldestin
di, OFFSET destin
stosw

;
;
;
;
;

Work upward
Load character to fill
Load length of string
Load address of destination
Store 'aa' into array

Comparing Arrays
The CMPS instruction compares two strings and points to the address after
which a match or nonmatch occurs. If the values are the same, the zero flag is
set. Either string can be considered the destination or the source unless a
segment override is used. This example using CMPSB assumes that the strings
are in different segments. Both segments must be initialized to the appropriate
segment register.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 115 of 11 Printed: 10/02/00 04:23 PM

116

Programmer’s Guide
.MODEL large, C
.DATA
string1 BYTE
"The quick brown fox jumps over the lazy dog"
.FARDATA
string2 BYTE
"The quick brown dog jumps over the lazy fox"
lstring EQU
LENGTHOF string2
.CODE
mov
ax, @data
; Load data segment
mov
ds, ax
;
into DS
mov
ax, @fardata
; Load far data segment
mov
es, ax
;
into ES
.
.
.
cld
; Work upward
mov
cx, lstring
; Load length of string
mov
si, OFFSET string1 ; Load offset of string1
mov
di, OFFSET string2 ; Load offset of string2
repe
cmpsb
; Compare
je
allmatch
; Jump if all match
.
.
.
allmatch:
; Special case for all match

Loading Data from Arrays
The LODS instruction loads a value from a string into the accumulator register.
This instruction is not used with a repeat instruction prefix, since continually
reloading the accumulator serves no purpose.
The code in this example loads, processes, and displays each byte in a string.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 116 of 12 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

info
linfo

.DATA
BYTE
WORD
.CODE
.
.
.
cld
mov
mov
mov

117

0, 1, 2, 3, 4, 5, 6, 7, 8, 9
LENGTHOF info

cx, linfo
si, OFFSET info
ah, 2

;
;
;
;

Work upward
Load length
Load offset of source
Display character function

al, '0'
dl, al
21h
get

;
;
;
;
;

Get a character
Convert to ASCII
Move to DL
Call DOS to display character
Repeat

get:
lodsb
add
mov
int
loop

Searching Arrays
The SCAS instruction compares the value pointed to by ES:DI with the value in
the accumulator. If both values are the same, it sets the zero flag.
A repeat prefix lets SCAS work on an entire string, scanning (from which SCAS
gets its name) for a particular value called the target. REPNE SCAS sets the
zero flag if it finds the target value in the array. REPE SCAS sets the zero flag
if the scanned array contains nothing but the target value.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 117 of 13 Printed: 10/02/00 04:23 PM

118

Programmer’s Guide

This example assumes that ES is not the same as DS and that the address of the
string is stored in a pointer variable. The LES instruction loads the far address
of the string into ES:DI.
.DATA
string BYTE
pstring PBYTE
lstring EQU
.CODE
.
.
.
cld
mov
les
mov
repne
jne
.
.
.
notfound:

"The quick brown fox jumps over the lazy dog"
string
; Far pointer to string
LENGTHOF string
; Length of string

cx, lstring
di, pstring
al, 'z'
scasb
notfound

;
;
;
;
;
;
;
;

Work upward
Load length of string
Load address of string
Load character to find
Search
Jump if not found
ES:DI points to character
after first 'z'

; Special case for not found

Translating Data in Byte Arrays
The XLAT (Translate) instruction copies a byte from an array of bytes into the
AL register. The instruction takes its name from its ability to translate an
element’s number into the element itself. For example, given the number 7,
XLAT returns byte #7 from the array. The array may hold byte-sized integers
or, very often, a table or list of characters. The syntax for XLAT is:
XLAT[[B]] [[[[segment:]]memory]]
The optional B suffix (for “byte”) reflects the size of data the instruction
handles. Both XLAT and XLATB assemble to exactly the same machine code.
To use XLAT, place the offset of the start of the array in the BX register and
the desired index value in AL. Array indexes always begin with 0 in assembly
language. To retrieve the first byte of the array, set AL to 0; to retrieve the
second byte, set AL to 1, and so forth. XLAT returns the byte element in AL,
overwriting the index number.
By default, the DS register contains the segment of the table, but you can use a
segment override to specify a different segment. You need not give an operand
except when specifying a segment override. (For information about the segment
override operator, see “Direct Memory Operands” in Chapter 3.)

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 118 of 14 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

119

This example illustrates XLAT by looking up hexadecimal characters in a list.
The code converts an eight-bit binary number to a string representing a
hexadecimal number.
; Table
hex
convert
key

of hexadecimal digits
BYTE
"0123456789ABCDEF"
BYTE
"You pressed the key with ASCII code "
BYTE
?,?,"h",13,10,"$"
.CODE
.
.
.
mov
ah, 8
; Get a key in AL
int
21h
; Call DOS
mov
bx, OFFSET hex
; Load table address
mov
ah, al
; Save a copy in high byte
and
al, 00001111y
; Mask out top character
xlat
; Translate
mov
key[1], al
; Store the character
mov
cl, 12
; Load shift count
shr
ax, cl
; Shift high char into position
xlat
; Translate
mov
key, al
; Store the character
mov
dx, OFFSET convert ; Load message
mov
ah, 9
; Display character
int
21h
; Call DOS

Although AL cannot contain an index value greater than 255, you can use
XLAT with arrays containing more than 256 elements. Simply treat each 256byte block of the array as a smaller sub-array. For example, to retrieve the
260th element of an array, add 256 to BX and set AL=3 (260-256-1).

Structures and Unions
A structure is a group of possibly dissimilar data types and variables that can be
accessed as a unit or by any of its components. The fields within the structure
can have different sizes and data types.
Unions are identical to structures, except that the fields of a union overlap in
memory, which allows you to define different data formats for the same
memory space. Unions can store different types of data depending on the
situation. They also can store data as one data type and retrieve it as another
data type.
Whereas each field in a structure has an offset relative to the first byte of the
structure, all the fields in a union start at the same offset. The size of a structure
is the sum of its components; the size of a union is the length of the longest
field.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 119 of 15 Printed: 10/02/00 04:23 PM

120

Programmer’s Guide

A MASM structure is similar to a struct in the C language, a STRUCTURE in
FORTRAN, and a RECORD in Pascal. Unions in MASM are similar to unions
in C and FORTRAN, and to variant records in Pascal.
Follow these steps when using structures and unions:
1. Declare a structure (or union) type.
2. Define one or more variables having that type.
3. Reference the fields directly or indirectly with the field (dot) operator.
You can use the entire structure or union variable or just the individual fields as
operands in assembler statements. This section explains the allocating,
initializing, and nesting of structures and unions.
MASM 6.1 extends the functionality of structures and also makes some changes
to MASM 5.1 behavior. If you prefer, you can retain MASM 5.1 behavior by
specifying OPTION OLDSTRUCTS in your program.

Declaring Structure and Union Types
When you declare a structure or union type, you create a template for data. The
template states the sizes and, optionally, the initial values in the structure or
union, but allocates no memory.
The STRUCT keyword marks the beginning of a type declaration for a
structure. (STRUCT and STRUC are synonyms.) The format for STRUCT
and UNION type declarations is:
name {STRUCT | UNION} [[alignment]] [[,NONUNIQUE ]]
fielddeclarations
name ENDS
The fielddeclarations is a series of one or more variable declarations. You can
declare default initial values individually or with the DUP operator. (See
“Defining Structure and Union Variables,” following.) “Referencing Structures,
Unions, and Fields,” later in this chapter, explains the NONUNIQUE keyword.
You can nest structures and unions, as explained in “Nested Structures and
Unions,” also later in this chapter.

Initializing Fields
If you provide initializers for the fields of a structure or union when you declare
the type, these initializers become the default value for the fields when you
define a variable of that type. “Defining Structure and Union Variables,”
following, explains default initializers.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 120 of 16 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

121

When you initialize the fields of a union type, the type and value of the first field
become the default value and type for the union. In this example of an initialized
union declaration, the default type for the union is DWORD:
DWB
d
w
b
DWB

UNION
DWORD
WORD
BYTE
ENDS

00FFh
?
?

If the size of the first member is less than the size of the union, the assembler
initializes the rest of the union to zeros. When initializing strings in a type, make
sure the initial values are long enough to accommodate the largest possible
string.

Field Names
Structure and union field names must be unique within a nesting level because
they represent the offset from the beginning of the structure to the
corresponding field.
A label elsewhere in the code may have the same name as a structure field, but
a text macro cannot. Also, field names between structures need not be unique.
Field names must be unique if you place OPTION M510 or OPTION
OLDSTRUCTS in your code or use the /Zm option from the command line,
since versions of MASM prior to 6.0 require unique field names. (See Appendix
A.)

Alignment Value and Offsets for Structures
Data access to structures is faster on aligned fields than on unaligned fields.
Therefore, alignment gains speed at the cost of space. Alignment improves
access on 16-bit and 32-bit processors but makes no difference in programs
executing on an 8-bit 8088 processor.
The way the assembler aligns structure fields determines the amount of space
required to store a variable of that type. Each field in a structure has an offset
relative to 0. If you specify an alignment in the structure declaration (or with the
/Zpn command-line option), the offset for each field may be modified by the
alignment (or n).
The only values accepted for alignment are 1, 2, and 4. The default is 1. If the
type declaration includes an alignment, each field is aligned to either the field’s
size or the alignment value, whichever is less. If the field size in bytes is greater
than the alignment value, the field is padded so that its offset is evenly divisible
by the alignment value. Otherwise, the field is padded so that its offset is evenly
divisible by the field size.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 121 of 17 Printed: 10/02/00 04:23 PM

122

Programmer’s Guide

Any padding required to reach the correct offset for the field is added prior to
allocating the field. The padding consists of zeros and always precedes the
aligned field. The size of the structure must also be evenly divisible by the
structure alignment value, so zeros may be added at the end of the structure.
If neither the alignment nor the /Zp command-line option is used, the offset is
incremented by the size of each data directive. This is the same as a default
alignment equal to 1. The alignment specified in the type declaration overrides
the /Zp command-line option.
These examples show how the assembler determines offsets:
STUDENT2
score
id
year
sname
STUDENT2

STRUCT
WORD
BYTE
DWORD
BYTE
ENDS

2
1
2
3
4

;
;
;
;
;

Alignment value is 2
Offset = 0
Offset = 2 (1 byte padding added)
Offset = 4
Offset = 8 (1 byte padding added)

One byte of padding is added at the end of the first byte-sized field. Otherwise,
the offset of the year field would be 3, which is not divisible by the alignment
value of 2. The size of this structure is now 9 bytes. Since 9 is not evenly
divisible by 2, 1 byte of padding is added at the end of student2.
STUDENT4
sname
score
year

STRUCT
BYTE
WORD
BYTE

id
STUDENT4

DWORD
ENDS

4
;
1
;
10 DUP (100) ;
2
;
;
;
3
;

Alignment value is 4
Offset = 0 (1 byte padding added)
Offset = 2
Offset = 22 (1 byte padding
added so offset of next field
is divisible by 4)
Offset = 24

The alignment value affects the alignment of structure variables, so adding an
alignment value affects memory usage. This feature provides compatibility with
structures in Microsoft C. MASM 6.1 provides an improved H2INC utility,
which C programmers can use to translate C structures to assembly. (See
Environment and Tools, Chapter 20.)
The ALIGN, EVEN, and ORG directives can modify how field offsets are
placed during structure definition. The EVEN and ALIGN directives insert
padding bytes to round the field offset up to the specified alignment boundary.
The ORG directive changes the offset of the next field to a given value, either
positive or negative. If you use ORG when declaring a structure, you cannot
define a structure of that type. ORG is useful when accessing existing data
structures, such as a stack frame created by a high-level language.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 122 of 18 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

123

Defining Structure and Union Variables
Once you have declared a structure or union type, you can define variables of
that type. For each variable defined, memory is allocated in the current segment
in the format declared by the type. The syntax for defining a structure or union
variable is:
[[name]] typename < [[initializer [[,initializer]]...]] >
[[name]] typename { [[initializer [[,initializer]]...]] }
[[name]] typename constant DUP ({ [[initializer [[,initializer]]...]] })
The name is the label assigned to the variable. If you do not provide a name, the
assembler allocates space for the variable but does not give it a symbolic name.
The typename is the name of a previously declared structure or union type.
You can give an initializer for each field. Each initializer must correspond in
type with the field defined in the type declaration. For unions, the type of the
initializer must be the same as the type for the first field. An initialization list can
also use the DUP operator.
The list of initializers can be broken only after a comma unless you end the line
with a continuation character (\). The last curly brace or angle bracket must
appear on the same line as the last initializer. You can also use the line
continuation character to extend a line as shown in the Item4 declaration that
follows. Angle brackets and curly braces can be intermixed in an initialization as
long as they match. This example illustrates the options for initializing lists in
structures of type ITEMS:

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 123 of 19 Printed: 10/02/00 04:23 PM

124

Programmer’s Guide
ITEMS
STRUCT
Iname
BYTE
'Item Name'
Inum
WORD
?
UNION
ITYPE
;
oldtype BYTE
0
;
newtype WORD
?
;
ENDS
;
ITEMS
ENDS
.
.
.
.DATA
Item1
ITEMS
< >
;
Item2
ITEMS
{ }
;
Item3
ITEMS
<'Bolts', 126>
;
;
;
Item4
ITEMS
{ \
'Bolts',
;
126 \
;
}

UNION keyword appears first
when nested in structure.
(See "Nested Structures
and Unions," following ).

Accepts default initializers
Accepts default initializers
Overrides default value of first
2 fields; use default of
the third field
Item name
Part number

The example defines — that is, allocates space for — four structures of the
ITEMS type. The structures are named Item1 through Item4. Each definition
requires the angle brackets or curly braces even when not initialized. If you
initialize more than one field, separate the values with commas, as shown in
Item3 and Item4.
You need not initialize all fields in a structure. If a field is blank, the assembler
uses the structure’s initial value given for that field in the declaration. If there is
no default value, the field value is left unspecified.
For nested structures or unions, however, these are equivalent:
Item5
Item6

ITEMS
ITEMS

{'Bolts', ,
}
{'Bolts', , { } }

A variable and an array of union type WB look like this:

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 124 of 20 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types
WB

?
?

UNION
WORD
BYTE
ENDS

num
array

WB
WB

{0Fh}
(40 / SIZEOF WB) DUP ({2})

w
b

125

; Store 0Fh
; Allocates and
;
initializes 20 unions

Arrays as Field Initializers
The size of the initializer determines the length of the array that can override the
contents of a field in a variable definition. The override cannot contain more
elements than the default. Specifying fewer override array elements changes the
first n values of the default where n is the number of values in the override. The
rest of the array elements take their default values from the initializer.

Strings as Field Initializers
If the override is shorter, the assembler pads the override with spaces to equal
the length of the initializer. If the initializer is a string and the override value is
not a string, the override value must be enclosed in angle brackets or curly
braces.
A string can override any member of type BYTE (or SBYTE). You need not
enclose the string in angle brackets or curly braces unless mixed with other
override methods.
If a structure has an initialized string field or an array of bytes, any new string
assigned to a variable of the field that is smaller than the default is padded with
spaces. The assembler adds four spaces at the end of 'Bolts' in the variables
of type ITEMS previously shown. The Iname field in the ITEMS structure
cannot contain a field initializer longer than 'Item Name'.

Structures as Field Initializers
Initializers for structure variables must be enclosed in curly braces or angle
brackets, but you can specify overrides with fewer elements than the defaults.
This example illustrates the use of default values with structures as field
initializers:

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 125 of 21 Printed: 10/02/00 04:23 PM

126

Programmer’s Guide
DISKDRIVES
a1
b1
c1
DISKDRIVES

STRUCT
BYTE ?
BYTE ?
BYTE ?
ENDS

INFO
buffer
crlf
query
endmark
drives
INFO

STRUCT
BYTE
100 DUP (?)
BYTE
13, 10
BYTE
'Filename: ' ; String <= can override
BYTE
36
DISKDRIVES <0, 1, 1>
ENDS

info1

{ , , 'Dir' }

INFO

; Next line illegal since name in query field is too long:
; info2 INFO
{"TESTFILE", , "DirectoryName"}
lotsof

INFO

{ , , 'file1', , {0,0,0} },
{ , , 'file2', , {0,0,1} },
{ , , 'file3', , {0,0,2} }

The following diagram shows how the assembler stores info1.

The initialization for drives gives default values for all three fields of the
structure. The fields left blank in info1 use the default values for those fields.
The info2 declaration is illegal because “DirectoryName” is longer than the
initial string for that field.

Arrays of Structures and Unions
You can define an array of structures using the DUP operator (see “Declaring
and Referencing Arrays,” earlier in this chapter) or by creating a list of
structures. For example, you can define an array of structure variables like this:

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 126 of 22 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types
Item7

ITEMS

127

30 DUP ({,,{10}})

The Item7 array defined here has 30 elements of type ITEMS, with the third
field of each element (the union) initialized to 10.
You can also list array elements as shown in the following example.
Item8

ITEMS

{'Bolts', 126, 10},
{'Pliers',139, 10},
{'Saws', 414, 10}

Redeclaring a Structure
The assembler generates an error when you declare a structure more than once
unless the following are the same:
u
u
u
u

Field names
Offsets of named fields
Initialization lists
Field alignment value

LENGTHOF, SIZEOF, and TYPE for Structures
The size of a structure determined by SIZEOF is the offset of the last field, plus
the size of the last field, plus any padding required for proper alignment. (For
information about alignment, see “Declaring Structure and Union Types,” earlier
in this chapter.)

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 127 of 23 Printed: 10/02/00 04:23 PM

128

Programmer’s Guide

This example, using the preceding data declarations, shows how to use the
LENGTHOF, SIZEOF, and TYPE operators with structures.
INFO
buffer
crlf
query
endmark
drives
INFO

STRUCT
BYTE
100 DUP (?)
BYTE
13, 10
BYTE
'Filename: '
BYTE
36
DISKDRIVES <0, 1, 1>
ENDS

info1
lotsof

INFO
INFO

{
{
{
{

sinfo1

EQU

SIZEOF

info1

linfo1
tinfo1

EQU
EQU

LENGTHOF
TYPE

info1
info1

slotsof EQU

SIZEOF

llotsof EQU
tlotsof EQU

LENGTHOF
TYPE

lotsof ; 116 * 3 = number of bytes in
;
initializers
lotsof ; 3 = number of items
lotsof ; 116 = same as size for structure
;
of type INFO

,
,
,
,

'Dir' }
'file1', , {0,0,0} },
'file2', , {0,0,1} },
'file3', , {0,0,2} }
; 116 = number of bytes in
;
initializers
; 1 = number of items
; 116 = same as size

LENGTHOF, SIZEOF, and TYPE for Unions
The size of a union determined by SIZEOF is the size of the longest field plus
any padding required. The length of a union variable determined by
LENGTHOF equals the number of initializers defined inside angle brackets or
curly braces. TYPE returns a value indicating the type of the longest field.
DWB
d
w
b
DWB

UNION
DWORD
WORD
BYTE
ENDS

?
?
?

num
array

DWB
DWB

{0FFFFh}
(100 / SIZEOF DWB) DUP ({0})

snum
lnum
tnum
sarray
larray
tarray

EQU
EQU
EQU
EQU
EQU
EQU

SIZEOF
LENGTHOF
TYPE
SIZEOF
LENGTHOF
TYPE

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 128 of 24 Printed: 10/02/00 04:23 PM

num
num
num
array
array
array

;
;
;
;
;
;

=
=
=
=
=
=

4
1
4
100 (4*25)
25
4

Chapter 5 Defining and Using Complex Data Types

129

Referencing Structures, Unions, and Fields
Like other variables, structure variables can be accessed by name. You can
access fields within structure variables with this syntax:
variable. field
References to fields must always be fully qualified, with the structure or union
names and the dot operator preceding the field name. The assembler requires
that you use the dot operator only with structure fields, not as an alternative to
the plus operator; nor can you use the plus operator as an alternative to the dot
operator.
The following example shows several ways to reference the fields of a structure
of type DATE.
DATE
month
day
year
DATE

STRUCT
BYTE
BYTE
WORD
ENDS

yesterday
.
.
.
mov
mov
mov
mov

; Defines structure type
?
?
?

DATE

al,
bx,
al,
al,

{1, 20, 1993}

yesterday.day
OFFSET yesterday
(DATE PTR [bx]).month
[bx].date.month

; Declare structure
;
variable

;
;
;
;
;
;
;

Use structure variables
Load structure address
Use as indirect operand
This is necessary only if
month is already a
field in a different
structure

Under OPTION M510 or OPTION OLDSTRUCTS, unique structure names
do not need to be qualified. However, if the NONUNIQUE keyword appears in
a structure definition, all fields of the structure must be fully qualified when
referenced, even if the OPTION OLDSTRUCTS directive appears in the code.
Also, you must qualify all references to a field. (For information on the
OPTION directive, see Chapter 1.)
Even if the initialized union is the size of a WORD or DWORD, members of
structures or unions are accessible only through the field’s names.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 129 of 25 Printed: 10/02/00 04:23 PM

130

Programmer’s Guide

In the following example, the two MOV statements show how you can access
the elements of an array of unions.
WB

?
?

UNION
WORD
BYTE
ENDS

array

(100 / SIZEOF WB) DUP ({0})

mov
mov

array[12].w, 40h
array[32].b, 2

w
b

As the preceding code illustrates, you can use unions to access the same data in
more than one form. One application of structures and unions is to simplify the
task of reinitializing a far pointer. For a far pointer declared as
FPWORD

TYPEDEF FAR PTR WORD

.DATA
WordPtr FPWORD ?

you must follow these steps to point WordPtr to a word value named
ThisWord in the current data segment.
mov
mov

WORD PTR WordPtr[2], ds
WORD PTR WordPtr, OFFSET ThisWord

The preceding method requires that you remember whether the segment or the
offset is stored first. However, if your program declares a union like this:
uptr
dwptr
STRUCT
offs
segm
ENDS
uptr

UNION
FPWORD

WORD
WORD

0
0

ENDS

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 130 of 26 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

131

You can initialize a far pointer with these steps:
.DATA
WrdPtr2 uptr
.
.
.
mov
mov

WrdPtr2.segm, ds
WrdPtr2.offs, OFFSET ThisWord

This code moves the segment and the offset into the pointer and then moves the
pointer into a register with the other field of the union. Although this technique
does not reduce the code size, it avoids confusion about the order for loading
the segment and offset.

Nested Structures and Unions
You can nest structures and unions in several ways. This section explains how
to refer to the fields in a nested structure or union. The following example
illustrates the four techniques for nesting, and how to reference the fields. Note
the syntax for nested structures. The techniques are reviewed following the
example.
ITEMS
Inum
Iname
ITEMS

STRUCT
WORD
BYTE
ENDS

INVENTORY
UpDate
oldItem

STRUCT
WORD
ITEMS

STRUCT ups
source
shipmode
ENDS
STRUCT
f1
f2
ENDS
INVENTORY

?
'Item Name'

ITEMS

?
{ \
100,
'AF8' \
}
{ ?, '94C' }

WORD
BYTE

?
?

WORD
WORD

?
?

; Unnamed nested structure

ENDS

.DATA
yearly

; Named variable of
;
existing structure
; Unnamed variable of
;
existing type
; Named nested structure

INVENTORY

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 131 of 27 Printed: 10/02/00 04:23 PM

{ }

132

Programmer’s Guide
; Referencing each type of data in the yearly structure:
mov
mov
mov
mov

ax, yearly.oldItem.Inum
yearly.ups.shipmode, 'A'
yearly.Inum, 'C'
ax, yearly.f1

To nest structures and unions, you can use any of these techniques:
u

The field of a structure or union can be a named variable of an existing
structure or union type, as in the oldItem field. Because INVENTORY
contains two structures of type ITEMS , the field names in oldItem are not
unique. Therefore, you must use the full field names when referencing those
fields, as in the statement
mov

To declare a named structure or union inside another structure or union, give
the STRUCT or UNION keyword first and then define a label for it. Fields
of the nested structure or union must always be qualified:
mov

ax, yearly.oldItem.Inum

yearly.ups.shipmode, 'A'

As shown in the Items field of Inventory, you also can use unnamed
variables of existing structures or unions inside another structure or union. In
these cases, you can reference fields directly:
mov
mov

yearly.Inum, 'C'
ax, yearly.f1

Records
Records are similar to structures, except that fields in records are bit strings.
Each bit field in a record variable can be used separately in constant operands or
expressions. The processor cannot access bits individually at run time, but it can
access bit fields with instructions that manipulate bits.
Records are bytes, words, or doublewords in which the individual bits or groups
of bits are considered fields. In general, the three steps for using record variables
are the same as those for using other complex data types:
1. Declare a record type.
2. Define one or more variables having the record type.
3. Reference record variables using shifts and masks.
Once it is defined, you can use the record variable as an operand in assembler
statements.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 132 of 28 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

133

This section explains the record declaration syntax and the use of the MASK
and WIDTH operators. It also shows some applications of record variables and
constants.

Declaring Record Types
A record type creates a template for data with the sizes and, optionally, the
initial values for bit fields in the record. It does not allocate memory space for
the
record.
The RECORD directive declares a record type for an 8-bit, 16-bit, or 32-bit
record that contains one or more bit fields. The maximum size is based on the
expression word size. See OPTION EXPR16 and OPTION EXPR32 in
Chapter 1. The syntax is:
recordname RECORD field [[, field]]...
The field declares the name, width, and initial value for the field. The syntax for
each field is:
fieldname:width[[=expression]]
Global labels, macro names, and record field names must all be unique, but
record field names can have the same names as structure field names. Width is
the number of bits in the field, and expression is a constant giving the initial (or
default) value for the field. Record definitions can span more than one line if the
continued lines end with commas.
If expression is given, it declares the initial value for the field. The assembler
generates an error message if an initial value is too large for the width of its field.
The first field in the declaration always goes into the most significant bits of the
record. Subsequent fields are placed to the right in the succeeding bits. If the
fields do not total exactly 8, 16, or 32 bits as appropriate, the entire record is
shifted right, so the last bit of the last field is the lowest bit of the record.
Unused bits in the high end of the record are initialized to 0.
The following example creates a byte record type COLOR having four fields:
blink, back, intense, and fore. The contents of the record type are shown
after the example. Since no initial values are given, all bits are set to 0. Note that
this is only a template maintained by the assembler. It allocates no space in the
data segment.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 133 of 29 Printed: 10/02/00 04:23 PM

134

Programmer’s Guide
COLOR

RECORD

blink:1, back:3, intense:1, fore:3

The next example creates a record type CW that has six fields. Each record
declared with this type occupies 16 bits of memory. Initial (default) values are
given for each field. You can use them when declaring data for the record. The
bit diagram after the example shows the contents of the record type.
CW

RECORD

r1:3=0, ic:1=0, rc:2=0, pc:2=3, r2:2=1, masks:6=63

Defining Record Variables
Once you have declared a record type, you can define record variables of that
type. For each variable, the assembler allocates memory in the format declared
by the type. The syntax is:
[[name]] recordname <[[initializer [[,initializer]]...]] >
[[name]] recordname { [[initializer [[,initializer]]...]] }
[[name]] recordname constant DUP ( [[initializer [[,initializer]]...]] )
The recordname is the name of a record type previously declared with the
RECORD directive.
A fieldlist for each field in the record can be a list of integers, character
constants, or expressions that correspond to a value compatible with the size of
the field. You must include curly braces or angle brackets even when you do not
specify an initial value.
If you use the DUP operator (see “Declaring and Referencing Arrays,” earlier in
this chapter) to initialize multiple record variables, only the angle brackets and

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 134 of 30 Printed: 10/02/00 04:23 PM

Chapter 5 Defining and Using Complex Data Types

135

any initial values need to be enclosed in parentheses. For example, you can
define an array of record variables with
xmas

COLOR

50 DUP ( <1, 2, 0, 4> )

You do not have to initialize all fields in a record. If an initial value is blank, the
assembler automatically stores the default initial value of the field. If there is no
default value, the assembler clears each bit in the field.
The definition in the following example creates a variable named warning
whose type is given by the record type COLOR. The initial values of the fields in
the variable are set to the values given in the record definition. The initial values
override any default record values given in the declaration.
COLOR

RECORD

warning COLOR

blink:1,back:3,intense:1,fore:3 ; Record
;
declaration
<1, 0, 1, 4>
; Record
;
definition

LENGTHOF, SIZEOF, and TYPE with Records
The SIZEOF and TYPE operators applied to a record name return the number
of bytes used by the record. SIZEOF returns the number of bytes a record
variable occupies. You cannot use LENGTHOF with a record declaration, but
you can use it with defined record variables. LENGTHOF returns the number
of records in an array of records, or 1 for a single record variable. The following
example illustrates these points.
; Record definition
; 9 bits stored in 2 bytes
RGBCOLOR
RECORD red:3,

;

mov
mov
mov
mov

green:3,

blue:3

ax, RGBCOLOR
; Equivalent to "mov ax, 01FFh"
ax, LENGTHOF RGBCOLOR ; Illegal since LENGTHOF can
;
apply only to data label
ax, SIZEOF
RGBCOLOR ; Equivalent to "mov ax, 2"
ax, TYPE
RGBCOLOR ; Equivalent to "mov ax, 2"

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 135 of 31 Printed: 10/02/00 04:23 PM

136

Programmer’s Guide
; Record instance
; 8 bits stored in 1 byte
RGBCOLOR2
RECORD red:3, green:3, blue:2
rgb
RGBCOLOR2 <1, 1, 1>
; Initialize to 00100101y
mov

ax, RGBCOLOR2

mov
mov
mov

ax, LENGTHOF rgb
ax, SIZEOF
rgb
ax, TYPE
rgb

; Equivalent
;
"mov
; Equivalent
; Equivalent
; Equivalent

to
ax, 00FFh"
to "mov ax, 1"
to "mov ax, 1"
to "mov ax, 1"

Record Operators
The WIDTH operator (used only with records) returns the width in bits of a
record or record field. The MASK operator returns a bit mask for the bit
positions occupied by the given record field. A bit in the mask contains a 1 if
that bit corresponds to a bit field. The following example shows how to use
MASK and WIDTH.
.DATA
COLOR
message
wblink
wback
wintens
wfore
wcolor

EQU
EQU
EQU
EQU
EQU
.CODE
.
.
.
mov
and

RECORD blink:1, back:3, intense:1, fore:3
COLOR
<1, 5, 1, 1>
WIDTH blink
; "wblink"
= 1
WIDTH back
; "wback"
= 3
WIDTH intense
; "wintens"
= 1
WIDTH fore
; "wfore"
= 3
WIDTH COLOR
; "wcolor"
= 8

ah, message
ah, NOT MASK back

ah, MASK blink

xor

ah, MASK intense

IF
mov
ELSE
mov
xor
ENDIF

(WIDTH COLOR) GT 8
ax, message
al, message
ah, ah

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 136 of 32 Printed: 10/02/00 04:23 PM

;
;
;
;
;
;
;
;
;
;

Load initial
1101 1001
Turn off
AND 1000 1111
"back"
--------1000 1001
Turn on
OR 1000 0000
"blink"
--------1000 1001
Toggle
XOR 0000 1000
"intense"
--------1000 0001

; If color is 16 bit, load
;
into 16-bit register
; else
;
load into low 8-bit register
;
and clear high 8-bits

Chapter 5 Defining and Using Complex Data Types

The example continues by illustrating several ways in which record fields can
serve as operands and expressions:
; Rotate "back" of "message" without changing other values
mov
mov
and

mov
shr
inc

al, message
;
ah, al
;
al, NOT MASK back;
;
;
cl, back
;
ah, cl
;
ah
;

value from memory
a copy for work
1101 1001=ah/al
out old bits
AND 1000 1111=mask
save old message
--------1000 1001=al
Load bit position
Shift to right
0000 1101=ah
Increment
0000 1110=ah

shl
and

ah, cl
ah, MASK back

Shift left again
Mask off extra bits
to get new message

ah, al

mov

message, ah

;
;
;
;
;
;
;

Load
Save
Mask
to

Combine old and new
Write back to memory

1110 0000=ah
AND 0111 0000=mask
--------0110 0000 ah
OR 1000 1001 al
--------1110 1001 ah

Record variables are often used with the logical operators to perform logical
operations on the bit fields of the record, as in the previous example using the
MASK operator.

Filename: LMAPGC05.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 137 of 33 Printed: 10/02/00 04:23 PM

137

135

C H A P T E R

Using Floating-Point and
Binary Coded Decimal Numbers

MASM requires different techniques for handling floating-point (real) numbers
and binary coded decimal (BCD) numbers than for handling integers. You have
two choices for working with real numbers — a math coprocessor or emulation
routines.
Math coprocessors — the 8087, 80287, and 80387 chips — work with the main
processor to handle real-number calculations. The 80486 processor performs
floating-point operations directly. All information in this chapter pertaining to the
80387 coprocessor applies to the 80486DX processor as well. It does not apply
to the 80486SX, which does not provide an on-chip coprocessor.
This chapter begins with a summary of the directives and formats of floatingpoint data that you need to allocate memory storage and initialize variables
before you can work with floating-point numbers.
The chapter then explains how to use a math coprocessor for floating-point
operations. It covers:
u
u
u
u

The architecture of the registers.
The operands for the coprocessor instruction formats.
The coordination of coprocessor and main processor memory access.
The basic groups of coprocessor instructions — for loading and storing data,
doing arithmetic calculations, and controlling program flow.

The next main section describes emulation libraries. The emulation routines
provided with all Microsoft high-level languages enable you to use coprocessor
instructions as though your computer had a math coprocessor. However, some
coprocessor instructions are not handled by emulation, as this section explains.
Finally, because math coprocessor and emulation routines can also operate on
BCD numbers, this chapter includes the instruction set for these numbers.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 135 of 1 Printed: 10/02/00 04:23 PM

136

Programmer’s Guide

Using Floating-Point Numbers
Before using floating-point data in your program, you need to allocate the
memory storage for the data. You can then initialize variables either as real
numbers in decimal form or as encoded hexadecimals. The assembler stores
allocated data in 10-byte IEEE format. This section covers floating-point
declarations and floating-point data formats.

Declaring Floating-Point Variables and Constants
You can allocate real constants using the REAL4, REAL8, and REAL10
directives. These directives allocate the following floating-point numbers:
Directive

Size

REAL4

Short (32-bit) real numbers

REAL8

Long (64-bit) real numbers

REAL10

10-byte (80-bit) real numbers and BCD numbers

Table 6.1 lists the possible ranges for floating-point variables. The number of
significant digits can vary in an arithmetic operation as the least-significant digit
may be lost through rounding errors. This occurs regularly for short and long
real numbers, so you should assume the lesser value of significant digits shown
in Table 6.1. Ten-byte real numbers are much less susceptible to rounding
errors for reasons described in the next section. However, under certain
circumstances, 10-byte real operations can have a precision of only 18 digits.
Table 6.1 Ranges of Floating-Point Variables
Data Type

Bits

Significant
Digits

Short real

6–7

Long real

15–16

10-byte real

Approximate Range
1.18 x 10- 38 to 3.40 x 1038
2.23 x 10- 308 to 1.79 x 10308
3.37 x 10- 4932 to 1.18 x 104932

With versions of MASM prior to 6.0, the DD, DQ, and DT directives could
allocate real constants. MASM 6.1 still supports these directives, but the
variables are integers rather than floating-point values. Although this makes no
difference in the assembly code, CodeView displays the values incorrectly.
You can specify floating-point constants either as decimal constants or as
encoded hexadecimal constants. You can express decimal real-number constants
in the form:
[[+ | –]] integer[[fraction]][[E[[+ | –]]exponent]]

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 136 of 2 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

137

For example, the numbers 2.523E1 and -3.6E-2 are written in the correct
decimal format. You can use these numbers as initializers for real-number
variables.
The assembler always evaluates digits of real numbers as base 10. It converts
real-number constants given in decimal format to a binary format. The sign,
exponent, and decimal part of the real number are encoded as bit fields within
the number.
You can also specify the encoded format directly with hexadecimal digits (0–9
plus A–F). The number must begin with a decimal digit (0–9) and end with the
real-number designator (R). It cannot be signed. For example, the hexadecimal
number 3F800000r can serve as an initializer for a doubleword-sized variable.
The maximum range of exponent values and the number of digits required in the
hexadecimal number depend on the directive. The number of digits for encoded
numbers used with REAL4, REAL8, and REAL10 must be 8, 16, and 20
digits, respectively. If the number has a leading zero, the number must be 9, 17,
or 21 digits.
Examples of decimal constant and hexadecimal specifications are shown here:
; Real numbers
short
REAL4
double REAL8
tenbyte REAL10

25.23
2.523E1
2523.0E-2

; IEEE format
; IEEE format
; 10-byte real format

; Encoded as hexadecimals
ieeeshort
REAL4
3F800000r
; 1.0 as IEEE short
ieeedouble
REAL8
3FF0000000000000r
; 1.0 as IEEE long
temporary
REAL10
3FFF8000000000000000r ; 1.0 as 10-byte
;
real

The section “Storing Numbers in Floating-Point Format,” following, explains the
IEEE formats — the way the assembler actually stores the data.
Pascal or C programmers may prefer to create language-specific TYPEDEF
declarations, as illustrated in this example:

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 137 of 3 Printed: 10/02/00 04:23 PM

138

Programmer’s Guide
; C-language specific
float
TYPEDEF REAL4
double
TYPEDEF REAL8
long_double
TYPEDEF REAL10
; Pascal-language specific
SINGLE
TYPEDEF REAL4
DOUBLE
TYPEDEF REAL8
EXTENDED
TYPEDEF REAL10

For applications of TYPEDEF, see “Defining Pointer Types with TYPEDEF,”
page 75.

Storing Numbers in Floating-Point Format
The assembler stores floating-point variables in the IEEE format. MASM 6.1
does not support .MSFLOAT and Microsoft binary format, which are available
in version 5.1 and earlier. Figure 6.1 illustrates the IEEE format for encoding
short (4-byte), long (8-byte), and 10-byte real numbers. Although this figure
places the most significant bit first for illustration, low bytes actually appear first
in memory.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 138 of 4 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

Figure 6.1

139

Encoding for Real Numbers in IEEE Format

The following list explains how the parts of a real number are stored in the IEEE
format. Each item in the list refers to an item in Figure 6.1.
u
u

Sign bit (0 for positive or 1 for negative) in the upper bit of the first byte.
Exponent in the next bits in sequence (8 bits for a short real number, 11 bits
for a long real number, and 15 bits for a 10-byte real number).

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 139 of 5 Printed: 10/02/00 04:23 PM

140

Programmer’s Guide
u

The integer part of the significand in bit 63 for the 10-byte real format. By
absorbing carry values, this bit allows 10-byte real operations to preserve
precision to 19 digits. The integer part is always 1 in short and long real
numbers; consequently, these formats do not provide a bit for the integer,
since there is no point in storing it.

Decimal part of the significand in the remaining bits. The length is 23 bits for
short real numbers, 52 bits for long real numbers, and 63 bits for 10-byte real
numbers.

The exponent field represents a multiplier 2n. To accommodate negative
exponents (such as 2-6), the value in the exponent field is biased; that is, the
actual exponent is determined by subtracting the appropriate bias value from the
value in the exponent field. For example, the bias for short real numbers is 127.
If the value in the exponent field is 130, the exponent represents a value of 2130127, or 23. The bias for long real numbers is 1,023. The bias for 10-byte real
numbers is 16,383.
Once you have declared floating-point data for your program, you can use
coprocessor or emulator instructions to access the data. The next section
focuses on the coprocessor architecture, instructions, and operands required for
floating-point operations.

Using a Math Coprocessor
When used with real numbers, packed BCD numbers, or long integers,
coprocessors (the 8087, 80287, 80387, and 80486) calculate many times faster
than the 8086-based processors. The coprocessor handles data with its own
registers. The organization of these registers can be one of the four formats for
using operands explained in “Instruction and Operand Formats,” later in this
section.
This section describes how the coprocessor transfers data to and from the
coprocessor, coordinates processor and coprocessor operations, and controls
program flow.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 140 of 6 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

141

Coprocessor Architecture
The coprocessor accesses memory as the CPU does, but it has its own data and
control registers — eight data registers organized as a stack and seven control
registers similar to the 8086 flag registers. The coprocessor’s instruction set
provides direct access to these registers.
The eight 80-bit data registers of the 8087-based coprocessors are organized as
a stack, although they need not be used as a stack. As data items are pushed
into the top register, previous data items move into higher-numbered registers,
which are lower on the stack. Register 0 is the top of the stack; register 7 is the
bottom. The syntax for specifying registers is:
ST [[(number)]]
The number must be a digit between 0 and 7 or a constant expression that
evaluates to a number from 0 to 7. ST is another way to refer to ST(0).
All coprocessor data is stored in registers in the 10-byte real format. The
registers and the register format are shown in Figure 6.2.

Figure 6.2

Coprocessor Data Registers

Internally, all calculations are done on numbers of the same type. Since 10-byte
real numbers have the greatest precision, lower-precision numbers are
guaranteed not to lose precision as a result of calculations. The instructions that
transfer values between the main memory and the coprocessor automatically
convert numbers to and from the 10-byte real format.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 141 of 7 Printed: 10/02/00 04:23 PM

142

Programmer’s Guide

Instruction and Operand Formats
Because of the stack organization of registers, you can consider registers either
as elements on a stack or as registers much like 8086-family registers. Table 6.2
lists the four main groups of coprocessor instructions and the general syntax for
each. The names given to the instruction format reflect the way the instruction
uses the coprocessor registers. The instruction operands are placed in the
coprocessor data registers before the instruction executes.
Table 6.2 Coprocessor Operand Formats
Instruction
Format

Syntax

Implied
Operands

Example

Classical stack

Finstruction

ST, ST(1)

fadd

Memory

Finstruction memory

fadd memloc

Finstruction ST(num), ST

—

fadd st(5), st
fadd st, st(3)

—

faddp st(4), st

Finstruction ST, ST(num)
Register pop

FinstructionP ST(num), ST

You can easily recognize coprocessor instructions because, unlike all 8086family instruction mnemonics, they start with the letter F. Coprocessor
instructions can never have immediate operands and, with the exception of the
FSTSW instruction, they cannot have processor registers as operands.

Classical-Stack Format
Instructions in the classical-stack format treat the coprocessor registers like items
on a stack — thus its name. Items are pushed onto or popped off the top
elements of the stack. Since only the top item can be accessed on a traditional
stack, there is no need to specify operands. The first (top) register (and the
second, if the instruction needs two operands) is always assumed.
ST (the top of the stack) is the source operand in coprocessor arithmetic
operations. ST(1), the second register, is the destination. The result of the
operation replaces the destination operand, and the source is popped off the
stack. This leaves the result at the top of the stack.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 142 of 8 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

143

The following example illustrates the classical-stack format; Figure 6.3 shows
the status of the register stack after each instruction.
fld1
fldpi
fadd

Figure 6.3

; Push 1 into first position
; Push pi into first position
; Add pi and 1 and pop

Status of the Register Stack

Memory Format
Instructions that use the memory format, such as data transfer instructions, also
treat coprocessor registers like items on a stack. However, with this format,
items are pushed from memory onto the top element of the stack, or popped
from the top element to memory. You must specify the memory operand.
Some instructions that use the memory format specify how a memory operand
is to be interpreted — as an integer (I) or as a binary coded decimal (B). The
letter I or B follows the initial F in the syntax. For example, FILD interprets its
operand as an integer and FBLD interprets its operand as a BCD number. If the
instruction name does not include a type letter, the instruction works on real
numbers.
You can also use memory operands in calculation instructions that operate on
two values (see “Using Coprocessor Instructions,” later in this section). The
memory operand is always the source. The stack top (ST) is always the implied
destination.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 143 of 9 Printed: 10/02/00 04:23 PM

144

Programmer’s Guide

The result of the operation replaces the destination without changing its stack
position, as shown in this example and in Figure 6.4:
m1
m2

Figure 6.4

.DATA
REAL4
REAL4
.CODE
.
.
.
fld
fld
fadd
fstp
fst

1.0
2.0

m1
m2
m1
m1
m2

;
;
;
;
;

Push m1 into first position
Push m2 into first position
Add m2 to first position
Pop first position into m1
Copy first position to m2

Status of the Register Stack and Memory Locations

Register Format
Instructions that use the register format treat coprocessor registers as registers
rather than as stack elements. Instructions that use this format require two
register operands; one of them must be the stack top (ST).
In the register format, specify all operands by name. The first operand is the
destination; its value is replaced with the result of the operation. The second
operand is the source; it is not affected by the operation. The stack positions of
the operands do not change.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 144 of 10 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

145

The only instructions that use the register operand format are the FXCH
instruction and arithmetic instructions for calculations on two values. With the
FXCH instruction, the stack top is implied and need not be specified, as shown
in this example and in Figure 6.5:
fadd

st(1), st

fadd

st, st(2)

fxch

st(1)

Figure 6.5

; Add second position to first ;
result goes in second position
; Add first position to third ;
result goes in first position
; Exchange first and second positions

Status of the Previously Initialized Register Stack

Register-Pop Format
The register-pop format treats coprocessor registers as a modified stack. The
source register must always be the stack top. Specify the destination with the
register’s name.
Instructions with this format place the result of the operation into the destination
operand, and the top pops off the stack. The register-pop format is used only
for instructions for calculations on two values, as in this example and in Figure
6.6:
faddp

st(2), st ; Add first and third positions and pop ;
first position destroyed;
;
third moves to second and holds result

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 145 of 11 Printed: 10/02/00 04:23 PM

146

Programmer’s Guide
Figure 6.6

Status of the Already Initialized Register Stack

Coordinating Memory Access
The math coprocessor and main processor work simultaneously. However,
since the coprocessor cannot handle device input or output, data originates in the
main processor.
The main processor and the coprocessor have their own registers, which are
separate and inaccessible to each other. They exchange data through memory,
since memory is available to both.
When using the coprocessor, follow these three steps:
1. Load data from memory to coprocessor registers.
2. Process the data.
3. Store the data from coprocessor registers back to memory.
Step 2, processing the data, can occur while the main processor is handling other
tasks. Steps 1 and 3 must be coordinated with the main processor so that the
processor and coprocessor do not try to access the same memory at the same
time; otherwise, problems of coordinating memory access can occur. Since the
processor and coprocessor work independently, they may not finish working on
memory in the order in which you give instructions. The two potential timing
conflicts that can occur are handled in different ways.
One timing conflict results from a coprocessor instruction following a processor
instruction. The processor may have to wait until the coprocessor finishes if the
next processor instruction requires the result of the coprocessor’s calculation.
You do not have to write your code to avoid this conflict, however. The
assembler coordinates this timing automatically for the 8088 and 8086
processors, and the processor coordinates it automatically on the 80186–80486
processors. This is the case shown in the first example that follows.
Another conflict results from a processor instruction that accesses memory
following a coprocessor instruction that accesses the same memory. The
processor can try to load a variable that is still being used by the coprocessor.
You need careful synchronization to control the timing, and this synchronization
is not automatic on the 8087 coprocessor. For code to run correctly on the
8087, you must include WAIT or FWAIT (mnemonics for the same instruction)
to ensure that the coprocessor finishes before the processor begins, as shown in
the second example.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 146 of 12 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

147

In this situation, the processor does not generate the FWAIT instruction
automatically.
; Processor instruction first - No wait needed
mov
WORD PTR mem32[0], ax
; Load memory
mov
WORD PTR mem32[2], dx
fild
mem32
; Load to register
; Coprocessor instruction first - Wait needed (for 8087)
fist
mem32
; Store to memory
fwait
; Wait until coprocessor
;
is done
mov
ax, WORD PTR mem32[0]
; Move to register
mov
dx, WORD PTR mem32[2]

When generating code for the 8087 coprocessor, the assembler automatically
inserts a WAIT instruction before the coprocessor instruction. However, if you
use the .286 or .386 directive, the compiler assumes that the coprocessor
instructions are for the 80287 or 80387 and does not insert the WAIT
instruction. If your code does not need to run on an 8086 or 8088 processor,
you can make your programs smaller and more efficient by using the .286 or
.386 directive.

Using Coprocessor Instructions
The 8087 family of coprocessors has separate instructions for each of the
following operations:
u
u
u

Loading and storing data
Doing arithmetic calculations
Controlling program flow

The following sections explain the available instructions and show how to use
them for each of these operations. For general syntax information, see
“Instruction and Operand Formats,” earlier in this section.

Loading and Storing Data
Data-transfer instructions copy data between main memory and the coprocessor
registers or between different coprocessor registers. Two principles govern data
transfers:
u

The choice of instruction determines whether a value in memory is
considered an integer, a BCD number, or a real number. The value is always
considered a 10-byte real number once transferred to the coprocessor.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 147 of 13 Printed: 10/02/00 04:23 PM

148

Programmer’s Guide
u

The size of the operand determines the size of a value in memory. Values in
the coprocessor always take up 10 bytes.

You can transfer data to stack registers using load commands. These commands
push data onto the stack from memory or from coprocessor registers. Store
commands remove data. Some store commands pop data off the register stack
into memory or coprocessor registers; others simply copy the data without
changing it on the stack.
If you use constants as operands, you cannot load them directly into
coprocessor registers. You must allocate memory and initialize a variable to a
constant value. That variable can then be loaded by using one of the load
instructions in the following list.
The math coprocessor offers a few special instructions for loading certain
constants. You can load 0, 1, pi, and several common logarithmic values
directly. Using these instructions is faster and often more precise than loading
the values from initialized variables.
All instructions that load constants have the stack top as the implied destination
operand. The constant to be loaded is the implied source operand.
The coprocessor data area, or parts of it, can also be moved to memory and
later loaded back. You may want to do this to save the current state of the
coprocessor before executing a procedure. After the procedure ends, restore the
previous status. Saving coprocessor data is also useful when you want to modify
coprocessor behavior by writing certain data to main memory, operating on the
data with 8086-family instructions, and then loading it back to the coprocessor
data area.
Use the following instructions for transferring numbers to and from
registers:
Instruction(s)

Description

FLD, FST, FSTP

Loads and stores real numbers

FILD, FIST, FISTP

Loads and stores binary integers

FBLD

Loads BCD

FBSTP

Stores BCD

FXCH

Exchanges register values

FLDZ

Pushes 0 into ST

FLD1

Pushes 1 into ST

FLDPI

Pushes the value of pi into ST

FLDCW mem2byte

Loads the control word into the coprocessor

F[[N]]STCW mem2byte

Stores the control word in memory

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 148 of 14 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers
FLDENV mem14byte

Loads environment from memory

F[[N]]STENV mem14byte

Stores environment in memory

Instruction(s)

Description

FRSTOR mem94byte

Restores state from memory

F[[N]]SAVE mem94byte

Saves state in memory

FLDL2E

Pushes the value of log2e into ST

FLDL2T

Pushes log210 into ST

FLDLG2

Pushes log102 into ST

FLDLN2

Pushes loge2 into ST

The following example and Figure 6.7 illustrate some of these instructions:
m1
m2

.DATA
REAL4
REAL4
.CODE
fld
fld
fst
fxch
fstp

Figure 6.7

1.0
2.0
m1
st(2)
m2
st(2)
m1

;
;
;
;
;

Push m1 into first item
Push third item into first
Copy first item to m2
Exchange first and third items
Pop first item into m1

Status of the Register Stack: Main Memory and Coprocessor

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 149 of 15 Printed: 10/02/00 04:23 PM

149

150

Programmer’s Guide

Doing Arithmetic Calculations
Most of the coprocessor instructions for arithmetic operations have several
forms, depending on the operand used. You do not need to specify the operand
type in the

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 150 of 16 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

151

instruction if both operands are stack registers, since register values are always
10-byte real numbers. In most of the arithmetic instructions listed here, the
result replaces the destination register. The instructions include:
Instruction

Description

FADD

Adds the source and destination

FSUB

Subtracts the source from the destination

FSUBR

Subtracts the destination from the source

FMUL

Multiplies the source and the destination

FDIV

Divides the destination by the source

FDIVR

Divides the source by the destination

FABS

Sets the sign of ST to positive

FCHS

Reverses the sign of ST

FRNDINT

Rounds ST to an integer

FSQRT

Replaces the contents of ST with its square root

FSCALE

Multiplies the stack-top value by 2 to the power contained in ST(1)

FPREM

Calculates the remainder of ST divided by ST(1)

80387 Only
Instruction

Description

FSIN

Calculates the sine of the value in ST

FCOS

Calculates the cosine of the value in ST

FSINCOS

Calculates the sine and cosine of the value in ST

FPREM1

Calculates the partial remainder by performing modulo division on the top
two stack registers

FXTRACT
F2XM1

Breaks a number down into its exponent and mantissa and pushes the
mantissa onto the register stack
Calculates 2x–1

FYL2X

Calculates Y * log2 X

FYL2XP1

Calculates Y * log2 (X+1)

FPTAN

Calculates the tangent of the value in ST

FPATAN

Calculates the arctangent of the ratio Y/X

F[[N]]INIT

Resets the coprocessor and restores all the default conditions in the control
and status words

F[[N]]CLEX

Clears all exception flags and the busy flag of the status word

FINCSTP

Adds 1 to the stack pointer in the status word

FDECSTP

Subtracts 1 from the stack pointer in the status word

FFREE

Marks the specified register as empty

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 151 of 17 Printed: 10/02/00 04:23 PM

152

Programmer’s Guide

The following example illustrates several arithmetic instructions. The code
solves quadratic equations, but does no error checking and fails for some values
because it attempts to find the square root of a negative number. Both Help and
the MATH.ASM sample file show a complete version of this procedure. The
complete form uses the FTST (Test for Zero) instruction to check for a negative
number or 0 before calculating the square root.
a
b
cc
posx
negx

.DATA
REAL4
REAL4
REAL4
REAL4
REAL4

3.0
7.0
2.0
0.0
0.0

.CODE
.
.
.
; Solve quadratic equation - no error checking
; The formula is: -b +/- squareroot(b2 - 4ac) / (2a)
fld1
; Get constants 2 and 4
fadd
st,st
; 2 at bottom
fld
st
; Copy it
fmul
a
; = 2a
fmul
fxch
fmul

st(1),st

fld
fmul
fsubr

b
st,st

fsqrt
fld
fchs
fxch

fld
fadd
fxch
fsubp

st
st,st(2)

fdiv
fstp
fdivr
fstp

st,st(2)
posx

st(2),st

negx

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 152 of 18 Printed: 10/02/00 04:23 PM

; = 4a
; Exchange
; = 4ac
;
;
;
;
;
;
;
;

Load b
= b2
= b2 - 4ac
Negative value here produces error
= square root(b2 - 4ac)
Load b
Make it negative
Exchange

;
;
;
;

Copy square root
Plus version = -b + root(b2 - 4ac)
Exchange
Minus version = -b - root(b2 - 4ac)

;
;
;
;

Divide plus version
Store it
Divide minus version
Store it

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

153

Controlling Program Flow
The math coprocessor has several instructions that set control flags in the status
word. The 8087-family control flags can be used with conditional jumps to
direct program flow in the same way that 8086-family flags are used. Since the
coprocessor does not have jump instructions, you must transfer the status word
to memory so that the flags can be used by 8086-family instructions.
An easy way to use the status word with conditional jumps is to move its upper
byte into the lower byte of the processor flags, as shown in this example:
fstsw
fwait
mov
sahf

mem16
ax, mem16

;
;
;
;

Store status word in memory
Make sure coprocessor is done
Move to AX
Store upper word in flags

The SAHF (Store AH into Flags) instruction in this example transfers AH into
the low bits of the flags register.
You can save several steps by loading the status word directly to AX on the
80287 with the FSTSW and FNSTSW instructions. This is the only case in
which data can be transferred directly between processor and coprocessor
registers, as shown in this example:
fstsw

The coprocessor control flags and their relationship to the status word are
described in “Control Registers,” following.
The 8087-family coprocessors provide several instructions for comparing
operands and testing control flags. All these instructions compare the stack top
(ST) to a source operand, which may either be specified or implied as ST(1).
The compare instructions affect the C3, C2, and C0 control flags, but not the
C1 flag. Table 6.3 shows the flags’ settings for each possible result of a
comparison or test.
Table 6.3 Control-Flag Settings After Comparison or Test
After FCOM

After FTEST

ST > source

ST is positive

ST < source

ST is negative

ST = source

ST is 0

Not comparable

ST is NAN or projective infinity

Variations on the compare instructions allow you to pop the stack once or twice
and to compare integers and zero. For each instruction, the stack top is always

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 153 of 19 Printed: 10/02/00 04:23 PM

154

Programmer’s Guide

the implied destination operand. If you do not give an operand, ST(1) is the
implied source. With some compare instructions, you can specify the source as
a memory or register operand.
All instructions summarized in the following list have implied operands: either
ST as a single-destination operand or ST as the destination and ST(1) as the
source. Each instruction in the list has implied operands. Some instructions have
a wait version and a no-wait version. The no-wait versions have N as the
second letter. The instructions for comparing and testing flags include:
Instruction

Description

FCOM

Compares the stack top to the source. The
source and destination are unaffected by the comparison.

FTST

Compares ST to 0.

FCOMP

Compares the stack top to the source and then pops the
stack.

FUCOM , FUCOMP,
FUCOMPP

Compares the source to ST and sets the condition codes of
the status word according to the result (80386/486 only).

F[[N]]STSW mem2byte

Stores the status word in memory.

FXAM

Sets the value of the control flags based on the type of the
number in ST.

FPREM

Finds a correct remainder for large operands. It uses the
C2 flag to indicate whether the remainder returned is partial
(C2 is set) or complete (C2 is clear). If the bit is set, the
operation should be repeated. It also returns the leastsignificant three bits of the quotient in C0, C3, and C1.

FNOP

Copies the stack top onto itself, thus padding the
executable file and taking up processing time without having
any effect on registers or memory.

FDISI , FNDISI , FENI, FNENI

Enables or disables interrupts (8087 only).

FSETPM

Sets protected mode. Requires a .286P or .386P directive
(80287, 80387, and 80486 only).

The following example illustrates some of these instructions. Notice how
conditional blocks are used to enhance 80287 code.
down
across
diamtr
status

.DATA
REAL4
REAL4
REAL4
WORD

10.35
13.07
12.93
?

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 154 of 20 Printed: 10/02/00 04:23 PM

; Sides of a rectangle
; Diameter of a circle

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers
P287

EQU
(@Cpu AND 00111y)
.CODE
.
.
.
; Get area of rectangle
fld
across
; Load one side
fmul
down
; Multiply by the other
; Get area of circle: Area = PI * (D/2)2
fld1
; Load one and
fadd
st, st
;
double it to get constant 2
fdivr
diamtr
; Divide diameter to get radius
fmul
st, st
; Square radius
fldpi
; Load pi
fmul
; Multiply it
; Compare area of circle and
fcompp
;
IF
p287
fstsw
ax
;
ELSE
fnstsw status
;
mov
ax, status ;
ENDIF
sahf
;
jp
nocomp
;
jz
same
;
jc
rectangle ;
jmp
circle
;
nocomp:

rectangle
Compare and throw both away
(For 287+, skip memory)
Load from coprocessor to memory
Transfer memory to register
Transfer AH to flags register
If parity set, can't compare
If zero set, they're the same
If carry set, rectangle is bigger
else circle is bigger

; Error handler
.
.
.

same:
.
.
.
rectangle:
.
.
.
circle:

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 155 of 21 Printed: 10/02/00 04:23 PM

; Both equal

; Rectangle bigger

; Circle bigger

155

156

Programmer’s Guide

Additional instructions for the 80387/486 are FLDENVD and FLDENVW for
loading the environment; FNSTENVD, FNSTENVW, FSTENVD, and
FSTENVW for storing the environment state; FNSAVED, FNSAVEW,
FSAVED, and FSAVEW for saving the coprocessor state; and FRSTORD and
FRSTORW for restoring the coprocessor state.
The size of the code segment, not the operand size, determines the number of
bytes loaded or stored with these instructions. The instructions ending with W
store the 16-bit form of the control register data, and the instructions ending
with D store the 32-bit form. For example, in 16-bit mode FSAVEW saves the
16-bit control register data. If you need to store the 32-bit form of the control
register data, use FSAVED.

Control Registers
Some of the flags of the seven 16-bit control registers control coprocessor
operations, while others maintain the current status of the coprocessor. In this
sense, they are much like the 8086-family flags registers (see Figure 6.8).

Figure 6.8

Coprocessor Control Registers

The status word register is the only commonly used control register. (The others
are used mostly by systems programmers.) The format of the status word
register is shown in Figure 6.9, which shows how the coprocessor control flags
align with the processor flags. C3 overwrites the zero flag, C2 overwrites the
parity flag, and C0 overwrites the carry flag. C1 overwrites an undefined bit, so
it cannot be used directly with conditional jumps, although you can use the
TEST instruction to

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 156 of 22 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

157

check C1 in memory or in a register. The status word register also overwrites
the sign and auxiliary-carry flags, so you cannot count on their being unchanged
after the operation.

Figure 6.9

Coprocessor and Processor Control Flags

Using An Emulator Library
If you do not have a math coprocessor or an 80486 processor, you can do most
floating-point operations by writing assembly-language procedures and accessing
an emulator from a high-level language. All Microsoft high-level languages come
with emulator libraries for all memory models.
To use emulator functions, first write your assembly-language procedure using
coprocessor instructions. Then assemble the module with the /FPi option and
link it with your high-level – language modules. You can enter options in the
Programmer’s WorkBench (PWB) environment, or you can use the OPTION
EMULATOR in your source code.
In emulation mode, the assembler generates instructions for the linker that the
Microsoft emulator can use. The form of the OPTION directive in the following
example tells the assembler to use emulation mode. This option (introduced in
Chapter 1) can be defined only once in a module.
OPTION EMULATOR

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 157 of 23 Printed: 10/02/00 04:23 PM

158

Programmer’s Guide

You can use emulator functions in a stand-alone assembler program by
assembling with the /Cx command-line option and linking with the appropriate
emulator library. The following fragment outlines a small-model program that
contains floating-point instructions served by an emulator:
.MODEL
OPTION
.
.
.
PUBLIC
.CODE

small, c
EMULATOR

main

main:
.STARTUP
.
fadd
st, st
fldpi

; Program entry point must
;
have name 'main'
; Floating-point instructions
;
emulated

Emulator libraries do not allow for all of the coprocessor instructions. The
following floating-point instructions are not emulated:
FBLD
FBSTP
FCOS
FDECSTP
FINCSTP
FINIT

FLDENV
FNOP
FPREM1
FRSTOR
FRSTORW
FRSTORD

FSAVE
FSAVEW
FSAVED
FSETPM
FSIN
FSINCOS

FSTENV
FUCOM
FUCOMP
FUCOMPP
FXTRACT

For information about writing assembly-language procedures for high-level
languages, see Chapter 12, “Mixed-Language Programming.”

Using Binary Coded Decimal Numbers
Binary coded decimal (BCD) numbers allow calculations on large numbers
without rounding errors. This characteristic makes BCD numbers a common
choice for monetary calculations. Although BCDs can represent integers of any
precision, the 8087-based coprocessors accommodate BCD numbers only in the
range ±999,999,999,999,999,999.
This section explains how to define BCD numbers, how to access them with a
math coprocessor or emulator, and how to perform simple BCD calculations on
the main processor.

Defining BCD Constants and Variables
Unpacked BCD numbers are made up of bytes containing a single decimal digit
in the lower 4 bits of each byte. Packed BCD numbers are made up of bytes

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 158 of 24 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

159

containing two decimal digits: one in the upper 4 bits and one in the lower 4 bits.
The leftmost digit holds the sign (0 for positive, 1 for negative).
Packed BCD numbers are encoded in the 8087 coprocessor’s packed BCD
format. They can be up to 18 digits long, packed two digits per byte. The
assembler zero-pads BCDs initialized with fewer than 18 digits. Digit 20 is the
sign bit, and digit 19 is reserved.
When you define an integer constant with the TBYTE directive and the current
radix is decimal (t), the assembler interprets the number as a packed BCD
number.
The syntax for specifying packed BCDs is the same as for other integers.
pos1
neg1

TBYTE
TBYTE

1234567890 ; Encoded as 00000000001234567890h
-1234567890 ; Encoded as 80000000001234567890h

Unpacked BCD numbers are stored one digit to a byte, with the value in the
lower
4 bits. They can be defined using the BYTE directive. For example, an
unpacked BCD number could be defined and initialized as follows:
unpackedr
unpackedf

BYTE
BYTE

1,5,8,2,5,2,9
9,2,5,2,8,5,1

; Initialized to 9,252,851
; Initialized to 9,252,851

As these two lines show, you can arrange digits backward or forward, depending
on how you write the calculation routines that handle the numbers.

BCD Calculations on a Coprocessor
As the previous section explains, BCDs differ from other numbers only in the
way a program stores them in memory. Internally, a math coprocessor does not
distinguish BCD integers from any other type. The coprocessor can load,
calculate, and store packed BCD integers up to 18 digits long.
The coprocessor instruction
fbld

bcd1

pushes the packed BCD number at bcd1 onto the coprocessor stack. When
your code completes calculations on the number, place the result back into
memory in BCD format with the instruction
fbstp

bcd1

which discards the variable from the stack top.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 159 of 25 Printed: 10/02/00 04:23 PM

160

Programmer’s Guide

BCD Calculations on the Main Processor
The 8086-family of processors can perform simple arithmetic operations on
BCD integers, but only one digit at a time. The main processor, like the
coprocessor, operates internally on the number’s binary value. It requires
additional code to translate the binary result back into BCD format.
The main processor provides instructions specifically designed to translate to
and from BCD format. These instructions are called “ASCII-adjust” and
“decimal-adjust” instructions. They get their names from Intel mnemonics that
use the term “ASCII” to refer to unpacked BCD numbers and “decimal” to
refer to packed BCD numbers.

Unpacked BCD Numbers
When a calculation using two one-digit values produces a two-digit result, the
instructions AAA, AAS, AAM, and AAD place the first digit in AL and the
second in AH. If the digit in AL needs to carry to or borrow from the digit in
AH, the instructions set the carry and auxiliary carry flags. The four ASCIIadjust instructions for unpacked BCDs are:
Instruction

Description

AAA

Adjusts after an addition operation.

AAS

Adjusts after a subtraction operation.

AAM

Adjusts after a multiplication operation. Always use with MUL, not with
IMUL.

AAD

Adjusts before a division operation. Unlike other BCD instructions, AAD
converts a BCD value to a binary value before the operation. After the
operation, use AAM to adjust the quotient. The remainder is lost. If you
need the remainder, save it in another register before adjusting the quotient.
Then move it back to AL and adjust if necessary.

For processor arithmetic on unpacked BCD numbers, you must do the 8-bit
arithmetic calculations on each digit separately, and assign the result to the AL
register. After each operation, use the corresponding BCD instruction to adjust
the result. The ASCII-adjust instructions do not take an operand and always
work on the value in the AL register.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 160 of 26 Printed: 10/02/00 04:23 PM

Chapter 6 Using Floating-Point and Binary Coded Decimal Numbers

161

The following examples show how to use each of these instructions in BCD
addition, subtraction, multiplication, and division.
; To add 9 and 3 as BCDs:
mov
ax, 9
mov
bx, 3
add
al, bl
aaa

; To subtract 4 from 13:
mov
ax, 103h
mov
bx, 4
sub
al, bl
aas

; To multiply 9 times 3:
mov
ax, 903h
mul
ah
aam

; To divide 25 by 2:
mov
ax, 205h
mov
bl, 2
aad
div

aam

;
;
;
;
;
;

Load 9
and 3 as unpacked BCDs
Add 09h and 03h to get 0Ch
Adjust 0Ch in AL to 02h,
increment AH to 01h, set carry
Result 12 (unpacked BCD in AX)

;
;
;
;
;
;

Load 13
and 4 as unpacked BCDs
Subtract 4 from 3 to get FFh (-1)
Adjust 0FFh in AL to 9,
decrement AH to 0, set carry
Result 9 (unpacked BCD in AX)

; Load 9 and 3 as unpacked BCDs
; Multiply 9 and 3 to get 1Bh
; Adjust 1Bh in AL
;
to get 27 (unpacked BCD in AX)

;
;
;
;
;
;
;
;
;
;

Load 25
and 2 as unpacked BCDs
Adjust 0205h in AX
to get 19h in AX
Divide by 2 to get
quotient 0Ch in AL
remainder 1 in AH
Adjust 0Ch in AL
to 12 (unpacked BCD in AX)
(remainder destroyed)

If you process multidigit BCD numbers in loops, each digit is processed and
adjusted in turn.

Packed BCD Numbers
Packed BCD numbers are made up of bytes containing two decimal digits: one
in the upper 4 bits and one in the lower 4 bits. The 8086-family processors
provide instructions for adjusting packed BCD numbers after addition and
subtraction. You must write your own routines to adjust for multiplication and
division.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 161 of 27 Printed: 10/02/00 04:23 PM

162

Programmer’s Guide

For processor calculations on packed BCD numbers, you must do the 8-bit
arithmetic calculations on each byte separately, placing the result in the AL
register. After each operation, use the corresponding decimal-adjust instruction
to adjust the result. The decimal-adjust instructions do not take an operand and
always work on the value in the AL register.
The 8086-family processors provide the instructions DAA (Decimal Adjust after
Addition) and DAS (Decimal Adjust after Subtraction) for adjusting packed
BCD numbers after addition and subtraction.
These examples use DAA and DAS to add and subtract BCDs.
;To add 88 and 33:
mov
ax, 8833h
add
al, ah
daa

;To subtract 38 from 83:
mov
ax, 3883h
sub
al, ah
das

; Load 88 and 33 as packed BCDs
; Add 88 and 33 to get 0BBh
; Adjust 0BBh to 121 (packed BCD:)
;
1 in carry and 21 in AL

; Load 83 and 38 as packed BCDs
; Subtract 38 from 83 to get 04Bh
; Adjust 04Bh to 45 (packed BCD:)
;
0 in carry and 45 in AL

Unlike the ASCII-adjust instructions, the decimal-adjust instructions never affect
AH. The assembler sets the auxiliary carry flag if the digit in the lower 4 bits
carries to or borrows from the digit in the upper 4 bits, and it sets the carry flag
if the digit in the upper 4 bits needs to carry to or borrow from another byte.
Multidigit BCD numbers are usually processed in loops. Each byte is processed
and adjusted in turn.

Filename: LMAPGC06.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 59 Page: 162 of 28 Printed: 10/02/00 04:23 PM

161

C H A P T E R

Controlling Program Flow

Very few programs execute all lines sequentially from .STARTUP to .EXIT.
Rather, complex program logic and efficiency dictate that you control the flow
of your program — jumping from one point to another, repeating an action until
a condition is reached, and passing control to and from procedures. This chapter
describes various ways for controlling program flow and several features that
simplify coding program-control constructs.
The first section covers jumps from one point in the program to another. It
explains how MASM 6.1 optimizes both unconditional and conditional jumps
under certain circumstances, so that you do not have to specify every attribute.
The section also describes instructions you can use to test conditional jumps.
The next section describes loop structures that repeat actions or evaluate
conditions. It discusses MASM directives, such as .WHILE and .REPEAT, that
generate appropriate compare, loop, and jump instructions for you, and the .IF,
.ELSE, and .ELSEIF directives that generate jump instructions.
The “Procedures” section in this chapter explains how to write an assemblylanguage procedure. It covers the extended functionality for PROC, a PROTO
directive that lets you write procedure prototypes similar to those used in C, an
INVOKE directive that automates parameter passing, and options for the stackframe setup inside procedures.
The last section explains how to pass program control to an interrupt routine.

Jumps
Jumps are the most direct way to change program control from one location to
another. At the processor level, jumps work by changing the value of the IP
(Instruction Pointer) register to a target offset and, for far jumps, by changing
the CS register to a new segment address. Jump instructions fall into only two
categories: conditional and unconditional.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 161 of 1 Printed: 10/02/00 04:23 PM

162

Programmer’s Guide

Unconditional Jumps
The JMP instruction transfers control unconditionally to another instruction.
JMP’s single operand contains the address of the target instruction.
Unconditional jumps skip over code that should not be executed, as shown here:
; Handle one case
label1: .
.
.
jmp continue
; Handle second case
label2: .
.
.
jmp continue
.
.
.
continue:

The distance of the target from the jump instruction and the size of the operand
determine the assembler’s encoding of the instruction. The longer the distance,
the more bytes the assembler uses to code the instruction. In versions of MASM
prior to 6.0, unconditional NEAR jumps sometimes generated inefficient code,
but MASM can now optimize unconditional jumps.

Jump Optimizing
The assembler determines the smallest encoding possible for the direct
unconditional jump. MASM does not require a distance operator, so you do not
have to determine the correct distance of the jump. If you specify a distance, it
overrides any assembler optimization. If the specified distance falls short of the
target address, the assembler generates an error. If the specified distance is
longer than the jump requires, the assembler encodes the given distance and
does not optimize it.
The assembler optimizes jumps when the following conditions are met:
u

You do not specify SHORT, NEAR, FAR, NEAR16, NEAR32, FAR16,
FAR32, or PROC as the distance of the target.
The target of the jump is not external and is in the same segment as the jump
instruction. If the target is in a different segment (but in the same group), it is
treated as though it were external.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 162 of 2 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

163

If these two conditions are met, MASM uses the instruction, distance, and size
of the operand to determine how to optimize the encoding for the jump. No
syntax changes are necessary.
Note This information about jump optimizing also applies to conditional jumps
on the 80386/486.

Indirect Operands
An indirect operand provides a pointer to the target address, rather than the
address itself. A pointer is a variable that contains an address. The processor
distinguishes indirect (pointer) operands from direct (address) operands by the
instruction’s context.
You can specify the pointer’s size with the WORD, DWORD, or FWORD
attributes. Default sizes are based on .MODEL and the default segment size.
jmp
jmp

[bx]
; Uses .MODEL and segment size defaults
WORD PTR [bx] ; A NEAR16 indirect call

If the indirect operand is a register, the jump is always a NEAR16 jump for a
16-bit register, and NEAR32 for a 32-bit register:
jmp
jmp

bx
ebx

; NEAR16 jump
; NEAR32 jump

A DWORD indirect operand, however, is ambiguous to the assembler.
jmp

DWORD PTR [var]

; A NEAR32 jump in a 32-bit segment;
; a FAR16 jump in a 16-bit segment

In this case, your code must clear the ambiguity with the NEAR32 or FAR16
keywords. The following example shows how to use TYPEDEF to define
NEAR32 and FAR16 pointer types.
NFP
FFP

TYPEDEF
TYPEDEF
jmp
jmp

PTR
PTR
NFP
FFP

NEAR32
FAR16
PTR [var] ; NEAR32 indirect jump
PTR [var] ; FAR16 indirect jump

You can use an unconditional jump as a form of conditional jump by specifying
the address in a register or indirect memory operand. Also, you can use indirect
memory operands to construct jump tables that work like C switch statements,
Pascal CASE statements, or Basic ON GOTO, ON GOSUB, or SELECT
CASE statements, as shown in the following example.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 163 of 3 Printed: 10/02/00 04:23 PM

164

Programmer’s Guide
NPVOID

TYPEDEF NEAR PTR
.DATA
ctl_tbl NPVOID extended,
ctrla,
ctrlb
.CODE
.
.
.
mov
ah, 8h
int
21h
cbw
mov
bx, ax
shl
bx, 1
jmp
ctl_tbl[bx]
extended:
mov
int
.
.
.
jmp
ctrla: .
.
.
jmp
ctrlb: .
.
.
jmp
.
.
next:
.

ah, 8h
21h

; Null key (extended code)
; Address of CONTROL-A key routine
; Address of CONTROL-B key routine

; Get a key
;
;
;
;

Stretch AL into AX
Copy
Convert to address
Jump to key routine

; Get second key of extended key
; Use another jump table
;
for extended keys

next
; CONTROL-A code here

next
; CONTROL-B code here

; Continue

In this instance, the indirect memory operands point to addresses of routines for
handling different keystrokes.

Conditional Jumps
The most common way to transfer control in assembly language is to use a
conditional jump. This is a two-step process:
1. First test the condition.
2. Then jump if the condition is true or continue if it is false.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 164 of 4 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

165

All conditional jumps except two (JCXZ and JECXZ) use the processor flags
for their criteria. Thus, any statement that sets or clears a flag can serve as a test
basis for a conditional jump. The jump statement can be any one of 30
conditional-jump instructions. A conditional-jump instruction takes a single
operand containing the target address. You cannot use a pointer value as a target
as you can with unconditional jumps.

Jumping Based on the CX Register
JCXZ and JECXZ are special conditional jumps that do not consult the
processor flags. Instead, as their names imply, these instructions cause a jump
only if the CX or ECX register is zero. The use of JCXZ and JECXZ with
program loops is covered in the next section, “Loops.”

Jumping Based on the Processor Flags
The remaining conditional jumps in the processor’s repertoire all depend on the
status of the flags register. As the following list shows, several conditional jumps
have two or three names — JE (Jump if Equal) and JZ (Jump if Zero), for
example. Shared names assemble to exactly the same machine instruction, so
you may choose whichever mnemonic seems more appropriate. Jumps that
depend on the status of the flags register include:
Instruction

Jumps if

JC/JB/JNAE

Carry flag is set

JNC/JNB/JAE

Carry flag is clear

JBE/JNA

Either carry or zero flag is set

JA/JNBE

Carry and zero flag are clear

JE/JZ

Zero flag is set

JNE/JNZ

Zero flag is clear

JL/JNGE

Sign flag ≠ overflow flag

JGE/JNL

Sign flag = overflow flag

JLE/JNG

Zero flag is set or sign ≠ overflow

JG/JNLE

Zero flag is clear and sign = overflow

Sign flag is set

JNS

Sign flag is clear

Overflow flag is set

JNO

Overflow flag is clear

JP/JPE

Parity flag is set (even parity)

JNP/JPO

Parity flag is clear (odd parity)

The last two jumps in the list, JPE (Jump if Parity Even) and JPO (Jump if
Parity Odd), are useful only for communications programs. The processor sets

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 165 of 5 Printed: 10/02/00 04:23 PM

166

Programmer’s Guide

the parity flag if an operation produces a result with an even number of set bits.
A communications program can compare the flag against the parity bit received
through the serial port to test for transmission errors.
The conditional jumps in the preceding list can follow any instruction that
changes the processor flags, as these examples show:
; Uses JO to handle overflow condition
add
ax, bx
; Add two values
jo
overflow
; If value too large, adjust
; Uses JNZ to check for zero as the result of subtraction
sub
ax, bx
; Subtract
mov
cx, Count
; First, initialize CX
jnz
skip
; If the result is not zero, continue
call
zhandler
; Else do special case

As the second example shows, the jump does not have to immediately follow
the instruction that alters the flags. Since MOV does not change the flags, it can
appear between the SUB instruction and the dependent jump.
There are three categories of conditional jumps:
u
u
u

Comparison of two values
Individual bit settings in a value
Whether a value is zero or nonzero

Jumps Based on Comparison of Two Values
The CMP instruction is the most common way to test for conditional jumps. It
compares two values without changing either, then sets or clears the processor
flags according to the results of the comparison.
Internally, the CMP instruction is the same as the SUB instruction, except that
CMP does not change the destination operand. Both set flags according to the
result of the subtraction.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 166 of 6 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

167

You can compare signed or unsigned values, but you must choose the
subsequent conditional jump to reflect the correct value type. For example, JL
(Jump if Less Than) and JB (Jump if Below) may seem conceptually similar,
but a failure to understand the difference between them can result in program
bugs. Table 7.1 shows the correct conditional jumps for comparisons of signed
and unsigned values. The table shows the zero, carry, sign, and overflow flags
as ZF, CF, SF, and OF, respectively.
Table 7.1 Conditional Jumps Based on Comparisons of Two Values
Signed Comparisons
Instruction
Jump if True

Unsigned Comparisons
Instruction
Jump if True

ZF = 1

JNE

ZF = 0

JNE

ZF = 0

JG/JNLE

ZF = 0 and SF = OF

JA/JNBE

CF = 0 and ZF = 0

JLE/JNG

ZF = 1 or SF ≠ OF

JBE/JNA

CF = 1 or ZF = 1

JL/JNGE

SF ≠ OF

JB/JNAE

CF = 1

JGE/JNL

SF = OF

JAE/JNB

CF = 0

The mnemonic names of jumps always refer to the comparison of CMP’s first
operand (destination) with the second operand (source). For instance, in this
example, JG tests whether the first operand is greater than the second.
cmp
jg
jl

ax, bx ; Compare AX and BX
next1 ; Equivalent to: If ( AX > BX ) goto next1
next2 ; Equivalent to: If ( AX < BX ) goto next2

Jumps Based on Bit Settings
The individual bit settings in a single value can also serve as the criteria for a
conditional jump. The TEST instruction tests whether specific bits in an operand
are on or off (set or clear), and sets the zero flag accordingly.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 167 of 7 Printed: 10/02/00 04:23 PM

168

Programmer’s Guide

The TEST instruction is the same as the AND instruction, except that TEST
changes neither operand. The following example shows an application of TEST.
.DATA
BYTE
?
.CODE
.
.
.
; If bit 2 or bit 4 is set, then call task_a
; Assume "bits" is 0D3h
test
bits, 10100y
; If 2 or 4 is set
AND
jz
skip1
;
call
task_a
; Then call task_a
skip1:
; Jump taken
.
.
.
; If bits 2 and 4 are clear, then call task_b
; Assume "bits" is 0E9h
test
bits, 10100y
; If 2 and 4 are clear AND
jnz
skip2
;
call
task_b
; Then call task_b
skip2:
; Jump taken
bits

11010011
00010100
-------00010000

11101001
00010100
-------00000000

The source operand for TEST is often a mask in which the test bits are the only
bits set. The destination operand contains the value to be tested. If all the bits
set in the mask are clear in the destination operand, TEST sets the zero flag. If
any of the flags set in the mask are also set in the destination operand, TEST
clears the zero flag.
The 80386/486 processors provide additional bit-testing instructions. The BT
(Bit Test) series of instructions copy a specified bit from the destination operand
to the carry flag. A JC or JNC can then route program flow depending on the
result. For variations on the BT instruction, see the Reference.

Jumps Based on a Value of Zero
A program often needs to jump based on whether a particular register contains a
value of zero. We’ve seen how the JCXZ instruction jumps depending on the
value in the CX register. You can test for zero in other data registers nearly as
efficiently with the OR instruction. A program can OR a register with itself
without changing the register’s contents, then act on the resulting flags status.
For example, the following example tests whether BX is zero:
or
jz

bx, bx
is_zero

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 168 of 8 Printed: 10/02/00 04:23 PM

; Is BX = 0?
; Jump if so

Chapter 7 Controlling Program Flow

169

This code is functionally equivalent to:
cmp
je

bx, 0
is_zero

; Is BX = 0?
; Jump if so

but produces smaller and faster code, since it does not use an immediate number
as an operand. The same technique also lets you test a register’s sign bit:
or
js

dx, dx
sign_set

; Is DX sign bit set?
; Jump if so

Jump Extending
Unlike an unconditional jump, a conditional jump cannot reference a label more
than 128 bytes away. For example, the following statement is valid as long as
target is within a distance of 128 bytes:
; Jump to target less than 128 bytes away
jz
target
; If previous operation resulted
;
in zero, jump to target

However, if target is too distant, the following sequence is necessary to
enable a longer jump. Note this sequence is logically equivalent to the preceding
example:
; Jumps to distant targets previously required two steps
jnz
skip
; If previous operation result is
;
NOT zero, jump to "skip"
jmp
target
;
Otherwise, jump to target
skip:

MASM can automate jump-extending for you. If you target a conditional jump
to a label farther than 128 bytes away, MASM rewrites the instruction with an
unconditional jump, which ensures that the jump can reach its target. If target
lies within a 128-byte range, the assembler encodes the instruction jz target
as is. Otherwise, MASM generates two substitute instructions:
jne $ + 2 + (length in bytes of the next instruction)
jmp NEAR PTR target

The assembler generates this same code sequence if you specify the distance
with NEAR PTR, FAR PTR, or SHORT. Therefore,
jz

NEAR PTR target

becomes

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 169 of 9 Printed: 10/02/00 04:23 PM

170

Programmer’s Guide
jne
jmp

$ + 5
NEAR PTR target

even if target is less than 128 bytes away.
MASM enables automatic jump expansion by default, but you can turn it off
with the NOLJMP form of the OPTION directive. For information about the
OPTION directive, see page 24.
If the assembler generates code to extend a conditional jump, it issues a level 3
warning saying that the conditional jump has been lengthened. You can set the
warning level to 1 for development and to level 3 for a final optimizing pass to
see if you can shorten jumps by reorganizing.
If you specify the distance for the jump and the target is out of range for that
distance, a “Jump out of Range” error results.
Since the JCXZ and JECXZ instructions do not have logical negations,
expansion of the jump instruction to handle targets with unspecified distances
cannot be performed for those instructions. Therefore, the distance must always
be short.
The size and distance of the target operand determines the encoding for
conditional or unconditional jumps to externals or targets in different segments.
The jump-extending and optimization features do not apply in this case.
Note Conditional jumps on the 80386 and 80486 processors can be to targets
up to 32K away, so jump extension occurs only for targets greater than that
distance.

Anonymous Labels
When you code jumps in assembly language, you must invent many label
names. One alternative to continually thinking up new label names is to use
anonymous labels, which you can use anywhere in your program. But because
anonymous labels do not provide meaningful names, they are best used for
jumping over only a few lines of code. You should mark major divisions of a
program with actual named labels.
Use two at signs (@@) followed by a colon (:) as an anonymous label. To jump
to the nearest preceding anonymous label, use @B (back) in the jump
instruction’s operand field; to jump to the nearest following anonymous label,
use @F (forward) in the operand field.
The jump in the following example targets an anonymous label:

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 170 of 10 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow
jge
.
.
.

171

@@:

The items @B and @F always refer to the nearest occurrences of @@:, so there
is never any conflict between different anonymous labels.

Decision Directives
The high-level structures you can use for decision-making are the .IF, .ELSEIF,
and .ELSE statements. These directives generate conditional jumps. The
expression following the .IF directive is evaluated, and if true, the following
instructions are executed until the next .ENDIF, .ELSE, or .ELSEIF directive is
reached. The .ELSE statements execute if the expression is false. Using the
.ELSEIF directive puts a new expression inside the alternative part of the
original .IF statement to be evaluated. The syntax is:
.IF condition1
statements
[[.ELSEIF condition2
statements]]
[[.ELSE
statements]]
.ENDIF
The decision structure
.IF
mov
.ELSE
mov
.ENDIF

cx == 20
dx, 20
dx, 30

generates this code:
.IF
0017
001A
001C

83 F9 14
75 05
BA 0014

001F
0021
0021

EB 03
BA 001E

0024

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 171 of 11 Printed: 10/02/00 04:23 PM

*
*

cx == 20
cmp
cx, 014h
jne
@C0001
mov
dx, 20

.ELSE
*
jmp
*@C0001:
mov
.ENDIF
*@C0003:

@C0003
dx, 30

172

Programmer’s Guide

Loops
Loops repeat an action until a termination condition is reached. This condition
can be a counter or the result of an expression’s evaluation. MASM 6.1 offers
many ways to set up loops in your programs. The following list compares
MASM loop structures:
Instructions

Action

LOOP

Automatically decrements CX. When CX = 0, the loop ends. The top
of the loop cannot be greater than 128 bytes from the LOOP
instruction. (This is true for all LOOP instructions.)

LOOPE/LOOPZ,
LOOPNE/LOOPNZ

Loops while equal or not equal. Checks both CX and the state of the
zero flag. LOOPZ ends when either CX=0 or the zero flag is clear,
whichever occurs first. LOOPNZ ends when either CX=0 or the zero
flag is set, whichever occurs first. LOOPE and LOOPZ assemble to
the same machine instruction, as do LOOPNE and LOOPNZ. Use
whichever mnemonic best fits the context of your loop. Set CX to a
number out of range if you don’t want a count to control the loop.

JCXZ, JECXZ

Branches to a label only if CX = 0 or ECX = 0. Unlike other
conditional-jump instructions, which can jump to either a near or a
short label under the 80386 or 80486, JCXZ and JECXZ always
jump to a short label.

Conditional jumps

Acts only if certain conditions met. Necessary if several conditions
must be tested. See “Conditional Jumps,” page 164.

The following examples illustrate these loop constructions.
; The LOOP instruction: For 200 to 0 do task
mov
cx, 200
; Set counter
next:
.
; Do the task here
.
.
loop
next
; Do again
; Continue after loop
; The LOOPNE instruction: While AX is not 'Y', do task
mov
cx, 256
; Set count too high to interfere
wend:
.
; But don't do more than 256 times
.
; Some statements that change AX
.
cmp
al, 'Y'
; Is it Y or too many times?
loopne wend
; No? Repeat
; Yes? Continue

The JCXZ and JECXZ instructions provide an efficient way to avoid executing
loops when the loop counter CX is empty. For example, consider the following
loops:

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 172 of 12 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow
mov
next:

cx, LoopCount
.
.
.
loop
next

173

; Load loop counter
; Iterate loop CX times

; Do again

If LoopCount is zero, CX decrements to -1 on the first pass. It then must
decrement 65,535 more times before reaching 0. Use a JCXZ to avoid this
problem:
mov
next:

cx, LoopCount
jcxz
done
.
.
.
loop
next

; Load loop counter
; Skip loop if count is 0
; Else iterate loop CX times

; Do again
; Continue after loop

done:

Loop-Generating Directives
The high-level control structures generate loop structures for you. These
directives are similar to the while and repeat loops of C or Pascal, and can
make your assembly programs easier to code and to read. The assembler
generates the appropriate assembly code. These directives are summarized as
follows:
Directives

Action

.WHILE ... .ENDW

The statements between .WHILE condition and .ENDW
execute while the condition is true.

.REPEAT ... .UNTIL

The loop executes at least once and continues until the condition
given after .UNTIL is true. Generates conditional jumps.

.REPEAT ... .UNTILCXZ

Compares label to an expression and generates appropriate
loop instructions.

.BREAK

End a .REPEAT or a .WHILE loop unconditionally.

.CONTINUE

Jump unconditionally past any remaining code to bottom of
loop.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 173 of 13 Printed: 10/02/00 04:23 PM

174

Programmer’s Guide

These constructs work much as they do in a high-level language such as C or
Pascal. Keep in mind the following points:
u

These directives generate appropriate processor instructions. They are not
new instructions.

They require proper use of signed and unsigned data declarations.

These directives cause a set of instructions to execute based on the evaluation of
some condition. This condition can be an expression that evaluates to a signed
or unsigned value, an expression using the binary operators in C (&&, ||, or !),
or the state of a flag. For more information about expression operators, see page
178.
The evaluation of the condition requires the assembler to know if the operands
in the condition are signed or unsigned. To state explicitly that a named memory
location contains a signed integer, use the signed data allocation directives
SBYTE, SWORD, and SDWORD.

.WHILE Loops
As with while loops in C or Pascal, the test condition for .WHILE is checked
before the statements inside the loop execute. If the test condition is false, the
loop does not execute. While the condition is true, the statements inside the loop
repeat.
Use the .ENDW directive to mark the end of the .WHILE loop. When the
condition becomes false, program execution begins at the first statement
following the .ENDW directive. The .WHILE directive generates appropriate
compare and jump statements. The syntax is:
.WHILE condition
statements
.ENDW
For example, this loop copies the contents of one buffer to another until a ‘$’
character (marking the end of the string) is found:
.DATA
buf1
buf2
.CODE
sub
.WHILE
mov
mov
inc
.ENDW

BYTE "This is a string",'$'
BYTE 100 DUP (?)
bx, bx
(buf1[bx] != '$')
al, buf1[bx]
buf2[bx], al
bx

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 174 of 14 Printed: 10/02/00 04:23 PM

; Zero out bx
; Get a character
; Move it to buffer 2
; Count forward

Chapter 7 Controlling Program Flow

175

.REPEAT Loops
MASM’s .REPEAT directive allows for loop constructions like the do loop of C
and the REPEAT loop of Pascal. The loop executes until the condition
following the .UNTIL (or .UNTILCXZ) directive becomes true. Since the
condition is checked at the end of the loop, the loop always executes at least
once. The .REPEAT directive generates conditional jumps. The syntax is:
.REPEAT
statements
.UNTIL condition
.REPEAT
statements
.UNTILCXZ [[condition]]
where condition can also be expr1 == expr2 or expr1 != expr2. When two
conditions are used, expr2 can be an immediate expression, a register, or (if
expr1 is a register) a memory location.
For example, the following code fills a buffer with characters typed at the
keyboard. The loop ends when the ENTER key (character 13) is pressed:
buffer

.DATA
BYTE
100 DUP (0)
.CODE
sub
bx, bx
.REPEAT
mov
ah, 01h
int
21h
mov
buffer[bx], al
inc
bx
.UNTIL (al == 13)

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 175 of 15 Printed: 10/02/00 04:23 PM

; Zero out bx

;
;
;
;

Get a key
Put it in the buffer
Increment the count
Continue until al is 13

176

Programmer’s Guide

The .UNTIL directive generates conditional jumps, but the .UNTILCXZ
directive generates a LOOP instruction, as shown by the listing file code for
these examples. In a listing file, assembler-generated code is preceded by an
asterisk.
ASSUME

bx:PTR SomeStruct
.REPEAT
*@C0001:
inc
ax
.UNTIL ax==6
*
cmp
ax, 006h
*
jne
@C0001

.REPEAT
*@C0003:

mov
.UNTILCXZ
loop

.REPEAT
*@C0004:
.UNTILCXZ
*
cmp
*
loope

ax, 1
@C0003

[bx].field != 6
[bx].field, 006h
@C0004

.BREAK and .CONTINUE Directives
The .BREAK and .CONTINUE directives terminate a .REPEAT or .WHILE
loop prematurely. These directives allow an optional .IF clause for conditional
breaks. The syntax is:
.BREAK [[.IF condition]]
.CONTINUE [[.IF condition]]
Note that .ENDIF is not used with the .IF forms of .BREAK and .CONTINUE
in this context. The .BREAK and .CONTINUE directives work the same way
as the break and continue instructions in C. Execution continues at the
instruction following the .UNTIL, .UNTILCXZ, or .ENDW of the nearest
enclosing loop.
Instead of ending the loop execution as .BREAK does, .CONTINUE causes
loop execution to jump directly to the code that evaluates the loop condition of
the nearest enclosing loop.
The following loop accepts only the keys in the range ‘0’ to ‘9’ and terminates
when you press ENTER.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 176 of 16 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

177

.WHILE 1
; Loop forever
mov
ah, 08h
; Get key without echo
int
21h
.BREAK .IF al == 13
; If ENTER, break out of the loop
.CONTINUE .IF (al < '0') || (al > '9')
; If not a digit, continue looping
mov
dl, al
; Save the character for processing
mov
ah, 02h
; Output the character
int
21h
.ENDW

If you assemble the preceding source code with the /Fl and /Sg command-line
options and then view the results in the listing file, you will see this code:
0017
0017
0019

B4 08
CD 21

001B
001D

3C 0D
74 10

001F
0021
0023
0025
0027
0029
002B

3C
72
3C
77
8A
B4
CD

002D
002F

EB E8

30
F4
39
F0
D0
02
21

.WHILE 1
*@C0001:
mov
int
.BREAK .IF al
*
cmp
*
je
.CONTINUE .IF
*
cmp
*
jb
*
cmp
*
ja
mov
mov
int
.ENDW
*
jmp
*@C0002:

ah, 08h
21h
== 13
al, 00Dh
@C0002
(al '0') || (al
al, '0'
@C0001
al, '9'
@C0001
dl, al
ah, 02h
21h

'9')

@C0001

The high-level control structures can be nested. That is, .REPEAT or .WHILE
loops can contain .REPEAT or .WHILE loops as well as .IF statements.
If the code generated by a .WHILE loop, .REPEAT loop, or .IF statement
generates a conditional or unconditional jump, MASM encodes the jump using
the jump extension and jump optimization techniques described in
“Unconditional Jumps,” page 162, and “Conditional Jumps,” page 164.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 177 of 17 Printed: 10/02/00 04:23 PM

178

Programmer’s Guide

Writing Loop Conditions
You can express the conditions of the .IF, .REPEAT, and .WHILE directives
using relational operators, and you can express the attributes of the operand with
the PTR operator. To write loop conditions, you also need to know how the
assembler evaluates the operators and operands in the condition. This section
explains the operators, attributes, precedence level, and expression evaluation
order for the conditions used with loop-generating directives.

Expression Operators
The binary relational operators in MASM 6.1 are the same binary operators
used in C. These operators generate MASM compare, test, and conditional
jump instructions. High-level control instructions include:
Operator

Meaning

Equal

Not equal

Greater than

Greater than or equal to

Less than

Less than or equal to

Bit test

Logical NOT

Logical AND

Logical OR

A condition without operators (other than !) tests for nonzero as it does in C.
For example, .WHILE (x) is the same as .WHILE (x != 0), and .WHILE
(!x) is the same as .WHILE (x == 0).
You can also use the flag names (ZERO?, CARRY?, OVERFLOW?, SIGN?,
and PARITY?) as operands in conditions with the high-level control structures.
For example, in .WHILE (CARRY?), the value of the carry flag determines the
outcome of the condition.

Signed and Unsigned Operands
Expression operators generate unsigned jumps by default. However, if either
side of the operation is signed, the assembler considers the entire operation
signed.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 178 of 18 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

179

You can use the PTR operator to tell the assembler that a particular operand in
a register or constant is a signed number, as in these examples:
.WHILE
.IF

SWORD PTR [bx] <= 0
SWORD PTR mem1 > 0

Without the PTR operator, the assembler would treat the contents of BX as an
unsigned value.
You can also specify the size attributes of operands in memory locations with
SBYTE, SWORD, and SDWORD, for use with .IF, .WHILE, and .REPEAT.
mem1
mem2

.DATA
SBYTE
WORD
.IF
.WHILE
.WHILE

?
?
mem1 > 0
mem2 < bx
SWORD PTR ax < count

Precedence Level
As with C, you can concatenate conditions with the && operator for AND, the
|| operator for OR, and the ! operator for negate. The precedence level is !, &&,
and ||, with ! having the highest priority. Like expressions in high-level languages,
precedence is evaluated left to right.

Expression Evaluation
The assembler evaluates conditions created with high-level control structures
according to short-circuit evaluation. If the evaluation of a particular condition
automatically determines the final result (such as a condition that evaluates to
false in a compound statement concatenated with AND), the evaluation does not
continue.
For example, in this .WHILE statement,
.WHILE (ax > 0) && (WORD PTR [bx] == 0)

the assembler evaluates the first condition. If this condition is false (that is, if
AX is less than or equal to 0), the evaluation is finished. The second condition is
not checked and the loop does not execute, because a compound condition
containing && requires both expressions to be true for the entire condition to be
true.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 179 of 19 Printed: 10/02/00 04:23 PM

180

Programmer’s Guide

Procedures
Organizing your code into procedures that execute specific tasks divides large
programs into manageable units, allows for separate testing, and makes code
more efficient for repetitive tasks.
Assembly-language procedures are similar to functions, subroutines, and
procedures in high-level languages such as C, FORTRAN, and Pascal. Two
instructions control the use of assembly-language procedures. CALL pushes the
return address onto the stack and transfers control to a procedure, and RET
pops the return address off the stack and returns control to that location.
The PROC and ENDP directives mark the beginning and end of a procedure.
Additionally, PROC can automatically:
u

Preserve register values that should not change but that the procedure might
otherwise alter.
Set up a local stack pointer, so that you can access parameters and local
variables placed on the stack.
Adjust the stack when the procedure ends.

Defining Procedures
Procedures require a label at the start of the procedure and a RET instruction at
the end. Procedures are normally defined by using the PROC directive at the
start of the procedure and the ENDP directive at the end. The RET instruction
normally is placed immediately before the ENDP directive. The assembler
makes sure the distance of the RET instruction matches the distance defined by
the PROC directive. The basic syntax for PROC is:
label PROC [[NEAR | FAR]]
.
.
.
RET [[constant]]
label ENDP
The CALL instruction pushes the address of the next instruction in your code
onto the stack and passes control to a specified address. The syntax is:
CALL {label | register | memory}
The operand contains a value calculated at run time. Since that operand can be a
register, direct memory operand, or indirect memory operand, you can write call
tables similar to the example code on page 164.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 180 of 20 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

181

Calls can be near or far. Near calls push only the offset portion of the calling
address and therefore must target a procedure within the same segment or
group. You can specify the type for the target operand. If you do not, MASM
uses the declared distance (NEAR or FAR) for operands that are labels and for
the size of register or memory operands. The assembler then encodes the call
appropriately, as it does with unconditional jumps. (See previous “Unconditional
Jumps” and “Conditional Jumps.”)
MASM optimizes a call to a far non-external label when the label is in the
current segment by generating the code for a near call, saving one byte.
You can define procedures without PROC and ENDP, but if you do, you must
make sure that the size of the CALL matches the size of the RET. You can
specify the RET instruction as RETN (Return Near) or RETF (Return Far) to
override the default size:
call
.
.
.

NEAR PTR task

task:

; Call is declared near
; Return comes to here

; Procedure begins with near label
.
.
.
retn

; Instructions go here
; Return declared near

The syntax for RETN and RETF is:
label: | label LABEL NEAR
statements
RETN [[constant]]
label LABEL FAR
statements
RETF [[constant]]
The RET instruction (and its RETF and RETN variations) allows an optional
constant operand that specifies a number of bytes to be added to the value of
the SP register after the return. This operand adjusts for arguments passed to
the procedure before the call, as shown in the example in “Using Local
Variables,” following.
When you define procedures without PROC and ENDP, you must make sure
that calls have the same size as corresponding returns. For example, RETF pops
two words off the stack. If a NEAR call is made to a procedure with a far
return, the popped value is meaningless, and the stack status may cause the
execution to return to a random memory location, resulting in program failure.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 181 of 21 Printed: 10/02/00 04:23 PM

182

Programmer’s Guide

An extended PROC syntax automates many of the details of accessing
arguments and saving registers. See “Declaring Parameters with the PROC
Directive,” later in this chapter.

Passing Arguments on the Stack
Each time you call a procedure, you may want it to operate on different data.
This data, called “arguments,” can be passed to the procedure in various ways.
Although you can pass arguments to a procedure in registers or in variables, the
most common method is the stack. Microsoft languages have specific
conventions for passing arguments. These conventions for assembly-language
modules shared with modules from high-level languages are explained in Chapter
12, “Mixed-Language Programming.”
This section describes how a procedure accesses the arguments passed to it on
the stack. Each argument is accessed as an offset from BP. However, if you use
the PROC directive to declare parameters, the assembler calculates these offsets
for you and lets you refer to parameters by name. The next section, “Declaring
Parameters with the PROC Directive,” explains how to use PROC this way.
This example shows how to pass arguments to a procedure. The procedure
expects to find those arguments on the stack. As this example shows, arguments
must be accessed as offsets of BP.
; C-style procedure call and definition

addup

mov
push
push
push
call
add
.
.
.
PROC

ax, 10
ax
arg2
cx
addup
sp, 6

;
;
;
;
;
;
;

Load and
push constant as third argument
Push memory as second argument
Push register as first argument
Call the procedure
Destroy the pushed arguments
(equivalent to three pops)

NEAR

push

mov
mov

bp, sp
ax, [bp+4]

add

ax, [bp+6]

add

ax, [bp+8]

pop
ret
ENDP

;
;
;
;
;
;
;
;
;
;
;
;
;

Return address for near call
takes two bytes
Save base pointer - takes two bytes
so arguments start at fourth byte
Load stack into base pointer
Get first argument from
fourth byte above pointer
Add second argument from
sixth byte above pointer
Add third argument from
eighth byte above pointer
Restore BP
Return result in AX

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 182 of 22 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

Figure 7.1 shows the stack condition at key points in the process.

Figure 7.1

Procedure Arguments on the Stack

Starting with the 80186 processor, the ENTER and LEAVE instructions
simplify the stack setup and restore instructions at the beginning and end of
procedures. However, ENTER uses a lot of time. It is necessary only with
nested, statically-scoped procedures. Thus, a Pascal compiler may sometimes
generate ENTER. The LEAVE instruction, on the other hand, is an efficient
way to do the stack cleanup. LEAVE reverses the effect of the last ENTER
instruction by restoring BP and SP to their values before the procedure call.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 183 of 23 Printed: 10/02/00 04:23 PM

183

184

Programmer’s Guide

Declaring Parameters with the PROC Directive
With the PROC directive, you can specify registers to be saved, define parameters to the procedure, and assign symbol names to parameters (rather than as
offsets from BP). This section describes how to use the PROC directive to
automate the parameter-accessing techniques described in the last section.
For example, the following diagram shows a valid PROC statement for a
procedure called from C. It takes two parameters, var1 and arg1, and uses
(and must save) the DI and SI registers:

The syntax for PROC is:
label PROC [[attributes]] [[USES reglist]] [[, ]] [[parameter[[:tag]]... ]]
The parts of the PROC directive include:
Argument

Description

label

The name of the procedure.

attributes

Any of several attributes of the procedure, including the distance, langtype, and
visibility of the procedure. The syntax for attributes is given on the following
page.

reglist

A list of registers following the USES keyword that the procedure uses, and that
should be saved on entry. Registers in the list must be separated by blanks or
tabs, not by commas. The assembler generates prologue code to push these
registers onto the stack. When you exit, the assembler generates epilogue code
to pop the saved register values off the stack.

parameter

The list of parameters passed to the procedure on the stack. The list can have a
variable number of parameters. See the discussion following for the syntax of
parameter. This list can be longer than one line if the continued line ends with a
comma.

This diagram shows a valid PROC definition that uses several attributes:

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 184 of 24 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

185

Attributes
The syntax for the attributes field is:
[[distance]] [[langtype]] [[visibility]] [[]]
The explanations for these options include:
Argument

Description

distance

Controls the form of the RET instruction generated. Can be NEAR or FAR. If
distance is not specified, it is determined from the model declared with the
.MODEL directive. NEAR distance is assumed for TINY, SMALL,
COMPACT, and FLAT. The assembler assumes FAR distance for MEDIUM ,
LARGE , and HUGE . For 80386/486 programming with 16- and 32-bit
segments, you can specify NEAR16, NEAR32, FAR16, or FAR32.

langtype

Determines the calling convention used to access parameters and restore the stack.
The BASIC, FORTRAN, and PASCAL langtypes convert procedure names to
uppercase, place the last parameter in the parameter list lowest on the stack, and
generate a RET num instruction to end the procedure. The RET adjusts the stack
upward by num, which represents the number of bytes in the argument list. This
step, called “cleaning the stack,” returns the stack pointer SP to the value it had
before the caller pushed any arguments.
The C and STDCALL langtype prefixes an underscore to the procedure name
when the procedure’s scope is PUBLIC or EXPORT and places the first
parameter lowest on the stack. SYSCALL is equivalent to the C calling convention
with no underscore prefixed to the procedure’s name. STDCALL uses caller
stack cleanup when :VARARG is specified; otherwise the called routine must
clean up the stack (see Chapter 12).

visibility

Indicates whether the procedure is available to other modules. The visibility can
be PRIVATE, PUBLIC, or EXPORT. A procedure name is PUBLIC unless it
is explicitly declared as PRIVATE. If the visibility is EXPORT, the linker places
the procedure’s name in the export table for segmented executables. EXPORT
also enables PUBLIC visibility.
You can explicitly set the default visibility with the OPTION directive. OPTION
PROC:PUBLIC sets the default to public. For more information, see Chapter 1,
“Using the Option Directive.”

prologuearg

Specifies the arguments that affect the generation of prologue and epilogue code
(the code MASM generates when it encounters a PROC directive or the end of a
procedure). For an explanation of prologue and epilogue code, see “Generating
Prologue and Epilogue Code,” later in this chapter.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 185 of 25 Printed: 10/02/00 04:23 PM

186

Programmer’s Guide

Parameters
The comma that separates parameters from reglist is optional, if both fields
appear on the same line. If parameters appears on a separate line, you must end
the reglist field with a comma. In the syntax:
parmname [[:tag]
parmname is the name of the parameter. The tag can be the qualifiedtype or
the keyword VARARG. However, only the last parameter in a list of parameters can use the VARARG keyword. The qualifiedtype is discussed in “Data
Types,” Chapter 1. An example showing how to reference VARARG parameters appears later in this section. You can nest procedures if they do not have
parameters or USES register lists. This diagram shows a procedure definition
with one parameter definition.

The procedure presented in “Passing Arguments on the Stack,” page 182, is
here rewritten using the extended PROC functionality. Prior to the procedure
call, you must push the arguments onto the stack unless you use INVOKE.
(See “Calling Procedures with INVOKE,” later in this chapter.)
addup

addup

PROC NEAR C,
arg1:WORD, arg2:WORD, count:WORD
mov
ax, arg1
add
ax, count
add
ax, arg2
ret
ENDP

If the arguments for a procedure are pointers, the assembler does not generate
any code to get the value or values that the pointers reference; your program
must still explicitly treat the argument as a pointer. (For more information about
using pointers, see Chapter 3, “Using Addresses and Pointers.”)

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 186 of 26 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

187

In the following example, even though the procedure declares the parameters as
near pointers, you must code two MOV instructions to get the values of the
parameters. The first MOV gets the address of the parameters, and the second MOV
gets the parameter.
; Call from C as a FUNCTION returning an integer

myadd

.MODEL medium, c
.CODE
PROC
arg1:NEAR PTR WORD, arg2:NEAR PTR WORD
mov
mov
mov
add

bx,
ax,
bx,
ax,

arg1
[bx]
arg2
[bx]

; Load first argument
; Add second argument

ret
myadd

ENDP

You can use conditional-assembly directives to make sure your pointer
parameters are loaded correctly for the memory model. For example, the
following version of myadd treats the parameters as FAR parameters, if
necessary.

myadd

.MODEL
.CODE
PROC

medium, c

IF
les
mov
les
add
ELSE
mov
mov
mov
add
ENDIF

@DataSize
bx, arg1
ax, es:[bx]
bx, arg2
ax, es:[bx]

arg1:PTR WORD,

bx,
ax,
bx,
ax,

ret
ENDP

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 187 of 27 Printed: 10/02/00 04:23 PM

arg1
[bx]
arg2
[bx]

; Could be any model
arg2:PTR WORD

; Far parameters

; Near parameters

188

Programmer’s Guide

Using VARARG
In the PROC statement, you can append the :VARARG keyword to the last
parameter to indicate that the procedure accepts a variable number of
arguments. However, :VARARG applies only to the C, SYSCALL, or
STDCALL calling conventions (see Chapter 12). A symbol must precede
:VARARG so the procedure can access arguments as offsets from the given
variable name, as this example illustrates:
addup3

addup3

PROTO NEAR C, argcount:WORD, arg1:VARARG
invoke

addup3, 3, 5, 2, 4

PROC
sub
sub

NEAR C, argcount:WORD, arg1:VARARG
ax, ax
; Clear work register
si, si

.WHILE
add
dec
inc
inc
.ENDW

argcount > 0
ax, arg1[si]
argcount
si
si

ret
ENDP

; Argcount has number of arguments
; Arg1 has the first argument
; Point to next argument

; Total is in AX

You can pass non-default-sized pointers in the VARARG portion of the
parameter list by separately passing the segment portion and the offset portion
of the address.
Note When you use the extended PROC features and the assembler encounters
a RET instruction, it automatically generates instructions to pop saved registers,
remove local variables from the stack, and, if necessary, remove parameters. It
generates this code for each RET instruction it encounters. You can reduce code
size by having only one return and jumping to it from various locations.

Using Local Variables
In high-level languages, local variables are visible only within a procedure. In
Microsoft languages, these variables are usually stored on the stack. In
assembly-language programs, you can also have local variables. These variables
should not be confused with labels or variable names that are local to a module,
as described in Chapter 8, “Sharing Data and Procedures Among Modules and
Libraries.”

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 188 of 28 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

189

This section outlines the standard methods for creating local variables. The next
section shows how to use the LOCAL directive to make the assembler

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 189 of 29 Printed: 10/02/00 04:23 PM

190

Programmer’s Guide

automatically generate local variables. When you use this directive, the
assembler generates the same instructions as those demonstrated in this section
but handles some of the details for you.
If your procedure has relatively few variables, you can usually write the most
efficient code by placing these values in registers. Use local (stack) data when
you have a large amount of temporary data for the procedure.
To use a local variable, you must save stack space for it at the start of the
procedure. A procedure can then reference the variable by its position in the
stack. At the end of the procedure, you must clean the stack by restoring the
stack pointer. This effectively throws away all local variables and regains the
stack space they occupied.
This example subtracts 2 bytes from the SP register to make room for a local
word variable, then accesses the variable as [bp-2].

task

push
call
.
.
.

ax
task

; Push one argument
; Call

PROC
push
mov
sub
.
.
.
mov
add
sub
.
.
.
mov
pop
ret
ENDP

NEAR
bp
bp, sp
sp, 2

; Save base pointer
; Load stack into base pointer
; Save two bytes for local variable

WORD PTR [bp-2], 3 ; Initialize local variable
ax, [bp-2]
; Add local variable to AX
[bp+4], ax
; Subtract local from argument
; Use [bp-2] and [bp+4] in
;
other operations
sp, bp
bp
2

; Clear local variables
; Restore base
; Return result in AX and pop
;
two bytes to clear parameter

Notice the instruction mov sp,bp at the end of the procedure restores the
original value of SP. The statement is required only if the value of SP changes
inside the procedure (usually by allocating local variables). The argument passed
to the procedure is removed with the RET instruction. Contrast this to the
example in “Passing Arguments on the Stack,” page 182, in which the calling
code adjusts the stack for the argument.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 190 of 30 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

191

Figure 7.2 shows the stack at key points in the process.

Figure 7.2

Local Variables on the Stack

Creating Local Variables Automatically
MASM’s LOCAL directive automates the process for creating local variables
on the stack. LOCAL frees you from having to count stack words, and it
makes your code easier to write and maintain. This section illustrates the
advantages of creating temporary data with the LOCAL directive.
To use the LOCAL directive, list the variables you want to create, giving a type
for each one. The assembler calculates how much space is required on the
stack. It also generates instructions to properly decrement SP (as described in
the previous section) and to reset SP when you return from the procedure.
When you create local variables this way, your source code can refer to each
local variable by name rather than as an offset of the stack pointer. Moreover,

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 191 of 31 Printed: 10/02/00 04:23 PM

192

Programmer’s Guide

the assembler generates debugging information for each local variable. If you
have programmed before in a high-level language that allows scoping, local
variables will seem familiar. For example, a C compiler sets up variables with
automatic storage class in the same way as the LOCAL directive.
We can simplify the procedure in the previous section with the following code:
task

task

PROC
LOCAL
.
.
.
mov
add
sub
.
.
.
ret
ENDP

NEAR
arg:WORD
loc:WORD

loc, 3
ax, loc
arg, ax

;
;
;
;

Initialize local variable
Add local variable to AX
Subtract local from argument
Use "loc" and "arg" in other operations

The LOCAL directive must be on the line immediately following the PROC
statement with the following syntax:
LOCAL vardef [[, vardef]]...
Each vardef defines a local variable. A local variable definition has this form:
label[[[count]]][[:qualifiedtype]]
These are the parameters in local variable definitions:
Argument

Description

label

The name given to the local variable. You can use this name to access the
variable.

count

The number of elements of this name and type to allocate on the stack. You
can allocate a simple array on the stack with count. The brackets around
count are required. If this field is omitted, one data object is assumed.

qualifiedtype

A simple MASM type or a type defined with other types and attributes. For
more information, see “Data Types” in Chapter 1.

If the number of local variables exceeds one line, you can place a comma at the
end of the first line and continue the list on the next line. Alternatively, you can
use several consecutive LOCAL directives.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 192 of 32 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

193

The assembler does not initialize local variables. Your program must include
code to perform any necessary initializations. For example, the following code
fragment sets up a local array and initializes it to zero:
arraysz EQU

aproc

PROC
USES di
LOCAL
var1[arraysz]:WORD, var2:WORD
.
.
.
; Initialize local array to zero
push
ss
pop
es
; Set ES=SS
lea
di, var1
; ES:DI now points to array
mov
cx, arraysz
; Load count
sub
ax, ax
rep
stosw
; Store zeros
; Use the array...
.
.
.
ret
aproc
ENDP

Even though you can reference stack variables by name, the assembler treats
them as offsets of BP, and they are not visible outside the procedure. In the
following procedure, array is a local variable.
index
test
LOCAL

;

EQU
10
PROC NEAR
array[index]:WORD
.
.
.
mov
bx, index
mov
array[bx], 5

; Not legal!

The second MOV statement may appear to be legal, but since array is an
offset of BP, this statement is the same as
;

mov [bp + bx + arrayoffset], 5

; Not legal!

BP and BX can be added only to SI and DI. This example would be legal,
however, if the index value were moved to SI or DI. This type of error in your
program can be difficult to find unless you keep in mind that local variables in
procedures are offsets of BP.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 193 of 33 Printed: 10/02/00 04:23 PM

194

Programmer’s Guide

Declaring Procedure Prototypes
MASM provides the INVOKE directive to handle many of the details important
to procedure calls, such as pushing parameters according to the correct calling
conventions. To use INVOKE, the procedure called must have been declared
previously with a PROC statement, an EXTERNDEF (or EXTERN) statement,
or a TYPEDEF. You can also place a prototype defined with PROTO before
the INVOKE if the procedure type does not appear before the INVOKE.
Procedure prototypes defined with PROTO inform the assembler of types and
numbers of arguments so the assembler can check for errors and provide
automatic conversions when INVOKE calls the procedure.
Declaring procedure prototypes is good programming practice, but is optional.
Prototypes in MASM perform the same function as prototypes in C and other
high-level languages. A procedure prototype includes the procedure name, the
types, and (optionally) the names of all parameters the procedure expects.
Prototypes usually are placed at the beginning of an assembly program or in a
separate include file so the assembler encounters the prototype before the actual
procedure.
Prototypes enable the assembler to check for unmatched parameters and are
especially useful for procedures called from other modules and other languages.
If you write routines for a library, you may want to put prototypes into an
include file for all the procedures used in that library. For more information
about using include files, see Chapter 8, “Sharing Data and Procedures among
Modules and Libraries.”
The PROTO directive provides one way to define a procedure prototype. The
syntax for a prototype definition is the same as for a procedure declaration (see
“Declaring Parameters with the PROC Directive,” earlier in this chapter), except
that you do not include the list of registers, prologuearg list, or the scope of the
procedure.
Also, the PROTO keyword precedes the langtype and distance attributes. The
attributes (like C and FAR) are optional. However, if they are not specified, the
defaults are based on any .MODEL or OPTION LANGUAGE statement. The
names of the parameters are also optional, but you must list parameter types. A
label preceding :VARARG is also optional in the prototype but not in the
PROC statement.
If a PROTO and a PROC for the same function appear in the same module,
they must match in attribute, number of parameters, and parameter types. The
easiest way to create prototypes with PROTO is to write your procedure and
then copy the first line (the line that contains the PROC keyword) to a location
in your program that follows the data declarations. Change PROC to PROTO
and remove the USES reglist, the prologuearg field, and the visibility field. It

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 194 of 34 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

195

is important that the prototype follow the declarations for any types used in it to
avoid any forward references used by the parameters in the prototype.
The following example illustrates how to define and then declare two typical
procedures. In both prototype and declaration, the comma before the argument
list is optional only when the list does not appear on a separate line:
; Procedure prototypes.
addup
myproc

PROTO NEAR C argcount:WORD, arg2:WORD, arg3:WORD
PROTO FAR C, argcount:WORD, arg2:VARARG

; Procedure declarations
addup
.
.
.
myproc

PROC NEAR C, argcount:WORD, arg2:WORD, arg3:WORD

PROC FAR C PUBLIC USES di si,
argcount:WORD,
arg2:VARARG

When you call a procedure with INVOKE, the assembler checks the arguments
given by INVOKE against the parameters expected by the procedure. If the
data types of the arguments do not match, MASM reports an error or converts
the type to the expected type. These conversions are explained in the next
section.

Calling Procedures with INVOKE
INVOKE generates a sequence of instructions that push arguments and call a
procedure. This helps maintain code if arguments or langtype for a procedure
are changed. INVOKE generates procedure calls and automatically:
u
u
u

Converts arguments to the expected types.
Pushes arguments on the stack in the correct order.
Cleans the stack when the procedure returns.

If arguments do not match in number or if the type is not one the assembler can
convert, an error results.
If the procedure uses VARARG, INVOKE can pass a number of arguments
different from the number in the parameter list without generating an error or
warning. Any additional arguments must be at the end of the INVOKE
argument list. All other arguments must match those in the prototype parameter
list.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 195 of 35 Printed: 10/02/00 04:23 PM

196

Programmer’s Guide

The syntax for INVOKE is:
INVOKE expression [[, arguments]]
where expression can be the procedure’s label or an indirect reference to a
procedure, and arguments can be an expression, a register pair, or an expression
preceded with ADDR. (The ADDR operator is discussed later in this chapter.)
Procedures with these prototypes
addup
myproc

PROTO NEAR C argcount:WORD, arg2:WORD, arg3:WORD
PROTO FAR C, argcount:WORD, arg2:VARARG

and these procedure declarations
addup
.
.
.
myproc

PROC NEAR C, argcount:WORD, arg2:WORD, arg3:WORD

PROC FAR C PUBLIC USES di si,
argcount:WORD,
arg2:VARARG

can be called with INVOKE statements like this:
INVOKE
INVOKE

addup,
myproc,

ax, x, y
bx, cx, 100, 10

The assembler can convert some arguments and parameter type combinations
so that the correct type can be passed. The signed or unsigned qualities of the
arguments in the INVOKE statements determine how the assembler converts
them to the types expected by the procedure.
The addup procedure, for example, expects parameters of type WORD, but
the arguments passed by INVOKE to the addup procedure can be any of these
types:
u
u

u
u
u

BYTE, SBYTE, WORD, or SWORD
An expression whose type is specified with the PTR operator to be one of
those types
An 8-bit or 16-bit register
An immediate expression in the range –32K to +64K
A NEAR PTR

If the type is smaller than that expected by the procedure, MASM widens the
argument to match.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 196 of 36 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

197

Widening Arguments
For INVOKE to correctly handle type conversions, you must use the signed
data types for any signed assignments. MASM widens an argument to match the
type expected by a procedure’s parameters in these cases:
Type Passed

Type Expected

BYTE, SBYTE

WORD, SWORD, DWORD, SDWORD

WORD, SWORD

DWORD, SDWORD

The assembler can extend a segment if far data is expected, and it can convert
the type given in the list to the types expected. If the assembler cannot convert
the type, however, it generates an error.

Detecting Errors
If the assembler needs to widen an argument, it first copies the value to AL or
AX. It widens an unsigned value by placing a zero in the higher register area,
and widens a signed value with a CBW, CWD, or CWDE instruction as
required. Similarly, the assembler copies a constant argument value into AL or
AX when the .8086 directive is in effect. You can see these generated
instructions in the listing file when you include the /Sg command-line option.
Using the accumulator register to widen or copy an argument may lead to an
error if you attempt to pass AX as another argument. For example, consider the
following INVOKE statement for a procedure with the C calling convention
INVOKE

myprocA, ax, cx, 100, arg

where arg is a BYTE variable and myproc expects four arguments of type
WORD. The assembler widens and then pushes arg like this:
mov
xor
push

al, DGROUP:arg
ah, ah
ax

The generated code thus overwrites the last argument (AX) passed to the
procedure. The assembler generates an error in this case, requiring you to
rewrite the INVOKE statement.
To summarize, the INVOKE directive overwrites AX and perhaps DX when
widening arguments. It also uses AX to push constants on the 8088 and 8086. If
you use these registers (or EAX and EDX on an 80386/486) to pass arguments,
they may be overwritten. The assembler’s error detection prevents this from
ever becoming a run-time bug, but AX and DX should remain your last choice
for holding arguments.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 197 of 37 Printed: 10/02/00 04:23 PM

198

Programmer’s Guide

Invoking Far Addresses
You can pass a FAR pointer in a segment::offset pair, as shown in the following.
Note the use of double colons to separate the register pair. The registers could
be any other register pair, including a pair that an MS-DOS call uses to return
values.
FPWORD
TYPEDEF FAR PTR WORD
SomeProc PROTO var1:DWORD, var2:WORD, var3:WORD
pfaritem
.
.
.
les
INVOKE

FPWORD

faritem

bx, pfaritem
SomeProc, ES::BX, arg1, arg2

However, INVOKE cannot combine into a single address one argument for the
segment and one for the offset.

Passing an Address
You can use the ADDR operator to pass the address of an expression to a
procedure that expects a NEAR or FAR pointer. This example generates code
to pass a far pointer (to arg1) to the procedure proc1.
PBYTE
arg1
proc1

INVOKE

TYPEDEF FAR PTR BYTE
BYTE
"This is a string"
PROTO
NEAR C fparg:PBYTE
.
.
.
proc1, ADDR arg1

For information on defining pointers with TYPEDEF, see “Defining Pointer
Types with TYPEDEF” in Chapter 3.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 198 of 38 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

199

Invoking Procedures Indirectly
You can make an indirect procedure call such as call [bx + si] by using a
pointer to a function prototype with TYPEDEF, as shown in this example:
FUNCPROTO
FUNCPTR

pfunc

TYPEDEF PROTO NEAR ARG1:WORD
TYPEDEF PTR FUNCPROTO

.DATA
FUNCPTR OFFSET proc1, OFFSET proc2
.CODE
.
.
.
mov
mov
INVOKE

bx, OFFSET pfunc
; BX points to table
si, Num
; Num contains 0 or 2
FUNCPTR PTR [bx+si], arg1 ; Call proc1 if Num=0
;
or proc2 if Num=2

You can also use ASSUME to accomplish the same task. The following
ASSUME statement associates the type FUNCPTR with the BX register.
ASSUME
mov
mov
INVOKE

BX:FUNCPTR
bx, OFFSET pfunc
si, Num
[bx+si], arg1

Checking the Code Generated
Code generated by the INVOKE directive may vary depending on the processor
mode and calling conventions in effect. You can check your listing files to see
the code generated by the INVOKE directive if you use the /Sg command-line
option.

Generating Prologue and Epilogue Code
When you use the PROC directive with its extended syntax and argument list,
the assembler automatically generates the prologue and epilogue code in your
procedure. “Prologue code” is generated at the start of the procedure. It sets up
a stack pointer so you can access parameters from within the procedure. It also
saves space on the stack for local variables, initializes registers such as DS, and
pushes registers that the procedure uses. Similarly, “epilogue code” is the code
at the end of the procedure that pops registers and returns from the procedure.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 199 of 39 Printed: 10/02/00 04:23 PM

200

Programmer’s Guide

The assembler automatically generates the prologue code when it encounters the
first instruction or label after the PROC directive. This means you cannot label
the prologue for the purpose of jumping to it. The assembler generates the
epilogue code when it encounters a RET or IRET instruction. Using the
assembler-generated prologue and epilogue code saves time and decreases the
number of repetitive lines of code in your procedures.
The generated prologue or epilogue code depends on the:
u

Local variables defined.

Arguments passed to the procedure.

Current processor selected (affects epilogue code only).
Current calling convention.
Options passed in the prologuearg of the PROC directive.
Registers being saved.

u
u
u

The prologuearg list contains options specifying how to generate the prologue
or epilogue code. The next section explains how to use these options, gives the
standard prologue and epilogue code, and explains the techniques for defining
your own prologue and epilogue code.

Using Automatic Prologue and Epilogue Code
The standard prologue and epilogue code handles parameters and local
variables. If a procedure does not have any parameters or local variables, the
prologue and epilogue code that sets up and restores a stack pointer is omitted,
unless
FORCEFRAME is included in the prologuearg list. (FORCEFRAME is
discussed later in this section.) Prologue and epilogue code also generates a push
and pop for each register in the register list.
The prologue code consists of three steps:
1. Point BP to top of stack.
2. Make space on stack for local variables.
3. Save registers the procedure must preserve.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 200 of 40 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

201

The epilogue cancels these three steps in reverse order, then cleans the stack, if
necessary, with a RET num instruction. For example, the procedure declaration
myproc

PROC NEAR PASCAL USES di si,
arg1:WORD, arg2:WORD, arg3:WORD
LOCAL local1:WORD, local2:WORD

generates the following prologue code:
push
mov
sub
push
push

bp
bp, sp
sp, 4
di
si

; Step 1:
;
point BP to stack top
; Step 2: space for 2 local words
; Step 3:
;
save registers listed in USES

The corresponding epilogue code looks like this:
pop
pop
mov
pop
ret

si
di
sp, bp
bp
6

; Undo Step 3
; Undo Step 2
; Undo Step 1
; Clean stack of pushed arguments

Notice the RET 6 instruction cleans the stack of the three word-sized
arguments. The instruction appears in the epilogue because the procedure does
not use the C calling convention. If myproc used C conventions, the epilogue
would end with a RET instruction without an operand.
The assembler generates standard epilogue code when it encounters a RET
instruction without an operand. It does not generate an epilogue if RET has a
nonzero operand. To suppress generation of a standard epilogue, use RETN or
RETF with or without an operand, or use RET 0.
The standard prologue and epilogue code recognizes two operands passed in the
prologuearg list, LOADDS and FORCEFRAME. These operands modify the
prologue code. Specifying LOADDS saves and initializes DS. Specifying
FORCEFRAME as an argument generates a stack frame even if no arguments
are sent to the procedure and no local variables are declared. If your procedure
has any parameters or locals, you do not need to specify FORCEFRAME.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 201 of 41 Printed: 10/02/00 04:23 PM

202

Programmer’s Guide

For example, adding LOADDS to the argument list for myproc creates this
prologue:
push
mov
sub
push
mov
mov
push
push

bp
bp,
sp,
ds
ax,
ds,
di
si

sp
4
DGROUP
ax

;
;
;
;
;
;
;
;

Step 1:
point BP to stack top
Step 2: space for 2 locals
Save DS and point it
to DGROUP, as
instructed by LOADDS
Step 3:
save registers listed in USES

The epilogue code restores DS:
pop
pop
pop
mov
pop
ret

si
di
ds
sp, bp
bp
6

; Undo Step 3
;
;
;
;

Restore DS
Undo Step 2
Undo Step 1
Clean stack of pushed arguments

User-Defined Prologue and Epilogue Code
If you want a different set of instructions for prologue and epilogue code in your
procedures, you can write macros that run in place of the standard prologue and
epilogue code. For example, while you are debugging your procedures, you may
want to include a stack check or track the number of times a procedure is called.
You can write your own prologue code to do these things whenever a procedure
executes. Different prologue code may also be necessary if you are writing
applications for Windows. User-defined prologue macros will respond correctly
if you specify FORCEFRAME in the prologuearg of a procedure.
To write your own prologue or epilogue code, the OPTION directive must
appear in your program. It disables automatic prologue and epilogue code
generation. When you specify
OPTION PROLOGUE : macroname
OPTION EPILOGUE : macroname
the assembler calls the macro specified in the OPTION directive instead of
generating the standard prologue and epilogue code. The prologue macro must
be a macro function, and the epilogue macro must be a macro procedure.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 202 of 42 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

203

The assembler expects your prologue or epilogue macro to have this form:
macroname MACRO procname, \
flag, \
parmbytes, \
localbytes, \
, \
userparms
Your macro must have formal parameters to match all the actual arguments
passed. The arguments passed to your macro include:
Argument

Description

procname

The name of the procedure.

flag

A 16-bit flag containing the following information:
Bit = Value

Description

Bit 0, 1, 2

For calling conventions (000=unspecified language type,
001=C, 010=SYSCALL, 011=STDCALL,
100=PASCAL, 101=FORTRAN, 110=BASIC).

Bit 3

Undefined (not necessarily zero).

Bit 4

Set if the caller restores the stack
(use RET, not RETn).

Bit 5

Set if procedure is FAR.

Bit 6

Set if procedure is PRIVATE.

Bit 7

Set if procedure is EXPORT.

Bit 8

Set if the epilogue is generated as a result of an IRET
instruction and cleared if the epilogue is generated as a
result of a RET instruction.

Bits 9–15

Undefined (not necessarily zero).

parmbytes

The accumulated count in bytes of all parameters given in the PROC
statement.

localbytes

The count in bytes of all locals defined with the LOCAL directive.

reglist

A list of the registers following the USES operator in the procedure
declaration. Enclose this list with angle brackets (< >) and separate each item
with commas. Reverse the list for epilogues.

userparms

Any argument you want to pass to the macro. The prologuearg (if there is
one) specified in the PROC directive is passed to this argument.

Your macro function must return the parmbytes parameter. However, if the
prologue places other values on the stack after pushing BP and these values are
not referenced by any of the local variables, the exit value must be the number
of bytes for procedure locals plus any space between BP and the locals.
Therefore, parmbytes is not always equal to the bytes occupied by the locals.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 203 of 43 Printed: 10/02/00 04:23 PM

204

Programmer’s Guide

The following macro is an example of a user-defined prologue that counts the
number of times a procedure is called.
ProfilePro

MACRO procname,
flag,
bytecount,
numlocals,
regs,
macroargs

inc

.DATA
WORD 0
.CODE
procname&count

push
mov

bp
bp, sp

procname&count

\
\
\
\
\

; Accumulates count of times the
;
procedure is called

; Other BP operations
IFNB
FOR r, regs
push r
ENDM
ENDIF
EXITM %bytecount
ENDM

Your program must also include this statement before calling any procedures
that use the prologue:
OPTION PROLOGUE:ProfilePro

If you define either a prologue or an epilogue macro, the assembler uses the
standard prologue or epilogue code for the one you do not define. The form of
the code generated depends on the .MODEL and PROC options used.
If you want to revert to the standard prologue or epilogue code, use
PROLOGUEDEF or EPILOGUEDEF as the macroname in the OPTION
statement.
OPTION EPILOGUE:EPILOGUEDEF

You can completely suppress prologue or epilogue generation with
OPTION PROLOGUE:None
OPTION EPILOGUE:None

In this case, no user-defined macro is called, and the assembler does not
generate a default code sequence. This state remains in effect until the next
OPTION
PROLOGUE or OPTION EPILOGUE is encountered.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 204 of 44 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

205

For additional information about writing macros, see Chapter 9, “Using
Macros.” The PROLOGUE.INC file provided in the MASM 6.1 distribution
disks can create the prologue and epilogue sequences for the Microsoft C
professional development system.

MS-DOS Interrupts
In addition to jumps, loops, and procedures that alter program execution,
interrupt routines transfer execution to a different location. In this case, control
goes to an interrupt routine.
You can write your own interrupt routines, either to replace an existing routine
or to use an undefined interrupt number. For example, you may want to replace
an MS-DOS interrupt handler, such as the Critical Error (Interrup 24h) and
CONTROL+C (Interrupt 23h) handlers. The BOUND instruction checks array
bounds and calls Interrupt 5 when an error occurs. If you use this instruction,
you need to write an interrupt handler for it.
This section summarizes the following:
u
u
u

How to call interrupts
How the processor handles interrupts
How to redefine an existing interrupt routine

The example routine in this section handles addition or multiplication overflow
and illustrates the steps necessary for writing an interrupt routine. For additional
information about MS-DOS and BIOS interrupts, see Chapter 11, “Writing
Memory-Resident Software.”

Calling MS-DOS and ROM-BIOS Interrupts
Interrupts provide a way to access MS-DOS and ROM-BIOS from assembly
language. They are called with the INT instruction, which takes an immediate
value between 0 and 255 as its only operand.
MS-DOS and ROM-BIOS interrupt routines accept data through registers. For
instance, most MS-DOS routines (and many BIOS routines) require a function
number in the AH register. Many handler routines also return values in registers.
To use an interrupt, you must know what data the handler routine expects and
what data, if any, it returns. For information, consult Help or one of the other
references mentioned in the Introduction.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 205 of 45 Printed: 10/02/00 04:23 PM

206

Programmer’s Guide

The following fragment illustrates a simple call to MS-DOS Function 9, which
displays the string msg on the screen:
msg

.DATA
BYTE
.CODE
mov
mov
mov
mov
int

"This writes to the screen$"
ax,
ds,
dx,
ah,
21h

SEG msg
ax
offset msg
09h

; Necessary only if DS does not
;
already point to data segment
; DS:DX points to msg
; Request Function 9

When the INT instruction executes, the processor:
1. Looks up the address of the interrupt routine in the Interrupt Vector Table.
This table starts at the lowest point in memory (segment 0, offset 0) and
consists of a series of far pointers called vectors. Each vector comprises a 4byte address (segment:offset) pointing to an interrupt handler routine. The
table sequence implies the number of the interrupt the vector references: the
first vector points to the Interrupt 0 handler, the second vector to the
Interrupt 1 handler, and so forth. Thus, the vector at 0000:i*4 holds the
address of the handler routine for Interrupt i.
2. Clears the trap flag (TF) and interrupt enable flag (IF).
3. Pushes the flags register, the current code segment (CS), and the current
instruction pointer (IP), in that order. (The current instruction is the one
following the INT statement.) As with a CALL, this ensures control returns
to the next logical position in the program.
4. Jumps to the address of the interrupt routine, as specified in the Interrupt
Vector Table.
5. Executes the code of the interrupt routine until it encounters an IRET
instruction.
6. Pops the instruction pointer, code segment, and flags.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 206 of 46 Printed: 10/02/00 04:23 PM

Chapter 7 Controlling Program Flow

207

Figure 7.3 illustrates how interrupts work.

Figure 7.3

Operation of Interrupts

Replacing an Interrupt Routine
To replace an existing interrupt routine, your program must:
u
u

Provide a new routine to handle the interrupt.
Replace the old routine’s address in the Interrupt Vector Table with the
address of your new routine.
Replace the old address back into the vector table before your program ends.

You can write an interrupt routine as a procedure by using the PROC and
ENDP directives. The routine should always be defined as FAR and should end
with an IRET instruction instead of a RET instruction.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 207 of 47 Printed: 10/02/00 04:23 PM

208

Programmer’s Guide

Note You can use the full extended PROC syntax (described in “Declaring
Parameters with the PROC Directive,” earlier in this chapter) to write interrupt
procedures. However, you should not make interrupt procedures NEAR or
specify arguments for them. You can use the USES keyword, however, to
correctly generate code to save and restore a register list in interrupt procedures.
The IRET instruction in MASM 6.1 has two forms that suppress epilogue code.
This allows an interrupt to have local variables or use a user-defined prologue.
IRETF pops a FAR16 return address, and IRETFD pops a FAR32 return
address.
The following example shows how to replace the handler for Interrupt 4. Once
registered in the Interrupt Vector Table, the new routine takes control when the
processor encounters either an INT 4 instruction or its special variation INTO
(Interrupt on Overflow). INTO is a conditional instruction that acts only when
the overflow flag is set. With INTO after a numerical calculation, your code can
automatically route control to a handler routine if the calculation results in a
numerical overflow. By default, the routine for Interrupt 4 simply consists of an
IRET, so it returns without doing anything. Using INTO is an alternative to
using JO (Jump on Overflow) to jump to another set of instructions.
The following example program first executes INT 21h to invoke MS-DOS
Function 35h (Get Interrupt Vector). This function returns the existing vector
for Interrupt 4. The program stores the vector, then invokes MS-DOS Function
25h (Set Interrupt Vector) to place the address of the ovrflow procedure in the
Interrupt Vector Table. From this point on, ovrflow gains control whenever
the processor executes INTO while the overflow flag is set. The new routine
displays a message and returns with AX and DX set to 0.
FPFUNC
msg
vector

.MODEL LARGE, C
TYPEDEF FAR PTR
.DATA
BYTE
"Overflow - result set to 0",13,10,'$'
FPFUNC ?
.CODE
.STARTUP
mov
int
mov
mov

ax, 3504h
21h
WORD PTR vector[2],es
WORD PTR vector[0],bx

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 208 of 48 Printed: 10/02/00 04:23 PM

; Load Interrupt 4 and call DOS
;
Get Interrupt Vector
; Save segment
;
and offset

Chapter 7 Controlling Program Flow
push
mov
mov
mov
mov
int
pop
.
.
.
add
into
.
.
.
lds
mov
int
mov
int

ds
ax,
ds,
dx,
ax,
21h
ds

ovrflow PROC
sti

FAR

mov
mov
int
sub
cwd
iret
ovrflow ENDP
END

ah, 09h
dx, OFFSET msg
21h
ax, ax

cs
ax
OFFSET ovrflow
2504h

209

; Save DS
; Load segment of new routine
; Load offset of new routine
; Load Interrupt 4 and call DOS
;
Set Interrupt Vector
; Restore

ax, bx

; Do arithmetic
; Call Interrupt 4 if overflow

dx, vector
ax, 2504h
21h
ax, 4C00h
21h

; Load original address
; Restore it to vector table
;
with DOS set vector function
; Terminate function

;
;
;
;
;
;
;
;

Enable interrupts
(turned off by INT)
Display string function
Load address
Call DOS
Set AX to 0
Set DX to 0
Return

Before the program ends, it again uses MS-DOS Function 25h to reset the
original Interrupt 4 vector back into the Interrupt Vector Table. This
reestablishes the original routine as the handler for Interrupt 4.
The first instruction of the ovrflow routine warrants further discussion. When
the processor encounters an INT instruction, it clears the interrupt flag before
branching to the specified interrupt handler routine. The interrupt flag serves a
crucial role in smoothing the processor’s tasks, but must not be abused. When
clear, the flag inhibits hardware interrupts such as the keyboard or system timer.
It should be left clear only briefly and only when absolutely necessary. Unless
you have a

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 209 of 49 Printed: 10/02/00 04:23 PM

210

Programmer’s Guide

compelling reason to leave the flag clear, always include an STI (Set Interrupt
Flag) instruction at the beginning of your interrupt handler routine to reenable
hardware interrupts.
CLI (Clear Interrupt Flag) and its corollary STI are designed to protect small
sections of time-dependent code from interruptions by the hardware. If you use
CLI in your program, be sure to include a matching STI instruction as well. The
sample interrupt handlers in Chapter 11, “Writing Memory-Resident Software,”
illustrate how to use these important instructions.

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 210 of 50 Printed: 10/02/00 04:23 PM

210

Programmer’s Guide

Filename: LMAPGC07.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 210 of 52 Printed: 10/02/00 04:23 PM

211

C H A P T E R

Sharing Data and Procedures
Among Modules and Libraries

To use symbols and procedures in more than one module, the assembler must
be able to recognize the shared data as global to all the modules where they are
used. MASM provides techniques to simplify data-sharing and give a high-level
interface to multiple-module programming. With these techniques, you can place
shared symbols in include files. This makes the data declarations in the file
available to all modules that use the include file.
This chapter explains the two data-sharing methods MASM 6.1 offers. The first
method simplifies data sharing between modules with include files. The second
does not involve include files. Instead, this method allows modules to share
procedures and data items using the PUBLIC and EXTERN directives.
The last section of this chapter explains how to create program libraries and
access their routines.

Selecting Data-Sharing Methods
If data defined in one module is to be used in other modules of a program, you
must declare the data public and external. MASM provides several ways to do
this:
Declare a symbol public with the PUBLIC directive in the module where it is
defined. This makes the symbol available to other modules. You must also
place an EXTERN statement for that symbol in all other modules that refer
to the public symbol. This statement informs the assembler that the symbol is
external — that is, defined in another module.
u
Declare the data communal with the COMM directive. However, communal
variables have limitations. You cannot depend on their location in memory
because they are allocated by the linker, and they cannot be initialized.
The EXTERNDEF directive declares a symbol either public or external, as
appropriate. EXTERNDEF simplifies the declarations for global (public and
external) variables and encourages the use of include files.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 211 of 1 Printed: 10/02/00 04:22 PM

212

Programmer’s Guide

The next section provides further details on using include files. For more
information on PUBLIC and EXTERN, see “Using Alternatives to Include
Files,” page 219.

Sharing Symbols with Include Files
Include files can contain any valid MASM statement, but typically consist of
type and symbol declarations. The assembler inserts the contents of the include
file into a module at the location of the INCLUDE directive. Include files are
optional, but can simplify project organization by eliminating the need to insert
common declarations into all modules of a program. An alternative to using
include files is described in “Using Alternatives to Include Files,” page 219.
This section explains how to organize symbol definitions and the declarations
that make them global (available to all modules); how to make both variables
and procedures public with EXTERNDEF, PROTO, and COMM.; and where
to place these directives in the modules and include files.

Organizing Modules
This section summarizes the organization of declarations and definitions in
modules and include files and the use of the INCLUDE directive.

Include Files
Type declarations that need to be identical in every module should be placed in
an include file. This ensures consistency and saves time when you update
programs. Include files should contain only symbol declarations and any other
declarations that are resolved at assembly time. (For a list of assembly-time
operations, see “Generating and Running Executable Programs” in Chapter 1.)
If more than one module accesses the include file, the file cannot contain
statements that define and allocate memory for symbols. Otherwise, the
assembler would attempt to allocate the same symbol more than once.
Note An include file used in two or more modules should not allocate data
variables.

Modules
An INCLUDE statement is usually placed before data and code segments in
your modules. When the assembler encounters an INCLUDE directive, it opens
the specified file and assembles all its statements. The assembler then returns to
the original module and continues the assembly.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 212 of 2 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

213

The INCLUDE directive takes the form:
INCLUDE filename
where filename is the full name of the include file. For example, the following
declaration inserts the contents of the include file SCREEN.INC in your
program:
INCLUDE SCREEN.INC

The filename in the INCLUDE directive must be fully specified; no extensions
are assumed. If a full pathname is not given, the assembler first searches the
directory of the source file containing the INCLUDE directive.
If the include file is not in the source file directory, the assembler searches the
paths specified in the assembler’s command-line option /I, or in PWB’s Include
Paths field in the MASM Option dialog box (accessed from the Option menu).
The /I option takes this form:
/I path
You can include more than one /I option on the command line. The assembler
then searches for include files within each specified path in the order given. If
none of these directories contains the include file, the assembler finally searches
in the paths specified in the INCLUDE environment variable. If the include file
still cannot be found, an assembly error occurs. (The /x command-line option
tells the assembler to ignore the INCLUDE environment variable when
searching for include files.)
An include file may specify another include file. The assembler processes the
second include file before returning to the first. Your program can nest include
files this way as deeply as the amount of free memory allows.

Include Files or Modules
You can use the EQU directive to create named constants that cannot be
redefined in your program. (For information about the EQU directive, see
“Integer Constants and Constant Expressions,” page 11.) Placing a constant
defined with EQU in an include file makes it available to all modules that use
that include file.
Placing TYPEDEF, STRUCT, UNION, and RECORD definitions in an include
file guarantees consistency in type definitions. If required, the variable instances
derived from these definitions can be made public among the modules with
EXTERNDEF declarations (see the next section). Macros, including macros
defined with TEXTEQU, must be placed in include files to make them visible in
other modules.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 213 of 3 Printed: 10/02/00 04:22 PM

214

Programmer’s Guide

If you elect to use full segment definitions with, or instead of, simplified
definitions, you can force a consistent segment order in all files by defining
segments in an include file. This technique is explained in “Controlling the
Segment Order,”
page 47.

Declaring Symbols Public and External
It is sometimes useful to make certain procedures and variables (such as status
flags) global to all program modules. Global variables are freely accessible within
all routines; you do not have to explicitly pass them to the routines that need
them. This section describes how to make variables and procedures global using
the EXTERNDEF, PROTO, or COMM declarations within include files.
When a procedure is defined in one module and called in another module, it
must be declared public in the defining module and external in the calling
module(s). MASM offers three ways to declare a procedure public and external:
u

u
u

Use the PUBLIC directive in the defining module and EXTERN in all other
modules that reference the procedure. The PUBLIC and EXTERN
directives are explained on page 220.
Declare the procedure with EXTERNDEF.
Prototype the procedure with the PROTO directive.

Using EXTERNDEF
MASM treats EXTERNDEF as a public declaration in the defining module, and
as an external declaration in the referencing module(s). You can use the
EXTERNDEF statement in your include file to make a variable common to two
or more modules. EXTERNDEF works with all types of variables, including
arrays, structures, unions, and records. It also works with procedures.
As a result, a single include file can contain an EXTERNDEF declaration that
works in both the defining module and any referencing module. It is ignored in
modules that neither define nor reference the variable. Therefore, an include file
for a library which is used in multiple .EXE files does not force the definition of
a symbol as EXTERN does.
The EXTERNDEF statement takes this form:
EXTERNDEF [[langtype]] name:qualifiedtype
The name is the variable’s identifier. The qualifiedtype is explained in detail in
“Data Types,” page 14.
The optional langtype specifier sets the naming conventions for the name it
precedes. It overrides any language specified in the .MODEL directive. The
specifier can be C, SYSCALL, STDCALL, PASCAL, FORTRAN, or

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 214 of 4 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

BASIC. For information on selecting the appropriate langtype type, see
“Naming and Calling Conventions,” page 308.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 215 of 5 Printed: 10/02/00 04:22 PM

215

216

Programmer’s Guide

The following diagram shows the statements that declare an array, make it
public, and use it in another module.

Figure 8.1

Using EXTERNDEF for Variables

The file position of EXTERNDEF directives is important. For more information,
see “Positioning External Declarations,” following.
You can also make procedures visible by using EXTERNDEF without PROTO
inside an include file. This method treats the procedure name as a simple
identifier, without the parameter list, so you forgo the assembler’s ability to
check for the correct parameters during assembly. Use EXTERNDEF with
procedures in the same way as variables:
EXTERNDEF MyProc:FAR

; Declare far procedure external

You can also use EXTERNDEF to make a code label global between modules
so that one module can reference a label in another module. Give the label
global scope with the double colon operator, like this:
EXTERNDEF codelabel:NEAR
.
.
.
codelabel::

Another module can reference codelabel like this:
EXTERNDEF codelabel:NEAR
.
.
.
jmp
codelabel

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 216 of 6 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

217

Using PROTO
This section describes how to prototype a procedure with the PROTO directive.
PROTO automatically issues an EXTERNDEF for the procedure unless the
PROC statement declares the procedure PRIVATE. Defining a prototype
enables type-checking for the procedure arguments.
Follow these steps to create an interface for a procedure defined in one module
and called from other modules:
1. Place the PROTO declaration in the include file.
2. Define the procedure with PROC in one module. The PROC directive
declares the procedure PUBLIC by default.
3. Call the procedure with the INVOKE statement (or with CALL). Make sure
that all calling modules access the include file.
For descriptions, syntax, and examples of PROTO, PROC, and INVOKE, see
Chapter 7, “Controlling Program Flow.”
The following example illustrates these three steps. In the example, a PROTO
statement defines the far procedure CopyFile, which uses the C parameterpassing and naming conventions, and takes the arguments filename and
numberlines. The diagram following the example shows the file placement
for these statements.
This definition goes into the include file:
CopyFile PROTO FAR C filename:BYTE, numberlines:WORD

The procedure definition for CopyFile is:
CopyFile PROC FAR C USES cx, filename:BYTE, numberlines:WORD

To call the CopyFile procedure, you can use this INVOKE statement:
INVOKE

CopyFile, NameVar, 200

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 217 of 7 Printed: 10/02/00 04:22 PM

218

Programmer’s Guide

Figure 8.2

Using PROTO and INVOKE

Using COMM
Another way to share variables among modules is to add the COMM
(communal) declaration to your include file. Since communal variables are
allocated by the linker and cannot be initialized, you cannot depend on their
location or sequence.
Communal variables are supported by MASM primarily for compatibility with
communal variables in Microsoft C. Communal variables are not used in any
other Microsoft language, and they are not compatible with C++ and some other
languages.
COMM declares a data variable external and instructs the linker to allocate the
variable if it has not been explicitly defined in a module. The memory space for
communal variables may not be assigned until load time, so using communal
variables may reduce the size of your executable file.
The COMM declaration has the syntax:
COMM [[langtype]] [[NEAR | FAR]] label:type[[:count]]
The label is the name of the variable. The langtype sets the naming conventions
for the name it precedes. It overrides any language specified in the .MODEL
directive.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 218 of 8 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

219

If NEAR or FAR is not specified, the variable determines the default from the
current memory model (NEAR for TINY, SMALL, COMPACT, and FLAT;
FAR for MEDIUM, LARGE, and HUGE). If you do not provide a memory
model with the .MODEL directive, you must specify a distance when accessing
a communal variable, like this:
mov
mov

ax, NEAR PTR CommNear
bx, FAR PTR CommFar

The type can be a constant expression, but it is usually a type such as BYTE,
WORD, or DWORD, or a structure, union, or record. If you first declare the
type with TYPEDEF, CodeView can provide type information. The count is the
number of elements. If no count is given, one element is assumed.
The following example creates the on far variable DataBlock, which is a
1,024-element array of uninitialized signed doublewords:
COMM FAR DataBlock:SDWORD:1024

Note C variables declared outside functions (except static variables) are
communal unless explicitly initialized; they are the same as assembly-language
communal variables. If you are writing assembly-language modules for C, you
can declare the same communal variables in both C and MASM include files.
However, communal variables in C do not have to be declared communal in
assembler. The linker will match the EXTERN, PUBLIC, and COMM
statements for the variable.
EXTERNDEF (explained in the previous section) is more flexible than COMM
because you can initialize variables defined with it, and your code can rely on
the position and sequence of the defined data.

Positioning External Declarations
Although LINK determines the actual address of an external symbol, the
assembler assumes a default segment for the symbol, based on the location of
the external directive in the source code. You should therefore position
EXTERN and
EXTERNDEF directives according to these rules:
u

If you know which segment defines an external symbol, put the EXTERN
statement in that segment.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 219 of 9 Printed: 10/02/00 04:22 PM

220

Programmer’s Guide
u

If you know the group but not the segment, position the EXTERN statement
outside any segment and reference the variable with the group name. For
example, if var1 is in DGROUP, reference the variable as
mov DGROUP:var1, 10

If you know nothing about the location of an external variable, put the
EXTERN statement outside any segment. You can use the SEG directive to
access the external variable like this:
mov
mov
mov

ax, SEG var1
es, ax
ax, es:var1

If the symbol is an absolute symbol or a far code label, you can declare it
external anywhere in the source code.

Always close any segments opened in include files so that external declarations
following an include statement are not incorrectly placed inside a segment. If
you want to be certain an external definition lies outside a segment, you can use
@CurSeg. The @CurSeg predefined symbol returns a blank if the definition is
not in a segment. For example,
.DATA
.
.
.
@CurSeg ENDS
EXTERNDEF var:WORD

; Close segment

For information about predefined symbols such as @CurSeg, see “Predefined
Symbols,” page 10.

Using Alternatives to Include Files
If your project uses only two modules (or if it is written with a version of
MASM prior to 6.0), you may want to continue using PUBLIC in the defining
module and EXTERN in the referencing module, and not create an include file
for the project. The EXTERN directive can be used in an include file, but the
include file containing EXTERN cannot be added to the module that contains
the corresponding PUBLIC directive for that symbol. This section assumes that
you are not using include files.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 220 of 10 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

221

PUBLIC and EXTERN
The PUBLIC and EXTERN directives are less flexible than EXTERNDEF and
PROTO because they are module-specific: PUBLIC must appear in the
defining module and EXTERN must appear in the calling modules. This section
shows how to use PUBLIC and EXTERN. Information on where to place the
external declarations in your file is in “Positioning External Declarations,”
previous.
The PUBLIC directive makes a name visible outside the module in which it is
defined. This gives other program modules access to that identifier.
The EXTERN directive performs the complementary function. It tells the
assembler that a name referenced within a particular module is actually defined
and declared public in another module that will be specified at link time.
A PUBLIC directive can appear anywhere in a file. Its syntax is:
PUBLIC [[langtype]] name[[, [[langtype]] name]]...
The name must be the name of an identifier defined within the current source
file. Only code labels, data labels, procedures, and numeric equates can be
declared public.
If you specify the langtype field here, it overrides the language specified by
.MODEL. The langtype field can be C, SYSCALL, STDCALL, PASCAL,
FORTRAN, or BASIC. For more information on specifying langtype types, see
“Declaring Parameters with the PROC Directive,” page 184, and “Naming and
Calling Conventions,” page 308.
The EXTERN directive tells the assembler that an identifier is external —
defined in some other module that will be supplied at link time. Its syntax is:
EXTERN [[langtype]] name:{ABS | qualifiedtype}
“Data Types,” page 14, describes qualifiedtype. You can use the ABS
(absolute) keyword only with external numeric constants. ABS causes the
identifier to be imported as a relocatable unsized constant. This identifier can
then be used anywhere a constant can be used. If the identifier is not found in
another module at link time, the linker generates an error.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 221 of 11 Printed: 10/02/00 04:22 PM

222

Programmer’s Guide

In the following example, the procedure BuildTable and the variable Var are
declared public. The procedure uses the Pascal naming and data-passing
conventions:

Figure 8.3

Using PUBLIC and EXTERN

Other Alternatives
You can also use the directives discussed earlier (EXTERNDEF, PROTO, and
COMM) without the include file. In this case, place the declarations to make a
symbol global in the same module where the symbol is defined. You might want
to use this technique if you are linking only a few modules that have very little
data in common.

Developing Libraries
As you create reusable procedures, you can place them in a library file for
convenient access. Although you can put any routine into a library, each library
file, recognizable by its .LIB extension, usually contains related routines. For
example, you might place string-manipulation functions in one library, matrix
calculations in another, and port communications in another. Do not place
communal variables (defined with the COMM directive) in a library.
A library consists of combined object modules, each created from a single
source file. The object module is the smallest independent unit in a library. If
you link with one symbol in a module, the linker adds the entire module to your
program, but not the entire library.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 222 of 12 Printed: 10/02/00 04:22 PM

Chapter 8 Sharing Data and Procedures Among Modules and Libraries

223

Associating Libraries with Modules
You can choose either of two methods for associating your libraries with the
modules that use them: you can use the INCLUDELIB directive inside your
source files, or link the modules from the command line.
To associate a specified library with your object code, use INCLUDELIB. You
can add this directive to the source file to specify the libraries you want linked,
rather than specifying them in the LINK command line. The INCLUDELIB
syntax is:
INCLUDELIB libraryname
The libraryname can be a file name or a complete path specification. If you do
not specify an extension, .LIB is assumed. The libraryname is placed in the
comment record of the object file. LINK reads this record and links with the
specified library file.
For example, the statement INCLUDELIB GRAPHICS passes a message from
the assembler to the linker telling LINK to use library routines from the file
GRAPHICS.LIB. If you place this statement in the source file DRAW.ASM and
GRAPHICS.LIB is in the same directory, you can assemble and link the
program with the following command:
ML DRAW.ASM

Without the INCLUDELIB directive, you must link the program DRAW.ASM
with either of the following commands:
ML DRAW.ASM GRAPHICS.LIB
ML DRAW /link GRAPHICS

If you want to assemble and link separately, type
ML /c DRAW.ASM
LINK DRAW,,,GRAPHICS

If you do not specify a complete path in the INCLUDELIB statement or at the
command line, LINK searches for the library file in the following order:
1. In the current directory.
2. In any directories in the library field of the LINK command line.
3. In any directories specified by the LIB environment variable.
The LIB.EXE utility helps you create, organize, and maintain run-time libraries.
Refer to Environment and Tools for instructions on LIB.EXE.

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 223 of 13 Printed: 10/02/00 04:22 PM

224

Programmer’s Guide

Using EXTERN with Library Routines
In some cases, EXTERN helps you limit the size of your executable file by
specifying in the syntax an alternative name for a procedure. You would use this
form of the EXTERN directive when declaring a procedure or symbol that may
not need to be used.
The syntax looks like this:
EXTERN [[langtype]] name [[ (altname) ]] :qualifiedtype
The addition of the altname to the syntax provides the name of an alternate
procedure that the linker uses to resolve the external reference if the procedure
given by name is not needed. Both name and altname must have the same
qualifiedtype.
When the linker encounters an external definition for a procedure that gives an
altname, the linker finishes processing that module before it links the object
module that contains the procedure given by name. If the program does not
reference any symbols in the name file’s object from any of the linked modules,
the linker uses altname to satisfy the external reference. This saves space
because the library object module is not brought in.
For example, assume that the contents of STARTUP.ASM include these
statements:
EXTERN

dummy

init(dummy):PROC
.
.
.
PROC
.
.
.
ret

; A procedure definition containing no
;
executable code

ENDP
.
.
.
call

; Defined in FLOAT.OBJ

init

In this example, the reference to the routine init (defined in FLOAT.OBJ)
does not force the module FLOAT.OBJ to be linked into the executable file. If
another reference causes FLOAT.OBJ to be linked into the executable file, then
init will refer to the init label in FLOAT.OBJ. If there are no references that
force linkage with FLOAT.OBJ, the linker will use the alternate name for
init(dummy).

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 224 of 14 Printed: 10/02/00 04:22 PM

224

Programmer’s Guide

Filename: LMAPGC08.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 224 of 16 Printed: 10/02/00 04:22 PM

225

C H A P T E R

Using Macros

A “macro” is a symbolic name you give to a series of characters (a text macro)
or to one or more statements (a macro procedure or function). As the assembler
evaluates each line of your program, it scans the source code for names of
previously defined macros. When it finds one, it substitutes the macro text for
the macro name. In this way, you can avoid writing the same code several
places in your program.
This chapter describes the following types of macros:
u
u

Text macros, which expand to text within a source statement.
Macro procedures, which expand to one or more complete statements and
can optionally take parameters.
Repeat blocks, which generate a group of statements a specified number of
times or until a specified condition becomes true.
Macro functions, which look like macro procedures and can be used like text
macros but which also return a value.
Predefined macro functions and string directives, which perform string
operations.

This chapter explains how to use macros for simple code substitutions and how
to write sophisticated macros with parameter lists and repeat loops. It also
describes how to use these features in conjunction with local symbols, macro
operators, and predefined macro functions.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 225 of 1 Printed: 10/02/00 04:22 PM

226

Programmer’s Guide

Text Macros
You can give a sequence of characters a symbolic name and then use the name
in place of the text later in the source code. The named text is called a text
macro.
The TEXTEQU directive defines a text macro, as these examples show:
name TEXTEQU
name TEXTEQU macroId | textmacro
name TEXTEQU %constExpr
In the previous lines, text is a sequence of characters enclosed in angle brackets,
macroId is a previously defined macro function, textmacro is a previously
defined text macro, and %constExpr is an expression that evaluates to text.
Here are some examples:
msg
string
msg
value

TEXTEQU
TEXTEQU
TEXTEQU
TEXTEQU

msg

%(3 + num)

;
;
;
;
;

Text assigned to symbol
Text macro assigned to symbol
New text assigned to symbol
Text representation of resolved
expression assigned to symbol

The first line assigns text to the symbol msg. The second line equates the text of
the msg text macro with a new text macro called string. The third line assigns
new text to msg. Although msg has new text, string retains its original text
value. The fourth line assigns 7 to value if num equals 4. If a text macro
expands to another text macro (or macro function, as discussed on page 248),
the resulting text macro will expand recursively.
Text macros are useful for naming strings of text that do not evaluate to
integers. For example, you might use a text macro to name a floating-point
constant or a bracketed expression. Here are some practical examples:
pi
WPT
arg1

TEXTEQU <3.1416>
TEXTEQU
TEXTEQU <[bp+4]>

; Floating point constant
; Sequence of key words
; Bracketed expression

Macro Procedures
If your program must perform the same task many times, you can avoid
repeatedly typing the same statements each time by writing a macro procedure.
Think of macro procedures (commonly called macros) as text-processing
mechanisms that automatically generate repeated text.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 226 of 2 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

227

This section uses the term “macro procedure” rather than “macro” when
necessary to distinguish between a macro procedure and a macro function.
Macro functions are described in “Returning Values with Macro Functions.”
Conforming to common usage, this chapter occasionally speaks of “calling” a
macro, a term that deserves further scrutiny. It’s natural to think of a program
calling a macro procedure in the same way it calls a normal subroutine
procedure, because they seem to perform identically. However, a macro is
simply a representative for real code. Wherever a macro name appears in your
program, so in reality does all the code the macro represents. A macro does not
cause the processor to vector off to a new location as does a normal procedure.
Thus, the expression “calling a macro” may imply the effect, but does not
accurately describe what actually occurs.

Creating Macro Procedures
You can define a macro procedure without parameters by placing the desired
statements between the MACRO and ENDM directives:
name MACRO
statements
ENDM
For example, suppose you want a program to beep when it encounters certain
errors. You could define a beep macro as follows:
beep
mov
mov
int

MACRO
ah, 2
dl, 7
21h

;; Select DOS Print Char function
;; Select ASCII 7 (bell)
;; Call DOS

ENDM

The double semicolons mark the beginning of macro comments. Macro
comments appear in a listing file only at the macro’s initial definition, not at the
point where the macro is referenced and expanded. Listings are usually easier to
read if the comments aren’t repeatedly expanded. However, regular comments
(those with a single semicolon) are listed in macro expansions. See Appendix C
for listing files and examples of how macros are expanded in listings.
Once you define a macro, you can call it anywhere in the program by using the
macro’s name as a statement. The following example calls the beep macro two
times if an error flag has been set.
.IF
beep
beep
.ENDIF

error

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 227 of 3 Printed: 10/02/00 04:22 PM

; If error flag is true
;
execute macro two times

228

Programmer’s Guide

During assembly, the instructions in the macro replace the macro reference. The
listing file shows:
.IF
0017
001C

80 3E 0000 R 00
74 0C

*
*

001E
0020
0022

B4 02
B2 07
CD 21

1
1
1

0024
0026
0028

B4 02
B2 07
CD 21

1
1
1

error
cmp
je

error, 000h
@C0001

beep
mov
mov
int

ah, 2
dl, 7
21h

mov
mov
int

ah, 2
dl, 7
21h

beep

002A

.ENDIF
*@C0001:

Contrast this with the results of defining beep as a procedure using the PROC
directive and then calling it with the CALL instruction.
Many such tasks can be handled as either a macro or a procedure. In deciding
which method to use, you must choose between speed and size. For repetitive
tasks, a procedure produces smaller code, because the instructions physically
appear only once in the assembled program. However, each call to the
procedure involves the additional overhead of a CALL and RET instruction.
Macros do not require a change in program flow and so execute faster, but
generate the same code multiple times rather than just once.

Passing Arguments to Macros
By defining parameters for macros, you can define a general task and then
execute variations of it by passing different arguments each time you call the
macro. The complete syntax for a macro procedure includes a parameter list:
name MACRO parameterlist
statements
ENDM
The parameterlist can contain any number of parameters. Use commas to
separate each parameter in the list. You cannot use reserved words as parameter
names unless you disable the keyword with OPTION NOKEYWORD. You
must also set the compatibility mode with OPTION M510 or the /Zm
command-line option.
To pass arguments to a macro, place the arguments after the macro name when
you call the macro:
macroname arglist

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 228 of 4 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

229

The assembler treats as one item all text between matching quotation marks in
an arglist.
The beep macro introduced in the previous section used the MS-DOS interrupt
to write only the bell character (ASCII 7). We can rewrite the macro with a
parameter that accepts any character:
writechar MACRO char
mov ah, 2
mov dl, char
int 21h
ENDM

;; Select DOS Print Char function
;; Select ASCII char
;; Call DOS

Whenever it expands the macro, the assembler replaces each instance of char
with the given argument value. The rewritten macro now writes any character to
the screen, not just ASCII 7:
writechar 7
writechar ‘A’

; Causes computer to beep
; Writes A to screen

If you pass more arguments than there are parameters, the additional arguments
generate a warning (unless you use the VARARG keyword; see page 242). If
you pass fewer arguments than the macro procedure expects, the assembler
assigns empty strings to the remaining parameters (unless you have specified
default values). This may cause errors. For example, a reference to the
writechar macro with no argument results in the following line:
mov

dl,

The assembler generates an error for the expanded statement but not for the
macro definition or the macro call.
You can make macros more flexible by leaving off arguments or adding
additional arguments. The next section tells some of the ways your macros can
handle missing or extra arguments.

Specifying Required and Default Parameters
Macro parameters can have special attributes to make them more flexible and
improve error handling. You can make parameters required, give them default
values, or vary their number. Variable parameters are used almost exclusively
with the FOR directive, so are covered in “FOR Loops and Variable-Length
Parameters,” later in this chapter.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 229 of 5 Printed: 10/02/00 04:22 PM

230

Programmer’s Guide

The syntax for a required parameter is:
parameter:REQ
For example, you can rewrite the writechar macro to require the char
parameter:
writechar MACRO char:REQ
mov ah, 2
mov dl, char
int 21h
ENDM

;; Select DOS Print Char function
;; Select ASCII char
;; Call DOS

If the call does not include a matching argument, the assembler reports the error
in the line that contains the macro reference. REQ can thus improve error
reporting.
You can also accommodate missing parameters by specifying a default value,
like this:
parameter:=textvalue
Suppose that you often use writechar to beep by printing ASCII 7. The
following macro definition uses an equal sign to tell the assembler to assume the
parameter char is 7 unless you specify otherwise:
writechar MACRO char:=<7>
mov ah, 2
mov dl, char
int 21h
ENDM

;; Select DOS Print Char function
;; Select ASCII char
;; Call DOS

If a reference to this macro does not include the argument char, the assembler
fills in the blank with the default value of 7 and the macro beeps when called.
Enclose the default parameter value in angle brackets so the assembler
recognizes the supplied value as a text value. This is explained in detail in “Text
Delimiters and the Literal-Character Operator,” later in this chapter.
Missing arguments can also be handled with the IFB, IFNB, .ERRB, and
.ERRNB directives. They are described in the section “Conditional Directives”
in chapter 1 and in Help. Here is a slightly more complex macro that uses some
of these techniques:

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 230 of 6 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

231

Scroll MACRO distance:REQ, attrib:=<7>, tcol, trow, bcol, brow
IFNB
;; Ignore arguments if blank
mov
cl, tcol
ENDIF
IFNB
mov
ch, trow
ENDIF
IFNB
mov
dl, bcol
ENDIF
IFNB
mov
dh, brow
ENDIF
IFDIFI ,
;; Don’t move BH onto itself
mov
bh, attrib
ENDIF
IF distance LE 0
;; Negative scrolls up, positive down
mov
ax, 0600h + (-(distance) AND 0FFh)
ELSE
mov
ax, 0700h + (distance AND 0FFh)
ENDIF
int
10h
ENDM

In this macro, the distance parameter is required. The attrib parameter has
a default value of 7 (white on black), but the macro also tests to make sure the
corresponding argument isn’t BH, since it would be inefficient (though legal) to
load a register onto itself. The IFNB directive is used to test for blank
arguments. These are ignored to allow the user to manipulate rows and columns
directly in registers CX and DX at run time.
The following shows two valid ways to call the macro:
; Assume DL and CL already loaded
dec
dh
; Decrement top row
inc
ch
; Increment bottom row
Scroll -3
; Scroll white on black dynamic
;
window up three lines
Scroll 5, 17h, 2, 2, 14, 12 ; Scroll white on blue constant
;
window down five lines

This macro can generate completely different code, depending on its arguments.
In this sense, it is not comparable to a procedure, which always has the same
code regardless of arguments.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 231 of 7 Printed: 10/02/00 04:22 PM

232

Programmer’s Guide

Defining Local Symbols in Macros
You can make a symbol local to a macro by identifying it at the start of the
macro with the LOCAL directive. Any identifier may be declared local.
You can choose whether you want numeric equates and text macros to be local
or global. If a symbol will be used only inside a particular macro, you can
declare it local so that the name will be available for other declarations outside
the macro.
You must declare as local any labels within a macro, since a label can occur
only once in the source. The LOCAL directive makes a special instance of the
label each time the macro appears. This prevents redefinition of the label when
expanding the macro. It also allows you to reuse the label elsewhere in your
code.
You must declare all local symbols immediately following the MACRO
statement (although blank lines and comments may precede the local symbol).
Separate each symbol with a comma. You can attach comments to the LOCAL
statement and list multiple LOCAL statements in the macro. Here is an example
macro that declares local labels:
power
MACRO
factor:REQ, exponent:REQ
LOCAL
again, gotzero
;; Local symbols
sub
dx, dx
;; Clear top
mov
ax, 1
;; Multiply by one on first loop
mov
cx, exponent
;; Load count
jcxz
gotzero
;; Done if zero exponent
mov
bx, factor
;; Load factor
again:
mul
bx
;; Multiply factor times exponent
loop
again
;; Result in AX
gotzero:
ENDM

If the labels again and gotzero were not declared local, the macro would
work the first time it is called, but it would generate redefinition errors on
subsequent calls. MASM implements local labels by generating different names
for them each time the macro is called. You can see this in listing files. The
labels in the power macro might be expanded to ??0000 and ??0001 on the
first call and to ??0002 and ??0003 on the second.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 232 of 8 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

233

You should avoid using anonymous labels in macros (see “Anonymous Labels”
in Chapter 7). Although legal, they can produce unwanted results if you expand
a macro near another anonymous label. For example, consider what happens in
the following:
Update MACRO arg1
@@: .
.
.
loop @B
ENDM
.
.
.
jcxz
@F
Update ax
@@:

Expanding Update places another anonymous label between the jump and its
target. The line
jcxz

consequently jumps to the start of the loop rather than over the loop — exactly
the opposite of what the programmer intended.

Assembly-Time Variables and Macro Operators
In writing macros, you will often assign and modify values assigned to symbols.
Think of these symbols as assembly-time variables. Like memory variables,
they are symbols that represent values. But since macros are processed at
assembly time, any symbol modified in a macro must be resolved as a constant
by the end of assembly.
The three kinds of assembly-time variables are:
u
u
u

Macro parameters
Text macros
Macro functions

When the assembler expands a macro, it processes the symbols in the order
shown here. MASM first replaces macro parameters with the text of their actual
arguments, then expands text macros.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 233 of 9 Printed: 10/02/00 04:22 PM

234

Programmer’s Guide

Macro parameters are similar to procedure parameters in some ways, but they
also have important differences. In a procedure, a parameter has a type and a
memory location. Its value can be modified within the procedure. In a macro, a
parameter is a placeholder for the argument text. The value can only be assigned
to another symbol or used directly; it cannot be modified. The macro may
interpret the argument text it receives either as a numeric value or as a text
value.
It is important to understand the difference between text values and numeric
values. Numeric values can be processed with arithmetic operators and assigned
to numeric equates. Text values can be processed with macro functions and
assigned to text macros.
Macro operators are often helpful when processing assembly-time variables.
Table 9.1 shows the macro operators that MASM provides.
Table 9.1 MASM Macro Operators
Symbol

Name

Description

Text Delimiters

Opens and closes a literal string.

Literal-Character Operator

Treats the next character as a literal character,
even if it would normally have another meaning.

Expansion Operator

Causes the assembler to expand a constant
expression or text macro.

Substitution Operator

Tells the assembler to replace a macro
parameter or text macro name with its
actual value.

The next sections explain these operators in detail.

Text Delimiters and the Literal-Character Operator
The angle brackets (< >) are text delimiters. A text value is usually delimited
when assigning a text macro. You can do this with TEXTEQU, as previously
shown, or with the SUBSTR and CATSTR directives discussed in “String
Directives and Predefined Functions,” later in this chapter.
By delimiting the text of macro arguments, you can pass text that includes
spaces, commas, semicolons, and other special characters. The following
example expands a macro called work in two different ways:
work
work

<1, 2, 3, 4, 5> ; Passes one argument with 13 chars,
;
including commas and spaces
1, 2, 3, 4, 5
; Passes five arguments, each
;
with 1 character

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 234 of 10 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

235

The literal-character operator (!) lets you include angle brackets as part of a
delimited text value, so the assembler does not interpret them as delimiters. The
assembler treats the character following ! literally rather than as a special
character, like this:
errstr

TEXTEQU 255>

; errstr = “Expression > 255”

Text delimiters also have a special use with the FOR directive, as explained in
“FOR Loops and Variable-Length Parameters,” later in this chapter.

Expansion Operator
The expansion operator (%) expands text macros or converts constant
expressions into their text representations. It performs these tasks differently in
different contexts, as discussed in the following.

Converting Numeric Expressions to Text
The expansion operator can convert numbers to text. The operator forces
immediate evaluation of a constant expression and replaces it with a text value
consisting of the digits of the result. The digits are generated in the current radix
(default decimal).
This application of the expansion operator is useful when defining a text macro,
as the following lines show. Notice how you can enclose expressions with
parentheses to make them more readable:
a
b
c

TEXTEQU <3 + 4>
TEXTEQU %3 + 4
TEXTEQU %(3 + 4)

; a = “3 + 4”
; b = “7”
; c = “7”

When assigning text macros, you can use numeric equates in the constant
expressions, but not text macros:
num
numstr
a
b

EQU
TEXTEQU
TEXTEQU
TEXTEQU

4
<4>
%3 + num
%3 + numstr

;
;
;
;

num = 4
numstr = <4>
a = <7>
b = <7>

The expansion operator gives you flexibility when passing arguments to macros.
It lets you pass a computed value rather than the literal text of an expression.
The following example illustrates by defining a macro
work

MACRO
arg
mov ax, arg * 4

ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 235 of 11 Printed: 10/02/00 04:22 PM

236

Programmer’s Guide

which accepts different arguments:
work

2 + 3

work

%2 + 3

work
work
work
work

2 + num
%2 + num
2 + numstr
%2 + numstr

;
;
;
;
;
;
;
;

Passes “2 + 3”
Code: mov ax, 2 + (3 * 4)
Passes 5
Code: mov ax, 5 * 4
Passes “2 + num”
Passes “6”
Passes “2 + numstr”
Passes “6”

You must consider operator precedence when using the expansion operator.
Parentheses inside the macro can force evaluation in a desired order:
work

MACRO
arg
mov ax, (arg) * 4

ENDM
work
work

2 + 3
%2 + 3

; Code: mov ax, (2 + 3) * 4
; Code: mov ax, (5) * 4

Several other uses for the expansion operator are reviewed in “Returning Values
with Macro Functions,” later in this chapter.

Expansion Operator as First Character on a Line
The expansion operator has a different meaning when used as the first character
on a line. In this case, it instructs the assembler to expand any text macros and
macro functions it finds on the rest of the line.
This feature makes it possible to use text macros with directives such as ECHO,
TITLE, and SUBTITLE, which take an argument consisting of a single text
value. For instance, ECHO displays its argument to the standard output device
during assembly. Such expansion can be useful for debugging macros and
expressions, but the requirement that its argument be a single text value may
have unexpected results. Consider this example:
ECHO

Bytes per element: %(SIZEOF array / LENGTHOF array)

Instead of evaluating the expression, this line echoes it:
Bytes per element: %(SIZEOF array / LENGTHOF array)

However, you can achieve the desired result by assigning the text of the
expression to a text macro and then using the expansion operator at the
beginning of the line to force expansion of the text macro.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 236 of 12 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros
temp
%

237

TEXTEQU %(SIZEOF array / LENGTHOF array)
ECHO
Bytes per element: temp

Note that you cannot get the same results simply by putting the % at the
beginning of the first echo line, because % expands only text macros, not
numeric equates or constant expressions.
Here are more examples of the expansion operator at the start of a line:
; Assume memmod, lang, and os specified with /D option
%
SUBTITLE Model: memmod Language: lang Operating System: os
; Assume num defined earlier
tnum
TEXTEQU %num
%
.ERRE
num LE 255, 255>

Substitution Operator
References to a parameter within a macro can sometimes be ambiguous. In such
cases, the assembler may not expand the argument as you intend. The
substitution operator (&) lets you identify unambiguously any parameter within a
macro.
As an example, consider the following macro:
errgen MACRO
num, msg
PUBLIC errnum
errnum BYTE
“Error num: msg”
ENDM

This macro is open to several interpretations:
u
u

Is errnum a distinct word or the word err next to the parameter num?
Should num and msg within the string be treated literally as part of the string
or as arguments?

In each case, the assembler chooses the most literal interpretation. That is, it
treats errnum as a distinct word, and num and msg as literal parts of the string.
The substitution operator can force different interpretations. If we rewrite the
macro with the & operator, it looks like this:
errgen MACRO
num, msg
PUBLIC err&num
err&num BYTE
“Error &num: &msg”
ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 237 of 13 Printed: 10/02/00 04:22 PM

238

Programmer’s Guide

When called with the following arguments,
errgen

the macro now generates this code:
err5

PUBLIC
BYTE

err5
“Error 5: Unreadable disk”

When it encounters the & operator, the assembler interprets subsequent text as a
parameter name until the next & or until the next separator character (such as a
space, tab, or comma). Thus, the assembler correctly parses the expression
err&num because num is delimited by & and a space. The expression could also
be written as err&num&, which again unambiguously identifies num as a
parameter.
The rule also works in reverse. You can delimit a parameter reference with & at
the end rather than at the beginning. For example, if num is 5, the expression
num&12 resolves to “512.”
The assembler processes substitution operators from left to right. This can have
unexpected results when you are pasting together two macro parameters. For
example, if arg1 has the value var and arg2 has the value 3, you could paste
them together with this statement:
&arg1&&arg2&

BYTE

“Text”

Eliminating extra substitution operators, you might expect the following to be
equivalent:
&arg1&arg2

BYTE

“Text”

However, this actually produces the symbol vararg2, because in processing
from left to right, the assembler associates both the first and the second &
symbols with the first parameter. The assembler replaces &arg1& by var,
producing vararg2. The arg2 is never evaluated. The correct abbreviation is:
arg1&&arg2

BYTE

“Text”

which produces the desired symbol var3. The symbol arg1&&arg2 is replaced
by var&arg2, which is replaced by var3.
The substitution operator is also necessary if you want to substitute a text macro
inside quotes. For example,

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 238 of 14 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros
arg
%echo
%echo

239

TEXTEQU
This is a string “&arg” ; Produces: This is a string “hello”
This is a string “arg” ; Produces: This is a string “arg”

You can also use the substitution operator in lines beginning with the expansion
operator (%) symbol, even outside macros (see page 236). It may be necessary
to use the substitution operator to paste text macro names to adjacent characters
or symbol names, as shown here:
text
value
%

TEXTEQU
TEXTEQU %5
ECHO
textvalue is text&&value

This echoes the message
textvalue is var5

Macro substitution always occurs before evaluation of the high-level control
structures. The assembler may therefore mistake a bit-test operator (&) in your
macro for a substitution operator. You can guarantee the assembler correctly
recognizes a bit-test operator by enclosing its operands in parentheses, as shown
here:
test

MACRO
x
.IF ax==&x
mov
ax, 10
.ELSEIF ax&(x)
mov ax, 20
.ENDIF

; &x substituted with parameter value
; & is bitwise AND

ENDM

The rules for using the substitution operator have changed significantly since
MASM 5.1, making macro behavior more consistent and flexible. If you have
macros written for MASM 5.1 or earlier, you can specify the old behavior by
using OLDMACROS or M510 with the OPTION directive (see page 24).

Defining Repeat Blocks with Loop Directives
A “repeat block” is an unnamed macro defined with a loop directive. The loop
directive generates the statements inside the repeat block a specified number of
times or until a given condition becomes true.
MASM provides several loop directives, which let you specify the number of
loop iterations in different ways. Some loop directives can also accept arguments
for each iteration. Although the number of iterations is usually specified in the
directive, you can use the EXITM directive to exit the loop early.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 239 of 15 Printed: 10/02/00 04:22 PM

240

Programmer’s Guide

Repeat blocks can be used outside macros, but they frequently appear inside
macro definitions to perform some repeated operation in the macro. Since repeat
blocks are macros themselves, they end with the ENDM directive.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 240 of 16 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

241

This section explains the following four loop directives: REPEAT, WHILE,
FOR, and FORC. In versions of MASM prior to 6.0, REPEAT was called
REPT, FOR was called IRP, and FORC was called IRPC. MASM 6.1
recognizes the old names.
The assembler evaluates repeat blocks on the first pass only. You should
therefore avoid using address spans as loop counters, as in this example:
REPEAT

(OFFSET label1 - OFFSET label2) ; Don't do this!

Since the distance between two labels may change on subsequent assembly
passes as the assembler optimizes code, you should not assume that address
spans remain constant between passes.
Note The REPEAT and WHILE directives should not be confused with the
REPEAT and WHILE directives (see “Loop-Generating Directives” in Chapter
7), which generate loop and jump instructions for run-time program control.

REPEAT Loops
REPEAT is the simplest loop directive. It specifies the number of times to
generate the statements inside the macro. The syntax is:
REPEAT constexpr
statements
ENDM
The constexpr can be a constant or a constant expression, and must contain no
forward references. Since the repeat block expands at assembly time, the
number of iterations must be known then.
Here is an example of a repeat block used to generate data. It initializes an array
containing sequential ASCII values for all uppercase letters.
alpha
LABEL
BYTE
letter =
‘A’
REPEAT 26
BYTE
letter
letter = letter + 1
ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 241 of 17 Printed: 10/02/00 04:22 PM

;
;
;;
;;
;;

Name the data generated
Initialize counter
Repeat for each letter
Allocate ASCII code for letter
Increment counter

242

Programmer’s Guide

Here is another use of REPEAT, this time inside a macro:
beep

MACRO
iter:=<3>
mov ah, 2
mov dl, 7
REPEAT iter
int 21h
ENDM

;;
;;
;;
;;

Character output function
Bell character
Repeat number specified by macro
Call DOS

ENDM

WHILE Loops
The WHILE directive is similar to REPEAT, but the loop continues as long as a
given condition is true. The syntax is:
WHILE expression
statements
ENDM
The expression must be a value that can be calculated at assembly time.
Normally, the expression uses relational operators, but it can be any expression
that evaluates to zero (false) or nonzero (true). Usually, the condition changes
during the evaluation of the macro so that the loop won’t attempt to generate an
infinite amount of code. However, you can use the EXITM directive to break
out of the loop.
The following repeat block uses the WHILE directive to allocate variables
initialized to calculated values. This is a common technique for generating
lookup tables. (A lookup table is any list of precalculated results, such as a table
of interest payments or trigonometric values or logarithms. Programs optimized
for speed often use lookup tables, since calculating a value often takes more
time than looking it up in a table.)
cubes
LABEL
BYTE
;;
root
=
1
;;
cube
=
root * root * root ;;
WHILE
cube LE 32767
;;
WORD
cube
;;
root
=
root + 1
;;
cube
=
root * root * root
ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 242 of 18 Printed: 10/02/00 04:22 PM

Name the data generated
Initialize root
Calculate first cube
Repeat until result too large
Allocate cube
Calculate next root and cube

Chapter 9 Using Macros

243

FOR Loops and Variable-Length Parameters
With the FOR directive you can iterate through a list of arguments, working on
each of them in turn. It has the following syntax:
FOR parameter,
statements
ENDM
The parameter is a placeholder that represents the name of each argument
inside the FOR block. The argument list must contain comma-separated
arguments and must always be enclosed in angle brackets. Here’s an example of
a FOR block:
series LABEL
BYTE
FOR
arg, <1,2,3,4,5,6,7,8,9,10>
BYTE arg DUP (arg)
ENDM

On the first iteration, the arg parameter is replaced with the first argument, the
value 1. On the second iteration, arg is replaced with 2. The result is an array
with the first byte initialized to 1, the next 2 bytes initialized to 2, the next 3
bytes initialized to 3, and so on.
The argument list is given specifically in this example, but in some cases the list
must be generated as a text macro. The value of the text macro must include the
angle brackets.
arglist TEXTEQU >
%FOR arg, arglist
.
.
.
ENDM

; Generate list as text macro
; Do something to arg

Note the use of the literal character operator (!) to identify angle brackets as
characters, not delimiters. See “Text Delimiters (< >) and the Literal-Character
Operator,” earlier in this chapter.
The FOR directive also provides a convenient way to process macros with a
variable number of arguments. To do this, add VARARG to the last parameter
to indicate that a single named parameter will have the actual value of all
additional arguments. For example, the following macro definition includes the
three possible parameter attributes — required, default, and variable.
work

MACRO

rarg:REQ, darg:=<5>, varg:VARARG

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 243 of 19 Printed: 10/02/00 04:22 PM

244

Programmer’s Guide

The variable argument must always be last. If this macro is called with the
statement
work 4, , 6, 7, a, b

the first argument is received as the value 4, the second is replaced by the
default value 5, and the last four are received as the single argument <6, 7,
a, b>. This is the same format expected by the FOR directive. The FOR
directive discards leading spaces but recognizes trailing spaces.
The following macro illustrates variable arguments:
show

MACRO chr:VARARG
mov
ah, 02h
FOR arg,
mov
dl, arg
int
21h
ENDM

ENDM

When called with
show

‘O’, ‘K’, 13, 10

the macro displays each of the specified characters one at a time.
The parameter in a FOR loop can have the required or default attribute. You
can modify the show macro to make blank arguments generate errors:
show

MACRO chr:VARARG
mov
ah, 02h
FOR arg:REQ,
mov
dl, arg
int
21h
ENDM

ENDM

The macro now generates an error if called with
show

‘O’,, ‘K’, 13, 10

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 244 of 20 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

245

Another approach would be to use a default argument:
show

MACRO chr:VARARG
mov
ah, 02h
FOR arg:=<‘ ’>,
mov
dl, arg
int
21h
ENDM

ENDM

Now calling the macro with
show

‘O’,, ‘K’, 13, 10

inserts the default character, a space, for the blank argument.

FORC Loops
The FORC directive is similar to FOR, but takes a string of text rather than a
list of arguments. The statements are assembled once for each character
(including spaces) in the string, substituting a different character for the
parameter each time through.
The syntax looks like this:
FORC parameter, < text>
statements
ENDM
The text must be enclosed in angle brackets. The following example illustrates
FORC:
FORC arg,
BYTE
BYTE
BYTE
ENDM

‘&arg’
;; Allocate uppercase letter
‘&arg’ + 20h
;; Allocate lowercase letter
‘&arg’ - 40h
;; Allocate ordinal of letter

Notice that the substitution operator must be used inside the quotation marks to
make sure that arg is expanded to a character rather than treated as a literal
string.
With versions of MASM earlier than 6.0, FORC is often used for complex
parsing tasks. A long sentence can be examined character by character. Each
character is then either thrown away or pasted onto a token string, depending on
whether it is a separator character. The new predefined macro functions and
string processing directives discussed in the following section are usually more
efficient for these tasks.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 245 of 21 Printed: 10/02/00 04:22 PM

246

Programmer’s Guide

String Directives and Predefined Functions
The assembler provides four directives for manipulating text:
Directive

Description

SUBSTR

Assigns part of string to a new symbol.

INSTR

Searches for one string within another.

SIZESTR

Determines the size of a string.

CATSTR

Concatenates one or more strings to a single string.

These directives assign a processed value to a text macro or numeric equate. For
example, the following lines
num
newstr

=
CATSTR

7
<3 + >, %num, < = > , %3 + num ; "3 + 7 = 10"

assign the string "3 + 7 = 10" to newstr. CATSTR and SUBSTR assign
text in the same way as the TEXTEQU directive. SIZESTR and INSTR assign
a number in the same way as the = operator. The four string directives take only
text values as arguments. Use the expansion operator (%) when you need to
make sure that constants and numeric equates expand to text, as shown in the
preceding lines.
Each of the string directives has a corresponding predefined macro function
version: @SubStr, @InStr, @SizeStr, and @CatStr. Macro functions are
similar to the string directives, but you must enclose their arguments in
parentheses. Macro functions return text values and can appear in any context
where text is expected. The following section, “Returning Values with Macro
Functions,” tells how to write your own macro functions. The following
example is equivalent to the previous CATSTR example:
num
newstr

=
7
TEXTEQU @CatStr( <3 + >, %num, < = > , %3 + num )

Macro functions are often more convenient than their directive counterparts
because you can use a macro function as an argument to a string directive or to
another macro function. Unlike string directives, predefined macro function
names are case sensitive when you use the /Cp command-line option.
Each string directive and predefined function acts on a string, which can be any
textItem. The textItem can be text enclosed in angle brackets (< >), the name of
a text macro, or a constant expression preceded by % (as in %constExpr).
Refer to Appendix B, “BNF Grammar,” for a list of types that textItem can
represent.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 246 of 22 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

247

The following sections summarize the syntax for each of the string directives
and functions. The explanations focus on the directives, but the functions work
the same except where noted.

SUBSTR
name SUBSTR string, start[[, length]]
@SubStr( string, start[[, length]] )
The SUBSTR directive assigns a substring from a given string to the symbol
name. The start parameter specifies the position in string, beginning with 1, to
start the substring. The length gives the length of the substring. If you do not
specify length, SUBSTR returns the remainder of the string, including the start
character.

INSTR
name INSTR [[start,]] string, substring
@InStr( [[start]], string, substring )
The INSTR directive searches a specified string for an occurrence of substring
and assigns its position number to name. The search is case sensitive. The start
parameter is the position in string to start the search for substring. If you do not
specify start, it is assumed to be position 1, the start of the string. If INSTR
does not find substring, it assigns position 0 to name.
The INSTR directive assigns the position value name as if it were a numeric
equate. In contrast, the @InStr returns the value as a string of digits in the
current radix.
The @InStr function has a slightly different syntax than the INSTR directive.
You can omit the first argument and its associated comma from the directive.
You can leave the first argument blank with the function, but a blank function
argument must still have a comma. For example,
pos

INSTR

is the same as
pos

= @InStr( , , )

You can also assign the return value to a text macro, like this:

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 247 of 23 Printed: 10/02/00 04:22 PM

248

Programmer’s Guide
strpos

TEXTEQU @InStr( , , )

SIZESTR
name SIZESTR string
@SizeStr( string )
The SIZESTR directive assigns the number of characters in string to name. An
empty string returns a length of zero. The SIZESTR directive assigns the size
value to a name as if it were a numeric equate. The @SizeStr function returns
the value as a string of digits in the current radix.

CATSTR
name CATSTR string[, string]...
@CatStr( string[, string]... )
The CATSTR directive concatenates a list of text values into a single text value
and assigns it to name. TEXTEQU is technically a synonym for CATSTR.
TEXTEQU is normally used for single-string assignments, while CATSTR is
used for multistring concatenations.
The following example pushes and pops one set of registers, illustrating several
uses of string directives and functions:

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 248 of 24 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

249

; SaveRegs - Macro to generate a push instruction for each
; register in argument list. Saves each register name in the
; regpushed text macro.
regpushed TEXTEQU <>
;; Initialize empty string
SaveRegs MACRO regs:VARARG
LOCAL reg
FOR reg,
;; Push each register
push reg
;;
and add it to the list
regpushed CATSTR
, <,>, regpushed
ENDM
;; Strip off last comma
regpushed CATSTR , regpushed
;; Mark start of list with <
regpushed SUBSTR regpushed, 1, @SizeStr( regpushed )
regpushed CATSTR regpushed, >
;; Mark end with >
ENDM
; RestoreRegs - Macro to generate a pop instruction for registers
; saved by the SaveRegs macro. Restores one group of registers.
RestoreRegs MACRO
LOCAL reg
%FOR reg, regpushed
pop reg
ENDM
ENDM

;; Pop each register

Notice how the SaveRegs macro saves its result in the regpushed text macro
for later use by the RestoreRegs macro. In this case, a text macro is used as a
global variable. By contrast, the reg text macro is used only in RestoreRegs.
It is declared LOCAL so it won’t take the name reg from the global name
space. The MACROS.INC file provided with MASM 6.1 includes expanded
versions of these same two macros.

Returning Values with Macro Functions
A macro function is a named group of statements that returns a value. When
calling a macro function, you must enclose its argument list in parentheses, even
if the list is empty. The function always returns text.
MASM 6.1 provides several predefined macro functions for common tasks. The
predefined macros include @Environ (see page 10) and the string functions
@SizeStr, @CatStr, @SubStr, and @InStr (discussed in the preceding
section).
You define macro functions in exactly the same way as macro procedures,
except that a macro function always returns a value through the EXITM
directive. Here is an example:

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 249 of 25 Printed: 10/02/00 04:22 PM

250

Programmer’s Guide
DEFINED MACRO
symbol:REQ
IFDEF symbol
EXITM <-1>
ELSE
EXITM <0>
ENDIF
ENDM

;; True
;; False

This macro works like the defined operator in the C language. You can use it to
test the defined state of several different symbols with a single statement, as
shown here:
IF DEFINED( DOS ) AND NOT DEFINED( XENIX )
;; Do something
ENDIF

Notice that the macro returns integer values as strings of digits, but the IF
statement evaluates numeric values or expressions. There is no conflict because
the assembler sees the value returned by the macro function exactly as if the
user had typed the values directly into the program:
IF -1 AND NOT 0

Returning Values with EXITM
The return value must be text, a text equate name, or the result of another
macro function. A macro function must first convert a numeric value — such as
a constant, a numeric equate, or the result of a numeric expression — before
returning it. The macro function can use angle brackets or the expansion
operator (%) to convert numbers to text. The DEFINED macro, for instance,
could have returned its value as
EXITM

%-1

Here is another example of a macro function that uses the WHILE directive to
calculate factorials:

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 250 of 26 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

251

factorial
MACRO
num:REQ
LOCAL
i, factor
factor =
num
i
=
1
WHILE
factor GT 1
i
=
i * factor
factor =
factor - 1
ENDM
EXITM
%i
ENDM

The integer result of the calculation is changed to a text string with the expansion
operator (%). The factorial macro can define data, as shown here:
var

WORD

factorial( 4 )

This statement initializes var with the number 24 (the factorial of 4).

Using Macro Functions with Variable-Length Parameter Lists
You can use the FOR directive to handle macro parameters with the VARARG
attribute. “FOR Loops and Variable-Length Parameters,” page 242, explains
how to do this in simple cases where the variable parameters are handled
sequentially, from first to last. However, you may sometimes need to process
the parameters in reverse order or nonsequentially. Macro functions make these
techniques possible.
For example, the following macro function determines the number of arguments
in a VARARG parameter:
@ArgCount MACRO arglist:VARARG
LOCAL count
count = 0
FOR arg,
count = count + 1
ENDM
EXITM %count
ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 251 of 27 Printed: 10/02/00 04:22 PM

;; Count the arguments

252

Programmer’s Guide

You can use @ArgCount inside a macro that has a VARARG parameter, as
shown here:
work
MACRO args:VARARG
%
ECHO Number of arguments is: @ArgCount( args )
ENDM

Another useful task might be to select an item from an argument list using an
index to indicate the item. The following macro simplifies this.
@ArgI MACRO index:REQ, arglist:VARARG
LOCAL count, retstr
retstr TEXTEQU <>
;; Initialize count
count = 0
;; Initialize return string
FOR arg,
count = count + 1
IF count EQ index
;; Item is found
retstr TEXTEQU ;; Set return string
EXITM
;;
and exit IF
ENDIF
ENDM
EXITM retstr
;; Exit function
ENDM

You can use @ArgI like this:
work
MACRO args:VARARG
%
ECHO Third argument is: @ArgI( 3, args )
ENDM

Finally, you might need to process arguments in reverse order. The following
macro returns a new argument list in reverse order.
@ArgRev MACRO arglist:REQ
LOCAL txt, arg
txt TEXTEQU <>
%
FOR arg,
txt CATSTR , <,>, txt
ENDM
txt SUBSTR
txt CATSTR
EXITM txt

;; Paste each onto list

;; Remove terminating comma
txt, 1, @SizeStr( %txt ) - 1
, txt, >
;; Add angle brackets

ENDM

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 252 of 28 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

253

Here is an example showing @ArgRev in use:
work
MACRO
args:VARARG
%
FOR arg, @ArgRev( )
ECHO
arg
ENDM
ENDM

;; Process in reverse order

These three macro functions appear in the MACROS.INC include file, located
on one of the MASM distribution disks.

Expansion Operator in Macro Functions
This list summarizes the behavior of the expansion operator (%) with macro
functions.
u

If a macro function is preceded by a %, it will be expanded. However, if it
expands to a text macro or a macro function call, it will not expand further.
If you use a macro function call as an argument for another macro function
call, a % is not needed.
If a macro function is called inside angle brackets and is preceded by %, it
will be expanded.

Advanced Macro Techniques
The concept of replacing macro names with predefined macro text is simple in
theory, but it has many implications and complications. Here is a brief summary
of some advanced techniques you can use in macros.

Defining Macros within Macros
Macros can define other macros, a technique called “nesting macros.” MASM
expands macros as it encounters them, so nested macros are always processed
in nesting order. You cannot reference a nested macro directly in your program,
since the assembler begins expansion from the outer macro. In effect, a nested
macro is local to the macro that defines it. Only the amount of available memory
limits the number of macros a program can nest.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 253 of 29 Printed: 10/02/00 04:22 PM

254

Programmer’s Guide

The following example demonstrates how one macro can define another. The
macro takes as an argument the name of a shift or rotate instruction, then
creates another macro that simplifies the instruction for 8088/86 processors.
shifts MACRO
opname
;; Macro generates macros
opname&s
MACRO operand:REQ, rotates:=<1>
IF rotates LE 2
;; One at a time is faster
REPEAT rotate
;;
for 2 or less
opname operand, 1
ENDM
ELSE
;; Using CL is faster for
mov
cl, rotates
;;
more than 2
opname operand, cl
ENDIF
ENDM
ENDM

Recall that the 8086 processor allows only 1 or CL as an operand for shift and
rotate instructions. Expanding shifts generates a macro for the shift
instruction that uses whichever operand is more efficient. You create the entire
series of macros, one for each shift instruction, like this:
; Call
shifts
shifts
shifts
shifts
shifts
shifts
shifts
shifts

macro repeatedly to make new macros
ror
; Generates rors
rol
; Generates rols
shr
; Generates shrs
shl
; Generates shls
rcl
; Generates rcls
rcr
; Generates rcrs
sal
; Generates sals
sar
; Generates sars

Then use the new macros as replacements for shift instructions, like this:
shrs
rols

ax, 5
bx, 3

Testing for Argument Type and Environment
Macros can expand conditional blocks of code by testing for argument type with
the OPATTR operator. OPATTR returns a single word constant that indicates
the type and scope of an expression, like this:
OPATTR expression
If expression is not valid or is forward-referenced, OPATTR returns a 0.
Otherwise, the return value incorporates the bit flags shown in the table below.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 254 of 30 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

255

OPATTR serves as an enhanced version of the .TYPE operator, which returns
only the low byte (bits 0 – 7) shown in the table. Bits 11 – 15 of the return value
are undefined.
Bit

Set If expression

References a code label

Is a memory variable or has a relocatable data label

Is an immediate value

Uses direct memory addressing

Is a register value

References no undefined symbols and is without error

Is relative to SS

References an external label

8 – 10

Has the following language type:
u

000 — No language type

001 — C

010 — SYSCALL

011 — STDCALL

100 — Pascal

101 — FORTRAN

110 — Basic

A macro can use OPATTR to determine if an argument is a constant, a register,
or a memory operand. With this information, the macro can conditionally
generate the most efficient code depending on argument type.
For example, given a constant argument, a macro can test it for 0. Depending on
the argument’s value, the code can select the most effective method to load the
value into a register:
IF CONST
mov
bx, CONST
ELSE
sub
bx, bx
ENDIF

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 255 of 31 Printed: 10/02/00 04:22 PM

; If CONST > 0, move into BX
; More efficient if CONST = 0

256

Programmer’s Guide

The second method is faster than the first, yet has the same result (with the
byproduct of changing the processor flags).
The following macro illustrates some techniques using OPATTR by loading an
address into a specified offset register:
load

MACRO reg:REQ, adr:REQ
IF (OPATTR (adr)) AND 00010000y
;; Register
IFDIFI reg, adr
;; Don’t load register
mov
reg, adr
;;
onto itself
ENDIF
ELSEIF (OPATTR (adr)) AND 00000100y
mov
reg, adr
;; Constant
ELSEIF (TYPE (adr) EQ BYTE) OR (TYPE (adr) EQ SBYTE)
mov
reg, OFFSET adr
;; Bytes
ELSEIF (SIZE (TYPE (adr)) EQ 2
mov
reg, adr
;; Near pointer
ELSEIF (SIZE (TYPE (adr)) EQ 4
mov
reg, WORD PTR adr[0]
;; Far pointer
mov
ds, WORD PTR adr[2]
ELSE
.ERR
ENDIF

ENDM

A macro also can generate different code depending on the assembly
environment. The predefined text macro @Cpu returns a flag for processor
type. The following example uses the more efficient constant variation of the
PUSH instruction if the processor is an 80186 or higher.
IF

@Cpu AND 00000010y
pushc MACRO op
push op
ENDM

;; 80186 or higher

ELSE
pushc MACRO op
mov ax, op
push ax
ENDM
ENDIF

;; 8088/8086

Another macro can now use pushc rather than conditionally testing for
processor type itself. Although either case produces the same code, using pushc
assembles faster because the environment is checked only once.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 256 of 32 Printed: 10/02/00 04:22 PM

Chapter 9 Using Macros

257

You can test the language and operating system using the @Interface text
macro. The memory model can be tested with the @Model, @DataSize, or
@CodeSize text macros.
You can save the contexts inside macros with PUSHCONTEXT and
POPCONTEXT. The options for these keywords are:
Option

Description

ASSUMES

Saves segment register information

RADIX

Saves current default radix

LISTING

Saves listing and CREF information

CPU

Saves current CPU and processor

ALL

All of the above

Using Recursive Macros
Macros can call themselves. In MASM 5.1 and earlier, recursion is an important
technique for handling variable arguments. MASM 6.1 handles variable
arguments much more cleanly with the FOR directive and the VARARG
attribute, as described in “FOR Loops and Variable-Length Parameters,” earlier
in this chapter. However, recursion is still available and may be useful for some
macros.

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 257 of 33 Printed: 10/02/00 04:22 PM

256

Programmer’s Guide

Filename: LMAPGC09.DOC Project:
Template: MSGRIDA1.DOT Author: rick debroux Last Saved By: Ruth L Silverio
Revision #: 86 Page: 256 of 34 Printed: 10/02/00 04:22 PM

257

C H A P T E R

1 0

Writing a Dynamic-Link Library
For Windows

The Windows operating system relies heavily on service routines and data
contained in special libraries called “dynamic-link libraries,” or DLLs for short.
Most of what Windows comprises, from the collections of screen fonts to the
routines that handle the graphical interface, is provided by DLLs. MASM 6.1
contains tools that you can use to write DLLs in assembly language. This
chapter shows you how.
DLLs do not run under MS-DOS. The information in this chapter applies only
to Windows, drawing in part on the chapter “Writing a Module-Definition File”
in Environment and Tools. The acronym API, which appears throughout this
chapter, refers to the application programming interface that Windows provides
for programs. For documentation of API functions, see the Programmer’s
Reference, Volume 2 of the Windows Software Development Kit (SDK).
The first section of this chapter gives an overview of DLLs and their similarities
to normal libraries. The next section explores the parts of a DLL and the rules
you must follow to create one. The third section applies this information to an
example DLL.

Overview of DLLs
A dynamic-link library is similar to a normal run-time library. Both types of
libraries contain a collection of compiled procedures, which serve one or more
calling modules. To link a normal library, the linker copies the required functions
from the library file (which usually has a .LIB extension) and combines them
with other modules to form an executable program in .EXE format. This
process is called static linking.
In dynamic linking, the library functions are not copied to an .EXE file. Instead,
they reside in a separate file in executable form, ready to serve any calling
program, called a “client.” When the first client requires the library, Windows
takes care of loading the functions into memory and establishing linkage. If

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 257 of 1 Printed: 10/02/00 04:22 PM

258

Programmer’s Guide

subsequent clients also need the library, Windows dynamically links them with
the proper library functions already in memory.

Loading a DLL
How Windows loads a DLL affects the client rather than the DLL itself.
Accordingly, this section focuses on how to set up a client program to use a
DLL. Since the client can itself be a DLL, this is information a DLL
programmer should know. However, MASM 6.1 does not provide all the tools
required to create a stand-alone program for Windows. To create such a
program, called an “application,” you must use tools in the Windows SDK.
Windows provides two methods for loading a dynamic-link library into memory:
Method

Description

Implicit loading

Windows loads the DLL along with the first client program and links it
before the client begins execution.

Explicit loading

Windows does not load the DLL until the first client explicitly requests it
during execution.

When you write a DLL, you do not need to know beforehand which of the two
methods will be used to load the library. The loading method is determined by
how the client is written, not the DLL.

Implicit Loading
The implicit method of loading a DLL offers the advantage of simplicity. The
client requires no extra programming effort and can call the library functions as
if they were normal run-time functions. However, implicit loading carries two
constraints:
u
u

The name of the library file must have a .DLL extension.
You must either list all DLL functions the client calls in the IMPORTS
section of the client’s module-definition file, or link the client with an import
library.

An import library contains no executable code. It consists of only the names and
locations of exported functions in a DLL. The linker uses the locations in the
import library to resolve references to DLL functions in the client and to build
an executable header. For example, the file LIBW.LIB provided with MASM
6.1 is the import library for the DLL files that contain the Windows API
functions.
The IMPLIB utility described in Environment and Tools creates an import
library. Run IMPLIB from the MS-DOS command line like this:

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 258 of 2 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

259

IMPLIB implibfile dllfile
where implibfile is the name of the import library you want to create from the
DLL file dllfile. Once you have created an import library from a DLL, link it
with a client program that relies on implicit loading, but does not list imported
functions in its module-definition file. Continuing the preceding example, here’s
the link step for a client program that calls library procedures in the DLL dllfile:
LINK client.OBJ, client.EXE, , implibfile, client.DEF
This simplified example creates the client program client.EXE, linking it with the
import library implibfile, which in turn was created from the DLL file dllfile.
To summarize implicit loading, a client program must either
u
u

List DLL functions in the IMPORTS section of its module-definition file, or
Link with an import library created from the DLL.

Implicit loading is best when a client always requires at least one procedure in
the library, since Windows automatically loads the library with the client. If the
client does not always require the library service, or if the client must choose at
run time between several libraries, you should use explicit loading, discussed
next.

Explicit Loading
To explicitly load a DLL, the client does not require linking with an import
library, nor must the DLL file have an extension of .DLL. Explicit loading
involves three steps in which the client calls Windows API functions:
1. The client calls LoadLibrary to load the DLL.
2. The client calls GetProcAddress to obtain the address of each DLL function
it requires.
3. When finished with the DLL, the client calls FreeLibrary to unload the DLL
from memory.
The following example fragment shows how a client written in assembly
language explicitly loads a DLL called SYSINFO.DLL and calls the DLL
function GetSysDate.
INCLUDE
.DATA
hInstance
szDLL
szDate
lpProc

windows.inc
HINSTANCE 0
BYTE
'SYSINFO.DLL', 0
BYTE
'GetSysDate', 0
DWORD
0

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 259 of 3 Printed: 10/02/00 04:22 PM

260

Programmer’s Guide
.CODE
.
.
.
INVOKE
mov
INVOKE
mov
mov
call
.
.
.
INVOKE

LoadLibrary, ADDR szDLL
; Load SYSINFO.DLL
hInstance, ax
; Save instance count
GetProcAddress, ax, ADDR szDate ; Get and save
lpProc, ax
;
far address of
lpProc[2], dx
;
GetSysDate
lpProc
; Call GetSysDate

FreeLibrary, hInstance

; Unload SYSINFO.DLL

For simplicity, the above example contains no error-checking code. An actual
program should check all values returned from the API functions.
The explicit method of loading a DLL requires more programming effort in the
client program. However, the method allows the client to control which (if any)
dynamic-link libraries to load at run time.

Searching for a DLL File
To load a DLL, whether implicitly or explicitly, Windows searches for the DLL
file in the following directories in the order shown:
1. The current directory
2. The Windows directory, which contains WIN.COM
3. The Windows system directory, which contains system files such as
GDI.EXE
4. The directory where the client program resides (except Windows 3.0 and
earlier)
5. Directories listed in the PATH environment string
6. Directories mapped in a network
If Windows does not locate the DLL in any of these directories, it prompts the
user with a message box.

Building a DLL
A DLL has additional programming requirements beyond those for a normal
run-time library. This section describes the requirements pertaining to the
library’s code, data, and stack. It also discusses the effects of the library’s
extension name.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 260 of 4 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

261

DLL Code
The code in a DLL consists of exported and nonexported functions. Exported
functions, listed in the EXPORTS section of the module-definition file, are
public routines serving clients. Nonexported functions provide private, internal
support for the exported procedures. They are not visible to a client.
Under Windows, an exported library routine must appear to the caller as a far
procedure. Your DLL routines can use any calling convention you wish,
provided the caller assumes the same convention. You can think of dynamiclink code as code for a normal run-time library with the following additions:
u
u
u

An entry procedure
A termination procedure
Special prologue and epilogue code

Entry Procedure
A DLL, like any Windows-based program, must have an entry procedure.
Windows calls the entry procedure only once when it first loads the DLL,
passing the following information in registers:
u
u
u

DS contains the library’s data segment address.
DI holds the library’s instance handle.
CX holds the library’s heap size in bytes.

Note Windows API functions destroy all registers except DI, SI, BP, DS, and
the stack pointer. To preserve the contents of other registers, your program
must save the registers before an API call and restore them afterwards.
This information corresponds to the data provided to an application. Since a
DLL has only one occurrence in memory, called an “instance,” the value in DI
is not usually important. However, a DLL can use its instance handle to obtain
resources from its own executable file.
The entry procedure does not need to record the address of the data segment.
Windows automatically ensures that each exported routine in the DLL has
access to the library’s data segment, as explained in “Prologue and Epilogue
Code,” on page 264.
The heap size contained in CX reflects the value provided in the HEAPSIZE
statement of the module-definition file. You need not make an accurate guess in
the HEAPSIZE statement about the library’s heap requirements, provided you
specify a moveable data segment. With a moveable segment, Windows
automatically allocates more heap when needed. However, Windows can

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 261 of 5 Printed: 10/02/00 04:22 PM

262

Programmer’s Guide

provide no more heap in a fixed data segment than the amount specified in the
HEAPSIZE statement. In any case, a library’s total heap cannot exceed 64K,
less the amount of static data. Static data and heap reside in the same segment.
Windows does not automatically deallocate unneeded heap while the DLL is in
memory. Therefore, you should not set an unnecessarily large value in the
HEAPSIZE statement, since doing so wastes memory.
The entry procedure calls the Windows API function LocalInit to allocate the
heap. The library must create a heap before its routines call any heap functions,
such as LocalAlloc. The following example illustrates these steps:
DLLEntry PROC FAR PASCAL PUBLIC

@@:

jcxz
INVOKE
.IF
INVOKE
call
mov
.ENDIF
ret

@F
LocalInit, ds, 0, cx
( ax )
UnlockSegment, -1
LibMain
ax, TRUE

; Entry point for DLL
;
;
;
;
;
;
;
;

If no heap, skip
Else set up the heap
If successful,
unlock the data segment
Call DLL's data init routine
Return AX = 1 if okay,
else if LocalInit error,
return AX = 0

DLLEntry ENDP

This example code is taken from the DLLENTRY.ASM module, contained in
the LIB subdirectory on one of the MASM 6.1 distribution disks. After
allocating the heap, the procedure calls the library’s initialization procedure —
called LibMain in this case. LibMain initializes the library’s static data (if
required), then returns to DLLEntry, which returns to Windows. If Windows
receives a return value of 0 (FALSE) from DLLEntry, it unloads the library and
displays an error message.
The process is similar to the way MS-DOS loads a terminate-and-stay-resident
program (TSR), described in the next chapter. Both the DLL and TSR return
control immediately to the operating system, then wait passively in memory to
be called.
The following section explains how a DLL gains control when Windows unloads
it from memory.

Termination Procedure
Windows maintains a DLL in memory until the last client program terminates or
explicitly unloads the library. When unloading a DLL, Windows first calls the
library’s termination procedure. This allows the DLL to return resources and do
any necessary cleanup operations before Windows unloads the library from
memory.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 262 of 6 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

263

Libraries that have registered window procedures with RegisterClass need not
call UnregisterClass to remove the class registration. Windows does this
automatically when it unloads the library.
You must name the library’s termination procedure WEP (for Windows Exit
Procedure) and list it in the EXPORTS section of the library’s module-definition
file. To ensure immediate operation, provide an ordinal number and use the
RESIDENTNAME keyword, as described in the chapter “Creating ModuleDefinition Files” in Environment and Tools. This keeps the name “WEP” in the
Windows-resident name table at all times.
Besides its name, the code for WEP should also remain constantly in memory.
To ensure this, place WEP in its own code segment and set the segment’s
attributes as PRELOAD FIXED in the SEGMENTS statement of the moduledefinition file. Thus, your DLL code should use a memory model that allows
multiple code segments, such as medium model. Since a termination procedure
is usually short, keeping it resident in memory does not burden the operating
system.
The termination procedure accepts a single parameter, which can have one of
two values. These values are assigned to the following symbolic constants in the
WINDOWS.INC file located in the LIB subdirectory:
u

WEP_SYSTEM_EXIT (value 1) indicates Windows is shutting down.

WEP_FREE_DLL (value 0) indicates the library’s last client has terminated or

has called FreeLibrary, and Windows is unloading the DLL.
The following fragment provides an outline for a typical termination procedure:
WEP

PROC FAR PASCAL EXPORT
wExitCode:WORD
Prolog
;
.IF
wExitCode == WEP_FREE_DLL
;
.
;
.
;
.
ELSEIF wExitCode == WEP_SYSTEM_EXIT
.
;
.
;
.
. ENDIF
;
;
mov
ax, TRUE
;
Epilog
;
ret
;

WEP

ENDP

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 263 of 7 Printed: 10/02/00 04:22 PM

Prologue macro,
discussed below
Get ready to
unload

Windows is
shutting down
If neither value,
take no action
Always return AX = 1
Epilogue code,
discussed below

264

Programmer’s Guide

Usually, the WEP procedure takes the same actions regardless of the parameter
value, since in either case Windows will unload the DLL.
Under Windows 3.0, the WEP procedure receives stack space of about 256
bytes. This allows the procedure to unhook interrupts, but little else. Any other
action, such as calling an API function, usually results in an unrecoverable
application error because of stack overflow. Later versions of Windows provide
at least 4K of stack to the WEP procedure, allowing it to call many API
functions.
However, WEP should not send or post a message to a client, because the client
may already be terminated. The WEP procedure should also not attempt file I/O,
since only application processes — not DLLs — can own files. When control
reaches WEP, the client may no longer exist and its files are closed.

Prologue and Epilogue Code
Exported procedures in a Windows-based program require special epilogue and
prologue code. (For a definition of these terms, see “Generating Prologue and
Epilogue Code” in Chapter 7.) The SAMPLES subdirectory on one of the
MASM 6.1 distribution disks contains macros you can use for far procedures in
your Windows-based programs. Here’s a listing of the prologue macro:
Prolog

MACRO
mov
nop
inc
push
mov
push
mov
ENDM

ax, ds
bp
bp
bp, sp
ds
ds, ax

;
;
;
;
;
;
;

Must be 1st, since Windows overwrites
Placeholder for 3rd byte
Push odd BP. Not required, but
allows CodeView to recognize frame
Set up stack frame to access params
Save DS
Point DS to DLL's data segment

The instruction
inc

marks the beginning of the stack frame with an odd number. This allows realmode Windows to locate segment addresses on the stack and update the
addresses when it moves or discards the corresponding segments. In protected
mode, selector values do not change when segments are moved, so marking the
stack frame is not required. However, certain debugging applications, such as
Microsoft Codeview for Windows and the Microsoft Windows 80386 Debugger
(both documented in Programming Tools of the SDK), search for a marked
frame to determine if the frame belongs to a far procedure. Without the mark,
these debuggers give meaningless information when backtracing through the
stack. Therefore, you should include the INC BP instruction for Windows-

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 264 of 8 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

265

based programs that may run in real mode or that require debugging with a
Microsoft debugger.
Another characteristic of the prologue macro may seem puzzling at first glance.
The macro moves DS into AX, then AX back into DS. This sequence of
instructions lets Windows selectively overwrite the prologue code in far
procedures. When Windows loads a program, it compares the names of far
procedures with the list of exported procedures in the module-definition file. For
procedures that do not appear on the list, Windows leaves their prologue code
untouched. However, Windows overwrites the first 3 bytes of all exported
procedures with
mov

ax, DGROUP

where DGROUP represents the selector value for the library’s data segment.
This explains why the prologue macro reserves the third byte with a NOP
instruction. The 1-byte instruction serves as padding to provide a 3-byte area for
Windows to overwrite.
The epilogue code returns BP to normal, like this:
Epilog

MACRO
pop
pop
dec
ENDM

ds
bp
bp

; Recover original DS
;
and BP+1
; Reset to original BP

DLL Data
A DLL can have its own local data segment up to 64K. Besides static data, the
segment contains the heap from which a procedure can allocate memory
through the LocalAlloc API function. You should minimize static data in a
DLL to reserve as much memory as possible for temporary allocations.
Furthermore, all procedures in the DLL draw from the same heap space. If
more than one procedure in the library accesses the heap, a procedure should
not hold allocated space unnecessarily at the expense of the other procedures.
A Windows-based program must reserve a “task header” in the first 16 bytes of
its data segment. If you link your program with a C run-time function, the C
startup code automatically allocates the task header. Otherwise, you must
explicitly reserve and initialize the header with zeros. The sample program
described in “Example of a DLL:SYSINFO,” page 267, shows how to allocate a
task header.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 265 of 9 Printed: 10/02/00 04:22 PM

266

Programmer’s Guide

DLL Stack
A DLL does not declare a stack segment and does not allocate stack space. A
client program calls a library’s exported procedure through a simple far call, and
the stack does not change. The procedure is, in effect, part of the calling
program, and therefore uses the caller’s stack.
This simple arrangement differs from that used in small and medium models, in
which many C run-time functions accept near pointers as arguments. Such
functions assume the pointer is relative to the current data segment. In
applications, the call works even if the argument points to a local variable on the
stack, since DS and SS contain the same segment address.
However, in a DLL, DS and SS point to different segments. Under small and
medium models, a library procedure must always pass pointers to static
variables located in the data segment, not to local variables on the stack.
When you write a DLL, include the FARSTACK keyword with the .MODEL
directive, like this:
.MODEL

small, pascal, farstack

This informs the assembler that SS points to a segment other than DGROUP.
With full segment definitions, also add the line:
ASSUME

DS:DGROUP, SS:NOTHING

DLL Extension Names
You can name an explicitly-loaded DLL file with any extension. The many files
in your Windows directory with extensions such as .DRV and .FON are almost
certainly DLLs. Many DLLs have an .EXE extension, though they are not true
executable files.
A library with an .EXE extension should always include stub code, specified by
the STUB statement in the module-definition file. The stub code activates when
run under MS-DOS, usually displaying a message to inform the user that the
program requires Windows. Without the stub code, the system hangs if a user
attempts to run a DLL with an .EXE extension.
Do not name a DLL with a .COM extension, since MS-DOS will give control to
the first byte of the program header. The header does not contain executable
instructions, and the system will hang even if the DLL has stub code.

Summary
Following is a summary of the previous information in this chapter.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 266 of 10 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

267

A dynamic-link library has only one instance — that is, it can load only once
during a Windows session.

A single DLL can service calls from many client programs. Windows takes
care of linkage between the DLL and each client.

Windows loads a DLL either implicitly (along with the first client) or
explicitly (when the first client calls LoadLibrary). It unloads the DLL when
the last client either terminates or calls FreeLibrary.

A client calls a DLL routine as a simple far procedure. The routine can use
any calling convention.

Windows ensures that the first instruction in a DLL procedure moves the
address of the library’s data segment into AX. You must provide the proper
prologue code to allow space for this 3-byte instruction and to copy AX to
DS.
All procedures in a DLL have access to a single common data segment. The
segment contains both static variables and heap space, and cannot exceed
64K.
A DLL procedure uses the caller’s stack.
All exported procedures in a DLL must appear in the EXPORTS list in the
library’s module-definition file.

u
u

Example of a DLL: SYSINFO
Like any library, a DLL should be as small and fast as possible — a good
argument for writing it in assembly language. This section describes an example
library called SYSINFO, written entirely in assembly language. The following
text applies previous information in this chapter to an actual DLL.
SYSINFO contains three callable procedures. The acronym ASCIIZ refers to a
string of ASCII characters terminated with a zero. The callable procedures are:
Procedure

Description

GetSysTime

Returns a far pointer to a 12-byte ASCIIZ string containing the current time
in hh:mm:ss format.

GetSysDate

Returns a far pointer to an ASCIIZ string containing the current date in any
of six languages.

GetSysInfo

Returns a far pointer to a structure containing the following system data:
u

ASCIIZ string of Windows version

ASCIIZ string of MS-DOS version

Current keyboard status

Current video mode

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 267 of 11 Printed: 10/02/00 04:22 PM

268

Programmer’s Guide
u

Math coprocessor flag

Processor type

ASCIIZ string of ROM-BIOS release date

To see SYSINFO in action, follow the steps below. The file SYSDATA.EXE
resides in the SAMPLES\WINDLL subdirectory of MASM if you requested
example files when installing MASM. Otherwise, you must first install the file
with the MASM 6.1 SETUP utility.
u

Create SYSINFO.DLL as described in the following section and place it in
the SAMPLES\WINDLL subdirectory for MASM 6.1.

From the Windows File Manager, make the SAMPLES\WINDLL
subdirectory the current directory.
In the Program Manager, choose Run from the File menu and type
SYSDATA
to run the example program SYSDATA.EXE. This program calls the
routines in SYSINFO.DLL and displays the returned data.

Entry Routine for SYSINFO
SYSINFO links with the DLLENTRY module, which serves as the library’s
entry point when Windows first loads the program. For a listing and description
of DLLENTRY.ASM, see the previous section, “Entry Procedure.”
DLLENTRY replaces the LIBENTRY module provided with the Windows
SDK, but unlocks the data segment after calling the API function InitTask.
LIBENTRY does not unlock the segment. DLLENTRY saves some space over
LIBENTRY, because it does not pass any arguments to LibMain.
The LibMain procedure handles the library’s initialization tasks. You can name
the procedure whatever you want, provided you make the same change in
DLLENTRY.ASM and reassemble both modules. You can even combine
DLLENTRY with LibMain to form one procedure, like this:

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 268 of 12 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows
DLLInit PROC FAR PASCAL PUBLIC

@@:

jcxz
INVOKE
.IF
INVOKE
.
.
.
mov
.ENDIF
ret

@F
LocalInit, ds, 0, cx
( ax )
UnlockSegment, -1

ax, TRUE

269

; Entry point for DLL
;
;
;
;
;
;
;
;
;
;

If no heap, skip
Else set up the heap
If successful,
unlock the data segment
Initialize DLL data. This
replaces the call to the
LibMain procedure.
Return AX = 1 if okay,
else if LocalInit error,
return AX = 0

DLLInit ENDP
END
DLLInit

Whatever you call your combined procedure (DLLInit in the preceding
example), place the name on the END statement as shown. This identifies the
procedure as the one that first executes when Windows loads the DLL.
SYSINFO accommodates several international languages. Currently, SYSINFO
recognizes English, French, Spanish, German, Italian, and Swedish, but you can
easily extend the code to include other languages. LibMain calls
GetProfileString to determine the current language, then initializes the variable
indx accordingly. The variable indirectly points to an array of strings containing
days and months in different languages. The GetSysDate procedure uses these
strings to create a full date in the correct language.

Static Data
SYSINFO stores the strings in its static data segment. This data remains in
memory along with the library’s code. All procedures have equal access to the
data segment.
Because the library does not call any C run-time functions, it explicitly allocates
the low paragraph of the data segment with the variable TaskHead. This 16byte area serves as the required Windows task header, described in “DLL
Data,” earlier in this chapter.

Module-Definition File
The library’s module-definition file, named SYSINFO.DEF, looks like this:

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 269 of 13 Printed: 10/02/00 04:22 PM

270

Programmer’s Guide
LIBRARY
DESCRIPTION
EXETYPE
CODE
DATA
SEGMENTS CODE2
EXPORTS

SYSINFO
'Sample assembly-language DLL'
WINDOWS
PRELOAD MOVEABLE DISCARDABLE
PRELOAD MOVEABLE SINGLE
PRELOAD FIXED
WEP
@1
RESIDENTNAME
GetSysTime
@2
GetSysDate
@3
GetSysInfo
@4

Note the following points about the module-definition file:
u
u

u
u

The LIBRARY statement identifies SYSINFO as a dynamic-link library.
SYSINFO places its termination procedure WEP in a separate code segment,
called CODE2, which the SEGMENTS statement declares as FIXED. This
keeps the WEP routine fixed in memory, while all other code remains
moveable.
The EXPORTS section lists all procedures the library exports, including WEP.
None of the library’s procedures require heap space, so SYSINFO.DEF
includes no HEAPSIZE statement.

Assembling and Linking SYSINFO
The following listing shows the description file for SYSINFO:
sysinfo.obj: sysinfo.asm dll.inc
ML /c /W3 sysinfo.asm
dllentry.obj: dllentry.asm dll.inc.
ML /c /W3 dllentry.asm
sysinfo.dll: dllentry.obj sysinfo.obj
LINK dllentry sysinfo, sysinfo.dll,, libw.lib mnocrtdw.lib,
sysinfo.def

To create SYSINFO.DLL, run the NMAKE utility described in Chapter 16 of
Environments and Tools. Or assemble and link SYSINFO with the three
command lines shown in the preceding listing. This does not require running
NMAKE.
SYSINFO links with the library modules MNOCRTDW.LIB and LIBW.LIB.
The first supplies the required Windows startup code for a medium-model DLL
that does not use any C run-time functions. LIBW.LIB is the Windows import
library, which contains no executable code. The import library provides linkage
information for the Windows API functions referenced in the DLL. Windows
establishes the final links when it loads the program.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 270 of 14 Printed: 10/02/00 04:22 PM

Chapter 10 Writing a Dynamic-Link Library for Windows

271

Expanding SYSINFO
SYSINFO is an example of how to write an assembly-language DLL without
overwhelming detail. It has plenty of room for expansion and improvements.
The following list may give you some ideas:
u

To create a heap area for the library, add the line
HEAPSIZE value

to the module-definition file, where value is an approximate guess for the
amount of heap required in bytes. The DLLEntry procedure automatically
allocates the indicated amount of heap. Keep the data segment moveable,
because Windows then provides more heap space if required by the DLL
procedures.
If you want to add a procedure that calls C run-time functions, you must
replace MNOCRTDW.LIB with MDLLCW.LIB, which is supplied with the
Windows SDK. The MDLLCW.LIB library contains the run-time functions
for medium-model DLLs.
Each time the GetSysInfo procedure is called, it retrieves the version
number of MS-DOS and Windows, gets the processor type, checks for a
coprocessor, and reads the ROM-BIOS release date. Since this information
does not change throughout a Windows session, it would be handled more
efficiently in the LibMain procedure, which executes only once. The code is
currently placed in GetSysInfo for the sake of clarity at the expense of
efficiency.
SYSINFO is not a true international program. You can easily add more
languages, extending the days and months arrays accordingly. Moreover,
for the sake of simplicity, the GetSysDate procedure arranges the date with
an American bias. For example, in many parts of the world, the date numeral
appears before the month rather than after. If you use SYSINFO in your
own applications, you should include code in LibMain to determine the
correct date format with additional calls to GetProfileString. You can find
more information on how to do this in Chapter 18 of the Microsoft Windows
Programmer’s Reference, Volume 1, supplied with the Windows SDK.

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 271 of 15 Printed: 10/02/00 04:22 PM

272

Programmer’s Guide

Filename: LMAPGC10.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 62 Page: 272 of 16 Printed: 10/02/00 04:22 PM

273

C H A P T E R

1 1

Writing Memory-Resident Software

Through its memory-management system, MS-DOS allows a program to remain
resident in memory after terminating. The resident program can later regain
control of the processor to perform tasks such as background printing or
“popping up” a calculator on the screen. Such a program is commonly called a
TSR, from the terminate-and-stay-resident function it uses to return to MSDOS.
This chapter explains the techniques of writing memory-resident software. The
first two sections present introductory material. Following sections describe
important MS-DOS and BIOS interrupts and focus on how to write safe,
compatible, memory-resident software. Two example programs illustrate the
techniques described in the chapter. The MASM 6.1 disks contain complete
source code for the two example TSR programs.

Terminate-and-Stay-Resident Programs
MS-DOS maintains a pointer to the beginning of unused memory. Programs
load into memory at this position and terminate execution by returning control to
MS-DOS. Normally, the pointer remains unchanged, allowing MS-DOS to reuse
the same memory when loading other programs.
A terminating program can, however, prevent other programs from loading on
top of it. These programs exit to MS-DOS through the terminate-and-stayresident function, which resets the free-memory pointer to a higher position.
This leaves the program resident in a protected block of memory, even though it
is no longer running.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 273 of 1 Printed: 10/02/00 04:22 PM

274

Programmer’s Guide

The terminate-and-stay-resident function (Function 31h) is one of the MS-DOS
services invoked through Interrupt 21h. The following fragment shows how a
TSR program terminates through Function 31h and remains resident in a 1000hbyte block of memory:
mov
mov
mov

ah, 31h
al, err
dx, 100h

int

21h

;
;
;
;
;

Request DOS Function 31h
Set return code
Reserve 100h paragraphs
(1000h bytes)
Terminate-and-stay-resident

Note In current versions of MS-DOS, Interrupt 27h also provides a terminateand-stay-resident service. However, Microsoft cannot guarantee future support
for Interrupt 27h and does not recommend its use.

Structure of a TSR
TSRs consist of two distinct parts that execute at different times. The first part
is the installation section, which executes only once, when MS-DOS loads the
program. The installation code performs any initialization tasks required by the
TSR and then exits through the terminate-and-stay-resident function.
The second part of the TSR, called the resident section, consists of code and
data left in memory after termination. Though often identified with the TSR
itself, the resident section makes up only part of the entire program.
The TSR’s resident code must be able to regain control of the processor and
execute after the program has terminated. Methods of executing a TSR are
classified as either passive or active.

Passive TSRs
The simplest way to execute a TSR is to transfer control to it explicitly from
another program. Because the TSR in this case does not solicit processor
control, it is said to be passive. If the calling program can determine the TSR’s
memory address, it can grant control via a far jump or call. More commonly, a
program activates a passive TSR through a software interrupt. The installation
section of the TSR writes the address of its resident code to the proper position
in the interrupt vector table (see “MS-DOS Interrupts” in Chapter 7). Any
subsequent program can then execute the TSR by calling the interrupt.
Passive TSRs often replace existing software interrupts. For example, a passive
TSR might replace Interrupt 10h, the BIOS video service. By intercepting calls
that read or write to the screen, the TSR can access the video buffer directly,
increasing display speed.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 274 of 2 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

275

Passive TSRs allow limited access since they can be invoked only from another
program. They have the advantage of executing within the context of the calling
program, and thus run no risk of interfering with another process. Such a risk
does exist with active TSRs.

Active TSRs
The second method of executing a TSR involves signaling it through some
hardware event, such as a predetermined sequence of keystrokes. This type of
TSR is “active” because it must continually search for its startup signal. The
advantage of active TSRs lies in their accessibility. They can take control from
any running application, execute, and return, all on demand.
An active TSR, however, must not seize processor control blindly. It must
contain additional code that determines the proper moment at which to execute.
The extra code consists of one or more routines called “interrupt handlers,”
described in the following section.

Interrupt Handlers in Active TSRs
The memory-resident portion of an active TSR consists of two parts. One part
contains the body of the TSR — the code and data that perform the program’s
main tasks. The other part contains the TSR’s interrupt handlers.
An interrupt handler is a routine that takes control when a specific interrupt
occurs. Although sometimes called an “interrupt service routine,” a TSR’s
handler usually does not service the interrupt. Instead, it passes control to the
original interrupt routine, which does the actual interrupt servicing. (See the
section “Replacing an Interrupt Routine” in Chapter 7 for information on how to
write an interrupt handler.)
Collectively, interrupt handlers ensure that a TSR operates compatibly with the
rest of the system. Individually, each handler fulfills one or more of the
following functions:
u
u
u

Auditing hardware events that may signal a request for the TSR
Monitoring system status
Determining whether a request for the TSR should be honored, based on
current system status

Auditing Hardware Events for TSR Requests
Active TSRs commonly use a special keystroke sequence or the timer as a
request signal. A TSR invoked through one of these channels must be equipped
with handlers that audit keyboard or timer events.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 275 of 3 Printed: 10/02/00 04:22 PM

276

Programmer’s Guide

A keyboard handler receives control at every keystroke. It examines each key,
searching for the proper signal or “hot key.” Generally, a keyboard handler
should not attempt to call the TSR directly when it detects the hot key. If the
TSR cannot safely interrupt the current process at that moment, the keyboard
handler is forced to exit to allow the process to continue. Since the handler
cannot regain control until the next keystroke, the user has to press the hot key
repeatedly until the handler can comply with the request.
Instead, the handler should merely set a request flag when it detects a hot-key
signal and then exit normally. Examples in the following paragraphs illustrate this
technique.
For computers other than MCA (IBM PS/2 and compatible), an active TSR
audits keystrokes through a handler for Interrupt 09, the keyboard interrupt:
Keybrd

PROC
sti
push
in
call
.IF
mov
.
.
.

FAR
ax
al, 60h
CheckHotKey
carry?
cs:TsrRequestFlag, TRUE

;
;
;
;
;
;
;

Interrupts are okay
Save AX register
AL = key scan code
Check for hot key
If hot key pressed,
raise flag and
set up for exit

A TSR running on a PS/2 computer cannot reliably read key scan codes using
this method. Instead, the TSR must search for its hot key through a handler for
Interrupt 15h (Miscellaneous System Services). The handler determines the
current keypress from the AL register when AH equals 4Fh, as shown here:
MiscServ PROC
sti
.IF
call
.IF
mov
.
.
.

FAR
ah == 4Fh
CheckHotKey
carry?
cs:TsrRequestFlag, TRUE

;
;
;
;
;
;

Interrupts okay
If Keyboard Intercept Service:
Check for hot key
If hot key pressed,
raise flag and
set up for exit

The example program on page 293 shows how a TSR tests for a PS/2 machine
and then sets up a handler for either Interrupt 09 or Interrupt 15h to audit
keystrokes.
Setting a request flag in the keyboard handler allows other code, such as the
timer handler (Interrupt 08), to recognize a request for the TSR. The timer
handler gains control at every timer interrupt, which occurs an average of 18.2
times per second.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 276 of 4 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

277

The following fragment shows how a timer handler tests the request flag and
continually polls until it can safely execute the TSR.
NewTimer PROC FAR
.
.
.
cmp
TsrRequestFlag, FALSE
.IF
!zero?
call
CheckSystem
.IF
carry?
call
ActivateTsr
.
.
.

; Has TSR been requested?
; If so, can system be
;
interrupted safely?
; If so,
;
activate TSR

Monitoring System Status
A TSR that uses a hardware device such as the video or disk must not interrupt
while the device is active. A TSR monitors a device by handling the device’s
interrupt. Each interrupt handler simply sets a flag to indicate the device is in
use, and then clears the flag when the interrupt finishes.
The following shows a typical monitor handler:
NewHandler PROC
FAR
mov
cs:ActiveFlag, TRUE
pushf
call
mov
iret
NewHandler ENDP

;
;
;
OldHandler
;
cs:ActiveFlag, FALSE ;
;

Set active flag
Simulate interrupt by
pushing flags, then
far-calling original routine
Clear active flag
Return from interrupt

Only hardware used by the TSR requires monitoring. For example, a TSR that
performs disk input/output (I/O) must monitor disk use through Interrupt 13h.
The disk handler sets an active flag that prevents the TSR from executing during
a read or write operation. Otherwise, the TSR’s own I/O would move the disk
head. This would cause the suspended disk operation to continue with the head
incorrectly positioned when the TSR returned control to the interrupted
program.
In the same way, an active TSR that displays to the screen must monitor calls to
Interrupt 10h. The Interrupt 10h BIOS routine does not protect critical sections
of code that program the video controller. The TSR must therefore ensure it
does not interrupt such nonreentrant operations.
The activities of the operating system also affect the system status. With few
exceptions, MS-DOS functions are not reentrant and must not be interrupted.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 277 of 5 Printed: 10/02/00 04:22 PM

278

Programmer’s Guide

However, monitoring MS-DOS is somewhat more complicated than monitoring
hardware. This subject is discussed in “Using MS-DOS in Active TSRs,” later
in this chapter.
Figure 11.1 illustrates the process described so far. It shows a time line for a
typical TSR signaled from the keyboard. When the keyboard handler detects the
proper hot key, it sets a request flag called TsrRequestFlag. Thereafter, the
timer handler continually checks the system status until it can safely call the
TSR.

Figure 11.1

Time Line of Interactions Between Interrupt Handlers for a Typical TSR

The following comments describe the chain of events depicted in Figure 11.1.
Each comment refers to one of the numbered pointers in the figure.
1. At time = t, the timer handler activates. It finds the flag TsrRequestFlag
clear, indicating the user has not requested the TSR. The handler terminates
without taking further action. Notice that Interrupt 13h is currently
processing a disk I/O operation.
2. Before the next timer interrupt, the keyboard handler detects the hot key,
signaling a request for the TSR. The keyboard handler sets
TsrRequestFlag and returns.
3. At time = t + 1/18 second, the timer handler again activates and finds
TsrRequestFlag set. The handler checks other active flags to determine if
the TSR can safely execute. Since Interrupt 13h has not yet completed its
disk operation, the timer handler finds DiskActiveFlag set. The handler
therefore terminates without activating the TSR.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 278 of 6 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

279

4. At time = t + 2/18 second, the timer handler again finds TsrRequestFlag
set and repeats its scan of the active flags. DiskActiveFlag is now clear,
but in the interim, Interrupt 10h has activated as indicated by the flag
VideoActiveFlag. The timer handler accordingly terminates without
activating the TSR.
5. At time = t + 3/18 second, the timer handler repeats the process. This time it
finds all active flags clear, indicating the TSR can safely execute. The timer
handler calls the TSR, which sets its own active flag to ensure it will not
interrupt itself if requested again.
6. The timer and other interrupts continue to function normally while the TSR
executes.
The timer itself can serve as the startup signal if the TSR executes periodically.
Screen clocks that continuously show seconds and minutes are examples of
TSRs that use the timer this way. ALARM.ASM, a program described in the
next section, shows another example of a timer-driven TSR.

Determining Whether to Invoke the TSR
Once a handler receives a request signal for the TSR, it checks the various
active flags maintained by the handlers that monitor system status. If any of the
flags are set, the handler ignores the request and exits. If the flags are clear, the
handler invokes the TSR, usually through a near or far call. Figure 11.1
illustrates how a timer handler detects a request and then periodically scans
various active flags until all the flags are clear.
A TSR that changes stacks must not interrupt itself. Otherwise, the second
execution would overwrite the stack data belonging to the first. A TSR prevents
this by setting its own active flag before executing, as shown in Figure 11.1. A
handler must check this flag along with the other active flags when determining
whether the TSR can safely execute.

Example of a Simple TSR: ALARM
This section presents a simple alarm clock TSR that demonstrates some of the
material covered so far. The program accepts an argument from the command
line that specifies the alarm setting in military form, such as 1635 for 4:35 P.M.
For simplicity, the argument must consist of four digits, including leading zeros.
To set the alarm at 7:45 A.M. , for example, enter the command:
ALARM 0745

The installation section of the program begins with the Install procedure.
Install computes the number of five-second intervals that must elapse before
the alarm sounds and stores this number in the word CountDown. The

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 279 of 7 Printed: 10/02/00 04:22 PM

280

Programmer’s Guide

procedure then obtains the vector for Interrupt 08 (timer) through MS-DOS
Function 35h and stores it in the far pointer OldTimer. Function 25h replaces
the vector with the far address of the new timer handler NewTimer. Once
installed, the new timer handler executes at every timer interrupt. These
interrupts occur 18.2 times per second or 91 times every five seconds.
Each time it executes, NewTimer subtracts one from a secondary counter called
Tick91. By counting 91 timer ticks, Tick91 accurately measures a period of
five seconds. When Tick91 reaches zero, it’s reset to 91 and CountDown is
decremented by one. When CountDown reaches zero, the alarm sounds.
;*
;*
;*
;*
;*
;*
;*

ALARM.ASM - A simple memory-resident program that beeps the speaker
at a prearranged time. Can be loaded more than once for multiple
alarm settings. During installation, ALARM establishes a handler
for the timer interrupt (Interrupt 08). It then terminates through
the terminate-and-stay-resident function (Function 31h). After the
alarm sounds, the resident portion of the program retires by setting
a flag that prevents further processing in the handler.

.MODEL tiny
.STACK
.CODE
ORG
5Dh
CountDown
LABEL

; Create ALARM.COM

WORD

.STARTUP
jmp
Install

; Location of time argument in PSP,
;
converted to number of 5-second
;
intervals to elapse
; Jump over data and resident code

; Data must be in code segment so it won’t be thrown away with Install code.
OldTimer
DWORD
?
; Address of original timer routine
tick_91
BYTE
91
; Counts 91 clock ticks (5 seconds)
TimerActiveFlag BYTE
0
; Active flag for timer handler
;* NewTimer - Handler routine for timer interrupt (Interrupt 08).
;* Decrements CountDown every 5 seconds. No other action is taken
;* until CountDown reaches 0, at which time the speaker sounds.
NewTimer PROC
.IF
jmp
.ENDIF
inc
pushf
call
sti
push
push
pop
dec

FAR
cs:TimerActiveFlag != 0 ; If timer busy or retired,
cs:OldTimer
;
jump to original timer routine
cs:TimerActiveFlag
cs:OldTimer
ds
cs
ds
tick_91

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 280 of 8 Printed: 10/02/00 04:22 PM

;
;
;
;
;
;
;
;

Set active flag
Simulate interrupt by pushing flags,
then far-calling original routine
Enable interrupts
Preserve DS register
Point DS to current segment for
further memory access
Count down for 91 ticks

Chapter 11 Writing Memory-Resident Software
.IF
mov
dec
.IF
call
inc
.ENDIF
.ENDIF
dec
pop
iret
NewTimer ENDP

zero?
tick_91, 91
CountDown
zero?
Sound
TimerActiveFlag

; If 91 ticks have elapsed,
;
reset secondary counter and
;
subtract one 5-second interval
; If CountDown drained,
;
sound speaker
; Alarm has sounded--inc flag
;
again so it remains set

TimerActiveFlag
ds

; Decrement active flag
; Recover DS
; Return from interrupt handler

;* Sound - Sounds speaker with the following tone and duration:
BEEP_TONE
BEEP_DURATION

EQU
EQU

440
6

; Beep tone in hertz
; Number of clocks during beep,
;
where 18 clocks = approx 1 second

Sound

PROC
USES ax bx cx dx es
; Save registers used in this routine
mov
al, 0B6h
; Initialize channel 2 of
out
43h, al
;
timer chip
mov
dx, 12h
; Divide 1,193,180 hertz
mov
ax, 34DCh
;
(clock frequency) by
mov
bx, BEEP_TONE
;
desired frequency
div
bx
; Result is timer clock count
out
42h, al
; Low byte of count to timer
mov
al, ah
out
42h, al
; High byte of count to timer
in
al, 61h
; Read value from port 61h
or
al, 3
; Set first two bits
out
61h, al
; Turn speaker on
; Pause for specified number of clock ticks
mov
sub
mov
add
adc
.REPEAT
mov
mov
sub
sbb
.UNTIL

dx,
cx,
es,
dx,
cx,

BEEP_DURATION
cx
cx
es:[46Ch]
es:[46Eh]

bx, es:[46Ch]
ax, es:[46Eh]
bx, dx
ax, cx
!carry?

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 281 of 9 Printed: 10/02/00 04:22 PM

;
;
;
;
;

Beep duration in clock
CX:DX = tick count for
Point ES to low memory
Add current tick count
Result is target count

ticks
pause
data
to CX:DX
in CX:DX

; Now repeatedly poll clock
;
count until the target
;
time is reached

281

282

Programmer’s Guide

Sound
;*
;*
;*
;*
;*
;*
;*
;*

in
xor
out
ret
ENDP

al, 61h
al, 3
61h, al

; When time elapses, get port value
; Kill bits 0-1 to turn
;
speaker off

Install - Converts ASCII argument to valid binary number, replaces
NewTimer as the interrupt handler for the timer, then makes program
memory-resident by exiting through Function 31h.
This procedure marks the end of the TSR's resident section and the
beginning of the installation section. When ALARM terminates through
Function 31h, the above code and data remain resident in memory. The
memory occupied by the following code is returned to DOS.

Install PROC
; Time argument is in hhmm military format. Converts ASCII digits to
; number of minutes since midnight, then converts current time to number
; of minutes since midnight. Difference is number of minutes to elapse
; until alarm sounds. Converts to seconds-to-elapse, divides by 5 seconds,
; and stores result in word CountDown.
DEFAULT_TIME
EQU
3600
; Default alarm setting = 1 hour
;
(in seconds) from present time
mov
ax, DEFAULT_TIME
cwd
; DX:AX = default time in seconds
.IF
BYTE PTR CountDown != ' ' ; If not blank argument,
xor
CountDown[0], '00'
;
convert 4 bytes of ASCII
xor
CountDown[2], '00'
;
argument to binary
mov
mul
add
mov
mov
mul
add
mov
mov
int

al, 10
BYTE PTR
al, BYTE
bh, al
al, 10
BYTE PTR
al, BYTE
bl, al
ah, 2Ch
21h

;
CountDown[0]
;
PTR CountDown[1]
;
;
CountDown[2]
;
PTR CountDown[3] ;
;
;
;

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 282 of 10 Printed: 10/02/00 04:22 PM

Multiply 1st hour digit by 10
and add to 2nd hour digit
BH = hour for alarm to go off
Repeat procedure for minutes
Multiply 1st minute digit by 10
and add to 2nd minute digit
BL = minute for alarm to go off
Request Function 2Ch
Get Time (CX = current hour/min)

Chapter 11 Writing Memory-Resident Software
mov
sub
push
mov
mul
sub
add

dl,
dh,
dx
al,
ch
ch,
cx,

mov
mul
sub
add

al, 60
bh
bh, bh
ax, bx

sub

ax, cx

.IF
add
.ENDIF

carry?
ax, 24 * 60

; AX = number of minutes since
;
midnight for alarm setting
; AX = time in minutes to elapse
;
before alarm sounds
; If alarm time is tomorrow,
;
add minutes in a day

mov
mul
pop
sub
sbb
.IF
mov
cwd
.ENDIF
.ENDIF
mov
div
mov

bx, 60
bx
bx
ax, bx
dx, 0
carry?
ax, 5

;
;
;
;
;
;

bx, 5
bx
CountDown, ax

; Divide result by 5 seconds
; AX = number of 5-second intervals
;
to elapse before alarm sounds

mov
int
mov
mov
mov
mov
int

ax, 3508h
21h
WORD PTR OldTimer[0], bx
WORD PTR OldTimer[2], es
ax, 2508h
dx, OFFSET NewTimer
21h

;
;
;
;
;
;
;

dx,
cl,
dx,
dx
ax,
21h

; DX = bytes in resident section

mov
mov
shr
inc
mov
int
Install ENDP
END

dh
dh
60
ch
ax

OFFSET Install
4
cl
3100h

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 283 of 11 Printed: 10/02/00 04:22 PM

; Save DX = current seconds
; Multiply current hour by 60
;
to convert to minutes
; Add current minutes to result
; CX = minutes since midnight
; Multiply alarm hour by 60
;
to convert to minutes

DX:AX = minutes-to-elapse-times-60
Recover current seconds
DX:AX = seconds to elapse before
alarm activates
If negative,
assume 5 seconds

Request Function 35h
Get Vector for timer (Interrupt 08)
Store address of original
timer interrupt
Request Function 25h
DS:DX points to new timer handler
Set Vector with address of NewTimer

; Convert to number of paragraphs
;
plus one
; Request Function 31h, error code=0
; Terminate-and-stay-resident

283

284

Programmer’s Guide

Note the following points about ALARM:
u

The constant BEEP_TONE specifies the alarm tone. Practical values for the
tone range from approximately 100 to 4,000 hertz.

The Install procedure marks the beginning of the installation section of
the program. Execution begins here when ALARM.COM is loaded. A TSR
generally places its installation code after the resident section. This allows the
terminating TSR to include the installation code with the rest of the memory
it returns to MS-DOS. Since the installation section executes only once, the
TSR can discard it after becoming resident.

You can install ALARM any number of times in quick succession, each time
with a new alarm setting. The timer handler does not restore the original
vector for Interrupt 08 after the alarm sounds. In effect, the multiple
installations remain daisy-chained in memory. The address in OldTimer for
one installation is the address of NewTimer in the preceding installation.
Until a system reboot, NewTimer remains in place as the Interrupt 08
handler, even after the alarm sounds. To save unnecessary activity, the byte
TimerActiveFlag remains set after the alarm sounds. This forces an
immediate jump to the original handler for all subsequent executions of
NewTimer.
NewTimer and Sound alter registers DS, AX, BX, CX, DX, and ES. To
preserve the original values in these registers, the procedures first push them
onto the stack and then restore the original values before exiting. This
ensures that the process interrupted by NewTimer continues with valid
registers after NewTimer returns.
ALARM requires little stack space. It assumes the current stack is adequate
and makes no attempt to set up a new one. More sophisticated TSRs,
however, should as a matter of course provide their own stacks to ensure
adequate stack depth. The example program presented in “Example of an
Advanced TSR: SNAP,” later in this chapter, demonstrates this safety
measure.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 284 of 12 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

285

Using MS-DOS in Active TSRs
This section explains how to write active TSRs that can safely call MS-DOS
functions. The material explores the problems imposed by the nonreentrant
nature of MS-DOS and explains how a TSR can resolve those problems. The
solution consists of four parts:
u

Understanding how MS-DOS uses stacks

Determining when MS-DOS is active

Determining whether a TSR can safely interrupt an active MS-DOS function

Monitoring the Critical Error flag

Understanding MS-DOS Stacks
MS-DOS functions set up their own stacks, which makes them nonreentrant. If
a TSR interrupts an MS-DOS function and then executes another function that
sets up the same stack, the second function will overwrite everything placed on
the stack by the first function. The problem occurs when the second function
returns and the first is left with unusable stack data. A TSR that calls an MSDOS function must not interrupt any function that uses the same stack.
MS-DOS versions 2.0 and later use three internal stacks: an I/O stack, a disk
stack, and an auxiliary stack. The current stack depends on the MS-DOS
function. Functions 01 through 0Ch set up the I/O stack. Functions higher than
0Ch (with few exceptions) use the disk stack, as do Interrupts 25h and 26h.
MS-DOS normally uses the auxiliary stack only when it executes Interrupt 24h
(Critical Error Handler).

Determining MS-DOS Activity
A TSR’s handlers can determine when MS-DOS is active by consulting a 1-byte
flag called the InDos flag. Every MS-DOS function sets this flag upon entry and
clears it upon termination. During installation, a TSR locates the flag through
Function 34h (Get Address of InDos Flag), which returns the address as ES:BX.
The installation portion then stores the address so the handlers can later find the
flag without again calling Function 34h.
Theoretically, a TSR can wait to execute until the InDos flag is clear, thus
sidestepping the entire issue of interrupting MS-DOS. However, several loworder functions — such as Function 0Ah (Get Buffered Keyboard Input) — wait
idly for an expected keystroke before they terminate. If a TSR were allowed to
execute only after MS-DOS returned, it too would have to wait for the
terminating event.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 285 of 13 Printed: 10/02/00 04:22 PM

286

Programmer’s Guide

The solution lies in determining when the low-order functions 01 through 0Ch
are active. MS-DOS provides another service for this purpose: Interrupt 28h,
the Idle Interrupt.

Interrupting MS-DOS Functions
MS-DOS continually calls Interrupt 28h from the low-order polling functions as
they wait for keyboard input. This signal says that MS-DOS is idle and that a
TSR may interrupt provided it does not overwrite the I/O stack. Put another
way, a TSR can safely interrupt MS-DOS Functions 01 through 0Ch provided it
does not call them.
An active TSR that calls MS-DOS must monitor Interrupt 28h with a handler.
When the handler gains control, it checks the TSR request flag. If the flag
indicates the TSR has been requested and if system hardware is inactive, the
handler executes the TSR. Since control must eventually return to the idle MSDOS function which has stored data on the I/O stack, the TSR in this case must
not call any MS-DOS function that also uses the I/O stack. Table 11.1 shows
which functions set up the I/O stack for various versions of MS-DOS.
Table 11.1

MS-DOS Internal Stacks
Critical
Error flag

MS-DOS
2.x

MS-DOS
3.0

MS-DOS
3.1+

01–0Ch

Clear
Set

I/O*
Aux*

I/O
Aux

33h

Clear
Set

Disk*
Disk

Disk
Disk

Caller*
Caller

50h–51h

Clear
Set

I/O
Aux

Caller
Caller

59h

Clear
Set

n/a*
n/a

I/O
Aux

Disk
Disk

5D0Ah

Clear
Set

n/a
n/a

Disk
Disk

62h

Clear
Set

n/a
n/a

Caller
Caller

All others

Clear
Set

Disk
Disk

Function

* I/O=I/O stack, Aux = auxiliary stack, Disk = disk stack, Caller = caller’s stack, n/a = function not
available.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 286 of 14 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

287

TSRs that perform tasks of long or indefinite duration should themselves call
Interrupt 28h. For example, a TSR that polls for keyboard input should include
an INT 28h instruction in the polling loop, as shown here:
poll:

int
mov
int
jnz
sub
int

28h
ah, 1
16h
poll
ah, ah
16h

; Signal idle state
; Key waiting?
; If not, repeat polling loop
; Otherwise, get key

This courtesy gives other TSRs a chance to execute if the InDos flag happens to
be set.

Monitoring the Critical Error Flag
MS-DOS sets the Critical Error flag to a nonzero value when it detects a critical
error. It then invokes Interrupt 24h (Critical Error Handler) and clears the flag
when Interrupt 24h returns. MS-DOS functions higher than 0Ch are illegal
during critical error processing. Therefore, a TSR that calls MS-DOS must not
execute while the Critical Error flag is set.
MS-DOS versions 3.1 and later locate the Critical Error flag in the byte
preceding the InDos flag. A single call to Function 34h (Get Address of InDos
Flag) thus effectively returns the addresses of both flags. For earlier versions of
MS-DOS or for the compatibility version of MS-DOS in OS/2 1.x, a TSR must
call Function 34h and then scan the segment returned in the ES register for one
of the two following sequences of instructions:
; Sequence of instructions in DOS Versions 2.0 - 3.0
cmp
ss:[CriticalErrorFlag], 0
jne
@F
int
28h
; Sequence of instructions in DOS compatibility version for OS/2 1.x
test
[CriticalErrorFlag], 0FFh
jnz
@F
push
ss:[ ? ]
int
28h

The question mark inside brackets in the preceding PUSH statement indicates
that the operand for the PUSH instruction can be any legal operand.
In either version of MS-DOS, the operand field in the first instruction gives the
flag’s offset. The value in ES determines the segment address. “Example of an
Advanced TSR: SNAP,” later in the chapter, presents a program that shows
how to locate the Critical Error flag with this technique.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 287 of 15 Printed: 10/02/00 04:22 PM

288

Programmer’s Guide

Preventing Interference
This section describes how an active TSR can avoid interfering with the process
it interrupts. Interference occurs when a TSR commits an error or performs an
action that affects the interrupted process after the TSR returns. Examples of
interference range from relatively harmless, such as moving the cursor, to
serious, such as overrunning a stack.
Although a TSR can interfere with another process in many different ways,
protection against interference involves only three steps:
1. Recording a current configuration
2. Changing the configuration so it applies to the TSR
3. Restoring the original configuration before terminating
The example program described on page 293 demonstrates all the
noninterference safeguards described in this section. These safeguards by no
means exhaust the subject of noninterference. More sophisticated TSRs may
require more sophisticated methods. However, noninterference methods
generally fall into one of the following categories:
u
u
u

Trapping errors
Preserving an existing condition
Preserving existing data

Trapping Errors
A TSR committing an error that triggers an interrupt must handle the interrupt to
trap the error. Otherwise, the existing interrupt routine, which belongs to the
underlying process, would attempt to service an error the underlying process did
not commit.
For example, a TSR that accepts keyboard input should include handlers for
Interrupts 23h and 1Bh to trap keyboard break signals. When MS-DOS detects
CTRL+C from the keyboard or input stream, it transfers control to Interrupt 23h
(CTRL+C Handler). Similarly, the BIOS keyboard routine calls Interrupt 1Bh
(CTRL+BREAK Handler) when it detects a CTRL+BREAK key combination. Both
routines normally terminate the current process.
A TSR that calls MS-DOS should also trap critical errors through Interrupt 24h
(Critical Error Handler). MS-DOS functions call Interrupt 24h when they
encounter certain hardware errors. The TSR must not allow the existing
interrupt routine to service the error, since the routine might allow the user to
abort service and return control to MS-DOS. This would terminate both the

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 288 of 16 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

289

TSR and the underlying process. By handling Interrupt 24h, the TSR retains
control if a critical error occurs.
An error-trapping handler differs in two ways from a TSR’s other handlers:
1. It is temporary, in service only while the TSR executes. At startup, the TSR
copies the handler’s address to the interrupt vector table; it then restores the
original vector before returning.
2. It provides complete service for the interrupt; it does not pass control on to
the original routine.
Error-trapping handlers often set a flag to let the TSR know the error has
occurred. For example, a handler for Interrupt 1Bh might set a flag when the
user presses CTRL+BREAK. The TSR can check the flag as it polls for keyboard
input, as shown here:
BrkHandler PROC FAR
; Handler for Interrupt 1Bh
.
.
.
mov
cs:BreakFlag, TRUE ; Raise break flag
iret
; Terminate interrupt
BrkHandler ENDP
.
.
.
mov
poll:
.
.
.
cmp
je
mov
int
jnz

BreakFlag, FALSE

; Initialize break flag

BreakFlag, TRUE
exit
ah, 1
16h
poll

; Keyboard break pressed?
; If so, break polling loop
; Key waiting?
; If not, repeat polling loop

Preserving an Existing Condition
A TSR and its interrupt handlers must preserve register values so that all
registers are returned intact to the interrupted process. This is usually done by
pushing the registers onto the stack before changing them, then popping the
original values before returning.
Setting up a new stack is another important safeguard against interference. A
TSR should usually provide its own stack to avoid the possibility of overrunning
the current stack. Exceptions to this rule are simple TSRs such as the sample
program ALARM that make minimal stack demands.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 289 of 17 Printed: 10/02/00 04:22 PM

290

Programmer’s Guide

A TSR that alters the video configuration should return the configuration to its
original state upon return. Video configuration includes cursor position, cursor
shape, and video mode. The services provided through Interrupt 10h enable a
TSR to determine the existing configuration and alter it if necessary.
However, some applications set video parameters by directly programming the
video controller. When this happens, BIOS remains unaware of the new
configuration and consequently returns inaccurate information to the TSR.
Unfortunately, there is no solution to this problem if the controller’s data
registers provide write-only access and thus cannot be queried directly. For
more information on video controllers, refer to Richard Wilton, Programmer’s
Guide to the PC & PS/2 Video Systems. (See “Books for Further Reading” in
the Introduction.)

Preserving Existing Data
A TSR requires its own disk transfer area (DTA) if it calls MS-DOS functions
that access the DTA. These include file control block functions and Functions
11h, 12h, 4Eh, and 4Fh. The TSR must switch to a new DTA to avoid
overwriting the one belonging to the interrupted process. On becoming active,
the TSR calls Function 2Fh to obtain the address of the current DTA. The TSR
stores the address and then calls Function 1Ah to establish a new DTA. Before
returning, the TSR again calls Function 1Ah to restore the address of the original
DTA.
MS-DOS versions 3.1 and later allow a TSR to preserve extended error
information. This prevents the TSR from destroying the original information if it
commits an MS-DOS error. The TSR retrieves the current extended error data
by calling MS-DOS Function 59h. It then copies registers AX, BX, CX, DX, SI,
DI, DS, and ES to an 11-word data structure in the order given. MS-DOS
reserves the last three words of the structure, which should each be set to zero.
Before returning, the TSR calls Function 5Dh with AL = 0Ah and DS:DX
pointing to the data structure. This call restores the extended error data to their
original state.

Communicating Through the Multiplex Interrupt
The Multiplex interrupt (Interrupt 2Fh) provides the Microsoft-approved way
for a program to verify the presence of an installed TSR and to exchange
information with it. MS-DOS version 2.x uses Interrupt 2Fh only as an interface
for the resident print spooler utility PRINT.COM. Later MS-DOS versions
standardize calling conventions so that multiple TSRs can share the interrupt.
A TSR chains to the Multiplex interrupt by setting up a handler. The TSR’s
installation code records the Interrupt 2Fh vector and then replaces it with the
address of the new multiplex handler.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 290 of 18 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

291

The Multiplex Handler
A program communicates with a multiplex handler by calling Interrupt 2Fh with
an identity number in the AH register. As each handler in the chain gains
control, it compares the value in AH with its own identity number. If the handler
finds that it is not the intended recipient of the call, it passes control to the
previous handler. The process continues until control reaches the target handler.
When the target handler finishes its tasks, it returns via an IRET instruction to
terminate the interrupt.
The target handler determines its tasks from the function number in AL.
Convention reserves Function 0 as a request for installation status. A multiplex
handler must respond to Function 0 by setting AL to 0FFh, to inform the caller
of the handler’s presence in memory. The handler should also return other
information to provide a completely reliable identification. For example, it might
return in ES:BX a far pointer to the TSR’s copyright notice. This assures the
caller it has located the intended TSR and not another TSR that has already
claimed the identity number in AH.
Identity numbers range from 192 to 255, since MS-DOS reserves lesser values
for its own use. During installation, a TSR must verify the uniqueness of its
number. It must not set up a multiplex handler identified by a number already in
use. A TSR usually obtains its identity number through one of the following
methods:
u
u

The programmer assigns the number in the program.
The user chooses the number by entering it as an argument in the command
line, placing it into an environment variable, or by altering the contents of an
initialization file.
The TSR selects its own number through a process of trial and error.

The last method offers the most flexibility. It finds an identity number not
currently in use among the installed multiplex handlers and does not require
intervention from the user.
To use this method, a TSR calls Interrupt 2Fh during installation with AH = 192
and AL = 0. If the call returns AL = 0FFh, the program tests other registers to
determine if it has found a prior installation of itself. If the test fails, the program
resets AL to zero, increments AH to 193, and again calls Interrupt 2Fh. The
process repeats with incrementing values in AH until the TSR locates a prior
installation of itself — in which case it should abort with an appropriate message
to the user — or until AL returns as zero. The TSR can then use the value in AH
as its identity number and proceed with installation.
The SNAP.ASM program in this chapter demonstrates how a TSR can use this
trial-and-error method to select a unique identity number. During installation, the
program calls Interrupt 2Fh to verify that SNAP is not already installed. When

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 291 of 19 Printed: 10/02/00 04:22 PM

292

Programmer’s Guide

deinstalling, the program again calls Interrupt 2Fh to locate the resident TSR in
memory. SNAP’s multiplex handler services the call and returns the address of
the resident code’s program-segment prefix. The calling program can then locate
the resident code and deinstall it, as explained in “Deinstalling a TSR,”
following.

Using the Multiplex Interrupt Under MS-DOS Version 2.x
A TSR can use the Multiplex interrupt under MS-DOS version 2.x, with certain
limitations. Under version 2.x, only MS-DOS’s print spooler PRINT, itself a
TSR program, provides an Interrupt 2Fh service. The Interrupt 2Fh vector
remains null until PRINT or another TSR is installed that sets up a multiplex
handler.
Therefore, a TSR running under version 2.x must first check the existing
Interrupt 2Fh vector before installing a multiplex handler. The TSR locates the
current Interrupt 2Fh handler through Function 35h (Get Interrupt Vector). If
the function returns a null vector, the TSR’s handler will be last in the chain of
Interrupt 2Fh handlers. The handler must terminate with an IRET instruction
rather than pass control to a nonexistent routine.
PRINT in MS-DOS version 2.x does not pass control to the previous handler. If
you intend to run PRINT under version 2.x, the program must be installed
before other TSRs that also handle Interrupt 2Fh. This places PRINT’s
multiplex handler last in the chain of handlers.

Deinstalling a TSR
A TSR should provide a means for the user to remove or “deinstall” it from
memory. Deinstallation returns occupied memory to the system, offering these
benefits:
u

The freed memory becomes available to subsequent programs that may
require additional memory space.
Deinstallation restores the system to a normal state. Thus, sensitive programs
that may be incompatible with TSRs can execute without the presence of
installed routines.

A deinstallation program must first locate the TSR in memory, usually by
requesting an address from the TSR’s multiplex handler. When it has located the
TSR, the deinstallation program should then compare addresses in the vector
table with the addresses of the TSR’s handlers. A mismatch indicates that
another TSR has chained a handler to the interrupt routine. In this case, the
deinstallation program should deny the request to deinstall. If the addresses of

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 292 of 20 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

the TSR’s handlers match those in the vector table, deinstallation can safely
continue.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 293 of 21 Printed: 10/02/00 04:22 PM

293

294

Programmer’s Guide

You can deinstall the TSR with these three steps:
1. Restore to the vector table the original interrupt vectors replaced by the
handler addresses.
2. Read the segment address stored at offset 2Ch of the resident TSR’s
program segment prefix (PSP). This address points to the TSR’s
“environment block,” a list of environment variables that MS-DOS copies
into memory when it loads a program. Place the block’s address in the ES
register and call MS-DOS Function 49h (Release Memory Block) to return
the block’s memory to the operating system.
3. Place the resident PSP segment address in ES and again call Function 49h.
This call releases the block of memory occupied by the TSR’s code and
data.
The example program in the next section demonstrates how to locate a resident
TSR through its multiplex handler, and deinstall it from memory.

Example of an Advanced TSR: SNAP
This section presents SNAP, a memory-resident program that demonstrates
most of the techniques discussed in this chapter. SNAP takes a snapshot of the
current screen and copies the text to a specified file. SNAP accommodates
screens with various column and line counts, such as CGA’s 40-column mode
or VGA’s 50-line mode. The program ignores graphics screens.
Once installed, SNAP occupies approximately 7.5K of memory. When it detects
the ALT+LEFT SHIFT+S key combination, SNAP displays a prompt for a file
specification. The user can type a new filename, accept the previous filename
by pressing ENTER, or cancel the request by pressing ESC.
SNAP reads text directly from the video buffer and copies it to the specified file.
The program sets the file pointer to the end of the file so that text is appended
without overwriting previous data. SNAP copies each line only to the last
character, ignoring trailing spaces. The program adds a carriage return–linefeed
sequence (0D0Ah) to the end of each line. This makes the file accessible to any
text editor that can read ASCII files.
To demonstrate how a program accesses resident data through the Multiplex
interrupt, SNAP can reset the display attribute of its prompt box. After installing
SNAP, run the main program with the /C option to change box colors:
SNAP /Cxx

The argument xx specifies the desired attribute as a two-digit hexadecimal
number — for example, 7C for red on white, or 0F for monochrome high

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 294 of 22 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

295

intensity. For a list of color and monochrome display attributes, refer to the
“Tables” section of the Reference.
SNAP can deinstall itself, provided another TSR has not been loaded after it.
Deinstall SNAP by executing the main program with the /D option:
SNAP /D

If SNAP successfully deinstalls, it displays the following message:
TSR deinstalled

Building SNAP.EXE
SNAP combines four modules: SNAP.ASM, COMMON.ASM,
HANDLERS.ASM, and INSTALL.ASM. Source files are located on one of
your distribution disks. Each module stores temporary code and data in the
segments INSTALLCODE and INSTALLDATA. These segments apply only to
SNAP’s installation phase; MS-DOS recovers the memory they occupy when
the program exits through the terminate-and-stay-resident function. The
following briefly describes each module:
u
u
u

SNAP.ASM contains the TSR’s main code and data.
COMMON.ASM contains procedures used by other example programs.
HANDLERS.ASM contains interrupt handler routines for Interrupts 08, 09,
10h, 13h, 15h, 28h, and 2Fh. It also provides simple error-trapping handlers
for Interrupts 1Bh, 23h, and 24h. Additional routines set up and deinstall the
handlers.
INSTALL.ASM contains an exit routine that calls the terminate-and-stayresident function and a deinstallation routine that removes the program from
memory. The module includes error-checking services and a command-line
parser.

This building-block approach allows you to create other TSRs by replacing
SNAP.ASM and linking with the HANDLERS and INSTALL object modules.
The library of routines accommodates both keyboard-activated and timeactivated TSRs. A time-activated TSR is a program that activates at a
predetermined time of day, similar to the example program ALARM introduced
earlier in this chapter. The header comments for the Install procedure in
HANDLERS.ASM explain how to install a time-activated TSR.
You can write new TSRs in assembly language or any high-level language that
conforms to the Microsoft conventions for ordering segments. Regardless of the
language, the new code must not invoke an MS-DOS function that sets up the
I/O stack (see “Interrupting MS-DOS Functions,” earlier in this chapter). Code

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 295 of 23 Printed: 10/02/00 04:22 PM

296

Programmer’s Guide

in Microsoft C, for example, must not call getche or kbhit, since these
functions in turn call MS-DOS Functions 01 and 0Bh.
Code written in a high-level language must not check for stack overflows.
Compiler-generated stack probes do not recognize the new stack setup when the
TSR executes, and therefore must be disabled. The example program BELL.C,
included on disk with the TSR library routines, demonstrates how to disable
stack checking in Microsoft C using the check_stack pragma.

Outline of SNAP
The following sections outline in detail how SNAP works. Each part of the
outline covers a specific portion of SNAP’s code. Headings refer to earlier
sections of this chapter, providing cross-references to SNAP’s key procedures.
For example, the part of the outline that describes how SNAP searches for its
startup signal refers to the section “Auditing Hardware Events for TSR
Requests,” earlier in this chapter.
Figures 11.2 through 11.4 are flowcharts of the SNAP program. Each chart
illustrates a separate phase of SNAP’s operation, from installation through
memory-residency to deinstallation.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 296 of 24 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

Figure 11.2

Flowchart for SNAP.EXE: Installation Phase

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 297 of 25 Printed: 10/02/00 04:22 PM

297

298

Programmer’s Guide

Figure 11.3

Flowchart for SNAP.EXE: Resident Phase

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 298 of 26 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

Figure 11.4

Flowchart for SNAP.EXE: Deinstallation Phase

Refer to the flowcharts as you read the following outline. They will help you
maintain perspective while exploring the details of SNAP’s operation. Text in
the outline cross-references the charts.
Note that information in both the outline and the flowcharts is generic. Except
for references to the SNAP procedure, all descriptions in the outline and the

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 299 of 27 Printed: 10/02/00 04:22 PM

299

300

Programmer’s Guide

flowcharts apply to any TSR created with the HANDLERS and INSTALL
modules.

Auditing Hardware Events for TSR Requests
To search for its startup signal, SNAP audits the keyboard with an interrupt
handler for either Interrupt 09 (keyboard) or Interrupt 15h (Miscellaneous
System Services). The Install procedure determines which of the two
interrupts to handle based on the following code:
.IF
mov
mov
call
mov

HotScan == 0
ah, HotShift
al, HotMask
GetTimeToElapse
CountDown, ax

;
;
;
;
;

Version, 031Eh
setup

; Force use of KeybrdMonitor as
;
keyboard handler
; DOS Version 3.3 or higher?
; No? Skip next step

.ELSE
cmp
jb
;
;
;
;
;
;
;

If valid scan code given:
AH = hour to activate
AL = minute to activate
Get number of 5-second intervals
to elapse before activation

Test for IBM PS/2 series. If not PS/2, use Keybrd and
SkipMiscServ as handlers for Interrupts 09 and 15h
respectively. If PS/2 system, set up KeybrdMonitor as the
Interrupt 09 handler. Audit keystrokes with MiscServ
handler, which searches for the hot key by handling calls
to Interrupt 15h (Miscellaneous System Services). Refer to
Section 11.2.1 for more information about keyboard handlers.
mov
int
sti

ax, 0C00h
15h

; Function 0Ch (Get System
;
Configuration Parameters)
; Compaq ROM may leave disabled

jc
or
jnz

setup
ah, ah
setup

; If carry set,
;
or if AH not 0,
;
services are not supported

; Test bit 4 to see if Intercept is implemented
test
BYTE PTR es:[bx+5], 00010000y
jz
setup
; If so, set up MiscServ as Interrupt 15h handler
mov
ax, OFFSET MiscServ
mov
WORD PTR intMisc.NewHand, ax
.ENDIF
; Set up KeybrdMonitor as Interrupt 09 handler
mov
ax, OFFSET KeybrdMonitor
mov
WORD PTR intKeybrd.NewHand, ax

The following describes the code’s logic:

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 300 of 28 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

301

If the program is running under MS-DOS version 3.3 or higher and if
Interrupt 15h supports Function 4Fh, set up handler MiscServ to search for
the hot key. Handle Interrupt 09 with KeybrdMonitor only to maintain the
keyboard active flag.

Otherwise, set up a handler for Interrupt 09 to search for the hot key. Handle
calls to Interrupt 15h with the routine SkipMiscServ, which contains this
single instruction:
jmp

cs:intMisc.OldHand

The jump immediately passes control to the original Interrupt 15h routine;
thus, SkipMiscServ has no effect. It serves only to simplify coding in
other parts of the program.
At each keystroke, the keyboard interrupt handler (either Keybrd or
MiscServ) calls the procedure CheckHotKey with the scan code of the
current key. CheckHotKey compares the scan code and shift status with the
bytes HotScan and HotShift. If the current key matches, CheckHotKey
returns the carry flag clear to indicate that the user has pressed the hot key.
If the keyboard handler finds the carry flag clear, it sets the flag
TsrRequestFlag and exits. Otherwise, the handler transfers control to the
original interrupt routine to service the interrupt.
The timer handler Clock reads the request flag at every occurrence of the timer
interrupt. Clock takes no action if it finds a zero value in TsrRequestFlag.
Figures 11.1 and 11.3 depict the relationship between the keyboard and timer
handlers.

Monitoring System Status
Because SNAP produces output to both video and disk, it avoids interrupting
either video or disk operations. The program uses interrupt handlers Video and
DiskIO to monitor Interrupts 10h (video) and 13h (disk). SNAP also avoids
interrupting keyboard use. The instructions at the far label KeybrdMonitor
serve as the monitor handler for Interrupt 09 (keyboard).
The three handlers perform similar functions. Each sets an active flag and then
calls the original routine to service the interrupt. When the service routine
returns, the handler clears the active flag to indicate that the device is no longer
in use.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 301 of 29 Printed: 10/02/00 04:22 PM

302

Programmer’s Guide

The BIOS Interrupt 13h routine clears or sets the carry flag to indicate the
operation’s success or failure. DiskIO therefore preserves the flags register
when returning, as shown here:
DiskIO

PROC
FAR
mov
cs:intDiskIO.Flag, TRUE ; Set active flag
; Simulate interrupt by pushing flags and far-calling old
; Int 13h routine
pushf
call
cs:intDiskIO.OldHand
; Clear active flag without disturbing flags register
mov
cs:intDiskIO.Flag, FALSE
sti
; Enable interrupts
; Simulate IRET without popping flags (since services use
; carry flag)
ret
2
DiskIO ENDP

The terminating RET 2 instruction discards the original flags from the stack
when the handler returns.

Determining Whether to Invoke the TSR
The procedure CheckRequest determines whether the TSR:
u
u

Has been requested.
Can safely interrupt the system.

Each time it executes, the timer handler Clock calls CheckRequest to read
the flag TsrRequestFlag. If CheckRequest finds the flag set, it scans other
flags maintained by the TSR’s interrupt handlers and by MS-DOS. These flags
indicate the current system status. As the flowchart in Figure 11.3 shows,
CheckRequest calls CheckDos (described following) to determine the status
of the operating system. CheckRequest then calls CheckHardware to check
hardware status.
CheckHardware queries the interrupt controller to determine if any device is

currently being serviced. It also reads the active flags maintained by the
KeybrdMonitor, Video, and DiskIO handlers. If the controller, keyboard,
video, and disk are all inactive, CheckHardware clears the carry flag and
returns.
CheckRequest indicates system status with the carry flag. If the procedure

returns the carry flag set, the caller exits without invoking the TSR. A clear
carry signals that the caller can safely execute the TSR.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 302 of 30 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

303

Determining MS-DOS Activity
As Figure 11.2 shows, the procedure GetDosFlags locates the InDos flag
during SNAP’s installation phase. GetDosFlags calls Function 34h (Get
Address of InDos Flag) and then stores the flag’s address in the far pointer
InDosAddr.
When called from the CheckRequest procedure, CheckDos reads InDos to
determine whether the operating system is active. Note that CheckDos reads
the flag directly from the address in InDosAddr. It does not call Function 34h
to locate the flag, since it has not yet established whether MS-DOS is active.
This follows from the general rule that interrupt handlers must not call any MSDOS function.
The next two sections more fully describe the procedure CheckDos.

Interrupting MS-DOS Functions
Figure 11.3 shows that the call to CheckDos can initiate either from Clock
(timer handler) or Idle (Interrupt 28h handler). If CheckDos finds the InDos
flag set, it reacts in different ways, depending on the caller:
u

If called from Clock, CheckDos cannot know which MS-DOS function is
active. In this case, it returns the carry flag set, indicating that Clock must
deny the request for the TSR.
If called from Idle, CheckDos assumes that one of the low-order polling
functions is active. It therefore clears the carry flag to let the caller know the
TSR can safely interrupt the function.

For more information on this topic, see the section “Interrupting MS-DOS
Functions,” earlier in this chapter.

Monitoring the Critical Error Flag
The procedure GetDosFlags (Figure 11.2) determines the address of the
Critical Error flag. The procedure stores the flag’s address in the far pointer
CritErrAddr.
When called from either the Clock or Idle handlers, CheckDos reads the
Critical Error flag. A nonzero value in the flag indicates that the Critical Error
Handler (Interrupt 24h) is processing a critical error and the TSR must not
interrupt. In this case, CheckDos sets the carry flag and returns, causing the
caller to exit without executing the TSR.

Trapping Errors
As Figure 11.3 shows, Clock and Idle invoke the TSR by calling the
procedure Activate. Before calling the main body of the TSR, Activate
sets up the following handlers:

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 303 of 31 Printed: 10/02/00 04:22 PM

304

Programmer’s Guide
Handler Name

For Interrupt

Receives Control When

CtrlBreak

1Bh (CTRL+BREAK Handler)

CTRL+BREAK sequence entered at

keyboard
CtrlC

23h (CTRL+C Handler)

MS-DOS detects a CTRL+C sequence
from the keyboard or input stream

CritError

24h (Critical Error Handler)

MS-DOS encounters a critical error

These handlers trap keyboard break signals and critical errors that would
otherwise trigger the original handler routines. The CtrlBreak and CtrlC
handlers contain a single IRET instruction, thus rendering a keyboard break
ineffective. The CritError handler contains the following instructions:
CritError PROC
sti
sub
.IF
mov
.ENDIF
iret
CritError ENDP

FAR
al, al
cs:major != 2
al, 3

;
;
;
;

Assume DOS 2.x
Set AL = 0 for ignore error
If DOS 3.x, set AL = 3
DOS call fails

The return code in AL stops MS-DOS from taking further action when it
encounters a critical error.
As an added precaution, Activate also calls Function 33h (Get or Set
to determine the current setting of the checking flag.
Activate stores the setting, then calls Function 33h again to turn off break
checking.
CTRL+BREAK Flag)

When the TSR’s main procedure finishes its work, it returns to Activate,
which restores the original setting for the checking flag. It also replaces the
original vectors for Interrupts 1Bh, 23h, and 24h.
SNAP’s error-trapping safeguards enable the TSR to retain control in the event
of an error. Pressing CTRL+BREAK or CTRL+C at SNAP’s prompt has no effect. If
the user specifies a nonexistent drive — a critical error — SNAP merely beeps the
speaker and returns normally.

Preserving an Existing Condition
Activate records the stack pointer SS:SP in the doubleword OldStackAddr.
The procedure then resets the pointer to the address of a new stack before
calling the TSR. Switching stacks ensures that SNAP has adequate stack depth
while it executes.

The label NewStack points to the top of the new stack buffer, located in the
code segment of the HANDLERS.ASM module. The equate constant

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 304 of 32 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

305

STACK_SIZ determines the size of the stack. The include file TSR.INC
contains the declaration for STACK_SIZ.
Activate preserves the values in all registers by pushing them onto the new
stack. It does not push DS, since that register is already preserved in the Clock
or Idle handler.

SNAP does not alter the application’s video configuration other than by moving
the cursor. Figure 11.3 shows that Activate calls the procedure Snap, which
executes Interrupt 10h to determine the current cursor position. Snap stores the
row and column in the word OldPos. The procedure restores the cursor to its
original location before returning to Activate.

Preserving Existing Data
Because SNAP does not call an MS-DOS function that writes to the DTA, it
does not need to preserve the DTA belonging to the interrupted process.
However, the code for switching and restoring the DTA is included within
IFDEF blocks in the procedure Activate. The equate constant DTA_SIZ,
declared in the TSR.INC file, governs the assembly of the blocks as well as the
size of the new DTA.
It is possible for SNAP to overwrite existing extended error information by
committing a file error. The program does not attempt to preserve the original
information by calling Functions 59h and 5Dh. In certain rare instances, this
may confuse the interrupted process after SNAP returns.

Communicating Through the Multiplex Interrupt
The program uses the Multiplex interrupt (Interrupt 2Fh) to
u
u
u

Verify that SNAP is installed.
Select a unique multiplex identity number.
Locate resident data.

For more information about Interrupt 2Fh, see the section “Communicating
through the Multiplex Interrupt,” earlier in this chapter.
SNAP accesses Interrupt 2Fh through the procedure CallMultiplex, as
shown in Figures 11.2 and 11.4. By searching for a prior installation,
CallMultiplex ensures that SNAP is not installed more than once. During
deinstallation, CallMultiplex locates data required to deinstall the resident
TSR.
The procedure Multiplex serves as SNAP’s multiplex handler. When it
recognizes its identity number in AH, Multiplex determines its tasks from the
function number in the AL register. The handler responds to Function 0 by

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 305 of 33 Printed: 10/02/00 04:22 PM

306

Programmer’s Guide

returning AL equalling 0FFh and ES:DI pointing to an identifier string unique to
SNAP.
CallMultiplex searches for the handler by invoking Interrupt 2Fh in a loop,

beginning with a trial identity number of 192 in AH. At the start of each iteration
of the loop, the procedure sets AL to zero to request presence verification from
the multiplex handler. If the handler returns 0FFh in AL, CallMultiplex
compares its copy of SNAP’s identifier string with the text at memory location
ES:DI. A failed match indicates that the multiplex handler servicing the call is
not SNAP’s handler. In this case, CallMultiplex increments AH and cycles
back to the beginning of the loop.
The process repeats until the call to Interrupt 2Fh returns a matching identifier
string at ES:DI, or until AL returns as zero. A matching string verifies that
SNAP is installed, since its multiplex handler has serviced the call. A return
value of zero indicates that SNAP is not installed and that no multiplex handler
claims the trial identity number in AH. In this case, SNAP assigns the number to
its own handler.

Deinstalling a TSR
During deinstallation, CallMultiplex locates SNAP’s multiplex handler as
described previously. The handler Multiplex receives the verification request
and returns in ES the code segment of the resident program.
Deinstall reads the addresses of the following interrupt handlers from the

data structure in the resident code segment:
Handler Name

Description

Clock

Timer handler

Keybrd

Keyboard handler (non-PS/2)

KeybrdMonitor

Keyboard monitor handler (PS/2)

Video

Video monitor handler

DiskIO

Disk monitor handler

SkipMiscServ

Miscellaneous Systems Services handler (non-PS/2)

MiscServ

Miscellaneous Systems Services handler (PS/2)

Idle

MS-DOS Idle handler

Multiplex

Multiplex handler

Deinstall calls MS-DOS Function 35h (Get Interrupt Vector) to retrieve the

current vectors for each of the listed interrupts. By comparing each handler
address with the corresponding vector, Deinstall ensures that SNAP can be
safely deinstalled. Failure in any of the comparisons indicates that another TSR
has been installed after SNAP and has set up a handler for the same interrupt. In

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 306 of 34 Printed: 10/02/00 04:22 PM

Chapter 11 Writing Memory-Resident Software

307

this case, Deinstall returns an error code, stopping the program with the
following message:
Can’t deinstall TSR

If all addresses match, Deinstall calls Interrupt 2Fh with SNAP’s identity
number in AH and AL set to 1. The handler Multiplex responds by returning
in ES the address of the resident code’s PSP. Deinstall then calls MS-DOS
Function 25h (Set Interrupt Vector) to restore the vectors for the original service
routines. This is called “unhooking” or “unchaining” the interrupt handlers.
After unhooking all of SNAP’s interrupt handlers, Deinstall returns with AX
pointing to the resident code’s PSP. The procedure FreeTsr then calls MSDOS Function 49h (Release Memory) to return SNAP’s memory to the
operating system. The program ends with the message
TSR deinstalled

to indicate a successful deinstallation.
Deinstalling SNAP does not guarantee more available memory space for the
next program. If another TSR loads after SNAP but handles interrupts other
than 08, 09, 10h, 13h, 15h, 28h, or 2Fh, SNAP still deinstalls properly. The
result is a harmless gap of deallocated memory formerly occupied by SNAP.
MS-DOS can use the free memory to store the next program’s environment
block. However, MS-DOS loads the program itself above the still-resident TSR.

Filename: LMAPGC11.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 66 Page: 307 of 35 Printed: 10/02/00 04:22 PM

307

C H A P T E R

1 2

Mixed-Language Programming

Mixed-language programming allows you to combine the unique strengths of
Microsoft Basic, C, C++, and FORTRAN with your assembly-language
routines. Any one of these languages can call MASM routines, and you can call
any of these languages from within your assembly-language programs. This
makes virtually all the routines from high-level–language libraries available to a
mixed-language program.
MASM 6.1 provides mixed-language features similar to those in high-level
languages. For example, you can use the INVOKE directive to call high-levellanguage procedures, and the assembler handles the argument-passing details for
you. You can also use H2INC to translate C header files to MASM include files,
as explained in Chapter 20 of Environment and Tools.
The mixed-language features of MASM 6.1 do not make older methods of
defining mixed-language interfaces obsolete. In most cases, mixed-language
programs written with earlier versions of MASM will assemble and link correctly
under MASM 6.1. (For more information, see Appendix A.)
This chapter explains how to write assembly routines that can be called from
high-level–language modules and how to call high-level language routines from
MASM. You should already understand the languages you want to combine and
should know how to write, compile, and link multiple-module programs with
these languages.
This chapter covers only assembly-language interface with C, C++, Basic, and
FORTRAN; it does not cover mixed-language programming between high-level
languages. The focus here is the Microsoft versions of C, C++, Basic, and
FORTRAN, but the same principles apply to other languages and compilers.
Many of the techniques used in this chapter are explained in the material in
Chapter 7 on writing procedures in assembly language, and in Chapter 8 on
multiple-module programming.
The first section of this chapter discusses naming and calling conventions. The
next section, “Writing an Assembly Procedure for a Mixed-Language Program,”
provides a template for writing an assembly-language procedure that can be

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 307 of 1 Printed: 10/02/00 04:21 PM

308

Programmer’s Guide

called from another module written in a high-level language. This represents the
essence of mixed-language programming. Assembly language is often used for
creating fast secondary routines in a large program written in a high-level
language.
The third section describes specific conventions for linking assembly-language
procedures with modules in C, C++, Basic, and FORTRAN. These languagespecific sections also provide details on how the language manages various data
structures so that your MASM programs are compatible with the data from the
high-level language.

Naming and Calling Conventions
Each language has its own set of conventions, which fall into two categories:
u

The “naming convention” specifies how or if the compiler or assembler alters
the name of an identifier before placing it into an object file.
The “calling convention” determines how a language implements a call to a
procedure and how the procedure returns to the caller.

MASM supports several different conventions. The assembler uses C
convention when you specify a language type (langtype) of C, and Pascal
convention for language types PASCAL, BASIC, or FORTRAN. To the
assembler, the keywords BASIC, PASCAL, and FORTRAN are synonymous.
MASM also supports the SYSCALL and STDCALL conventions, which mix
elements of the C and Pascal conventions.
MASM gives you several ways to set the naming and calling conventions in your
assembly-language program. Using .MODEL with a langtype sets the default
for the module. This can also be done with the OPTION directive. This is
equivalent to the /Gc or /Gd option from the command line. Procedure
prototypes and declarations can specify a langtype to override the default.
When you write mixed-language routines, the easiest way to ensure convention
compatibility is to adopt the conventions of the called procedure’s language.
However, Microsoft languages can change the naming and calling conventions
for different procedures. If your program must call a procedure that uses an
argument-passing method different from that of the default language, prototype
the procedure first with the desired language type. This tells the assembler to
override the conventions of the default language and assume the proper
conventions for the prototyped procedure. “The MASM/High-Level–Language
Interface” section in this chapter explains how to change the default
conventions. The following sections provide more detail on the information
summarized in Table 12.1.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 308 of 2 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

309

Table 12.1 Naming and Calling Conventions
Convention

Leading
underscore

SYSCALL

STDCALL

BASIC

FORTRAN

PASCAL

Capitalize all

Arguments
pushed left to
right

Arguments
pushed right to
left

Caller stack
cleanup

:VARARG
allowed

* The STDCALL language type uses caller stack cleanup if the :VARARG parameter is used. Otherwise, the called routine must
clean up the stack.

Naming Conventions
“Naming convention” refers to the way a compiler or assembler stores the
names of identifiers. The first two rows of Table 12.1 show how each language
type affects symbol names. SYSCALL leaves symbol names as they appear in
the source code, but C and STDCALL add an underscore prefix. PASCAL,
BASIC, and FORTRAN change symbols to all uppercase.
The following list describes how these naming conventions affect a variable
called Big Time in your source code:
Langtype Specified
SYSCALL

Characteristics
Leaves the name unmodified. The linker sees the variable as
Big Time.

C, STDCALL

The assembler (or compiler) adds a leading underscore to the
name, but does not change case. The linker sees the variable as
_Big Time.

PASCAL, FORTRAN,
BASIC

Converts all names to uppercase. The linker sees the variable as
Big Time.

The C Calling Convention
Specify the C language type for assembly-language procedures called from
programs that assume the C calling convention. Note that such programs are not
necessarily written in C, since other languages can mimic C conventions.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 309 of 3 Printed: 10/02/00 04:21 PM

310

Programmer’s Guide

Argument Passing
With the C calling convention, the caller pushes arguments from right to left as
they appear in the caller’s argument list. The called procedure returns without
removing the arguments from the stack. It is the caller’s responsibility to clean
the stack after the call, either by popping the arguments or by adding an
appropriate value to the stack pointer SP.

Register Preservation
The called routine must return with the original values in BP, SI, DI, DS, and
SS. It must also preserve the direction flag.

Varying Number of Arguments
The additional overhead of cleaning the stack after each call has compensations.
It frees the caller from having to pass a set number of arguments to the called
procedure each time. Because the first argument in the list is always the last one
pushed, it is always on the top of the stack. Thus, it has the same address
relative to the frame pointer, regardless of how many arguments were actually
passed.
For example, consider the C library function printf, which accepts different
numbers of arguments. A C program calls the function like this:
printf( "Numbers:
printf( "Also:

%f %f %.2f\n", n1, n2, n3 );
%f", n4 );

The first line passes four arguments (including the string in quotes) and the
second line passes only two arguments. Notice that printf has no reliable way of
determining how many arguments the caller has pushed. Therefore, the function
returns without adjusting the stack. The C calling convention requires the caller
to take responsibility for removing the arguments from the stack, since only the
caller knows how many arguments it passed.
Use INVOKE to call a C-callable function from your assembly-language
program, since INVOKE automatically generates the necessary stack-cleaning
code after the call. You must also prototype the function with the VARARG
keyword if appropriate, as explained in “Procedures,” Chapter 7. Similarly,
when you write a C-callable procedure that accepts a varying number of
arguments, include VARARG in the procedure’s PROC statement.

The Pascal Calling Convention
By default, the langtype for FORTRAN, BASIC, and PASCAL selects the
Pascal calling convention. This convention pushes arguments left to right so that
the last argument is lowest on the stack, and it requires that the called routine
remove arguments from the stack.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 310 of 4 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

311

Argument Passing
Arguments are placed on the stack in the same order in which they appear in the
source code. The first argument is highest in memory (because it is also the first
argument to be placed on the stack), and the stack grows downward.

Register Preservation
A routine that uses the Pascal calling convention must preserve SI, DI, BP, DS,
and SS. For 32-bit code, the EBX, ES, FS, and GS registers must be preserved
as well as EBP, ESI, and EDI. The direction flag is also cleared upon entry and
must be preserved.

Varying Number of Arguments
Passing a variable number of arguments is not possible with the Pascal calling
convention.

The STDCALL and SYSCALL Calling Conventions
A STDCALL procedure adopts the C name and calling conventions when
prototyped with the VARARG keyword. Refer to the section “Declaring
Parameters with the PROC Directive” in Chapter 7. Without VARARG, the
procedure uses the C naming and Pascal calling conventions. STDCALL
provides compatibility with 32-bit versions of Microsoft compilers.
As Table 12.1 shows, SYSCALL is identical to the C calling convention, but
does not add an underscore prefix to symbols.

Argument Passing
Argument passing order for both STDCALL and SYSCALL is the same as the
C calling convention. The caller pushes the arguments from right to left and
must remove the parameters from the stack after the call. However, STDCALL
requires the called procedure to clean the stack if the procedure does not accept
a variable number of arguments.

Register Preservation
Both conventions require the called procedure to preserve the registers BP, SI,
DI, DS, and SS. Under STDCALL, the direction flag is clear on entry and must
be returned clear.

Varying Number of Arguments
SYSCALL allows a variable number of arguments in the same way as the C
calling convention. STDCALL also mimics the C convention when VARARG
appears in the called procedure’s declaration or definition. It allows a varying
number of arguments and requires the caller to clean the stack. If not declared

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 311 of 5 Printed: 10/02/00 04:21 PM

312

Programmer’s Guide

or defined with VARARG, the called procedure does not accept a variable
argument list and must clean the stack before it returns.

Writing an Assembly Procedure
For a Mixed-Language Program
MASM 6.1 simplifies the coding required for linking MASM routines to highlevel– language routines. You can use the PROTO directive to write procedure
prototypes, and the INVOKE directive to call external routines. MASM
simplifies procedure-related tasks in the following ways:
The PROTO directive improves error checking on argument types.
u INVOKE pushes arguments onto the stack and converts argument types to
types expected when possible. These arguments can be referenced by their
parameter label, rather than as offsets of the stack pointer.
u
The LOCAL directive following the PROC statement saves places on the
stack for local variables. These variables can also be referenced by name,
rather than as offsets of the stack pointer.
u PROC sets up the appropriate stack frame according to the processor mode.
u
The USES keyword preserves registers given as arguments.
u
The C calling conventions specified in the PROC syntax allow for a variable
number of arguments to be passed to the procedure.
u
The RET keyword adjusts the stack upward by the number of bytes in the
argument list, removes local variables from the stack, and pops saved
registers.
u
The PROC statement lists parameter names and types. The parameters can
be referenced by name inside the procedure.

The complete syntax and parameter descriptions for these procedure directives
are explained in “Procedures” in Chapter 7. This section provides a template
that you can use for writing a MASM routine to be called from a high-level
language.
The template looks like this:
Label PROC [[distance langtype visibility USES reglist
parmlist]]
LOCAL varlist
.
.
.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 312 of 6 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

313

RET
Label ENDP
Replace the italicized words with appropriate keywords, registers, or variables as
defined by the syntax in “Declaring Parameters with the PROC Directive” in
Chapter 7.
The distance (NEAR or FAR) and visibility (PUBLIC, PRIVATE, or
EXPORT) that you give in the procedure declaration override the current
defaults. In some languages, the model can also be specified with command-line
options.
The langtype determines the calling convention for accessing arguments and
restoring the stack. For information on calling conventions, see “Naming and
Calling Conventions” earlier in this chapter.
The types for the parameters listed in the parmlist must be given. Also, if any of
the parameters are pointers, the assembler does not generate code to get the
value of the pointer references. You must write this code yourself. An example
of how to write such code is provided in “Declaring Parameters with the PROC
Directive” in Chapter 7.
If you need to code your own stack-frame setup manually, or if you do not
want the assembler to generate the standard stack setup and cleanup, see
“Passing Arguments on the Stack” and “User-Defined Prologue and Epilogue
Code” in Chapter 7.

The MASM/High-Level–Language Interface
Since high-level–language programs require initialization, you must write the
main routine of a mixed-language program in the high-level language, or link
with the startup code supplied by the high-level–language compiler. This gives
the assembly code access to high-level routines or library functions. The next
section explains how to link an assembly-language program with C-language
startup code.
For procedures with prototypes, INVOKE makes calls from MASM to highlevel–
language programs, much like procedure or function calls in the high-level
language. INVOKE calls procedures and generates the code to push arguments
in the order specified by the procedure’s calling convention, and to remove
arguments from the stack at the end of the procedure.
INVOKE can also do type checking and data conversion for the argument types
so that the procedure receives compatible data. For explanations of how to write
procedure prototypes and several examples of procedure declarations and the
corresponding prototypes, see “Declaring Procedure Prototypes” in Chapter 7.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 313 of 7 Printed: 10/02/00 04:21 PM

314

Programmer’s Guide

For programs that mix assembly language and C, the H2INC utility makes it
easy to write prototypes and data declarations for the C procedures you want to
call from MASM. H2INC translates the C prototypes and declarations into the
corresponding MASM prototypes and declarations, which INVOKE can use to
call the procedure. The use of H2INC is explained in Chapter 20 in
Environment and Tools.
Mixed-language programming also allows the main program or a routine to use
external data — data defined in the other module. External data is the data that is
stored in a set place in memory (unlike dynamic and local data, which is
allocated on the stack and heap) and is visible to other modules.
External data is shared by all routines. One of the modules must define the static
data, which causes the compiler to allocate storage for the data. The other
modules that access the data must declare the data as external.

Argument Passing
Each language has its own convention for how an argument is actually passed. If
the argument-passing conventions of your routines do not agree, then a called
routine receives bad data. Microsoft languages support three different methods
for passing an argument:
u

Near reference. Passes a variable’s near (offset) address, expressed as an
offset from the default data segment. This method gives the called routine
direct access to the variable itself. Any change the routine makes to the
parameter is reflected in the calling routine.
Far reference. Passes a variable’s far (segmented) address. Though slower
than passing a near reference, this method is necessary for passing data that
lies outside the default data segment. (This is not an issue in Basic unless you
have specifically requested far memory.)
Value. Passes only a copy of the variable, not its address. With this method,
the called routine gets a copy of the argument on the stack, but has no access
to the original variable. The copy is discarded when the routine returns, and
the variable retains its original value.

When you pass arguments between routines written in different languages, you
must ensure that the caller and the called routine use the same conventions for
passing and receiving arguments. In most cases, you should check the argumentpassing defaults used by each language and make any necessary adjustments.
Most languages have features that allow you to change argument-passing
methods.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 314 of 8 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

315

Register Preservation
A procedure called from any high-level language should preserve the direction
flag and the values of BP, SI, DI, SS, and DS. Routines called from MASM
must not alter SI, DI, SS, DS, or BP.

Pushing Addresses
Microsoft high-level languages push segment addresses before offsets. This lets
the called routine use the LES and LDS instructions to read far addresses from
the stack. Furthermore, each word of an argument is placed on the stack in
order of significance. Thus, the high word of a long integer is pushed first,
followed by the low word.

Array Storage
Most high-level-language compilers store arrays in row-major order. This means
that all elements of a row are stored consecutively. The first five elements of an
array with four rows and three columns are stored in row-major order as
A[1, 1], A[1, 2], A[1, 3], A[2, 1], A[2, 2]

In column-major order, the column elements are stored consecutively. For
example, this same array would be stored in column-major order as
A[1, 1], A[2, 1], A[3, 1], A[4, 1], A[1, 2], A[2, 2]

The C/MASM Interface
This section summarizes the characteristics of the interface between MASM and
Microsoft C and QuickC compilers. With the default naming and calling
convention, the assembler (or compiler) pushes arguments right to left and adds
a leading underscore to routine names.

Compatible Data Types
This list shows the 16-bit C data types and equivalent data types in MASM 6.1.
For 32-bit C compilers, int and unsigned int are equivalent to the MASM types
SDWORD and DWORD, respectively.
C Type

Equivalent MASM Type

unsigned char

BYTE

char

SBYTE

unsigned short, unsigned int

WORD

int, short

SWORD

unsigned long

DWORD

long

SDWORD

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 315 of 9 Printed: 10/02/00 04:21 PM

316

Programmer’s Guide
float

REAL4

double

REAL8

long double

REAL10

Naming Restrictions
C is case-sensitive and does not convert names to uppercase. Since C normally
links with the /NOI command-line option, you should assemble MASM modules
with the /Cx or /Cp option to prevent the assembler from converting names to
uppercase.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 316 of 10 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

317

Argument-Passing Defaults
C always passes arrays by reference and all other variables (including structures)
by value. C programs in tiny, small, and medium model pass near addresses for
arrays, unless another distance is specified. Compact-, large-, and huge-model
programs pass far addresses by default. To pass by reference a variable type
other than array, use the C-language address-of operator (&).
If you need to pass an array by value, declare the array as a structure member
and pass a copy of the entire structure. However, this practice is rarely
necessary and usually impractical except for very small arrays, since it can make
substantial demands on stack space. If your program must maintain an array
through a procedure call, create a temporary copy of the array in heap and
provide the copy to the procedure by reference.

Changing the Calling Convention
Put _pascal or _fortran in the C function declaration to specify the Pascal
calling convention.

Array Storage
Array declarations give the number of elements. A1[a][b] declares a twodimensional array in C with a rows and b columns. By default, the array’s
lower bound is zero. Arrays are stored by the compiler in row-major order. By
default, passing arrays from C passes a pointer to the first element of the array.

String Format
C stores strings as arrays of bytes and uses a null character as the end-of-string
delimiter. For example, consider the string declared as follows:
char msg[] = "string of text"

The string occupies 15 bytes of memory as:

Figure 12.1

C String Format

Since msg is an array of characters, it is passed by reference.

External Data
In C, the extern keyword tells the compiler that the data or function is external.
You can define a static data object in a C module by defining a data object
outside all functions and subroutines. Do not use the static keyword in C with a
data object that you want to be public.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 317 of 11 Printed: 10/02/00 04:21 PM

318

Programmer’s Guide

Structure Alignment
By default, C uses word alignment (unpacked storage) for all data objects longer
than 1 byte. This storage method specifies that occasional bytes may be added
as padding, so that word and doubleword objects start on an even boundary. In
addition, all nested structures and records start on a word boundary. MASM
aligns on byte boundaries by default.
When converting .H files with H2INC, you can use the /Zp command-line
option to specify structure alignment. If you do not specify the /Zp option,
H2INC uses word-alignment. Without H2INC, set the alignment to 2 when
declaring the MASM structure, compile the C module with /Zp1, or assemble
the MASM module with /Zp2.

Compiling and Linking
Use the same memory model for both C and MASM.

Returning Values
The assembler returns simple data types in registers. Table 12.2 shows the
register conventions for returning simple data types to a C program.
Table 12.2

Data Type

Registers

char

short, near, int (16-bit)

short, near, int (32-bit)

EAX

long, far (16-bit)

High-order portion (or segment address) in DX;
low-order portion (or offset address) in AX

long, far (32-bit)

High-order portion (or segment address) in EDX;
low-order portion (or offset address) in EAX

Procedures using the C calling convention and returning type float or type
double store their return values into static variables. In multi-threaded
programs, this could mean that the return value may be overwritten. You can
avoid this by using the Pascal calling convention for multi-threaded programs so
float or double values are passed on the stack.
Structures less than 4 bytes long are returned in DX:AX. To return a longer
structure from a procedure that uses the C calling convention, you must copy
the structure to a global variable and then return a pointer to that variable in the
AX register (DX:AX, if you compiled in compact, large, or huge model or if the
variable is declared as a far pointer).

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 318 of 12 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

319

Structures, Records, and User-Defined Data Types
You can pass structures, records, and user-defined types as arguments by value
or by reference.

Writing Procedure Prototypes
The H2INC utility simplifies the task of writing prototypes for the C functions
you want to call from MASM. The C prototype converted by H2INC into a
MASM prototype allows INVOKE to correctly call the C function. Here are
some examples of C functions and the MASM prototypes created with H2INC.
/* Function Prototype Declarations to Convert with H2INC */
long checktypes (
char *name,
unsigned char a,
int b,
float d,
unsigned int *num );
my_func (float fNum, unsigned int x);
extern my_func1 (char *argv[]);
struct videoconfig _far * _far pascal my_func2 (int, scri );

For these C prototypes, H2INC generates this code:
@proto_0

TYPEDEF

checktypes

PROTO

PROTO C :PTR SBYTE, :BYTE,
:SWORD, :REAL4, :PTR WORD
@proto_0

@proto_1
my_func

TYPEDEF
PROTO

PROTO C :REAL4, :WORD
@proto_1

@proto_2
my_func1

TYPEDEF
PROTO

PROTO C :PTR PTR SBYTE
@proto_2

@proto_3
my_func2

TYPEDEF
PROTO

PROTO FAR PASCAL :SWORD, :scri
@proto_3

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 319 of 13 Printed: 10/02/00 04:21 PM

320

Programmer’s Guide

Example
As shown in the following short example, the main module (written in C) calls
an assembly routine, Power2.
#include
extern int Power2( int factor, int power );
void main()
{
printf( "3 times 2 to the power of 5 is %d\n", Power2( 3, 5 ) );
}

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 320 of 14 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

321

Figure 12.2 shows how functions that observe the C calling convention use the
stack frame.

Figure 12.2

C Stack Frame

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 321 of 15 Printed: 10/02/00 04:21 PM

322

Programmer’s Guide

The MASM module that contains the Power2 routine looks like this:
.MODEL

small, c

Power2

PROTO C factor:SWORD, power:SWORD
.CODE

Power2

PROC
mov
mov
shl

Power2

C factor:SWORD,
ax, factor
cx, power
ax, cl

power:SWORD
; Load Arg1 into AX
; Load Arg2 into CX
; AX = AX * (2 to power of CX)
; Leave return value in AX

ret
ENDP
END

The MASM procedure declaration for the Power2 routine specifies the C
langtype and the parameters expected by the procedure. The langtype specifies
the calling and naming conventions for the interface between MASM and C.
The routine is public by default. When the C module calls Power2, it passes
two arguments, 3 and 5 by value.

Using the C Startup Code
This section explains how to write an assembly-language program that can call C
library functions. It links with the C startup module, which performs the
necessary initialization required by the library functions.
You must follow these steps when writing such a program:
1. Specify the C convention in the .MODEL statement.
2. Include the following (optional) statement to note linkage with the C startup
module:
EXTERN

_acrtused:abs

3. Prototype or declare as external all C functions the program references.
4. Include a public procedure called main in your assembly-language module.
The C startup code calls _main (which is why all C programs begin with a
main function). This procedure serves as the effective entry point for your
program.
5. Omit an entry point in the program’s END directive. The C startup code
serves as the true entry point when the program runs.
6. Assemble with ML’s /Cx switch to preserve the case of nonlocal names.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 322 of 16 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

323

The following example serves as a template for these steps. The program calls
the C run-time function printf to display two variables.
.MODEL
EXTERN

printf

format

small, c
_acrtused:abs
.
.
.
PROTO
NEAR,
pstring:NEAR PTR BYTE,
num1:WORD, num2:VARARG
.DATA
BYTE
'%i
%i', 13, 0

; Step 1: declare C conventions
; Step 2: bring in C startup

; Step 3: prototype
;
external C
;
routines

.CODE
main

PROC
.
.
.
INVOKE
.
.
.
END

PUBLIC

; Step 4: C startup calls here

printf, OFFSET format, ax, bx

; Step 5: no label on END

The C++/MASM Interface
C++ can apply a protocol called a “linkage specification” to mixed-language
procedures. This lets you link C++ code in the same way as C code. All
information in the preceding section applies when linking assembly-language and
C++ routines through the C linkage specification.
The C linkage specification forces the C++ compiler to adopt C conventions —
which are not the same as C++ conventions — for listed routines. Since MASM
does not specifically support C++ conventions, set the C linkage specification in
your C++ code for all mixed-language routines, as shown here:
extern “C” declaration
where declaration is the prototype of an exported C++ function or an imported
assembly-language procedure. You can bracket a list of declarations:
extern "C"
{
int
WriteLine( short attr, char *string );
void
GoExit( int err );
}

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 323 of 17 Printed: 10/02/00 04:21 PM

324

Programmer’s Guide

or apply the specification to individual prototypes:
extern "C" int
extern "C" void

WriteLine( short attr, char *string );
GoExit( int err );

Note the syntax remains the same whether WriteLine and GoExit are
exported C++ functions or imported assembly-language routines. The linkage
specification applies only to called routines, not to external variables. Use the
extern keyword (without the “C”) as you normally would when identifying
objects external to the C++ module.

The FORTRAN/MASM Interface
This section summarizes the specific details important to calling FORTRAN
procedures or receiving arguments from FORTRAN routines that call MASM
routines. It includes a sample MASM and FORTRAN module.
A FORTRAN procedure follows the Pascal calling convention by default. This
convention passes arguments in the order listed, and the calling procedure
removes the arguments from the stack. The naming convention converts all
exported names to uppercase.

Compatible Data Types
This list shows the FORTRAN data types that are equivalent to the MASM 6.1
data types.
FORTRAN Type

Equivalent MASM Type

CHARACTER*1

BYTE

INTEGER*1

SBYTE

INTEGER*2

SWORD

REAL*4

REAL4

INTEGER*4

SDWORD

REAL*8, DOUBLE PRECISION

REAL8

Naming Restrictions
FORTRAN allows 31 characters for identifier names. A digit or an underscore
cannot be the first character in an identifier name.

Argument-Passing Defaults
By default, FORTRAN passes arguments by reference as far addresses if the
FORTRAN module is compiled in large or huge memory model. It passes them
as near addresses if the FORTRAN module is compiled in medium model.
Versions of FORTRAN prior to Version 4.0 always require large model.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 324 of 18 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

325

The FORTRAN compiler passes an argument by value when declared with the
VALUE attribute. This declaration can occur either in a FORTRAN
INTERFACE block (which determines how to pass an argument) or in a
function or subroutine declaration (which determines how to receive an
argument).
In FORTRAN you can apply the NEAR (or FAR) attribute to reference
parameters. These keywords override the default. They have no effect when
they specify the same method as the default.

Changing the Calling Convention
A call to a FORTRAN function or subroutine declared with the PASCAL or C
attribute passes all arguments by value in the parameter list (except for
parameters declared with the REFERENCE attribute). This change in default
passing method applies to function and subroutine definitions as well as to the
functions and subroutines described by INTERFACE blocks.

Array Storage
When you declare FORTRAN arrays, you can specify any integer for the lower
bound (the default is 1). The FORTRAN compiler stores all arrays in columnmajor order — that is, the leftmost subscript increments most rapidly. For
example, the first seven elements of an array defined as A[3,4] are stored as
A[1,1], A[2,1], A[3,1], A[1,2], A[2,2], A[3,2], A[1,3]

String Format
FORTRAN stores strings as a series of bytes at a fixed location in memory, with
no delimiter at the end of the string. When passing a variable-length FORTRAN
string to another language, you need to devise a method by which the target
routine can find the end of the string.
Consider the string declared as
CHARACTER*14 MSG
MSG = 'String of text'

The string is stored in 14 bytes of memory like this:

Figure 12.3

FORTRAN String Format

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 325 of 19 Printed: 10/02/00 04:21 PM

326

Programmer’s Guide

Strings are passed by reference. Although FORTRAN has a method for passing
length, the variable-length FORTRAN strings cannot be used in a mixedlanguage interface because other languages cannot access the temporary variable
that FORTRAN uses to communicate string length. However, fixed-length
strings can be passed if the FORTRAN INTERFACE statement declares the
length of the string in advance.

External Data
FORTRAN routines can directly access external data. In FORTRAN you can
declare data to be external by adding the EXTERN attribute to the data
declaration. You can also access a FORTRAN variable from MASM if it is
declared in a COMMON block.
A FORTRAN program can call an external assembly procedure with the use of
the INTERFACE statement. However, the INTERFACE statement is not
strictly necessary unless you intend to change one of the FORTRAN defaults.

Structure Alignment
By default, FORTRAN uses word alignment (unpacked storage) for all data
objects larger than 1 byte. This storage method specifies that occasional bytes
may be added as padding, so that word and doubleword objects start on an even
boundary. In addition, all nested structures and records start on a word
boundary. The MASM default is byte-alignment, so you should specify an
alignment of 2 for MASM structures or use the /Zp1 option when compiling in
FORTRAN.

Compiling and Linking
Use the same memory model for the MASM and FORTRAN modules.

Returning Values
You must use a special convention to return floating-point values, records, userdefined types, arrays, and values larger than 4 bytes to a FORTRAN module
from an assembly procedure. The FORTRAN module creates space in the stack
segment to hold the actual return value. When the call to the assembly
procedure is made, an extra parameter is passed. This parameter is the last one
pushed. The segment address of the return value is contained in SS.
In the assembly procedure, put the data for the return value at the location
pointed to by the return value offset. Then copy the return-value offset (located
at BP + 6) to AX, and copy SS to DX. This is necessary because the calling
module expects DX:AX to point to the return value.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 326 of 20 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

327

Structures, Records, and User-Defined Data Types
The FORTRAN structure variable, defined with the STRUCTURE keyword
and declared with the RECORD statement, is equivalent to the Pascal
RECORD and the C struct. You can pass structures as arguments by value or
by reference (the default).
The FORTRAN types COMPLEX*8 and COMPLEX*16 are not directly
implemented in MASM. However, you can write structures that are equivalent.
The type COMPLEX*8 has two fields, both of which are 4-byte floating-point
numbers; the first contains the real component, and the second contains the
imaginary component. The type COMPLEX is equivalent to the type
COMPLEX*8.
The type COMPLEX*16 is similar to COMPLEX*8. The only difference is
that each field of the former contains an 8-byte floating-point number.
A FORTRAN LOGICAL*2 is stored as a 1-byte indicator value (1=true,
0=false) followed by an unused byte. A FORTRAN LOGICAL*4 is stored as a
1-byte indicator value followed by three unused bytes. The type LOGICAL is
equivalent to LOGICAL*4, unless $STORAGE:2 is in effect.
To pass or receive a FORTRAN LOGICAL type, declare a MASM structure
with the appropriate fields.

Varying Number of Arguments
In FORTRAN, you can call routines with a variable number of arguments by
including the VARYING attribute in your interface to the routine, along with the
C attribute. You must use the C attribute because a variable number of
arguments is possible only with the C calling convention. The VARYING
attribute prevents FORTRAN from enforcing a matching number of parameters.

Pointers and Addresses
FORTRAN programs can determine near and far addresses with the
LOCNEAR and LOCFAR functions. Store the result as INTEGER*2 (with
the LOCNEAR function) or as INTEGER*4 (with the LOCFAR function). If
you pass the result of LOCNEAR or LOCFAR to another language, be sure to
pass by value.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 327 of 21 Printed: 10/02/00 04:21 PM

328

Programmer’s Guide

Example
In the following example, the FORTRAN module calls an assembly procedure
that calculates A*2^B, where A and B are the first and second parameters,
respectively. This is done by shifting the bits in A to the left B times.
INTERFACE TO INTEGER*2 FUNCTION POWER2(A, B)
INTEGER*2 A, B
END
PROGRAM MAIN
INTEGER*2 POWER2
INTEGER*2 A, B
A = 3
B = 5
WRITE (*, *) '3 TIMES 2 TO THE B OR 5 IS ',POWER2(A, B)
END

To understand the assembly procedure, consider how the parameters are placed
on the stack, as illustrated in Figure 12.4.

Figure 12.4

FORTRAN Stack Frame

Figure 12.4 assumes that the FORTRAN module is compiled in large model. If
you compile the FORTRAN module in medium model, then each argument is
passed as a 2-byte, not 4-byte, address. The return address is 4 bytes long
because procedures called from FORTRAN must always be FAR.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 328 of 22 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

329

The assembler code looks like this:
.MODEL LARGE, FORTRAN
Power2

PROTO

FORTRAN, pFactor:FAR PTR SWORD, pPower:FAR PTR SWORD

.CODE
Power2

Power2

PROC

FORTRAN, pFactor:FAR PTR SWORD, pPower:FAR PTR SWORD

les
mov
les
mov
shl
ret
ENDP
END

bx,
ax,
bx,
cx,
ax,

pFactor
es:[bx]
pPower
es:[bx]
cl

;
;
;
;
;
;

ES:BX points to factor
AX = value of factor
ES:BX points to power
CX = value of power
Multiply by 2^power
Return result in AX

The Basic/MASM Interface
This section explains how to call MASM procedures or functions from Basic
and how to receive Basic arguments for the MASM procedure. Pascal is the
default naming and calling convention, so all lowercase letters are converted to
uppercase. Routines defined with the FUNCTION keyword return values, but
routines defined with SUB do not. Basic DEF FN functions and GOSUB
routines cannot be called from another language.
The information provided pertains to Microsoft’s Basic and QuickBasic
compilers. Differences between the two compilers are noted when necessary.

Compatible Data Types
The following list shows the Basic data types that are equivalent to the MASM
6.1 data types.
Basic Type

Equivalent MASM Type

STRING*1

WORD

INTEGER (X%)

SWORD

SINGLE (X!)

REAL4

LONG (X&),
CURRENCY

SDWORD

DOUBLE (X#)

REAL8

Naming Conventions
Basic recognizes up to 40 characters of a name. In the object code, Basic also
drops any of its reserved characters: %, &, !, #, @, &.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 329 of 23 Printed: 10/02/00 04:21 PM

330

Programmer’s Guide

Argument-Passing Defaults
Basic can pass data in several ways and can receive it by value or by near
reference. By default, Basic arguments are passed by near reference as 2-byte
addresses. To pass a near address, pass only the offset; if you need to pass a far
address, pass the segment and offset separately as integer arguments. Pass the
segment address first, unless you have specified C compatibility with the
CDECL keyword.
Basic passes each argument in a call by far reference when CALLS is used to
invoke a routine. You can also use SEG to modify a parameter in a preceding
DECLARE statement so that Basic passes that argument by far reference. To
pass any other variable type by value, apply the BYVAL keyword to the
argument in the DECLARE statement. You cannot pass arrays and userdefined types by value.
DECLARE SUB Test(BYVAL a%, b%, SEG c%)
CALL Test(x%, y%, z%)
CALLS Test(x%, y%, z%)

This CALL statement passes the first argument (a%) by value, the second
argument (b%) by near reference, and the third argument (c%) by far reference.
The statement
CALLS Test2(x%, y%, z%)

passes each argument by far reference.

Changing the Calling Convention
Including the CDECL keyword in the Basic DECLARE statement enables the
C calling and naming conventions. This also allows a call to a MASM procedure
with a varying number of arguments.

Array Storage
The DIM statement sets the number of dimensions for a Basic array and also
sets the array’s maximum subscript value. In the array declaration DIM
x(a,b), the upper bounds (the maximum number of values possible) of the
array are a and b. The default lower bound is 0. The default upper bound for an
array subscript is 10.
The default for column storage in Basic is column-major order, as in
FORTRAN. For an array defined as DIM Arr%(3,3), reference the last
element as Arr%(3,3). The first five elements of Arr (3,3) are
Arr(0,0), Arr(1,0), Arr(2,0), Arr(0,1), Arr(1,1)

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 330 of 24 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

331

When you pass an array from Basic to a language that stores arrays in rowmajor order, use the command-line option /R when compiling the Basic module.
Most Microsoft languages permit you to reference arrays directly. Basic uses an
array descriptor, however, which is similar in some respects to a Basic string
descriptor. The array descriptor is necessary because Basic handles memory
allocation for arrays dynamically, and thus may shift the location of the array in
memory.
A reference to an array in Basic is really a near reference to an array descriptor.
Array descriptors are always in DGROUP, even though the data may be in far
memory. Array descriptors contain information about type, dimensions, and
memory locations of data. You can safely pass arrays to MASM routines only if
you follow three rules:
u

Pass the array’s address by applying the VARPTR function to the first
element of the Basic array and passing the result by value. To pass the far
address of the array, apply both the VARPTR and VARSEG functions and
pass each result by value. The receiving language gets the address of the first
element and considers it to be the address of the entire array. It can then
access the array with its normal array-indexing syntax.
The MASM routine that receives the array should not call back to one of the
calling program’s routines before it has finished processing the array.
Changing data within the caller’s heap — even data unrelated to the array —
may change the array’s location in the heap. This would invalidate any
further work the called routine performs, since the routine would be
operating on the array’s old location.
Basic can pass any member of an array by value. When passing individual
array elements, these restrictions do not apply.

You can apply LBOUND and UBOUND to a Basic array to determine lower
and upper bounds, and then pass the results to another routine. This way, the
size of the array does not need to be determined in advance.

String Format
Basic maintains a 4-byte string descriptor for each string, as shown in the
following. The first field of the string descriptor contains a 2-byte integer
indicating the length of the actual string text. The second field contains the offset
address of this text within the caller’s data segment.

Figure 12.5

Basic String Descriptor Format

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 331 of 25 Printed: 10/02/00 04:21 PM

332

Programmer’s Guide

An assembly-language procedure can store a Basic string descriptor as a simple
structure, like this:
DESC
len
off
DESC

STRUCT
WORD
WORD
ENDS

?
?

string
sdesc

BYTE
DESC

"This text referenced by a string descriptor"
(LENGTHOF string, string)

; Length of string
; Offset of string

Version 7.0 or later of the Microsoft Basic Compiler provides new functions that
access string descriptors. These functions simplify the process of sharing Basic
string data with routines written in other languages.
Earlier versions of Basic offer the LEN (Length) and SADD (String Address)
functions, which together obtain the information stored in a string descriptor.
LEN returns the length of a string in bytes. SADD returns the offset address of
a string in the data segment. The caller must provide both pieces of information
so the called procedure can locate and read the entire string. The address
returned by SADD is declared as type INTEGER but is actually equivalent to a
C near pointer.
If you need to pass the far address of a string, use the SSEGADD (String
Segment Address) function of Microsoft Basic version 7.0 or later. You can also
determine the segment address of the first element with VARSEG.

External Data
Declaring global data in Basic follows the same two-step process as in other
languages:
1. Declare shareable data in Basic with the COMMON statement.
2. Identify the shared variables in your assembly-language procedures with the
EXTERN keyword. Place the EXTERN statement outside of a code or data
segment when declaring far data.

Structure Alignment
Basic packs user-defined types. For MASM structures to be compatible, select
byte-alignment.

Compiling and Linking
Always use medium model in assembly-language procedures linked with Basic
modules. If you are listing other libraries on the LINK command line, specify
Basic libraries first. (There are differences between the QBX and command-line
compilation. See your Basic documentation.)

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 332 of 26 Printed: 10/02/00 04:21 PM

Chapter 12 Mixed-Language Programming

333

Returning Values
Basic follows the usual convention of returning values in AX or DX:AX. If the
value is not floating point, an array, or a structured type, or if it is less than 4
bytes long, then the 2-byte integers should be returned from the MASM
procedure in AX and 4-byte integers should be returned in DX:AX. For all other
types, return the near offset in AX.

User-Defined Data Types
The Basic TYPE statement defines structures composed of individual fields.
These types are equivalent to the C struct, FORTRAN record (declared with
the STRUCTURE keyword), and Pascal Record types.
You can use any of the Basic data types except variable-length strings or
dynamic arrays in a user-defined type. Once defined, Basic types can be passed
only by reference.

Varying Number of Arguments
You can vary the number of arguments in Basic when you change the calling
convention with CDECL. To call a function with a varying number of
arguments, you also need to suppress the type checking that normally forces a
call to be made with a fixed number of arguments. In Basic, you can remove
this type checking by omitting a parameter list from the DECLARE statement.

Pointers and Addresses
VARSEG returns a variable’s segment address, and VARPTR returns a
variable’s offset address. These intrinsic Basic functions enable your program to
pass near or far addresses.

Example
This example calls the Power2 procedure in the MASM 6.1 module.
DEFINT A-Z
DECLARE FUNCTION Power2 (A AS INTEGER, B AS INTEGER)
PRINT "3 times 2 to the power of 5 is ";
PRINT Power2(3, 5)
END

The first argument, A, is higher in memory than B because Basic pushes
arguments in the same order in which they appear.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 333 of 27 Printed: 10/02/00 04:21 PM

334

Programmer’s Guide

Figure 12.6 shows how the arguments are placed on the stack.

Figure 12.6

Basic Stack Frame

The assembly procedure can be written as follows:

Power2
Power2

Power2

.MODEL

medium

PROTO
.CODE
PROC

PASCAL, factor:PTR WORD, power:PTR WORD

mov
mov
mov
mov
shl
ret
ENDP
END

PASCAL, factor:PTR WORD, power:PTR WORD
bx,
ax,
bx,
cx,
ax,

WORD PTR factor
[bx]
WORD PTR power
[bx]
cl

;
;
;
;
;

BX points to factor
Load factor into AX
BX points to power
Load power into CX
AX = AX * (2 to power of CX)

Note that each parameter must be loaded in a two-step process because the
address of each is passed rather than the value. The return address is 4 bytes
long because procedures called from Basic must be FAR.

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 334 of 28 Printed: 10/02/00 04:21 PM

334

Programmer’s Guide

Filename: LMAPGC12.DOC Project:
Template: MSGRIDA1.DOT Author: Mike Eddy Last Saved By: Ruth L Silverio
Revision #: 71 Page: 334 of 30 Printed: 10/02/00 04:21 PM

335

C H A P T E R

1 3

Writing 32-Bit Applications

This chapter is an introduction to 32-bit programming for the 80386. The
guidelines in this chapter also apply to the 80486 processor, which is basically a
faster 80386 with the equivalent of a 80387 floating-point processor. Since you
are already familiar with 16-bit real-mode programming, this chapter covers the
differences between 16-bit programming and 32-bit protected-mode
programming.
The 80386 processor (and its successors such as the 80486) can run in real
mode, virtual-86 mode, and in protected mode. In real and virtual-86 modes, the
80386 can run 8086/8088 programs. In protected mode, it can run 80286
programs. The 386 also extends the features of protected mode to include 32-bit
operations and segments larger than 64K.
The MS-DOS operating system directly supports 8086/8088 programs, which it
runs either in real mode or virtual-86 mode. Native 32-bit 80386 programs can
be run by using a “DOS extender,” by using the WINMEM32.DLL facility of
Microsoft Windows 3.x, or by running a native 32-bit operating system, such as
Microsoft Windows NT. You can use MASM to generate object code (OMF or
COFF) for 32-bit programs. To do this, you will need a software development
kit such as the Windows SDK for the target environment. Such kits include the
linker and other components specific to your chosen operating environment.

32-Bit Memory Addressing
The 80386 has six segment registers. Four of these are familiar to 8086/8088
programmers: CS (Code Segment), SS (Stack Segment), DS (Data Segment),
and ES (Extra Segment). The two additional registers, FS and GS, are used as
data segment registers.
Memory addresses on 80x86 machines consist of two parts — a segment and an
offset. In real-mode programs, the segment is a 16-bit number and the offset is a
16-bit number. Effective addresses are calculated by multiplying the segment by

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 335 of 1 Printed: 10/02/00 04:21 PM

336

Programmer’s Guide

16 and adding the offset to it. In protected mode, the segment value is not used
directly as a number, but instead is an index to a table of “selectors.” Each
selector describes a block of memory, including attributes such as the size and
location of the block, and the access rights the program has to it (read, write,
execute). The effective address is calculated by adding the offset to the base
address of the memory block described by the selector.
All segment registers are 16 bits wide. The offset in a 32-bit protected-mode
program is itself 32 bits wide, which means that a single segment can address up
to 4 gigabytes of memory. Because of this large range, there is little need to use
segment registers to extend the range of addresses in 32-bit programs. If all six
segment registers are initially set to the same value, then the rest of the program
can ignore them and treat the processor as if it used a 32-bit linear address
space. This is called 0:32, or flat, addressing. (The full segmented 32-bit
addressing mode, in which the segment registers can contain different values, is
called 16:32 addressing.) Flat addressing is used by the Windows NT operating
system.

Figure 13.1

32-Bit Register Set

MASM Directives for 32-Bit Programming
If you use the simplified segment directives, a 32-bit program is surprisingly
similar to a program for MS-DOS. Here are the differences:

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 336 of 2 Printed: 10/02/00 04:21 PM

Chapter 13 Writing 32-Bit Applications

337

Supply the .386 directive, which enables the 32-bit programming features of
the 386 and its successors. The .386 directive must precede the .MODEL
directive.

For flat-model programming, use the directive
.MODEL flat, stdcall
which tells the assembler to assume flat model (0:32) and to use the
Windows NT standard calling convention for subroutine calls.

Precede your data declarations with the .DATA directive.

Precede your instruction codes with the .CODE directive.
At the end of the source file, place an END directive.

Sample Program
The following sample is a 32-bit assembly language subroutine, such as might be
called from a 32-bit C program written for the Windows NT operating system.
The program illustrates the use of a variety of directives to make assembly
language easier to read and maintain. Note that with 32-bit flat model
programming, there is no longer any need to refer to segment registers, since
these are artifacts of segmented addressing.

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 337 of 3 Printed: 10/02/00 04:21 PM

338

Programmer’s Guide
;* szSearch - An example of 32-bit assembly programming using MASM 6.1
;*
;* Purpose: Search a buffer (rgbSearch) of length cbSearch for the
;*
first occurrence of szTok (null terminated string).
;*
;* Method: A variation of the Boyer-Moore method
;*
1. Determine length of szTok (n)
;*
2. Set array of flags (rgfInTok) to TRUE for each character
;*
in szTok
;*
3. Set current position of search to rgbSearch (pbCur)
;*
4. Compare current position to szTok by searching backwards
;*
from the nth position. When a comparison fails at
;*
position (m), check to see if the current character
;*
in rgbSearch is in szTok by using rgfInTok. If not,
;*
set pbCur to pbCur+(m)+1 and restart compare. If
;*
pbCur reached, increment pbCur and restart compare.
;*
5. Reset rgfInTok to all 0 for next instantiation of the
;*
routine.

FALSE
TRUE

.386
.MODEL

flat, stdcall

EQU
EQU

0
NOT FALSE

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 338 of 4 Printed: 10/02/00 04:21 PM

Chapter 13 Writing 32-Bit Applications
.DATA
; Flags buffer - data initialized to FALSE. We will
; set the appropriate flags to TRUE during initialization
; of szSearch and reset them to FALSE before exit.
rgfInTok
BYTE
256 DUP (FALSE);
.CODE
PBYTE

TYPEDEF PTR BYTE

szSearch PROC PUBLIC USES esi edi,
rgbSearch:PBYTE,
cbSearch:DWORD,
szTok:PBYTE
; Initialize flags buffer. This tells us if a character is in
; the search token - Note how we use EAX as an index
; register. This can be done with all extended registers.
mov
esi, szTok
xor
eax, eax
.REPEAT
lodsb
mov
BYTE PTR rgfInTok[eax], TRUE
.UNTIL (!AL)
; Save count of szTok bytes in EDX
mov
edx, esi
sub
edx, szTok
dec
edx
; ESI will always point to beginning of szTok
mov
esi, szTok
; EDI will point to current search position
; and will also contain the return value
mov
edi, rgbSearch
; Store pointer
mov
add
sub

to end of rgbSearch in EBX
ebx, edi
ebx, cbSearch
ebx, edx

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 339 of 5 Printed: 10/02/00 04:21 PM

339

340

Programmer’s Guide
; Initialize ECX with length of szTok
mov
ecx, edx
.WHILE ( ecx != 0 )
dec
ecx
; Move index to current
mov
al, [edi+ecx]
;
characters to compare
;
;
;
;

;
;

;
;
;

If the current byte in the buffer doesn't exist in the
search token, increment buffer pointer to current position
+1 and start over. This can skip up to 'EDX'
bytes and reduce search time.
.IF
!(rgfInTok[eax])
add
edi, ecx
inc
edi
; Initialize ECX with
mov
ecx, edx
;
length of szTok
Otherwise, if the characters match, continue on as if
we have a matching token
.ELSEIF (al == [esi+ecx])
.CONTINUE
Finally, if we have searched all szTok characters,
and land here, we have a mismatch and we increment
our pointer into rgbSearch by one and start over.
.ELSEIF (!ecx)
inc
edi
mov
ecx, edx
.ENDIF

; Verify that we haven't searched beyond the buffer.
.IF
(edi > ebx)
mov
edi, 0
; Error value
.BREAK
.ENDIF
.ENDW
; Restore flags
mov
xor
.REPEAT
lodsb
mov
.UNTIL

in rgfInTok to 0 (for next time).
esi, szTok
eax, eax

BYTE PTR rgfInTok[eax], FALSE
!AL

; Put return value in eax
mov
eax, edi
ret
szSearch ENDP
end

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 340 of 6 Printed: 10/02/00 04:21 PM

340

Programmer’s Guide

Filename: LMAPGC13.DOC Project:
Template: MSGRIDA1.DOT Author: Terri Sharkey Last Saved By: Ruth L Silverio
Revision #: 27 Page: 340 of 8 Printed: 10/02/00 04:21 PM

341

A P P E N D I X

Differences Between
MASM 6.1 and 5.1

For the many users who come to version 6.1 of the Microsoft Macro Assembler
directly from the popular MASM 5.1, this appendix describes the differences
between the two versions. Version 6.1 contains significant changes, including:
u

u
u

An integrated development environment called Programmer’s WorkBench
(PWB) from which you can write, edit, debug, and execute code.
Expanded functionality for structures, unions, and type definitions.
New directives for generating loops and decision statements, and for
declaring and calling procedures.
Simplified methods for applying public attributes to variables and routines in
multiple-module programs.
Enhancements for writing and using macros.
Flat-model support for Windows NT and new instructions for the 80486
processor.

The OPTION M510 directive (or the /Zm command-line switch) assures nearly
complete compatibility between MASM 6.1 and MASM 5.1. However, to take
full advantage of the enhancements in MASM 6.1, you will need to rewrite
some code written for MASM 5.1.
The first section of this appendix describes the new or enhanced features in
MASM 6.1. The second section, “Compatibility Between MASM 5.1 and 6.1,”
explains how to:
u
u

Minimize the number of required changes with the OPTION directive.
Rewrite your existing assembly code, if necessary, to take advantage of the
assembler’s enhancements.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 341 of 1 Printed: 10/02/00 04:18 PM

342

Programmer’s Guide

New Features of Version 6.1
This section gives an overview of the new features of MASM 6.1 and provides
references to more detailed information elsewhere in the documentation. For full
explanations and coding examples, see the documentation listed in the crossreferences.

The Assembler, Environment, and Utilities
Most of the executable files provided with MASM 6.1 are new or revised. For a
complete list of these files, read the PACKING.TXT file on the distribution
disk. The book Getting Started also provides information about setting up the
environment, assembler, and Help system.

The Assembler
The macro assembler, named ML.EXE, can assemble and link in one step. Its
new 32-bit operation gives ML.EXE the ability to handle much larger source
files than MASM 5.1. The command-line options are new. For example, the /Fl
and /Sc options generate instruction timings in the listing file. Command-line
options are case-sensitive and must be separated by spaces.
For backward compatibility with MASM 5.1 makefiles, MASM 6.1 includes the
MASM.EXE utility. MASM.EXE translates MASM 5.1 command-line options
to the new MASM 6.1 command-line options and calls ML.EXE. See the
Reference book for details.

H2INC
H2INC converts C include files to MASM include files. It translates data
structures and declarations but does not translate executable code. For more
information, see Chapter 20 of Environment and Tools.

NMAKE
NMAKE replaces the MAKE utility. NMAKE provides new functions for
evaluating target files and more flexibility with macros and command-line
options. For more information, see Environment and Tools.

Integrated Environment
PWB is an integrated development environment for writing, developing, and
debugging programs. For information on PWB and the CodeView debugging
application, see Environment and Tools.

Online Help
MASM 6.1 incorporates the Microsoft Advisor Help system. Help provides a
vast database of online help about all aspects of MASM, including the syntax

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 342 of 2 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

343

and timings for processor and coprocessor instructions, directives, commandline options, and support programs such as LINK and PWB.
For information on how to set up the help system, see Getting Started. You can
invoke the help system from within PWB or from the QuickHelp program
(QH).

HELPMAKE
You can use the HELPMAKE utility to create additional help files from ASCII
text files, allowing you to customize the online help system. For more
information, see Environment and Tools.

Other Programs
MASM 6.1 contains the most recent versions of LINK, LIB, BIND, CodeView,
and the mouse driver. The CREF program is not included in MASM 6.1. The
Source Browser provides the information that CREF provided under MASM
5.1. For more information on the Source Browser, see Chapter 5 of
Environment and Tools or Help.

Segment Management
This section lists the changes and additions to memory-model support and
directives that relate to memory model.

Predefined Symbols
The following predefined symbols (also called predefined equates) provide
information about simplified segments:
Predefined Symbol

Value

@stack

DGROUP for near stacks, STACK for far stacks

@Interface

Information about language parameters

@Model

Information about the current memory model

@Line

The source line in the current file

@Date

The current date

@FileCur

The current file

@Time

The current time

@Environ

The current environment variables

For more information about predefined symbols, see “Predefined Symbols” in
Chapter 1.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 343 of 3 Printed: 10/02/00 04:18 PM

344

Programmer’s Guide

Enhancements to the ASSUME Directive
MASM automatically generates ASSUME values for the code segment register
(CS). It is no longer necessary to include lines such as
ASSUME CS:MyCodeSegment

in your programs. In addition, the ASSUME directive can include ERROR,
FLAT, or register:type. MASM 6.1 issues a warning when you specify
ASSUME values for CS other than the current segment or group.
For more information, see “Setting the ASSUME Directive for Segment
Registers” in Chapter 2 and “Defining Register Types with ASSUME” in
Chapter 3.

Relocatable Offsets
For compatibility with applications for Windows, the LROFFSET operator can
calculate a relocatable offset, which is resolved by the loader at run time. See
Help for details.

Flat Model
MASM 6.1 supports the flat-memory model of Windows NT, which allows
segments as large as 4 gigabytes. All other memory models limit segment size to
64K for MS-DOS and Windows. For more information about memory models,
see “Defining Basic Attributes with .MODEL” in Chapter 2.

Data Types
MASM 6.1 supports an improved data typing. This section summarizes the
improved forms of data declarations in MASM 6.1.

Defining Typed Variables
You can now use the type names as directives to define variables. Initializers are
unsigned by default. The following example lines are equivalent:
var1
var1

DB
BYTE

25
25

Signed Types
You can use the SBYTE, SWORD, and SDWORD directives to declare signed
data. For more information about these directives, see “Allocating Memory for
Integer Variables” in Chapter 4.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 344 of 4 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

345

Floating-Point Types
MASM 6.1 provides the REAL4, REAL8, and REAL10 directives for
declaring floating-point variables. For information on these type directives, see
“Declaring Floating-Point Variables and Constants” in Chapter 6 .

Qualified Types
Type definitions can now include distance and language type attributes.
Procedures, procedure prototypes, and external declarations let you specify the
type as a qualified type. A complete description of qualified types is provided in
the section “Data Types” in Chapter 1.

Structures
Changes to structures since MASM 5.1 include:
u
u

Structures can be nested.
The names of structure fields need not be unique. As a result, you must
qualify references to field names.
Initialization of structure variables can continue over multiple lines provided
the last character in the line before the comment field is a comma.
Curly braces and angle brackets are equivalent.

For example, this code works in MASM 6.1:
SCORE
team1
score1
team2
score2
SCORE
first

STRUCT
BYTE
BYTE
BYTE
BYTE
ENDS
SCORE

mov

10 DUP (?)
?
10 DUP (?)
?

{"BEARS", 20,
"CUBS", 10 }

; This comment is allowed.

al, [bx].score.team1 ; Field name must be qualified
;
with structure name.

You can use OPTION OLDSTRUCTS or OPTION M510 to enable MASM
5.1 behavior for structures. See “Compatibility between MASM 5.1 and 6.1,”
later in this appendix. For more information on structures and unions, see
“Structures and Unions” in Chapter 5.

Unions
MASM 6.1 allows the definition of unions with the UNION directive. Unions
differ from structures in that all fields within a union occupy the same data
space. For more information, see “Structures and Unions” in Chapter 5.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 345 of 5 Printed: 10/02/00 04:18 PM

346

Programmer’s Guide

Types Defined with TYPEDEF
The TYPEDEF directive defines a type for use later in the program. It is most
useful for defining pointer types. For more information on defining types, see
“Data Types” in Chapter 1, and “Defining Pointer Types with TYPEDEF” in
Chapter 3.

Names of Identifiers
MASM 6.1 accepts identifier names up to 247 characters long. All characters
are significant, whereas under MASM 5.1, names are significant to 31
characters only. For more information on identifiers, see “Identifiers” in Chapter
1.

Multiple-Line Initializers
In MASM 6.1, a comma at the end of a line (except in the comment field)
implies that the line continues. For example, the following code is legal in
MASM 6.1:
longstring

BYTE

bitmasks

BYTE

"This string ",
"continues over two lines."
80h, 40h, 20h, 10h,
08h, 04h, 02h, 01h

For more information, see “Statements” in Chapter 1.

Comments in Extended Lines
MASM 5.1 allows a backslash ( \ ) as the line-continuation character if it is the
last nonspace character in the line. MASM 6.1 permits a comment to follow the
backslash.

Determining Size and Length of Data Labels
The LENGTHOF operator returns the number of data items allocated for a data
label. MASM 6.1 also provides the SIZEOF operator. When applied to a type,
SIZEOF returns the size attribute of the type expression. When applied to a data
label, SIZEOF returns the number of bytes used by the initializer in the label’s
definition. In this case, SIZEOF for a variable equals the number of bytes in the
type multiplied by LENGTHOF for the variable.
MASM 6.1 recognizes the LENGTH and SIZE operators for backward
compatibility. For a description of the behavior of SIZE under OPTION M510,
see “Length and Size of Labels with OPTION M510,” later in this appendix.
For obsolete behavior with the LENGTH operator, see also “LENGTH
Operator Applied to Record Types,” page 356.
For information on LENGTHOF and SIZEOF, see the following sections in
chapter 5: “Declaring and Referencing Arrays,” “Declaring and Initializing

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 346 of 6 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

347

Strings,” “Declaring Structure and Union Variables,” and “Defining Record
Variables.”

HIGHWORD and LOWWORD Operators
These operators return the high and low words for a given 32-bit operand. They
are similar to the HIGH and LOW operators of MASM 5.1 except that
HIGHWORD and LOWWORD can take only constants as operands, not
relocatables (labels).

PTR and CodeView
Under MASM 5.1, applying the PTR operator to a data initializer determines the
size of the data displayed by CodeView. You can still use PTR in this manner in
MASM 6.1, but it does not affect CodeView typing. Defining pointers with the
TYPEDEF directive allows CodeView to generate correct information. See
“Defining Pointer Types with TYPEDEF” in Chapter 3.

Procedures, Loops, and Jumps
With its significant improvements for procedure and jump handling, MASM 6.1
closely resembles high-level – language implementations of procedure calls.
MASM 6.1 generates the code to correctly handle argument passing, check type
compatibility between parameters and arguments, and process a variable number
of arguments. MASM 6.1 can also automatically recast jump instructions to
correct for insufficient jump distance.

Function Prototypes and Calls
The PROTO directive lets you prototype procedures in the same way as highlevel languages. PROTO enables type-checking and type conversion of
arguments when calling the procedure with INVOKE. For more information,
see “Declaring Procedure Prototypes” in Chapter 7.
The INVOKE directive sets up code to call a procedure and correctly pass
arguments according to the prototype. MASM 6.1 also provides the VARARG
keyword to pass a variable number of arguments to a procedure with INVOKE.
For more information about INVOKE and VARARG, see “Calling Procedures
with INVOKE” and “Declaring Parameters with the PROC Directive” in
Chapter 7.
The ADDR keyword is new since MASM 5.1. When used with INVOKE, it
provides the address of a variable, in the same way as the address-of operator
(&) in C. This lets you conveniently pass an argument by reference rather than
value. See “Calling Procedures with INVOKE” in Chapter 7.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 347 of 7 Printed: 10/02/00 04:18 PM

348

Programmer’s Guide

High-Level Flow-Control Constructions
MASM 6.1 contains several directives that generate code for loops and decisions
depending on the status of a conditional statement. The conditions are tested at
run time rather than at assembly time.
Directives new since MASM 5.1 include .IF, .ELSE, .ELSEIF, .REPEAT,
.UNTIL, .UNTILCXZ, .WHILE, and .ENDW. MASM 6.1 also provides the
associated .BREAK and .CONTINUE directives for loops and IF statements.
For more information, see “Loops” in Chapter 7 and “Decision Directives” on
page 171.

Automatic Optimization for Unconditional Jumps
MASM 6.1 automatically determines the smallest encoding for direct
unconditional jumps. See “Unconditional Jumps” in Chapter 7.

Automatic Lengthening for Conditional Jumps
If a conditional jump cannot reach its target destination, MASM automatically
recasts the code to use an unconditional jump to the target. See “Jump
Extending,” page 169.

User-Defined Stack Frame Setup and Cleanup
The prologue code generated immediately after a PROC statement sets up the
stack for parameters and local variables. The epilogue code handles stack
cleanup. MASM 6.1 allows user-defined prologues and epilogues, as described
in “Generating Prologue and Epilogue Code” in Chapter 7.

Simplifying Multiple-Module Projects
MASM 6.1 simplifies the sharing of code and data among modules and makes
the use of include files more efficient.

EXTERNDEF in Include Files
MASM 5.1 requires that you declare public and external all data and routines
used in more than one module. With MASM 6.1, a single EXTERNDEF
directive accomplishes the same task. EXTERNDEF lets you put global data
declarations within an include file, making the data visible to all source files that
include the file. For more information, see “Using EXTERNDEF” in Chapter 8.

Search Order for Include Files
MASM 6.1 searches for include files in the directory of the main source file
rather than in the current directory. Similarly, it searches for nested include files
in the directory of the include file. You can specify additional paths to search
with the /I command-line option. For more information on include files, see
“Organizing Modules” in Chapter 8.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 348 of 8 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

349

Enforcing Case Sensitivity
In MASM 5.1, sensitivity to case is influenced only by command-line options
such as /MX, not the language type given with the .MODEL directive. In
MASM 6.1, the language type takes precedence over the command-line options
in specifying case sensitivity.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 349 of 9 Printed: 10/02/00 04:18 PM

350

Programmer’s Guide

Alternate Names for Externals
The syntax for EXTERN allows you to specify an alternate symbol name,
which the linker can use to resolve an external reference to an unused symbol.
This prevents linkage with unneeded library code, as explained in “Using
EXTERN with Library Routines,” Chapter 8.

Expanded State Control
Several directives in MASM 6.1 enable or disable various aspects of the
assembler control. These include 80486 coprocessor instructions and use of
compatibility options.

The OPTION Directive
The new OPTION directive allows you to selectively define the assembler’s
behavior, including its compatibility with MASM 5.1. See “Using the OPTION
Directive” in Chapter 1 and “Compatibility between MASM 5.1 and 6.1,” later
in this appendix.

The .NO87 Directive
The .NO87 directive disables all coprocessor instructions. For more information,
see Help.

The .486 and .486P Directives
MASM 6.1 can assemble instructions specific to the 80486, enabled with the
.486 directive. The .486P directive enables 80486 instructions at the highest
privilege level (recommended for systems-level programs only). For more
information, see Help.

The PUSHCONTEXT and POPCONTEXT Directives
The directive PUSHCONTEXT saves the assembly environment, and
POPCONTEXT restores it. The environment includes the segment register
assumes, the radix, the listing and CREF flags, and the current processor and
coprocessor. Note that .NOCREF (the MASM 6.1 equivalent to .XCREF) still
determines whether information for a given symbol will be added to Browser
information and to the symbol table in the listing file. For more information on
listing files, see Appendix C or Help.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 350 of 10 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

351

New Processor Instructions
MASM 6.1 supports these instructions for the 80486 processor:
80486 Instruction

Description

BSWAP

Byte swap

CMPXCHG

Compare and exchange

INVD

Invalidate data cache

INVLPG

Invalidate Translation Lookaside Buffer entry

WBINVD

Write back and invalidate data cache

XADD

Exchange and add

For full descriptions of these instructions, see the Reference or Help.

Renamed Directives
Although MASM 6.1 still supports the old names in MASM 5.1, the following
directives have been renamed for language consistency:
MASM 6.1

MASM 5.1

.DOSSEG

DOSSEG

.LISTIF

.LFCOND

.LISTMACRO

.XALL

.LISTMACROALL

.LALL

.NOCREF

.XCREF

.NOLIST

.XLIST

.NOLISTIF

.SFCOND

.NOLISTMACRO

.SALL

ECHO

%OUT

EXTERN

EXTRN

FOR

IRP

FORC

IRPC

REPEAT

REPT

STRUCT

STRUC

SUBTITLE

SUBTTL

Specifying 16-Bit and 32-Bit Instructions
MASM 6.1 supports all instructions that work with the extended 32-bit registers
of the 80386/486. For certain instructions, you can override the default operand

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 351 of 11 Printed: 10/02/00 04:18 PM

352

Programmer’s Guide

size with the W (word) and the D (doubleword) suffixes. For details, see the
Reference or Help.

Macro Enhancements
There are significant enhancements to macro functions in MASM 6.1.
Directives provide for a variable number of arguments, loop constructions,
definitions of text equates, and macro functions.

Variable Arguments
MASM 5.1 ignores extra arguments passed to macros. In MASM 6.1, you can
pass a variable number of arguments to a macro by appending the VARARG
keyword to the last macro parameter in the macro definition. The macro can
then reference additional arguments relative to the last declared parameter. This
procedure is explained in “Returning Values with Macro Functions” in Chapter
9.

Required and Default Macro Arguments
With MASM 6.1, you can use REQ or the := operator to specify required or
default arguments. See “Specifying Required and Default Parameters” in
Chapter 9.

New Directives for Macro Loops
Within a macro definition, WHILE repeats assembly as long as a condition
remains true. Other macro loop directives, IRP, IRPC, and REPT, have been
renamed FOR, FORC, and REPEAT. For more information, see “Defining
Repeat Blocks with Loop Directives” in Chapter 9.

Text Macros
The EQU directive retains its old functionality, but MASM 6.1 also incorporates
a TEXTEQU directive for defining text macros. TEXTEQU allows greater
flexibility than EQU. For example, TEXTEQU can assign to a label the value
calculated by a macro function. For more information, see “Text Macros” in
Chapter 9.

The GOTO Directive for Macros
Within a macro definition, GOTO transfers assembly to a line labeled with a
leading colon(:). For more information on GOTO, see Help.

Macro Functions
At assembly time, macro functions can determine and return a text value using
EXITM. Predefined macro string functions concatenate strings, return the size
of a string, and return the position of a substring within a string. For information

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 352 of 12 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

353

on writing your own macro functions, see “Returning Values with Macro
Functions” in Chapter 9.

Predefined Macro Functions
MASM 6.1 provides the following predefined text macro functions:
Symbol

Value Returned

@CatStr

A concatenated string

@InStr

The position of one string within another

@SizeStr

The size of a string

@SubStr

A substring

For more information on predefined macros, see “String Directives and
Predefined Functions” in Chapter 9.

MASM 6.1 Programming Practices
MASM 6.1 provides many features that make it easier for you to write assembly
code. If you are familiar with MASM 5.1 programming, you may find it helpful
to adopt the following list of new programming practices for programming with
MASM 6.1. The list summarizes many of the changes covered in the following
section, “Compatibility Between MASM 5.1 and 6.1.”
u
u

u
u

Select identifier names that do not begin with the dot operator (.).
Use the dot operator (.) only to reference structure fields, and the plus
operator (+) when not referencing structures.
Different structures can have the same field names. However, the assembler
does not allow ambiguous references. You must include the structure type
when referring to field names common to two or more structures.
Separate macro arguments with commas, not spaces.
Avoid adding extra ampersands in macros. For a list of the new rules about
using ampersands in macros, see “Substitution Operator” in Chapter 9 and
“OPTION OLDMACROS,” page 372.
By default, code labels defined with a colon are local. Place two colons after
code labels if you want to reference the label outside the procedure.

Compatibility Between MASM 5.1 and 6.1
MASM 6.1 provides a “compatibility mode,” making it easy for you to transfer
existing MASM 5.1 code to the new version. You invoke the compatibility mode
through the OPTION M510 directive or the /Zm command-line switch. This

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 353 of 13 Printed: 10/02/00 04:18 PM

354

Programmer’s Guide

section explains the changes you may need to make to get your MASM 5.1 code
to run under MASM 6.1 in compatibility mode.

Rewriting Code for Compatibility
In some cases, MASM 6.1 with OPTION M510 does not support MASM 5.1
behavior. In several cases, this is because bugs in MASM 5.1 were corrected.
To update your code to MASM 6.1, use the instructions in this section. This
usually requires only minor changes.
Many of the topics listed here will not apply to your code. This section discusses
topics in order of likelihood, beginning with the most common. In addition, you
may have conflicts between identifier names and new reserved words. OPTION
NOKEYWORD resolves errors generated from the use of reserved words as
identifiers. See “OPTION NOKEYWORD,” page 376, for more information.

Bug Fixes Since MASM 5.1
This section lists the differences between MASM 5.1 and MASM 6.1 due to
bug corrections since MASM 5.1.

Invalid Use of LOCK, REPNE, and REPNZ
Except in compatibility mode, MASM 6.1 flags illegal uses of the instruction
prefixes LOCK, REPNE, and REPNZ. The error generated for invalid uses of
the LOCK, REPNE, and REPNZ prefixes is error A2068:
instruction prefix not allowed

Table A.1 summarizes the correct use of the instruction prefixes. It lists each
string instruction with the type of repeat prefix it uses, and indicates whether the
instruction works on a source, a destination, or both.
Table A.1 Requirements for String Instructions
Instruction

Repeat Prefix

Source/Destination

MOVS

REP

Both

DS:SI, ES:DI

SCAS

REPE/REPNE

Destination

ES:DI

CMPS

REPE/REPNE

Both

DS:SI, ES:DI

LODS

Source

DS:SI

STOS

REP

Destination

ES:DI

INS

REP

Destination

ES:DI

OUTS

REP

Source

DS:SI

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 354 of 14 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

355

No Closing Quotation Marks in Macro Arguments
In MASM 5.1, you can use both single and double quotation marks (' and ") to
begin strings in macro arguments. The assembler does not generate an error or
warning if the string does not end with quotation marks on a macro call. Instead,
MASM 5.1 considers the remainder of the line to be part of the macro argument
containing the opening quote, as if there were a closing quotation mark at the
end of the line.
By default, MASM 6.1 now generates error A2046:
missing single or double quotation mark in string

so all single and double quotation marks in macro arguments must be matched.
To correct such errors in MASM 6.1, either end the string with a closing
quotation mark as shown in the following example, or use the macro escape
character (!) to treat the quotation mark literally.
; MASM 5.1 code
MyMacro
"all this in one argument
; Default MASM 6.1 code
MyMacro
"all this in one argument"

Making a Scoped Label Public
MASM 5.1 considers code labels defined with a single colon inside a procedure
to be local to that procedure if the module contains a .MODEL directive with a
language type. Although the label is local, MASM 5.1 does not generate an error
if it is also declared PUBLIC. MASM 6.1 generates error A2203:
cannot declare scoped code label as PUBLIC

If you want to make a label PUBLIC, it must not be local. You can use the
double colon operator to define a non-scoped label, as shown in this example:
PUBLIC
publicLabel::

publicLabel
; Non-scoped label MASM 6.1

Byte Form of BT, BTS, BTC, and BTR Instructions
MASM 5.1 allows a byte argument for the 80386 bit-test instructions, but
encodes it as a word argument. The byte form is not supported by the
processor.
MASM 6.1 does not support this behavior and generates error A2024:
invalid operand size for instruction

Rewrite your code to use a word-sized argument.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 355 of 15 Printed: 10/02/00 04:18 PM

356

Programmer’s Guide

Default Values for Record Fields
In MASM 5.1, default values for record fields can range down to –2n , where n
is the number of bits in the field. This results in the loss of the sign bit.
MASM 6.1 allows a range of –2n–1 to 2n–1 for default values. Illegal initializers
generate error A2071:
initializer too large for specified size

Design Change Issues
MASM 6.1 includes design changes that make the language more consistent.
These changes are not affected by the OPTION directive, discussed later in
this appendix. Therefore, the changes require revisions in your code. In most
cases, the necessary revisions are minor and the circumstances requiring
changes are rare.

Operands of Different Size
MASM 5.1 does not require operands to agree in size, as the following code
illustrates:
.DATA?
var1
var2
.CODE
.
.
.
mov

DB
DB

var1, ax

?
?

; Copy AX to word at var1

The operands for the MOV instruction do not match in size, yet the instruction
assembles correctly. It places the contents of AL into var1 and AH into var2,
moving a word of data in one step. If the code defined var1 as a word value,
the instruction
mov

var1, al

would also assemble correctly, copying AL into the low byte of var1 while
leaving the high byte unaffected. Except at warning level 0, MASM 5.1 issues a
warning to inform you of the size mismatch, but both scenarios are legal.
MASM 6.1 does not accept instructions with operands that do not agree in size.
You must specifically “coerce” the size of the memory operand, like this:

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 356 of 16 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1
mov

357

BYTE PTR var1, al

Conflicting Structure Declarations
MASM 5.1 allows you to declare two or more structures with the same name.
Each declaration replaces the previous declaration. However, the field names
from previous declarations still remain in the assembler’s list of declared values.
MASM 6.1 does not allow conflicting declarations of a structure. It generates
errors A2160 through A2165 for each conflicting declaration. The errors note a
specific conflict, such as conflicting number of fields, conflicting names of fields,
or conflicting initializers.

Forward References to Text Macros Outside of Expressions
MASM 5.1 allows forward references to text macros in specialized cases.
MASM 6.1 with OPTION M510 also permits forward references, but only
when the text macro is referenced in an expression. To revise your code, place
all macro definitions at the beginning of the file.

HIGH and LOW Applied to Relocatable Operands
In some cases, MASM 5.1 accepts HIGH and LOW applied to relocatable
memory expressions. For example, MASM 5.1 allows this code sequence:
; MASM 5.1 code
EXTRN
var1:WORD
var2
DW
0
mov
al, LOW var1
mov
ah, HIGH var1

; These two instructions yield the
; same as mov ax, OFFSET var1

However, the instruction
mov ax, LOW var2

is not legal. MASM 6.1 generates error A2105:
HIGH and LOW require immediate operands

The OFFSET operator is required on these operands in MASM 6.1, as shown in
the following. Rewrite your code if necessary.
; MASM 6.1 code
mov
al, LOW OFFSET var1
mov
ah, HIGH OFFSET var2

OFFSET Applied to Group Names and Indirect Memory Operands
In MASM 6.1, you cannot apply OFFSET to a group name, indirect argument,
or procedure argument. Doing so generates error A2098:

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 357 of 17 Printed: 10/02/00 04:18 PM

358

Programmer’s Guide
invalid operand for OFFSET

LENGTH Operator Applied to Record Types
In MASM 5.1, the LENGTH operator, when applied to a record type, returns
the total number of bits in a record definition.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 358 of 18 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

359

In MASM 6.1, the statement LENGTH recordName returns error A2143:
expected data label

Rewrite your code if necessary. The new SIZEOF operator returns information
about records in MASM 6.1. For more information, see “Defining Record
Variables” in Chapter 5.

Signed Comparison of Hexadecimal Values Using GT, GE, LE, or LT
The rules for two’s-complement comparisons have changed. In MASM 5.1, the
expression
0FFFFh GT -1

is false because the two’s-complement values are equal. However, because
hexadecimal numbers are now treated as unsigned, the expression is true in
MASM 6.1. To update, rewrite the affected code.

RET Used with a Constant in Procedures with Epilogues
By default in MASM 6.1, the RET instruction followed by a constant
suppresses automatic generation of epilogue code. MASM 5.1 ignores the
operand and generates the epilogue. Remove the argument if necessary. See
“Generating Prologue and Epilogue Code” in Chapter 7.

Code Labels at Top of Procedures with Prologues
By default in MASM 5.1, a code label defined on the same line as the first
procedure instruction refers to the first byte of the prologue.
In MASM 6.1, a code label defined at the beginning of a procedure refers to the
first byte of the procedure after the prologue. If you need to label the start of the
prologue code, place the label before the PROC statement. For more
information, see “Generating Prologue and Epilogue Code” in Chapter 7.

Use of % as an Identifier Character
MASM 5.1 allows % as an identifier character. This behavior leads to
ambiguities when % is used as the expansion operator in macros. Since % is not
allowed as a character in MASM 6.1 identifiers, you must change the names of
any identifiers containing the % character. For a list of legal identifier characters,
see “Identifiers” in Chapter 1.

ASSUME CS Set to Wrong Value
With MASM 6.1 you do not need to use the ASSUME statement for the CS
register. Instead, MASM 6.1 generates an automatic ASSUME statement for
the code segment register to the current segment or group, as explained in
“Setting the ASSUME Directive for Segment Registers” in Chapter 2.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 359 of 19 Printed: 10/02/00 04:18 PM

360

Programmer’s Guide

Additionally, MASM 6.1 does not allow explicit ASSUME statements for CS
that contradict the automatically set ASSUME statement.
MASM 5.1 allows CS to be assumed to the current segment, even if that
segment is a member of a group. With MASM 6.1, this results in warning
A4004:
cannot ASSUME CS

To avoid this warning with MASM 6.1, delete the ASSUME statement for CS.

Code Requiring Two-Pass Assembly
Unlike version 5.1, MASM 6.1 does most of its work on its first pass, then
performs as many subsequent passes as necessary. In contrast, MASM 5.1
always assembles in two source passes. As a result, you may need to revise or
delete some pass-dependent constructs under MASM 6.1.

Two-Pass Directives
To assure compatibility, MASM 6.1 supports 5.1 directives referring to two
passes. These include .ERR1, .ERR2, IF1, IF2, ELSEIF1, and ELSEIF2. For
second-pass constructs, you must specify OPTION SETIF2, as discussed in
“OPTION SETIF2,” page 377. Without OPTION SETIF2, the IF2 and .ERR2
directives cause error A2061:
[[ELSE]]IF2/.ERR2 not allowed : single-pass assembler

MASM 6.1 handles first-pass constructs differently. It treats the .ERR1
directive as .ERR, and the IF1 directive as IF.
The following examples show you how you can rewrite typical pass-sensitive
code for MASM 6.1:
u

Declare var external only if not defined in current module:
; MASM 5.1:
IF2
IFNDEF var
EXTRN var:far
ENDIF
ENDIF
; MASM 6.1:
EXTERNDEF

var:far

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 360 of 20 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1
u

361

Include a file of definitions only once to speed assembly:
; MASM 5.1:
IF1
INCLUDE file1.inc
ENDIF
; MASM 6.1:
INCLUDE FILE1.INC

Generate a %OUT or .ERR message only once:
; MASM 5.1:
IF2
%OUT This is my message
ENDIF
IF2
.ERRNZ A NE B
ENDIF
; MASM 6.1:
ECHO This is my message
.ERRNZ A NE B

Generate an error if a symbol is not defined but may be forward referenced:
; MASM 5.1:
IF2
.ERRNDEF
ENDIF
; MASM 6.1:
.ERRNDEF

var

For information on conditional directives, see “Conditional Directives,” Chapter
1.

IFDEF and IFNDEF with Forward-Referenced Identifiers
If you use a symbol name that has not yet been defined in an IFDEF or
IFNDEF expression, MASM 6.1 returns FALSE for the IFDEF expression and
TRUE for the IFNDEF expression. When OPTION M510 is enabled, the
assembler generates warning A6005:
expression condition may be pass-dependent

To resolve the warning, place the symbol definition before the conditional test.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 361 of 21 Printed: 10/02/00 04:18 PM

362

Programmer’s Guide

Address Spans as Constants
The value of offsets calculated on the first assembly pass may not be the same
as those calculated on later passes. Therefore, you should avoid comparisons
with an address span, as in the following examples:
IF (OFFSET var1 - OFFSET var2) EQ 10
WHILE dx LT (OFFSET var1 - OFFSET var2)
REPEAT OFFSET var1 - OFFSET var2

However, the DUP operator allows such an expression as its count value. The
assembler evaluates the DUP count on every pass, so even expressions
involving forward references assemble correctly.
You can also use expressions containing span distances with the .ERR
directives, since the assembler evaluates these directives after calculating all
offsets:
.ERRE OFFSET var1 - OFFSET var2 - 10,

.TYPE with Forward References
MASM 5.1 evaluates .TYPE on both assembly passes. This means it yields zero
on the first pass and nonzero on the second pass, if applied to an expression that
forward-references a symbol.
MASM 6.1 evaluates .TYPE only on the first assembly pass. As a result, if the
operand references a symbol that has not yet been defined, .TYPE yields a
value of zero. This means that .TYPE, if used in a conditional-assembly
construction, may yield different results in MASM 6.1 than in MASM 5.1.

Obsolete Features No Longer Supported
The following two features are no longer supported by MASM 6.1. Because
both are obscure features provided by early versions of the assembler, they
probably do not affect your MASM 5.1 code.

The ESC Instruction
MASM 6.1 no longer supports the ESC instruction, which was used to send
hand-coded commands to the coprocessor. Because MASM 6.1 recognizes and
assembles the full set of coprocessor mnemonics, the ESC instruction is not
necessary. Using the ESC instruction generates error A2205:
ESC instruction is obsolete: ignored

To update MASM 5.1 code, use the coprocessor instructions instead of ESC.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 362 of 22 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

363

The MSFLOAT Binary Format
MASM 6.1 does not support the .MSFLOAT directive, which provided the
Microsoft Binary Format (MSB) for floating-point numbers in variable
initializers. Using the .MSFLOAT directive generates error A2204:
.MSFLOAT directive is obsolete: ignored

Use IEEE format or, if MSB format is necessary, initialize variables with
hexadecimal values. See “Storing Numbers in Floating-Point Format” in Chapter
6.

Using the OPTION Directive
The OPTION directive lets you control compatibility with MASM 5.1 code.
This section explains the differences in MASM 5.1 and MASM 6.1 behavior
that the OPTION directive can influence.
The OPTION M510 directive (or /Zm command-line option) initiates all aspects
of 5.1 compatibility mode. You can select from among specific characteristics of
MASM 5.1 behavior with the OPTION arguments discussed in following
sections. Each section also explains how to revise your code if you want to
remove OPTION directives from your MASM 5.1 code.
Note If your code includes both .MODEL and OPTION M510, the OPTION
M510 statement must appear first. Wherever this appendix suggests using
OPTION M510 in your code, you can set the /Zm command-line option
instead.

OPTION M510
This section discusses the M510 argument to the OPTION directive, which
selects the MASM 5.1 compatibility mode. In this mode, MASM 6.1
implements MASM 5.1 behavior relating to macros, offsets, scope of code
labels, structures, identifier names, identifier case, and other behaviors.
The OPTION M510 directive automatically sets the following:
OPTION
OPTION
OPTION
OPTION

OLDSTRUCTS
OLDMACROS
DOTNAME
SETIF2:TRUE

;
;
;
;

MASM 5.1 structures
MASM 5.1 macros
Identifiers may begin with a dot (.)
Two-pass code activates on every pass

If you do not have a .386, 386P .486, or 486P directive in your module, then
OPTION M510 adds:

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 363 of 23 Printed: 10/02/00 04:18 PM

364

Programmer’s Guide
OPTION EXPR16

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 364 of 24 Printed: 10/02/00 04:18 PM

; 16-bit expression precision
;
See "OPTION EXPR16," following

Appendix A Differences Between MASM 6.1 and 5.1

365

If you do not have a .MODEL directive in your module, OPTION M510 adds:
OPTION OFFSET:SEGMENT

; OFFSET operator defaults to
;
segment-relative
;
See "OPTION OFFSET," following

If you do not have a .MODEL directive with a language specifier in your
module, OPTION M510 also adds:
OPTION NOSCOPED

OPTION PROC:PRIVATE

; Code labels are not local inside
;
procedures
;
See "OPTION NOSCOPED," following
; Labels defined with PROC are not
;
public by default
;
See "OPTION PROC," following

If you want to remove OPTION M510 from your code (or /Zm from the
command line), add the OPTION directive arguments to your module according
to the conditions stated earlier.
There may be compatibility issues affecting your code that are supported under
OPTION M510, but are not covered by the other OPTION directive
arguments. Once you have modified your source code so it no longer requires
behavior supported by OPTION M510, you can replace OPTION M510 with
other OPTION directive arguments. These compatibility issues are discussed in
following sections.
Once you have replaced OPTION M510 with other forms of the OPTION
directive and your code works correctly, try removing the OPTION directives,
one at a time. Make appropriate source modifications as necessary, until your
code uses only MASM 6.1 defaults.

Reserved Keywords Dependent on CPU Mode with OPTION M510
With OPTION M510, keywords and instructions not available in the current
CPU mode (such as ENTER under .8086) are not treated as keywords. This
also means the USE32, FLAT, FAR32, and NEAR32 segment types and the
80386/486 registers are not keywords with a processor selection less than .386.
If you remove OPTION M510, any reserved word used as an identifier
generates a syntax error. You can either rename the identifiers or use OPTION
NOKEYWORD. For more information on OPTION NOKEYWORD, see
“OPTION NOKEYWORD,” later in this appendix.

Invalid Use of Instruction Prefixes with OPTION M510
Code without OPTION M510 generates errors for all invalid uses of the
instruction prefixes. OPTION M510 suppresses some of these errors to match
MASM 5.1 behavior. MASM 5.1 does not check for illegal usage of the
instruction prefixes LOCK, REP, REPE, REPZ, REPNE, and REPNZ.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 365 of 25 Printed: 10/02/00 04:18 PM

366

Programmer’s Guide

Illegal usage of these prefixes results in error A2068:
instruction prefix not allowed

For more information on these instruction prefixes, see “Overview of String
Instructions” in Chapter 5. See also “Bug Fixes from MASM 5.1,” earlier in this
appendix.

Size of Constant Operands with OPTION M510
In MASM 5.1, a large constant value that can fit only in the processor’s default
word (4 bytes for .386 and .486, 2 bytes otherwise) is assigned a size attribute
of the default word size. The value of the constant affects the number of bytes
changed by the instruction. For example,
; Legal only with OPTION M510
mov
[bx], 0100h

is legal in OPTION M510 mode. Since 0100h cannot fit in a byte, the
assembler interprets the value as a word.
Without OPTION M510, the assembler never assigns a size automatically. You
must state it explicitly with the PTR operator, as shown in the following
example:
; Without OPTION M510
mov
[bx], WORD PTR 0100h

Code Labels when Defining Data with OPTION M510
MASM 5.1 allows a code label definition in a data definition statement if that
statement does not also define a data label. MASM 6.1 also allows such
definitions if OPTION M510 is enabled; otherwise it is illegal.
; Legal only with OPTION M510
MyCodeLabel:
DW
0

SEG Operator with OPTION M510
In MASM 5.1, the SEG operator returns a label’s segment address unless the
frame is explicitly specified, in which case it returns the segment address of the
frame. A statement such as SEG DGROUP:var always returns DGROUP,
whereas SEG var always returns the segment address of var. OPTION M510
forces this same behavior in MASM 6.1.
If you do not use OPTION M510, the behavior of the SEG operator is
determined by the OPTION OFFSET directive, as described in “OPTION
OFFSET,” later in this appendix.
In MASM 6.1, the value returned by the SEG operator applied to a nonexternal
variable depends on compatibility mode:

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 366 of 26 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

367

Without OPTION M510, SEG returns the address of the frame (the
segment, group, or the value assumed to the segment register) if one has
been explicitly set.

With OPTION M510, SEG returns the group if one has been specified. In
the absence of a defined group, SEG returns the segment where the variable
is defined.

Expression Evaluation with OPTION M510
By default, MASM 6.1 changes the way expressions are evaluated. In MASM
5.1,
var-2[bx]

is parsed as
(var-2)[bx]

Without OPTION M510, you must rewrite the statement, since the assembler
parses it as
var-(2[bx])

which generates an error.

Length and Size of Labels with OPTION M510
With OPTION M510, you can apply the LENGTH and SIZE operators to any
label. For a code label, SIZE returns a value of 0FFFFh for NEAR and 0FFFEh
for FAR. LENGTH always returns a value of 1. For strings, SIZE and
LENGTH both return 1.
Without OPTION M510, SIZE returns values of 0FF01h, 0FF02h, 0FF04h,
0FF05h, and 0FF06h for SHORT, NEAR16, NEAR32, FAR16, and FAR32
labels, respectively. LENGTH returns 1 except when used with DUP, in which
case it returns the outermost count. For arrays initialized with DUP, SIZE
returns the length multiplied by the size of the type.
The LENGTHOF and SIZEOF operators in MASM 6.1 handle arrays much
more consistently. These operators return the number of data items and the
number of bytes in an initializer. For a description of SIZEOF and
LENGTHOF, see the following sections in Chapter 5: “Declaring and
Referencing Arrays,” “Declaring and Initializing Strings,” “Defining Structure
and Union Variables,” and “Defining Record Variables.”

Comparing Types Using EQ and NE with OPTION M510
With OPTION M510, the assembler converts types to a constant value before
comparisons with EQ and NE. Code types are converted to values of 0FFFFh

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 367 of 27 Printed: 10/02/00 04:18 PM

368

Programmer’s Guide

(near) or 0FFFEh (far). If OPTION M510 is not enabled, the assembler
converts types to constants only when comparing them with constants. Thus,
MASM 6.1 recognizes only equivalent qualified types as equal expressions.
For existing MASM 5.1 code, these distinctions affect only the use of the TYPE
operator in conjunction with EQ and NE. The following example illustrates how
the assembler compares types with and without compatibility mode:
MYSTRUCT
f1
f2
MYSTRUCT

STRUC
DB
DB
ENDS

0
0

; With OPTION M510
val
val
val
val

=
=
=
=

(TYPE MYSTRUCT) EQ WORD
2 EQ WORD
WORD EQ WORD
SWORD EQ WORD

;
;
;
;

True:
True:
True:
True:

;
;
;
;

False:
True:
True:
False:

2
2
2
2

EQ
EQ
EQ
EQ

2
2
2
2

; Without OPTION M510
val
val
val
val

=
=
=
=

(TYPE MYSTRUCT) EQ WORD
2 EQ WORD
WORD EQ WORD
SWORD EQ WORD

MyStruct NE WORD
2 EQ 2
WORD EQ WORD
SWORD NE WORD

Use of Constant and PTR as a Type with OPTION M510
You can use a constant as the left operand to PTR in compatibility mode.
Otherwise, you must use a type expression. With OPTION M510, a constant
must have a value of 1 (BYTE), 2 (WORD), 4 (DWORD), 6 (FWORD), 8
(QWORD) or 10 (TBYTE). The assembler treats the constant as the

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 368 of 28 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

369

parenthesized type. Note that the TYPE operator yields a type expression, but
the SIZE operator yields a constant.
; With OPTION M510
MyData

mov
mov
mov
mov

WORD PTR [bx], 10
(TYPE MyData) PTR [bx], 10
(SIZE MyData) PTR [bx], 10
2 ptr [bx], 10

;
;
;
;

Legal
Legal
Legal
Legal

;
;
;
;

Legal
Legal
Illegal
Illegal

; Without OPTION M510
MyData

WORD

;
;

mov
mov
mov
mov

WORD PTR [bx], 10
(TYPE MyData) PTR [bx], 10
(SIZE MyData) PTR [bx], 10
2 PTR [bx], 10

Structure Type Cast on Expressions with OPTION M510
In compatibility mode, use the PTR operator to type-cast a constant to a
structure type. This is most often done in data initializers to affect the CodeView
information of the data label. Without OPTION M510, the assembler generates
an error.
MYSTRC
f1
MYSTRC

STRUC
DB
ENDS

MyPtr

MYSTRC PTR 0

; Illegal without OPTION M510

In MASM 6.1, the initializer type does not influence CodeView’s type
information.

Hidden Coercion of OFFSET Expression Size with OPTION M510
When programming for the 80386 or 80486, the size of an OFFSET expression
can be 2 bytes for a symbol in a USE16 segment, or 4 bytes for a symbol in a
USE32 or FLAT segment. With OPTION M510, you can use a 32-bit
OFFSET expression in a 16-bit context. Without OPTION M510, you must use
the LOWWORD operator to convert the offset size.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 369 of 29 Printed: 10/02/00 04:18 PM

370

Programmer’s Guide
.386
; With OPTION M510
seg32
SEGMENT USE32
MyLabel WORD
0
seg32
ENDS
seg16

seg16

SEGMENT
mov
mov
mov
ENDS

USE16 'code'
ax, OFFSET MyLabel
ax, LOWWORD OFFSET MyLabel
eax, OFFSET MyLabel

;
;
;
;

With OPTION M510:
Legal
Legal
Legal

; Without OPTION M510
seg32
SEGMENT USE32
MyLabel WORD
0
seg32
ENDS
seg16
;

seg16

SEGMENT
mov
mov
mov
ENDS

USE16 'code'
;
ax, OFFSET MyLabel
;
ax, LOWWORD offset MyLabel ;
eax, OFFSET MyLabel
;

Without OPTION M510:
Illegal
Legal
Legal

Specifying Radixes with OPTION M510
If the current radix in your code is greater than 10 decimal, MASM 6.1 allows
the radix specifiers B (binary) and D (decimal) only in compatibility mode. You
must change B to Y for binary, and D to T for decimal, since both B and D are
legitimate hexadecimal values, making numbers such as 12D ambiguous. If you
want to keep B and D as radix specifiers when the current radix is greater than
10, you must specify OPTION M510. For more information about radixes, see
“Integer Constants and Constant Expressions” in Chapter 1.

Naming Conventions with OPTION M510
By default, MASM 5.1 does not write the names of public variables in
uppercase to the object file, even when a language type of PASCAL,
FORTRAN, or BASIC is specified.
Unless you use OPTION M510, these language types in MASM 6.1 write
identifier names in uppercase, even with the /Cp or /Cx command-line options.
When you link with /NOI, case must match in the object files to resolve
externals.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 370 of 30 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

371

Length Significance of Symbol Names with OPTION M510
With MASM 5.1, only the first 31 characters of a symbol name are considered
significant, and only the first 31 characters of a public or external symbol name
are placed in the object file.
Without OPTION M510, the entire name is considered significant. The
maximum number of characters placed in the object file is controlled with the
/Hnumber command-line option, with a default of 247 (the maximum length of
an identifier in MASM 6.1).

String Defaults in Structure Variables with OPTION M510
In compatibility mode, a constant initializer can override a structure field
initialized with a string value. Without OPTION M510, only another string or a
list can override a string initializer. To update your code, surround the constant
override value with angle brackets or curly braces to indicate a list with one
element.
MTSTRUCT
MyString
MTSTRUCT

STRUCT
BYTE
ENDS

"This is a string"

; With OPTION M510
MyInst

MTSTRUCT

<0>

; Without OPTION M510, either of these statements is correct
MyInst
MyInst

MTSTRUCT
MTSTRUCT

<<0>>
{<0>}

Effects of the ? Initializer in Data Definitions with OPTION M510
As described in “Declaring and Initializing Strings” in Chapter 5, the assembler
treats the ? initializer as either zero or as an unspecified value. In compatibility
mode, however, the assembler always treats the ? initializer as zero unless it is
used with the DUP operator. In this case, the assembler allocates space, but
does not initialize it with any value.

Current Address Operator with OPTION M510
In compatibility mode, the current address operator ($) applied to a structure
returns the offset of the first byte of the structure. When OPTION M510 is not
enabled, $ returns the offset of the current field in the structure.

Segment Association for FAR Externals with OPTION M510
In MASM 5.1, you must place an EXTRN directive for a variable in the same
segment that holds the variable. For far data, this often entails opening and
closing a segment just to place the EXTRN statement.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 371 of 31 Printed: 10/02/00 04:18 PM

372

Programmer’s Guide

MASM 6.1 offers much greater flexibility in where EXTERN and
EXTERNDEF statements can appear, as described in “Positioning External
Declarations” in Chapter 8. However, in compatibility mode, MASM 6.1
emulates the behavior of MASM 5.1.

Defining Aliases Using EQU with OPTION M510
In MASM 5.1, you can equate one symbol with another. These equates are
called “aliases.”
Unless you specify OPTION M510, MASM 6.1 does not allow aliases defined
with EQU. An immediate expression or text must appear as the right operand of
an EQU directive. Change aliases to use the TEXTEQU directive, described in
“Text Macros” in Chapter 9. This change may cause an expression to evaluate
differently.
The following examples illustrate the differences between MASM 5.1 code,
MASM 6.1 code with OPTION M510, and MASM 6.1 code without OPTION
M510:
; MASM 5.1 code
var1
EQU
3
var2
EQU
var1

; var2 taken as an alias
; var2 references var1 anywhere var2 is
;
used as a symbol

; MASM 6.1 with OPTION M510
var1
EQU
3
var2
EQU
var1
; var2 taken as a var2 EQU
; var2 substituted for var1 whenever
;
text macros substituted
; MASM 6.1 without OPTION M510
var1
EQU
3
var2
EQU
var1
; Treated as var2 EQU 3

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 372 of 32 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

373

Difference in Text Macro Expansions with OPTION M510
MASM 6.1 recursively expands text macros used as values, whereas MASM 5.1
simply replaces the text macro with its value. The following example illustrates
the difference:
; With OPTION M510
tm1
tm2

EQU
EQU

tm3

CATSTR

tm1

; ==

; Without OPTION M510
tm3

CATSTR

tm1

; ==

Conditional Directives and Missing Operands with OPTION M510
MASM 5.1 considers a missing argument to be a zero. MASM 6.1 requires an
argument unless OPTION M510 is enabled.

OPTION OLDSTRUCTS
This section describes changes in MASM 6.1 that apply to structures. With
OPTION OLDSTRUCTS or OPTION M510:
u
u

You can use plus operator (+) in structure field references.
Labels and structure field names cannot have the same name with OPTION
OLDSTRUCTS.

Plus Operator Not Allowed with Structures
By default, each reference to structure member names must use the dot
operator (.) to separate the structure variable name from the field name. You
cannot use the dot operator as the plus operator (+) or vice versa.
To convert your code so that it does not need OPTION OLDSTRUCTS:
u
u

Qualify all structure field references.
Change all uses of the dot operator ( . ) that occur outside of structure
references to use the plus operator ( + ).

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 373 of 33 Printed: 10/02/00 04:18 PM

374

Programmer’s Guide

If you remove OPTION OLDSTRUCTS from your code, the assembler
generates errors for all lines requiring change. Using the dot operator in any
context other than for a structure field results in error A2166:
structure field expected

Unqualified structure references result in error A2006:
undefined symbol : identifier

The following example illustrates how to change MASM 5.1 code from the old
structure references to the new type in MASM 6.1:
; OPTION OLDSTRUCTS (simulates MASM 5.1)
structname
STRUC
a
BYTE ?
b
WORD ?
structname
ENDS
structinstance
mov
mov
mov

structname <>
ax, [bx].b
al, structinstance.a
ax, [bx].4

; This code assembles
;
correctly only with
;
OPTION OLDSTRUCTS
;
or OPTION M510

; OPTION NOOLDSTRUCTS (the MASM 6.1 default)
structname
STRUCT
a
BYTE ?
b
WORD ?
structname
ENDS
structinstance
mov
mov
mov

structname <>
ax, [bx].structname.b
al, structinstance.a
ax, [bx]+4

; Add qualifying type
; No change needed
; Change dot to plus

; Alternative methods in MASM 6.1
; Either this:
ASSUME bx:PTR structname
mov
ax, [bx]
; or this:
mov
ax, (structname PTR[bx]).b

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 374 of 34 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

375

Duplicate Structure Field Names
With the default, OPTION NOOLDSTRUCTS, label and structure field names
may have the same name. With OPTION OLDSTRUCTS (the MASM 5.1
default), labels and structure fields cannot have the same name. For more
information, see “Structures and Unions” in Chapter 5.

OPTION OLDMACROS
This section describes how MASM 5.1 and 6.1 differ in their handling of
macros. Without OPTION OLDMACROS or OPTION M510, MASM 6.1
changes the behavior of macros in several ways. If you want the MASM 5.1
macro behavior, add OPTION OLDMACROS or OPTION M510 to your
MASM 5.1 code.

Separating Macro Arguments with Commas
MASM 5.1 allows white spaces or commas to separate arguments to macros.
MASM 6.1 with OPTION NOOLDMACROS (the default) requires commas
between arguments. For example, in the macro call
MyMacro

var1 var2 var3, var4

OPTION OLDMACROS causes the assembler to treat all four items as
separate arguments. With OPTION NOOLDMACROS, the assembler treats
var1 var2 var3

as one argument, since the items are not separated with commas. To convert
your macro code, replace spaces between macro arguments with a single
comma.

New Behavior with Ampersands in Macros
The default OPTION NOOLDMACROS causes the assembler to interpret
ampersands (&) within a macro differently than does MASM 5.1. MASM 5.1
requires one ampersand for each level of macro nesting. OPTION
OLDMACROS emulates this behavior.
Without OPTION OLDMACROS, MASM 6.1 removes ampersands only once
no matter how deeply nested the macro. To update your MASM 5.1 macros,
follow this simple rule: replace every sequence of ampersands with a single
ampersand. The only exception is when macro parameters immediately precede
and follow the ampersand, and both require substitution. In this case, use two
ampersands. For a description of the new rules, see “Substitution Operator” in
Chapter 9.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 375 of 35 Printed: 10/02/00 04:18 PM

376

Programmer’s Guide

This example shows how to update a MASM 5.1 macro:
; OPTION OLDMACROS (emulates MASM 5.1 behavior)
createNames
macro
irp
tail,
irp
num,
; Define more
arg&&tail&&&num&&&?
ENDM
ENDM
ENDM

arg

<1, 2>
names of the form: abcNext1?
label
BYTE

; OPTION NOOLDMACROS (the MASM 6.1 default)
createNames
macro
arg
for
tail, ; FOR is the MASM 6.1
for
num, <1, 2>
;
synonym for irp
; Define more names of the form: abcNext1?
arg&&tail&&num&?
label
BYTE
ENDM
ENDM
ENDM

OPTION DOTNAME
MASM 5.1 allows names of identifiers to begin with a period. The MASM 6.1
default is OPTION NODOTNAME. Adding OPTION DOTNAME to your
code enables the MASM 5.1 behavior.
If you don’t want to use this directive in your source code, rename the
identifiers whose names begin with a period.

OPTION EXPR16
MASM 5.1 treats expressions as 16-bit words if you do not specify .386 or
.386P directives. MASM 6.1 by default treats expressions as 32-bit words,
regardless of the CPU type. You can force MASM 6.1 to use the smaller
expression size with the OPTION EXPR16 statement.
Unless your MASM 5.1 code specifies .386 or .386P, OPTION M510 also sets
16-bit expression size. You can selectively disable this by following OPTION
M510 with the OPTION EXPR32 directive, which sets the size back to 32 bits.
You cannot have both OPTION EXPR32 and OPTION EXPR16 in your
program.
It may not be easy to determine the effect of changing from 16-bit internal
expression size to 32-bit size. In most cases, the 32-bit word size does not affect
the MASM 5.1 code. However, problems may arise because of differences in

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 376 of 36 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

377

intermediate values during evaluation of expressions. You can compare the files
for differences by generating listing files with the /Fl and /Sa command-line
options with and without OPTION EXPR16.

OPTION OFFSET
The information in this section is relevant only if your MASM 5.1 code does not
use the .MODEL directive. With no .MODEL, MASM 5.1 computes offsets
from the start of the segment, whereas MASM 6.1 computes offsets from the
start of the group. (With .MODEL, MASM 5.1 also computes offsets from the
start of the group.)
To force MASM 6.1 to emulate 5.1 behavior, specify either
OFFSET:SEGMENT or OPTION M510. Both directives cause the assembler
to compute offsets relative to the segment if you do not include .MODEL.
To selectively enable MASM 6.1 behavior, place the directive OPTION
OFFSET:GROUP after OPTION M510. In this case, you should ensure each
OFFSET statement has a segment override where appropriate. The following
example shows how OPTION OFFSET:SEGMENT affects code written for
MASM 5.1:
OPTION OFFSET:SEGMENT
MyGroup GROUP
MySeg
MySeg
SEGMENT 'data'
MyLabel LABEL
BYTE
DW
OFFSET MyLabel
DW
OFFSET MyGroup:MyLabel
DW
OFFSET MySeg:MyLabel
MySeg
ENDS

; Relative to MySeg
; Relative to MyGroup
; Relative to MySeg

In the preceding example, the first OFFSET statement computes the offset of
MyLabel relative to MySeg. Without OFFSET:SEGMENT, MASM 6.1 returns
the offset relative to MyGroup. To maintain the correct behavior with
OFFSET:GROUP, specify a segment override, as shown in the following. The
other two OFFSET statements already include overrides, and so do not require
modification.
OPTION OFFSET:GROUP
MyGroup GROUP
MySeg
MySeg
SEGMENT 'data'
MyLabel LABEL
BYTE
DW
OFFSET MySeg:MyLabel
DW
OFFSET MyGroup:MyLabel
DW
OFFSET MySeg:MyLabel
MySeg
ENDS

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 377 of 37 Printed: 10/02/00 04:18 PM

; Relative to MySeg
; Relative to MyGroup
; Relative to MySeg

378

Programmer’s Guide

When not in compatibility mode, the OPTION OFFSET directive determines
whether the SEG operator returns a value relative to the group or segment. With
OPTION M510, SEG is always segment-relative by default, regardless of the
current value of OPTION OFFSET.

OPTION NOSCOPED
The information in this section applies only if the .MODEL directive in your
MASM 5.1 code does not specify a language type. Without a language type,
MASM 5.1 assumes code labels in procedures have no “scope” — that is, the
labels are not local to the procedure. When not in compatibility mode, MASM
6.1 always gives scope to code labels, even without a language type.
To force MASM 5.1 behavior, specify either OPTION M510 or OPTION
NOSCOPED in your code. To selectively enable MASM 6.1 behavior, place
the directive OPTION SCOPED after OPTION M510.
To determine which labels require change, assemble the module without the
OPTION NOSCOPED directive. For each reference to a label that is not local,
the assembler generates error A2006:
undefined symbol : identifier

OPTION PROC
The information in this section applies only if the .MODEL directive in your
MASM 5.1 code does not specify a language type. Without a language type,
MASM 5.1 makes procedures private to the module. By default, MASM 6.1
makes procedures public. You can explicitly change the default visibility to
private with either OPTION M510, OPTION PROC:PRIVATE, or OPTION
PROC:EXPORT.
To selectively enable MASM 6.1 behavior, place the directive OPTION
PROC:PUBLIC after OPTION M510. You can override the default by adding
the PUBLIC or PRIVATE keyword to selected procedures. The following
example shows how to change MASM 5.1 code to keep a procedure private:
; MASM 5.1 (OPTION PROC:PRIVATE)
MyProc PROC NEAR
; MASM 6.1 (OPTION PROC:PUBLIC)
MyProc PROC NEAR PRIVATE

This is necessary only to avoid naming conflicts between public names in
multiple modules or libraries. The symbol table in a listing file shows the
visibility (public, private, or export) of each procedure.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 378 of 38 Printed: 10/02/00 04:18 PM

Appendix A Differences Between MASM 6.1 and 5.1

379

OPTION NOKEYWORD
MASM 6.1 has several new keywords that MASM 5.1 does not recognize as
reserved. To resolve any conflicts, you can:
u

Rename any offending symbols in your code.

Selectively disable keywords with the OPTION NOKEYWORD directive.

The second option lets you retain the offending symbol names in your code by
forcing MASM 6.1 to not recognize them as keywords. For example,
OPTION NOKEYWORD:

removes the keywords INVOKE and STRUCT from the assembler’s list of
reserved words. However, you cannot then use the keywords in their intended
function, since the assembler no longer recognizes them.
The following list shows MASM 6.1 reserved words new since MASM 5.1:
.BREAK
.CONTINUE
.DOSSEG
.ELSE
.ELSEIF
.ENDIF
.ENDW
.EXIT
.IF
.LISTALL
.LISTIF
.LISTMACRO
.LISTMACROALL
.NO87
.NOCREF
.NOLIST
.NOLISTIF
.NOLISTMACRO
.REPEAT
.STARTUP
.UNTIL

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 379 of 39 Printed: 10/02/00 04:18 PM

.UNTILCXZ
.WHILE
ADDR
ALIAS
BSWAP
CARRY?
CMPXCHG
ECHO
EXTERN
EXTERNDEF
FAR16
FAR32
FLAT
FLDENVD
FLDENVW
FNSAVED
FNSAVEW
FNSTENVD
FNSTENVW
FOR
FORC

FRSTORD
FRSTORW
FSAVED
FSAVEW
FSTENVD
FSTENVW
GOTO
HIGHWORD
INVD
INVLPG
INVOKE
IRETDF
IRETF
LENGTHOF
LOOPD
LOOPED
LOOPEW
LOOPNED
LOOPNEW
LOOPNZD
LOOPNZW

380

Programmer’s Guide

LOOPW

PUSHCONTEXT

SWORD

LOOPZW

PUSHD

SYSCALL

LOWWORD

PUSHW

TEXTEQU

LROFFSET

REAL10

TR3

NEAR16

REAL4

TR4

NEAR32

REAL8

TR5

OPATTR

REPEAT

TYPEDEF

OPTION

SBYTE

UNION

OVERFLOW?
PARITY?
POPAW
POPCONTEXT
PROTO
PUSHAW

SDWORD
SIGN?
SIZEOF
STDCALL
STRUCT
SUBTITLE

VARARG
WBINVD
WHILE
XADD
ZERO?

OPTION SETIF2
By default, MASM 6.1 does not recognize pass-dependent constructs. Both the
OPTION M510 and OPTION SETIF2 statements force MASM 6.1 to handle
MASM 5.1 constructs that activate on the second assembly pass, such as
.ERR2, IF2, and ELSEIF2.
Invoke the option like this:
OPTION SETIF2: {TRUE | FALSE}
When set to TRUE, OPTION SETIF2 forces all second-pass constructs to
activate on every assembly pass. When set to FALSE, second-pass constructs
do not activate on any pass. OPTION M510 implies OPTION SETIF2:TRUE.

Changes to Instruction Encodings
MASM 6.1 contains changes to the encodings for several instructions. In some
cases, the changes help optimize code size.

Coprocessor Instructions
For the 8087 coprocessor, MASM 5.1 adds an extra NOP before the no-wait
versions of coprocessor instructions. MASM 6.1 does not. In the rare case that
the missing NOP affects timing, insert NOP.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 380 of 40 Printed: 10/02/00 04:18 PM

Chapter 1 Chapter Head

381

For the 80287 coprocessor or better, MASM 5.1 inserts FWAIT before certain
instructions. MASM 6.1 does not prefix any 80287, 80387, or 80486
coprocessor instruction with FWAIT, except for wait forms of instructions that
have a no-wait form.

RET Instruction
MASM 5.1 generates a 3-byte encoding for RET, RETN, or RETF instructions
with an operand value of zero, unless the operand is an external absolute. In this
case, MASM 5.1 ignores the parameter and generates a 1-byte encoding.
MASM 6.1 does the opposite. It ignores a zero operand for the return
instructions and generates a 1-byte encoding, unless the operand is an external
absolute. In this case, MASM 6.1 generates a 3-byte encoding.
Thus, you can suppress epilogue code in a procedure but still specify the default
size for RET by coding the return as
ret

Arithmetic Instructions
Versions 5.1 and 6.1 differ in the way they encode the arithmetic instructions
ADC, ADD, AND, CMP, OR, SUB, SBB, and XOR, under the following
conditions:
u
u

The first operand is either AX or EAX.
The second operand is a constant value between 0 and 127.

For the AX register, there is no size or speed difference between the two
encodings. For the EAX register, the encoding in MASM 6.1 is 2 bytes smaller.
The OPTION NOSIGNEXTEND directive forces the MASM 5.1 behavior for
AND, OR,
and XOR.

Filename: LMAPGAPA.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 381 of 41 Printed: 10/02/00 04:18 PM

379

A P P E N D I X

BNF Grammar

This appendix provides a complete description of symbols, operators, and
directives for MASM 6.1. It uses the Backus-Naur Form (BNF) for grammar
notation. You can use BNF grammar to determine the exact syntax for any
language component and find all available options for any MASM command.
BNF definitions consist of “nonterminals” and “terminals.” Nonterminals are
placeholders within a BNF definition, defined elsewhere in the BNF grammar.
Terminals are endpoints in a BNF definition, consisting of MASM 6.1
keywords. In this Appendix, all nonterminals appear in italics type and all
terminals appear in bold type.

BNF Conventions
The conventions use different font attributes for different items in the BNF. The
symbols and formats are as follows:
Attribute

Description

nonterminal

Italic type indicates nonterminals.

RESERVED

Terminals in boldface type are literal reserved words and
symbols that must be entered as shown. Characters in this
context are always case insensitive.

[[ ]]

Objects enclosed in double brackets ([[ ]]) are optional. The
brackets do not actually appear in the source code.

A vertical bar indicates a choice between the items on each side
of the bar.

.8086

Underlined items indicate the default option if one is given.

default typeface

Characters in the set described or listed can be used as
terminals in MASM statements.

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 379 of 1 Printed: 10/02/00 04:19 PM

380

Programmer’s Guide

How to Use the BNF Grammar
To illustrate the use of the BNF, Figure B.1 diagrams the definition of the
TYPEDEF directive, starting with the nonterminal typedefDir.
The entries under each horizontal brace in Figure B.1 are terminals (such as
NEAR16, NEAR32, FAR16, and FAR32) or nonterminals (such as qualifier,
qualifiedType, distance, and protoSpec) that can be further defined. Each
italicized nonterminal in the typedefDir definition is also an entry in the BNF.
Three vertical dots indicate a branching definition for a nonterminal that, for the
sake of simplicity, this figure does not illustrate.
The BNF grammar allows recursive definitions. For example, the grammar uses
qualifiedType as a possible definition for qualifiedType, which is also a
component of the definition for qualifier.

Figure B.1

BNF Definition of the TYPEDEF Directive

Nonterminal

Definition

;;

endOfLine
| comment

=Dir

id = immExpr ;;

addOp

+|-

aExpr

term
| aExpr && term

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 380 of 2 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

altId

arbitraryText

charList

asmInstruction

mnemonic [[ exprList ]]

assumeDir

ASSUME assumeList ;;
| ASSUME NOTHING ;;

assumeList

assumeRegister
| assumeList , assumeRegister

assumeReg

assumeRegister

assumeSegReg
| assumeReg

assumeSegReg

segmentRegister : assumeSegVal

assumeSegVal

frameExpr
| NOTHING | ERROR

assumeVal

qualifiedType
| NOTHING | ERROR

bcdConst

[[ sign ]] decNumber

binaryOp

== | != | >= | <= | > | < | &

bitDef

bitFieldId : bitFieldSize [[ = constExpr ]]

bitDefList

bitDef
| bitDefList , [[ ;; ]] bitDef

bitFieldId

bitFieldSize

constExpr

blockStatements

directiveList
| .CONTINUE [[ .IF cExpr ]]
| .BREAK [[ .IF cExpr ]]

bool

TRUE | FALSE

byteRegister

AL | AH | BL | BH | CL | CH | DL | DH

cExpr

aExpr
| cExpr || aExpr

character

Any character with ordinal in the range 0–255
except linefeed (10)

charList

character
| charList character

className

string

commDecl

[[ nearfar ]] [[ langType ]] id : commType
[[ : constExpr ]]

commDir

COMM commList ;;

comment

; text ;;

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 381 of 3 Printed: 10/02/00 04:19 PM

381

382

Programmer’s Guide
Nonterminal

Definition

commentDir

COMMENT delimiter
text
text delimiter text ;;

commList

commDecl
| commList , commDecl

commType

type
| constExpr

constant

digits [[ radixOverride ]]

constExpr

expr

contextDir

PUSHCONTEXT contextItemList ;;
| POPCONTEXT contextItemList ;;

contextItem

ASSUMES | RADIX | LISTING | CPU | ALL

contextItemList

contextItem
| contextItemList , contextItem

controlBlock

whileBlock
| repeatBlock

controlDir

controlIf
| controlBlock

controlElseif

.ELSEIF cExpr ;;

directiveList
[[ controlElseif ]]
controlIf

.IF cExpr ;;

directiveList
[[ controlElseif ]]
[[ .ELSE ;;
directiveList ]]
.ENDIF ;;
coprocessor

.8087 | .287 | .387 | .NO87

crefDir

crefOption ;;

crefOption

.CREF

| .XCREF [[ idList ]]
| .NOCREF [[ idList ]]
cxzExpr

expr
| ! expr
| expr == expr
| expr != expr

dataDecl

DB | DW | DD | DF | DQ | DT | dataType | typeId

dataDir

[[ id ]] dataItem ;;

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 382 of 4 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

dataItem

dataDecl scalarInstList
| structTag structInstList
| typeId structInstList
| unionTag structInstList
| recordTag recordInstList

dataType

decdigit

0|1|2|3|4|5|6|7|8|9

decNumber

decdigit
| decNumber decdigit

delimiter

Any character except whiteSpaceCharacter

digits

decdigit
| digits decdigit
| digits hexdigit

directive

generalDir
| segmentDef

directiveList

directive
| directiveList directive

distance

nearfar
| NEAR16 | NEAR32 | FAR16 | FAR32

e01

e01 orOp e02
| e02

e02

e02 AND e03
| e03

e03

NOT e04

| e04
e04

e04 relOp e05
| e05

e05

e05 addOp e06
| e06

e06

e06 mulOp e07
| e06 shiftOp e07
| e07

e07

e07 addOp e08
| e08

e08

HIGH e09
| LOW e09
| HIGHWORD e09
| LOWWORD e09

| e09

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 383 of 5 Printed: 10/02/00 04:19 PM

383

384

Programmer’s Guide
Nonterminal

Definition

e09

| e09 : e10
| e10
e10

e10 . e11
| e10 [[ expr ]]
| e11

e11

echoDir

ECHO arbitraryText ;;
%OUT arbitraryText ;;

elseifBlock

elseifStatement ;;
directiveList
[[ elseifBlock ]]

elseifStatement

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 384 of 6 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

endDir

END [[ immExpr ]] ;;

endpDir

procId ENDP ;;

endsDir

id ENDS ;;

equDir

textMacroId EQU equType ;;

equType

immExpr
| textLiteral

errorDir

errorOpt ;;

errorOpt

exitDir

.EXIT [[ expr ]] ;;

exitmDir:

EXITM
| EXITM textItem

exponent

E [[ sign ]] decNumber

expr

SHORT e05
| .TYPE e01
| OPATTR e01

| e01
exprList

expr
| exprList , expr

externDef

[[ langType ]] id [[ ( altId ) ]] : externType

externDir

externKey externList ;;

externKey

EXTRN | EXTERN | EXTERNDEF

externList

externDef
| externList , [[ ;; ]] externDef

externType

ABS

| qualifiedType
fieldAlign

constExpr

fieldInit

[[ initValue ]]
| structInstance

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 385 of 7 Printed: 10/02/00 04:19 PM

385

386

Programmer’s Guide
Nonterminal

Definition

fieldInitList

fieldInit
| fieldInitList , [[ ;; ]] fieldInit

fileChar

delimiter

fileCharList

fileChar
| fileCharList fileChar

fileSpec

fileCharList
| textLiteral

flagName

ZERO? | CARRY? | OVERFLOW?
| SIGN? | PARITY?

floatNumber

[[ sign ]] decNumber . [[ decNumber ]] [[ exponent ]]
| digits R
| digits r

forcDir

FORC | IRPC

forDir

FOR | IRP

forParm

id [[ : forParmType ]]

forParmType

REQ

| = textLiteral
frameExpr

SEG id

| DGROUP : id
| segmentRegister : id
| id
generalDir

gpRegister

AX | EAX | BX | EBX | CX | ECX | DX | EDX
| BP | EBP | SP | ESP | DI | EDI | SI | ESI

groupDir

groupId GROUP segIdList

groupId

hexdigit

a|b|c|d|e|f
|A|B|C|D|E|F

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 386 of 8 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

alpha
| id alpha
| id decdigit

idList

id
| idList , id

ifDir

ifStatement ;;
directiveList
[[ elseifBlock ]]
[[ ELSE ;;
directiveList ]]
ENDIF ;;

ifStatement

immExpr

expr

includeDir

INCLUDE fileSpec ;;

includeLibDir

INCLUDELIB fileSpec ;;

initValue

inSegDir

[[ labelDef ]] inSegmentDir

inSegDirList

inSegDir
| inSegDirList inSegDir

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 387 of 9 Printed: 10/02/00 04:19 PM

387

388

Programmer’s Guide
Nonterminal

Definition

inSegmentDir

instrPrefix

instruction

[[ instrPrefix ]] asmInstruction

invokeArg

invokeDir

INVOKE expr [[ , [[ ;; ]] invokeList ]] ;;

invokeList

invokeArg
| invokeList , [[ ;; ]] invokeArg

keyword

Any reserved word

keywordList

keyword
| keyword keywordList

labelDef

id :
| id ::
| @@:

labelDir

id LABEL qualifiedType ;;

langType

listDir

listOption ;;

listOption

localDef

LOCAL idList ;;

localDir

LOCAL parmList ;;

localDirList

localDir
| localDirList localDir

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 388 of 10 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

localList

localDef
| localList localDef

macroArg

macroArgList

macroArg
| macroArgList , macroArg

macroBody

[[ localList ]]
macroStmtList

macroCall

id macroArgList ;;
| id ( macroArgList )

macroDir

id MACRO [[ macroParmList ]] ;;
macroBody
ENDM ;;

macroFor

forDir forParm , < macroArgList > ;;
macroBody
ENDM ;;

macroForc

forcDir id , textLiteral ;;
macroBody
ENDM ;;

macroFuncId

macroId

macroProcId
| macroFuncId

macroIdList

macroId
| macroIdList , macroId

macroLabel

macroParm

id [[ : parmType ]]

macroParmList

macroParm
| macroParmList , [[ ;; ]] macroParm

macroProcId

macroRepeat

repeatDir constExpr ;;
macroBody
ENDM ;;

macroStmt

directive
| exitmDir
| : macroLabel
| GOTO macroLabel

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 389 of 11 Printed: 10/02/00 04:19 PM

389

390

Programmer’s Guide
Nonterminal

Definition

macroStmtList

macroStmt ;;
| macroStmtList macroStmt ;;

macroWhile

WHILE constExpr ;;

macroBody
ENDM ;;
mapType

ALL | NONE | NOTPUBLIC

memOption

mnemonic

Instruction name

modelDir

.MODEL memOption [[ , modelOptlist ]] ;;

modelOpt

langType
| stackOption

modelOptlist

modelOpt
| modelOptlist , modelOpt

module

[[ directiveList ]] endDir

mulOp

* | / | MOD

nameDir

NAME id ;;

nearfar

NEAR | FAR

nestedStruct

structHdr [[ id ]] ;;
structBody
ENDS ;;

offsetDir

offsetDirType ;;

offsetDirType

EVEN
| ORG immExpr
| ALIGN [[ constExpr ]]

offsetType

GROUP | SEGMENT | FLAT

oldRecordFieldList

[[ constExpr ]]
| oldRecordFieldList , [[ constExpr ]]

optionDir

OPTION optionList ;;

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 390 of 12 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

optionItem

optionList

optionItem
| optionList , [[ ;; ]] optionItem

optText

, textItem

orOp

OR | XOR

oVisibility

PUBLIC | PRIVATE | EXPORT

pageDir

PAGE [[ pageExpr ]] ;;

pageExpr

+
| [[ pageLength ]] [[ , pageWidth ]]

pageLength

constExpr

pageWidth

constExpr

parm

parmId [[ : qualifiedType ]]
| parmId [[ constExpr ]] [[ : qualifiedType ]]

parmId

parmList

parm
| parmList , [[ ;; ]] parm

parmType

REQ

| = textLiteral
| VARARG
pOptions

[[ distance ]] [[ langType ]] [[ oVisibility ]]

primary

expr binaryOp expr
| flagName
| expr

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 391 of 13 Printed: 10/02/00 04:19 PM

391

392

Programmer’s Guide
Nonterminal

Definition

procDir

procId PROC [[ pOptions ]] [[ < macroArgList > ]]
[[ usesRegs ]] [[ procParmList ]]

processor

.8086
| .186
| .286 | .286C | .286P
| .386 | .386C | .386P
| .486 | .486P

processorDir

processor ;;
| coprocessor ;;

procId

procParmList

[[ , [[ ;; ]] parmList ]]
[[ , [[ ;; ]] parmId :VARARG]]

protoArg

[[ id ]] : qualifiedType

protoArgList

[[ , [[ ;; ]] protoList ]]
[[ , [[ ;; ]] [[ id ]] :VARARG ]]

protoList

protoArg
| protoList , [[ ;; ]] protoArg

protoSpec

[[ distance ]] [[ langType ]] [[ protoArgList ]]
| typeId

protoTypeDir

id PROTO protoSpec

pubDef

[[ langType ]] id

publicDir

PUBLIC pubList ;;

pubList

pubDef
| pubList , [[ ;; ]] pubDef

purgeDir

PURGE macroIdList

qualifiedType

type
| [[ distance ]] PTR [[ qualifiedType ]]

qualifier

qualifiedType
| PROTO protoSpec

quote

“
|‘

radixDir

.RADIX constExpr ;;

radixOverride

h|o|q|t|y
|H|O|Q|T|Y

recordConst

recordTag { oldRecordFieldList }
| recordTag < oldRecordFieldList >

recordDir

recordTag RECORD bitDefList ;;

recordFieldList

[[ constExpr ]]
| recordFieldList , [[ ;; ]] [[ constExpr ]]

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 392 of 14 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

recordInstance

{ [[ ;; ]] recordFieldList [[ ;; ]] }
| < oldRecordFieldList >
| constExpr DUP ( recordInstance )

recordInstList

recordInstance
| recordInstList , [[ ;; ]] recordInstance

recordTag

specialRegister
| gpRegister
| byteRegister

regList

relOp

EQ | NE | LT | LE | GT | GE

repeatBlock

.REPEAT ;;

blockStatements ;;
untilDir ;;
repeatDir

REPEAT | REPT

scalarInstList

initValue
| scalarInstList , [[ ;; ]] initValue

segAlign

BYTE | WORD | DWORD | PARA | PAGE

segAttrib

segDir

segId

segIdList

segId
| segIdList , segId

segmentDef

segmentDir [[ inSegDirList ]] endsDir
| simpleSegDir [[ inSegDirList ]] [[ endsDir ]]

segmentDir

segId SEGMENT [[ segOptionList ]] ;;

segmentRegister

CS | DS | ES | FS | GS | SS

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 393 of 15 Printed: 10/02/00 04:19 PM

393

394

Programmer’s Guide
Nonterminal

Definition

segOption

segAlign
| segRO
| segAttrib
| segSize
| className

segOptionList

segOption
| segOptionList segOption

segOrderDir

.ALPHA | .SEQ | .DOSSEG | DOSSEG

segRO

READONLY

segSize

USE16 | USE32 | FLAT

shiftOp

SHR | SHL

sign

-|+

simpleExpr

( cExpr )
| primary

simpleSegDir

segDir ;;

sizeArg

id
| type
| e10

specialChars

: | . | [[ | ]] | ( | ) | < | > | { | }
|+|-|/|*|&|%|!
|’|\|=|;|,|“
| whiteSpaceCharacter
| endOfLine

specialRegister

CR0 | CR2 | CR3
| DR0 | DR1 | DR2 | DR3 | DR6 | DR7
| TR3 | TR4 | TR5 | TR6 | TR7

stackOption

NEARSTACK | FARSTACK

startupDir

.STARTUP ;;

stext

stringChar
| stext stringChar

string

quote [[ stext ]] quote

stringChar

quote quote
| Any character except quote

structBody

structItem ;;
| structBody structItem ;;

structDir

structTag structHdr [[ fieldAlign ]]
[[, NONUNIQUE ]] ;;
structBody
structTag ENDS ;;

structHdr

STRUC | STRUCT | UNION

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 394 of 16 Printed: 10/02/00 04:19 PM

Appendix B BNF Grammar
Nonterminal

Definition

structInstance

< [[ fieldInitList ]] >
| { [[ ;; ]] [[ fieldInitList ]] [[ ;; ]] }
| constExpr DUP ( structInstList )

structInstList

structInstance
| structInstList , [[ ;; ]] structInstance

structItem

dataDir
| generalDir
| offsetDir
| nestedStruct

structTag

term

simpleExpr
| ! simpleExpr

text

textLiteral
| text character
| ! character text
| character
| ! character

textDir

id textMacroDir ;;

textItem

textLiteral
| textMacroId
| % constExpr

textLen

constExpr

textList

textItem
| textList , [[ ;; ]] textItem

textLiteral

< text >;;

textMacroDir

CATSTR [[ textList ]]
| TEXTEQU [[ textList ]]
| SIZESTR textItem
| SUBSTR textItem , textStart [[ , textLen ]]
| INSTR [[ textStart , ]] textItem , textItem

textMacroId

textStart

constExpr

titleDir

titleType arbitraryText ;;

titleType

TITLE | SUBTITLE | SUBTTL

type

typedefDir

typeId TYPEDEF qualifier

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 395 of 17 Printed: 10/02/00 04:19 PM

395

396

Programmer’s Guide
Nonterminal

Definition

typeId

unionTag

untilDir

.UNTIL cExpr ;;
.UNTILCXZ [[ cxzExpr ]] ;;

usesRegs

USES regList

whileBlock

.WHILE cExpr ;;

blockStatements ;;
.ENDW

whiteSpaceCharacter

Filename: LMAPGAPB.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 396 of 18 Printed: 10/02/00 04:19 PM

ASCII 8, 9, 11–13, 26, 32

397

A P P E N D I X

Generating and Reading
Assembly Listings

A listing file shows precisely how the assembler translates your source file into
machine code. The listing documents the assembler’s assumptions, memory
allocations, and optimizations.
MASM creates an assembly listing of your source file whenever you do one of
the following:
u
u
u

Select the appropriate option in PWB.
Use one of the related source code directives.
Specify the /Fl option on the MASM command line.

The assembly listing contains both the statements in the source file and the
binary code (if any) generated for each statement. The listing also shows the
names and values of all labels, variables, and symbols in your file.
The assembler creates tables for macros, structures, unions, records, segments,
groups, and other symbols, and places the tables at the end of the assembly
listing. Only the types of symbols encountered in the program are included. For
example, if your program has no macros, the symbol table does not have a
macros section.

Generating Listing Files
To generate a listing file from within PWB, follow these steps:
1. From the Options menu, choose MASM Options.
2. In the MASM Options dialog box, choose Set Debug or Release Options.
The dialog box for Set Debug or Release Options lists the choices summarized
in Table C.1. This table also shows the equivalent source code directives and
command-line options.

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 397 of 1 Printed: 10/02/00 04:19 PM

398

Programmer’s Guide

Table C.1

Options for Generating or Modifying Listing Files

To generate this information:
To generate this
information:

In PWB1, select:

In source
code, enter:

From command
line, enter:

Default listing — includes all
assembled lines

Generate Listing File

.LIST (default)

/Fl

Turn off all source
listings (overrides all listing
directives)

Generate Listing File (turn
off)

.NOLIST
(synonym = .SFCOND)

—

List all source lines, including
false conditionals and
generated code

Include All Source Lines

.LISTALL

/Fl /Sa

Show instruction timings

List Instruction
Timings

—

/Fl /Sc

Show assembler-generated
code

List Generated
Instructions

—

/Fl /Sg

Include false conditionals2

List False
Conditionals

.LISTIF
(synonym = .LFCOND)

/Fl /Sx

Suppress listing of any
subsequent conditional blocks
whose condition is false

List False Conditionals
(turn off)

.NOLISTIF
(synonym = .SFCOND)

—

Toggle between .LISTIF and
.NOLISTIF

—

.TFCOND

—

Suppress symbol table
generation

Generate Symbol Table
(turn off the default)

—

/Fl /Sn

List all processed macro
statements

—

.LISTMACROALL
(synonym = .LALL)

—

List only instructions, data,
and segment directives in
macros

—

.LISTMACRO
(default)
(synonym = .XALL)

—

Turn off all listing during
macro expansion

—

.NOLISTMACRO
(synonym = .SALL)

—

Specify title for each page
(use only once per file)

—

TITLE name

/St name

Specify subtitle for page

—

SUBTITLE name

/Ss name

Designate page length and line
width, increment section
number, or generate page
breaks

—

PAGE
[[length,width]][[+]]

/Sp length
/Sl width

Generate first-pass listing

—

/Ep

1 Select MASM Options from the Options menu, then choose Set Dialog Options from the MASM Options dialog box.

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 398 of 2 Printed: 10/02/00 04:19 PM

Appendix C Generating and Reading Assembly Listings

399

2 See “Conditional Directives” in Chapter 1

Precedence of Command-Line Options and Listing Directives
Since command-line options and source code directives can specify opposite
behavior for the same listing file option, the assembler interprets the commands
according to the following precedence levels. Selecting PWB options is
equivalent to specifying /Fl /Sx on the command line:
u

/Sa overrides any source code directives that suppress listing.

Source code directives override all command-line options except /Sa.

.NOLIST overrides other listing directives such as .NOLISTIF and
.LISTMACROALL.
The /Sx, /Ss, /Sp, and /Sl options set initial values for their respective
features. Directives in the source file can override these command-line
options.

Reading the Listing File
The first half of the listing shows macros from the include file DOS.MAC,
structure declarations, and data. After the .DATA directive, the columns on the
left show offsets and initialized byte values within the data segment.
Instructions begin after the .CODE directive. The three columns on the left
show offsets, instruction timings, and binary code generated by the assembler.
The columns on the right list the source statements exactly as they appear in the
source file or as expanded by a macro. Various symbols and abbreviations in the
middle column provide information about the code, as explained in the following
section. The subsequent section, “Symbols and Abbreviations,” explains the
meanings of listing symbols.

Generated Code
The assembler lists the code generated from the statements of a source file.
With the /Sc command-line switch, which generates instruction timings, each
line has this syntax:
offset [[timing]] [[code]]
The offset is the offset from the beginning of the current code segment. The
timing shows the number of cycles the processor needs to execute the
instruction. The value of timing reflects the CPU type; for example, specifying
the .386 directive produces instruction timings for the 80386 processor. If the
statement generates code or data, code shows the numeric value in hexadecimal

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 399 of 3 Printed: 10/02/00 04:19 PM

400

Programmer’s Guide

notation if the value is known at assembly time. If the value is calculated at run
time, the assembler indicates what action is necessary to compute the value.
When assembling under the default .8086 directive, timing includes an effective
address value if the instruction accesses memory. The 80186/486 processors do
not use effective address values. For more information on effective address
timing, see the “Processor” section in the Reference book.

Error Messages
If any errors occur during assembly, each error message and error number
appears directly below the statement where the error occurred. An example of
an error line and message is:
mov
ax, [dx][di]
listtst.asm(77): error A2031: must be index or base register

Symbols and Abbreviations
The assembler uses the symbols and abbreviations shown in Table C.2 to
indicate addresses that need to be resolved by the linker or values that were
generated in a special way. The example in this section illustrates many of these
symbols.
The example listing was produced using “List Generated Instructions” and “List
Instruction Timings” in PWB. These options correspond to the ML commandline switches /Fl /Sg /Sc.
Table C.2

Symbols and Abbreviations in Listings

Character

Meaning

Line from include file

EQU or equal-sign (=) directive

nn[xx]

DUP expression:

----

Segment/group address (linker must resolve)

Relocatable address (linker must resolve)

Assembler-generated code

External address (linker must resolve)

Macro-expansion nesting level (+ if more than 9)

Operator size override

nn copies of the value xx

Address size override

nn:

Segment override in statement

nn/

REP

or LOCK prefix instruction

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 400 of 4 Printed: 10/02/00 04:19 PM

Appendix C Generating and Reading Assembly Listings

Table C.3 explains the five symbols that may follow timing values in your
listing. The Reference book will help you determine correct timings for those
values marked with a symbol.

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 401 of 5 Printed: 10/02/00 04:19 PM

401

402

Programmer’s Guide
Table C.3 Symbols in Timing Column
Symbol

Meaning

Add cycles depending on next executed instruction.

Add cycles depending on number of iterations or size of data.

Different timing value in protected mode.

Add cycles depending on operands or combination of the preceding.

Separates two values for “jump taken” and “jump not taken.”

Microsoft (R) Macro Assembler Version 6.10
listtst.asm

.MODEL
.386
.DOSSEG
.STACK
INCLUDE
C StrDef MACRO
C name1
BYTE
C
BYTE
C l&name1 EQU
C
ENDM
C
C Display MACRO
C
mov
C
mov
C
int
C
ENDM
num
EQU
COLOR
RECORD
value
TEXTEQU
tnum
TEXTEQU
strpos TEXTEQU

= 0020
= 35
= 32
= 04
)

0004
0000
0001
0002

09/20/00

01
01
0000

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 402 of 6 Printed: 10/02/00 04:19 PM

12:00:00
Page 1 - 1

small, c

256
dos.mac
name1, text
&text
13d, 10d, '$'
LENGTHOF name1

string
ah, 09h
dx, OFFSET string
21h
20h
b:1, r:3=1, i:1=1, f:3=7
%3 + num
%num
@InStr( , ,

PutStr

PROTO

pMsg:PTR BYTE

DATE
month
day
year
DATE

STRUCT
BYTE
BYTE
WORD
ENDS

1
1
?

Appendix C Generating and Reading Assembly Listings
0002
0000

U1
fsize
bsize
U1

0028

0000
0000
0004
0005
0009
000A

UNION
WORD
BYTE
ENDS

40
60

.DATA
00000000
1F
01 14 07C9
00
001E [
0000
]

0046 46 69 6E 69
65 64
004F 0D 0A 24
= 0009
0052 54 68 69 73
string","0"
73 20
73 74
6E 67

73 68
2E

20 69

ddData
text
today
flag
buffer

DWORD
COLOR
DATE
BYTE
WORD

ending

StrDef
BYTE

1
1

BYTE
lending EQU
Msg

?
<>
<01, 20, 1993>
0
30 DUP (0)

ending, "Finished."
"Finished."
13d, 10d, '$'
LENGTHOF ending
BYTE
"This is a

61 20
72 69
30

0063 ---- 0052 R

float
FPBYTE
FPMSG
PBYTE
NPWORD
PVOID
PPBYTE

0000

TYPEDEF
TYPEDEF FAR
FPBYTE
TYPEDEF
TYPEDEF NEAR
TYPEDEF
TYPEDEF

REAL4
PTR BYTE
Msg
PTR BYTE
PTR WORD
PTR
PTR PBYTE

.CODE
.STARTUP

0000
0000
0003
0005
0007
0009
000C
000E

2
2p
2
2
3
2p
2

B8
8E
8C
2B
C1
8E
03

0010

E8 0000 E

---- R
D8
D3
D8
E3 04
D0
E3

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 403 of 7 Printed: 10/02/00 04:19 PM

*@Startup:
*
mov
*
mov
*
mov
*
sub
*
shl
*
mov
*
add

ax,
ds,
bx,
bx,
bx,
ss,
sp,

EXTERNDEF
call

DGROUP
ax
ss
ax
004h
ax
bx
work:NEAR
work

403

404

Programmer’s Guide

0013
0016
0019

2
7m
2

68 0052 R
E8 0029
83 C4 02

001C
001F
0021
0023

2
2p
2
4

0028
002B
002D
0031

2
7n
4
6

B8 ---- R
8E C0
B0 63
26: 8B 0E
0020
BF 0052
F2/ AE
66| A1 0000 R
67& FE 03

0034

E8 0000 E

0037
0039
003C

2
2
37

B4 09
BA 0046 R
CD 21

003E
0040

2
37

B4 4C
CD 21

*
*
*

push
call
add

55
8B
B4
8B
8A

004C
0059
0059
005B
005C

EB 10

ax,
es,
al,
cx,

mov
repne
mov
inc

di, 82
scasb
eax, ddData
BYTE PTR [ebx]

Display
mov
mov
int

1
1
1

*
*

mov
int

37
2
4

005E
005E
2
0060 7m,3

CD 21
47
8A 15

0A D2
75 F7

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 404 of 8 Printed: 10/02/00 04:19 PM

*
*

morework:NEAR
morework
ending
ah, 09h
dx, OFFSET ending
21h

PROC

pMsg:PTR BYTE

bp
bp, sp
mov
ah,
mov
di,
mov
dl,
mov
ax,
listtst.asm(77): error A2031: must be index or base
EC
02
7E 04
15

@data
ax
'c'
es:num

.EXIT
ah, 04Ch
021h

PutStr
2
4
2
4
4

mov
mov
mov
mov

EXTERNDEF
call

0042
0042
0043
0045
0047
004A

INVOKE PutStr, ADDR msg
OFFSET Msg
PutStr
sp, 00002h

push
mov

*
jmp
*@C0002:

02H
pMsg
[di]
[dx][di]
register

.WHILE (dl)
@C0001
int
inc
mov
.ENDW

21h
di
dl, [di]

*@C0001:
*
or dl, dl
*
jne
@C0002
ret

Appendix C Generating and Reading Assembly Listings
0062
0063
0064

4
10m

5D
C3

*
*

405

pop
bp
ret
00000h
PutStr ENDP
END

Reading Tables in a Listing File
The tables at the end of a listing file list the macros, structures, unions, records,
segments, groups, and symbols that appear in a source file. These tables are not
printed in the previous sample listing, but are summarized as follows.

Macro Table
Lists all macros in the main file or the include files. Differentiates between
macro functions and macro procedures.

Structures and Unions Table
Provides the size in bytes of the structure or union and the offset of each field.
The type of each field is also given.

Record Table
“Width” gives the number of bits of the entire record. “Shift” provides the offset
in bits from the low-order bit of the record to the low-order bit of the field.
“Width” for fields gives the number of bits in the field. “Mask” gives the
maximum value of the field, expressed in hexadecimal notation. “Initial” gives
the initial value supplied for the field.

Type Table
The “Size” column in this table gives the size of the TYPEDEF type in bytes,
and the “Attr” column gives the base type for the TYPEDEF definition.

Segment and Group Table
“Size” specifies whether the segment is 16 bit or 32 bit. “Length” gives the size
of the segment in bytes. “Align” gives the segment alignment (WORD, PARA,
and so on). “Combine” gives the combine type (PUBLIC, STACK, and so on).
“Class” gives the segment’s class (CODE, DATA, STACK, or CONST).

Procedures, Parameters, and Locals
Gives the types and offsets from BP of all parameters and locals defined in each
procedure, as well as the size and memory location of each procedure.

Symbol Table
All symbols (except names for macros, structures, unions, records, and
segments) are listed in a symbol table at the end of the listing. The “Name”

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 405 of 9 Printed: 10/02/00 04:19 PM

406

Programmer’s Guide

column lists the names in alphabetical order. The “Type” column lists each
symbol’s type.
The length of a multiple-element variable, such as an array or string, is the
length of a single element, not the length of the entire variable.
If the symbol represents an absolute value defined with an EQU or equal sign
(=) directive, the “Value” column shows the symbol’s value. The value may be
another symbol, a string, or a constant numeric value (in hexadecimal),
depending on the type. If the symbol represents a variable or label, the “Value”
column shows the symbol’s hexadecimal offset from the beginning of the
segment in which it is defined.
The “Attr” column shows the attributes of the symbol. The attributes include
the name of the segment (if any) in which the symbol is defined, the scope of
the symbol, and the code length. A symbol’s scope is given only if the symbol is
defined using the EXTERN and PUBLIC directives. The scope can be external,
global, or communal. The “Attr” column is blank if the symbol has no attribute.

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 406 of 10 Printed: 10/02/00 04:19 PM

406

Programmer’s Guide

Filename: LMAPGAPC.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 406 of 12 Printed: 10/02/00 04:19 PM

407

A P P E N D I X

MASM Reserved Words

This appendix lists the reserved words recognized by MASM. They are divided
primarily by their use in the language. The primary categories are:
u
u
u
u
u

Operands and symbols
Registers
Operators and directives
Processor instructions
Coprocessor instructions

Reserved words in MASM 6.1 are reserved under all CPU modes. Words
enabled in .8086 mode, the default, can be used in all higher CPU modes. To
use words from subcategories such as “Special Operands for the 80386” (later
in this appendix) requires .386 mode or higher.
You can disable the recognition of any reserved word specified in this appendix
by setting the NOKEYWORD option for the OPTION directive. Once
disabled, the word can be used in any way as a user-defined symbol (provided
the word is a valid identifier). If you want to remove the STR instruction, the
MASK operator, and the NAME directive, for instance, from the set of words
MASM recognizes as reserved, add this statement to your program:
OPTION NOKEYWORD:

Words in this appendix identified with an asterisk (*) are new since MASM 5.1.

Operands and Symbols
The words on the two lists in this section are the operands to certain directives.
They have special meaning to the assembler. The words on the first list are not
reserved words. They can be used in every way as normal identifiers, without
affecting their use as operands to directives. The assembler interprets their use
from context.

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 407 of 1 Printed: 10/02/00 04:24 PM

408

Programmer’s Guide

Even though the words on the first list are not reserved, they should not be
defined to be text macros or text macro functions. If they are, they will not be
recognized in their special contexts. The assembler does not give a warning if
such a redefinition occurs.
ABS

LARGE

NOTHING

ALL

LISTING*

NOTPUBLIC*

ASSUMES

LJMP*

OLDMACROS*

LOADDS*

OLDSTRUCTS*

CASEMAP*

M510*

OS_DOS*

COMMON
COMPACT
CPU*
DOTNAME*
EMULATOR*
EPILOGUE*
ERROR*
EXPORT*
EXPR16*
EXPR32*
FARSTACK*
FLAT
FORCEFRAME
HUGE
LANGUAGE*

MEDIUM
MEMORY
NEARSTACK*
NODOTNAME*
NOEMULATOR*
NOKEYWORD*
NOLJMP*
NOM510*
NONE
NONUNIQUE*
NOOLDMACROS*
NOOLDSTRUCTS*
NOREADONLY*
NOSCOPED*
NOSIGNEXTEND*

PARA
PRIVATE*
PROLOGUE*
RADIX*
READONLY*
REQ*
SCOPED*
SETIF2*
SMALL
STACK
TINY
USE16
USE32
USES

These operands are reserved words. Reserved words are not case sensitive.
$
?
@B
@F
ADDR*
BASIC
BYTE
C
CARRY?*

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 408 of 2 Printed: 10/02/00 04:24 PM

DWORD
FAR
FAR16*
FORTRAN
FWORD
NEAR
NEAR16*
OVERFLOW?*
PARITY?*

PASCAL
QWORD
REAL4*
REAL8*
REAL10*
SBYTE*
SDWORD*
SIGN?*
STDCALL*

Appendix D MASM Reserved Words

SWORD*

TBYTE

WORD

SYSCALL*

VARARG*

ZERO?*

Special Operands for the 80386/486
FLAT*

NEAR32*

FAR32*

Predefined Symbols
Unlike most MASM reserved words, predefined symbols are case sensitive.
@Model*
@SizeStr*
@stack*
@SubStr*
@Time*
@Version
@WordSize

@Environ*
@fardata
@fardata?
@FileCur*
@FileName
@InStr*
@Interface*
@Line*

@CatStr*
@code
@CodeSize
@Cpu
@CurSeg
@data
@DataSize
@Date*

Registers
AH
AL
AX
BH
BL
BP
BX
CH
CL
CR0
CR2
CR3

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 409 of 3 Printed: 10/02/00 04:24 PM

CS
CX
DH
DI
DL
DR0
DR1
DR2
DR3
DR6
DR7
DS

DX
EAX
EBP
EBX
ECX
EDI
EDX
ES
ESI
ESP
FS
GS

SI
SP
SS
ST
TR3*
TR4*
TR5*
TR6
TR7

409

410

Programmer’s Guide

Operators and Directives
.186

.ERRDIFI

.TYPE

.286

.ERRE

.UNTIL*

.286C

.ERRIDN

.UNTILCXZ*

.286P

.ERRIDNI

.WHILE*

.287

.ERRNB

.XALL

.386

.ERRNDEF

.XCREF

.386C

.ERRNZ

.XLIST

.386P
.387
.486*
.486P*
.8086
.8087
.ALPHA
.BREAK*
.CODE
.CONST
.CONTINUE*
.CREF
.DATA
.DATA?
.DOSSEG*
.ELSE*
.ELSEIF*
.ENDIF*
.ENDW*
.ERR
.ERR1
.ERR2
.ERRB
.ERRDEF
.ERRDIF

.EXIT*
.FARDATA
.FARDATA?
.IF*
.LALL
.LFCOND
.LIST
.LISTALL*
.LISTIF*
.LISTMACRO*
.LISTMACROALL*
.MODEL
.NO87*
.NOCREF*
.NOLIST*
.NOLISTIF*
.NOLISTMACRO*
.RADIX
.REPEAT*
.SALL
.SEQ
.SFCOND
.STACK
.STARTUP*
.TFCOND

ALIAS*
ALIGN
ASSUME
CATSTR
COMM
COMMENT
DB
DD
DF
DOSSEG
DQ
DT
DUP
DW
ECHO*
ELSE
ELSEIF
ELSEIF1
ELSEIF2
ELSEIFB
ELSEIFDEF
ELSEIFDIF
ELSEIFDIFI
ELSEIFE
ELSEIFIDN

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 410 of 4 Printed: 10/02/00 04:24 PM

Appendix D MASM Reserved Words

ELSEIFIDNI

IFIDN

POPCONTEXT*

ELSEIFNB

IFIDNI

PROC

ELSEIFNDEF

IFNB

PROTO*

END

IFNDEF

PTR

ENDIF

INCLUDE

PUBLIC

ENDM

INCLUDELIB

PURGE

ENDP

INSTR

PUSHCONTEXT*

ENDS

INVOKE*

RECORD

EQ
EQU
EVEN
EXITM
EXTERN*
EXTERNDEF*
EXTRN
FOR*
FORC*
GE
GOTO*
GROUP
GT
HIGH
HIGHWORD*
IF
IF1
IF2
IFB
IFDEF
IFDIF
IFDIFI
IFE

IRP
IRPC
LABEL
LE
LENGTH
LENGTHOF*
LOCAL
LOW
LOWWORD*
LROFFSET*
LT
MACRO
MASK
MOD
.MSFLOAT
NAME
NE
OFFSET
OPATTR*
OPTION*
ORG
%OUT
PAGE

REPEAT*
REPT
SEG
SEGMENT
SHORT
SIZE
SIZEOF*
SIZESTR
STRUC
STRUCT*
SUBSTR
SUBTITLE*
SUBTTL
TEXTEQU*
THIS
TITLE
TYPE
TYPEDEF*
UNION*
WHILE*
WIDTH

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 411 of 5 Printed: 10/02/00 04:24 PM

411

412

Processor Instructions
Processor instructions are not case sensitive.

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 412 of 6 Printed: 10/02/00 04:24 PM

Appendix D MASM Reserved Words

8086/8088 Processor Instructions
AAA

JAE

LDS

AAD

LEA

AAM

JBE

LES

AAS

LODS

ADC

JCXZ

LODSB

ADD

LODSW

AND

LOOP

CALL
CBW
CLC
CLD
CLI
CMC
CMP
CMPS
CMPSB
CMPSW
CWD
DAA
DAS
DEC
DIV
ESC
HLT
IDIV
IMUL
IN
INC
INT
INTO
IRET
JA

JGE
JL
JLE
JMP
JNA
JNAE
JNB
JNBE
JNC
JNE
JNG
JNGE
JNL
JNLE
JNO
JNP
JNS
JNZ
JO
JP
JPE
JPO
JS
JZ
LAHF

LOOPE
LOOPEW*
LOOPNE
LOOPNEW*
LOOPNZ
LOOPNZW*
LOOPW*
LOOPZ
LOOPZW*
MOV
MOVS
MOVSB
MOVSW
MUL
NEG
NOP
NOT
OR
OUT
POP
POPF
PUSH
PUSHF
RCL
RCR

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 413 of 7 Printed: 10/02/00 04:24 PM

413

414

Programmer’s Guide

RET

SCAS

STOSB

RETF

SCASB

STOSW

RETN

SCASW

SUB

ROL

SHL

TEST

ROR

SHR

WAIT

SAHF

STC

XCHG

SAL

STD

XLAT

SAR

STI

XLATB

SBB

STOS

XOR

INSW
LEAVE
OUTS
OUTSB

OUTSW
POPA
PUSHA
PUSHW*

SIDT
SLDT
SMSW
STR

VERR
VERW

80186 Processor Instructions
BOUND
ENTER
INS
INSB

80286 Processor Instructions
ARPL
LAR
LSL
SGDT

80286 and 80386 Privileged-Mode Instructions
CLTS
LGDT

LIDT
LLDT

LMSW
LTR

CDQ
CMPSD
CWDE
INSD
IRETD
IRETDF*

IRETF*
JECXZ
LFS
LGS
LODSD
LOOPD*

80386 Processor Instructions
BSF
BSR
BT
BTC
BTR
BTS

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 414 of 8 Printed: 10/02/00 04:24 PM

Appendix D MASM Reserved Words

LOOPED*

SETAE

SETNGE

LOOPNED*

SETB

SETNL

LOOPNZD*

SETBE

SETNLE

LOOPZD*

SETC

SETNO

LSS

SETE

SETNP

MOVSD

SETG

SETNS

MOVSX

SETGE

SETNZ

MOVZX

SETL

SETO

OUTSD
POPAD
POPFD
PUSHAD
PUSHD*
PUSHFD
SCASD
SETA

SETLE
SETNA
SETNAE
SETNB
SETNBE
SETNC
SETNE
SETNG

SETP
SETPE
SETPO
SETS
SETZ
SHLD
SHRD
STOSD

INVD*
INVLPG*

WBINVD*
XADD*

REPE
REPNE

REPNZ
REPZ

80486 Processor Instructions
BSWAP*
CMPXCHG*

Instruction Prefixes
LOCK
REP

Coprocessor Instructions
Coprocessor instructions are not case sensitive.

8087 Coprocessor Instructions
F2XM1
FABS
FADD

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 415 of 9 Printed: 10/02/00 04:24 PM

FADDP
FBLD
FBSTP

FCHS
FCLEX
FCOM

415

416

Programmer’s Guide

FCOMP

FLDCW

FRNDINT

FCOMPP

FLDENV

FRSTOR

FDECSTP

FLDENVW*

FRSTORW*

FDISI

FLDL2E

FSAVE

FDIV

FLDL2T

FSAVEW*

FDIVP

FLDLG2

FSCALE

FDIVR

FLDLN2

FSQRT

FDIVRP

FLDPI

FST

FENI
FFREE
FIADD
FICOM
FICOMP
FIDIV
FIDIVR
FILD
FIMUL
FINCSTP
FINIT
FIST
FISTP
FISUB
FISUBR
FLD
FLD1

FLDZ
FMUL
FMULP
FNCLEX
FNDISI
FNENI
FNINIT
FNOP
FNSAVE
FNSAVEW*
FNSTCW
FNSTENV
FNSTENVW*
FNSTSW
FPATAN
FPREM
FPTAN

FSTCW
FSTENV
FSTENVW*
FSTP
FSTSW
FSUB
FSUBP
FSUBR
FSUBRP
FTST
FWAIT
FXAM
FXCH
FXTRACT
FYL2X
FYL2XP1

80287 Privileged-Mode Instruction
FSETPM

80387 Instructions
FCOS
FLDENVD*
FNSAVED*
FNSTENVD*

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 416 of 10 Printed: 10/02/00 04:24 PM

FPREM1
FRSTORD*
FSAVED*
FSIN

FSINCOS
FSTENVD*
FUCOM
FUCOMP

Appendix D MASM Reserved Words

FUCOMPP

Filename: LMAPGAPD.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 4 Page: 417 of 11 Printed: 10/02/00 04:24 PM

417

A P P E N D I X

Default Segment Names

If you use simplified segment directives by themselves, you do not need to
know the names assigned for each segment. However, it is possible to mix full
segment definitions with simplified segment directives, in which case you need
to know the segment names.
Table E.1 shows the default segment names created by each directive.
If you use .MODEL, a _TEXT segment is always defined, even if all .CODE
directives specify a name. The default segment name used as part of far-code
segment names is the filename of the module. The default name associated with
the .CODE directive can be overridden, as can the default names for
.FARDATA and .FARDATA?.
The segment and group table at the end of listings always shows the actual
segment names. However, the GROUP and ASSUME statements generated by
the .MODEL directive are not shown in listing files. For a program that uses all
possible segments, group statements equivalent to the following would be
generated:
DGROUP

GROUP

_DATA, CONST, _BSS, STACK

For the tiny model, these ASSUME statements would be generated:
ASSUME

cs:DGROUP, ds:DGROUP, ss:DGROUP

For small and compact models with NEARSTACK, these ASSUME statements
would be generated:
ASSUME

cs: _TEXT, ds:DGROUP, ss:DGROUP

Filename: LMAPGAPE.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 417 of 1 Printed: 10/02/00 04:24 PM

418

Programmer’s Guide

For medium, large, and huge models with NEARSTACK, these ASSUME
statements would be generated:
ASSUME

Table E.1
Model
Tiny

Small

Medium

Compact

cs:name_TEXT, ds:DGROUP, ss:DGROUP

Default Segments and Types for Standard Memory Models
Directive

Name

Align

Combine

Class

Group
DGROUP

.CODE

_TEXT

WORD

PUBLIC

'CODE'

.FARDATA

FAR_DATA

PARA

PRIVATE

'FAR_DATA'

.FARDATA?

FAR_BSS

PARA

PRIVATE

'FAR_BSS'

.DATA

_DATA

WORD

PUBLIC

'DATA'

DGROUP

.CONST

CONST

WORD

PUBLIC

'CONST'

DGROUP

.DATA?

_BSS

WORD

PUBLIC

'BSS'

DGROUP

.CODE

_TEXT

WORD

PUBLIC

'CODE'

.FARDATA

FAR_DATA

PARA

PRIVATE

'FAR_DATA'

.FARDATA?

FAR_BSS

PARA

PRIVATE

'FAR_BSS'

.DATA

_DATA

WORD

PUBLIC

'DATA'

DGROUP

.CONST

CONST

WORD

PUBLIC

'CONST'

DGROUP

.DATA?

_BSS

WORD

PUBLIC

'BSS'

DGROUP

.STACK

STACK

PARA

STACK

'STACK'

DGROUP *

.CODE

name_TEXT

WORD

PUBLIC

'CODE'

.FARDATA

FAR_DATA

PARA

PRIVATE

'FAR_DATA'

.FARDATA?

FAR_BSS

PARA

PRIVATE

'FAR_BSS'

.DATA

_DATA

WORD

PUBLIC

'DATA'

DGROUP

.CONST

CONST

WORD

PUBLIC

'CONST'

DGROUP

.DATA?

_BSS

WORD

PUBLIC

'BSS'

DGROUP

.STACK

STACK

PARA

STACK

'STACK'

DGROUP*

.CODE

_TEXT

WORD

PUBLIC

'CODE'

.FARDATA

FAR_DATA

PARA

PRIVATE

'FAR_DATA'

.FARDATA?

FAR_BSS

PARA

PRIVATE

'FAR_BSS'

.DATA

_DATA

WORD

PUBLIC

'DATA'

DGROUP

.CONST

CONST

WORD

PUBLIC

'CONST'

DGROUP

.DATA?

_BSS

WORD

PUBLIC

'BSS'

DGROUP

.STACK

STACK

PARA

STACK

'STACK'

DGROUP*

Filename: LMAPGAPE.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 418 of 2 Printed: 10/02/00 04:24 PM

Appendix E Default Segment Names
Table E.1

419

(continued)

Model

Directive

Name

Align

Combine

Class

Large or huge

.CODE

name_TEXT

WORD

PUBLIC

'CODE'

.FARDATA

FAR_DATA

PARA

PRIVATE

'FAR_DATA'

.FARDATA?

FAR_BSS

PARA

PRIVATE

'FAR_BSS'

.DATA

_DATA

WORD

PUBLIC

'DATA'

DGROUP

.CONST

CONST

WORD

PUBLIC

'CONST'

DGROUP

.DATA?

_BSS

WORD

PUBLIC

'BSS'

DGROUP

.STACK

STACK

PARA

STACK

'STACK'

DGROUP*

.CODE

_TEXT

DWORD

PUBLIC

'CODE'

.FARDATA

_DATA

DWORD

PUBLIC

'DATA'

.FARDATA?

_BSS

DWORD

PUBLIC

'BSS'

.DATA

_DATA

DWORD

PUBLIC

'DATA'

.CONST

CONST

DWORD

PUBLIC

'CONST'

.DATA?

_BSS

DWORD

PUBLIC

'BSS'

.STACK

STACK

DWORD

PUBLIC

'STACK'

Flat

* unless the stack type is FARSTACK

Filename: LMAPGAPE.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 419 of 3 Printed: 10/02/00 04:24 PM

Group

420

Programmer’s Guide

Filename: LMAPGAPE.DOC Project:
Template: MSGRIDA1.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 420 of 4 Printed: 10/02/00 04:24 PM

421

Glossary
8087, 80287, or 80387 coprocessor Intel chips that
perform high-speed floating-point and binary
coded decimal number processing. Also called
math coprocessors. Floating-point instructions are
supported directly by the 80486 processor.

A
address The memory location of a data item or
procedure. The expression can represent just the
offset (in which case the default segment is
assumed), or it can be in segment:offset format.
address constant In an assembly-language
instruction, an immediate operand derived by
applying the SEG or OFFSET operator to an
identifier.
address range A range of memory bounded by
two addresses.
addressing modes The various ways a memory
address or device I/O address can be generated.
See “far address,” “near address.”
aggregate types Data types containing more
than one element, such as arrays, structures, and
unions.
animate A debugging feature in which each line
in a running program is highlighted as it executes.
The Animate command from the CodeView
debugger Run menu turns on animation.
API (application programming interface) A set of
system-level routines that can be used in an
application program for tasks such as basic
input/output and file management. In a graphicsoriented operating environment like Microsoft
Windows, high-level support for video graphics
output is part of the Windows graphical API.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 421 of 16 Printed: 10/02/00 04:21 PM

arg In PWB, a function modifier that introduces
an argument or an editing function. The
argument may be of any type and is passed to
the next function as input. For example, the
PWB command Arg textarg Copy passes
the text argument textarg to the function
Copy.
argument A value passed to a procedure or
function. See “parameter.”
array An ordered set of continuous elements of
the same type.
ASCII (American Standard Code for Information
Interchange) A widely used coding scheme
where 1-byte numeric values represent letters,
numbers, symbols, and special characters. There
are 256 possible codes. The first 128 codes are
standardized; the remaining 128 are special
characters defined by the computer
manufacturer.
assembler A program that converts a text file
containing mnemonically coded microprocessor
instructions into the corresponding binary
machine code. MASM is an assembler. See
“compiler.”
assembly language A programming language in
which each line of source code corresponds to a
specific microprocessor instruction. Assembly
language gives the programmer full access to the
computer’s hardware and produces the most
compact, fastest executing code. See “high-level
language.”
assembly mode The mode in which the
CodeView debugger displays the assemblylanguage equivalent of the high-level code being
executed. CodeView obtains the assembly-

422

base address

language code by disassembling the executable
file. See “source mode.”

B
base address The starting address of a stack
frame. Base addresses are usually stored in the
BP register.
base name The portion of the filename that
precedes the extension. For example, SAMPLE
is the base name of the file SAMPLE.ASM.
BCD (binary coded decimal) A way of
representing decimal digits where 4 bits of 1 byte
are a decimal digit, coded as the equivalent
binary number.
binary Referring to the base-2 counting system,
whose digits are 0 and 1.
binary expression A Boolean expression
consisting of two operands joined by a binary
operator and resolving to a binary number.
binary file A file that contains numbers in binary
form (as opposed to ASCII characters
representing the same numbers). For example, a
program file is a binary file.
binary operator A Boolean operator that takes
two arguments. The AND and OR operators in
assembly language are examples of binary
operators.
BIOS (Basic Input/Output System) The software
in a computer’s ROM which forms a hardwareindependent interface between the CPU and its
peripherals (for example, keyboard, disk drives,
video display, I/O ports).
bit Short for binary digit. The basic unit of
binary counting. Logically equivalent to decimal
digits, except that bits can have a value of 0 or 1,
whereas decimal digits can range from 0 through
9.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 422 of 16 Printed: 10/02/00 04:21 PM

breakpoint A user-defined condition that pauses
program execution while debugging. CodeView
can set breakpoints at a specific line of code, for
a specific value of a variable, or for a
combination of these two conditions.
buffer A reserved section of memory that holds
data temporarily, most often during input/output
operations.
byte The smallest unit of measure for computer
memory and data storage. One byte consists of 8
bits and can store one 8-bit character (a letter,
number, punctuation mark, or other symbol). It
can represent unsigned values from 0 to 255 or
signed values between –128 and +127.

C
C calling convention The convention that follows
the C standard for calling a procedure—that is,
pushing arguments onto the stack from right to
left (in reverse order from the way they appear
in the argument list). The C calling convention
permits a variable number of arguments to be
passed.
chaining (to an interrupt) Installing an interrupt
handler that shares control of an interrupt with
other handlers. Control passes from one handler
to the next until a handler breaks the chain by
terminating through an IRET instruction. See
“interrupt handler,” “hooking (an interrupt).”
character string See “string.”
clipboard In PWB, a section of memory that
holds text deleted with the Copy, Ldelete, or
Sdelete functions. Any text attached to the
clipboard deletes text already there. The Paste
function inserts text from the clipboard at the
current cursor position.
.COM The filename extension for executable
files that have a single segment containing both
code and data. Tiny model produces .COM files.

device driver

combine type The segment-declaration specifier
(AT, COMMON, MEMORY, PUBLIC, or
STACK) which tells the linker to combine all
segments of the same type. Segments without a
combine type are private and are placed in
separate physical segments.
compact A memory model with multiple data
segments but only one code segment.
compiler A program that translates source code
into machine language. Usually applied only to
high-level languages such as Basic, FORTRAN,
or C. See “assembler.”
constant A value that does not change during
program execution. A variable, on the other
hand, is a value that can—and usually does—
change. See “symbolic constant.”
constant expression Any expression that
evaluates to a constant. It may include integer
constants, character constants, floating-point
constants, or other constant expressions.

D
debugger A utility program that allows the
programmer to execute a program one line at a
time and view the contents of registers and
memory in order to help locate the source of
bugs or other problems. Examples are CodeView
and Symdeb.
declaration A construct that associates the name
and the attributes of a variable, function, or type.
See “variable declaration.”
default A setting or value that is assumed unless
specified otherwise.
definition A construct that initializes and
allocates storage for a variable, or that specifies
either code labels or the name, formal
parameters, body, and return type of a
procedure. See “type definition.”

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 423 of 16 Printed: 10/02/00 04:21 PM

423

description file A text file used as input for the
NMAKE utility.

424

device driver

device driver A program that transforms I/O
requests into the operations necessary to make a
specific piece of hardware fulfill that request.
Dialog Command window The window at the
bottom of the CodeView screen where dialog
commands can be entered, and previously
entered dialog commands can be reviewed.
direct memory operand In an assembly-language
instruction, a memory operand that refers to the
contents of an explicitly specified memory
location.
directive An instruction that controls the
assembler’s state.

subroutines to be packaged, distributed, and
maintained independently of their callers.
Windows extends the dynamic-link mechanism
to serve as the primary method by which all
system and nonsystem services are obtained. See
“linking.”
dynamic-link library (DLL) A library file that
contains the executable code for a group of
dynamically linked routines.
dynamic-link routine A routine in a dynamic-link
library that can be linked at load time or run
time.

displacement In an assembly-language
instruction, a constant value added to an
effective address. This value often specifies the
starting address of a variable, such as an array or
multidimensional table.

element A single member variable of an array of
like variables.

DLL See “dynamic-link library.”

errorlevel code See “exit code.”

double-click To rapidly press and release a
mouse button twice while pointing the mouse
cursor at an object on the screen.

.EXE The filename extension for a program that
can be loaded and executed by the computer.
The small, compact, medium, large, huge, and
flat models generate .EXE files. See “.COM,”
“tiny.”

double precision A real (floating-point) value
that occupies 8 bytes of memory (MASM type
REAL8). Double-precision values are accurate
to 15 or 16 digits.
doubleword A 4-byte word (MASM type
DWORD).
drag To move the mouse while pointing at an
object and holding down one of the mouse
buttons.
dump To display or print the contents of
memory in a specified memory range.
dynamic linking The resolution of external
references at load time or run time (rather than
link time). Dynamic linking allows the called

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 424 of 16 Printed: 10/02/00 04:21 PM

environment block The section of memory
containing the MS-DOS environment variables.

exit code A code returned by a program to the
operating system. This usually indicates whether
the program ran successfully.
expanded memory Increased memory available
after adding an EMS (Expanded Memory
Specification) board to an 8086 or 80286
machine. Expanded memory can be simulated in
software. The EMS board can increase memory
from 1 megabyte to 8 megabytes by swapping
segments of high-end memory into lower
memory. Applications must be written to the
EMS standard in order to make use of expanded
memory. See “extended memory.”
expression Any valid combination of

forward declaration

mathematical or logical variables, constants,
strings, and operators that yields a single value.

425

extended memory Physical memory above 1
megabyte that can be addressed by 80286–
80486 machines in protected mode. Adding a
memory card adds extended memory. On
80386-based machines, extended memory can be
made to simulate expanded memory by using a
memory-management program.
extension The part of a filename (of up to three
characters) that follows the period (.). An
extension is not required but is usually added to
differentiate similar files. For example, the
source-code file MYPROG.ASM is assembled
into the object file MYPROG.OBJ, which is
linked to produce the executable file
MYPROG.EXE.
external variable A variable declared in one
module and referenced in another module.

F
far address A memory location specified with a
segment value plus an offset from the start of
that segment. Far addresses require 4 bytes—
two for the segment and two for the offset. See
“near address.”
field One of the components of a structure,
union, or record variable.
fixup The linking process that supplies addresses
for procedure calls and variable references.
flags register A register containing information
about the status of the CPU and the results of
the last arithmetic operation performed by the
CPU.
flat A nonsegmented linear address space.
Selectors in flat model can address the entire 4
gigabytes of addressable memory space. See
“segment,” “selector.”
formal parameters The variables that receive
values passed to a function when the function is
called.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 425 of 16 Printed: 10/02/00 04:21 PM

426

forward declaration

forward declaration A function declaration that
establishes the attributes of a symbol so that it
can be referenced before it is defined, or called
from a different source file.
frame The segment, group, or segment register
that specifies the segment portion of an address.

H
handle An arbitrary value that an operating
system supplies to a program (or vice versa) so
that the program can access system resources,
files, peripherals, and so forth, in a controlled
fashion.
handler See “interrupt handler.”

G
General-Protection (GP) fault An error that
occurs in protected mode when a program
accesses invalid memory locations or accesses
valid locations in an invalid way (such as writing
into ROM areas).
gigabyte 1,024 megabytes, or 1,073,741,824
bytes.
global See “visibility.”
global constant A constant available throughout
a module. Symbolic constants defined in the
module-level code are global constants.
global data segment A data segment that is
shared among all instances of a dynamic-link
routine; in other words, a single segment that is
accessible to all processes that call a particular
dynamic-link routine.
global variable A variable that is available
(visible) across multiple modules.
granularity The degree to which library
procedures can be linked as individual blocks of
code. In Microsoft libraries, granularity is at the
object-file level. If a single object file containing
three procedures is added to a library, all three
procedures will be linked with the main program
even if only one of them is actually called.
group A collection of individually defined
segments that have the same segment base
address.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 426 of 16 Printed: 10/02/00 04:21 PM

hexadecimal The base-16 numbering system
whose digits are 0 through F (the letters A
through F represent the decimal numbers 10
through 15). This is often used in computer
programming because it is easily converted to
and from the binary (base-2) numbering system
the computer itself uses.
high-level language A programming language
that expresses operations as mathematical or
logical relationships, which the language’s
compiler then converts into machine code. This
contrasts with assembly language, in which the
program is written directly as a sequence of
explicit microprocessor instructions. Basic, C,
COBOL, and FORTRAN are examples of highlevel languages. See “assembly language,”
“compiler.”
hooking (an interrupt) Replacing an address in
the interrupt vector table with the address of
another interrupt handler. See “interrupt
handler,” “interrupt vector table,” “unhooking
(an interrupt).”
huge A memory model (similar to large model)
with more than one code segment and more than
one data segment. However, individual data
items can be larger than 64K, spanning more
than one segment. See “large.”

I
identifier A name that identifies a register or
memory location.

linked list

IEEE format A standard created by the Institute
of Electrical and Electronics Engineers for
representing floating-point numbers, performing
math with them, and handling
underflow/overflow conditions. The 8087 family
of coprocessors and the emulator package
implement this format.
immediate expression An expression that
evaluates to a number that can be either a
component of an address or the entire address.
immediate operand In an assembly-language
instruction, a constant operand that is specified at
assembly time and stored in the program file as
part of the instruction opcode.
import library A pseudo library that contains
addresses rather than executable code. The linker
reads the addresses from an import library to
resolve references to external dynamic-link
library routines.
include file A text file with the .INC extension
whose contents are inserted into the source-code
file and immediately assembled.
indirect memory operand In an assemblylanguage instruction, a memory operand whose
value is treated as an address that points to the
location of the desired data. See “pointer.”
instruction The unit of binary information that a
CPU decodes and executes. In assembly
language, instruction refers to the mnemonic
(such as LDS or SHL) that the assembler
converts into machine code.
instruction prefix See “prefix.”
interrupt A signal to the processor to halt its
current operation and immediately transfer
control to an interrupt handler. Interrupts are
triggered either by hardware, as when the
keyboard detects a keypress, or by software, as
when a program executes the INT instruction.
See “interrupt handler.”

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 427 of 16 Printed: 10/02/00 04:21 PM

427

interrupt handler A routine that receives
processor control when a specific interrupt
occurs.
interrupt service routine See “interrupt handler.”
interrupt vector An address that points to an
interrupt handler.
interrupt vector table A table maintained by the
operating system. It contains addresses (vectors)
of current interrupt handlers. When an interrupt
occurs, the CPU branches to the address in the
table that corresponds to the interrupt’s number.
See “interrupt handler.”

K
keyword A word with a special, predefined
meaning for the assembler. Keywords cannot be
used as identifiers.
kilobyte (K) 1,024 bytes.

L
label A symbol (identifier) representing the
address of a code label or data objects.
language type The specifier that establishes the
naming and calling conventions for a procedure.
These are BASIC, C, FORTRAN, PASCAL,
STDCALL, and SYSCALL.
large A memory model with more than one
code segment and more than one data segment,
but with no individual data item larger than 64K
(a single segment). See “huge.”
library A file that contains modules of compiled
code. MS-DOS programs use normal run-time
libraries, from which the linker extracts modules
and combines them with other object modules to
create executable program files. Windows-based
programs can use dynamic-link libraries (see),
which the operating system loads and links to
calling programs. See also “import library.”

428

linked list

linked list A data structure in which each entry
includes a pointer to the location of the adjoining
entries.
linking In normal static linking, the process in
which the linker resolves all external references
by searching run-time and user libraries, and then
computes absolute offset addresses for these
references. Static linking results in a single
executable file. In dynamic linking (see), the
operating system, rather than the linker, provides
the addresses after loading the modules into
separate parts of memory.
local constant A constant whose scope is limited
to a procedure or a module.

macro A block of text or instructions that has
been assigned an identifier. When the assembler
sees this identifier in the source code, it
substitutes the related text or instructions and
assembles them.
main module The module containing the point
where program execution begins (the program’s
entry point). See “module.”
math coprocessor See “8087, 80287, or 80387
coprocessor.”
medium A memory model with multiple code
segments but only one data segment.
megabyte 1,024 kilobytes or 1,048,576 bytes.

local variable A variable whose scope is confined
to a particular unit of code, such as module-level
code, or a procedure. See “module-level code.”

member One of the elements of a structure or
union; also called a field.

logical device A symbolic name for a device that
can be mapped to a physical (actual) device.

memory address A number through which a
program can reference a location in memory.

logical line A complete program statement in
source code, including the initial line of code and
any extension lines.

memory map A representation of where in
memory the computer expects to find certain
types of information.

logical segment A memory area in which a
program stores code, data, or stack information.
See “physical segment.”

memory model A convention for specifying the
number and types of code and data segments in
a module. See “tiny,” “small,” “medium,”
“compact,” “large,” “huge,” and “flat.”

low-level input and output routines Run-time
library routines that perform unbuffered,
unformatted input/output operations.

memory operand An operand that specifies a
memory location.

LSB (least-significant bit) The bit lowest in
memory in a binary number.

meta A prefix that modifies the subsequent
PWB function.

M
machine code The binary numbers that a
microprocessor interprets as program
instructions. See “instruction.”

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 428 of 16 Printed: 10/02/00 04:21 PM

mnemonic A word, abbreviation, or acronym
that replaces something too complex to
remember or type easily. For example, ADC is
the mnemonic for the 8086’s add-with-carry
instruction. The assembler converts it into
machine (binary) code, so it is not necessary to
remember or calculate the binary form.

overlay

module A discrete group of statements. Every
program has at least one module (the main
module). In most cases, a module is the same as
a source file.
module-definition file A text file containing
information that the linker uses to create a
Windows-based program.
module-level code Program statements within
any module that are outside procedure
definitions.
MSB (most-significant bit) The bit farthest to the
left in a binary number. It represents 2(n-1) ,
where n is the number of bits in the number.
multitasking operating system An operating
system in which two or more programs,
processes, or threads can execute
simultaneously.

N
naming convention The way the compiler or
assembler alters the name of a routine before
placing it into an object file.
NAN Acronym for “not a number.” Math
coprocessors generate NANs when the result of
an operation cannot be represented in IEEE
format. For example, if two numbers being
multiplied have a product larger than the
maximum value permitted, the coprocessor
returns a NAN instead of the product.
near address A memory location specified by
the offset from the start of the value in a
segment register. A near address requires only 2
bytes. See “far address.”
nonreentrant See “reentrant procedure.”
null character The ASCII character encoded as
the value 0.
null pointer A pointer to nothing, expressed as

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 429 of 16 Printed: 10/02/00 04:21 PM

429

the value 0.

O
.OBJ Default filename extension for an object
file.
object file A file (normally with the extension
.OBJ) produced by assembling source code. It
contains relocatable machine code. The linker
combines object files with run-time and library
code to create an executable file.
offset The number of bytes from the beginning
of a segment to a particular byte within that
segment.
opcode The binary number that represents a
specific microprocessor instruction.
operand A constant or variable value that is
manipulated in an expression or instruction.
operator One or more symbols that specify how
the operand or operands of an expression are
manipulated.
option A variable that modifies the way a
program performs. Options can appear on the
command line, or they can be part of an
initialization file (such as TOOLS.INI). An
option is sometimes called a switch.
output screen The CodeView screen that
displays program output. Choosing the Output
command from the View menu or pressing F4
switches to this screen.
overflow An error that occurs when the value
assigned to a numeric variable is larger than the
allowable limit for that variable’s type.
overlay A program component loaded into
memory from disk only when needed. This
technique reduces the amount of free RAM
needed to run the program.

430

parameter

P
parameter The name given in a procedure
definition to a variable that is passed to the
procedure. See “argument.”

passing by reference Transferring the address of
an argument to a procedure. This allows the
procedure to modify the argument’s value.
passing by value Transferring the value (rather
than the address) of an argument to a procedure.
This prevents the procedure from changing the
argument’s original value.
physical segment The true memory address of a
segment, referenced through a segment register.
pointer A variable containing the address of
another variable. See “indirect memory
operand.”
precedence The relative position of an operator
in the hierarchy that determines the order in
which expression elements are evaluated.
preemptive Having the power to take precedence
over another event.
prefix A keyword (LOCK, REP, REPE,
REPNE, REPNZ, or REPZ) that modifies the
behavior of an instruction. MASM 6.1 ensures
the prefix is compatible with the instruction.
private Data items and routines local to the
module in which they are defined. They cannot
be accessed outside that module. See “public.”
privilege level A hardware-supported feature of
the 80286–80486 processors that allows the
programmer to specify the exclusivity of a
program or process. Programs running at lownumbered privilege levels can access data or
resources at higher-numbered privilege levels,
but the reverse is not true. This feature reduces
the possibility that malfunctioning code will
corrupt data or crash the operating system.
privileged mode The term applied to privilege
level 0. This privilege level should be used only
by a protected-mode operating system. Special
privileged instructions are enabled by .286P,
.386P, and .486P. Privileged mode should not be
confused with protected mode.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 430 of 16 Printed: 10/02/00 04:21 PM

ROM (read-only memory)

procedure call An expression that invokes a
procedure and passes actual arguments (if any)
to the procedure.
procedure definition A definition that specifies a
procedure’s name, its formal parameters, the
declarations and statements that define what it
does, and (optionally) its return type and storage
class.
procedure prototype A procedure declaration
that includes a list of the names and types of
formal parameters following the procedure name.
process Generally, any executing program or
code unit. This term implies that the program or
unit is one of a group of processes executing
independently.
Program Segment Prefix (PSP) A 256-byte data
structure at the base of the memory block
allocated to a transient program. It contains data
and addresses supplied by MS-DOS that a
program can read during execution.
protected mode The 80286–80486 operating
mode that permits multiple processes to run and
not interfere with each other. This feature should
not be confused with privileged mode.

431

from. RAM data is volatile; it is usually lost
when the computer is turned off. Programs are
loaded into and executed from RAM. See
“ROM.”
real mode The normal operating mode of the
8086 family of processors. Addresses correspond
to physical (not mapped) memory locations, and
there is no mechanism to keep one application
from accessing or modifying the code or data of
another. See “protected mode.”
record A MASM variable that consists of a
sequence of bit values.
reentrant procedure A procedure that can be
safely interrupted during execution and restarted
from its beginning in response to a call from a
preemptive process. After servicing the
preemptive call, the procedure continues
execution at the point at which it was interrupted.
register operand In an assembly-language
instruction, an operand that is stored in the
register specified by the instruction.
register window The optional CodeView window
in which the CPU registers and the flag register
bits are displayed.

public Data items and procedures that can be
accessed outside the module in which they are
defined. See “private.”

registers Memory locations in the processor that
temporarily store data, addresses, and processor
flags.

regular expression A text expression that
specifies a pattern of text to be matched (as
opposed to matching specific characters).

qualifiedtype A user-defined type consisting of
an existing MASM type (intrinsic, structure,
union, or record), or a previously defined
TYPEDEF type, together with its language or
distance attributes.

R
radix The base of a number system. The default
radix for MASM and CodeView is 10.
RAM (random-access memory) Computer
memory that can be both written to and read

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 431 of 16 Printed: 10/02/00 04:21 PM

relocatable Not having an absolute address. The
assembler does not know where the label, data,
or code will be located in memory, so it
generates a fixup record. The linker provides the
address.
return value The value returned by a function.
ROM (read-only memory) Computer memory that
can only be read from and cannot be modified.
ROM data is permanent; it is not lost when the

432

routine

machine is turned off. A computer’s ROM often
contains BIOS routines and parts of the
operating system. See “RAM.”

after they scroll off the top. This mode is
required with computers that are not IBM
compatible.

routine A generic term for a procedure or
function.

selector A value that indirectly references a
segment address. A protected-mode operating
system, such as Windows, assigns selector
values to programs, which use them as segment
addresses. If a program attempts to use an
unassigned selector, it triggers a GeneralProtection fault (see).

run-time dynamic linking The act of establishing
a link when a process is running. See “dynamic
linking.”
run-time error A math or logic error that can be
detected only when the program runs. Examples
of run-time errors are dividing by a variable
whose value is zero or calling a DLL function
that doesn’t exist.

S
scope The range of statements over which a
variable or constant can be referenced by name.
See “global constant,” “global variable,” “local
constant,” “local variable.”
screen swapping A screen-exchange method that
uses buffers to store the debugging and output
screens. When you request the other screen, the
two buffers are exchanged. This method is
slower than flipping (the other screen-exchange
method), but it works with most adapters and
most types of programs.
scroll bars The bars that appear at the right side
and bottom of a window and some list boxes.
Dragging the mouse on the scroll bars allows
scrolling through the contents of a window or
text box.

shared memory A memory segment that can be
accessed simultaneously by more than one
process.
shell escape A method of gaining access to the
operating system without leaving CodeView or
losing the current debugging context. It is
possible to execute MS-DOS commands, then
return to the debugger.
sign extended The process of widening an
integer (for example, going from a byte to a
word, or a word to a doubleword) while retaining
its correct value and sign.
signed integer An integer value that uses the
most-significant bit to represent the value’s sign.
If the bit is one, the number is negative; if zero,
the number is positive. See “two’s complement,”
“unsigned integer,” “MSB.”
single precision A real (floating-point) value that
occupies 4 bytes of memory. Single-precision
values are accurate to six or seven decimal
places.

segment A section of memory, limited to 64K
with 16-bit segments or 4 gigabytes with 32-bit
segments, containing code or data. Also refers to
the starting address of that memory area.

single-tasking environment An environment in
which only one program runs at a time. MSDOS is a single-tasking environment.

sequential mode The mode in CodeView in
which no windows are available. Input and
output scroll down the screen, and the old output
scrolls off the top of the screen when the screen
is full. You cannot examine previous commands

source file A text file containing symbols that
define the program.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 432 of 16 Printed: 10/02/00 04:21 PM

small A memory model with only one code
segment and only one data segment.

source mode The mode in which CodeView

status bar

displays the assembly-language source code that

433

represents the machine code currently being
executed.
stack An area of memory in which data items
are consecutively stored and removed on a lastin, first-out basis. A stack can be used to pass
parameters to procedures.
stack frame The portion of a stack containing a
particular procedure’s local variables and
parameters.
stack probe A short routine called on entry to a
function to verify that there is enough room in
the program stack to allocate local variables
required by the function.
stack switching Changing the stack pointers to
point to another stack area.
stack trace A symbolic representation of the
functions that are being executed to reach the
current instruction address. As a function is
executed, the function address and any function
arguments are pushed on the stack. Therefore,
tracing the stack shows the active functions and
their arguments.
standard error The device to which a program
can send error messages. The display is normally
standard error.
standard input The device from which a
program reads its input. The keyboard is
normally standard input.
standard output The device to which a program
can send its output. The display is normally
standard output.
statement A combination of labels, data
declarations, directives, or instructions that the
assembler can convert into machine code.
status bar See “linking.”

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 433 of 16 Printed: 10/02/00 04:21 PM

434

static linking

static linking The line at the bottom of the PWB
or CodeView screen. The status bar displays text
position, keyboard status, current context of
execution, and other program information.

text box In PWB, a box where you type
information needed to carry out a command. A
text box appears within a dialog box. The text
box may be blank or contain a default entry.

STDCALL A calling convention that uses caller
stack cleanup if the VARARG keyword is
specified. Otherwise the called routine must
clean up the stack.

tiny Memory model with a single segment for
both code and data. This limits the total program
size to 64K. Tiny programs have the filename
extension .COM.

string A contiguous sequence of characters
identified with a symbolic name.

toggle A function key or menu selection that
turns a feature off if it is on, or on if it is off.
Used as a verb, “toggle” means to reverse the
status of a feature.

string literal A string of characters and escape
sequences delimited by single quotation marks
(' ') or double quotation marks (" ").
structure A set of variables that may be of
different types, grouped under a single name.
structure member One of the elements of a
structure. Also called a field.
switch See “option.”
symbol A name that identifies a memory
location (usually for data).
symbolic constant A constant represented by a
symbol rather than the constant itself. Symbolic
constants are defined with EQU statements.
They make a program easier to read and modify.

TOOLS.INI A file containing initialization
information for many of the Microsoft utilities,
including PWB.
two’s complement A form of base-2 notation in
which negative numbers are formed by inverting
the bit values of the equivalent positive number
and adding 1 to the result.
type A description of a set of values and a valid
set of operations on items of that type. For
example, a variable of type BYTE can have any
of a set of integer values within the range
specified for the type on a particular machine.

SYSCALL A language type for a procedure. Its
conventions are identical to C’s, except no
underscore is prefixed to the name.

type checking An operation in which the
assembler verifies that the operands of an
operator are valid or that the actual arguments in
a function call are of the same types as the
function definition’s parameters.

type definition The storage format and attributes
for a data unit.

tag The name assigned to a structure, union, or
enumeration type.
task See “process.”
text Ordinary, readable characters, including the
uppercase and lowercase letters of the alphabet,
the numerals 0 through 9, and punctuation
marks.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 434 of 16 Printed: 10/02/00 04:21 PM

U
unary expression An expression consisting of a
single operand preceded or followed by a unary
operator.
unary operator An operator that acts on a single
operand, such as NOT.

word

underflow An error condition that occurs when a
calculation produces a result too small for the
computer to represent.
unhooking (an interrupt) The act of removing
your interrupt handler and restoring the original
vector. See “hooking (an interrupt).”
union A set of values (in fields) of different
types that occupy the same storage space.
unresolved external See “unresolved reference.”
unresolved reference A reference to a global or
external variable or function that cannot be
found, either in the modules being linked or in
the libraries linked with those modules. An
unresolved reference causes a fatal link error.
unsigned integer An integer in which the mostsignificant bit serves as part of the number,
rather than as an indication of sign. For example,
an unsigned byte integer can have a value from 0
to 255. A signed byte integer, which reserves its
eighth bit for the sign, can range from -127 to
+128. See “signed integer,” “MSB.”
user-defined type A data type defined by the
user. It is usually a structure, union, record, or
pointer.

V
variable declaration A statement that initializes
and allocates storage for a variable of a given
type.
virtual disk A portion of the computer’s random
access memory reserved for use as a simulated
disk drive. Also called an electronic disk or RAM
disk. Unless saved to a physical disk, the
contents of a virtual disk are lost when the
computer is turned off.
virtual memory Memory space allocated on a
disk, rather than in RAM. Virtual memory allows
large data structures that would not fit in
conventional memory, at the expense of slow
access.

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 435 of 16 Printed: 10/02/00 04:21 PM

435

visibility The characteristic of a variable or
function that describes the parts of the program
in which it can be accessed. An item has global
visibility if it can be referenced in every source
file constituting the program. Otherwise, it has
local visibility.

W
watch window The window in CodeView that
displays watch statements and their values. A
variable or expression is watchable only while
execution is occurring in the section of the
program (context) in which the item is defined.
window A discrete area of the screen in PWB or
CodeView used to display part of a file or to
enter statements.
window commands Commands that work only in
CodeView’s window mode. Window commands
consist of function keys, mouse selections, CTRL
and ALT key combinations, and selections from
pop-up menus.
window mode The mode in which CodeView
displays separate windows, which can change
independently. CodeView has mouse support
and a wide variety of window commands in
window mode.
word A data unit containing 16 bits (2 bytes). It
can store values from 0 to 65,535 (or -32,768 to
+32,767).

436

Filename: LMAPGGLO.DOC Project: Glossary
Template: GLOSS.DOT Author: Ruth L Silverio Last Saved By: Ruth L Silverio
Revision #: 2 Page: 436 of 16 Printed: 10/02/00 04:21 PM

435

Index
! (literal-character operator) 235
!= (not equal operator) 178
” (double quotation marks) 109, 353
$ (current address operator) 368
% (expansion operator) 235, 248, 357
& (substitution operator) 238, 372
&& (logical AND operator) 178
’ (single quotation mark) 109, 353
( ) (parentheses) 106
+ (plus operator) 63, 66, 352, 370
. (dot operator) 126
. (structure-member operator) 64, 67, 352, 370
.186 directive 38
.286 directive 38
.286P directive 38
.287 directive 38
.386 directive
FLAT, with 26, 36
processor mode, specifying 38, 336
segment mode, setting 46, 68
.386P directive 38
.387 directive 38
.486 directive
FLAT, with 36
processor mode, specifying 38
segment mode, setting 46, 68
.486P directive 38
.8087 directive 38
: (colon) 22, 352, 354
: (segment-override operator) 50, 59– 60, 64
:: (double colon) 197, 215, 352– 354
; (semicolon) 21
;; (double semicolon) 227
< (less than operator) 178
< > (angle brackets) See Angle brackets
= = (equal operator) 178
> (greater than operator) 178
? (question mark initializer)
array elements 109
described 368
variables 87
@ (at sign) 10
@@: (anonymous label) 170
[ ] (brackets) 107
[ ] (index operator) 63
\ (backslash character), MASM code 22
\ (line-continuation character) 121
{} (curly braces) 121, 131

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 435 of 1 Printed: 10/02/00 04:20 PM

|| (logical OR operator) 178
32-bit programming 335
80186 processor 3
80188 processor 3
80286 processor 3
80287 math coprocessor 3, 135
80386 processor 3, 335
80387 math coprocessor 3, 135
80486 processor 3, 135
8086-based processors 2– 3
8087 math coprocessor 3, 135
8088 processor 3

A
AAD instruction 160
AAM instruction 160
ABS operand 220
Accessing data with pointers See Pointer variables
ADC instruction 92– 94
ADD instruction 92– 94
ADDR operator 197
Addresses
displacement of 65
dynamic 79
effective 65
errors in 54
far 57, 74, 80
near 57, 80
physical 7
registers, loading into 80
relocatable 57
segmented 7– 8, 53
Addressing
direct registers, used in 62– 63
indirect registers, used in 65, 68
scaling operands 70
specifying 60
Aliases 87, 369
ALIGN directive 3
Align types 45
See also individual entries
.ALPHA directive 47
AND instruction 27, 99, 100
Angle brackets (< >)
default parameters 230
epilogues 202
FOR loops 242

436

Index
FORC loops 244

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 436 of 2 Printed: 10/02/00 04:20 PM

Index
Angle brackets (< >) (continued)
macro text delimiters 234
prologues 202
records 131
structures and unions 121
Anonymous label (@@) 170
API (Application Programming Interface) 257
Architecture, segmented 2, 5
Architecture, unsegmented 5
Arguments
errors 196
macro 252
mixed-language programs, passing in 314
qualifiedtypes, with 16
stack, on 182
Arrays
accessing elements in 105
declaring 105
defined 105
defining 15
DUP, declaring with 106, 124
instructions for processing 110
length of 108
multiple-line declarations for 105
number of bytes in 108
referencing 108, 316
size of elements 108
with DUP operator See DUP operator
with SIZEOF operator See SIZEOF operator
with TYPE operator See TYPE operator
ASCIIZ 267
Assembly
actions during 23
conditional See Conditional assembly
INCLUDE files 212
language
book list xviii
mixed-language programs 312
listing files See Listing files
two-pass 358
Assembly pointers See Conditional assembly
Assembly-time variables 233
ASSUME directive
.MODEL, generated with 37
code segments, changing 357
enhancements 344
general-purpose registers 77
segment registers, setting 49– 55, 58– 59, 357
AT address combine type 46
/AT command-line option, ML 36
At sign (@) 10

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 437 of 3 Printed: 10/02/00 04:20 PM

437

438

Index

OPTION directive 25

Backslash character (\) 22
Backus-Naur Form See BNF grammar
Base Pointer (BP) register 73
Basic calling conventions 308– 310
Basic/MASM programs 328– 332
Binary Coded Decimals
calculating with 156– 160
defining 156
instructions for 156– 160
packed 158
unpacked 159– 160
Bits
mask 99– 102
rotating 100
shifting 100
BNF grammar 16, 379– 380
BOUND instruction 108, 204
BP (Base Pointer) register 73
Brackets ([ ]) 107
.BREAK directive 173, 176
BSF instruction 100
BSR instruction 100
BYTE
align type 45
directive 86

C
C calling convention 309
C++/MASM programs 322– 323
C/MASM programs 315– 321
CALL instruction 180
Calling conventions 309
Basic 308– 310
directives, specifying 37
FORTRAN 308– 310
(list) 308
mixed-language programming 308– 309
Pascal 310
STDCALL 311
SYSCALL 308– 311
CARRY? flag as operand 178
Case sensitivity
enforcing 348
macro functions, predefined 245
MASM statements 22
radix specifiers 11
reserved words 9, 407
specifying
command-line options, in 25
language type 348

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 438 of 4 Printed: 10/02/00 04:20 PM

Index
Case sensitivity (continued)
symbols, predefined 10
CASEMAP
ALL argument, OPTION directive 25
NONE argument, OPTION directive 25
NOTPUBLIC argument, OPTION directive 25
CATSTR directive 245– 247
CATSTR, compared with TEXTEQU directive See
TEXTEQU directive
@CatStr predefined string function 245– 247
CBW instruction 90
CDQ instruction 90
CLC instruction 104
Cleaning the stack 185
CLI instruction 5, 209
Client program 257, 266
CMC instruction 104
CMP instruction 166
CMPS instruction 110– 114, 353
CMPSB instruction 114
.CODE directive 33, 40– 42
Code segment See Segments, code
Code, near or far 57
@CodeSize predefined symbol 40
CodeView for Windows 264
Combine types
(list) 46
See also individual entries
.COM files
relocatable segment expression, lacking 62
starting address 56
tiny model, using 36, 46– 47
COMM directive 16, 211, 217– 218
Command-line driver, ML xvi
Command-line options See ML command-line options
COMMENT directive 22
Comments
extended lines, in 346
macros, in 227
source code 21– 22
COMMON combine type 46
Communal variables 217
Compact model See Memory models, compact
Compatibility, MASM 5.1 See MASM 5.1 compatibility
Conditional assembly
assembly behavior, changing 23
conditions, testing for 28
directives 28
pointers 83, 187
Conditional-error directives (table) 29
Conditional jumps 164– 170
Conditions, testing for conditional assembly See Conditional
assembly

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 439 of 5 Printed: 10/02/00 04:20 PM

Constants
defined 11
expressions 12
immediate 61
integer 11– 12
size 363
size of 12
symbolic 12
.CONST directive 33, 39– 40
.CONTINUE directive 173, 176
Coprocessors
architecture 140– 144
control registers 156
data format in registers 140
defined 135
described 3, 139
instructions
arithmetic 148– 150
data transfer 146
described 146
(list) 414
overview 141
program control 151– 155
memory access 145
operand formats
classical stack 141
memory 142
overview 141
register 143
register-pop 144
specifying 37, 140
status word register 156
steps for using 145
/Cp command-line option, ML 10, 245
@Cpu predefined symbol 254
Curly braces ({})
records 131
structures and unions 121
Current address operator ($) 368
@CurSeg predefined symbol 39, 219
CWD instruction 90
CWDE instruction 90
/Cx command-line option, ML 158

D
DAA instruction 162
DAS instruction 162
.DATA directive 33, 39– 40
.DATA? directive 33, 39– 40
@data predefined symbol 39
Data segment See Segments, data
@DataSize predefined symbol 39, 83

439

440

Index

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 440 of 6 Printed: 10/02/00 04:20 PM

Index
Data types
arrays See Arrays
attributes for 15
Binary Coded Decimals 159
defined 14
defining 87
directives 14
floating-point 136
initializers, as 14
integers, allocating memory for 85– 86
new features, MASM 6.1 344
qualifiedtypes 15, 214
real 14, 136
signed 14, 86
strings See Strings
structures 117
unions 117
user-defined 15
Data, near or far 57, 58
Data-sharing methods 211
Data-sharing methods, multiple-module programs See
Multiple-module programs
Date, system 11
DB directive 86
DD directive 86
DEC instruction 92– 94
DF directive 86
DGROUP group name
.MODEL, defined by 34, 39, 51
DS registers, initializing to 56
MS-DOS programs, for 41– 42
near data, accessing 57– 58
segment 35– 37, 51– 52, 57
Direct memory operands
loading offset of 82
overview 60– 64
Directives
.286P 38
.287 38
.386 See .386 directive
.386P 38
.387 38
.486P 38
.8087 38
.ALPHA 47
ALIGN 3
.BREAK 173, 176
BYTE 86
CATSTR 245– 247
.CODE 33, 40– 42
COMM 16, 211, 217– 218
COMMENT 22
Conditional assembly 28
Conditional error 29, 358

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 441 of 7 Printed: 10/02/00 04:20 PM

441

442

Index

Directives (continued)
.CONST 33, 39– 40
.CONTINUE 173, 176
.DATA 33, 39– 40
.DATA? 33, 39– 40
Data declarations, for 87
Data types, for 14
Data-sharing See EXTERN directive
DB 86
DD 86, 136
Decision 171
DF 86
.DOSSEG 47
DQ 86, 136
DT 86, 136
DW 86
DWORD 86
ECHO 236
.ELSE 171
ELSE 28
.ELSEIF 171
ELSEIF 28
ELSEIF1 29, 358
ELSEIF2 29, 358
END 33, 56
.ENDIF 171
ENDIF 28
ENDM 227– 239
ENDP 180– 181, 206
ENDS 44
.ENDW 173
EQU 12, 369
.ERR 30
.ERR1 30, 358
.ERR2 30, 358
.ERRB 30, 231
.ERRDEF 30
.ERRDIF 30
.ERRE 30
.ERRIDN 29
.ERRNB 29, 231
.ERRNDEF 29
.ERRNZ 29
EVEN 3
.EXIT 33, 41– 43
EXITM 248
EXTERN See EXTERN directive
EXTERNDEF See EXTERNDEF directive
FARDATA 33, 39– 40
.FARDATA 39– 40
.FARDATA? 33, 39– 40
Floating-point 136
FOR 242– 243, 249
FORC 244

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 442 of 8 Printed: 10/02/00 04:20 PM

Index
Directives (continued)
FWORD 86
GROUP 51– 52
.IF 171
IF 28– 29
IF1 29, 358
IF2 29, 358
IFB 29, 231
IFDEF 29, 359
IFDIF 29
IFE 29
IFIDN 29
IFNB 29, 231
IFNDEF 29, 359
INCLUDE 212
INCLUDELIB 222
INSTR 245– 246
INVOKE See INVOKE directive
LABEL 16
LOCAL 188– 191, 232
loop-generating 173
.MODEL See .MODEL directive
.MSFLOAT 361
Naming conventions 37
.NO87 38, 349
obsolete 361
OPTION See OPTION directive
ORG 56
POPCONTEXT 255, 349
PROC 180– 184, 193, 206, 312
PUBLIC 185, 211, 220
PUSHCONTEXT 255, 349
QWORD 86
.RADIX 11
REAL4 136– 137
REAL8 136– 137
REAL10 136– 137
RECORD 130– 131
Renamed since MASM 5.1 350
.REPEAT 173– 177
REPEAT 240
SBYTE 86
SDWORD 86
SEGMENT 44– 47
Segment order, controlling 47
.SEQ 47
SIZESTR 245– 246
STACK See STACK directive
.STARTUP See .STARTUP directive
STARTUP See .STARTUP directive
STRUCT 118– 129
SUBSTR 245– 246
SWORD 86
TBYTE 86, 159

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 443 of 9 Printed: 10/02/00 04:20 PM

443

444

Index

Directives (continued)
TEXTEQU See TEXTEQU directive
UNION 118– 119, 122, 125– 129
.UNTIL 173
.UNTILCXZ 173
.WHILE 173– 177
WHILE 241
WORD 86
Directives: 36– 38, 46
Displacement 66
Distance attributes 15
DIV instruction 97– 98
Division 97, 102
DLLs
client program 257, 266
data segment 265– 269
defined 257, 266
example 267– 268
extension name 266
heap 261– 262, 265– 267
IMPLIB utility 258
initialization 261– 262, 268– 269
loading 258– 260
programming requirements 260– 261, 267
prologue and epilogue 264– 267
stacks in 46, 264– 267
summary 266
termination 262– 264, 270
Document conventions vii
DOS See MS-DOS
.DOSSEG directive 47
Dot (.) operator See Structure-member operator
DOTNAME argument, OPTION directive 25
Double colon (::) 197, 215
Double quotation marks (”) 109
Double semicolon (;;) 227
Doublewords 86
DQ directive 86
DT directive 86
DUP operator
arrays, with 106, 124
record variables, with 131
structures and unions, with 121
DW directive 86
DWORD
align type 45
directive 86
Dynamic -link libraries See DLLs

E
ECHO directive 236
.ELSE directive 171

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 444 of 10 Printed: 10/02/00 04:20 PM

ELSE directive 28

Index
.ELSEIF directive 171
ELSEIF directive 28
ELSEIF1 directive 358
ELSEIF2 directive 29, 358
EMULATOR argument, OPTION directive 27, 157
Emulator libraries 155– 156
END directive 33, 56
.ENDIF directive 171
ENDIF directive 28
ENDM directive 227– 239
ENDP directive 180– 181, 206
ENDS directive 44
.ENDW directive 173
ENTER instruction 183
Environment
target 4
variables
INCLUDE 213
LIB 222
returning values of 10
/EP command-line option, ML 342
EPILOGUE argument, OPTION directive 26, 201– 203
Epilogue code
defined 198
macros 201– 202, 264– 265
PROC statement, specifying arguments in 185
procedures, with 26
RET instruction 357
standard 199
user-defined 201
EQ operator 365
EQU directive 12, 369
Equal directive (=) 12
Equates, predefined See Predefined symbols
.ERR directive 29
.ERR1 directive 30, 358
.ERR2 directive 30, 358
.ERRB directive 29, 231
.ERRDEF directive 29
.ERRDIF directive 29
.ERRE directive 29
.ERRIDN directive 29
.ERRNB directive 29, 231
.ERRNDEF directive 29
.ERRNZ directive 29
Error detection 196
ERROR operand 49–50
Errors, argument passing 196
ESC instruction 360
EVEN directive 3
Executable (.EXE) files, controlling size of 223
Exit codes, Windows operating system 263
.EXIT directive 33, 41– 43
EXITM directive 248

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 445 of 11 Printed: 10/02/00 04:20 PM

445

446

Index

Expansion operator (%) 235– 236, 248, 357
Explicit loading 258
Exponent bias 139
EXPORT operand 185
EXPORTS statement 261, 270
EXPR16 argument, OPTION directive 13, 26, 361, 373
EXPR32 argument, OPTION directive 13, 26, 373
Expression operators 178
Expressions
assembly-time evaluation 23
constant 12
loop conditions, evaluating 179
OPTION M510 behavior 364, 373
order of evaluation 14
size 366, 373
word size 13, 26
Extension, filename 266
EXTERN directive
data-sharing 211
executable file size, limiting 223
module-specific 220
overview 16
positioning 218
procedure prototypes, declaring 193
External declarations 216– 218
External variables 217, 369
EXTERNDEF directive
data-sharing 211
overview 16
positioning 218
procedure prototypes, declaring 193
symbols, declaring 214– 215

F
Far addresses, invoking 57, 74, 80– 81, 197
Far code 57
Far data 58– 60
.FARDATA directive 33, 39– 40
.FARDATA? directive 39– 40
FAR operator 169, 185
Far pointer 74, 80– 81
FARSTACK operand
example 35
grouping 34
in Windows-based programs 266
MS-DOS program, initializing 43
special cases, setting for 37
Farwords 86
FCOM instruction 153
Fields, statements in 21– 22
Files
.COM

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 446 of 12 Printed: 10/02/00 04:20 PM

relocatable segment expression, lacking 62

Index
Files (continued)
.COM (continued)
starting address 56
tiny model, using 36, 46– 47
executable 24
include 212– 213, 348
line numbers 11
naming 11
Flags
CARRY? 178
operands, as 178
OVERFLOW? 178
PARITY? 178
SIGN? 178
stack, saving on 73
ZERO? 178
Flags register See Registers, flags
Flat model See Memory models, flat
FLAT operand 46, 49– 50
FLD1 instruction 147
FLDZ instruction 147
Floating-point
calculations 3
constants
decimal form 137
encoded hexadecimal format 137
syntax for defining 136
emulation 157– 158
IEEE format 139
instructions
arithmetic 148– 149
controlling 26
data transfer 147
not emulated (list) 158
program control 152– 153, 156
operations 146
values
double precision 139
single precision 139
variables
IEEE format 138
Microsoft binary format 138
.MSFLOAT format 138
ranges 136
FOR directive 242– 243, 249
FORC directive 244
FORCEFRAME operand 200– 201
FORTRAN calling convention 308– 310
FORTRAN/MASM programs 323– 326
/Fpi command-line option, ML 26, 157
Frame 62
FS register 17
FTST instruction 153

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 447 of 13 Printed: 10/02/00 04:20 PM

447

448

Index

Full segment definitions
described 32
segment registers, initializing 54– 56
using 44– 51
Full segment defintions See .STARTUP directive
FWORD directive 86
FXCH instruction 144

G
Global variables 211
GROUP directive 51– 52
Groups
defined 51
DGROUP 51
SEG operator, returned by 62
GS register 17

H
H2INC 318
Heap space 261– 262, 265– 267
HEAPSIZE statement 261, 271
Help, online See Microsoft Advisor
HIGH operator 356
HIGHWORD operator 346
Huge model See Memory models, huge

I
/I command-line option, ML 213
Identifiers
ABS, using 220
naming restrictions 9, 346, 353, 357, 368
OPTION DOTNAME 373
OPTION NOKEYWORD 376
IDIV instruction 97– 98
IEEE format 139
.IF directive 171
IF directive 28– 29
IF1 directive 29, 358
IF2 directive 29, 358
IFB directive 29, 231
IFDEF directive 29, 359
IFDIF directive 29
IFE directive 29
IFIDN directive 29
IFNB directive 29, 231
IFNDEF directive 29, 359
Immediate operands 60– 62
IMPLIB utility 258
Implicit loading 258
Import libraries 258

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 448 of 14 Printed: 10/02/00 04:20 PM

Index
IMPORTS statement 266
IMUL instruction 95– 96
IN instruction 5
INC instruction 92– 94
INCLUDE directive 212
INCLUDE environment variable 213
Include files
assembling 213
nested 213
overview 212, 348
INCLUDELIB directive 222
Index operator ([ ]) 63
Indirect memory operands 60, 64– 70
Indirect procedure calls See INVOKE directive
Initializers
allocating 87
directives for 15
multiple-line 346
Instance 261, 266
INSTR directive 245– 246
@InStr predefined string function 245– 246
Instruction Pointer (IP) register 20, 57, 161
Instructions
ADC 92– 94
ADD 92– 94
AND 26, 99– 100
arithmetic 378
bit-test 354
BOUND 108, 204
BSF 100
BSR 100
CALL 180
CBW 90
CDQ 90
CLC 104
CLI 5, 209
CMC 104
CMP 166
CMPS 110– 114, 353
CMPSB 114
conditional-jump 165– 167
coprocessor 377
CWD 90
CWDE 90
DAA 162
DAS 162
DEC 92– 94
default segments, requiring 49
DIV 97– 98
encodings, changes to 377– 378
ENTER 183
ESC 360
FCOM 153
FLD1 147

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 449 of 15 Printed: 10/02/00 04:20 PM

Instructions (continued)
FLDZ 147
floating-point See Floating-point instructions
FTST 153
FXCH 144
IDIV 97– 98
IMUL 95– 96
IN 5
INC 92– 94
INT 204– 205
INTO 207
JCXZ 170– 173
JECXZ 170– 173
JMP 49, 162
JO 165
jump 165–167, 170, 173
LAHF 73
LDS 81
LEA 82, 104
LEAVE 183
LES 81
(list) 412
LOCK 353, 363
LODS 110– 115, 353
logical 99– 102
LOOP 172
LOOPE 172
LOOPNE 172
LOOPNZ 172
LOOPZ 172
MOV 49, 82, 89
MOVS 110– 113, 353
MOVSX 92
MOVZX 92
MUL 95– 96
NOP 377
NOT 99– 100
obsolete 360
operands for 60
OR 26, 99– 100, 168
OUT 5
POP 49, 71
POPA 74
POPAD 74
POPF 73
POPFD 73
privileged 2, 38
PUSH 49, 71
PUSHA 74
PUSHAD 74
PUSHF 73
PUSHFD 73
RCL 101– 104
RCR 101– 104

449

450

Index

Instructions (continued)
REP 110– 112, 363
REPE 110– 112, 363
REPNE 110– 112, 353, 363
REPNZ 110– 112, 353, 363
REPZ 110– 112, 363
RET 378
RETF 181, 378
RETN 181, 378
ROL 101– 104
ROR 101– 104
SAL 101– 104
SAR 101– 104
SBB 92– 94
SCAS 110– 115, 353
SHL 101– 104
SHR 101– 104
STC 104
STI 5, 209
STOS 110– 113, 353
SUB 92– 94
TEST 167– 168
timing xvii, 399– 400
XCHG 90
XLAT 116
XLATB 116
XOR 26, 99– 100
Integers
adding 92– 94
allocating memory for 85– 86
Binary Coded Decimal (BCD) 159
bit operations on 99
constants, defining 11– 12
dividing 97– 98
exchanging 90
hexadecimal 12
initializing 87
memory format 86
moving 89
multiplying 95– 96
operations with 88
popping off stack 71
pushing onto stack 71
radix specifiers for 11
sign-extending 90
signed 86
size of 86
stack 71
subtracting 92– 94
translating 116
types, defining 14, 86
value range 86
@Interface predefined symbol 37
Interrupt vector 205

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 450 of 16 Printed: 10/02/00 04:20 PM

Index
Interrupt-enable flag 205
Interrupts
CLI instruction 209
handlers 206– 207
INT instruction 204– 205
MS-DOS 204, 285
operation 206
overview 204
redefining 207
STI instruction 209
vector table 205
INTO instruction 207
INVOKE directive
actions 194
ADDR, invoking 197
arguments, widening 196
error detection 196
far addresses, invoking 197
generated code, checking 198
indirect procedure calls 198
mixed-language programs 312– 313
procedures, calling 193– 197, 216
type conversions 194– 195

J
JCXZ instruction 170– 173
JECXZ instruction 170– 173
JMP instruction 49, 162
JO instruction 165
Jumps
anonymous 170
automatic 169
conditional
bit status 167
comparisons 166
extending 26, 169
flag status 165– 166
instructions (list) 165– 167
overview 164
zero value 168
directives for 171
extension, automatic 26, 169
instructions 165– 167
optimization, automatic 162
overview 161
unconditional
indirect operands 163
jump tables 163
overview 162

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 451 of 17 Printed: 10/02/00 04:20 PM

L
LABEL directive 16

451

452

Index

Labels
anonymous 170
code
length 346
OPTION M510 behavior 363
OPTION NOSCOPED 375
procedures, in 357
referencing 352
size 346
visibility 354
LAHF instruction 73
LANGUAGE
BASIC argument, OPTION directive 26
C argument, OPTION directive 26
FORTRAN argument, OPTION directive 26
PASCAL argument, OPTION directive 26
STDCALL argument, OPTION directive 26
SYSCALL argument, OPTION directive 26
LANGUAGE argument, OPTION directive 193
Language attributes
.MODEL directive, with 34, 37
OPATTR operator 253
OPTION directive, with 26
Large model See Memory models, large
LDS instruction 81
LEA instruction 82, 104
LEAVE instruction 183
Length of strings See LENGTHOF operator
LENGTH operator 356– 357, 364
LENGTHOF operator
number of items, returning 110, 124, 132, 346
structures, defining 108
unions, with 125
LES instruction 81
Libraries
C run-time 271
emulator 155– 156
overview 221
source files, specifying in 222
LIBRARY statement 270
Line-continuation character (\) 121
LINK, command-line options See individual entries
Linkage specification 322– 323
Linking
actions during 24, 45
segment order in 48
Listing files
code generated 399
command-line options 397– 399
error messages 400
examples 401
generating 397
PWB options 397– 399
reading 399, 405

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 452 of 18 Printed: 10/02/00 04:20 PM

Index
Listing files (continued)
symbols used in (list) 400
tables in 405– 406
Literal-character operator (!) 235
LJMP argument, OPTION directive 27
LOADDS operand 200– 201
Loading local address variables See Local variables
Loading, actions during 24
Local addresses, loading See Local variables
LOCAL directive 188– 191, 232
Local variables
creating 188
loading addresses of 82
procedures, in 188
LOCK instruction 353, 363
LODS instruction 110– 115, 353
Logical AND 178
Logical instruction 99– 100
Logical line 22
Lookup tables 241
LOOP instruction 172
LOOPE instruction 172
LOOPNE instruction 172
LOOPNZ instruction 172
Loops
conditions
expression evaluation 179
precedence 179
PTR operator in 178
relational operators for (list) 178
signed operands 178
writing 178
controlling 176
directives
.REPEAT 173– 177
.WHILE 173– 177
instructions (list) 172
macros
FOR 242– 243, 249
FORC 244
REPEAT 240
WHILE 241
LOOPZ instruction 172
LOW operator 356
LOWWORD operator 346, 366
LROFFSET operator 344

M
M510 argument, OPTION directive
compatibility with MASM 5.1 26, 353– 370
expression word size, setting 13
structures, with 119

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 453 of 19 Printed: 10/02/00 04:20 PM

453

454

Index

Macros
arguments
commas 352, 372
quotation marks 353
testing 29, 252
variable 242, 249
calling 227
checking argument types with 253
comments (;;) 227
expansion 23
functions
defined 248
epilogues 201
EXITM 248
prologues 201
returning values 248
local symbols in 232
loops
FOR 242– 243, 249
FORC 244
REPEAT 240
WHILE 242– 243
MASM 5.1 behavior 25, 356, 372
nested 251
new features 351
operators
behavior in macro functions 251
expansion (%) 235– 236, 248, 357
(list) 234
literal-character (!) 235
substitution (&) 238, 352, 372
OPTION OLDMACROS 372
parameters
default values 230
procedure parameters, compared to 234
required 229
substitution 238
passing arguments to 228, 235
predefined string functions 11
procedures
defined 226
functions, compared to 228
recursive 255
redefining 251
text
defined 226
forward referencing 356
numeric equates, compared to 234
OPTION M510 behavior 370
syntax 226
VARARG keyword 242, 249, 351
writing 227
Mask
defined 99

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 454 of 20 Printed: 10/02/00 04:20 PM

Index
Mask (continued)
logic instructions, with 102
record operators, with 133
MASK operator 133
MASM 5.1 compatibility
address fixups 26
macro behavior 25, 356, 372
OPTION directive, specifying 25
overview xvi
structures 25
updating code 353– 360
MASM utility xvi, 342
Math coprocessor See Coprocessors
Medium model See Memory models, medium
Memory
access 64
allocation 24
virtual 5
MEMORY combine type 46
Memory models
attributes 35
compact 36
described 34
determining 10
far code segments 40
far data segments 40
flat 36, 58, 336
huge 36
large 36
medium 36
model-independent code 83
near code segments 40
small 36
specifying in PROC statement 185
tiny 36, 46– 47
Memory-resident programs See TSRs
Microsoft Advisor xiii, 342
Minus operator (– ) 64
Mixed-language programming
argument passing 314
assembly procedures 312
Basic/MASM programs 328– 332
C prototypes, converting with H2INC 318
C++/MASM programs 322– 323
C/MASM programs 315– 321
calling conventions
Basic 308– 310
FORTRAN 308– 310
(list) 308
Pascal 310
STDCALL 311
SYSCALL 308– 311
column-major order 315

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 455 of 21 Printed: 10/02/00 04:20 PM

455

456

Index

Mixed-language programming (continued)
compatible data types
Basic (list) 328
C (list) 315
FORTRAN (list) 323
external data 314
FORTRAN/MASM programs 323– 326
initialization code 313, 321
INVOKE, using 312– 313
naming conventions 308– 309
overview 307
register preservation 314
row-major order 315
ML command-line options
/AT 36
/Cp 10, 245
described xvi
/EP 342
/Fpi 26, 157
/I 213
listing options (list) 397
overview xvi
/X 213
/Zm 62, 119
/Zp 119
Mode, real, protected See Real mode; Protected mode
.MODEL directive
attributes 34– 35
DGROUP 51
language types, specifying 26, 308
memory model, defining 35– 36
mode default 46
overview 34
positioning 46
simplified segment directives 33
@Model predefined symbol 35, 83
Module-definition file
described 270
statements
EXPORTS 261, 270
HEAPSIZE 261, 271
IMPORTS 266
LIBRARY 270
STUB 266
Module-specific EXTERN directive See EXTERN directive
MOV instruction 49, 82, 89
MOVS instruction 110– 113, 353
MOVSX instruction 92
MOVZX instruction 92
MS-DOS interrupts 204, 285
MS-DOS operating system 2– 6
MUL instruction 95– 96
Multiple-module programs
alternatives to include files 219

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 456 of 22 Printed: 10/02/00 04:20 PM

Index
Multiple-module programs (continued)
COMM, using 217
data-sharing methods 211
declaring symbols public and external 214
EXTERN with library routines 223
external declarations, positioning 218
EXTERNDEF, using 214
include files 212– 213
libraries 221– 222
modules 212
PROTO, using 216
PUBLIC and EXTERN, using 220
sharing symbols with include files 212
Multiplex interrupt 291, 304
Multiplication
instructions 95
shift operations 102

N
Naming conventions
directives 37
(list) 308
mixed-language programming 308– 309
Naming restrictions 9
Naming restrictions, identifers See Identifiers
NE operator 365
Near address 57, 80
NEAR operator 169, 185
NEARSTACK operand
ASSUME statement 54
default stack type 37, 42
described 35
New features, MASM 6.1 xiv– xv, 342– 351
NMAKE 270
.NO87 directive 38, 349
NODOTNAME argument, OPTION directive 25
NOEMULATOR argument, OPTION directive 27
NOKEYWORD argument, OPTION directive 9, 27, 353, 376
NOLJMP argument, OPTION directive 27, 170
NOM510 argument, OPTION directive 25
NONUNIQUE operand 118, 126
NOOLDMACROS argument, OPTION directive 26
NOOLDSTRUCTS argument, OPTION directive 26
NOP instruction 377
NOREADONLY argument, OPTION directive 27
NOSCOPED argument, OPTION directive 26, 362, 375
NOSIGNEXTEND argument, OPTION directive 27, 378
NOT instruction 99– 100
NOTHING operand 49– 50
Number of items with LENGTHOF operator See LENGTHOF
operator
Numeric equates, compared to text macros 234

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 457 of 23 Printed: 10/02/00 04:20 PM

457

458

Index

O
OFFSET
FLAT argument, OPTION directive 27
GROUP argument, OPTION directive 27
SEGMENT argument, OPTION directive 27, 62
OFFSET operator 61, 82, 356, 374
Offsets
accessing data with 74
addresses 7
described 5– 7
determining 23– 24, 360, 374
fixups for 26
OLDMACROS argument, OPTION directive 25, 239, 361,
372
OLDSTRUCTS argument, OPTION directive
MASM 5.1 compatibility 25, 361, 370– 372
structures, with 119, 126
Online help See Microsoft Advisor
OPATTR operator 252– 253
Operands
ABS 220
direct memory 60– 64
EXPORT 185
FAR 15
FARSTACK See FARSTACK operand
FLAT 46, 49– 50
FORCEFRAME 244
immediate 60– 62
indirect memory 60, 64– 70
NEAR 15
PRIVATE READONLY 44– 45
registers 61
size 66, 355
USE16 44– 46
USE32 44– 46
Operating systems
(list) 4
.MODEL, specifying with 34
multitasking 6
types See MS-DOS, Windows operating systems
Operators
ADDR 197
current address ($) 368
dot (.) 126, 352, 370
EQ 365
expansion (%) 235– 236, 248, 357
expressions, in 12– 13
FAR 169, 185
HIGH 356
HIGHWORD 346
index ([ ]) 63
instructions, compared to 13
LENGTH 356– 357, 364

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 458 of 24 Printed: 10/02/00 04:20 PM

LENGTHOF 346
Operators (continued)
LOW 356
LOWWORD 346, 366
LROFFSET 344
macro 251
MASK 133
minus (– ) 64
NE 365
NEAR 169, 185
OFFSET 61, 82, See OFFSET operator
OPATTR 252– 254
plus (+) 63, 66
precedence 14
PTR See PTR operator
PTR, example See PTR operator
relational 357, 365
relational (list) 178
SEG 50, 62, 363
segment-override (:) 59, 64
SHORT 169
SIZE 364– 365
size See PTR operator
SIZEOF 86, 346
structure-member (.) 64– 67, 126, 352, 370
substitution (&) 238
.TYPE 252, 360
TYPE 86
WIDTH 133
OPTION directive
CASEMAP 25
described 23
DOTNAME 25, 361, 373
emulation mode 157
EMULATOR 26, 157
EPILOGUE 26, 201– 203
EXPR16 OPTION directive 13, 26, 361, 373
EXPR32 OPTION directive 13, 26, 373
LANGUAGE 26, 193
language types, specifying 308
list of arguments for 25
LJMP 26
M510 See M510 argument, OPTION directive
NODOTNAME 25
NOEMULATOR 26
NOKEYWORD See NOKEYWORD argument, OPTION
directive
NOLJMP 27, 170
NOM510 25
NOOLDMACROS 26
NOOLDSTRUCTS 26
NOREADONLY 27
NOSCOPED 26, 362, 375
NOSIGNEXTEND 27, 378

Index
OPTION directive (continued)
OFFSET 26, 62, 362, 374– 375
OLDMACROS 25, 237
OLDSTRUCTS See OLDSTRUCTS argument, OPTION
directive
PROC 185, 375
procedure use 26
PROLOGUE 26, 201– 203
READONLY 26
SCOPED 25
SETIF2 25, 29– 30
using 25, 361
OR instruction 27, 99– 100, 168
ORG directive 56
OUT instruction 5
OVERFLOW? flag as operand 178

P
PAGE align type 45
PARA align type 45
Parentheses [( )] 106
PARITY? flag as operand 178
Pascal convention 310
Physical line 22
Plus operator (+) 66, 352, 370
Pointer variables 74– 78
Pointers
accessing data with 74
arguments, as 80
copying 79
far 74, 80– 81
initializing 78
location 74
operations 78
TYPEDEF, defined with 15, 75– 78
types, to 15
Pointers and conditional Assembly See Conditional assembly
Pointers defined by TYPEDEF See TYPEDEF directive
POP instruction 49, 71
POPA instruction 74
POPAD instruction 74
POPCONTEXT directive 255, 349
POPF instruction 73
POPFD instruction 73
Positioning
EXTERN directive See EXTERN directive
EXTERNDEF directive See EXTERNDEF directive
Precedence operators 14
Predefined equates See Predefined symbols
Predefined functions for macros 11
Predefined string functions
@CatStr 245– 247

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 459 of 25 Printed: 10/02/00 04:20 PM

@InStr 245– 246

459

460

Index

Predefined string functions (continued)
@SizeStr 245– 246
@SubStr 245– 246
Predefined symbols 39, 83
@Codesize 40
@Cpu 254
@CurSeg 39, 219
@Data 39
@DataSize 39, 83
@Interface 37
(list) 10, 409
@Model 35, 83
@stack 37
@Wordsize 39
case sensitivity 9– 10
new to MASM 6.1 (list) 343
PRIVATE operand 185
Privilege levels 5
Problems, reporting xx
PROC
EXPORT argument, OPTION directive 25
PRIVATE argument, OPTION directive 25, 362
PUBLIC argument, OPTION directive 25, 185
PROC directive 180– 184, 193, 206, 312
PROC statements with visibility See also Visibility
PROC with RET instruction See RET instruction
Procedure prototypes
declaring See EXTERNDEF directive
defined with See PROTO directive
defined with PROTO directive See PROTO directive
writing See PROTO directive
Procedures
arguments
far pointers 197
near addresses 197
passing 182
pointers 80
type conversions 195, 196
CALL instruction 180
calling See INVOKE directive
calls
indirect 198
optimizing 181
defining 180
epilogues 26
EXTERNDEF directive 214–215
See also EXTERNDEF directive,
include files 214
INVOKE directive 193– 197, 216
libraries 221
local variables 188– 192
See also Local variables
Macro See Macros, procedures
new features 347

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 460 of 26 Printed: 10/02/00 04:20 PM

Index
Procedures (continued)
OPTION PROC 375
overview 180
parameters
declaring 184– 186
variable numbers of 186– 188, 194
PROC attributes, specifying 185
prologues 26
PROTO directive 193, 214, 216
See also PROTO directive
prototypes, writing 193
RET instruction 180
RETF instruction 181
RETN instruction 181
syntax description 184
VARARG keyword 186– 188, 194
visibility 25, 375
Processors
See also Real mode; Protected mode
8086-based 2– 3
.MODEL directive 37
modes, determining 10
target 2
timing xvii, 399– 400
Product assistance xx
Program Segment Prefix (PSP) 56
Programming, MASM 6.1 practices 352
Programs
exiting 41
mixed-language 307
starting 41
PROLOGUE argument, OPTION directive 25, 201– 203
Prologue code
arguments, specifying 185
code labels in 357
defined 198
macros for 201– 203, 264– 265
standard 199
user-defined 26, 201
Protected mode
described 2– 7, 335
flat model 335
read-only segments 45
PROTO directive
include files 211, 214– 216
procedure prototypes, defined with 193
procedure prototypes, writing 312
Prototypes
procedure
directives for 193
overview 193
qualifiedtypes, defined with 15
PTR operator
example 92

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 461 of 27 Printed: 10/02/00 04:20 PM

461

462

Index

PTR operator (continued)
OPTION M510 behavior 365
pointer to type, as 15
signed number, specifying 178
size 66, 88
TYPEDEF, used with 75
PUBLIC combine type 45
PUBLIC directive 185, 211, 220
PUSH instruction 49, 71
PUSHA instruction 74
PUSHAD instruction 74
PUSHCONTEXT directive 255, 349
PUSHF instruction 73
PUSHFD instruction 73

Q
Quadwords 86
Qualifiedtypes
BNF grammar 16
defined 15
pointers, defining 75– 76
prototypes, as 15
rules for use 15– 16
Question mark initializer ( ? )
array elements 109
described 368
variables 87
Quotation marks (' or ") 109
QWORD directive 86

R
.RADIX directive 11
Radix specifiers
(list) 11
OPTION M510 behavior 367
RCL instruction 101– 104
RCR instruction 101– 104
Read-only code 27
READONLY argument, OPTION directive 26
READONLY operand 44– 45
Real mode 2, 4, 7
Real numbers See Floating-point
REAL4 directive 136– 137
REAL8 directive 136– 137
REAL10 directive 136– 137
RECORD directive 130– 131
Records
defined 129
field ranges 354
LENGTH operator 357
operators 133– 134

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 462 of 28 Printed: 10/02/00 04:20 PM

RECORD syntax 130– 131

Index
Records with SIZEOF operator See SIZEOF operator
Records with TYPE operator See TYPE operator
Recursive macros 255
Register operands 61
Registers
16-bit 16– 17, 67
32-bit 335
base 65– 70
coprocessor 140
copying pairs of 82
division (table) 98
Eflags 20
extended 17
flags 20
FS 17
general purpose 19
GS 17
index 65– 69
indirect addressing 65
indirect operands 67– 68
initializing 44
Instruction Pointer (IP) 20, 57, 161
(list) 409
loading addresses into 80
mixed 16-bit, 32-bit 70
pointers as 77
scaling 67– 69
segment See Segment registers
Stack Pointer (SP) 19
Stack Segment (SS) 73
stacks, saving on 74
types, defined with ASSUME 77
Relational operators (list) 178
Relocatable
addresses 57
expressions 62, 65
REP instruction 110– 112, 363
REPE instruction 110– 112, 363
Repeat blocks 239
.REPEAT directive 173
REPEAT directive 240
REPNE instruction 110– 112, 353, 363
REPNZ instruction 110– 112, 353, 363
Reporting problems xx
REPZ instruction 110– 112
Reserved words
described 8, 26
(list) 407
OPTION M510 behavior 362
OPTION NOKEYWORD 376
RET instruction
epilogue code, generating 200, 378
instruction encodings, changes to 357
PROC, with 180

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 463 of 29 Printed: 10/02/00 04:20 PM

463

464

Index

RETF instruction 181, 378
RETN instruction 181, 378
ROL instruction 101– 104
ROM-BIOS interrupts See Interrupts
ROR instruction 101– 104
Rotate instructions 101
Routines, interrupt 206

S
SAL instruction 101– 104
SAR instruction 101– 104
SBB instruction 92– 94
SBYTE directive 86
Scaling factor 107
Scaling index registers 67– 69
SCAS instruction 110– 112, 115, 353
Scope within visibility See also Visibility
SCOPED argument, OPTION directive 26
SDWORD directive 86
SEG operator 49, 62, 363
SEGMENT
FLAT argument, OPTION directive 27
USE16 argument, OPTION directive 27
USE32 argument, OPTION directive 27
Segment arithmetic 7
SEGMENT directive 44– 47
Segment mode, setting See .386 directive; .486 directive
Segment registers
32-bit 335
assigning 59, 62
ASSUME directive 49– 55, 58– 59, 357
changing 57
default 60, 64
described 18
FS 18
GS 18
initializing 43, 54– 57
MS-DOS, under 24, 43
near code 57
restoring 59
segment-override operator (:) 50, 59– 60, 64
Segment registers
initializing See STARTUP directive
setting See STACK directive
Segment selectors 5
Segment-override operator (:) 50, 59– 60, 64
Segmented architecture 2, 5
Segments
32-bit 36, 335
accessing data 74
aligning 44– 45
class types 44, 47– 48

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 464 of 30 Printed: 10/02/00 04:20 PM

Index
Segments (continued)
code
creating 40
far 40
memory model support for 36
near 40
combining 40, 44– 46
current 10
data
creating 39
default 49, 54– 55, 59
far 40
memory model support for 36
near 39
defined 31
described 5– 7
determining order of 47– 48
determining position of 23– 24
determining size of 44
fixups for 26
full segment definitions, defining 32
groups, defining 51
initializing 55
location of 6
naming 40
ordering with the linker 48
protection 6
READONLY 45
simplified segment directives 37– 42
size, determining 10
types 44
USE16 44
USE32 44
values 55
word size, setting 46
Selector 335
Semicolon (;), comments 21
.SEQ directive 47
SETIF2 argument, OPTION directive 25, 29– 30
Shift instructions 100
SHL instruction 101– 104
SHORT operator 169
SHR instruction 101– 104
Sign-extending integers 90
SIGN? flag as operand 178
Signed data 14, 91
Signed numbers, specifying See PTR operator
Significand 139
Simplified segment directives
code segments 41
code, starting and ending 42
data segments 40
described 32
language convention 36

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 465 of 31 Printed: 10/02/00 04:20 PM

465

466

Index

Simplified segment directives (continued)
memory model 35
.MODEL, defining with 34
operating system 35
processor 38
segment registers, initializing 54– 56
stack 39
stack distance 37
using 33
Single quotation mark (') 109
Size attribute, segments
FLAT 46
USE16 46
USE32 46
Size mismatch 355
Size of strings See SIZEOF operator
SIZE operator 364, 365
@SizeStr predefined string function 245– 246
SIZEOF operator
arrays, with 108
described 346
records, with 132
strings, with 110
structures, with 124
types 86
unions, with 125
SIZESTR directive 245– 246
Small model See Memory models, small
Source code, statements in 21
SP (Stack Pointer) register 19, 71– 73
SS (Stack Segment) register 73
STACK combine type 45
.STACK directive
described 33
segment registers, setting 56
Stack distance 37
Stack frame 73, 200, 264– 265
Stack Pointer (SP) register 19
@stack predefined symbol 37
Stack Segment (SS) register 73
Stacks
cleaning 185
creating 38
described 71
distance 37
far 10
FARSTACK 35, 37
in DLLs 264– 267
local variables on 188– 191
near 10
NEARSTACK 33, 35– 37
operations with 72– 74
operators 71
passing arguments 182

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 466 of 32 Printed: 10/02/00 04:20 PM

Index
Stacks (continued)
pointer 71– 73
POP instructions 71
PUSH instructions 71
saving flags 73
saving registers 74
segment register 18
separate 46
trace 264
.STARTUP directive
described 33
initializing segments 54– 56
program, starting 41– 42
segment address 37
Statements
case sensitivity 22
syntax 21
Status flags, saving 73
STC instruction 104
STDCALL calling convention 311, 336
STI instruction 5, 209
STOS instruction 110– 113, 353
Strings
declaring 109
defined 105
defining 15
initializing 109
instructions
processing, for 110
requirements (table) 112, 353
length of 110
multiple-line declarations for 109
overview 111
predefined functions for macros 11
See also Predefined string functions
size of 110
type of 110
STRUCT directive 118– 129
Structure-member operator (.) 64– 67, 126, 352, 370
Structures
alignment of fields 118– 119
array initializers 122
arrays 124
compatibility with MASM 5.1 25, 118
current address operator ($) 368
default field values 122
defined 117
fields
accessing 64, 67, 371
initializing 118
naming 119, 352, 372
initializers, as 123
MASM 5.1 behavior 25, 355, 370
memory allocation 117

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 467 of 33 Printed: 10/02/00 04:20 PM

467

468

Index

Structures (continued)
nested 128– 129
new features 345
operators 124
OPTION M510 behavior 366
OPTION OLDSTRUCTS 370
redeclaration 124, 355
referencing fields in 126
steps for using 118
string initializers 122, 368
syntax
types 118
variables 121
Structures with LENGTHOF operator See LENGTHOF
operator
Structures with SIZEOF operator See SIZEOF operator
Structures with TYPE operator See TYPE operator
STUB statement 266
SUB instruction 92– 94
Substitution operator (&) 238, 372
SUBSTR directive 245– 246
@SubStr predefined string function 245– 246
SWORD directive 86
Symbol table, listing files 405
Symbols
declaring public and external 214, 220
external 369
naming 346, 368
predefined 9– 11
Symbols, declaring by EXTERNDEF directive See
EXTERNDEF directive
Syntax, MASM 6.1 statements 21
SYSCALL calling convention 308– 311
System date 11
System time 11

T
Tables, lookup 241
Target environment 4
TBYTE directive 86, 159
Terminate-and-Stay-Resident programs See TSRs
TEST instruction 167– 168
Testing for zero 168
Text delimiters See Angle brackets
Text macros See Macros, text
TEXTEQU directive
aliases 369
CATSTR, compared with 247
syntax 226
Time, system 11
Timing (cycle/second) xvii, 399– 400
Tiny model See Memory models, tiny

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 468 of 34 Printed: 10/02/00 04:20 PM

Trap flag 205

Index
TSRs
active
described 275
interrupt handlers in 275
MS-DOS functions, calling 285
MS-DOS functions, interrupting 286, 302
deinstalling 292, 305
described 273
errors, trapping 288– 289
examples
ALARM.ASM 279– 280, 284
SNAP.ASM 293– 305
existing data, preserving 290, 303
hardware events, auditing 275– 276, 299
interrupt handlers 275
monitoring
Critical Error flag 287
system status 277, 300
MS-DOS internal stacks (lists) 286
multiplex interrupt 290, 304
passive 274
Type conversions See INVOKE directive
Type of strings See TYPE operator
TYPE operator
and OPATTR 252– 253
arrays, with 108
compatibility 360, 365
records, with 132
string, with 110
structures, with 124
types 86
unions, with 125
TYPEDEF directive
aliases, created by 87, 137
BNF, from 380
data types, defining 87
indirect operands, defining 163
pointers, defined by 15, 75– 78
procedure declarations 193
procedure prototypes 193
qualifiedtypes 16
TYPEDEF, used with PTR operator See PTR operator
Types, data See Data types

memory allocation 117
Unions (continued)
nested 128– 129
operators 125
referencing fields in 126
steps for using 118
strings as initializers 122
types 118
variables 121, 127
Unpacked BCD numbers 160
Unsegmented architecture 5
Unsigned data 91
.UNTIL directive 173
.UNTILCXZ directive 173
USE16 operand 44– 46
USE32 operand 44– 46
USES in PROC statement 184
Utilities
IMPLIB 258
MASM 342
ML xvi

Unconditional jumps 162
UNION directive 118– 119, 122, 125– 129
Unions
arrays as initializers 122
arrays of 124
defined 117
fields 119, 127– 129

WDEB386 debugger 264
WEP (Windows Exit Procedure) 263– 264, 270
.WHILE directive 173
WHILE directive 241
WIDTH operator 133
Windows operating system
API 257, 262

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 469 of 35 Printed: 10/02/00 04:20 PM

V
VARARG keyword
macros, used in 242, 249, 351
procedures, used with 186– 188, 194
Variables
assembly-time 233
communal 217
environment 10, 213, 222
external 217, 369
floating-point 136– 138
global 211
initializing 87
integers, allocating memory for 85– 86
local address, loading 82
naming restrictions 9
Virtual memory 5
Virtual-86 mode 2, 335
Visibility
PROC statement 25, 185
scope, within 9

469

470

Index

applications 258, 261
DLLs 261
Windows operating system (continued)
exit codes 263
MS-DOS, compared 4
programming for 4
protected mode 2, 6
SDK 268
task header 265, 269
Windows NT 3– 5
WORD align type 45
WORD directive 86
Word size
default 13, 363, 373
expressions, in 13, 26
@WordSize predefined symbol 39
Words, reserved See Reserved words

X
XCHG instruction 90
/X command-line option, ML 213
XLAT instruction 116
XLATB instruction 116
XOR instruction 27, 99– 100

Z
ZERO? flag as operand 178
/Zm command-line option, ML 62, 119
/Zp command-line option, ML 119

Filename: LMAPGINX.DOC Project:
Template: INDEX.DOT Author: Samuel G. Dawson Last Saved By: Ruth L Silverio
Revision #: 20 Page: 470 of 36 Printed: 10/02/00 04:20 PM

Documentation Feedback — Microsoft ® Macro Assembler Version 6.1
Please help us improve our documentation. When you have used MASM 6.1 for a while, please complete and return this form. Use the
back of the form for additional suggestions and comments. Suggestions and comments become the property of Microsoft Corporation.

Rate each component of the document set:
Rate each from 1 (never use) to 5 (often use).
1 2 3 4 5

Getting Started

1 2 3 4 5

Programmer’s Guide

1 2 3 4 5

Environment and Tools

1 2 3 4 5

Reference

1 2 3 4 5

Online Help

List what stands out most about each component:
List the one or two things that you like, dislike, or both. Or, use
the space to comment on the rating you gave the component in
the previous section.
Getting Started______________________________________
__________________________________________________
__________________________________________________
__________________________________________________
Programmer’s Guide _________________________________
__________________________________________________
__________________________________________________
__________________________________________________
Environment and Tools _______________________________
__________________________________________________
__________________________________________________
__________________________________________________
Reference __________________________________________
__________________________________________________
__________________________________________________
__________________________________________________
Online Help_________________________________________
__________________________________________________
__________________________________________________
__________________________________________________

Filename: LMAPGDFD.DOC Project: Documentation Feedback Form MASM 6.1
Template: FEEDBACK.DOT Author: Ruth L Silverio Last Saved By: Mike Eddy
Revision #: 5 Page: 1 of 1 Printed: 10/02/00 04:21 PM

What one thing would you like to see added to or removed from
each component?
Getting Started _____________________________________
_________________________________________________
_________________________________________________
_________________________________________________
Programmer’s Guide _________________________________
_________________________________________________
_________________________________________________
_________________________________________________
Environment and Tools _______________________________
_________________________________________________
_________________________________________________
_________________________________________________
Reference__________________________________________
_________________________________________________
_________________________________________________
_________________________________________________
Online Help ________________________________________
_________________________________________________
_________________________________________________
_________________________________________________
How many years of programming experience do you have?
With assembly language
With other programming languages (including application
macro languages)

Name _____________________________________________
Address ___________________________________________
City/State/Zip _______________________________________
Home Phone ( _____ ) ________________________________
Work Phone ( _____ ) ________________________________
May we contact you for additional information about your comments?

Yes

Additional comments:

......................................................................................................................Fold......................................................................................................................

Place stamp
here.
Post Office
will not deliver
without proper
postage.

Microsoft Corporation
Languages — MASM 6.1
One Microsoft Way
Redmond WA 98052-9953

Filename: LMAPGDFD.DOC Project: Documentation Feedback Form MASM 6.1
Template: FEEDBACK.DOT Author: Ruth L Silverio Last Saved By: Mike Eddy
Revision #: 5 Page: 2 of 2 Printed: 10/02/00 04:21 PM

Source Exif Data:

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.4
Linearized                      : No
Modify Date                     : 2003:09:27 01:22:37+02:00
Create Date                     : 2003:09:27 01:22:12+02:00
Page Count                      : 520
Mod Date                        : 2003:09:27 01:22:37+02:00
Creation Date                   : 2003:09:27 01:22:12+02:00
Metadata Date                   : 2003:09:27 01:22:37+02:00
Has XFA                         : No

EXIF Metadata provided by EXIF.tools

Programmer's Guide

Programmer's%20Guide

Navigation menu

Versions of this User Manual:

Views

Navigation