DSP56800E and DSP56800EX - Reference Manual

DSP56800E and
DSP56800EX
Reference Manual
Digital Signal Controller
Cores
DSP56800ERM
Rev. 3
09/2011
freescale.com
Contents
About This Book
Audience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii
Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii
Suggested Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxviii
Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix
Definitions, Acronyms, and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx
Chapter 1
Introduction
1.1
1.2
1.3
1.4
Key Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Architectural Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example DSP56800EX Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction to Digital Signal Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1-1
1-3
1-4
1-5
Chapter 2
Core Architecture Overview
2.1
Extending DSP56800E Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1
2.2
Extending DSP56800 Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1
2.3
Core Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
2.4
Dual Harvard Memory Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5
2.5
System Architecture and Peripheral Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6
2.5.1
Core Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7
2.5.2
Address Buses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-8
2.5.3
Data Buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-8
2.5.4
Data Arithmetic Logic Unit (ALU) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
2.5.5
Address Generation Unit (AGU) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9
2.5.6
Program Controller and Hardware Looping Unit . . . . . . . . . . . . . . . . . . . . . . . . . . 2-10
2.5.7
Bit-Manipulation Unit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11
2.5.8
Enhanced On-Chip Emulation (Enhanced OnCE) Unit . . . . . . . . . . . . . . . . . . . . . 2-11
2.6
Blocks Outside the Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11
2.6.1
Program Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11
2.6.2
Data Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-11
2.6.3
Bootstrap Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-12
2.6.4
External Bus Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-12
Chapter 3
Data Types and Addressing Modes
3.1
Core Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1
Freescale Semiconductor
Table of Contents
iii
3.2
Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-5
3.2.1
Data Formats. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
3.2.1.1
Signed Integer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
3.2.1.2
Unsigned Integer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
3.2.1.3
Signed Fractional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6
3.2.1.4
Unsigned Fractional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
3.2.2
Understanding Fractional and Integer Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
3.3
Memory Access Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8
3.3.1
Move Instruction Syntax. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8
3.3.1.1
Ordering Source and Destination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.3.1.2
Memory Space Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.3.1.3
Specifying Data Size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.3.2
Instructions That Access Data Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.3.2.1
Signed and Unsigned Moves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
3.3.2.2
Moving Words from Memory to a Register . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
3.3.2.3
Accessing Peripheral Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
3.3.3
Instructions That Access Program Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
3.3.4
Instructions with an Operand in Data Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
3.3.5
Parallel Moves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
3.3.5.1
Single Parallel Move. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
3.3.5.2
Dual Parallel Read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
3.4
Data Alignment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
3.4.1
Data Alignment in Accumulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
3.4.2
Data Alignment in Data Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14
3.4.3
Data Alignment in 24-Bit AGU and Control Registers . . . . . . . . . . . . . . . . . . . . . 3-14
3.4.4
Data Alignment in 16-Bit AGU and Control Registers . . . . . . . . . . . . . . . . . . . . . 3-15
3.4.5
Data Alignment in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
3.4.5.1
Byte and Word Addresses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
3.4.5.2
Byte Variable Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
3.4.5.3
Word Variable Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
3.4.5.4
Long-Word Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
3.5
Memory Access and Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
3.5.1
Word and Byte Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
3.5.2
Accessing Word Values Using Word Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
3.5.3
Accessing Long-Word Values Using Word Pointers . . . . . . . . . . . . . . . . . . . . . . . 3-19
3.5.4
Accessing Byte Values Using Word Pointers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
3.5.5
Accessing Byte Values Using Byte Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
3.6
Addressing Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.6.1
Addressing Mode Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
3.6.2
Register-Direct Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
3.6.3
Address-Register-Indirect Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
3.6.3.1
No Update: (Rn) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-29
3.6.3.2
Post-Increment: (Rn)+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-30
3.6.3.3
Post-Decrement: (Rn)– . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-31
3.6.3.4
Post-Update by Offset N: (Rn)+N, (R3)+N3 . . . . . . . . . . . . . . . . . . . . . . . . . . 3-32
3.6.3.5
Index by Offset N: (Rn+N). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-33
3.6.3.6
Index by 3-Bit Displacement: (RRR+x), (SP–x) . . . . . . . . . . . . . . . . . . . . . . . 3-34
iv
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
3.6.3.7
3.6.3.8
3.6.3.9
3.6.4
3.6.4.1
3.6.4.2
3.6.4.3
3.6.4.4
3.6.4.5
3.6.4.6
3.6.5
3.6.5.1
3.6.5.2
3.6.5.3
3.6.5.4
3.6.6
3.6.7
Index by 6-Bit Displacement: (SP–xx) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Index by 16-Bit Displacement: (Rn+xxxx) . . . . . . . . . . . . . . . . . . . . . . . . . . .
Index by 24-Bit Displacement: (Rn+xxxxxx) . . . . . . . . . . . . . . . . . . . . . . . . .
Immediate Address Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4-Bit Immediate Data: #x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5-Bit Immediate Data: #xx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6-Bit Immediate Data: #xx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7-Bit Immediate Data: #xx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16-Bit Immediate Data: #xxxx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32-Bit Immediate Data: #xxxxxxxx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Absolute Address Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Absolute Short Address: aa. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
I/O Short Address: <<pp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16-Bit Absolute Address: xxxx. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24-Bit Absolute Address: xxxxxx. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Implicit Address Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bit-Reverse Address Mode (DSP56800EX Core only) . . . . . . . . . . . . . . . . . . . . .
3-35
3-36
3-37
3-38
3-38
3-38
3-38
3-39
3-39
3-41
3-41
3-42
3-43
3-44
3-45
3-45
3-45
Chapter 4
Instruction Set Introduction
4.1
Instruction Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1
4.1.1
Multiplication Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2
4.1.2
Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
4.1.3
Shifting Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6
4.1.4
Logical Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7
4.1.5
AGU Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7
4.1.6
Bit-Manipulation Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
4.1.7
Looping Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
4.1.8
Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10
4.1.9
Program Control Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10
4.2
Instruction Aliases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12
4.2.1
The ANDC, EORC, ORC, and NOTC Aliases. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12
4.2.2
Instruction Operand Remapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
4.2.2.1
Duplicate Operand Remapping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
4.2.2.2
Addressing Mode Remapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
4.3
Delayed Flow Control Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14
4.3.1
Using Delayed Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14
4.3.2
Delayed Instruction Restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
4.3.3
Delayed Instructions and Interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
4.4
Instruction Set Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
4.4.1
Using the Instruction Summary Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
4.4.2
Register Field Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
4.4.3
Immediate Value Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
4.4.4
Instruction Summary Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
4.4.5
Parallel Move Summary Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-49
4.5
Register-to-Register Moves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-51
Freescale Semiconductor
Table of Contents
v
Chapter 5
Data Arithmetic Logic Unit
5.1
Data ALU Overview and Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2
5.1.1
Data Registers (X0, Y1, Y0). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3
5.1.2
Accumulator Registers (A, B, C, D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-4
5.1.3
Multiply-Accumulator (MAC) and Logic Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5
5.1.4
Single-Bit Accumulator Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5
5.1.5
Arithmetic and Logical Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6
5.1.6
Data Limiter and MAC Output Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6
5.2
Accessing the Accumulator Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6
5.2.1
Accessing an Entire Accumulator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
5.2.1.1
Writing an Accumulator with a Small Operand . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
5.2.1.2
Using the Extension Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-9
5.2.2
Accessing Portions of an Accumulator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-9
5.2.3
Reading and Writing Integer Data to an Accumulator . . . . . . . . . . . . . . . . . . . . . . 5-12
5.2.4
Reading 16-Bit Results of DSC Algorithms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-12
5.2.5
Converting a 36-Bit Accumulator to a 16-Bit Value . . . . . . . . . . . . . . . . . . . . . . . 5-13
5.2.6
Saving and Restoring Accumulators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-13
5.2.7
Bit-Manipulation Operations on Accumulators . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-14
5.3
Fractional and Integer Arithmetic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-14
5.3.1
DSP56800E Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-15
5.3.2
Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16
5.3.3
Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-18
5.3.3.1
Fractional Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-18
5.3.3.2
Integer Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-19
5.3.3.3
Operand Re-Ordering for Multiplication Instructions . . . . . . . . . . . . . . . . . . . 5-20
5.3.4
Division. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-21
5.3.4.1
General-Purpose Four-Quadrant Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-22
5.3.4.2
Positive Dividend and Divisor with Remainder . . . . . . . . . . . . . . . . . . . . . . . . 5-23
5.3.4.3
Signed Dividend and Divisor with No Remainder . . . . . . . . . . . . . . . . . . . . . . 5-23
5.3.4.4
Division Overflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-24
5.3.5
Logical Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
5.3.6
Shifting Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
5.3.6.1
Shifting 16-Bit Words. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
5.3.6.2
Shifting 32-Bit Long Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-26
5.3.6.3
Shifting Accumulators by 16 Bits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
5.3.6.4
Shifting with Accumulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
5.4
Unsigned Arithmetic Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
5.4.1
Condition Codes for Unsigned Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-27
5.4.2
Unsigned Single-Precision Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-28
5.5
Extended- and Multi-Precision Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29
5.5.1
Extended-Precision Addition and Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29
5.5.2
Multi-Precision Fractional Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29
5.5.3
Multi-Precision Integer Multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-32
5.5.3.1
Signed 32-Bit × Signed 32-Bit with 32-Bit Result . . . . . . . . . . . . . . . . . . . . . . 5-33
5.5.3.2
Unsigned 32-Bit × Unsigned 32-Bit with 32-Bit Result. . . . . . . . . . . . . . . . . . 5-34
5.5.3.3
Signed 32-Bit × Signed 32-Bit with 64-Bit Result . . . . . . . . . . . . . . . . . . . . . . 5-34
vi
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
5.5.3.4
Other Applications of Multi-Precision Integer Multiplication . . . . . . . . . . . . .
5.6
Normalizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.6.1
Normalized Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.6.2
Normalizing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7
Condition Code Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7.1
Condition Code Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7.2
Condition Codes and Data Sizes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8
Saturation and Data Limiting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.1
Data Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.2
MAC Output Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.3
Instructions Not Affected by the MAC Output Limiter . . . . . . . . . . . . . . . . . . . . .
5.9
Rounding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.9.1
Convergent Rounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.9.2
Two’s-Complement Rounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.9.3
Rounding Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5-35
5-36
5-36
5-37
5-38
5-38
5-38
5-39
5-39
5-41
5-42
5-43
5-44
5-46
5-46
Chapter 6
Address Generation Unit
6.1
AGU Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1
6.1.1
Primary Address Arithmetic Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
6.1.2
Secondary Address Adder Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-3
6.1.3
Single-Bit Shifting Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-3
6.2
AGU Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-3
6.2.1
Address Registers (R0–R5, N) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-4
6.2.2
Stack Pointer Register (SP). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-4
6.2.3
Offset Register (N) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-5
6.2.4
Secondary Read Offset Register (N3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-5
6.2.5
Modifier Register (M01). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-5
6.2.6
Shadow Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-5
6.3
Using Address Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-6
6.4
Byte and Word Addresses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7
6.5
Word Pointer Memory Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-8
6.5.1
Accessing Bytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-10
6.5.2
Accessing Long Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-10
6.5.3
Accessing Data Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-11
6.5.4
Accessing Program Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-13
6.6
Byte Pointer Memory Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-13
6.6.1
Byte Pointers vs. Word Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-14
6.6.2
Byte Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-15
6.7
AGU Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-18
6.8
Linear and Modulo Address Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-20
6.8.1
Linear Address Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-20
6.8.2
Understanding Modulo Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-20
6.8.3
Configuring Modulo Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-22
6.8.3.1
Configuring for Byte and Word Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-22
6.8.3.2
Configuring for Long Word Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-23
6.8.4
Base Pointer and Offset Values in Modulo Instructions. . . . . . . . . . . . . . . . . . . . . 6-26
Freescale Semiconductor
Table of Contents
vii
6.8.4.1
6.8.4.2
6.8.4.3
6.8.4.3.1
6.8.4.3.2
6.8.4.4
6.8.5
6.8.5.1
6.8.5.2
6.8.5.3
6.8.6
6.8.7
6.8.8
6.8.9
6.8.9.1
6.8.9.2
6.8.9.3
Operand Placement Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Example of Incorrect Modulo Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Special Case - ADDA Instructions in Modulo Arithmetic . . . . . . . . . . . . . . . .
Case 1. Adding a Positive Immediate Offset to a Pointer . . . . . . . . . . . . .
Case 2. Adding a Negative Immediate Offset to a Pointer . . . . . . . . . . . . .
Restrictions on the Offset Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Supported Memory Access Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Modulo Addressing for Word Memory Accesses . . . . . . . . . . . . . . . . . . . . . .
Modulo Addressing for Byte and Long Memory Accesses . . . . . . . . . . . . . . .
Modulo Addressing for AGU Arithmetic Instructions . . . . . . . . . . . . . . . . . . .
Simple Circular Buffer Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Setting Up a Modulo Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Wrapping to a Different Bank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Side Effects of Modulo Arithmetic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
When a Pointer Lies Outside a Modulo Buffer . . . . . . . . . . . . . . . . . . . . . . . .
Restrictions on the Offset Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Memory Locations Not Accessible Using Modulo Arithmetic . . . . . . . . . . . .
6-26
6-27
6-28
6-28
6-28
6-28
6-29
6-29
6-29
6-30
6-30
6-32
6-33
6-34
6-34
6-34
6-34
Chapter 7
Bit-Manipulation Unit
7.1
Bit-Manipulation Unit Overview and Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-2
7.1.1
8-Bit Mask Shift Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-2
7.1.2
16-Bit Masking Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-3
7.1.3
16-Bit Testing Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-3
7.1.4
16-Bit Logic Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.2
Bit-Manipulation Unit Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.2.1
Testing Bits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.2.2
Conditional Branching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
7.2.3
Modifying Selected Bits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5
7.3
ANDC, EORC, ORC, and NOTC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5
7.4
Other Bit-Manipulation Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
7.5
Programming Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
7.5.1
Bit-Manipulation Operations on Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
7.5.2
Bit-Manipulation Operations on Byte Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-6
7.5.2.1
Absolute Addresses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-7
7.5.2.2
Word Pointers with Byte Offsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-7
7.5.3
Using Complex Addressing Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-8
7.5.4
Synthetic Conditional Branch and Jump Operations . . . . . . . . . . . . . . . . . . . . . . . . 7-8
7.5.4.1
JRSET and JRCLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
7.5.4.2
BR1SET and BR1CLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
7.5.4.3
JR1SET and JR1CLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-10
Chapter 8
Program Controller
8.1
8.1.1
viii
Program Controller Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-1
Instruction Latch and Decoder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
8.1.2
Program Counter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.1.3
Looping Control Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.1.4
Hardware Stack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.1.5
Interrupt Control Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.1.6
Interrupt Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-3
8.2
Program Controller Programming Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
8.2.1
Operating Mode Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
8.2.1.1
Operating Mode (MA and MB)—Bits 0–1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.2
External X Memory (EX)—Bit 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.3
Saturation (SA)—Bit 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.4
Rounding (R)—Bit 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.5
Stop Delay (SD)—Bit 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.6
X or P Memory (XP)—Bit 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-6
8.2.1.7
Condition Code Mode (CM)—Bit 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
8.2.1.8
Nested Looping (NL)—Bit 15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
8.2.2
Status Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-7
8.2.2.1
Carry (C)—Bit 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
8.2.2.2
Overflow (V)—Bit 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
8.2.2.3
Zero (Z)—Bit 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
8.2.2.4
Negative (N)—Bit 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
8.2.2.5
Unnormalized (U)—Bit 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-9
8.2.2.6
Extension in Use (E)—Bit 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-10
8.2.2.7
Limit (L)—Bit 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-10
8.2.2.8
Size (SZ)—Bit 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-10
8.2.2.9
Interrupt Mask (I0–I1)—Bits 8–9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-10
8.2.2.10
Program Counter Extension (P0–P4)—Bits 10–14 . . . . . . . . . . . . . . . . . . . . . 8-11
8.2.2.11
Loop Flag (LF)—Bit 15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
8.2.3
Loop Count Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
8.2.4
Loop Count Register 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-11
8.2.5
Loop Address Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-12
8.2.6
Loop Address Register 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-12
8.2.7
Hardware Stack Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-12
8.2.8
Fast Interrupt Status Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-12
8.2.9
Fast Interrupt Return Address. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-14
8.3
Software Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-14
8.3.1
Pushing and Popping Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-14
8.3.2
Subroutines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15
8.3.3
Interrupt Service Routines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15
8.3.4
Parameter Passing and Local Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-16
8.4
Hardware Stack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-17
8.5
Hardware Looping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-18
8.5.1
Repeat (REP) Looping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-18
8.5.2
DO Looping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-19
8.5.3
Specifying a Loop Count of Zero . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-20
8.5.4
Terminating a DO Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-20
8.5.4.1
Allowing Current Block to Finish and Then Exiting . . . . . . . . . . . . . . . . . . . . 8-20
8.5.4.2
Immediate Exit from a Hardware Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-21
Freescale Semiconductor
Table of Contents
ix
8.5.5
Specifying a Large Immediate Loop Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.6
Nested Hardware Looping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.6.1
Nesting a REP Loop Within a DO Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.6.2
Nesting a DO Loop Within a DO Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.6.3
Nesting a DO Loop Within a Software Loop . . . . . . . . . . . . . . . . . . . . . . . . . .
8.6
Executing Programs from Data Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.6.1
Entering Data-Memory Execution Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.6.2
Exiting Data-Memory Execution Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.6.3
Interrupts in Data-Memory Execution Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.6.4
Restrictions on Data-Memory Execution Mode . . . . . . . . . . . . . . . . . . . . . . . . . . .
8-21
8-22
8-22
8-22
8-23
8-23
8-25
8-26
8-28
8-28
Chapter 9
Processing States
9.1
Normal Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-1
9.2
Reset Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-1
9.3
Exception Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-2
9.3.1
Interrupt Priority Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
9.3.2
Interrupt and Exception Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-4
9.3.2.1
Normal Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-5
9.3.2.2
Fast Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-6
9.3.3
Interrupt Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-8
9.3.3.1
External Hardware Interrupt Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-8
9.3.3.2
Hardware Interrupt Sources Within the Core . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-8
9.3.3.2.1
Illegal Instruction Interrupt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
9.3.3.2.2
Hardware Stack Overflow Interrupt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
9.3.3.2.3
Misaligned Data Access Interrupt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
9.3.3.2.4
Debugging (Enhanced OnCE) Interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-9
9.3.3.3
Software Interrupt Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.3.3.3.1
SWI Instruction—Level 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.3.3.3.2
SWI #x Instructions—Levels 0–2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.3.3.3.3
SWILP Instruction—Lowest Priority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.3.4
Non-Interruptible Instruction Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-10
9.4
Wait Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-11
9.4.1
Wait Mode Timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-12
9.4.2
Disabling Wait Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-12
9.5
Stop Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-12
9.5.1
Stop Mode Timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-13
9.5.2
Disabling Stop Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-13
9.6
Debug Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-13
Chapter 10
Instruction Pipeline
10.1 Pipeline Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2 Normal Pipeline Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2.1
General Pipeline Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2.2
Data ALU Execution Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
x
DSP56800E and DSP56800EX Core Reference Manual
10-2
10-3
10-3
10-4
Freescale Semiconductor
10.3 Pipeline During Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-7
10.3.1
Standard Interrupt Processing Pipeline. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-7
10.3.2
The RTID Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-9
10.3.3
Nested Interrupts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-11
10.3.4
SWI and Illegal Instructions During Interrupt Processing . . . . . . . . . . . . . . . . . . 10-11
10.3.5
Fast Interrupt Processing Pipeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-13
10.3.6
Interrupting a Fast Interrupt Service Routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-14
10.3.7
FIRQ Followed by Another Interrupt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-16
10.3.8
Interrupt Latency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-22
10.3.8.1
Interrupt Latency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-22
10.3.8.2
Re-Enabling Interrupt Arbitration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-23
10.3.8.3
Cases That Increase Interrupt Latency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-23
10.3.8.4
Delay When Enabling Interrupts via CCPL . . . . . . . . . . . . . . . . . . . . . . . . . . 10-24
10.4 Pipeline Dependencies and Interlocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-26
10.4.1
Data ALU Pipeline Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-26
10.4.2
AGU Pipeline Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-28
10.4.3
Instructions with Inherent Stalls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-30
10.4.3.1
Dependencies with Hardware Looping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-31
Chapter 11
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.1 Enhanced OnCE Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-1
11.1.1
Enhanced OnCE Module Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-2
11.2 Enhanced OnCE System Level View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-2
11.3 Accessing the Enhanced OnCE Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-4
11.3.1
External Interaction via JTAG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-4
11.3.2
Core Access to the Enhanced OnCE Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-5
11.3.3
Other Supported Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
11.4 Enhanced OnCE and the Processing States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
11.4.1
Using the Debug Processing State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
11.4.2
Debugging and the Other Processing States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-7
11.4.3
Enhanced OnCE Module Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-8
11.4.3.1
Command, Status, and Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-8
11.4.3.2
Breakpoint Unit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-9
11.4.3.2.1
Trigger Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-9
11.4.3.2.2
16-bit Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-10
11.4.3.2.3
Combining Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-11
11.4.3.3
Step Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-11
11.4.3.4
Change-of-Flow Trace Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-11
11.4.3.5
Realtime Data Transfer Unit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-11
11.4.4
Effectively Using the Debug Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-13
11.4.4.1
Using the Step Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-13
11.4.4.1.1
Usage upon Exiting the Debug Processing State . . . . . . . . . . . . . . . . . . . 11-13
11.4.4.1.2
Step Counter Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-13
11.4.4.1.3
Other Step Counter Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-14
11.4.4.2
Using the Breakpoint Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-14
11.4.4.2.1
Listing the Breakpoint Unit Triggers Available . . . . . . . . . . . . . . . . . . . . 11-16
Freescale Semiconductor
Table of Contents
xi
11.4.4.2.2
Breakpoint Unit Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.4.4.2.3
Combining the Breakpoint Unit with the Step Counter . . . . . . . . . . . . . .
11.4.4.2.4
Breakpoint Unit — Step Counter Actions . . . . . . . . . . . . . . . . . . . . . . . .
11.4.4.3
Capture Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.4.4.3.1
16-Bit Capture Counter (Non-Cascaded) . . . . . . . . . . . . . . . . . . . . . . . . .
11.4.4.3.2
Actions for 16-Bit Capture Counter (Non-Cascaded) . . . . . . . . . . . . . . .
11.4.4.3.3
Using the Capture Counter with the Step Counter . . . . . . . . . . . . . . . . . .
11.4.4.3.4
16-bit Capture Counter — Step Counter Actions . . . . . . . . . . . . . . . . . . .
11.4.4.3.5
40-Bit Capture Counter (Cascaded) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.4.4.3.6
Actions for 40-Bit Capture Counter (Cascaded). . . . . . . . . . . . . . . . . . . .
11.4.4.4
Programmable Trace Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.4.5
Example Breakpoint Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5 JTAG Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.1
JTAG Capabilities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2
JTAG Port Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2.1
JTAG Terminal Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2.2
Core JTAG Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2.3
Core JTAG Port Block Diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2.4
Core TAP Controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.3
JTAG Port Restriction — STOP Processing State . . . . . . . . . . . . . . . . . . . . . . . .
11-18
11-19
11-19
11-20
11-20
11-22
11-23
11-23
11-24
11-24
11-24
11-26
11-27
11-27
11-27
11-28
11-29
11-29
11-30
11-32
Appendix A
Instruction Set Details
A.1
A.2
A.3
A.3.1
A.4
A.5
A.5.1
A.5.2
A.5.3
A.5.4
A.5.5
A.5.6
A.5.7
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-1
Instruction Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-7
32 x 32 to 32/64 Multiply and MAC Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . A-314
32 x 32 to 32/64 Multiplication and MAC Instruction Details. . . . . . . . . . . . . . . A-316
Test Bitfield and Set/Clear (BFSC) Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-327
Instruction Opcode Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-329
Register Operand Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-330
MOVE Instruction Register Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-336
Encodings for Instructions that Support the Entire Register Set . . . . . . . . . . . . . A-338
Parallel Move Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-343
Addressing Mode Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-345
Conditional Instruction Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-345
Immediate and Absolute Address Encoding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-346
Appendix B
Condition Code Calculation
B.1
Factors Affecting Condition Code Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-1
B.1.1
Operand Size and Type. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-1
B.1.2
MAC Output Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-3
B.1.3
Condition Code Mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-3
B.2
Condition Code Register. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-4
B.2.1
Size Bit (SZ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-5
B.2.2
Limit Bit (L) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-5
xii
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
B.2.3
Extension in Use Bit (E) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-6
B.2.4
Unnormalized Bit (U) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-6
B.2.5
Negative Bit (N) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-6
B.2.6
Zero Bit (Z) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-7
B.2.7
Overflow Bit (V). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-7
B.2.8
Carry Bit (C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-7
B.3
Condition Code Summary by Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-8
B.3.1
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-8
B.3.2
Condition Code Calculation Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-8
B.3.3
Special Calculation Rules for Certain Instructions. . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.1
ASL and ASL.W. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.2
ASLL.W and ASLL.L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.3
ASRAC and LSRAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.4
BFCHG, BFCLR, BFSET, BFTSTH, and BRSET . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.5
BFTSTL and BRCLR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.6
BFSC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-14
B.3.3.7
IMPY.W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-15
B.3.3.8
NORM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-15
Appendix C
Glossary
Freescale Semiconductor
Table of Contents
xiii
xiv
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
List of Figures
Figure 1-1
Figure 1-2
Figure 1-3
Figure 1-4
Figure 1-5
Figure 2-1
Figure 2-2
Figure 2-3
Figure 2-4
Figure 3-1
Figure 3-2
Figure 3-3
Figure 3-4
Figure 3-5
Figure 3-6
Figure 3-7
Figure 3-8
Figure 3-9
Figure 3-10
Figure 3-11
Figure 3-12
Figure 3-13
Figure 3-14
Figure 3-15
Figure 3-16
Figure 3-17
Figure 3-18
Figure 3-19
Figure 3-20
Figure 3-21
Figure 3-22
Figure 3-23
Figure 3-24
Figure 3-25
DSP56800EX/DSP56800E Core Block Diagram. . . . . . . . . . . . . . . . . . . . . . . . 1-3
Example of Chip Based on DSP56800EX Core . . . . . . . . . . . . . . . . . . . . . . . . . 1-4
Analog Signal Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-5
Digital Signal Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-6
Mapping DSC Algorithms into Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-7
Core Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-3
Dual Harvard Memory Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-5
DSC Chip Architecture with External Bus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6
Core Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-7
Core Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-2
Single Parallel Move. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
Dual Parallel Read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-12
Data Alignment in Accumulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13
Supported Data Types in Data Registers (X0, Y1, Y0) . . . . . . . . . . . . . . . . . . 3-14
Data Alignment in 24-bit AGU Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
Data Alignment in 16-Bit AGU Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15
Structure of Byte and Word Addresses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16
Accessing a Word with a Word Pointer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
Correct Storage of 32-Bit Value in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . 3-19
Accessing a Long Word Using an Address Register . . . . . . . . . . . . . . . . . . . . 3-20
Accessing a Long Word Using the SP Register . . . . . . . . . . . . . . . . . . . . . . . . 3-20
Accessing a Byte with a Word Pointer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-21
Accessing a Byte with a Byte Pointer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-22
Address Register Indirect: No Update . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-29
Address Register Indirect: Post-Increment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-30
Address Register Indirect: Post-Decrement . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-31
Address Register Indirect: Post-Update by Offset N . . . . . . . . . . . . . . . . . . . . 3-32
Address Register Indirect: Indexed by Offset N. . . . . . . . . . . . . . . . . . . . . . . . 3-33
Address Register Indirect: Indexed by 3-Bit Displacement . . . . . . . . . . . . . . . 3-34
Address Register Indirect: Indexed by 6-Bit Displacement . . . . . . . . . . . . . . . 3-35
Address Register Indirect: Indexed by 16-Bit Displacement . . . . . . . . . . . . . . 3-36
Address Register Indirect: Indexed by 24-Bit Displacement . . . . . . . . . . . . . . 3-37
Immediate Addressing: 5-Bit Immediate Data to Accumulator . . . . . . . . . . . . 3-38
Immediate Addressing: 7-Bit Immediate Data to Address Register. . . . . . . . . 3-39
Freescale Semiconductor
List of Figures
xv
Figure 3-26
Figure 3-27
Figure 3-28
Figure 3-29
Figure 3-30
Figure 3-31
Figure 3-32
Figure 3-33
Figure 4-1
Figure 5-1
Figure 5-2
Figure 5-3
Figure 5-4
Figure 5-5
Figure 5-6
Figure 5-7
Figure 5-8
Figure 5-9
Figure 5-10
Figure 5-11
Figure 5-12
Figure 5-13
Figure 5-14
Figure 5-15
Figure 5-16
Figure 5-17
Figure 5-18
Figure 5-19
Figure 5-20
Figure 5-21
Figure 5-22
Figure 5-23
Figure 5-24
Figure 5-25
Figure 5-26
Figure 5-27
Figure 5-28
Figure 5-29
Figure 6-1
xvi
Immediate Addressing: 7-Bit Immediate Data to Data ALU Register . . . . . . . 3-39
Immediate Addressing: 16-Bit Immediate Data to AGU Register . . . . . . . . . . 3-40
Immediate Addressing: 16-Bit Immediate Data to Data ALU Register . . . . . . 3-40
Immediate Addressing: 32-Bit Immediate Data . . . . . . . . . . . . . . . . . . . . . . . . 3-41
Absolute Addressing: 6-Bit Absolute Short Address . . . . . . . . . . . . . . . . . . . . 3-42
Absolute Addressing: 6-Bit I/O Short Address . . . . . . . . . . . . . . . . . . . . . . . . 3-43
Absolute Addressing: 16-Bit Absolute Address . . . . . . . . . . . . . . . . . . . . . . . . 3-44
Absolute Addressing: 24-Bit Absolute Address . . . . . . . . . . . . . . . . . . . . . . . . 3-45
Moving Data in the Register Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-51
Data ALU Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2
Data ALU Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-3
The 32-Bit Y Register—Composed of Y1 Concatenated with Y0. . . . . . . . . . . 5-4
Different Components of an Accumulator (Using “FF” Notation) . . . . . . . . . . 5-4
Writing the Accumulator as a Whole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-8
Writing the Accumulator by Portions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10
Writing the Accumulator Extension Registers (FF2) . . . . . . . . . . . . . . . . . . . . 5-10
Reading the Accumulator Extension Registers (FF2) . . . . . . . . . . . . . . . . . . . 5-11
Integer Word Addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16
Fractional Word Addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16
Adding a Word Integer to a Long-Word Integer . . . . . . . . . . . . . . . . . . . . . . . 5-17
Adding a Word Fractional to a Long-Word Fractional . . . . . . . . . . . . . . . . . . 5-17
Comparison of Integer and Fractional Multiplication . . . . . . . . . . . . . . . . . . . 5-18
Fractional Multiplication (MPY) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-19
Integer Multiplication with Word-Sized Result (IMPY.W) . . . . . . . . . . . . . . . 5-20
Integer Multiplication with Long-Word-Sized Result (IMPY.L). . . . . . . . . . . 5-20
16- and 32-Bit Logical Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25
Arithmetic Shifts on 16-Bit Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-26
Arithmetic Shifts on 32-Bit Long Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-26
Single-Precision-Times-Double-Precision Signed Multiplication . . . . . . . . . . 5-30
Double-Precision-Times-Double-Precision Signed Multiplication . . . . . . . . . 5-31
32-Bit × 32-Bit –> 32-Bit Signed Integer Multiplication . . . . . . . . . . . . . . . . . 5-33
32-Bit × 32-Bit –> 32-Bit Unsigned Integer Multiplication. . . . . . . . . . . . . . . 5-34
32-Bit × 32-Bit –> 64-Bit Signed Integer Multiplication . . . . . . . . . . . . . . . . . 5-35
Normalizing a Small Negative Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-36
Normalizing a Large Positive Value. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-37
Example of Saturation Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-40
Convergent Rounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-45
Two’s-Complement Rounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-46
Address Generation Unit Block Diagram (DSP56800E Core). . . . . . . . . . . . . . 6-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Figure 6-2
Figure 6-3
Figure 6-4
Figure 6-5
Figure 6-6
Figure 6-7
Figure 6-8
Figure 6-9
Figure 6-10
Figure 7-1
Figure 8-1
Figure 8-2
Figure 8-3
Figure 8-4
Figure 8-5
Figure 9-1
Figure 9-2
Figure 9-3
Figure 10-1
Figure 10-2
Figure 10-3
Figure 10-4
Figure 10-5
Figure 10-6
Figure 10-7
Figure 10-8
Figure 10-9
Figure 10-10
Figure 10-11
Figure 10-12
Figure 10-13
Figure 11-1
Figure 11-2
Figure 11-3
Figure 11-4
Figure 11-5
Figure 11-6
Figure 11-7
Figure 11-8
Dual Parallel Read Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
Address Generation Unit Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . 6-4
Word vs. Byte Addresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-8
Executing the MOVE.L X:(R3+2),D Instruction . . . . . . . . . . . . . . . . . . . . . . . 6-12
Executing the MOVEU.BP X:(R1+7),B Instruction . . . . . . . . . . . . . . . . . . . . . 6-17
Circular Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-21
37-Location Circular Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-22
Simple Five-Location Circular Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-31
Linear Addressing with a Modulo Modifier . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-33
Bit-Manipulation Unit Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-2
Program Controller Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-2
Program Controller Programming Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-4
Effects of the JSR Instruction on the Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15
Example Stack Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-17
Example Data-Memory Execution Mode Memory Map . . . . . . . . . . . . . . . . . 8-24
Interrupt Vector Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-5
Control Flow in Normal Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . 9-6
Control Flow in Fast Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-7
DSP56800E Eight-Stage Pipeline. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-1
Standard Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-8
Execution of the RTID Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-10
Interrupting an Interrupt Handler (Nested Interrupt) . . . . . . . . . . . . . . . . . . . 10-12
Fast Interrupt Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-13
Interrupting a Fast Interrupt Routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-15
Interrupting After Completing the Fastest Fast Interrupt Routine . . . . . . . . . 10-17
Interruption by Level 3 Interrupt During FRTID Execution . . . . . . . . . . . . . 10-19
Second Interrupt Case with 4 Cycles Executed in FRTID Delay Slots . . . . . 10-21
Interrupt Latency Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-22
Interrupt Latency Calculation—Non-Interruptible Instructions . . . . . . . . . . . 10-23
Interrupt Latency and the REP Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-24
Delay When Updating the CCPL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-25
DSP56800E On-Chip System with Debug Port . . . . . . . . . . . . . . . . . . . . . . . . 11-3
JTAG/Enhanced OnCE Interface Block Diagram . . . . . . . . . . . . . . . . . . . . . . 11-5
Breakpoint Unit Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-9
Trigger 1 Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-10
Trigger 2 Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-10
Realtime Data Transfer Unit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-12
Step Counter — Started upon Exiting Debug State . . . . . . . . . . . . . . . . . . . . 11-13
Step Counter — Started upon Exiting Debug State with Breakpoint Active . 11-13
Freescale Semiconductor
List of Figures
xvii
Figure 11-9
Figure 11-10
Figure 11-11
Figure 11-12
Figure 11-13
Figure 11-14
Figure 11-15
Figure 11-16
Figure 11-17
Figure A-1
Figure A-2
Figure B-1
Figure B-2
xviii
Breakpoint Unit Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-15
Triggering the Step Counter with the Breakpoint Unit. . . . . . . . . . . . . . . . . . 11-19
Capture Counter — 16-bit Configuration (Non-Cascaded) . . . . . . . . . . . . . . 11-20
Triggering the Step Counter with the Capture Counter . . . . . . . . . . . . . . . . . 11-23
Capture Counter — 40-bit Configuration (Cascaded) . . . . . . . . . . . . . . . . . . 11-24
Programmable Trace Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-25
JTAG Port Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-29
Core JTAG Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-30
TAP Controller State Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-31
Example Instruction Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-329
Encoding for the MPY Y1,B1,A X:(R1)+,Y1 Instruction . . . . . . . . . . . . . A-329
Internal Data ALU Alignment and Extension . . . . . . . . . . . . . . . . . . . . . . . . . . B-2
Internal AGU Alignment and Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
List of Tables
Table 2-1
Table 3-1
Table 3-2
Table 3-3
Table 3-4
Table 3-5
Table 3-6
Table 3-7
Table 3-8
Table 3-9
Table 3-10
Table 3-11
Table 3-12
Table 3-13
Table 3-14
Table 4-1
Table 4-2
Table 4-3
Table 4-4
Table 4-5
Table 4-6
Table 4-7
Table 4-8
Table 4-9
Table 4-10
Table 4-11
Table 4-12
Table 4-13
Table 4-14
Table 4-15
Table 4-16
Table 4-17
Table 4-18
Table 4-19
Example for Chip I/O and On-Chip Peripheral Memory Map . . . . . . . . . . . . . . 2-4
Core Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-3
Interpretation of 16-Bit Data Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7
Interpretation of 36-Bit Data Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8
Memory Space Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-9
Suffixes for Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
Typical 16-Bit-Word Register Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
Useful Built-In Assembler Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-22
Notation for AGU Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-23
Register-Direct Addressing Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-24
Address-Register-Indirect Addressing Modes . . . . . . . . . . . . . . . . . . . . . . . . . 3-25
Immediate Addressing Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-25
Absolute Addressing Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
Assembler Operator Syntax for Immediate Data Sizes . . . . . . . . . . . . . . . . . . 3-26
Assembler Operator Syntax for Branch and Jump Addresses . . . . . . . . . . . . . 3-27
Multiplication Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2
Additional 32-Bit DSP56800EX Multiplication Instructions. . . . . . . . . . . . . . . 4-2
Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
Shifting Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6
Logical Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7
AGU Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-8
Bitfield Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
Additional DSP56800EX Bitfield Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
Looping Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9
Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-10
Program Control and Change-of-Flow Instructions . . . . . . . . . . . . . . . . . . . . . 4-11
Miscellaneous Program Control Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . 4-11
Aliases for Logical Instructions with Immediate Data . . . . . . . . . . . . . . . . . . . 4-12
Instructions with Alternate Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13
Delayed Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14
Sample Instruction Summary Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
Register Fields for General-Purpose Writes and Reads . . . . . . . . . . . . . . . . . . 4-18
Address Generation Unit (AGU) Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
Data ALU Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
Freescale Semiconductor
List of Tables
xix
Table 4-20
Table 4-21
Table 4-22
Table 4-23
Table 4-24
Table 4-25
Table 4-26
Table 4-27
Table 4-28
Table 4-29
Table 4-30
Table 4-31
Table 4-32
Table 4-33
Table 4-34
Table 4-35
Table 4-36
Table 4-37
Table 4-38
Table 4-39
Table 4-40
Table 4-41
Table 4-42
Table 4-43
Table 4-44
Table 5-1
Table 5-2
Table 5-3
Table 5-4
Table 5-5
Table 6-1
Table 6-2
Table 6-3
Table 6-4
Table 6-5
Table 6-6
Table 6-7
Table 6-8
xx
Additional Register Sets for Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . 4-20
Immediate Value Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
Move Byte Instructions—Byte Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
Move Byte Instructions—Word Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-22
Move Long Word Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-23
Move Word Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-24
Memory-to-Memory Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-26
Immediate Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-27
Register-to-Register Move Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Conditional Register Transfer Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Move Word Instructions—Program Memory. . . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Data ALU Multiply Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Data ALU Extended-Precision Multiplication Instructions . . . . . . . . . . . . . . . 4-30
Data ALU Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-31
Data ALU Shifting Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-39
Data ALU Logical Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-41
Miscellaneous Data ALU Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-41
AGU Arithmetic and Shifting Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-42
Bit-Manipulation Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-43
Branch-on-Bit-Manipulation Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-45
Change-of-Flow Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-46
Looping Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-47
Control Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-48
Single Parallel Move Instructions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-49
Dual Parallel Read Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-50
Accessing the Accumulator Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-7
Data Types and Range of Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-15
Data Limiter Saturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-39
MAC Unit Outputs with Saturation Enabled . . . . . . . . . . . . . . . . . . . . . . . . . . 5-42
Rounding Results for Different Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-47
Capabilities of the Address Pointer Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . 6-6
Hardware Implementation of Addressing Mode Arithmetic—
Word Pointers to Data Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-9
Addressing Mode Arithmetic—Program Memory . . . . . . . . . . . . . . . . . . . . . . 6-13
Addressing Mode Arithmetic—Byte Pointers to Data Memory . . . . . . . . . . . 6-14
AGU Address Arithmetic Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-18
Programming the M01 Register—Byte and Word Accesses . . . . . . . . . . . . . . 6-23
Programming the M01 Register—Long-Word Accesses . . . . . . . . . . . . . . . . . 6-25
Base Pointer and Offset/Update for DSP56800E Instructions . . . . . . . . . . . . . 6-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Table 7-1
Table 8-1
Table 8-2
Table 8-3
Table 8-4
Table 8-5
Table 9-1
Table 9-2
Table 10-1
Table 10-2
Table 10-3
Table 10-4
Table 10-5
Table 10-6
Table 10-7
Table 11-1
Table 11-2
Table 11-3
Table 11-4
Table 11-5
Table 11-6
Table 11-7
Table 11-8
Table 11-9
Table 11-10
Table 11-11
Table 11-12
Table 11-13
Table 11-14
Table 11-15
Table A-1
Table A-2
Table A-3
Table A-4
Table A-5
Table A-6
Table A-7
Table A-8
Operations Synthesized Using DSP56800E Instructions . . . . . . . . . . . . . . . . . . 7-8
OMR Bit Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-5
SR Bit Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-8
Interrupt Mask Bits Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-10
FISR Bit Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-13
Hardware Stack Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-18
Interrupt Priority Level Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
Current Core Interrupt Priority Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-3
Mapping Fundamental Operations to Pipeline Stages . . . . . . . . . . . . . . . . . . . 10-3
Instruction Pipelining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-4
Execution of Data ALU Instructions in the Pipeline . . . . . . . . . . . . . . . . . . . . 10-6
Data ALU Operand Dependency Pipeline . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-27
Data ALU Pipeline with No Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . 10-28
AGU Write Dependency Pipeline. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-29
AGU Pipeline With No Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-30
Processing States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-6
Step Counter Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-14
Notation used in Breakpoint Unit Triggering . . . . . . . . . . . . . . . . . . . . . . . . . 11-16
First Part of Breakpoint Unit Trigger(s)— 16-bit Counter Available for
Triggering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-17
Breakpoint Unit Trigger — 16-bit Counter Available for Triggering . . . . . . 11-18
Possible Breakpoint Unit Actions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-18
Breakpoint Unit — Step Counter Operation. . . . . . . . . . . . . . . . . . . . . . . . . . 11-19
Starting and Stopping the Capture Counter — Non-Cascaded. . . . . . . . . . . . 11-20
First Part of Breakpoint Unit Trigger— 16-bit Counter in Capture Mode . . . 11-21
Breakpoint Unit Trigger — for 16-bit Capture Counter. . . . . . . . . . . . . . . . . 11-22
Possible Capture Counter Actions — Non-Cascaded. . . . . . . . . . . . . . . . . . . 11-22
Possible Capture Counter Actions — Non-Cascaded. . . . . . . . . . . . . . . . . . . 11-23
Starting and Stopping Trace Buffer Capture . . . . . . . . . . . . . . . . . . . . . . . . . 11-25
Possible Actions on Trace Buffer Full . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-25
JTAG Pin Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-28
Register Fields for General-Purpose Writes and Reads . . . . . . . . . . . . . . . . . . . A-2
Address Generation Unit (AGU) Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-3
Data ALU Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-3
Additional Register Fields for Move Instructions . . . . . . . . . . . . . . . . . . . . . . . A-4
Opcode Encoding Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-5
Instruction Field Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-6
Data ALU Register Operand Encodings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-330
Three-Operand Data ALU Instruction Register Encodings . . . . . . . . . . . . . . A-332
Freescale Semiconductor
List of Tables
xxi
Table A-9
Table A-10
Table A-11
Table A-12
Table A-13
Table A-14
Table A-15
Table A-16
Table A-17
Table A-18
Table A-19
Table B-1
Table B-2
Table B-3
xxii
Register Op Codes for DALU Instructions with Parallel Moves . . . . . . . . . . A-336
Register Encodings for MOVE Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . A-336
Encodings for Instructions with Different Load and Store Register Sets. . . . A-339
Bit-Manipulation Register Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-340
Size-Dependent Register Encodings for MOVE Instructions . . . . . . . . . . . . A-342
Single Parallel Move Register Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-344
Dual Parallel Read Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-344
Addressing Mode Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-345
Condition Encoding for the Tcc Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . A-345
Condition Encoding for Jump and Branch Instructions . . . . . . . . . . . . . . . . . A-346
Offset Values for iii Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-347
Condition Code Bit Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-4
Condition Code Summary Table Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-8
Condition Code Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-9
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
List of Examples
Example 3-1
Example 3-2
Example 3-3
Example 3-4
Example 3-5
Example 3-6
Example 3-7
Example 3-8
Example 3-9
Example 3-10
Example 4-1
Example 4-2
Example 4-3
Example 4-4
Example 4-5
Example 4-6
Example 5-1
Example 5-2
Example 5-3
Example 5-4
Example 5-5
Example 5-6
Example 5-7
Example 5-8
Example 5-9
Example 5-10
Example 5-11
Example 5-12
Demonstrating Source and Destination Operands . . . . . . . . . . . . . . . . . . . . . . . 3-9
Program Memory Accesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
Examples of Operands in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-11
Loading Accumulators with Different Data Types. . . . . . . . . . . . . . . . . . . . . . 3-13
Storing Accumulators with Different Data Types . . . . . . . . . . . . . . . . . . . . . . 3-14
Allocation of 2 Bytes Globally . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
Allocation of a Character String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-17
Using the Register-Direct Addressing Mode . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
Effects of Data Types on AGU Arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28
Effects of Data Types on Address Displacements . . . . . . . . . . . . . . . . . . . . . . 3-28
Logical OR with a Data Memory Location . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12
Code Fragment with Regular Branch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
Code Fragment with Delayed Branch. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
Valid Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
Invalid Instruction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
Examples of Single Parallel Moves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-50
X0 Register Used in Operation and Loaded in Parallel . . . . . . . . . . . . . . . . . . . 5-4
Accumulator A Used in Operation and Stored in Parallel . . . . . . . . . . . . . . . . . 5-5
Unsigned Load of a Long Word to an Accumulator . . . . . . . . . . . . . . . . . . . . . 5-9
Reading the Contents of the C2 Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11
Writing a Value into the C2 Register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11
Loading an Accumulator with an Integer Word . . . . . . . . . . . . . . . . . . . . . . . . 5-12
Loading an Accumulator with a Long Integer . . . . . . . . . . . . . . . . . . . . . . . . . 5-12
Reading an Integer Value from an Accumulator . . . . . . . . . . . . . . . . . . . . . . . 5-12
Reading a Word from an Accumulator with Saturation . . . . . . . . . . . . . . . . . . 5-12
Reading a Long Value from an Accumulator with Limiting . . . . . . . . . . . . . . 5-13
Converting a 36-Bit Accumulator to a 16-Bit Value . . . . . . . . . . . . . . . . . . . . 5-13
Saving and Restoring an Accumulator—Word Accesses. . . . . . . . . . . . . . . . . 5-13
Example 5-13
Example 5-14
Example 5-15
Example 5-16
Saving and Restoring an Accumulator—Long Accesses . . . . . . . . . . . . . . . . .
Bit Manipulation on a DSP56800E Accumulator. . . . . . . . . . . . . . . . . . . . . . .
Signed Division with Remainder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Unsigned Division with Remainder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Freescale Semiconductor
List of Examples
5-14
5-14
5-22
5-23
xxiii
Example 5-17
Example 5-18
Example 5-19
Example 5-20
Example 5-21
Example 5-22
Example 5-23
Example 5-24
Example 5-25
Example 5-26
Example 5-27
Example 5-28
Example 5-29
Example 5-30
Example 5-31
Example 5-32
Example 5-33
Example 6-1
Example 6-2
Example 6-3
Example 6-4
Example 6-5
Example 6-6
Example 6-7
Example 6-8
Example 6-9
Example 6-10
Example 6-11
Example 6-12
Example 7-1
Example 7-2
Example 7-3
Example 7-4
Signed DIvision Without Remainder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-24
Multiplication of 2 Unsigned Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-28
64-Bit Addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29
64-Bit Subtraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29
Fractional Single-Precision Times Double-Precision—Both Signed . . . . . . . . 5-30
Multiplying Two Fractional Double-Precision Values. . . . . . . . . . . . . . . . . . . 5-32
Multiplying Two Signed Long Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33
Multiplying Two Unsigned Long Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-34
Multiplying Two Signed Long Integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-35
Multiplying Signed 16-Bit Word with Signed 32-Bit Long . . . . . . . . . . . . . . . 5-36
Normalizing with the NORM Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-37
Normalizing with a Shift Instruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-37
Demonstrating the Data Limiter—Positive Saturation . . . . . . . . . . . . . . . . . . . 5-40
Demonstrating the Data Limiter—Negative Saturation . . . . . . . . . . . . . . . . . . 5-41
Demonstrating the MAC Output Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-42
Example Code for Two’s-Complement Rounding . . . . . . . . . . . . . . . . . . . . . . 5-46
Example Code for Convergent Rounding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-47
Accessing Bytes with the MOVE.B Instruction . . . . . . . . . . . . . . . . . . . . . . . . 6-10
Addressing Mode Examples for Long Memory Accesses . . . . . . . . . . . . . . . . 6-11
Accessing Elements in a Data Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-11
Comparison of MOVE.BP and MOVE.B Instructions. . . . . . . . . . . . . . . . . . . 6-15
Accessing Elements in an Array of Bytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-16
Invalid Use of the Modulo Addressing Mode . . . . . . . . . . . . . . . . . . . . . . . . . 6-28
Adding Positive Offset to a Modulo Pointer . . . . . . . . . . . . . . . . . . . . . . . . . . 6-28
Adding “–2” to a Modulo Pointer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-28
Correct Usage - Offset Values Satisfying Restriction . . . . . . . . . . . . . . . . . . . 6-29
Initializing the Circular Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-31
Accessing the Circular Buffer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-31
Accessing the Circular Buffer with Post-Update by Three . . . . . . . . . . . . . . . 6-32
Examples of Byte Masks in BRSET and BRCLR Instructions . . . . . . . . . . . . . 7-3
Using a Mask to Operate on Bits 7–4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-3
Testing Bits in an Operand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-4
Branching on Bits in an Operand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-5
Example 7-5
Example 7-6
Example 7-7
Example 7-8
Clearing Bits in an Operand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Logical Operations on Bytes in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Logical Operations on Bytes Using Word Pointers . . . . . . . . . . . . . . . . . . . . . .
Bit-Manipulation Operations Using Complex Addressing Modes. . . . . . . . . . .
xxiv
DSP56800E and DSP56800EX Core Reference Manual
7-5
7-7
7-8
7-8
Freescale Semiconductor
Example 7-9
Example 7-10
Example 7-11
Example 8-1
Example 8-2
Example 8-3
Example 8-4
Example 8-5
Example 8-6
Example 8-7
Example 8-8
Example 8-9
Example 8-10
Example 8-11
Example 8-12
Example 8-13
Example 8-14
Example 8-15
Example 8-16
Example 9-1
Example 10-1
Example 10-2
Example 10-3
Example 10-4
Example 10-5
Example 10-6
Example 10-7
Example 10-8
Example 10-9
JRSET and JRCLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
BR1SET and BR1CLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-9
JR1SET and JR1CLR Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-10
Pushing a Value on the Software Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-14
Pushing Multiple Values on the Software Stack . . . . . . . . . . . . . . . . . . . . . . . 8-14
Popping Values from the Software Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-15
Subroutine Call with Passed Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-16
Repeat Loop Example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-18
DO Loop Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-19
DO Loop Special Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-20
Immediate Exit from Hardware Loop. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-21
Using the DOSLC Instruction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-22
Example of a REP Loop Nested Within a DO Loop . . . . . . . . . . . . . . . . . . . . 8-22
Example of Nested DO Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-23
Example of Nested Looping in Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-23
Entering Data Memory Execution, 19-Bit Target Address . . . . . . . . . . . . . . . 8-25
Entering Data Memory Execution, 21-Bit Target Address . . . . . . . . . . . . . . . 8-26
Exiting Data-Memory Execution Mode, 19-Bit Target Address . . . . . . . . . . . 8-27
Exiting Data-Memory Execution Mode, 21-Bit Target Address . . . . . . . . . . . 8-27
BRSET Non-Interruptible Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-11
Example Code to Demonstrate Pipeline Flow . . . . . . . . . . . . . . . . . . . . . . . . . 10-4
Demonstrating the Data ALU Execution Stages . . . . . . . . . . . . . . . . . . . . . . . 10-6
Data ALU Operand Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-27
Case with No Data ALU Pipeline Dependencies . . . . . . . . . . . . . . . . . . . . . . 10-27
Pipeline Dependency with AGU Registers. . . . . . . . . . . . . . . . . . . . . . . . . . . 10-28
Case Without AGU Pipeline Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . 10-29
MOVE Instructions That Introduce Stalls . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-30
Instructions with No Stalls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-31
Dependency with Load of LC and Start of Hardware Loop. . . . . . . . . . . . . . 10-31
Freescale Semiconductor
List of Examples
xxv
xxvi
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
About This Book
This manual describes the central processing unit of the DSP56800E and DSP56800EX in detail. It is
intended to be used with the appropriate DSP56800E or DSP56800EX family member reference manual,
which describes the specific chip architecture, peripheral definitions, and programming models. The
appropriate DSP56800E or DSP56800EX family member’s technical data sheet provides timing, pinout,
and packaging descriptions.
This manual provides practical information to help the user accomplish the following:
•
Understand the operation and instruction set of the DSP56800E and DSP56800EX families
•
Write code for DSC algorithms
•
Write code for general control tasks
•
Write code for communication routines
•
Write code for data-manipulation algorithms
Audience
The information in this manual is intended to assist software engineers with developing application
software for DSP56800E and DSP56800EX family devices.
Organization
Information in this manual is organized into chapters by topic. The contents of the chapters are as follows:
Chapter 1, “Introduction.” This chapter introduces the DSP56800E and DSP56800EX core architecture
and its application. It also provides the novice with a brief overview of digital signal processing.
Chapter 2, “Core Architecture Overview.” The DSP56800E and DSP56800EX core architecture
consists of the data arithmetic logic unit (ALU), address generation unit (AGU), bit-manipulation unit, and
program controller. This chapter describes each subsystem and the buses that interconnect the major
components in the DSC core central processing module.
Chapter 3, “Data Types and Addressing Modes.” This chapter presents the programming model,
introduces the MOVE instructions and their syntax, and presents the data types and addressing modes
found on the core.
Chapter 4, “Instruction Set Introduction.” This chapter presents register notation and summarizes the
instruction set. It shows the registers and addressing modes available to each instruction as well as the
number of execution cycles and program words required.
Chapter 5, “Data Arithmetic Logic Unit.” This chapter describes the data ALU architecture, its
programming model, methods for accessing the accumulators, and data types. The chapter also provides an
introduction to fractional and integer arithmetic on the core and discusses other topics such as unsigned
and multi-precision arithmetic.
Freescale Semiconductor
About This Book
xxvii
Chapter 6, “Address Generation Unit.” This chapter describes in detail the AGU architecture, its
programming model, its addressing modes, and its address modifiers.
Chapter 7, “Bit-Manipulation Unit.” This chapter describes in detail the bit-manipulation unit’s
architecture and capabilities.
Chapter 8, “Program Controller.” This chapter describes in detail the program controller architecture,
its programming model, the hardware and software stacks, subroutines, and hardware looping.
Chapter 9, “Processing States.” This chapter introduces the different processing states of the core
(normal, reset, exception, wait, stop, and debug).
Chapter 10, “Instruction Pipeline.” This chapter describes the pipeline of the DSP56800E and
DSP56800EX architecture.
Chapter 11, “JTAG and Enhanced On-Chip Emulation (Enhanced OnCE).” This chapter provides an
overview of the JTAG test interface and the integrated emulation and debugging module (Enhanced
OnCE™).
Appendix A, “Instruction Set Details.” This appendix presents a detailed description of each DSC core
instruction, its use, and its effect on the processor.
Appendix B, “Condition Code Calculation.” This appendix presents a detailed description of condition
code computation.
Appendix C, “Glossary.” The Glossary defines useful DSC, electronics, and communications terms.
Suggested Reading
The following DSC-related books may aid an engineer who is new to the field of digital signal processing:
Advanced Topics in Signal Processing, Jae S. Lim and Alan V. Oppenheim (Prentice-Hall: 1988)
Applications of Digital Signal Processing, A. V. Oppenheim (Prentice-Hall: 1978)
Digital Processing of Signals: Theory and Practice, Maurice Bellanger (John Wiley and Sons: 1984)
Digital Signal Processing, Alan V. Oppenheim and Ronald W. Schafer (Prentice-Hall: 1975)
Digital Signal Processing: A System Design Approach, David J. DeFatta, Joseph G. Lucas, and William S.
Hodgkiss (John Wiley and Sons: 1988)
Discrete-Time Signal Processing, A. V. Oppenheim and R.W. Schafer (Prentice-Hall: 1989)
Foundations of Digital Signal Processing and Data Analysis, J. A. Cadzow (Macmillan: 1987)
Handbook of Digital Signal Processing, D. F. Elliott (Academic Press: 1987)
Introduction to Digital Signal Processing, John G. Proakis and Dimitris G. Manolakis (Macmillan: 1988)
Multirate Digital Signal Processing, R. E. Crochiere and L. R. Rabiner (Prentice-Hall: 1983)
Signal Processing Algorithms, S. Stearns and R. Davis (Prentice-Hall: 1988)
Signal Processing Handbook, C. H. Chen (Marcel Dekker: 1988)
Signal Processing: The Modern Approach, James V. Candy (McGraw-Hill: 1988)
Theory and Application of Digital Signal Processing, Lawrence R. Rabiner and Bernard Gold
(Prentice-Hall: 1975)
xxviii
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Conventions
This document uses the following notational conventions:
•
Bits within registers are always listed from most significant bit (MSB) to least significant bit (LSB).
•
Bits within a register are formatted AA[n:0] when more than one bit is involved in a description.
For purposes of description, the bits are presented as if they are contiguous within a register.
However, they are not always contiguous. Refer to the programming-model diagrams or to the
programmer’s sheets to find the exact location of bits within a register.
•
When a bit is described as set, its value is set to one. When a bit is described as cleared, its value is
set to zero.
•
In graphic displays of registers, the following definitions of notation apply:
— Grey bit: An unimplemented bit that always reads as zero. Writing has no effect.
— TYPE: The bit’s type defines its behavior. Possible values include:
– r: Read-only. Writing this bit has no effect.
– w: Write-only.
– rw: Standard read/write bit. Only software (or a hardware reset) can change the bit’s value.
— RESET: The reset value of the bit. Possible values include:
– 0: Will reset to a logic 0.
– 1: Will reset to a logic 1.
– ?: The reset state is undefined.
– —: The reset state depends on individual chip implementation.
•
A pin or signal that is asserted low (made active when pulled to ground) has a bar over its name.
For example, the SS0 pin is asserted low.
•
Hexadecimal values are preceded by a dollar sign ($), as follows: $FFFB is the X memory address
for the interrupt priority register (IPR).
•
Unless noted otherwise, M designates the value 220 and K designates the value 210.
•
Memory addresses in the separate program and data memory spaces are differentiated by a
one-letter prefix. Data memory addresses have an X: prefix, while program memory addresses have
a P: prefix. For example, P:$0200 indicates a location in program memory. The terms data memory
and X memory are used interchangeably, and the terms program memory and P memory are used
interchangeably.
•
Code examples are displayed in a monospaced font, as follows:
BFSET #$0007,X:PCC ; Configure:
; MISO0, MOSI0, SCK0 for SPI master
; ~SS0 as PC3 for GPIO
Freescale Semiconductor
About This Book
line 1
line 2
line 3
xxix
Definitions, Acronyms, and Abbreviations
The following terms appear frequently in this manual:
DSC
digital signal controller
JTAG
Joint Test Action Group
Enhanced OnCE
Enhanced On-Chip Emulation
ALU
arithmetic logic unit
AGU
address generation unit
IP-BUS
Freescale standard on-chip peripheral interface bus
A complete list of relevant terms and their definitions appears in Appendix C, “Glossary.”
xxx
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 1
Introduction
The 32-bit DSP56800EX core represents the next step in the evolution of Freescale’s families of digital
signal controllers (DSCs). The DSP56800EX core extends the capabilities of the DSP56800E core
architecture.
The DSP56800EX core has all DSP56800E core features and adds new enhancements, including:
•
32-bit x 32-bit multiply and MAC operations
•
all registers in the Address Generation Unit (AGU) have shadowed registers that effectively reduce
the context save/restore time during exception processing, reducing latency
•
bit-reverse addressing mode, supporting Fast Fourier Transform (FFT)
•
new bit manipulation instruction (BFSC) that integrates test-bitfield and a set/clear-bitfield
operations into a single instruction
Both the DSP56800EX and DSP56800E cores provide low-cost, low-power computing, combining DSC
power and parallelism with MCU-like programming simplicity. Each core is a general-purpose central
processing unit, designed for both efficient digital signal processing and a variety of controller operations.
The veteran DSC programmer recognizes a powerful DSC instruction set in these DSC cores.
Microcontroller programmers have access to a rich set of controller and general processing instructions. A
powerful multiply-accumulate (MAC) unit, with optional rounding and negation, enables the efficient
coding of DSC and digital filtering algorithms. The DSC cores’ large register set, powerful addressing
modes, and bit-manipulation unit allow traditional control tasks to be performed with ease, without the
complexity and limitations normally associated with DSCs. Assisting in the coding of general-purpose
programs is support for a software stack; flexible addressing modes; and byte, word, and long-word data
types.
1.1 Key Features
The DSP56800EX and DSP56800E architecture provides a variety of features that enhance performance,
reduce application cost, and ease product development. The architectural features that make these benefits
possible include the following:
•
High Performance—support for all digital signal processing applications.
•
Compatibility—The DSP56800EX is source-code compatible with the Freescale DSP56800E
family, making it a logical upgrade for performance-hungry applications. DSP56800 and
DSP56800E software can be run on the DSP56800EX by simply recompiling or reassembling it.
Freescale Semiconductor
Introduction
1-1
Introduction
•
Ease of Programming—The instruction mnemonics are designed to resemble the mnemonics of
MCUs, simplifying the transition from programming traditional microprocessors. Instruction-set
support for both fractional and integer data types provides the flexibility that is necessary for
optimal algorithm implementation.
•
Support for High-Level Languages—The C programming language is well suited to the DSC
core architecture. The majority of an application can be written in a high-level language without
compromising DSC performance. A flexible instruction set and programming model enable the
efficient generation of compiled code.
•
Rich Instruction Set—In addition to supporting instructions that support DSC algorithms, the
DSP56800EX and DSP56800E provide control, bit-manipulation, and integer processing
instructions. Powerful addressing modes and a range of data sizes are also provided. The result is
compact, efficient code.
•
High Code Density—The base instruction word size for the DSC cores is only 16 bits, with
multi-word instructions for more complex operations, resulting in optimal code density. The
instruction set emphasizes efficient control programming, which accounts for the largest portion of
an application.
•
Multi-Tasking Support—Implementing a real-time operating system or simple multi-tasking is
much easier on the DSP56800EX and DSP56800E than on most DSCs. The architecture provides
full support for a software stack, fast 32-bit context saves and restores to and from the system stack,
atomic test-and-set operations, and four prioritized software interrupts.
•
Precision—The DSP56800EX and DSP56800E cores enable precise DSC calculations. Enough
precision for 96 dB of dynamic range is provided by 16-bit data paths. Intermediate values in the
36-bit accumulators can range over 216 dB.
•
Hardware Looping—Two types of zero-overhead hardware looping are provided, enhancing
performance and making loop-unrolling techniques unnecessary.
•
Parallelism—Each on-chip execution unit, memory device, and peripheral operates independently
and in parallel. Because of the high level of parallelism, the following can be executed in a single
instruction:
— Fetching the next instruction
— A 16-bit × 16-bit multiplication with 36-bit accumulation
— Optional negation, rounding, and saturation of the result
— Two 16-bit data moves
— No-overhead hardware looping
— Two address pointer updates
1-2
•
Invisible Instruction Pipeline—The eight-stage instruction pipeline provides enhanced
performance while remaining essentially invisible to the programmer. Developers can program in
high-level languages such as C without being concerned about the pipeline, even as they benefit
from the pipeline’s throughput of one instruction per cycle.
•
Low Power Consumption—Implemented in CMOS, the DSC cores inherently consume very little
power. In addition, the core architecture supports low-power modes, including STOP and WAIT,
which can provide even more power savings. The power management implementation can shut off
unused sections of logic.
•
Real-Time Debugging—Freescale’s Enhanced On-Chip Emulation technology (Enhanced
OnCE™) allows simple, inexpensive, non-intrusive, and speed-independent access to the internal
state of the DSC core. By using Enhanced OnCE, programmers have full control over the
processor’s operation, simplifying and speeding debugging tasks without having to halt the core.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Architectural Overview
The DSP56800EX and DSP56800E’s efficient instruction set and bus structure, extensive parallelism,
on-chip program and data memories, and advanced debugging and test features make the core an excellent
solution for real-time, embedded DSC and control tasks. It is the perfect choice for wireless and wireline
DSC applications, digital and industrial control, or any other embedded-controller application that needs
high-performance processing.
1.2 Architectural Overview
The DSP56800EX and DSP56800E cores each consist of a data arithmetic logic unit (ALU), an address
generation unit (AGU), a program controller, a bit-manipulation unit, an Enhanced On-Chip Emulation
module (Enhanced OnCE), and associated buses. The following diagram shows the core architecture.
DSC Core
Program Control Unit
PC
LA
LA2
HWS0
HWS1
FIRA
OMR
SR
LC
LC2
FISR
ALU1
Address
Generation
Unit
(AGU)
Instruction
Decoder
Interrupt
Unit
ALU2
R0
R1
Program
Memory
R2
R3
R4
R5
N
M01
N3
Looping
Unit
SP
XAB1
Data
Memory
XAB2
PAB
PDB
CDBW
CDBR
XDB2
A2
B2
C2
D2
BitManipulation
Unit
Enhanced
OnCE™
JTAG TAP
Y
A1
B1
C1
D1
Y1
Y0
X0
MAC and ALU
A0
B0
C0
D0
Data
Arithmetic
Logic Unit
(ALU)
IP-BUS
Interface
External
Bus
Interface
Multi-Bit Shifter
Figure 1-1. DSP56800EX/DSP56800E Core Block Diagram
Freescale Semiconductor
Introduction
1-3
Introduction
Flexible memory support is one of the strengths of the DSC architecture. Supported memories include:
•
Program RAM and ROM modules.
•
Data RAM and ROM modules.
•
Non-volatile memory (NVM) modules.
•
Bootstrap ROM for devices that execute code from RAM.
The Freescale IP-BUS architecture supports a variety of on-chip peripherals. Among the peripherals
available on some devices that are based on the DSP56800EX and DSP56800E cores are the following:
•
Phase-locked loop (PLL) module
•
16-bit timer module
•
Computer operating properly (COP) and real-time timer module
•
Synchronous serial interface (SSI) module
•
Serial peripheral interface (SPI) module
•
Programmable general-purpose I/O (GPIO) module
1.3 Example DSP56800EX Device
Figure 1-2 shows an example device that is built around the DSP56800EX core.
IRQA
RAM
Flash
ADR
24
External Bus
Interface
DATA
IRQB
DSP56800EX
COP & Real-
32-Bit
Time Timer
DSC
Core
32
Timers
PLL
Serial
JTAG
GPIO
AA0002
Figure 1-2. Example of Chip Based on DSP56800EX Core
The DSC core architecture optionally supports chips with external bus interfaces. For chips with an
external bus, the core architecture supports an external address bus that is up to 24 bits wide and data bus
widths of 8, 16, or 32 bits.
1-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Introduction to Digital Signal Processing
1.4 Introduction to Digital Signal Processing
Digital signal processing (DSC) is the arithmetic processing of real-time signals that are sampled and
digitized at regular intervals. Examples of DSC processing include the following:
•
Filtering
•
Convolution (mixing two signals)
•
Correlation (comparing two signals)
•
Rectification, amplification, and transformation
Figure 1-3 shows an example of analog signal processing. The circuit in the illustration filters a signal from
a sensor using an operational amplifier and then controls an actuator with the result. Since the ideal filter is
impossible to design, the engineer must design the filter for acceptable response, considering variations in
temperature, component aging, power-supply variation, and component accuracy. The resulting circuit
typically has low noise immunity, requires adjustments, and is difficult to modify.
Analog Filter
Rf
Cf
x(t)
x(t)
Input
from
Sensor
Ri
y(t)
+
t
y(t)
Output
to
Actuator
Rf
1
y(t)
--------- = – ------ ---------------------------R
1
+
jωR
x(t)
i
f Cf
Frequency Characteristics
Gain
Ideal
Filter
Frequency
fc
Actual
Filter
f
AA0003
Figure 1-3. Analog Signal Processing
The equivalent circuit using a DSC is shown in Figure 1-4 on page 1-6. This application requires an
analog-to-digital (A/D) converter and digital-to-analog (D/A) converter in addition to the DSC.
Freescale Semiconductor
Introduction
1-5
Introduction
Low-Pass
Anti-Aliasing
Filter
Sampler And
Analog-to-Digital
Converter
Digital-to-Analog
Converter
c(k) × (n – k)
D/A
k=0
x(t)
x(n)
y(t)
y(n)
Finite Impulse
Response
Analog Out
Gain
A
Ideal
Filter
Reconstruction
Low-Pass
FIR Filter
N–1

A/D
Analog In
DSC Operation
f
fc
Frequency
Analog
Filter
Gain
A
f
fc
Frequency
Digital
Filter
Gain
A
f
Frequency
fc
AA0004
Figure 1-4. Digital Signal Processing
Processing in this circuit begins with band limiting the input signal with an anti-alias filter, which
eliminates out-of-band signals that can be aliased back into the pass band due to the sampling process. The
signal is then sampled, digitized with an A/D converter, and sent to the DSC. The DSC output is processed
by a D/A converter and is low-pass filtered to remove the effects of digitizing.
The particular filter implemented by the DSC is strictly a matter of software. The DSC can implement any
filter that can be implemented using analog techniques. Moreover, adaptive filters, which are extremely
difficult to implement using analog techniques, can easily be created using DSC.
In summary, the advantages of using the DSC include the following:
1-6
•
Fewer components
•
Stable, deterministic performance
•
No filter adjustments
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Introduction to Digital Signal Processing
•
Wide range of applications
•
Filters with sharper filtering characteristics
•
High noise immunity
•
Adaptive filters are easily implemented
•
Self-test can be built in
•
Better power-supply rejection
The DSP56800EX and DSP56800E families do not consist of custom chips designed for a particular
application; they are designed with a general-purpose DSC architecture to efficiently execute common
DSC algorithms and controller code in minimal time.
As Figure 1-5 shows, the key attributes of a DSC are as follows:
•
Multiply-accumulate (MAC) operation
•
Fetching up to two operands per instruction cycle for the MAC
•
Flexibility in implementation through a powerful instruction set
•
Input/output capability to move data in and out of the DSC
FIR Filter
N–1

A/D
x(t)
x(n)
c(k) × (n – k)
k=0
D/A
y(t)
y(n)
X
Memory
X
Program
Σ
MAC
AA0005
Figure 1-5. Mapping DSC Algorithms into Hardware
The multiply-accumulate (MAC) operation is the fundamental operation used in DSC. The DSP56800EX
and DSP56800E families of processors have a dual Harvard architecture that is optimized for MAC
operations. Figure 1-5 shows how the DSC architecture matches the shape of the MAC operation. The two
operands, C() and X(), are directed to a multiply operation, and the result is summed. This process is built
into the chip in that two separate data memory accesses are allowed to feed a single-cycle MAC. The entire
process must occur under program control to direct the correct operands to the multiplier and to save the
accumulated result, as needed. Since the memory and the MAC are independent, the DSC can perform two
memory moves, a multiply and an accumulate, and two address updates in a single operation. As a result,
many DSC benchmarks execute very efficiently for a single-multiplier architecture.
Freescale Semiconductor
Introduction
1-7
Introduction
1-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 2
Core Architecture Overview
This chapter presents the core’s architecture and programming model as well as the overall system
architecture for devices based on the DSP56800EX and DSP56800E cores. It introduces the different
blocks and data paths within the core and their functions. More detailed information on the individual
blocks within the core, such as the data ALU, AGU, and program controller, appears in later chapters.
2.1 Extending DSP56800E Architecture
The DSP56800EX core architecture extends Freescale’s DSP56800E family architecture. It is source-code
compatible with DSP56800E devices and adds the following new features:
•
32-bit x 32-bit multiply and MAC operations with 32-bit and 64-bit results
•
all registers in the Address Generation Unit (AGU) have shadowed registers that effectively reduce
the context save/restore time during exception processing, reducing latency
•
bit-reverse addressing mode, supporting Fast Fourier Transform (FFT)
•
new bit manipulation instruction (BFSC) that integrates test-bitfield and a set/clear-bitfield
operations into a single instruction
2.2 Extending DSP56800 Architecture
The DSP56800E and DSP56800EX core architecture extends Freescale’s DSP56800 family architecture.
It is source-code compatible with DSP56800 devices and adds the following new features:
•
Byte and long data types, supplementing the DSP56800’s word data type
•
24-bit data memory address space
•
21-bit program memory address space
•
Three additional 24-bit pointer registers (one of which can be used as an offset register)
•
A secondary 16-bit offset register to further enhance the dual parallel data ALU instructions
•
Two additional 36-bit accumulator registers
•
Full-precision integer multiplication
•
32-bit logical and shifting operations
•
Second read in dual read instruction can access off-chip memory
•
Loop count (LC) register extended to 16 bits
Freescale Semiconductor
Core Architecture Overview
2-1
Core Architecture Overview
•
Full support for nested DO looping through additional loop address and count registers
•
Loop address and hardware stack extended to 24 bits
•
Three additional interrupt levels with a software interrupt for each level
•
Enhanced On-Chip Emulation (Enhanced OnCE) with three debugging modes:
— Non-intrusive real-time debugging
— Minimally intrusive real-time debugging
— Breakpoint and step mode (core is halted)
2.3 Core Programming Model
The registers in the core that are considered part of the core programming model are shown in Figure 2-1.
Registers for on-chip peripherals are mapped into a 64-location block of data memory. An example for this
block of memory is shown in Table 2-1 on page 2-4. Consult a specific device’s user’s manual for details
on the peripherals that are implemented, their function, the registers that are defined for them in this
memory area, and their location in memory.
2-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Core Programming Model
Data Arithmetic Logic Unit (ALU)
Data Registers
35 32 31
16 15
0
A
A2
A1
A0
B
B2
B1
B0
C
C2
C1
C0
D
D2
D1
D0
15
0
Y1
Y
Y0
X0
Address Generation Unit (AGU)
Pointer Registers
23
0
Secondary Offset Register
15
0
R0
N3
R1
Modifier Registers
R2
15
0
R3
M01
R4
R5
N
SP
Program Control Unit
20
Program Counter
0
PC
Loop Address
23
Operating Mode Register
and Status Register
(OMR, SR)
15
0
OMR
0
SR
LA
LA2
Hardware Stack
23
Fast Interrupt Status
Register
12
0
0
FISR
HWS0
HWS1
20
Fast Interrupt Return Address
15
Loop Counter
0
FIRA
0
LC
LC2
Figure 2-1. Core Programming Model
Freescale Semiconductor
Core Architecture Overview
2-3
Core Architecture Overview
Table 2-1. Example for Chip I/O and On-Chip Peripheral Memory Map
X:$xxFFFF
(Reserved for DSC Core)
X:$xxFFFE
(Reserved for DSC Core)
X:$xxFFFD
(Reserved for DSC Core)
X:$xxFFFC
(Reserved for DSC Core)
X:$xxFFFB
(Reserved for Interrupt Priority)
X:$xxFFFA
(Reserved for Interrupt Priority)
X:$xxFFF9
(Reserved for Bus Control)
X:$xxFFF8
(Reserved for Bus Control)
X:$xxFFF7
(Reserved for DSC Core)
X:$xxFFF6
(Reserved for DSC Core)
X:$xxFFF5
(Reserved for DSC Core)
X:$xxFFF4
(Reserved for DSC Core)
X:$xxFFF3
(Available for Peripherals)
X:$xxFFF2
(Available for Peripherals)
X:$xxFFF1
(Available for Peripherals)
X:$xxFFF0
(Available for Peripherals)
.
.
.
.
.
.
X:$xxFFC3
(Available for Peripherals)
X:$xxFFC2
(Available for Peripherals)
X:$xxFFC1
(Available for Peripherals)
X:$xxFFC0
(Available for Peripherals)
NOTE:
Peripherals can be located anywhere in data memory and are defined by
the specific device’s user’s manual.
2-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Dual Harvard Memory Architecture
2.4 Dual Harvard Memory Architecture
The DSC core has a dual Harvard architecture with separate program and data memory spaces, as shown in
Figure 2-2. This architecture allows for simultaneous program and data memory accesses. The data
memory interface also supports two simultaneous read operations, enabling up to three simultaneous
memory accesses.
15
0
$FFFFFF
15
16M (32 Mbyte)
0
$1FFFFF
2M (4 Mbyte)
Data
Memory
Space
Program
Memory
Space
$xxFFFF
$xxFFC0
Optimized for
IP-BUS
Peripherals
$0
0
(64K – 64)
Accessible with
X:<<pp Addressing
(Relocatable)
(Relocatable)
Interrupt
Vectors
64K
$0
0
Figure 2-2. Dual Harvard Memory Architecture
The block of memory containing reset and interrupt vectors can be any size and can be located anywhere in
program memory. Peripheral registers are memory mapped into a 64-location region in the data memory
space.
A 64-word block of data memory allocated for memory-mapped IP-BUS peripheral registers can be
located anywhere in data memory. Usually the location of this memory block is chosen so that it does not
overlap with RAM or ROM data memory. The X:<<pp addressing mode (see Section 3.6.5.2, “I/O Short
Address: <<pp,” on page 3-43) provides efficient access to this memory range, enabling single-word,
single-cycle move and bit-manipulation instructions.
Note that the top 12 locations in the peripheral register area ($xxFFF4 through $xxFFFF) are reserved for
use by the core, interrupt priority functions, and bus control functions, as shown in Table 2-1 on page 2-4.
The compiler has access only to the lower 16 Mbyte of data memory.
Freescale Semiconductor
Core Architecture Overview
2-5
Core Architecture Overview
2.5 System Architecture and Peripheral Interface
The DSC system architecture encompasses all the on-chip components, including the core, on-chip
memory, peripherals, and the buses that are necessary to connect them. Figure 2-3 shows the overall
system architecture for a device with an external bus.
Peripheral
Peripheral
Peripheral
IP-BUS
Program
Memory
Data
Memory
PAB
PDB
XAB1
DSC
CDBR
Core
CDBW
XAB2
IP-BUS
Interface
External
Bus
Interface
XDB2
External
Address
External
Data
Figure 2-3. DSC Chip Architecture with External Bus
The complete architecture includes the following components:
•
DSP56800EX or DSP56800E core
•
On-chip program memory
•
On-chip data memory
•
On-chip peripherals
•
Freescale IP-BUS peripheral interface
•
External bus interface
Some DSC devices might not implement an external bus interface. Regardless of the implementation, all
peripherals communicate with the core via the IP-BUS interface. The IP-BUS–interface standard connects
the two data address buses and the CDBR, CDBW, and XDB2 uni-directional data buses to the
corresponding bus interfaces on the peripheral devices. The program memory buses are not connected to
peripherals.
2-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
System Architecture and Peripheral Interface
2.5.1 Core Block Diagram
The DSC core is composed of several independent functional units. The program controller, address
generation unit (AGU), and data arithmetic logic unit (ALU) contain their own register sets and control
logic, allowing them to operate independently and in parallel, which increases throughput. There is also an
independent bit-manipulation unit, which enables efficient bit-manipulation operations. Each functional
unit interfaces with the other units, memory, and the memory-mapped peripherals over the core’s internal
address and data buses. See Figure 2-4.
DSC Core
Program Control Unit
PC
LA
LA2
HWS0
HWS1
FIRA
OMR
SR
LC
LC2
FISR
Address
Generation
Unit
(AGU)
Instruction
Decoder
Interrupt
Unit
ALU1
ALU2
R0
R1
Program
Memory
R2
R3
R4
R5
N
M01
N3
Looping
Unit
SP
XAB1
Data
Memory
XAB2
PAB
PDB
CDBW
CDBR
XDB2
A2
B2
C2
D2
BitManipulation
Unit
Enhanced
OnCE™
JTAG TAP
Y
A1
B1
C1
D1
Y1
Y0
X0
MAC and ALU
A0
B0
C0
D0
Data
Arithmetic
Logic Unit
(ALU)
IP-BUS
Interface
External
Bus
Interface
Multi-Bit Shifter
Figure 2-4. Core Block Diagram
Instruction execution is pipelined to take advantage of the parallel units, significantly decreasing the
execution time for each instruction. For example, all within a single execution cycle, it is possible for the
data ALU to perform a multiplication operation, for the AGU to generate up to two addresses, and for the
program controller to prefetch the next instruction.
Freescale Semiconductor
Core Architecture Overview
2-7
Core Architecture Overview
The major components of the core are the following:
•
Address buses
•
Data buses
•
Data arithmetic logic unit (ALU)
•
Address generation unit (AGU)
•
Program controller
•
Bit-manipulation unit
•
Enhanced OnCE debugging module
The following sections describe these components.
2.5.2 Address Buses
The core contains three address buses: the program memory address bus (PAB), the primary data address
bus (XAB1), and the secondary data address bus (XAB2). The program address bus is 21 bits wide and is
used to address (16-bit) words in program memory. The two 24-bit data address buses allow for two
simultaneous read accesses to data (X) memory. The XAB1 bus can address byte, word, and long data
types. The XAB2 bus is limited to (16-bit) word accesses.
All three buses address on-chip memory. They can also address off-chip memory on devices that contain
an external bus interface unit.
2.5.3 Data Buses
Data transfers inside the chip occur over the following buses:
•
Two uni-directional 32-bit buses:
— Core data bus for reads (CDBR)
— Core data bus for writes (CDBW)
•
Two uni-directional 16-bit buses:
— Secondary X data bus (XDB2)
— Program data bus (PDB)
•
IP-BUS interface
Data transfers between the data ALU and data memory use the CDBR and CDBW when a single memory
read or write is performed. When two simultaneous memory reads are performed, the transfers use the
CDBR and XDB2 buses. All other data transfers to core blocks occur using the CDBR and CDBW buses.
Peripheral transfers occur through the IP-BUS interface. Instruction word fetches occur over the PDB.
This bus structure supports up to three simultaneous 16-bit transfers. Any one of the following can occur in
a single clock cycle:
2-8
•
One instruction fetch
•
One read from data memory
•
One write to data memory
•
Two reads from data memory
•
One instruction fetch and one read from data memory
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
System Architecture and Peripheral Interface
•
One instruction fetch and one write to data memory
•
One instruction fetch and two reads from data memory
An instruction fetch will take place on every clock cycle, although it is possible for data memory accesses
to be performed without an instruction fetch. Such accesses typically occur when a hardware loop is
executed and the repeated instruction is only fetched on the first loop iteration. See Section 8.5, “Hardware
Looping,” on page 8-18 for more information on hardware loops.
2.5.4 Data Arithmetic Logic Unit (ALU)
The data arithmetic logic unit (ALU) performs all of the arithmetic, logical, and shifting operations on data
operands. The data ALU contains the following components:
•
Three 16-bit data registers (X0, Y0, and Y1)
•
Four 36-bit accumulator registers (A, B, C, and D)
•
One multiply-accumulator (MAC) unit
•
A single-bit accumulator shifter
•
One arithmetic and logical multi-bit shifter
•
One MAC output limiter
•
One data limiter
All in a single instruction cycle, the data ALU can perform multiplication, multiply-accumulation (with
positive or negative accumulation), addition, subtraction, shifting, and logical operations. Division and
normalization operations are provided by iteration instructions. Signed and unsigned multi-precision
arithmetic is also supported. All operations are performed using two’s-complement fractional or integer
arithmetic.
Data ALU source operands can be 8, 16, 32, or 36 bits in size and can be located in memory, in immediate
instruction data, or in the data ALU registers. Arithmetic operations and shifts can have 16-, 32-, or 36-bit
results. The instruction set also supports 8-bit results for some arithmetic operations. Logical operations
are performed on 16- or 32-bit operands and yield results of the same size. The results of data ALU
operations are stored either in one of the data ALU registers or directly in memory.
Chapter 5, “Data Arithmetic Logic Unit,” contains a detailed description of the data ALU.
2.5.5 Address Generation Unit (AGU)
The address generation unit (AGU) performs all of the calculations of effective addresses for data operands
in memory. It contains two address ALUs, allowing up to two 24-bit addresses to be generated every
instruction cycle: one for either the primary data address bus (XAB1) or the program address bus (PAB),
and one for the secondary data address bus (XAB2). The address ALU can perform both linear and modulo
address arithmetic. The AGU operates independently of the other core units, minimizing
address-calculation overhead.
The AGU can directly address 224 (16M) words on the XAB1 and XAB2 buses. It can access 221 (2M)
words on the PAB. The XAB1 bus can address byte, word, and long data operands. The PAB and XAB2
buses can only address words in memory.
The AGU consists of the following registers and functional units:
•
Seven 24-bit address registers (R0–R5 and N)
Freescale Semiconductor
Core Architecture Overview
2-9
Core Architecture Overview
•
Four shadow registers (for R0, R1, N, and M01) on the DSP56800E core, or nine shadow registers
(for all Rn, N, N3, and M01) on the DSP56800EX core
•
A 24-bit dedicated stack pointer register (SP)
•
Two offset registers (N and N3)
•
A 16-bit modifier register (M01)
•
A 24-bit adder unit
•
A 24-bit modulo arithmetic unit
Each of the address registers, R0–R5 and N, can contain either data or an address. All of these registers can
provide an address for the XAB1 and PAB address buses; addresses on the XAB2 bus are provided by the
R3 register. The N offset register can be used either as a general-purpose address register or as an offset or
update value for the addressing modes that support those values. The second 16-bit offset register, N3, is
used only for offset or update values. The modifier register, M01, selects between linear and modulo
address arithmetic.
See Chapter 6, “Address Generation Unit,” for a complete discussion of the AGU.
2.5.6 Program Controller and Hardware Looping Unit
The program controller is responsible for instruction fetching and decoding, interrupt processing, hardware
interlocking, and hardware looping. Actual instruction execution takes place in the other core units, such as
in the data ALU, AGU, or bit-manipulation unit.
The program controller contains the following:
•
An instruction latch and decoder
•
The hardware looping control unit
•
Interrupt control logic
•
A program counter (PC)
•
Two special registers for fast interrupts:
— Fast interrupt return address register (FIRA)
— Fast interrupt status register (FISR)
•
Seven user-accessible status and control registers:
— Two-level-deep hardware stack
— Loop address register (LA)
— Loop address register 2 (LA2)
— Loop count register (LC)
— Loop count register 2 (LC2)
— Status register (SR)
— Operating mode register (OMR)
The operating mode register (OMR) is a programmable register that controls the operation of the core,
including the memory-map configuration. The initial operating mode is typically latched on reset from an
external source; it can subsequently be altered under program control.
The loop address register (LA) and loop count register (LC) work in conjunction with the hardware stack
to support no-overhead hardware looping. The hardware stack is an internal last-in-first-out (LIFO) buffer
that consists of two 24-bit words and that stores the address of the first instruction of a hardware DO loop.
2-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Blocks Outside the Core
When the execution of the DO instruction begins a new hardware loop, the address of the first instruction
in the loop is pushed onto the hardware stack. When a loop finishes normally or an ENDDO instruction is
encountered, the value is popped from the hardware stack. This process allows for one hardware DO loop
to be nested inside another.
The program controller is described in more detail in Chapter 8, “Program Controller.” For more
information on hardware looping, see Section 8.5, “Hardware Looping,” on page 8-18. Information on
interrupt processing is contained in Chapter 9, “Processing States.”
2.5.7 Bit-Manipulation Unit
The bit-manipulation unit performs bitfield operations on data memory words, peripheral registers, and
registers within the DSC core. It is capable of testing, setting, clearing, or inverting individual or multiple
bits within a 16-bit word. The bit-manipulation unit can also test bytes for branch-on-bitfield instructions.
See Chapter 7, “Bit-Manipulation Unit,” for a detailed description of the bitfield unit.
2.5.8 Enhanced On-Chip Emulation (Enhanced OnCE) Unit
The Enhanced On-Chip Emulation (Enhanced OnCE) unit provides a non-intrusive debugging
environment. It is capable of examining and changing core or peripheral registers and memory values. It
can also be used to set breakpoints in program or data memory and step or trace instruction execution.
Refer to Chapter 11, “JTAG and Enhanced On-Chip Emulation (Enhanced OnCE),” for an overview of the
Enhanced OnCE unit’s capabilities.
2.6 Blocks Outside the Core
Devices based on the DSC core contain several additional memory and peripheral blocks. These blocks
provide the functionality that is necessary for a complete working system on a chip. Typical blocks include
those outlined in the following subsections.
2.6.1 Program Memory
Program memory (RAM and/or flash memory) can be provided on-chip with the DSC architecture. The
PAB bus is used to select program memory addresses; instruction fetches are performed over the PDB.
Writes of 16-bit data to program memory are supported over the CDBW bus.
The interrupt and reset vector table can be any size and located anywhere in program memory. The size of
the table is determined by the number of peripherals on the device and by the requirements of the particular
application.
Program memory can be expanded off-chip, with a maximum of 221 (2M) addressable locations.
2.6.2 Data Memory
On-chip data memory (RAM or flash memory) can be implemented on a DSC device. Addresses in data
memory are selected on the XAB1 and XAB2 buses. Byte, word, and long data transfers occur on the
CDBR and CDBW buses. A second 16-bit read operation can be performed in parallel on the XDB2 bus.
Freescale Semiconductor
Core Architecture Overview
2-11
Core Architecture Overview
Peripheral registers are memory mapped into the data memory space. The instruction set optimizes access
to the peripheral registers with a special peripheral addressing mode that makes access to a 64-location
peripheral address space more efficient. Although the peripheral register address range is typically from
$00FFC0 to $00FFFF, individual DSC devices may locate it anywhere in the data memory address space.
The top 12 locations of the peripheral register address space are reserved by the system architecture for the
core, interrupt priority, and bus control configuration registers.
A special addressing mode also exists for the first 64 locations in data memory. Like the peripheral
addressing mode, these locations can be accessed using single word, single cycle instructions. For more
information on these and other addressing modes used to access data memory, see Section 3.6.5.1,
“Absolute Short Address: aa,” on page 3-42.
Data memory can be expanded off-chip, with a maximum of 224 (16M) addressable locations.
2.6.3 Bootstrap Memory
A program bootstrap ROM is typically provided for devices that execute programs from on-chip RAM
instead of ROM. The bootstrap ROM is used to load the application into RAM on reset. The DSC
architecture provides a bootstrapping mode, which fetches instructions from ROM and configures the
RAM as read-only. The operating mode register can then be reprogrammed to fetch instructions from
RAM. See the specific device’s user’s manual for information on bootstrapping mode.
2.6.4 External Bus Interface
An external bus interface extends the data and address buses off the chip, allowing access to external data
and program memory, I/O devices, or other peripherals. The external-bus-interface timing is
programmable, allowing for a wide variety of external devices. These devices can include slow memory
devices, other DSCs, MPUs in master/slave system configurations, or any number of other peripherals.
All three sets of buses (PAB and PDB; XAB1, CDBW, and CDBR; and XAB2 and XDB2) can be
extended to access external devices. Refer to the specific device’s user’s manual for information on
implementing the external bus interface.
2-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 3
Data Types and Addressing Modes
The core contains a large register set and a variety of data types, enabling the efficient implementation of
digital signal processing and general-purpose control algorithms. Byte, word, and long memory accesses
are supported, as are instructions in which a memory access can occur in parallel with an arithmetic
operation. A powerful set of addressing modes also improves execution speed and reduces code size.
3.1 Core Programming Model
The registers in the DSC core programming model are shown in Figure 3-1 on page 3-2. The programming
model is divided into three major blocks in the DSC core.
Registers in the data ALU are used for operations within that block, such as arithmetic operations. More
information on these registers can be found in Section 5.1, “Data ALU Overview and Architecture,” on
page 5-2.
Registers in the address generation unit (AGU) are used as pointers and for operations within that block,
such as computations of effective addresses. More information on these registers can be found in
Section 6.1, “AGU Architecture,” on page 6-1.
Registers in the program control unit are used for instruction fetching, hardware looping, interrupt
handling, status, and control. More information on these registers can be found in Section 8.1, “Program
Controller Architecture,” on page 8-1.
Freescale Semiconductor
Data Types and Addressing Modes
3-1
Data Types and Addressing Modes
Data Arithmetic Logic Unit (ALU)
Data Registers
35 32 31
16 15
0
A
A2
A1
A0
B
B2
B1
B0
C
C2
C1
C0
D
D2
D1
D0
15
0
Y1
Y
Y0
X0
Address Generation Unit (AGU)
Pointer Registers
23
0
Secondary Offset Register
15
0
R0
N3
R1
Modifier Registers
R2
15
0
R3
M01
R4
R5
N
SP
Program Control Unit
20
Program Counter
0
PC
Loop Address
23
Operating Mode Register
and Status Register
(OMR, SR)
15
0
OMR
0
SR
LA
LA2
Hardware Stack
23
Fast Interrupt Status
Register
12
0
0
FISR
HWS0
HWS1
20
Fast Interrupt Return Address
15
Loop Counter
0
FIRA
0
LC
LC2
Figure 3-1. Core Programming Model
Table 3-1 on page 3-3 contains summary descriptions of all the registers in the core.
3-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Core Programming Model
Table 3-1. Core Registers
Unit
Name
Size
(Bits)
Data ALU
Y1
16
Data register (upper 16 bits of 32-bit Y register).
Y0
16
Data register (lower 16 bits of 32-bit Y register).
Y
32
One long register containing two concatenated 16-bit registers, Y1:Y0. This register is pushed to the stack when a fast interrupt is processed.
X0
16
Data register.
A2
4
Accumulator extension register—Bits 35 to 32 of an accumulator.
A1
16
Accumulator most significant product (MSP) register—Bits 31 to 16 of an accumulator.
A0
16
Accumulator least significant product (LSP) register—Bits 15 to 0 of an accumulator.
A10
32
Accumulator long portion—Bits 31 to 0 of an accumulator, containing concatenated registers: A1:A0.
A
36
Accumulator—Contains three concatenated registers: A2:A1:A0.
B2
4
Accumulator extension register—Bits 35 to 32 of an accumulator.
B1
16
Accumulator most significant product (MSP) register—Bits 31 to 16 of an accumulator.
B0
16
Accumulator least significant product (LSP) register—Bits 15 to 0 of an accumulator.
B10
32
Accumulator long portion—Bits 31 to 0 of an accumulator, containing concatenated registers: B1:B0.
B
36
Accumulator—Contains three concatenated registers: B2:B1:B0.
C2
4
Accumulator extension register—Bits 35 to 32 of an accumulator.
C1
16
Accumulator most significant product (MSP) register—Bits 31 to 16 of an accumulator.
C0
16
Accumulator least significant product (LSP) register—Bits 15 to 0 of an accumulator.
Freescale Semiconductor
Description
Data Types and Addressing Modes
3-3
Data Types and Addressing Modes
Table 3-1. Core Registers (Continued)
Unit
Name
Size
(Bits)
Data ALU
C10
32
Accumulator long portion—Bits 31 to 0 of an accumulator, containing concatenated registers: C1:C0.
C
36
Accumulator—Contains three concatenated registers: C2:C1:C0.
D2
4
Accumulator extension register—Bits 35 to 32 of an accumulator.
D1
16
Accumulator most significant product (MSP) register—Bits 31 to 16 of an accumulator.
D0
16
Accumulator least significant product (LSP) register—Bits 15 to 0 of an accumulator.
D10
32
Accumulator long portion—Bits 31 to 0 of an accumulator, containing concatenated registers: D1:D0.
D
36
Accumulator—Contains three concatenated registers: D2:D1:D0.
R0
24
Address register—This register is also shadowed for fast interrupt processing.
R1
24
Address register—This register is also shadowed for fast interrupt processing.
R2
24
Address register—On the DSP56800EX core, this register is also shadowed for
fast interrupt processing.
R3
24
Address register—On the DSP56800EX core, this register is also shadowed for
fast interrupt processing.
R4
24
Address register—On the DSP56800EX core, this register is also shadowed for
fast interrupt processing.
R5
24
Address register—On the DSP56800EX core, this register is also shadowed for
fast interrupt processing.
N
24
Offset register, may also be used as a pointer or index—This register is also
shadowed for fast interrupt processing.
SP
24
Stack pointer.
N3
16
Second read offset register—Sign extended to 24 bits and used as an offset in
updating the R3 pointer in dual read instructions. On the DSP56800EX core, this
register is also shadowed for fast interrupt processing.
M01
16
Modifier register—Used for enabling modulo arithmetic on the R0 and R1
address registers. This register is also shadowed for fast interrupt processing.
AGU
3-4
Description
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data Types
Table 3-1. Core Registers (Continued)
Unit
Name
Size
(Bits)
Program
Controller
PC
21
Program counter—Composed of a dedicated 16-bit register (bits 15-0 of the program counter) as well as 5 bits stored in the upper byte of the status register.
LA
24
Loop address—Contains address of the last instruction word in a hardware DO
loop.
LA2
24
Loop address 2—Saves loop address for outer loop.
HWS
24
Hardware stack—Provides access to the hardware stack as a two-location LIFO
buffer.
FIRA
21
Fast interrupt return address—Saves a 21-bit copy of the return address upon
entering a level 2 fast interrupt service routine.
FISR
13
Fast interrupt status register—Saves a copy of the condition code register, the
stack alignment state, and the hardware looping status upon entering a level 2
fast interrupt service routine.
OMR
16
Operating mode register—Sets up modes for the core.
SR
16
Status register—Contains status, control, and the 5 MSBs of the program counter
register.
LC
16
Loop counter—Contains loop count when hardware looping.
LC2
16
Loop counter 2—Saves loop count for outer loop.
Description
3.2 Data Types
The DSC architecture supports byte (8-bit), word (16-bit), and long-word (32-bit) integer data types. It also
supports word, long-word, and accumulator (36-bit) fractional data types.
Fractional and integer representations differ in the location of the decimal (or binary) point. For fractional
arithmetic, the decimal (or binary) point is always located immediately to the right of the MSP’s most
significant bit. For integer values, the decimal is always located immediately to the right of the value’s
least significant bit. Table 3-2 on page 3-7 shows the location of the decimal point (binary point), bit
weightings, and operand alignment for different fractional and integer representations.
The interpretation of a data value (fractional or integer) is determined by the instruction that uses it. In
some cases, the same instruction can operate on both types of data, with identical results. In others,
different instructions are used for processing fractional numbers and integer numbers. Multiplication, for
example, is performed with the MPY instruction for fractional values and with IMPY.L for integer values.
The following subsections describe the data types and their interpretation.
Freescale Semiconductor
Data Types and Addressing Modes
3-5
Data Types and Addressing Modes
3.2.1 Data Formats
The DSC core supports four types of two’s-complement data formats:
•
Signed integer
•
Unsigned integer
•
Signed fractional
•
Unsigned fractional
Signed and unsigned integer data types are useful for general-purpose computing; they are familiar to
microprocessor and microcontroller programmers. Fractional data types allow for powerful numeric and
digital-signal-processing algorithms to be implemented.
3.2.1.1 Signed Integer
This format is used for processing data as integers. In this format, the N-bit operand is represented using
the N.0 format (N integer bits). Signed integer numbers lie in the following range:
–2[N–1] ≤ SI ≤ [2[N–1] – 1]
This data format is available for bytes, words, and longs. The most negative, signed word that can be
represented is –32,768 ($8000), and the most negative, signed long word is –2,147,483,648 ($8000_0000).
The most positive signed word is 32,767 ($7FFF), and the most positive signed long word is 2,147,483,647
($7FFF_FFFF).
3.2.1.2 Unsigned Integer
Unsigned integer numbers are positive only, and they have nearly twice the magnitude of a signed number
of the same size. Unsigned integer numbers lie in the following range:
0 ≤ UI ≤ [2N – 1]
The binary word is interpreted as having a binary point immediately to the right of the integer’s least
significant bit.
This data format is available for bytes, words, and long words. The most positive, 16-bit, unsigned integer
is 65,535 ($FFFF), and the most positive, 32-bit, unsigned integer is 4,294,967,295 ($FFFF_FFFF). The
smallest unsigned integer number is zero ($0000), regardless of size.
3.2.1.3 Signed Fractional
In this format, the N bit operand is represented using the 1.[N–1] format (1 sign bit, N–1 fractional bits).
Signed fractional numbers lie in the following range:
–1.0 ≤ SF ≤ +1.0 – 2–[N–1]
This data format is available for words and long words. For both word and long-word signed fractions, the
most negative number that can be represented is –1.0, whose internal representation is $8000 (word) or
$80000000 (long word). The most positive word is $7FFF (1.0 – 2–15), and the most positive long word is
$7FFF_FFFF (1.0 – 2–31).
3-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data Types
3.2.1.4 Unsigned Fractional
Unsigned fractional numbers may be thought of as positive only, and they have nearly twice the magnitude
of a signed number with the same number of bits. Unsigned fractional numbers lie in the following range:
0.0 ≤ UF ≤ 2.0 – 2–[N–1]
The binary word is interpreted as having a binary point after the MSB.
This data format is available for words and longs. The most positive, 16-bit, unsigned number is $FFFF, or
{1.0 + (1.0 – 2–[N–1])} = 1.99997. The smallest unsigned fractional number is zero ($0000).
3.2.2 Understanding Fractional and Integer Data
Data in a memory location or register can be interpreted as fractional or integer, depending on a program’s
needs. Table 3-2 shows how a 16-bit value can be interpreted as either fractional or integer, depending on
the location of the binary point.
Table 3-2. Interpretation of 16-Bit Data Values
Integer
Hexadecimal
Representation
Fraction
Binary
Decimal
Binary
Decimal
$7FFF
0111 1111 1111 1111.
32767
0.111 1111 1111 1111
0.99997
$7000
0111 0000 0000 0000.
28672
0.111 0000 0000 0000
0.875
$4000
0100 0000 0000 0000.
16384
0.100 0000 0000 0000
0.5
$2000
0010 0000 0000 0000.
8192
0.010 0000 0000 0000
0.25
$1000
0001 0000 0000 0000.
4096
0.001 0000 0000 0000
0.125
$0000
0000 0000 0000 0000.
0
0.000 0000 0000 0000
0.0
$C000
1100 0000 0000 0000.
–16384
1.100 0000 0000 0000
–0.5
$E000
1110 0000 0000 0000.
–8192
1.110 0000 0000 0000
–0.25
$F000
1111 0000 0000 0000.
–4096
1.111 0000 0000 0000
–0.125
$9000
1001 0000 0000 0000.
–28672
1.001 0000 0000 0000
–0.875
$8000
1000 0000 0000 0000.
–32768
1.000 0000 0000 0000
–1.0
The relationship between the integer interpretation of a 16-bit value and the corresponding fractional
interpretation is:
Fractional Value = Integer Value / (215)
There is a similar relationship between 32-bit integers and fractional values:
Fractional Value = Integer Value / (231)
Table 3-3 on page 3-8 shows how a 36-bit value can be interpreted as either an integer or fractional value,
depending on the location of the binary point.
Freescale Semiconductor
Data Types and Addressing Modes
3-7
Data Types and Addressing Modes
Table 3-3. Interpretation of 36-Bit Data Values
Decimal Representation
Hexadecimal
Representation
36-Bit Integer in
Entire Accumulator
16-Bit Integer in MSP of
Accumulator
Fraction
$2 0000 0000
8589934592
(Overflows)
4.0
$0 8000 0000
2147483648
(Overflows)
1.0
$0 4000 0000
1073741824
16384
0.5
$0 2000 0000
536870912
8192
0.25
$0 0000 0000
0
0
0.0
$F E000 0000
–536870912
–8192
–0.25
$F C000 0000
–1073741824
–16384
–0.5
$F 8000 0000
–2147483648
–32768
–1.0
$E 0000 0000
–8589934592
(Overflows)
–4.0
3.3 Memory Access Overview
The core implements a powerful set of memory-access operations that eases the task of programming the
CPU, decreases program code size, improves efficiency, and decreases the power consumption and
processing power that are required to perform a given task.
Memory is accessed in a variety of ways. Examples include the following types of instructions:
•
Move instructions that access data or program memory
•
Arithmetic or bit-manipulation instructions where one operand is located in data memory
•
Parallel move instructions that perform an operation and move data to or from memory
simultaneously
Each of these memory accesses can be performed both on different sizes of data and with a number of
different addressing modes. Byte, word, and long-word memory accesses, on both signed and unsigned
data, are supported. The provided addressing modes make it easy to access memory quickly and
efficiently.
3.3.1 Move Instruction Syntax
The core supports memory moves to and from both data and program memory, multiple data sizes, and a
variety of addressing modes. Understanding the syntax for each of these options is essential to
understanding and taking advantage of this flexibility.
3-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Memory Access Overview
3.3.1.1 Ordering Source and Destination
The syntax and sequence for all move instructions on the core are as follows: SRC,DST. The source and
destination are separated by a comma, with no spaces either before or after the comma, as shown in
Example 3-1.
Example 3-1. Demonstrating Source and Destination Operands
MOVE.W X0,R3
; X0 is the source operand
; R3 is the destination operand
3.3.1.2 Memory Space Syntax
Each instruction that accesses memory must specify the particular memory address space (data or
program) that is being referenced. Addresses in memory should be prefixed with either X: to indicate the
data memory space or with P: to indicate the program memory space. Table 3-4 shows the address space
prefixes and their use.
Table 3-4. Memory Space Symbols
Symbol
Examples
Description
P:
P:(R2)+
Program memory access
X:
X:(R0)
X:$C000
Data memory access
To avoid confusion, specify all addresses with one of these prefixes. Instructions that do not have this
requirement include jump and branch instructions, whose target addresses always access program memory.
3.3.1.3 Specifying Data Size
The size of data accessed from memory is indicated by a suffix:
•
“.W” suffix—indicates word memory accesses
•
“.L” suffix—indicates long memory accesses
•
“.B” suffix—indicates byte memory accesses
•
“.BP” suffix—indicates byte memory accesses
The difference between the two byte accesses is explained in Section 3.5, “Memory Access and Pointers.”
3.3.2 Instructions That Access Data Memory
Instructions access data memory in one of three ways: using a MOVE instruction with a parameter that
refers to data memory, using an arithmetic instruction that has a parameter in data memory, or using a
bitfield manipulation instruction.
3.3.2.1 Signed and Unsigned Moves
The core provides separate move instructions to ensure that the destination register is zero extended or sign
extended, as appropriate. For unsigned register loads from memory, the letter “U” immediately follows the
“MOVE” portion of the instruction. Unsigned moves are important only when a register is being written
and are not required when a register is being read.
Freescale Semiconductor
Data Types and Addressing Modes
3-9
Data Types and Addressing Modes
Table 3-5 summarizes the various move instructions.
Table 3-5. Suffixes for Move Instructions
Suffix
.W
Examples
Description
MOVE.W X:(R0),A
Load register with 1 word from memory, with sign extension
MOVEU.W X:(R0),R5
Load register with 1 word from memory, with zero extension
.L
MOVE.L X:(R0),A
Load register with 1 long from memory, with sign extension; note
that no extension is performed when moving to 24-bit AGU registers
.B
MOVE.B X:(R0),X0
Load register with 1 byte from memory, with sign extension
U.B
MOVEU.B X:(R0),X0
Load register with 1 byte from memory, with zero extension
.BP
MOVE.BP X:(R0),X0
Load register with 1 byte from memory, with sign extension
MOVEU.BP X:(R0),X0
Load register with 1 byte from memory, with zero extension
U.W
U.BP
3.3.2.2 Moving Words from Memory to a Register
Data ALU registers are typically used to hold signed or fractional data because these data types are the
ones that are most often used in DSC algorithms. In contrast, the AGU and program controller registers
almost always manipulate unsigned values because addresses are always positive integer values. When
loading word values into any of these registers, be sure to use the correct type of MOVE instruction to fit
the use of the value.
For loading data ALU registers, the MOVE.W instruction is most frequently used. This instruction loads
the value into the register and sign extends it correctly. Use the MOVEU.W instruction when loading word
values into the AGU and program controller registers. Using this instruction ensures that the word value is
zero extended to the full register width. Table 3-6 shows how MOVE instructions are typically used to load
registers with 16-bit data.
Table 3-6. Typical 16-Bit-Word Register Loads
Instruction
MOVE.W
Destination
Description
Data ALU registers
Signed words loaded to data ALU registers
MOVEU.W
AGU registers
Unsigned words loaded to AGU pointer registers
MOVEU.W
LA, LC, HWS, OMR, and SR
Unsigned words loaded to other control registers
The MOVE.W instruction is always used to store any register to a word location in memory.
3.3.2.3 Accessing Peripheral Registers
The rules for accessing peripheral registers are the same as the rules for data memory accesses because
peripheral registers are memory mapped in the data memory space.
3-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Memory Access Overview
3.3.3 Instructions That Access Program Memory
The size of data that is accessed from program memory is always 16 bits, so the “.W” suffix is used at all
times. Accesses to program memory follow the same rules that are used for data memory accesses.
Example 3-2 shows examples of valid program memory accesses.
Example 3-2. Program Memory Accesses
MOVE.W P:(R0)+,X0
MOVEU.WP:(R0)+,R3
MOVE.W R2,P:(R0)+
; Read 16-bit signed word from program memory
; Read 16-bit unsigned word from program memory
; Write 16-bit word to memory
3.3.4 Instructions with an Operand in Data Memory
In some arithmetic instructions, one operand is located in data memory. This operand value must be moved
into a temporary register in the data ALU or in the AGU before the instruction can use it. If the instruction
modifies the operand value, the value must then be written back to data memory.
When loaded into a temporary register, the value is aligned and extended in the same way that it would be
if it were placed in a register with a MOVE instruction. For example, the instruction ADD.B X:$4000,A
uses the same method for loading the value at byte address $4000 into a temporary register that the
instruction MOVE.B X:$4000,A uses to load the value into A. For more information on this loading
method, see Section 3.4.1, “Data Alignment in Accumulators.”
Example 3-3 shows some instructions with an operand in data memory.
Example 3-3. Examples of Operands in Memory
;memory location as source operand
ADD.BP X:$4000,A
; Add byte in memory to accumulator
ADD.W X:$2000,A
; Add word in memory to accumulator
ADD.L X:$2000,A
; Add long in memory to accumulator
;memory location with read-modify-write
DEC.BP X:$4000
; Decrement byte in
DEC.W X:$2000
; Decrement word in
DEC.L X:$2000
; Decrement long in
instruction
memory
memory
memory
3.3.5 Parallel Moves
The core implements two additional types of memory moves: the single parallel move and the dual parallel
read. Both are considered “parallel move” instructions and are extremely powerful in DSC algorithms and
numeric computation. Parallel moves are restricted to arithmetic operations in the data ALU. A parallel
move is not permitted, for example, with a JMP or BFSET instruction.
Freescale Semiconductor
Data Types and Addressing Modes
3-11
Data Types and Addressing Modes
3.3.5.1 Single Parallel Move
The single parallel move allows an arithmetic operation and 1 memory word access to be completed with
1 instruction, in 1 clock cycle. For example, all in the same instruction, it is possible to add two numbers
while writing a value from a data ALU register to memory.
Figure 3-2 illustrates a single parallel move that uses 1 program word and executes in 1 instruction cycle.
In this example, the following events occur:
1. Register X0 is added to the A accumulator, and the result is stored back in A.
2. The contents of the Y0 register is stored as a word in data memory at the address contained
in the R1 register.
3. When the memory move is completed, the R1 register is post-updated by the value of
R1+N.
ADD X0,A
Opcode and Operands
Y0,X:(R1)+N
; Example parallel move instruction
Single Parallel Move
(Uses XAB1 and CDBW)
Figure 3-2. Single Parallel Move
3.3.5.2 Dual Parallel Read
With a single instruction, in 1 instruction cycle, the dual parallel read performs an arithmetic operation and
reads two word values from data memory. For example, a dual parallel read can multiply two numbers
while reading two values from data memory to two of the data ALU registers.
Figure 3-3 illustrates a dual parallel read that also uses 1 program word and executes in 1 instruction cycle.
In this example, the following events occur:
1. The original contents of the X0 and Y0 registers are multiplied, and the result is added to and
stored in the A accumulator.
2. The contents of the data memory location pointed to by the R0 register are moved into the
Y1 register. The size of the access is 1 memory word.
3. The contents of the data memory location pointed to by the R3 register are moved into the
X0 register. The size of the access is 1 memory word.
4. After completing the memory moves, the R0 register is post-updated with the value R0+N,
and R3 is decremented.
MAC X0,Y0,A
Opcode and Operands
X:(R0)+N,Y1
Primary Read
(Uses XAB1 and CDBR)
X:(R3)-,X0
Secondary Read
(Uses XAB2 and XDB2)
Figure 3-3. Dual Parallel Read
3-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data Alignment
3.4 Data Alignment
This section discusses how data is aligned in registers and memory.
3.4.1 Data Alignment in Accumulators
Figure 3-4 shows the alignment of different-size data values when they are located in an accumulator. Byte
and word values are located in the FF1 portion of an accumulator, while 32-bit values occupy both the FF1
and FF0 portions.
24 23
35
MOVE.B, MOVE.BP (Signed Byte Move)
SXT.B (Force to Signed Byte)
16 15
Sign Extension
35
MOVEU.B, MOVEU.BP (Unsigned Byte Move)
ZXT.B (Force to Unsigned Byte)
Zero Fill
24 23
16 15
Zero Extension
0
Zero Fill
35
32 31
Sign
Extension
MOVE.W (Signed Word Move)
0
16 15
0
Zero Fill
35
32 31
Sign
Extension
MOVE.L (Signed Long Move)
SXT.L (Force to Signed Long)
0
NOTE: Instructions SXT.B and ZXT.B do not change the LSP of a 32- or 36-bit register destination,
unless the source is a 16-bit register. In this case, the LSP is cleared.
Figure 3-4. Data Alignment in Accumulators
When a byte or word value is moved into an accumulator using one of the MOVE instructions, the FF0
portion is always cleared. Values can be loaded into an accumulator as either signed or unsigned, using the
MOVE or MOVEU mnemonics, respectively. When a signed move is performed, the value is sign
extended through bit 35 of the accumulator. Unsigned moves cause the value to be zero extended.
Move instructions that place a value in an accumulator are shown in Example 3-4.
Example 3-4. Loading Accumulators with Different Data Types
MOVE.B X:(R0+88),A
MOVE.BPX:(R0),A
MOVEU.BX:(R0+3),A
MOVEU.BPX:(R0)+,A
MOVE.W X:(R0),A
MOVE.L X:(R0),A
;
;
;
;
;
;
accumulator
accumulator
accumulator
accumulator
accumulator
accumulator
loaded
loaded
loaded
loaded
loaded
loaded
with
with
with
with
with
with
signed byte
signed byte
unsigned byte
unsigned byte
signed word
signed long
Moves from an accumulator register to memory use only the portions of the accumulator that are identified
in Figure 3-4. Saturation is allowed only on word data types (MOVE.W) and occurs only when an entire
accumulator (A, B, C, or D) is the source operand. See Example 3-5 on page 3-14.
Freescale Semiconductor
Data Types and Addressing Modes
3-13
Data Types and Addressing Modes
Example 3-5. Storing Accumulators with Different Data Types
MOVE.B A1,X:(R0+3)
MOVE.BPA1,X:(R0)
MOVE.W A1,X:(R0)
MOVE.W A,X:(R0)
MOVE.L A10,X:(R0)
;
;
;
;
;
store
store
store
store
store
accumulator
accumulator
accumulator
accumulator
accumulator
byte
byte
word
word
long
(no saturation)
(no saturation)
(no saturation)
(saturation)
(no saturation)
When a MOVE.W or MOVE.L instruction is used to write an accumulator extension register to memory,
the value is sign extended to 16 or 32 bits before it is written.
3.4.2 Data Alignment in Data Registers
The alignment of data within the 16-bit data registers is shown in Figure 3-5. Moves of words (MOVE.W)
from memory (integer or fractional) fill the entire 16-bit register. Signed moves of bytes from memory
(MOVE.B or MOVE.BP) are put in the lower 8 bits of the data register and are sign extended in the upper
8 bits. Unsigned moves are marked with “U” (MOVEU.B or MOVEU.BP) and place zero extension into
the upper 8 bits of the data register.
MOVE.B (Signed Byte Move)
MOVEU.B (Unsigned Byte Move)
8 7
Sign
Extension
15
0
15
0
15
0
8 7
Zero
Extension
MOVE.W (Signed Word Move)
Figure 3-5. Supported Data Types in Data Registers (X0, Y1, Y0)
The Y register, the combination of the Y0 and Y1 registers, can hold a full 32-bit value. It is always read or
written with a long-word move instruction (MOVE.L), and it is never sign extended or zero extended
because a 32-bit value completely fills it.
3.4.3 Data Alignment in 24-Bit AGU and Control Registers
The 24-bit registers in the AGU include the address pointer registers (R0–R5, N, and SP), loop address
registers (LA and LA2), and the hardware stack register (HWS). All values (byte, word, and long word) are
right aligned in the destination register. When an unsigned move instruction is used to load one of these
registers, the value is zero extended to the full register width. Signed moves cause the value to be sign
extended. The placement of data in AGU registers from memory appears in Figure 3-6 on page 3-15.
3-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data Alignment
23
16 15
Sign
Extension
SXTA.B (Force to Signed Byte)
23
16 15
Zero
Extension
ZXTA.B (Force to Unsigned Byte)
23
0
87
Sign
Extension
87
0
Zero
Extension
16 15
0
Sign
Extension
MOVE.W (Signed Word Move)
16 15
Zero
Extension
0
16 15
0
23
MOVEU.W (Unsigned Word Move)
23
MOVE.L (Long Move)
Figure 3-6. Data Alignment in 24-bit AGU Registers
When a MOVE.L instruction is used to write a value to an AGU pointer register, the lower 24 bits are
written and the upper 8 bits are discarded. Using MOVE.L to store a register in memory stores the register
value in the lower 24 bits and fills the upper 8 bits with zero. Sixteen-bit accesses (such as using
MOVE.W) always access the low-order sixteen bits. Although there are no instructions that move bytes to
or from the AGU registers, byte data types can be used with the AGU’s SXTA.B and ZXTA.B
instructions.
Note that accessing the HWS register also pushes and pops values onto the hardware stack. Refer to
Section 8.1.4, “Hardware Stack,” on page 8-3 for details.
3.4.4 Data Alignment in 16-Bit AGU and Control Registers
The alignment of data within the AGU’s 16-bit registers (N3, M01, LC, and LC2) is shown in Figure 3-7.
When these registers are written to with a MOVE.L instruction, the upper 16 bits are discarded. Reading
this register with a MOVE.L instruction places the register contents on the lowest 16 bits, and the upper 16
bits are filled with zero extension. Byte accesses are not supported with these registers.
15
0
15
0
MOVEU.W (Unsigned Word Move)
MOVE.L (Long Move)
Figure 3-7. Data Alignment in 16-Bit AGU Registers
3.4.5 Data Alignment in Memory
The DSC core architecture requires that variables in data memory be aligned to byte, word, or long-word
address boundaries according to the type of data being accessed.
Freescale Semiconductor
Data Types and Addressing Modes
3-15
Data Types and Addressing Modes
3.4.5.1 Byte and Word Addresses
In order to access the different sizes of data that are supported by the core, the instruction set supports two
types of addresses: byte and word. Word addresses can be used to access byte (8-bit), word (16-bit), or
long-word (32-bit) values in memory. Byte addresses are used only for accessing bytes.
In general, the core data memory map can be thought of as 224 contiguous 16-bit words. When word
pointers are used, the address selects for use one of the bytes of the word, the complete word, or two words
when a long-word access is performed. Byte pointers select both the word in the memory map to access
and the desired byte within the word. Figure 3-8 shows the two types of pointers.
Word Pointer
23
16 15
8
7
1
0
1
0
Word Address
Word pointers can access bytes, words, or long words.
Byte Pointer
23
16 15
Word Address
8
7
Upper or Lower Byte Select
Byte pointers can access bytes only.
Figure 3-8. Structure of Byte and Word Addresses
Bits 23–1 of a byte address select the word in the memory map that is to be accessed. The LSB selects the
byte within that word. If the LSB is zero, the lower byte within the word is selected; if the LSB is one, the
upper byte is selected.
Note that, because there are only 23 word-select bits in a byte pointer, byte variables can only be located in
the lower 223 locations in the data memory map.
3.4.5.2 Byte Variable Alignment
Byte variables can be allocated anywhere in the lower half of the 24-bit data memory space, since the
24-bit address used for accessing bytes can only access the lower 223 words in data memory.
Although byte variables can be located at any address, the core assembler only allows byte labels on word
(even) address boundaries. When a label is used to name a byte variable location, the assembler will force
the address of the variable to be even. When a byte is allocated statically or globally using a label, it will
use up 16 bits and will be located in the least significant 8 bits.
Example 3-6 on page 3-17 shows the ds assembler directive being used to allocate 1 word of uninitialized
data memory. The variables thus created are referenced by the word address labels X:BYTVAR1 and
X:BYTVAR2. For each variable, the byte is located in the least significant 8 bits of the word.
3-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Memory Access and Pointers
Example 3-6. Allocation of 2 Bytes Globally
BYTVAR1
BYTVAR2
org
ds
ds
X:$100
1
1
; Allocate 2 bytes at word address = $100
; DS directive allocates 1 word at $100
; DS directive allocates a second word at $102
Arrays of bytes and structures containing bytes correctly allocate a byte as 8 bits rather than 16 bits. An
array of bytes can begin at any byte address. In Example 3-7, a string containing the characters “my world”
is allocated in data memory, where each character is stored in a byte. The string uses 8 bytes of data
memory (four 16-bit words).
Example 3-7. Allocation of a Character String
STRING1
org
dcb
dcb
dcb
dcb
X:$200
‘ym’
‘w ’
‘ro’
‘dl’
;
;
;
;
;
Allocate 8 bytes at word address = $200
‘m’ in lower byte, ‘y’ in upper byte
‘ ’ in lower byte, ‘w’ in upper byte
‘o’ in lower byte, ‘r’ in upper byte
‘l’ in lower byte, ‘d’ in upper byte
Data is organized in the memory map with the least significant byte occupying the lowest address in
memory—so-called little-endian byte ordering. This organization accounts for why the pairs of characters
are reversed in Example 3-7.
3.4.5.3 Word Variable Alignment
Word (16-bit) variables are naturally aligned correctly using word addressing—each address is treated as
referring to a 16-bit data value (see Section 3.5.2, “Accessing Word Values Using Word Pointers,” for
information on word addressing). Data accesses to program memory are always treated as word accesses
and behave the same as word accesses to data memory.
3.4.5.4 Long-Word Alignment
The core architecture requires that long-word variables be allocated on even word addresses, as illustrated
in Figure 3-10 on page 3-19. In general, a long word is accessed using the (lower) even word address.
Long-word accesses using the stack pointer work somewhat differently. See Section 3.5.3, “Accessing
Long-Word Values Using Word Pointers,” for more information.
3.5 Memory Access and Pointers
The DSP56800 core was designed to operate as a word-addressable machine, in which each address
represents one 16-bit word value. The core instruction set has been enhanced to access byte, word, and
long-word memory accesses while maintaining compatibility with the DSP56800 architecture. This
section introduces the concept of word and byte pointers and shows how they are used to access byte,
word, and long values in memory.
3.5.1 Word and Byte Pointers
As described in Section 3.4.5.1, “Byte and Word Addresses,” the core architecture supports both byte and
word addresses. Byte pointers are used to access byte values in memory, while word pointers are used to
access byte, word, or long-word data types in memory.
Freescale Semiconductor
Data Types and Addressing Modes
3-17
Data Types and Addressing Modes
There is no inherent difference between a byte address and a word address—they are both simply 24-bit
quantities. Individual instructions determine how an address is used: an address in an AGU register is
considered a byte pointer when it is used by instructions that expect byte pointers, and it is considered a
word pointer when it is used by instructions expecting word pointers.
Instructions use the “.BP” suffix to indicate that an address register is to be used as a byte pointer. The
“.B”, “.W”, and “.L” suffixes indicate that an address register represents a word pointer. The suffix “.BP”
is also used to indicate that an absolute address is a byte address.
Characteristics of word pointers include the following:
•
They indicate that an address register (R0–R5, N, SP) points to a word address in memory.
•
They can be used for byte, word, or long data memory accesses.
•
Immediate offsets are in bytes (for byte instructions) or in words (for word and long instructions).
Offsets in the N register are expressed in words (for word instructions) or in longs (for long
instructions).
•
They provide efficient accesses to structures.
•
They are fully compatible with the DSP56800 architecture, which only supports word accesses.
Characteristics of byte pointers include the following:
•
They indicate that an address register (R0–R5, N) points to a byte address in data memory.
•
They are used for byte accesses only.
•
Offsets are always in bytes.
•
They can only access the lower half of the 24-bit data memory space (the lowest 223 words).
•
They are extremely efficient for accessing arrays of bytes in memory.
•
They cannot access program memory.
•
Several instructions use address registers as byte pointers, including the following:
— MOVE.BP, MOVEU.BP
— ADD.BP, SUB.BP, CMP.BP
— INC.BP, DEC.BP, NEG.BP
— CLR.BP, TST.BP
NOTE:
The SP register cannot be used as a byte pointer. The SP register is always
used as a stack pointer, so it must always be word aligned for the correct
operation of instructions such as JSR, RTS, and RTI. However, it is
possible to place and access bytes on the stack with the (SP – offset)
addressing modes.
Byte pointers are used exclusively for accessing byte values in data memory. Word pointers, however, can
be used for accessing data of any size: byte, word, or long word. The instruction itself determines if an
address is used as a word or byte pointer.
A word pointer can be converted to a byte pointer by left shifting the value 1 bit, using the ASLA
instruction. Similarly, a byte pointer can be converted to a word pointer by logically right shifting the value
1 bit, using the LSRA instruction (the LSB is lost).
Examples of byte and word pointers are shown in the following sections. More detailed examples of byte
and word pointers appear in Section 6.5, “Word Pointer Memory Accesses,” on page 6-8 and Section 6.6,
“Byte Pointer Memory Accesses,” on page 6-13.
3-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Memory Access and Pointers
3.5.2 Accessing Word Values Using Word Pointers
Word values are accessed from program or data memory with the MOVE.W or MOVEU.W instructions or
with any of the data ALU instructions that access an operand from data memory, such as
ADD.W X:(R0),A or DEC.W X:$C200. Word memory accesses always use an address as a word pointer.
Figure 3-9 shows an example of a word access using a word pointer. The example executes the
MOVE.W A1,X:(R0) instruction. This instruction uses the value in the R0 register, $1000, as the address
in X memory to which the value in A1 ($ABCD) is written.
X Memory
Word
Address
15
$001000
A
R0
0
B
C
D
$001000
Instruction: MOVE.W A1,X:(R0)
Access Size: Word
Figure 3-9. Accessing a Word with a Word Pointer
3.5.3 Accessing Long-Word Values Using Word Pointers
Long-word values are accessed from data memory with the MOVE.L instruction or with any data ALU
instruction that accesses a long-word operand from data memory, such as ADD.L X:$1000,A. Long-word
memory accesses always use a word address. Each long-word value occupies two memory word locations,
as shown in Figure 3-10, and is always aligned on an even word address except when SP is used. The even
address holds the lower word, and the odd address holds the upper word.
Storage of $12345678 in Data Memory
X Memory
Word
Address
15
$001000
0
1
2
3
4
Odd Address: Always holds upper word
5
6
7
8
Even Address: Always holds lower word
Figure 3-10. Correct Storage of 32-Bit Value in Memory
Freescale Semiconductor
Data Types and Addressing Modes
3-19
Data Types and Addressing Modes
Although a long-word value is always located on an even word address boundary, the effective address
used to access the value is not always that even word address. For all registers and addressing modes other
than the stack pointer (SP), the lower even address is used when accessing a long word. In an addressing
mode that uses the stack pointer, the effective address is the odd address that contains the upper word of
the 32-bit value. An attempt to access a long word in any other way generates a misaligned data access
exception. Refer to Section 9.3.3.2.3, “Misaligned Data Access Interrupt,” on page 9-9 for more
information.
Figure 3-11 shows a long-word access using an AGU pointer register. The example executes the
MOVE.L A10,X:(R0) instruction, which uses the value in the R0 register, $1000, as a word address. The
32-bit value contained in the A accumulator, $12345678, is written to this location and the following one.
X Memory
Word
Address
15
$001000
0
1
2
3
4
5
6
7
8
R0
Note: Even Effective Address
$001000
Instruction: MOVE.L A10,X:(R0)
Access Size: Long
Effective Address: Even Value
Figure 3-11. Accessing a Long Word Using an Address Register
Figure 3-12 shows a long-word access using the stack pointer. The example executes the
MOVE.L A10,X:(SP) instruction, which uses the value in the SP register, $1001, as a word address. The
32-bit value contained in the A accumulator, $12345678, is written to addresses $1000 and $1001.
X Memory
Word
Address
15
$001000
SP
0
1
2
3
4
5
6
7
8
Note: Odd Effective Address
$001001
Instruction: MOVE.L A10,X:(SP)
Access Size: Long
Effective Address: Odd Value
Figure 3-12. Accessing a Long Word Using the SP Register
3-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Memory Access and Pointers
Note that, if the stack pointer addressing mode is used, each long value must still be aligned on an even
word address boundary even though the effective address that is used to access the value is odd.
3.5.4 Accessing Byte Values Using Word Pointers
The MOVE.B and MOVEU.B instructions are useful for accessing structures or unions containing bytes as
well as for accessing bytes in a stack frame. These instructions use the address registers (R0–R5, N, SP) as
word pointers and use an offset value to select the upper or lower byte.
Figure 3-13 shows an example of a byte access using a word pointer. The example executes the
MOVE.B A1,X:(R0+3) instruction. In this case, the address contained in R0, $1000, is added to an
immediate offset after the offset has been arithmetically right shifted 1 bit to give the correct word address:
(3>>1) + $1000 = $1001. The least significant bit (LSB) of the immediate offset selects which byte at the
word address is accessed. In this example, the LSB of the immediate offset (3) is set, so the upper byte of
the memory word is accessed. The lowest 8 bits of the A1 register, $CD, are then written to this location.
The lower byte of the memory location $1001 is not modified.
X Memory
Word
Address
15
Byte address: $2003
0
Word Address: $1001
Byte Select: 1 (Upper)
$001001
C
D
X
X
$001000
LSB of Offset
Short Immediate Value “3”
from the Instruction Word
>>1
R0
Instruction: MOVE.B
Access Size: Byte
Byte Selected: Upper
+
$001000
A1,X:(R0+3)
Figure 3-13. Accessing a Byte with a Word Pointer
3.5.5 Accessing Byte Values Using Byte Pointers
Byte pointers are useful for accessing byte variables or arrays of bytes. Instructions that use addresses as
byte pointers include the MOVE.BP and MOVEU.BP instructions as well as data ALU instructions that
access byte operands from data memory using the “.BP” suffix, such as ADD.BP X:$2001,A.
When a byte pointer is used, the value in the selected address register is a byte address. The byte address is
specified using the following:
•
The contents of a register: MOVE.BP X:(R2),A
•
The result of an AGU calculation: MOVE.BP X:(R1+$A701),A
Freescale Semiconductor
Data Types and Addressing Modes
3-21
Data Types and Addressing Modes
•
An absolute address (upper byte): MOVE.BP X:@hb($F000),X0
•
An absolute address (lower byte): MOVE.BP X:@lb(VAR_LABEL),X0
•
An absolute address (upper byte): MOVE.BP X:$108001,X0
Two of the functions in the preceding list are built into the assembler. These functions, described in
Table 3-7, are useful for converting a word address or label into a byte address for instructions that expect
to receive a byte address.
Table 3-7. Useful Built-In Assembler Functions
Assembler Function
Computation
Performed
@hb(value)
(value<<1) + 1
Function is used to generate a byte address from a word
address or label for the upper byte of a word
@lb(value)
(value<<1) + 0
Function is used to generate a byte address from a word
address or label for the lower byte of a word
Comments
NOTE:
The stack pointer register is always used as a word pointer.
Figure 3-14 shows a byte access using a byte pointer. The example executes the MOVE.BP A1,X:(R0)
instruction. The address contained in R0, $2001, is logically right shifted to give the correct word address,
$1000. The LSB of the R0 register selects which byte at the word address is accessed. In this example, the
LSB determines that the upper byte is to be accessed at location $1000. The lowest 8 bits of the A1
register, $CD, are then written to this location. The lower byte of memory location $1000 is not modified.
X Memory
Word
Address
15
0
Word Address: $1000
Byte Select: 1 (Upper)
$001000
C
D
X
X
>>1
LSB
R0
$002001
Instruction: MOVE.BP
Access Size: Byte
Byte Selected: Upper
Byte address: $2001
A1,X:(R0)
Figure 3-14. Accessing a Byte with a Byte Pointer
3-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6 Addressing Modes
Addressing modes specify where the operands for an instruction can be found (in an immediate value, in a
register, or in memory) and provide the exact addresses of the operands. The core instruction set contains a
full set of operand addressing modes, which are optimized for high-performance signal processing as well
as for efficient controller code. All address calculations are performed in the address generation unit to
minimize execution time.
The addressing modes are grouped into categories:
•
Register direct—directly references the registers on the chip as operands
•
Address register indirect—uses an address register as a pointer to reference a location in memory
as an operand
•
Immediate—operand is contained as a value within the instruction itself
•
Absolute—uses the address contained within the instruction itself to reference a location in memory
as an operand
•
Bit reverse (reverse carry)—applies only to address register indirect indexed by N = (Rn)+N
address calculations and to word-sized or longword-sized operands
These addressing modes are referred to extensively in Section 4.4.4, “Instruction Summary Tables,” on
page 4-20.
An effective address in an instruction specifies the addressing mode. In some addressing modes, the
effective address further specifies an address register that points to a location in memory, how the address
is calculated, and how the register is updated.
3.6.1 Addressing Mode Summary
This section contains a series of tables that summarize the addressing modes in the core. The notation used
in these tables to reference AGU registers is summarized in Table 3-8.
Table 3-8. Notation for AGU Registers
Register Field
Registers
Comments
Rn
R0–R5, N, SP
Eight AGU address registers
Rk
R0–R3, N, SP
Six AGU address registers (DSP56800 registers)
RRR
R0–R5, N
Rj
R0, R1, R2, R3
Seven AGU address registers
Four pointer registers available for addressing
Table 3-9 on page 3-24 shows all accessible core registers (register direct).
Table 3-10 on page 3-25 shows data and program memory accesses (address register indirect).
Table 3-11 on page 3-25 shows all immediate addressing modes.
Table 3-12 on page 3-26 shows all absolute addressing modes.
Freescale Semiconductor
Data Types and Addressing Modes
3-23
Data Types and Addressing Modes
Table 3-9. Register-Direct Addressing Mode
Addressing Mode
Any register
Notation in the Instruction Set Summary1
Examples
dd
dddd.L
DD
DDDDD
HHH
HHH.L
HHHH
HHHH.L
HHHHH
fff
F
F1
FF
FFF1
FFF
EEE
Rj
Rn
RRR
SSSS
A, A2, A1, A0
B, B2, B1, B0
C, C2, C1, C0
D, D2, D1, D0
Y, Y1, Y0, X0
R0, R1, R2, R3
R4, R5
SP
N
N3
M01
PC
OMR, SR
LA, LA2, LC, LC2
HWS
FISR, FIRA
1.The register field notations found in the middle column are explained in more detail in Table 4-17 on
page 4-18, Table 4-16 on page 4-17, and Table 4-18 on page 4-19.
3-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
Table 3-10. Address-Register-Indirect Addressing Modes
Notation in the Instruction Set
Summary
Addressing Mode
Examples
Accessing Program Memory
Post-increment
Post-update by offset N
P:(Rj)+
P:(R0)+
P:(Rj)+N
P:(R3)+N
Accessing Data Memory
No update
X:(Rn)
X:(R5)
X:(N)
X:(SP)
Post-increment
X:(Rn)+
X:(R1)+
X:(SP)+
Post-decrement
X:(Rn)–
X:(R5)–
X:(N)–
Post-update by offset N or N3; available for word
accesses only
X:(Rn)+N
X:(R3)+N3
X:(R1)+N
X:(R3)+N3
Indexed by offset N
X:(Rn+N)
X:(R4+N)
X:(SP+N)
Indexed by 3-bit displacement
X:(RRR+x)
X:(SP–x)
X:(R1+7)
X:(N+3)
X:(SP–8)
Indexed by 6-bit displacement—SP register only
X:(SP–xx)
X:(SP+15)
X:(SP–$1E)
Indexed by 16-bit displacement
X:(Rn+xxxx)
X:(R4–97)
X:(N+1234)
X:(SP+$03F7)
Indexed by 24-bit displacement
X:(Rn+xxxxxx)
X:(Rn+$408001)
X:(SP–$10ABCD)
X:(N+$C08000)
Table 3-11. Immediate Addressing Modes
Notation in the Instruction Set
Summary
Addressing Mode
Immediate short data—5-, 6-, and 7-bit (unsigned
and signed)
Immediate data—16-bit (unsigned and signed)
Long immediate data—24- and 32-bit
Freescale Semiconductor
Examples
#xx
#14
#<3
#xxxx
#$369C
#>1234
#xxxxxxxx
#$12345678
#>>$00001234
Data Types and Addressing Modes
3-25
Data Types and Addressing Modes
Table 3-12. Absolute Addressing Modes
Addressing Mode
Absolute short address—6 bit
(direct addressing)
Notation in the Instruction Set
Summary
Examples
X:aa
X:$0002
X:<$02
I/O short address—6 bit
(direct addressing)
X:<<pp
X:<<$FFE3
Absolute address—16-bit
(extended addressing)
X:xxxx
X:$00F001
X:>$C002
X:xxxxxx
X:$18FC04
X:>>$804001
Absolute long address—24-bit
(long extended addressing)
Several of the examples in Table 3-11 on page 3-25 and Table 3-12 demonstrate the use of assembler
forcing operators. These operators can be used in an instruction to force a desired addressing mode, as
shown in Table 3-13.
Table 3-13. Assembler Operator Syntax for Immediate Data Sizes
Desired Action
Forcing Operator Syntax
Example
Force short immediate data
#<xx
#<$07
Force 9-bit immediate data
#>xxx
#>$07
Force 16-bit immediate data
#>xxxx
#>$07
#>xxxxxx
#>$07
Force absolute short address
X:<xx
X:<$02
Force I/O short address
X:<<xx
X:<<$FFE3
Force 16-bit absolute address
X:>xxxx
X:>$02
Force 24-bit absolute long address
X:>xxxxxx
X:>$02
Force short offset
X:(Rn+<x)
X:(SP-<x)
X:(SP-<xx)
X:(SP-<$02)
X:(R0+<3)
Force 16-bit offset
X:(Rn+>xxxx)
X:(SP->$02)
Force 24-bit offset
X:(Rn+>>xxxxxx)
X:(SP->>$02)
Force 24- or 32-bit immediate data
Other assembler forcing operators are available for hardware looping, jump and branch instructions as
shown in Table 3-14.
3-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
Table 3-14. Assembler Operator Syntax for Branch and Jump Addresses
Desired Action
Forcing Operator Syntax
Example
Force 7-bit relative branch offset
<xx
<LABEL1
Force 18-bit relative branch offset
>xxxxx
>LABEL2
Force 21-bit relative branch offset
>>xxxxxx
>>LABEL3
Force 19-bit absolute loop address
>xxxxx
>LABEL4
Force 21-bit absolute loop address
>>xxxxxx
>>LABEL5
Force 19-bit absolute jump address
>xxxxx
>LABEL4
Force 21-bit absolute jump address
>>xxxxxx
>>LABEL5
Freescale Semiconductor
Data Types and Addressing Modes
3-27
Data Types and Addressing Modes
3.6.2 Register-Direct Modes
The register-direct addressing modes specify that each of up to three operands is in either the AGU, data
ALU, or control registers. This type of reference is classified as a register reference.
NOTE:
There can be pipeline dependencies when a data ALU, AGU, or control
register is being accessed. Refer to Section 10.4, “Pipeline Dependencies
and Interlocks,” on page 10-26 to understand dependencies when
accessing these registers.
In Example 3-8, two operands are specified with the register-direct addressing mode. The source operand,
R0, is in the AGU, and the destination operand, X0, is in the data ALU.
Example 3-8. Using the Register-Direct Addressing Mode
MOVE.W R0,X0
; Operands are registers
3.6.3 Address-Register-Indirect Modes
In the address-register-indirect addressing modes, the operand is not the address register itself, but consists
of the contents of the memory location that is pointed to by the address register. Most
address-register-indirect modes also allow the pointer register to be updated in some way. The X:(Rn)addressing mode, for example, accesses the memory location indicated by the address register and then
subtracts one from the register, when the register is used as a word pointer accessing a 16-bit word.
Note that the arithmetic performed can differ depending on the data type. In Example 3-9, the R5 register
is post-incremented by one for a byte or word access and by two for a long memory access.
Example 3-9. Effects of Data Types on AGU Arithmetic
MOVE.BPX:(R5)+,A
MOVE.W X:(R5)+,A
MOVE.L X:(R5)+,A
; Byte Access: R5 <= R5 + 1
; Word Access: R5 <= R5 + 1
; Long Access: R5 <= R5 + 2
In the MOVE.L instruction in Example 3-10, the assembler right shifts the offset of “6” when encoding the
value. When executing the instruction, the AGU unit then left shifts the value in hardware to generate a
displacement of 6 (that is, 3 long words) from the SP. See Section 6.7, “AGU Arithmetic Instructions,” on
page 6-18 for detailed information on how arithmetic is performed for different data types and addressing
modes.
Example 3-10. Effects of Data Types on Address Displacements
MOVE.W X:(SP–3),A
MOVE.L X:(SP–6),A
; Access 3rd word from SP
; Access 3rd long from SP
The type of arithmetic (linear or modulo) used for calculating the effective address in R0 or R1 is specified
in the modifier register (M01) rather than encoded in the instruction. Modulo arithmetic is covered in detail
in Section 6.8, “Linear and Modulo Address Arithmetic,” on page 6-20.
The remainder of this section illustrates each address-register-indirect addressing mode.
3-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.3.1 No Update: (Rn)
The address of the operand is in the address register Rn, N, or SP. The contents of the address register are
unchanged. Figure 3-15 demonstrates this addressing mode.
No Update Example: MOVE.W A1,X:(R2)
After Execution
Before Execution
A2
A
0
A1
1
2
A0
3
35 32 31
4
5
6
16 15
A2
7
8
A
0
0
A1
1
2
35 32 31
R2
X
X
X
$001000
23
5
6
7
8
0
X Memory
0
X
4
16 15
X Memory
15
$001000
A0
3
15
$001000
1
R2
0
0
2
3
4
$001000
23
0
Available for: Byte (Byte Pointer [Word Pointer for SP]), Word, Long
Assembler Syntax: X:(Rn), X:(N), X:(SP)
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-15. Address Register Indirect: No Update
Freescale Semiconductor
Data Types and Addressing Modes
3-29
Data Types and Addressing Modes
3.6.3.2 Post-Increment: (Rn)+
The address of the operand is in the address register Rn, N, or SP. After the operand address is used, it is
incremented and stored in the same address register. When a long 32-bit memory location is accessed, the
pointer is incremented by two.
Figure 3-16 demonstrates this addressing mode.
Post-Increment Example: MOVE.W B0,X:(R2)+
Before Execution
B2
B
A
After Execution
B1
6
5
B0
4
35 32 31
3
F
E
16 15
B2
D
C
B
0
A
B1
6
5
B0
4
35 32 31
3
F
16 15
X Memory
0
15
C
0
0
$002501
X
X
X
X
$002501
X
X
X
X
$002500
X
X
X
X
$002500
F
E
D
C
$002500
23
D
X Memory
15
R2
E
R2
0
$002501
23
0
Available for: Byte (Byte Pointer), Word, Long
Assembler Syntax: X:(Rn)+, X:(N)+, X:(SP)+, P:(Rj)+
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-16. Address Register Indirect: Post-Increment
3-30
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.3.3 Post-Decrement: (Rn)–
The address of the operand is in the address register Rn, N, or SP. After the operand address is used, it is
decremented and stored in the same address register. When a long 32-bit memory location is accessed, the
pointer is decremented by two.
Figure 3-17 demonstrates this addressing mode.
Post-Decrement Example: MOVE.W B,X:(R2)Before Execution
B2
B
0
After Execution
B1
6
5
B0
4
35 32 31
3
F
E
B2
D
16 15
C
B
0
0
B1
6
5
B0
4
35 32 31
3
F
16 15
X Memory
0
15
C
0
0
$004735
X
X
X
X
$004735
6
5
4
3
$004734
X
X
X
X
$004734
X
X
X
X
$004735
23
D
X Memory
15
R2
E
R2
0
$004734
23
0
Available for: Byte (Byte Pointer), Word, Long
Assembler Syntax: X:(Rn)–, X:(N)–, X:(SP)–
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-17. Address Register Indirect: Post-Decrement
Freescale Semiconductor
Data Types and Addressing Modes
3-31
Data Types and Addressing Modes
3.6.3.4 Post-Update by Offset N: (Rn)+N, (R3)+N3
The address of the operand is in the address register Rn, N, or SP. After the operand address is used, the
contents of the offset register (N or N3) are added to the address register and stored in the same address
register. In the addressing update, the contents of the offset register are treated as a signed, 16-bit,
two’s-complement number (the offset register itself remains unchanged). The lower 16 bits of the offset
register are sign extended to 24 bits and used in the addition to the address register. The 24-bit result is then
stored back to the address register.
NOTE:
The upper 8 bits of the N register are ignored in this addressing mode.
Figure 3-18 demonstrates this addressing mode.
Post-Update by Offset N Example: MOVE.W Y1,X:(R2)+N
Before Execution
After Execution
Y1
Y
5
5
Y0
5
31
5
A
A
16 15
Y1
A
A
Y
0
5
5
Y0
5
31
5
A
A
16 15
X Memory
0
0
15
0
$003204
X
X
X
X
$003204
X
X
X
X
$003200
X
X
X
X
$003200
5
5
5
5
+
$003200
23
N
R2
0
$F00004
23
A
X Memory
15
R2
A
0
$003204
23
Sign Extend
from Bit 15
N
0
$F00004
23
0
Available for: Word
Assembler Syntax: X:(Rn)+N, X:(R3)+N3, X:(N)+N, X:(SP)+N, P:(Rj)+N
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-18. Address Register Indirect: Post-Update by Offset N
3-32
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.3.5 Index by Offset N: (Rn+N)
The address of the operand is the sum of the contents of the address register Rn, N, or SP and the contents
of the address offset register N. The content of N is treated as a signed, two’s-complement, 24-bit number.
The contents of the address register and N register are unchanged by this addressing mode. When a long
32-bit memory location is accessed, the N register is left shifted 1 bit before the addition.
Figure 3-19 demonstrates this addressing mode.
Indexed by Offset N Example: MOVE.W A1,X:(R2+N)
Before Execution
A2
A
F
After Execution
A1
E
D
A0
C
35 32 31
B
A
9
A2
8
7
16 15
A
0
F
A1
E
D
A0
C
35 32 31
B
A
16 15
X Memory
0
15
X
X
X
X
$007003
E
D
C
B
$007000
X
X
X
X
$007000
X
X
X
X
$007000
N
R2
0
0
$007000
23
+
$000003
23
7
0
$007003
23
8
X Memory
15
R2
9
N
0
0
$000003
23
0
Available for: Byte (Byte Pointer), Word, Long
Assembler Syntax: X:(Rn+N), X:(N+N), X:(SP+N)
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 0
Figure 3-19. Address Register Indirect: Indexed by Offset N
Freescale Semiconductor
Data Types and Addressing Modes
3-33
Data Types and Addressing Modes
3.6.3.6 Index by 3-Bit Displacement: (RRR+x), (SP–x)
This addressing mode contains the 3-bit immediate displacement within the instruction word. This field is
always one extended to form a negative offset from –1 to –8 when the SP register is used. The field is
always zero extended to form a positive offset from 0 to 7 when R0, R1, R2, R3, R4, R5, or the N register
is used.
Figure 3-20 demonstrates this addressing mode.
Indexed by 3-Bit Displacement Example: MOVE.W A1,X:(R4+3)
Before Execution
A2
A
F
After Execution
A1
E
D
A0
C
35 32 31
B
A
9
A2
8
16 15
7
A
0
F
A1
E
D
A0
C
35 32 31
B
A
16 15
X Memory
0
15
7
0
0
$007003
X
X
X
X
$007003
E
D
C
B
$007000
X
X
X
X
$007000
X
X
X
X
$007000
23
8
X Memory
15
R4
9
R4
0
+
$007000
23
0
Zero Extend for (RRR+x)
One Extend for (SP–x)
3-Bit Immediate Value
from the Instruction Word
Available for: Byte (Word Pointer), Word
Assembler Syntax: X:(Rn+x), X:(N+x), X:(SP–x)
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 0
Figure 3-20. Address Register Indirect: Indexed by 3-Bit Displacement
3-34
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.3.7 Index by 6-Bit Displacement: (SP–xx)
This addressing mode contains the 6-bit immediate displacement within the instruction word. This field is
always one extended to form a negative offset from –1 to –64. When a long 32-bit memory location is
accessed, the 6-bit displacement is left shifted 1 bit before the addition.
Figure 3-21 demonstrates this addressing mode.
Indexed by 6-Bit Displacement Example: MOVE.W A1,X:(SP-32)
Before Execution
A2
A
F
After Execution
A1
E
D
A0
C
35 32 31
B
A
9
A2
8
16 15
7
A
0
F
A1
E
D
A0
C
35 32 31
B
A
16 15
15
0
15
7
0
0
$007020
X
X
X
X
$007020
X
X
X
X
$007000
X
X
X
X
$007000
E
D
C
B
$007020
23
8
X Memory
X Memory
SP
9
SP
0
+
$007020
23
0
One Extend for (SP–xx)
6-Bit Immediate Value
from the Instruction Word
Available for: Word, Long
Assembler Syntax: X:(SP–xx)
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 0
Figure 3-21. Address Register Indirect: Indexed by 6-Bit Displacement
Freescale Semiconductor
Data Types and Addressing Modes
3-35
Data Types and Addressing Modes
3.6.3.8 Index by 16-Bit Displacement: (Rn+xxxx)
This addressing mode contains the 16-bit immediate displacement in the second instruction word. This
second word is treated as a signed, two’s-complement, 16-bit value except when byte pointers (MOVE.BP
and MOVEU.BP) are used, in which case the second word is zero extended. This addressing mode is
available for the move instructions. When a long 32-bit memory location is accessed, the 16-bit
displacement is left shifted 1 bit before the addition. When byte values are accessed, the displacement is
given in bytes.
Figure 3-22 demonstrates this addressing mode.
Indexed by 16-Bit Displacement Example: MOVE.W A1,X:(R2+$10CF)
Before Execution
A2
A
F
After Execution
A1
E
D
A0
C
35 32 31
B
A
9
A2
8
16 15
7
A
0
F
A1
E
D
A0
C
35 32 31
B
A
X Memory
0
15
7
0
0
$0080CF
X
X
X
X
$0080CF
E
D
C
B
$007000
X
X
X
X
$007000
X
X
X
X
$007000
23
8
X Memory
15
R2
9
16 15
R2
0
+
$007000
23
0
Zero Extend for MOVE.BP, MOVEU.BP
One Extend for All Other Instructions
16-Bit Immediate Value
from the Instruction Word
Available for: Byte (Byte and Word Pointer), Word, Long
Assembler Syntax: X:(Rn+xxxx), X:(N+xxxx), X:(SP+xxxx)
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 1
Figure 3-22. Address Register Indirect: Indexed by 16-Bit Displacement
3-36
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.3.9 Index by 24-Bit Displacement: (Rn+xxxxxx)
This addressing mode contains the 24-bit immediate displacement in 2 of the 3 instruction words. The
24-bit displacement is treated as a signed, two’s-complement value. This addressing mode is available for
move instructions. When a long-word (32-bit) memory location is accessed, the 24-bit displacement is left
shifted 1 bit before the addition. When a byte is accessed, the displacement value is given in bytes.
Figure 3-23 demonstrates this addressing mode.
Indexed by 24-Bit Long Displacement Example: MOVE.W A1,X:(R2+$40100F)
Before Execution
A2
A
F
After Execution
A1
E
D
A0
C
35 32 31
B
A
9
A2
8
16 15
7
A
0
F
A1
E
D
A0
C
35 32 31
B
A
7
0
X Memory
0
15
0
$40800F
X
X
X
X
$40800F
E
D
C
B
$007000
X
X
X
X
$007000
X
X
X
X
$007000
23
8
16 15
X Memory
15
R2
9
R2
0
+
$007000
23
0
24-Bit Immediate Value
from the Instruction Word
Available for: Byte (Byte and Word Pointer), Word, Long
Assembler Syntax: X:(Rn+xxxxxx), X:(N+xxxxxx), X:(SP+xxxxxx)
Additional Instruction Execution Cycles: 2
Additional Effective Address Program Words: 2
Figure 3-23. Address Register Indirect: Indexed by 24-Bit Displacement
Freescale Semiconductor
Data Types and Addressing Modes
3-37
Data Types and Addressing Modes
3.6.4 Immediate Address Modes
The immediate address modes do not use an address register to specify an effective address. These modes
specify the value of the operand directly in a field of the instruction.
3.6.4.1 4-Bit Immediate Data: #x
The 4-bit immediate data operand is located in the instruction operation word. In the ADDA instruction,
the 4-bit unsigned value is zero extended to form a 24-bit value. In data ALU shifting instructions, the 4-bit
value is zero extended to form a data ALU operand.
3.6.4.2 5-Bit Immediate Data: #xx
The 5-bit immediate data operand is located in the instruction operation word. When the MOVE.L
instruction is used to write an accumulator, the 5-bit value is sign extended to form a 36-bit value. In data
ALU instructions, the 5-bit value is zero extended to form a data ALU operand.
Figure 3-24 demonstrates this addressing mode.
5-Bit Immediate into Full 36-Bit Accumulator Example: MOVE.L #-4,B
Before Execution
B2
B
X
After Execution
B1
X
35 32 31
X
B0
X
X
X
X
16 15
B2
X
X
B
0
F
B1
F
F
B0
F
35 32 31
F
F
F
F
16 15
C
0
Available for: Long
Assembler Syntax: #xx
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-24. Immediate Addressing: 5-Bit Immediate Data to Accumulator
3.6.4.3 6-Bit Immediate Data: #xx
The 6-bit immediate data operand is located in the instruction operation word. The 6-bit unsigned value is
zero extended to form a 16-bit loop count. It is used by the DO and REP instructions when the loop count
is specified with an immediate value.
3-38
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.4.4 7-Bit Immediate Data: #xx
The 7-bit immediate data operand is located in the instruction operation word. The 7-bit signed value is
sign extended to the appropriate size of the register. It is used by the MOVE.W instruction. Figure 3-25
and Figure 3-26 demonstrate this addressing mode.
7-Bit Immediate Into 24-Bit Address Register Example: MOVE.W #-2,R0
Before Execution
R0
After Execution
R0
XXXXXX
23
0
$FFFFFE
23
0
Available for: Word
Assembler Syntax: #xx
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-25. Immediate Addressing: 7-Bit Immediate Data to Address Register
7-Bit Immediate into 16-Bit Data Register Example: MOVE.W #$0006,X0
Before Execution
X0
After Execution
XXXX
X0
15
$0006
0
15
0
7-Bit Immediate into 36-Bit Accumulator Example: MOVE.W #-58,B
Before Execution
B2
B
X
After Execution
B1
X
X
35 32 31
B0
X
X
X
X
16 15
B2
X
X
B
0
F
B1
F
F
35 32 31
B0
C
6
0
0
0
0
16 15
0
Available for: Word
Assembler Syntax: #xx
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-26. Immediate Addressing: 7-Bit Immediate Data to Data ALU Register
See Section 5.2.3, “Reading and Writing Integer Data to an Accumulator,” on page 5-12 for more details
on correctly loading the accumulator registers.
3.6.4.5 16-Bit Immediate Data: #xxxx
There are two instructions available for writing 16-bit immediate data to an AGU register. The MOVEU.W
instruction loads an AGU register with an unsigned 16-bit value, and the MOVE.L instruction loads an
AGU register with a signed 16-bit value. Figure 3-27 on page 3-40 demonstrates these two instructions.
Freescale Semiconductor
Data Types and Addressing Modes
3-39
Data Types and Addressing Modes
Immediate into 24-Bit Address Register Example: MOVE.L #$FF8001,R5
Before Execution
X
R5
X
23
After Execution
X
X
X
X
16 15
F
R5
0
F
23
8
0
0
1
16 15
0
Immediate into 24-Bit Address Register Example: MOVEU.W #$8001,R5
After Execution
Before Execution
X
R5
X
23
X
X
X
X
16 15
0
R5
0
0
23
8
0
0
1
16 15
0
Available for: Word
Assembler Syntax: #xxxx
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 1
Figure 3-27. Immediate Addressing: 16-Bit Immediate Data to AGU Register
Sixteen-bit immediate data can also be moved to the data ALU registers. When the MOVE.W instruction is
used, the 16-bit value is loaded into the MSP of the accumulator, the value is sign extended into the
extension register, and the LSP is cleared. If the MOVE.L instruction is used, the value is moved into the
LSP of an accumulator and is sign extended through the upper 20 bits. These two cases are shown in
Figure 3-28.
Positive Immediate into 36-Bit Accumulator Example: MOVE.W #$1234,B
Before Execution
B2
B
X
After Execution
B1
X
X
B0
X
35 32 31
X
X
X
B2
X
X
16 15
B
0
0
B1
1
2
B0
3
35 32 31
4
0
0
0
0
16 15
0
Negative Immediate into Full 36-Bit Accumulator Example: MOVE.L #$FFFFB000,B
Before Execution
B2
B
X
After Execution
B1
X
35 32 31
X
B0
X
X
X
X
16 15
B2
X
X
B
0
F
B1
F
F
B0
F
35 32 31
F
B
0
0
0
16 15
0
Available for: Word, Long
Assembler Syntax: #xxxx
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 1
Figure 3-28. Immediate Addressing: 16-Bit Immediate Data to Data ALU Register
Sixteen-bit immediate data is also used to specify the mask for the bit-manipulation instructions.
3-40
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.4.6 32-Bit Immediate Data: #xxxxxxxx
Figure 3-29 demonstrates using 32-bit immediate data to load a register. The immediate data value is
truncated to 24 bits when it is written to one of the 24-bit AGU registers. The value is sign extended when
it is moved to a 36-bit accumulator.
Immediate into 24-Bit Address Register Example: MOVE.L #$12345678,R5
Before Execution
X
R5
X
23
X
X
After Execution
X
X
16 15
3
R5
0
4
23
5
6
7
8
16 15
0
Negative Immediate into 36-Bit Accumulator Example: MOVE.L #$800CF001,B
Before Execution
B2
B
X
After Execution
B1
X
X
B0
X
35 32 31
X
X
X
B2
X
X
16 15
B
0
F
B1
8
0
B0
0
35 32 31
C
F
0
0
1
16 15
0
Positive Immediate into Full 36-Bit Accumulator Example: MOVE.L #$A987,B
Before Execution
B2
B
X
After Execution
B1
X
X
35 32 31
B0
X
X
X
X
16 15
B2
X
X
B
0
0
B1
0
0
35 32 31
B0
0
0
A
16 15
9
8
7
0
Available for: Long
Assembler Syntax: #xxxxxxxx
Additional Instruction Execution Cycles: 2
Additional Effective Address Program Words: 2
Figure 3-29. Immediate Addressing: 32-Bit Immediate Data
3.6.5 Absolute Address Modes
The absolute address modes do not use an address register to specify an effective address. These modes
specify the address of the operand directly in a field of the instruction. This category includes direct
addressing, extended addressing, and immediate data.
Freescale Semiconductor
Data Types and Addressing Modes
3-41
Data Types and Addressing Modes
3.6.5.1 Absolute Short Address: aa
For the absolute short addressing mode, the address of the operand occupies 6 bits in the instruction
operation word and is zero extended to 24 bits. This scheme allows direct access to the first 64 locations in
X memory. No registers are used to form the address of the operand.
Figure 3-30 demonstrates this addressing mode. Note the use of the assembler forcing operator (<) in this
example (see Table 3-13 on page 3-26).
Absolute Short Address Example: MOVE.W R2,X:<$0003
Before Execution
R2
$ABCD
After Execution
R2
15
$ABCD
0
15
0
15
X Memory
X Memory
15
$000003
X
X
X
0
X
$000000
$000003
A
0
B
C
D
$000000
Available for: Word
Assembler Syntax: X:aa
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-30. Absolute Addressing: 6-Bit Absolute Short Address
3-42
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.5.2 I/O Short Address: <<pp
In this addressing mode, the instruction specifies only the 6 LSBs of the effective address. The upper 18
bits are hard-wired to a specific area of memory, which varies depending on the specific implementation of
the chip. This scheme allows efficient access to a 64-location area in data memory, which may be
dedicated to on-chip peripheral registers.
Figure 3-31 demonstrates the I/O short addressing mode. Note the use of the assembler forcing operator
(<<) in this example, indicating that the I/O short addressing mode is in use (see Table 3-13 on page 3-26).
I/O Short Address Example: MOVEU.W X:<<$FFFB,R3
Before Execution
R3
XXXX
After Execution
R3
15
$5678
0
15
0
15
X Memory
X Memory
15
$00FFFF
$00FFFB
0
0
$00FFFF
5
6
7
8
$00FFFB
5
6
7
8
Available for: Word
Assembler Syntax: X:<<pp
Additional Instruction Execution Cycles: 0
Additional Effective Address Program Words: 0
Figure 3-31. Absolute Addressing: 6-Bit I/O Short Address
Freescale Semiconductor
Data Types and Addressing Modes
3-43
Data Types and Addressing Modes
3.6.5.3 16-Bit Absolute Address: xxxx
The address of the operand is zero extended to 24 bits. No registers are used to form the address of the
operand. When a long 32-bit memory location is accessed, the 16-bit absolute address is left shifted 1 bit
before the access occurs.
Figure 3-32 demonstrates the 16-bit absolute addressing mode.
Absolute Address Example: MOVE.W X:$8079,X0
Before Execution
X0
XXXX
After Execution
X0
15
$1234
0
15
0
15
X Memory
X Memory
15
$008079
1
2
3
0
4
$008079
1
0
2
3
4
Available for: Byte (BP), Word, Long
Assembler Syntax: X:xxxx
Additional Instruction Execution Cycles: 1
Additional Effective Address Program Words: 1
Figure 3-32. Absolute Addressing: 16-Bit Absolute Address
3-44
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
3.6.5.4 24-Bit Absolute Address: xxxxxx
This addressing mode requires 2 words of instruction extension. The address of the operand is located in
the extension words. No registers are used to form the address of the operand. When a long 32-bit memory
location is accessed, the 24-bit absolute address is left shifted 1 bit before the access occurs.
Figure 3-33 demonstrates the 24-bit absolute addressing mode.
Absolute Address Example: MOVE.W X:$418003,X0
Before Execution
X0
XXXX
After Execution
X0
15
$1234
0
15
0
15
X Memory
X Memory
15
$418003
1
2
3
0
4
$418003
1
0
2
3
4
Available for: Byte (BP), Word, Long
Assembler Syntax: X:xxxxxx
Additional Instruction Execution Cycles: 2
Additional Effective Address Program Words: 2
Figure 3-33. Absolute Addressing: 24-Bit Absolute Address
3.6.6 Implicit Address Modes
Some instructions make implicit reference to the program counter (PC), software stack, hardware stack,
loop address register (LA), loop counter (LC), or status register (SR). For example, the DO instruction
accesses the LA and LC registers without explicitly referencing them in the instruction. Similarly, the JSR,
RTI, and RTS instructions access the PC, SR, and SP registers without explicitly referencing them in the
instruction. The implied registers and their use are described in the individual instruction descriptions in
Appendix A, “Instruction Set Details.”
3.6.7 Bit-Reverse Address Mode (DSP56800EX Core only)
The bit-reverse address mode, which is also known as reverse carry address mode, is useful for many DSC
applications. It is available only on the DSP56800EX core.
Reverse carry arithmetic is enabled for the R0 and R1 registers through programming the Modifier
Register (M01). Reverse carry addressing is not available for the R2-R5, N, or SP registers. The default
addressing mode for the R0 and R1 registers is linear addressing. Linear arithmetic is enabled for the R0
and R1 registers by programming the M01 register to 0xFFFF. The M01 register is set to 0xFFFF at reset.
Freescale Semiconductor
Data Types and Addressing Modes
3-45
Data Types and Addressing Modes
For both the DSP56800E and DSP56800EX cores, an M01 register with M01[15:14] = 0b00 configures R0
for modulo arithmetic, and an M01 register with M01[15:14] = 0b10 configures both R0 and R1 registers
for modulo arithmetic. For the DSP56800E core, M01 register settings with M01[15:14] = 0b01 or 0b11
(but not M01 = 0xFFFF) are reserved. For the DSP56800EX core, an M01 register setting with
M01[15:0] = 0x4000 (M01[15:14] = 0b01; M01[13:0] = 0x0000) configures R0 for reverse carry
addressing, and an M01 register setting with M01[15:0] = 0xc000 (M01[15:14] = 0b11;
M01[13:0] = 0x0000) configures R0 and R1 for reverse carry addressing.
NOTE:
Modulo address arithmetic applies to certain instructions that operate on
R0 and R1 as well as certain address calculations that use and/or update R0
and R1. In contrast, reverse carry addressing applies only to address
register indirect indexed by N = (Rn)+N address calculations. Also,
reverse carry addressing applies only to word-sized or longword-sized
operands.
Reverse carry address modification is useful for bit-reversed FFT buffers. Reverse carry address
modification is designed to work on a buffer that is aligned on a 0-modulo-(power-of-two size) address
(word or longword). It is designed to start at the beginning of the buffer and step through the entire buffer.
The user has the responsibility to loop through the buffer the correct number of times. Performing reverse
carry address modification beyond this number of times will simply repeat the loop through the buffer.
Reverse carry addressing is performed by doing the (Rn)+N next address calculation and propagating the
carry in the reverse direction modulo the buffer size. That is, the carry is propagated from the MSB of the
buffer address to the LSB.
Reverse carry addressing works as follows. A power-of-two buffer size must be used = 2**k where k < 13.
•
The buffer must be aligned on a 0-modulo (2**k) address.
•
The initial value of Rn is the start of the buffer and N must be 2**(k-1).
When (Rn)+N addressing is used, the next Rn is calculated as follows:
1. The lower-order 14 bits of Rn and N are reversed:
Rn_reversed[13:0] = Rn[0:13]
N_reversed[13:0] = N[0:13]
2. The next Rn with lower-order bits [13:0] reversed is calculated (carry is ignored):
next_Rn_reversed[13:0]
= Rn_reversed[13:0] + N_reversed[13:0]
3. The next Rn is built by reversing the lower-order 14 bits of this result and appending it to
the upper bits of Rn:
next_Rn[23:0] = {Rn[23:14], next_Rn_reversed[0:13]}
The user is responsible for stepping through the buffer for the correct number of times.
Example
In the following example:
•
8 word buffer = 2**3; k = 3; base at 0x00_BDC8
•
initial: Rn = 0x00_BDC8, N = 0x00_0004 = 2**(k-1) = 2**2
•
do 8 iterations; within the buffer the reference order is 0,4,2,6,1,5,3,7
3-46
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Addressing Modes
ITERATION 1: Rn = 0x00_BDC8, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_84EF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_8CEF
current_address[23:0]
= 0x00_BDC8 <- 1st reference
next_Rn[23:0]
= 0x00_BDCC
ITERATION 2: Rn = 0x00_BDCC, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_8CEF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_94EF
current_address[23:0]
= 0x00_BDCC <- 2nd reference
next_Rn[23:0]
= 0x00_BDCA
ITERATION 3: Rn = 0x00_BDCA, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_94EF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_9CEF
current_address[23:0]
= 0x00_BDCA <- 3rd reference
next_Rn[23:0]
= 0x00_BDCE
ITERATION 4: Rn = 0x00_BDCE, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_9CEF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_A4EF
current_address[23:0]
= 0x00_BDCE <- 4th reference
next_Rn[23:0]
= 0x00_BDC9
ITERATION 5: Rn = 0x00_BDC9, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_A4EF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_ACEF
current_address[23:0]
= 0x00_BDC9 <- 5th reference
next_Rn[23:0]
= 0x00_BDCD
ITERATION 6: Rn = 0x00_BDCD, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_ACEF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_B4EF
current_address[23:0]
= 0x00_BDCD <- 6th reference
next_Rn[23:0]
= 0x00_BDCB
ITERATION 7: Rn = 0x00_BDCB, N = 0x00_0004
Rn_reversed[23:0]
= 0x00_B4EF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_BCEF
current_address[23:0]
= 0x00_BDCB <- 7th reference
next_Rn[23:0]
= 0x00_BDCF
ITERATION 8: Rn = 0x00_BDCF, N = 0x00_0004
Freescale Semiconductor
Data Types and Addressing Modes
3-47
Data Types and Addressing Modes
Rn_reversed[23:0]
= 0x00_BCEF
+ N_reversed[23:0]
= 0x00_0800
------------------------------------next_Rn_reversed[23:0] = 0x00_84EF <<< not used, would be 9th reference
current_address[23:0]
= 0x00_BDCF <- 8th reference
next_Rn[23:0]
= 0x00_BDC8 <<< not used, would be 9th reference
3-48
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 4
Instruction Set Introduction
The DSP56800E and DSP56800EX provide a powerful instruction set, enabling the efficient
implementation of digital signal processing and general-purpose computing algorithms. The instruction set
is designed around a large register set, with support for byte, word, and long memory accesses. It also has
special support for powerful DSC capabilities, such as instructions with data moves that occur in parallel
and hardware looping capabilities.
The core architecture contains several functional units that operate in parallel:
•
Data ALU
•
AGU
•
Program controller
•
Bit-manipulation unit
The instruction set is designed to keep each of these units busy in every instruction cycle. Often a single
instruction activates more than one functional unit, enabling the parallel execution of operations. This
arrangement helps to achieve maximum speed, minimum power consumption, and minimum use of
program memory.
This chapter provides an introduction to the core instruction set. The instruction set has been divided into
functional groups, simplifying how to locate the instructions that implement a particular function. The
instructions, their parameters, and their use are summarized at the end of this chapter. For a full description
of each instruction, consult Appendix A, “Instruction Set Details.”
4.1 Instruction Groups
The core instruction set can be divided into several general categories that are based on function:
•
Multiplication—integer and fractional multiplication and multiply-accumulate operations.
•
Arithmetic—all arithmetic operations other than multiplication.
•
Shifting—shift and rotate operations.
•
Logic—Boolean logic functions, such as AND, OR, and NOT.
•
AGU arithmetic—address calculation operations.
•
Bit manipulation—instructions for manipulating values at the bit level.
•
Looping—instructions that support iterative loops.
•
Move—data movement operations.
•
Program control—instructions that control execution flow.
Each instruction group is described in the following sections.
Freescale Semiconductor
Instruction Set Introduction
4-1
Instruction Set Introduction
4.1.1 Multiplication Instructions
These instructions perform all of the multiplication operations within the data ALU. Optional data
transfers (parallel moves) can be specified with some of the multiplication instructions. These transfers
allow new data to be pre-fetched for use in instructions that follow, or they allow results calculated by
previous instructions to be stored.
Multiplication instructions execute in 1 instruction cycle. They may affect one or more of the condition
code register bits.
Table 4-1 lists the multiplication instructions available on both the DSP56800E core and the DSP56800EX
core.
Table 4-1. Multiplication Instructions
Instruction
Parallel
Move?
IMAC.L
—
Signed integer multiply-accumulate with full precision
IMACUS
—
Unsigned/signed integer multiply-accumulate with full precision
IMACUU
—
Unsigned/unsigned integer multiply-accumulate with full precision
IMPY.L
—
Signed integer multiply with full precision
IMPY.W
—
Signed integer multiply with integer result
IMPYSU
—
Signed/unsigned integer multiply with full precision
IMPYUU
—
Unsigned/unsigned integer multiply with full precision
MAC
Yes
Signed fractional multiply-accumulate
MACR
Yes
Signed fractional multiply-accumulate and round
MACSU
—
Signed/unsigned fractional multiply-accumulate
MPY
Yes
Signed fractional multiply
MPYR
Yes
Signed fractional multiply and round
MPYSU
—
Signed/unsigned fractional multiply
Description
Table 4-2 lists additional 32-bit multiplication instructions available on the DSP56800EX core.
Table 4-2. Additional 32-Bit DSP56800EX Multiplication Instructions
4-2
Instruction
Parallel
Move?
IMAC32
—
Integer multiply-accumulate 32 bits
IMPY32
—
Integer multiply 32 bits x 32 bits → 32 bits
IMPY64
—
Integer multiply 32 bits x 32 bits → 64 bits
IMPY64UU
—
Unsigned integer multiply 32bits x 32 bits → 64 bits
MAC32
—
Fractional multiply-accumulate 32 bits x 32 bits → 32 bits
Description
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Groups
Table 4-2. Additional 32-Bit DSP56800EX Multiplication Instructions (Continued)
Instruction
Parallel
Move?
MPY32
—
Fractional multiply 32 bits x 32 bits → 32 bits
MPY64
—
Fractional multiply 32 bits x 32 bits → 64 bits
Description
4.1.2 Arithmetic Instructions
This group consists of all non-multiplication mathematical instructions. These instructions can operate on
values located either in registers or in memory, although using register-based operands allows data move
operations to be executed in parallel.
The arithmetic instructions typically execute in 1 instruction cycle, although instructions that use more
complex addressing modes may take longer. The instructions may affect one or more of the condition code
register bits.
Table 4-3 on page 4-3 lists the arithmetic instructions.
Table 4-3. Arithmetic Instructions
Instruction
Parallel
Move?
ABS
Yes
ADC
—
ADD
Yes
ADD.B
—
Add byte value from memory to register
ADD.BP
—
Add byte value from memory to register
ADD.L
—
Add long value from memory (or immediate) to register
ADD.W
—
Add word value from memory (or immediate) to register
CLR
Yes
CLR.B
—
Clear a byte value in memory
CLR.BP
—
Clear a byte value in memory
CLR.L
—
Clear a long value in memory
CLR.W
—
Clear a word value in memory or in a register
CMP
Yes
Compare a word value from memory (or immediate) with an accumulator; also
compare two registers, where the second is always an accumulator; comparison
done on 36 bits
CMP.B
—
Compare the byte portions of two registers or an immediate with the byte portion
of a register; comparison done on 8 bits
CMP.BP
—
Compare a byte value from memory with a register; comparison done on 8 bits
Freescale Semiconductor
Description
Absolute value
Add long with carry
Add two registers
Clear a 36-bit register value
Instruction Set Introduction
4-3
Instruction Set Introduction
Table 4-3. Arithmetic Instructions (Continued)
Instruction
Parallel
Move?
CMP.L
—
Compare a long value from memory (or an immediate value) with a register; also
compare the long portions of two registers; comparison done on 32 bits
CMP.W
—
Compare a word value from memory (or immediate) with a register; also compare the word portions of two registers; comparison done on 16 bits
DEC.BP
—
Decrement byte in memory
DEC.L
—
Decrement an accumulator or a long in memory
DEC.W
Yes
DIV
—
Divide iteration
INC.BP
—
Increment byte in memory
INC.L
—
Increment an accumulator or a long in memory
INC.W
Yes
Increment upper word of accumulator, word register, or a word in memory
NEG
Yes
Negate an accumulator
NEG.BP
—
Negate byte in memory
NEG.L
—
Negate a long word in memory
NEG.W
—
Negate a word in memory
NORM
—
Normalize
RND
Yes
Round
SAT
Yes
Saturate a value in an accumulator and store in destination
SBC
—
SUB
Yes
SUB.B
—
Subtract byte value from memory to register
SUB.BP
—
Subtract byte value from memory to register
SUB.L
—
Subtract long value from memory to register
SUB.W
—
Subtract word value from memory (or immediate) to register
SUBL
Yes
SXT.B
—
Sign extend a byte value in a register and store in destination
SXT.L
—
Sign extend a value in an accumulator and store in destination
SWAP
—
Swap R0, R1, N, and M01 registers—as well as R2, R3, R4, R5, and N3 registers for the DSP56800EX core—with corresponding shadows
Tcc
—
Conditionally transfer one or two registers to other registers
4-4
Description
Decrement upper word of accumulator, word register, or a word in memory
Subtract long with carry
Subtract two registers
Shift accumulator left and subtract word value
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Groups
Table 4-3. Arithmetic Instructions (Continued)
Instruction
Parallel
Move?
TFR
Yes
Transfer data ALU register to an accumulator
TST
Yes
Test a 36-bit accumulator
TST.B
—
Test byte in memory or in a register
TST.BP
—
Test byte in memory
TST.L
—
Test an accumulator or a long in memory
TST.W
—
Test a word in memory or in a register
ZXT.B
—
Zero extend a byte value in an register and store in destination
Freescale Semiconductor
Description
Instruction Set Introduction
4-5
Instruction Set Introduction
4.1.3 Shifting Instructions
The shifting instructions are used to perform shift and rotate operations within the data ALU. They
generally execute in 1 instruction cycle, except for the multi-bit shift instructions (ASLL.L, ASRR.L, and
LSRR.L), which execute in 2 cycles. These instructions may affect one or more of the condition code
register bits.
Table 4-4 lists the shifting instructions.
Table 4-4. Shifting Instructions
Instruction
Parallel
Move?
ASL1
Yes
ASL16
—
Arithmetic left shift a register or accumulator by 16 bits
ASL.W
—
Arithmetic shift left a 16-bit register (shift register 1 bit)
ASLL.L
—
Arithmetic multi-bit shift left a long value
ASLL.W
—
Arithmetic multi-bit shift left a word value
ASR
Yes
Arithmetic shift right (shift register 1 bit)
ASR16
—
Arithmetic right shift a register or accumulator by 16 bits
ASRAC
—
Arithmetic multi-bit shift right with accumulate
ASRR.L
—
Arithmetic multi-bit shift right a long value
ASRR.W
—
Arithmetic multi-bit shift right a word value
LSL.W
—
Logical shift left a word-sized register
LSR.W
—
Logical shift right (shift word-sized register 1 bit)
LSR16
—
Logical right shift a register or accumulator by 16 bits
LSRAC
—
Logical multi-bit shift right with accumulate
LSRR.L
—
Logical multi-bit shift right a long value
LSRR.W
—
Logical multi-bit shift right a word value
ROL.L
—
Rotate left on long register
ROL.W
—
Rotate left on word register
ROR.L
—
Rotate right on long register
ROR.W
—
Rotate right on word register
Description
Arithmetic shift left (shift register 1 bit)
1.ASL should not be used to shift the 16-bit X0, Y0, and Y1 registers because the condition codes might not
be calculated as expected. The ASL.W instruction should be used instead.
4-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Groups
4.1.4 Logical Instructions
The instructions in this group perform Boolean logic operations. Optional data transfers are not permitted
with logical instructions, except with the EOR.L instruction, which permits a single parallel move. These
instructions execute in 1 cycle.
Table 4-5 lists the logical instructions.
Table 4-5. Logical Instructions
Instruction1
Parallel
Move?
AND.L
—
Logical AND on long registers
AND.W
—
Logical AND on word registers
ANDC
—
Logical AND immediate data on word in memory
CLB
—
Count leading zeros or ones
EOR.L
Yes
Logical exclusive OR on long registers
EOR.W
—
Logical exclusive OR on word registers
EORC
—
Logical exclusive OR immediate data on word in memory
NOT.W
—
Logical complement on word registers
NOTC
—
Logical complement on word in memory
OR.L
—
Logical OR on long registers
OR.W
—
Logical OR on word registers
ORC
—
Logical OR immediate data on word in memory
Description
1.Note that ANDC, EORC, ORC, and NOTC are not true instructions, but are aliases to bit-manipulation instructions that perform the same function. See Section 4.2.1, “The ANDC, EORC, ORC, and NOTC Aliases,”
for more information.
4.1.5 AGU Arithmetic Instructions
These instructions perform all of the address-calculation arithmetic operations within the address
generation unit. AGU arithmetic instructions typically use AGU registers for operands, although some
instructions can operate on immediate data. Only the CMPA, CMPA.W, DECTSTA, TSTA.B, TSTA.W,
TSTA.L, and TSTDECA.W instructions modify the condition code register bits.
No optional data transfers (parallel moves) can be specified with the AGU arithmetic instructions.
Arithmetic instructions typically execute in 1 instruction cycle, although some of the operations may take
additional cycles depending on the operand addressing mode.
Table 4-6 on page 4-8 lists the AGU arithmetic instructions.
Freescale Semiconductor
Instruction Set Introduction
4-7
Instruction Set Introduction
Table 4-6. AGU Arithmetic Instructions
Instruction
ADDA
ADDA.L
ALIGNSP
Add register or immediate to AGU register
Add to AGU register with 1 bit left shift of source operand
Save old value of stack pointer onto stack, aligning SP for long memory accesses before performing the save
ASLA
Arithmetic 1 bit left shift an AGU register
ASRA
Arithmetic 1 bit right shift an AGU register
CMPA
Compare two AGU registers; comparison done on 24 bits
CMPA.W
Compare two AGU registers; comparison done on 16 bits
DECA
Decrement an AGU register by one
DECA.L
Decrement an AGU register by two
DECTSTA
Decrement and test an AGU register
LSRA
Logical 1 bit right shift an AGU register
NEGA
Negate an AGU register
SUBA
Subtract register or immediate from AGU register
SXTA.B
Sign extend a byte value in an AGU register
SXTA.W
Sign extend a word value in an AGU register
TFRA
Transfer one AGU register to another
TSTA.B
Test the byte portion of an AGU register
TSTA.L
Test the long portion of an AGU register
TSTA.W
Test the word portion of an AGU register
TSTDECA.W
4-8
Description
Test and decrement the word portion of an AGU register
ZXTA.B
Zero extend a byte value in an AGU register
ZXTA.W
Zero extend a word value in an AGU register
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Groups
4.1.6 Bit-Manipulation Instructions
The bit-manipulation instructions are used to test or modify a set of one or more bits, called a bitfield,
within a word. They can operate on any data memory location, peripheral, or register. The carry bit in the
status register is the only condition code affected by these instructions. They all execute in 2, 3, or 4
instruction cycles.
For similar instructions that change execution flow based on a bitfield test, see Section 4.1.9, “Program
Control Instructions.”
Table 4-7 lists the bit-manipulation instructions available on both the DSP56800E core and the
DSP56800EX core.
Table 4-7. Bitfield Instructions
Instruction
Description
BFCHG
Bitfield test and change
BFCLR
Bitfield test and clear
BFSET
Bitfield test and set
BFTSTH
Bitfield test for on condition
BFTSTL
Bitfield test for off condition
Table 4-8 lists the additional bit-manipulation instruction available on the DSP56800EX core.
Table 4-8. Additional DSP56800EX Bitfield Instruction
Instruction
Description
BFSC
Bitfield test and set/clear
Using the bit-manipulation instructions to modify AGU registers (Rn, N, SP, or M01) can result in pipeline
dependencies. See Section 10.4.2, “AGU Pipeline Dependencies,” on page 10-28 for more information.
4.1.7 Looping Instructions
The looping instructions are used to perform program looping with minimal overhead. The core
architecture supports efficient hardware looping on a single instruction (using REP) or on a block of
instructions (using DO). Using these instructions can dramatically increase the performance of iterative
algorithms. For a full discussion of hardware looping and the looping instructions, see Section 8.5,
“Hardware Looping,” on page 8-18.
Table 4-9 lists the loop instructions.
Table 4-9. Looping Instructions
Instruction
DO
DOSLC
Freescale Semiconductor
Description
Load LC register with unsigned 16-bit loop count and start hardware loop
Start hardware loop with signed 16-bit loop count already in LC register
Instruction Set Introduction
4-9
Instruction Set Introduction
Table 4-9. Looping Instructions (Continued)
Instruction
ENDDO
REP
Description
Terminate current hardware DO loops
Repeat immediately following instruction
4.1.8 Move Instructions
The move instructions transfer data between core registers and memory or peripherals, or between two
memory or peripheral locations. Move instructions that write an accumulator register to memory or a
peripheral can also automatically saturate, limiting the value written. In addition to the following move
instructions, there are also parallel moves that can be used simultaneously with many of the arithmetic
instructions. The parallel moves appear in Table 4-43 on page 4-49 and Table 4-44 on page 4-50 and are
discussed in detail in Section 3.3.5, “Parallel Moves,” on page 3-11 and in Appendix A, “Instruction Set
Details.”
Table 4-10 lists the move instructions.
Table 4-10. Move Instructions
Instruction
Description
MOVE.B
Move (signed) byte using word pointers and byte addresses
MOVE.BP
Move (signed) byte using byte pointers and byte addresses
MOVEU.B
Move unsigned byte using word pointers and byte addresses
MOVEU.BP
Move unsigned byte using byte pointers and byte addresses
MOVE.L
Move long using word pointers
MOVE.W
Move (signed) word using word pointers and word addresses
(data or program memory)
MOVEU.W
Move unsigned word using word pointers and word addresses
(data or program memory)
Writing AGU registers (Rn, N, SP, or M01) with a MOVE instruction can result in an execution pipeline
stall. See Section 10.4.2, “AGU Pipeline Dependencies,” on page 10-28 for more information.
4.1.9 Program Control Instructions
The program control instructions include branches, jumps, conditional branches, conditional jumps, and
other instructions that affect the program counter and software stack. Also included in this instruction
group are the STOP and WAIT instructions, which place the DSC chip in a low-power state.
Table 4-11 lists the change-of-flow instructions.
4-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Groups
Table 4-11. Program Control and Change-of-Flow Instructions
Instruction
Description
Bcc
Branch conditionally
BRA
Branch
BRAD
Delayed branch
BRCLR
Branch if selected bits are clear
BRSET
Branch if selected bits are set
BSR
FRTID
ILLEGAL
Branch to subroutine
Delayed return from fast interrupt
Generate an illegal instruction exception
Jcc
Jump conditionally
JMP
Jump
JMPD
Delayed jump
JSR
Jump to subroutine
RTI
Return from interrupt
RTID
Delayed return from interrupt
RTS
Return from subroutine
RTSD
SWI
SWI #<0–2>
SWILP
Delayed return from subroutine
Software interrupt at highest priority level
Software interrupt at specified priority level
Software interrupt at lowest priority level
See Section 7.5, “Programming Considerations,” on page 7-6 for other program control instructions that
can be synthesized from existing core instructions. For information on the delayed program control
instructions (BRAD, FRTID, JMPD, RTID, and RTSD), see Section 4.3, “Delayed Flow Control
Instructions.”
Table 4-12 lists the miscellaneous program control instructions.
Table 4-12. Miscellaneous Program Control Instructions
Instruction
Description
DEBUGEV
Generate debug event
DEBUGHLT
Enter debug mode
NOP
No operation
STOP
Stop processing (lowest power standby)
Freescale Semiconductor
Instruction Set Introduction
4-11
Instruction Set Introduction
Table 4-12. Miscellaneous Program Control Instructions (Continued)
Instruction
WAIT
Description
Wait for interrupt (low power standby)
4.2 Instruction Aliases
The DSC core assembler provides a number of additional, useful instruction mnemonics that are actually
aliases to other instructions. Each of these instructions is mapped to one of the core instructions and
dis-assembles as such.
4.2.1 The ANDC, EORC, ORC, and NOTC Aliases
The core instruction set does not support logical operations using 16-bit immediate data. It is possible to
achieve the same result, however, using the bit-manipulation instructions. To simplify implementing these
operations, the core assembler provides the following operations:
•
ANDC—logically AND a 16-bit immediate value with a destination
•
EORC—logically exclusive OR a 16-bit immediate value with a destination
•
NOTC—take the logical one’s-complement of a 16-bit destination
•
ORC—logically OR a 16-bit immediate value with a destination
These operations are not new instructions, but aliases to existing bit-manipulation instructions. They are
mapped as indicated in Table 4-13.
Table 4-13. Aliases for Logical Instructions with Immediate Data
Operands
Remapped
DSP56800E/
DSP56800EX
Instruction
Operands
ANDC
#xxxx,DST
BFCLR
#xxxx,DST
EORC
#xxxx,DST
BFCHG
#xxxx,DST
NOTC
DST
BFCHG
#$FFFF,DST
ORC
#xxxx,DST
BFSET
#xxxx,DST
Desired
Instruction
Note that for the ANDC instruction, a one’s-complement of the mask value is used when remapping to the
BFCLR instruction. For the NOTC instruction, all bits in the 16-bit mask are set to one.
In Example 4-1, a logical OR operation is performed on an immediate value with a location in memory.
Example 4-1. Logical OR with a Data Memory Location
ORC
#$00FF,X:$400; Set all bits of lower byte in X:$400
The assembler translates this instruction into BFSET #$00FF,X:$400, which performs the same
operation. If the assembled code is later dis-assembled, the instruction appears as a BFSET instruction.
4-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Aliases
4.2.2 Instruction Operand Remapping
The core assembler performs a few additional mapping functions, either to allow an alternative syntax for
certain instructions or to simplify the addressing mode used by an instruction. These remapping functions
are discussed in the following sections.
4.2.2.1 Duplicate Operand Remapping
Several instructions, such as the ADDA, SAT, and ZXT.B instructions, allow different source and
destination register operands to be specified. Often, however, the source and destination registers are the
same. For situations when they are the same, the core assembler provides an alternate syntax in which the
operand is only specified once. Table 4-14 lists the standard and duplicate-operand syntaxes for these
instructions.
Table 4-14. Instructions with Alternate Syntax
Standard Syntax
Operation
ADDA
Alternate Syntax
Operands
#xxxx,Rn,Rn
Operation
ADDA
#xxxxxx,Rn,Rn
ADDA.L
#xxxx,Rn,Rn
Operands
#xxxx,Rn
#xxxxxx,Rn
ADDA.L
#xxxxxx,Rn,Rn
#xxxx,Rn
#xxxxxx,Rn
ASL16
FFF,FFF
ASL16
FFF
ASLA
Rn,Rn
ASLA
Rn
ASR16
FFF,FFF
ASR16
FFF
LSR16
FFF,FFF
LSR16
FFF
SAT
FF,FFF
SAT
FF
SXT.B
FFF,FFF
SXT.B
FFF
SXT.L
FF,FFF
SXT.L
FF
ZXT.B
FFF,FFF
ZXT.B
FFF
Note that the alternate syntax is merely an alias to the regular instruction syntax. When dis-assembled, the
instruction appears with the standard syntax, with the register operand repeated.
4.2.2.2 Addressing Mode Remapping
When an instruction operand uses the index-by-6-bit-displacement or index-by-3-bit-displacement
addressing modes, the core assembler examines the effective address calculation to see if the operand can
be mapped to one that uses a simpler addressing mode. Specifically, when the assembler detects
occurrences of the following addressing modes, it remaps them:
•
X:(SP–xx) where the value of the 6-bit offset is “0”
•
X:(SP–x) where the value of the 3-bit offset is “0”
Freescale Semiconductor
Instruction Set Introduction
4-13
Instruction Set Introduction
In both cases, the operand addressing mode is remapped to the X:(SP) addressing mode.
4.3 Delayed Flow Control Instructions
One particular class of instructions merits additional attention: the delayed flow control instructions. These
instructions are designed to increase throughput by eliminating execution cycles that are wasted when
program flow changes.
An instruction that affects normal program flow (such as branch or jump instruction) requires 2 or 3
additional instruction cycles to flush the execution pipeline. The program controller stops fetching
instructions at the current location and begins to fill the pipeline from the target address. The execution
pipeline is said to stall while this switch occurs. The additional cycles required to flush the pipeline are
reflected in the total cycle count for each change-of-flow instruction. A special group of instructions
referred to as “delayed” instructions provide a mechanism for executing useful tasks during these normally
wasted cycles.
4.3.1 Using Delayed Instructions
The delayed instructions use the execution pipeline more efficiently by executing one or more of the
instructions following the delayed instruction before execution is switched to the target address. The
number of instructions is limited by the number of delay slots that are available with a given delayed
instruction, where each delay slot consists of 1 program word.
The delayed instructions, and the number of delay slots for each, are shown in Table 4-15.
Table 4-15. Delayed Instructions
Delayed Instructions
Number of Delay Slots
BRAD
2
JMPD
2
RTID
3
RTSD
3
FRTID
2
The delay slots following each of these instructions must be filled with exactly the same number of
instruction words as there are delay slots. If not all delay slots can be filled with valid instructions, then
each unused delay slot must be filled with an NOP instruction. If a pipeline dependency occurs due to
instructions executed in the delay slots, the appropriate amount of interlock cycles are inserted by the core,
reducing the number of delay-slot cycles that are available for instructions by the same number of cycles.
See Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26 for more information.
Example 4-2 shows a code fragment that reverses the contents of a buffer in memory that starts at the
address contained in R0. Note the BRA instruction that is used to return to the top of the loop. Due to the
design of the execution pipeline, the pipeline stalls for 2 cycles while control is transferred back to the top
of the loop.
4-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Delayed Flow Control Instructions
Example 4-2. Code Fragment with Regular Branch
ADDA
#buflen-1,R0,R1
; put end of buffer in R1
SWAP_LOOP
CMPA R0,R1
BLE
DONE
MOVE.W X:(R0)+,X0
MOVE.W X0,X:(R1)BRA
SWAP_LOOP
;
;
;
;
;
;
check if done yet
if R0 >= R1, we’re done
perform the swap
"
"
"
branch back to top of loop
pipeline stalls for 2 cycles
...
; subsequent code...
DONE
A more efficient way to implement the reversal algorithm is to use the BRAD instruction instead of BRA.
Example 4-3 on page 4-15 shows BRAD being used, with the code rearranged appropriately. By using
BRAD, we can execute the two MOVE.W instructions during the 2 cycles that would normally be wasted
due to the branch.
Example 4-3. Code Fragment with Delayed Branch
ADDA
#buflen-1,R0,R1
; put end of buffer in R1
SWAP_LOOP
CMPA R0,R1
BLE
DONE
BRAD SWAP_LOOP
MOVE.W X:(R0)+,X0
MOVE.W X0,X:(R1)-
;
;
;
;
check if done yet
if R0 >= R1, we’re done
delayed branch to top of loop
swap occurs in the delay slots!
...
; subsequent code...
DONE
Similar strategies can be used on subroutines and interrupt handlers, where employing the RTSD and
RTID instructions can eliminate the wasted cycles associated with the RTS and RTI instructions.
4.3.2 Delayed Instruction Restrictions
Not all instructions are allowed in delay slots. The following instructions cannot be executed in a delay
slot. The assembler detects these instructions and flags them as illegal.
•
DO, DOSLC, REP, ENDDO
•
JMP, JMPD, Jcc, JSR, BRA, BRAD, Bcc, BSR, RTS, RTSD, RTI, RTID, FRTID
•
ADD.W with the following operands: ADD.W EEE,X:(SP-xx)
•
SWILP, SWI #0, SWI #1, SWI #2, SWI
•
STOP, WAIT
•
SWAP SHADOWS
•
Move instructions that access program memory
•
Any move clear or test instruction that accesses the SP, N3, M01, LA, LA2, LC, LC2, HWS, SR, or
OMR registers
•
The BFCHG, BFCLR, and BFSET instructions (and the aliases to them: ANDC, EORC, NOTC, and
ORC) that access the SP, N3, M01, LA, LA2, LC, LC2, HWS, SR, or OMR registers
•
The clear or test instructions (except TSTA.B, TSTA.W, TSTA.L, DECTSTA, or TSTDECA.W)
that access the SP,N3,M01,LA,LA2,LC,LC2, HWS, SR, or OMR registers
•
ALIGNSP
•
Tcc
Freescale Semiconductor
Instruction Set Introduction
4-15
Instruction Set Introduction
•
DEBUGHLT, DEBUGEV
There are additional restrictions on instructions that are allowed in delay slots for the RTID instruction.
Because this instruction restores the value of the status register, instructions that update the status register
are not allowed. The assembler detects these cases, which appear in the following list, and flags them as
illegal.
•
ADC, SBC, ROL.L, ROR.L, ROL.W, ROR.W
In addition to all of the preceding restrictions, the instructions that can be in the delay slots for the FRTID
instruction are further limited. The assembler dis-allows the following:
•
ADC, SBC, ROL.L, ROR.L, ROL.W, ROR.W
•
Any instruction in which the SP register is used as an address pointer, in an addressing mode, or in
an AGU calculation
•
Move instructions where the source or destination is the R0, R1, or N register
•
BFCHG, BFCLR, or BFSET instructions (including the ANDC, EORC, NOTC, and ORC
instruction aliases) that operate on the R0, R1, or N registers
•
CLR.W or TST.W on either the R0, R1, or N registers
•
Any two instructions in the delay slots (including any hardware interlocks) with a total execution
time greater than 3 cycles
4.3.3 Delayed Instructions and Interrupts
Instructions that are executed in delay slots are not interruptible. From the time that execution begins for a
delayed instruction to the end of execution for the instruction that occupies the last delay slot, no interrupts
are serviced. Any interrupt that occurs during this time is latched as pending and is not serviced until after
the final delay-slot instruction. See Section 9.3.4, “Non-Interruptible Instruction Sequences,” on page 9-10
for more information.
4.4 Instruction Set Summary
This section presents the entire core instruction set in tabular form. The tables show the instruction
mnemonics, supported operands, and addressing modes for each instruction. The number of instruction
cycles that each operation takes to execute and the number of program words that it occupies is also listed.
With these tables, it is easy to determine the appropriate instruction for a given application.
4.4.1 Using the Instruction Summary Tables
The entries in the instruction summary tables give the name of the operation (the instruction mnemonic),
the legal operands, cycle and word counts, and a brief description of the operation. The general form
appears in Table 4-16.
Table 4-16. Sample Instruction Summary Table
Operation
MAC
4-16
Operands
C
W
(±)FFF1,FFF1,FFF
1
1
Comments
Fractional multiply-accumulate; multiplication result
optionally negated before accumulation
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
The operands are specified using the register and immediate values that are allowed, or, when there are a
number of options, using shorthand notation. This notation, which is used to describe a set of registers, is
explained in Section 4.4.2, “Register Field Notation.”
The summary tables and the notation definitions make it possible to determine whether or not a particular
instruction is legal. For the MAC instruction in Table 4-16, for example, we can determine that the
following are valid core instructions:
MAC
MAC
MAC
X0,Y0,A
+X0,Y0,A
-X0,Y0,A
; A + X0*Y0 -> A
; A + X0*Y0 -> A
; A - (X0*Y0) -> A
The (+) in the operand entry for MAC indicates that an optional “+” or “–” sign can be specified before the
input register combination. If a “–” is specified, the multiplication result is negated.
Table 4-44 on page 4-50 shows all the registers and addressing modes that are allowed in a dual read
instruction, one of the core’s parallel move instructions. Based on the entries in the summary tables for the
MOVE, MACR, and ADD instructions, as well as the information contained in Table 4-44, we see that the
instructions in Example 4-4 are allowed.
Example 4-4. Valid Instructions
MOVE.W
MACR X0,Y1,A
ADD
Y0,B
X:(R0)+,Y0
X:(R1)+N,Y1
X:(R1)+N,Y0
X:(R3)+,X0
X:(R3)-,X0
X:(R3)+,X0
The instruction summary tables can also be used to determine if a particular instruction is not allowed.
Consider the instruction in Example 4-5.
Example 4-5. Invalid Instruction
ADD
X0,Y1,A
X:(R2)-,X0
X:(R3)+N,Y0
Using the information in Table 4-33 on page 4-31 and Table 4-44 on page 4-50, we know that this
instruction is invalid for the following reasons:
•
The ADD instruction only takes two operands, not three.
•
The pointer R2 is not allowed for the first memory read.
•
The post-decrement addressing mode is not available for the first memory read.
•
The X0 register cannot be a destination for the first memory read.
•
The post-update–by–N addressing mode is not allowed for the second memory read; only the
post-increment, post-decrement, and post-update–with–N3 addressing modes are allowed.
•
The Y0 register cannot be a destination for the second memory read.
4.4.2 Register Field Notation
There are many different register fields that are used within the instruction summary tables. The following
tables outline the notation that is used to specify legal registers.
Table 4-17 shows the register sets that are available for the most important move instructions. Whenever
the supported set of registers varies due to whether the set is the source or destination of an operation, the
difference is noted. Register fields that are used in conjunction with AGU move instructions are listed in
Table 4-18 on page 4-19.
Freescale Semiconductor
Instruction Set Introduction
4-17
Instruction Set Introduction
In some cases, the notation that is used for specifying an accumulator determines whether or not saturation
is enabled when the accumulator is being used as a source in a move or parallel move instruction. Refer to
Section 5.8.1, “Data Limiter,” on page 5-39 and Section 5.2, “Accessing the Accumulator Registers,” on
page 5-6 for more information.
Table 4-17. Register Fields for General-Purpose Writes and Reads
Register Field
Registers in This
Field
HHH
(source)
A1, B1, C1, D1
X0, Y0, Y1
Comments
Seven data ALU registers—four 16-bit MSP portions of the accumulators and three 16-bit data registers used as source registers. Note the
usage of A1, B1, C1, and D1.
This field is identical to the FFF1 field.
HHH
(destination)
A, B, C, D
Y
X0, Y0, Y1
Seven data ALU registers—four 16-bit MSP portions of the accumulators and three 16-bit data registers used as destination registers. Note
the usage of A, B, C, and D. Writing word data to the 32-bit Y register
clears the Y0 portion.
HHH.L
(source)
A10, B10, C10, D10
Y
Five data ALU registers—four 32-bit MSP:LSP portions of the accumulators and one 32-bit Y data register (Y1:Y0) used as source register.
Used for long memory accesses.
HHH.L
(destination)
A, B, C, D
Y
Five data ALU registers—four 32-bit MSP:LSP portions of the accumulators and one 32-bit Y data register (Y1:Y0) used as destination register.
Used for long memory accesses.
HHHH
(source)
A1, B1, C1, D1
X0, Y0, Y1
R0–R5, N
HHHH
(destination)
A, B, C, D
Y
X0, Y0, Y1
R0–R5, N
HHHH.L
(source)
A10, B10, C10, D10
Y
R0–R5, N
HHHH.L
(destination)
A, B, C, D
Y
R0–R5, N
Seven data ALU and seven AGU registers used as source registers.
Note the usage of A1, B1, C1, and D1.
Seven data ALU and seven AGU registers used as destination registers. Note the usage of A, B, C, and D. Writing word data to the 32-bit Y
register clears the Y0 portion.
Five data ALU and seven AGU registers used as source registers.
Used for long memory accesses. Also see dddd.L.
Five data ALU and seven AGU registers used as destination registers.
Used for long memory accesses. Also see dddd.L.
Table 4-18 shows the register sets that are available for use for pointers in address-register-indirect
addressing modes. The most commonly used fields in this table are Rn and RRR. This table also shows the
notation that is used for AGU registers in AGU arithmetic operations.
4-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-18. Address Generation Unit (AGU) Registers
Registers in This
Field
Comments
Rn
R0–R5
N
SP
Eight AGU registers available as pointers for addressing and address
calculations
RRR
R0–R5
N
Seven AGU registers available as pointers for addressing and as
sources and destinations for move instructions
Rj
R0, R1, R2, R3
N3
N3
M01
M01
Address modifier register
FIRA
FIRA
Fast interrupt return register
Register Field
Four pointer registers available as pointers for addressing
One index register available only for the second access in dual parallel
read instructions
Table 4-19 shows the register sets that are available for use in data ALU arithmetic operations. The most
commonly used fields in this table are EEE and FFF.
Table 4-19. Data ALU Registers
Register
Field
Registers in This
Field
Comments
FFF
A, B, C, D
Y
X0, Y0, Y1
Eight data ALU registers—four 36-bit accumulators, one 32-bit long register Y, and three 16-bit data registers accessible during data ALU operations.
FFF1
A1, B1, C1, D1
X0, Y0, Y1
Seven data ALU registers—four 16-bit MSP portions of the accumulators
and three 16-bit data registers accessible during data ALU operations.
This field is identical to the HHH (source) field. It is very similar to FFF, but
indicates that the MSP portion of the accumulator is in use. Note the usage
of A1, B1, C1, and D1.
EEE
A, B, C, D
X0, Y0, Y1
Seven data ALU registers—four accumulators and three 16-bit data registers accessible during data ALU operations.
This field is similar to FFF but is missing the 32-bit Y register. Used for
instructions where Y is not a useful operand (use Y1 instead).
fff
A, B, C, D, Y
FF
A, B, C, D
Four 36-bit accumulators accessible during data ALU operations.
DD
X0, Y0, Y1
Three 16-bit data registers.
F
A, B
Freescale Semiconductor
Four 36-bit accumulators and one 32-bit long register accessible during
data ALU operations.
Two 36-bit accumulators accessible during parallel move instructions and
some data ALU operations.
Instruction Set Introduction
4-19
Instruction Set Introduction
Table 4-19. Data ALU Registers (Continued)
Register
Field
Registers in This
Field
F1
A1, B1
Comments
The 16-bit MSP portion of two accumulators accessible as source operands in parallel move instructions.
Table 4-20 shows additional register fields that are available for move instructions.
Table 4-20. Additional Register Sets for Move Instructions
Register Field
DDDDD
Registers in This
Field
A, A2, A1, A0
B, B2, B1, B0
C, C1
D, D1
Y
Y1, Y0, X0
R0, R1, R2, R3
R4, R5, N, SP
M01, N3
Comments
This table lists the CPU registers. It contains the contents of the
HHHHH and SSSS register fields.
Y is permitted only as a destination, not as a source.
Writing word data to the 32-bit Y register clears the Y0 portion.
Note that the C2, C0, D2, and D0 registers are not available within this
field. See the dd register field for these registers
OMR, SR
LA, LC
HWS
dd
C2, D2, C0, D0
Extension and LS portion of the C and D accumulators.
This register set supplements the DDDDD field.
HHHHH
SSSS
A, A2, A1, A0
B, B2, B1, B0
C, C1
D, D1
Y
Y1, Y0, X0
R0, R1, R2, R3
R4, R5, N, SP
M01, N3
LA, LC, HWS
OMR, SR
dddd.L
4-20
A2, B2, C2, D2
Y0, Y1, X0
SP, M01, N3,
LA, LA2, LC, LC2,
HWS, OMR, SR
This set designates registers that are written with signed values when
written with word values.
Y is permitted only as a destination, not as a source.
The registers in this field and SSSS combine to make the DDDDD register field.
This set designates registers that are written with unsigned values when
written with word values.
The registers in this field and HHHHH combine to make the DDDDD
register field.
Miscellaneous set of registers that can be placed onto or removed from
the stack 32 bits at a time.
This list supplements the registers in the HHHH.L field, which can also
access the stack via the MOVE.L instruction.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
4.4.3 Immediate Value Notation
Immediate values, including absolute and offset addresses, are presented in the instruction set summary
using the notation presented in Table 4-21.
Table 4-21. Immediate Value Notation
Immedate Value Field
Description
<MASK16>
16-bit mask value
<MASK8>
8-bit mask value
<OFFSET18>
18-bit signed PC-relative offset
<OFFSET22>
22-bit signed PC-relative offset
<OFFSET7>
7-bit signed PC-relative offset
<ABS16>
16-bit absolute address
<ABS19>
19-bit absolute address
<ABS21>
21-bit absolute address
4.4.4 Instruction Summary Tables
A summary of the entire core instruction set is presented in this section in tabular form. In these tables, the
instructions are broken into several different categories and then listed alphabetically.
The tables specify the operation, operands, and any relevant comments. There are separate fields for
sources and destinations of move instructions. In addition, each instruction has two fields:
•
C—Number of clock cycles that are required to execute the instruction
•
W—Number of program words that are required by the instruction
Descriptions of the parallel move instruction syntax (for those operations that support them) are located at
the end of this section. See Table 4-43 on page 4-49 and Table 4-44 on page 4-50 for information on
parallel moves.
Table 4-22. Move Byte Instructions—Byte Pointers
Operation
MOVE.BP
Source
Destination
C
W
X:(RRR)
X:(RRR)+
X:(RRR)–
HHH
1
1
Move signed byte from memory
X:(RRR+N)
HHH
2
1
Address = Rn+N
X:(RRR+xxxx)
HHH
2
2
Unsigned 16-bit offset
X:(RRR+xxxxxx)
HHH
3
3
24-bit offset
X:xxxx
HHH
2
2
Unsigned 16-bit absolute address
X:xxxxxx
HHH
3
3
24-bit absolute address
Freescale Semiconductor
Comments
Instruction Set Introduction
4-21
Instruction Set Introduction
Table 4-22. Move Byte Instructions—Byte Pointers (Continued)
Operation
MOVEU.BP
MOVE.BP
Source
Destination
C
W
Comments
X:(RRR)
X:(RRR)+
X:(RRR)–
HHH
1
1
Move unsigned byte from memory
X:(RRR+N)
HHH
2
1
Address = Rn+N
X:(RRR+xxxx)
HHH
2
2
Unsigned 16-bit offset
X:(RRR+xxxxxx)
HHH
3
3
24-bit offset
X:xxxx
HHH
2
2
Unsigned 16-bit absolute address
X:xxxxxx
HHH
3
3
24-bit absolute address
HHH
X:(RRR)
X:(RRR)+
X:(RRR)–
1
1
Move signed byte to memory
HHH
X:(RRR+N)
2
1
Address = Rn+N
HHH
X:(RRR+xxxx)
2
2
Unsigned 16-bit offset
HHH
X:(RRR+xxxxxx)
3
3
24-bit offset
HHH
X:xxxx
2
2
Unsigned 16-bit absolute address
HHH
X:xxxxxx
3
3
24-bit absolute address
Table 4-23. Move Byte Instructions—Word Pointers
Operation
MOVE.B
MOVEU.B
MOVE.B
4-22
Source
Destination
C
W
Comments
X:(Rn+xxxx)
HHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHH
3
3
24-bit offset
X:(SP)
HHH
1
1
Pointer is SP
X:(RRR+x)
HHH
2
1
x: offset ranging from 0 to 7
X:(Rn+xxxx)
HHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHH
3
3
24-bit offset
X:(SP–x)
HHH
2
1
x: offset ranges from 1 to 8
X:(SP)
HHH
1
1
Pointer is SP
HHH
X:(RRR+x)
2
1
x: offset ranges from 0 to 7
HHH
X:(Rn+xxxx)
2
2
Signed 16-bit offset
HHH
X:(Rn+xxxxxx)
3
3
24-bit offset
HHH
X:(SP–x)
2
1
x: offset ranges from 1 to 8
HHH
X:(SP)
1
1
Pointer is SP
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-24. Move Long Word Instructions
Operation
MOVE.L
MOVE.L
Source
Destination
C
W
Comments
X:(Rn)
X:(Rn)+
X:(Rn)–
HHHH.L
1
1
Move signed 32-bit long word from memory;
note that Rn includes SP
X:(SP)–
dddd.L
1
1
Pop 32 bits from stack; does not modify bits
14–10 in SR
X:(Rn+N)
HHHH.L
2
1
Address = Rn+N
X:(Rn+xxxx)
HHHH.L
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHHH.L
3
3
24-bit offset
X:(SP–xx)
HHHH.L
2
1
Unsigned 6-bit offset, left shifted 1 bit
X:xxxx
HHHH.L
2
2
Unsigned 16-bit address
X:xxxxxx
HHHH.L
3
3
24-bit address
HHHH.L
X:(Rn)
X:(Rn)+
X:(Rn)–
1
1
Move signed 32-bit long word to memory;
note that Rn includes SP
dddd.L
X:(SP)+
1
1
Push 32 bits onto stack; SP not permitted in
dddd.L
HHHH.L
X:(Rn+N)
2
1
Address = Rn+N
HHHH.L
X:(Rn+xxxx)
2
2
Signed 16-bit offset
HHHH.L
X:(Rn+xxxxxx)
3
3
24-bit offset
HHHH.L
X:(SP–xx)
2
1
Unsigned 6-bit offset, left shifted 1 bit
HHHH.L
X:xxxx
2
2
Unsigned 16-bit address
HHHH.L
X:xxxxxx
3
3
24-bit address
Freescale Semiconductor
Instruction Set Introduction
4-23
Instruction Set Introduction
Table 4-25. Move Word Instructions
Operation
MOVE.W
Source
Destination
C
W
Comments
X:(Rn)
X:(Rn)+
X:(Rn)–
HHHHH
1
1
Move signed 16-bit integer word from memory
X:(Rn+N)
HHHHH
2
1
Address = Rn+N
X:(Rn)+N
HHHHH
1
1
Post-update of Rn register
X:(Rn+x)
HHH
2
1
x: offset ranging from 0 to 7
X:(Rn+xxxx)
HHHHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHHHH
3
3
24-bit offset
X:(SP–xx)
HHH
2
1
Unsigned 6-bit offset
X:xxxx
HHHHH
2
2
Unsigned 16-bit address
X:xxxxxx
HHHHH
3
3
24-bit address
X:<<pp
X0, Y1, Y0
A, B, C, A1, B1
1
1
6-bit peripheral address
X:aa
X0, Y1, Y0
A, B, C, A1, B1
1
1
6-bit absolute short address
1
1
Refer to Table 4-44 on page 4-50.
(parallel)
MOVEU.W
4-24
X:(Rn)
X:(Rn)+
X:(Rn)–
SSSS
1
1
Move signed 16-bit integer word from memory
X:(Rn+N)
SSSS
2
1
Address = Rn+N
X:(Rn)+N
SSSS
1
1
Post-update of Rn register
X:(Rn+xxxx)
SSSS
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
SSSS
3
3
24-bit offset
X:(SP–xx)
RRR
2
1
Unsigned 6-bit offset
X:xxxx
SSSS
2
2
Unsigned 16-bit address
X:xxxxxx
SSSS
3
3
24-bit address
X:<<pp
RRR
1
1
6-bit peripheral address
X:aa
RRR
1
1
6-bit absolute short address
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-25. Move Word Instructions (Continued)
Operation
MOVE.W
Source
Destination
C
W
DDDDD
X:(Rn)
X:(Rn)+
X:(Rn)–
1
1
Move signed 16-bit integer word to memory
DDDDD
X:(Rn+N)
2
1
Address = Rn+N
DDDDD
X:(Rn)+N
1
1
Post-update of Rn register
HHH
X:(Rn+x)
2
1
x: offset ranging from 0 to 7
DDDDD
X:(Rn+xxxx)
2
2
Signed 16-bit offset
DDDDD
X:(Rn+xxxxxx)
3
3
24-bit offset
HHHH
X:(SP–xx)
2
1
Unsigned 6-bit offset
DDDDD
X:xxxx
2
2
Unsigned 16-bit address
DDDDD
X:xxxxxx
3
3
24-bit address
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
X:<<pp
1
1
6-bit peripheral address
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
X:aa
1
1
6-bit absolute short address
Freescale Semiconductor
Comments
Instruction Set Introduction
4-25
Instruction Set Introduction
Table 4-26. Memory-to-Memory Move Instructions
Operation
Source
Destination
C
W
X:(RRR)
X:(RRR)+
X:(RRR)–
X:xxxx
2
2
Move byte from one memory location to another;
RRR used as a byte pointer
X:(RRR+N)
X:xxxx
3
2
RRR used as a byte pointer
X:(RRR+xxxx)
X:xxxx
3
3
Unsigned 16-bit offset; RRR used as a byte
pointer
X:xxxx
X:xxxx
3
3
16-bit absolute address
X:(RRR+x)
X:xxxx
3
2
x: offset ranges from 0 to 7
X:(SP)
X:xxxx
2
2
Signed 16-bit offset
X:(SP–x)
X:xxxx
3
2
x: offset ranges from 1 to 8
MOVE.B1
X:(Rn+xxxx)
X:xxxx
3
3
Signed 16-bit offset
MOVE.W
X:(Rn+x)
X:xxxx
3
2
Move word from one memory location to another;
x: offset ranges from 0 to 7
X:(SP–xx)
X:xxxx
3
2
X:(Rn)
X:(Rn)+
X:(Rn)–
X:xxxx
2
2
X:(Rn+N)
X:xxxx
3
2
X:(Rn)+N
X:xxxx
2
2
X:(Rn+xxxx)
X:xxxx
3
3
Signed 16-bit offset
X:xxxx
X:xxxx
3
3
16-bit absolute address
X:(SP–xx)
X:xxxx
3
2
Move long from one memory location to another
X:(Rn)
X:(Rn)+
X:(Rn)–
X:xxxx
2
2
X:(Rn+N)
X:xxxx
3
2
X:(Rn+xxxx)
X:xxxx
3
3
Signed 16-bit offset
X:xxxx
X:xxxx
3
3
16-bit absolute address
MOVE.BP1
MOVEU.B1
MOVE.L
Comments
1.The destination operand X:xxxx is always specified as a byte address for the MOVE.BP, MOVEU.B, and
MOVE.B instructions. The upper 15 bits of the address select the appropriate word location in memory, and
the LSB selects the upper or lower byte of that word.
4-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-27. Immediate Move Instructions
Operation
MOVE.W
Source
Destination
C
W
Comments
#<–64,63>
HHHH
1
1
Signed 7-bit integer data (data is put in the lowest 7
bits of the word portion of any accumulator, upper 8
bits and extension register are sign extended, LSP
portion is set to 0).
X:xxxx
2
2
Signed 7-bit integer data (data put in the low portion of the word).
HHHHH
2
2
Signed 16-bit immediate data.
dd
2
2
Move to C2, D2, C0, D0 registers.
X:(Rn)
2
2
X:(Rn+xxxx)
3
3
X:(SP–xx)
2
2
X:<<pp
2
2
Move 16-bit immediate data to the one of 64 locations in X data memory—peripheral registers.
X:aa
2
2
Move 16-bit immediate data to the first 64 locations
of X data memory.
X:xxxx
3
3
X:xxxxxx
4
4
#xxxx
MOVEU.W
#xxxx
SSSS
2
2
Unsigned 16-bit immediate data.
MOVE.L
#xxxx
X:xxxx
3
3
Sign extend 16-bit value and move to 32-bit memory location.
#xxxxxxxx
X:xxxx
4
4
Move to 32-bit memory location.
#<–16,15>
HHH.L
1
1
Signed 5-bit integer data (data is put in the lowest 5
bits of the word portion of the register; upper bits
are sign extended).
#xxxx
HHHH.L
2
2
Sign extend the 16-bit immediate data to 36 bits
when moving to an accumulator; sign extend to
24 bits when moving to an AGU register.
Use MOVEU.W for moves to the AGU with
unsigned 16-bit immediate data.
#xxxxxxxx
HHH.L
3
3
Move signed 32-bit immediate data to a 32-bit
accumulator.
#xxxxxx
RRR
3
3
Move unsigned 24-bit immediate value to AGU register.
Freescale Semiconductor
Instruction Set Introduction
4-27
Instruction Set Introduction
Table 4-28. Register-to-Register Move Instructions
Operation
MOVE.W
MOVEU.W
MOVE.L
SWAP
Source
Destination
C
W
Comments
DDDDD
HHHHH
1
1
Move signed word to register.
HHH
RRR
DDDDD
SSSS
1
1
HHH.L
RRR
1
1
RRR
HHH.L
Move signed word to register.
Move unsigned word to register.
MOVEU.W HWS,HWS is not supported.
Move pointer register to data ALU register.
Zero extend the 24-bit value contained in the
RRR register.
SHADOWS
1
1
This instruction swaps the value in the R0, R1,
N, and M01 registers with their shadow registers. It is the only instruction that accesses the
shadow registers.
NOTE:
Additional register-to-register move instructions include the TFR and
SXT.L instructions for data ALU registers (see Table 4-33 on page 4-31)
and the TFRA instruction for AGU registers (see Table 4-37 on
page 4-42).
Table 4-29. Conditional Register Transfer Instructions
Data ALU Transfer
AGU Transfer
Operation
Tcc1
Source
Destination
Source
DD
F
(No transfer)
A
B
(No transfer)
B
A
(No transfer)
DD
F
R0
R1
A
B
R0
R1
B
A
R0
R1
C
W
Comments
1
1
Conditionally transfer one
register
Destination
Conditionally transfer one
data ALU register and
one AGU register
1.The Tcc instruction does not support the HI, LS, NN, and NR conditions.
4-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-30. Move Word Instructions—Program Memory
Operation1
Source
Destination
C
W
MOVE.W
P:(Rj)+
P:(Rj)+N
X0, Y1, Y0,
A, B, C,
A1 or B1
5
1
Read signed word from program memory
MOVEU.W
P:(Rj)+
P:(Rj)+N
RRR
5
1
Read unsigned word from program memory
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
P:(Rj)+
P:(Rj)+N
5
1
Write word to program memory
MOVE.W
Comments
1.These instructions are not allowed when the XP bit in the OMR is set (that is, when the instructions are executing from data memory).
Table 4-31. Data ALU Multiply Instructions
Operation
Operands
C
W
Comments
IMAC.L
FFF1,FFF1,fff
1
1
Integer 16 × 16 multiply-accumulate with 36-bit result.
IMPY.L
FFF1,FFF1,fff
1
1
Integer 16 × 16 multiply with 32-bit result.
IMPY.W
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
1
1
Integer 16 × 16 multiply with 16-bit result.
(±)FFF1,FFF1,FFF
1
MAC
When the destination is the Y register or an accumulator, the LSP portion is unchanged by the instruction.
Note: Assembler also accepts first two operands
when they are specified in opposite order.
1
(parallel)
MACR
Refer to Table 4-43 and Table 4-44.
(±)FFF1,FFF1,FFF
1
1
(parallel)
MPY
(parallel)
Freescale Semiconductor
Fractional MAC with round; multiplication result
optionally negated before addition.
Refer to Table 4-43 and Table 4-44.
FFF1,FFF1,FFF
–Y1,X0,FFF
–Y0,X0,FFF
–Y1,Y0,FFF
–Y0,Y0,FFF
–A1,Y0,FFF
–B1,Y1,FFF
–C1,Y0,FFF
–C1,Y1,FFF
Fractional multiply-accumulate; multiplication result
optionally negated before accumulation.
1
1
Fractional multiply.
Fractional multiply where one operand negated
before multiplication.
Note: Assembler also accepts first two operands
when they are specified in opposite order.
Refer to Table 4-43 and Table 4-44.
Instruction Set Introduction
4-29
Instruction Set Introduction
Table 4-31. Data ALU Multiply Instructions (Continued)
Operation
MPYR
Operands
FFF1,FFF1,FFF
C
W
1
1
–Y1,X0,FFF
–Y0,X0,FFF
–Y1,Y0,FFF
–Y0,Y0,FFF
–A1,Y0,FFF
–B1,Y1,FFF
–C1,Y0,FFF
–C1,Y1,FFF
Comments
Fractional multiply; result rounded.
Fractional multiply where one operand negated
before multiplication. The result is rounded.
Note: Assembler also accepts first two operands
when they are specified in opposite order.
(parallel)
Refer to Table 4-43 and Table 4-44.
Table 4-32. Data ALU Extended-Precision Multiplication Instructions
Operation
IMACUS
IMACUU
IMPYSU
4-30
Operands
C
W
A0,A1,Y
A0,B1,Y
A0,C1,Y
A0,D1,Y
B0,C1,Y
B0,D1,Y
C0,C1,Y
C0,D1,Y
1
1
A0,A1,Y
A0,B1,Y
A0,C1,Y
A0,D1,Y
B0,C1,Y
B0,D1,Y
C0,C1,Y
C0,D1,Y
1
A1,A0,Y
A1,B0,Y
A1,C0,Y
A1,D0,Y
B1,C0,Y
B1,D0,Y
C1,C0,Y
C1,D0,Y
1
Comments
Integer 16 × 16 multiply accumulate:
F0 (unsigned) × F1 (signed).
This instruction is described in more detail in
Section 5.5.3, “Multi-Precision Integer Multiplication,” on
page 5-32.
1
Integer 16 × 16 multiply accumulate:
F0 (unsigned) × F1 (unsigned).
This instruction is described in more detail in
Section 5.5.3, “Multi-Precision Integer Multiplication,” on
page 5-32.
1
Integer 16 × 16 multiply:
F1 (signed) × F0 (unsigned).
This instruction is described in more detail in
Section 5.5.3, “Multi-Precision Integer Multiplication,” on
page 5-32.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-32. Data ALU Extended-Precision Multiplication Instructions (Continued)
Operation
Operands
IMPYUU
MACSU
MPYSU
C
W
A1,A0,Y
A1,B0,Y
A1,C0,Y
A1,D0,Y
B1,C0,Y
B1,D0,Y
C1,C0,Y
C1,D0,Y
1
1
A0,A0,FF
A0,B0,FF
A0,C0,FF
A0,D0,FF
B0,C0,FF
B0,D0,FF
C0,C0,FF
C0,D0,FF
1
Comments
Integer 16 × 16 multiply:
F1 (unsigned) × F0 (unsigned).
This instruction is described in more detail in
Section 5.5.3, “Multi-Precision Integer Multiplication,” on
page 5-32.
1
Integer 16 × 16 multiply:
F0 (unsigned) × F0 (unsigned).
This instruction is described in more detail in
Section 5.5.3, “Multi-Precision Integer Multiplication,” on
page 5-32.
X0,Y1,EEE
X0,Y0,EEE
Y0,Y1,EEE
Y0,Y0,EEE
Y0,A1,EEE
Y1,B1,EEE
Y0,C1,EEE
Y1,C1,EEE
1
X0,Y1,EEE
X0,Y0,EEE
Y0,Y1,EEE
Y0,Y0,EEE
Y0,A1,EEE
Y1,B1,EEE
Y0,C1,EEE
Y1,C1,EEE
1
1
16 × 16 => 32-bit unsigned/signed fractional MAC.
The first operand is treated as signed and the second as
unsigned.
1
16 × 16 => 32-bit signed/unsigned fractional multiply.
The first operand is treated as signed and the second as
unsigned.
Table 4-33. Data ALU Arithmetic Instructions (Sheet 1 of 8)
Operation
ABS
Operands
C
W
FFF
1
1
(parallel)
Comments
Absolute value.
Refer to Table 4-43 on page 4-49.
ADC
Y,F
1
1
Add with carry (set C bit also).
ADD
FFF,FFF
1
1
36-bit addition of two registers.
(parallel)
ADD.B
ADD.BP
Freescale Semiconductor
Refer to Table 4-43 and Table 4-44.
#xxx,EEE
2
2
Add 9-bit signed immediate.
X:xxxx,EEE
2
2
Add memory byte to register.
X:xxxxxx,EEE
3
3
Instruction Set Introduction
4-31
Instruction Set Introduction
Table 4-33. Data ALU Arithmetic Instructions (Sheet 2 of 8)
Operation
ADD.L
ADD.W
CLR
Operands
C
W
X:xxxx,fff
2
2
X:xxxxxx,fff
3
3
#xxxx,fff
2
2
Add a 16-bit immediate value sign extended to 32 bits
to a data register.
X:(Rn),EEE
2
1
Add memory word to register.
X:(Rn+xxxx),EEE
3
2
X:(SP–xx),EEE
3
1
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
EEE,X:(SP–xx)
4
2
EEE,X:xxxx
3
2
#<0–31>,EEE
1
1
Add an immediate integer 0–31.
#xxxx,EEE
2
2
Add a signed 16-bit immediate.
F
1
1
Clear 36-bit accumulator and set condition codes.
Also see CLR.W.
(parallel)
CLR.B
CLR.BP
4-32
Comments
Add memory long to register.
Add register to memory word, storing the result back
to memory.
Refer to Table 4-43 and Table 4-44.
X:(SP)
1
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:(RRR)
1
1
X:(RRR)+
1
1
X:(RRR)–
1
1
X:(RRR+N)
2
1
X:(RRR+xxxx)
2
2
X:(RRR+xxxxxx)
3
3
X:xxxx
2
2
X:xxxxxx
3
3
Clear a byte in memory.
Rn may be SP.
Clear a byte in memory.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-33. Data ALU Arithmetic Instructions (Sheet 3 of 8)
Operation
Operands
C
W
X:(Rn)
1
1
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:xxxx
2
2
X:xxxxxx
3
3
DDDDD
(except Y)
1
1
CLR.L
CLR.W
Comments
Clear a long in memory.
Clear a register. Clear an entire accumulator when FF
specified; clear an entire AGU register when Rn is
specified.
Note: When clearing an AGU register, it is
recommended to use MOVE.W #0,Rn. This is
beneficial because it clears the register without
introducing any dependencies due to the pipeline.
Not permitted for the 32-bit Y register—instead use
MOVE.W #0,Y.
X:(Rn)
1
1
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn)+N
1
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:aa
1
1
X:<<pp
1
1
X:xxxx
2
2
X:xxxxxx
3
3
Freescale Semiconductor
Clear a word in memory.
Instruction Set Introduction
4-33
Instruction Set Introduction
Table 4-33. Data ALU Arithmetic Instructions (Sheet 4 of 8)
Operation
CMP
CMP.B
CMP.BP
CMP.L
4-34
Operands
C
W
Comments
EEE,EEE
1
1
36-bit compare of two accumulators or data registers.
X:(Rn),FF
2
1
Compare memory word with 36 bit accumulator.
X:(Rn+xxxx),FF
3
2
Also see CMP.W.
X:(SP–xx),FF
3
1
X:xxxx,FF
2
2
X:xxxxxx,FF
3
3
#<0–31>,FF
1
1
Compare accumulator with an immediate integer
0–31.
#xxxx,FF
2
2
Compare accumulator with a signed 16-bit immediate.
(parallel)
1
1
Refer to Table 4-43 on page 4-49.
EEE,EEE
1
1
Compare the 8-bit byte portions of two data registers.
#<0–31>,EEE
1
1
Compare the byte portion of a data register with an
immediate integer 0–31.
#xxx,EEE
2
2
Compare with a 9-bit signed immediate integer.
X:xxxx,EEE
2
2
Compare memory byte with register.
X:xxxxxx,EEE
3
3
FFF,FFF
1
1
Compare the 32-bit long portions of two data registers
or accumulators.
X:xxxx,fff
2
2
Compare memory long with a data register.
X:xxxxxx,fff
3
3
#xxxx,fff
2
2
Note: Condition codes set based on 36-bit result.
Also see CMP.W for condition codes on 16 bits.
Compare a 16-bit immediate value sign extended to
32 bits with a data register.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-33. Data ALU Arithmetic Instructions (Sheet 5 of 8)
Operation
Operands
C
W
EEE,EEE
1
1
Compare the 16-bit word portions of two data registers or accumulators.
X:(Rn),EEE
2
1
Compare memory word with a data register or the
word portion of an accumulator.
X:(Rn+xxxx),EEE
3
2
X:(SP–xx),EEE
3
1
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
#<0–31>,EEE
1
1
Compare the word portion of a data register with an
immediate integer 0–31.
#xxxx,EEE
2
2
Compare the word portion of a data register with a
signed 16-bit immediate.
X:xxxx
3
2
Decrement byte in memory.
X:xxxxxx
4
3
fff
1
1
Decrement long.
X:xxxx
3
2
Decrement long in memory.
X:xxxxxx
4
3
EEE
1
1
Decrement word.
X:(Rn)
3
1
Decrement word in memory using appropriate
addressing mode.
X:(Rn+xxxx)
4
2
X:(SP–xx)
4
1
X:xxxx
3
2
X:xxxxxx
4
3
(parallel)
1
1
Refer to Table 4-43 on page 4-49.
DIV
FFF1,fff
1
1
Divide iteration.
INC.BP
X:xxxx
3
2
Increment byte in memory.
X:xxxxxx
4
3
fff
1
1
Increment long.
X:xxxx
3
2
Increment long in memory.
X:xxxxxx
4
3
CMP.W
DEC.BP
DEC.L
DEC.W
INC.L
Freescale Semiconductor
Comments
Instruction Set Introduction
4-35
Instruction Set Introduction
Table 4-33. Data ALU Arithmetic Instructions (Sheet 6 of 8)
Operation
INC.W
NEG
Operands
C
W
EEE
1
1
Increment word.
X:(Rn)
3
1
Increment word in memory using appropriate
addressing mode.
X:(Rn+xxxx)
4
2
X:(SP–xx)
4
1
X:xxxx
3
2
X:xxxxxx
4
3
(parallel)
1
1
Refer to Table 4-43 on page 4-49.
FFF
1
1
Two’s-complement negation.
(parallel)
NEG.BP
NEG.L
NEG.W
RND
Refer to Table 4-43 on page 4-49.
X:xxxx
3
2
X:xxxxxx
4
3
X:xxxx
3
2
X:xxxxxx
4
3
X:(Rn)
3
1
X:(Rn+xxxx)
4
2
X:(SP–xx)
4
1
X:xxxx
3
2
X:xxxxxx
4
3
fff
1
1
(parallel)
SAT
FF,FFF
Comments
Negate byte in memory.
Negate long in memory.
Negate word in memory using appropriate addressing
mode.
Round.
Refer to Table 4-43 on page 4-49.
1
1
(parallel)
Saturate and transfer 32 bits independent of SA bit.
Refer to Table 4-43 on page 4-49.
SBC
Y,F
1
1
Subtract with carry (set C bit also).
SUB
FFF,FFF
1
1
36-bit subtraction of two registers.
(parallel)
SUB.B
SUB.BP
4-36
Refer to Table 4-43 and Table 4-44.
#xxx,EEE
2
2
Subtract 9-bit signed immediate.
X:xxxx,EEE
2
2
Subtract memory byte from register.
X:xxxxxx,EEE
3
3
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-33. Data ALU Arithmetic Instructions (Sheet 7 of 8)
Operation
Operands
C
W
X:xxxx,fff
2
2
X:xxxxxx,fff
3
3
#xxxx,fff
2
2
Subtract a 16-bit immediate value, sign extended to
32 bits, from a data register.
X:(Rn),EEE
2
1
Subtract memory word from register.
X:(Rn+xxxx),EEE
3
2
X:(SP–xx),EEE
3
1
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
#<0–31>,EEE
1
1
Subtract an immediate value 0–31.
#xxxx,EEE
2
2
Subtract a signed 16-bit immediate.
SXT.B
FFF,FFF
1
1
Sign extend byte.
SXT.L
FF,FFF
1
1
Sign extend long and transfer without saturating.
TFR
FFF,fff
1
1
Transfer register to register, 36 bits. Also see SXT.L.
SUB.L
SUB.W
(parallel)
TST
Subtract memory long from register.
Refer to Table 4-43 and Table 4-44.
FF
1
1
(parallel)
TST.B
Comments
Test 36-bit accumulator.
Refer to Table 4-43 on page 4-49.
EEE
1
1
Test 8-bit byte in register.
X:(SP)
1
1
Test a byte in memory using appropriate addressing
mode.
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:(RRR)
1
1
X:(RRR)+
1
1
X:(RRR)–
1
1
X:(RRR+N)
2
1
X:(RRR+xxxx)
2
2
X:(RRR+xxxxxx)
3
3
X:xxxx
2
2
X:xxxxxx
3
3
TST.BP
Freescale Semiconductor
Test a byte in memory using appropriate addressing
mode.
Instruction Set Introduction
4-37
Instruction Set Introduction
Table 4-33. Data ALU Arithmetic Instructions (Sheet 8 of 8)
Operation
TST.L
TST.W
ZXT.B
4-38
Operands
C
W
Comments
fff
1
1
Test 32-bit long in register.
X:(Rn)
1
1
Test a long in memory using appropriate addressing
mode.
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:(SP–xx)
2
1
X:xxxx
2
2
X:xxxxxx
3
3
DDDDD
(except HWS and Y)
1
1
Test 16-bit word in register.
All registers are allowed except HWS and Y.
Limiting is not performed if an accumulator is specified.
X:(Rn)
1
1
Test a word in memory using appropriate addressing
mode.
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn)+N
1
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:(SP–xx)
2
1
X:aa
1
1
X:<<pp
1
1
X:xxxx
2
2
X:xxxxxx
3
3
FFF,FFF
1
1
Zero extend byte.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-34. Data ALU Shifting Instructions
Operation
Operands
C
W
fff
1
1
ASL
(parallel)
Comments
Arithmetic shift left entire register by 1 bit
Refer to Table 4-43 and Table 4-44.
ASL.W
DD
1
1
Arithmetic shift left 16-bit register by 1 bit
ASL16
FFF,FFF
1
1
Arithmetic shift left of the first operand by 16 bits, placing
result in the destination operand
ASLL.L
#<0–31>,fff
2
1
Arithmetic shift left by a 5-bit positive immediate integer
EEE,FFF
ASLL.W
#<0–15>,FFF
Bi-directional arithmetic shift of destination by value in the
first operand: positive –> left shift
1
1
EEE,FFF
Arithmetic shift left of destination by value specified in 4
LSBs of the first operand
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
ASR
FFF
Arithmetic shift left by a 4-bit positive immediate integer
Arithmetic shift left of the first operand by value specified in
4 LSBs of the second operand; place result in FFF
1
1
(parallel)
Arithmetic shift right entire register by 1 bit
Refer to Table 4-43 and Table 4-44.
ASR16
FFF,FFF
1
1
Arithmetic shift right of the first operand by 16 bits, placing
result in the destination operand
ASRAC
Y1,X0,FF
Y0,X0,FF
Y1,Y0,FF
Y0,Y0,FF
A1,Y0,FF
B1,Y1,FF
C1,Y0,FF
C1,Y1,FF
1
1
Arithmetic word shifting with accumulation
ASRR.L
#<0–31>,fff
2
1
Arithmetic shift right by a 5-bit positive immediate integer
EEE,FFF
Freescale Semiconductor
Bi-directional arithmetic shift of destination by value in the
first operand: positive –> right shift
Instruction Set Introduction
4-39
Instruction Set Introduction
Table 4-34. Data ALU Shifting Instructions (Continued)
Operation
ASRR.W
Operands
C
W
#<0–15>,FFF
1
1
Comments
Arithmetic shift right by a 4-bit positive immediate integer
EEE,FFF
Arithmetic shift right of destination by value specified in 4
LSBs of the first operand
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
Arithmetic shift right of the first operand by value specified
in 4 LSBs of the second operand; places result in FFF
LSL.W
EEE
1
1
1-bit logical shift left of word
LSR.W
EEE
1
1
1-bit logical shift right of word
LSR16
FFF,FFF
1
1
Logical shift right of the first operand by 16 bits, placing
result in the destination operand (new bits zeroed)
LSRAC
Y1,X0,FF
Y0,X0,FF
Y1,Y0,FF
Y0,Y0,FF
A1,Y0,FF
B1,Y1,FF
C1,Y0,FF
C1,Y1,FF
1
1
Logical word shifting with accumulation
LSRR.L
#<0–31>,fff
2
1
Logical shift right by a 5-bit positive immediate integer
EEE,FFF
LSRR.W
#<0–15>,FFF
Bi-directional logical shift of destination by value in the first
operand: positive –> right shift
1
1
Logical shift right by a 4-bit positive immediate integer
(sign extends into FF2)
EEE,FFF
Logical shift right of destination by value specified in 4 LSBs
of the first operand (sign extends into FF2)
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
Logical shift right of the first operand by value specified in 4
LSBs of the second operand; places result in FFF (sign
extends into FF2)
ROL.L
F
1
1
Rotate 32-bit register left by 1 bit through the carry bit
ROL.W
EEE
1
1
Rotate 16-bit register left by 1 bit through the carry bit
ROR.L
F
1
1
Rotate 32-bit register right by 1 bit through the carry bit
ROR.W
EEE
1
1
Rotate 16-bit register right by 1 bit through the carry bit
4-40
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-34. Data ALU Shifting Instructions (Continued)
Operation
SUBL
Operands
C
W
(parallel)
1
1
Comments
Refer to Table 4-43 on page 4-49.
Table 4-35. Data ALU Logical Instructions
Operation
AND.L
Operands
C
W
#<0–31>,fff
1
1
FFF,fff
AND.W
1
1
EEE,EEE
EOR.L
AND with a 5-bit positive immediate integer (0–31)
32-bit logical AND
#<0–31>,EEE
CLB
Comments
AND with a 5-bit positive immediate integer (0–31)
16-bit logical AND
FFF,EEE
1
1
Count leading bits (minus 1); designed to operate with
the ASLL and ASRR instructions
FFF,fff
1
1
32-bit exclusive OR (XOR)
(parallel)
Refer to Table 4-43 on page 4-49.
EOR.W
EEE,EEE
1
1
16-bit exclusive OR (XOR)
NOT.W
EEE
1
1
One’s-complement (bit-wise) negation
OR.L
FFF,fff
1
1
32-bit logical OR
OR.W
EEE,EEE
1
1
16-bit logical OR
ANDC, EORC, ORC, and NOTC can also be used to perform logical operations with an immediate value
on registers and data memory locations. See Section 4.2.1, “The ANDC, EORC, ORC, and NOTC
Aliases,” for additional information.
Table 4-36. Miscellaneous Data ALU Instructions
Operation
NORM
Operands
C
W
Comments
R0,F
4
1
Normalization iteration instruction for normalizing the F accumulator
Freescale Semiconductor
Instruction Set Introduction
4-41
Instruction Set Introduction
Table 4-37. AGU Arithmetic and Shifting Instructions
Operation
Operands
C
W
Rn,Rn
1
1
Add first operand to the second and store the result in the
second operand.
Rn,Rn,N
1
1
Add first operand to the second and store result in the N register.
#<0–15>,Rn
1
1
Add unsigned 4-bit value to Rn.
#<0–15>,Rn,N
1
1
Add an unsigned 4-bit value to an AGU register and store
result in the N register.
#xxxxx,Rn,Rn
2
2
Add first register with a signed 17-bit immediate value and
store the result in Rn.
#xxxxxx,Rn,Rn
3
3
Add first register with a 24-bit immediate value and store the
result in Rn.
#xxxx,HHH,Rn
4
2
Add a data register with an unsigned 16-bit value and store
the result in Rn. HHH is accessed as a signed 16-bit word.
#xxxxxx,HHH,Rn
5
3
Add a data register with a 24-bit immediate value and store
the result in Rn. HHH is accessed as a signed 16-bit word.
Rn,Rn
1
1
Add first operand left shifted 1 bit to the second and store the
result in the second operand.
Rn,Rn,N
1
1
Add first operand left shifted 1 bit to the second and store
result in the N register.
#xxxx,Rn,Rn
2
2
Add first register left shifted 1 bit with an unsigned 16-bit
immediate value and store the result in Rn.
#xxxxxx,Rn,Rn
3
3
Add first register left shifted 1 bit with a 24-bit immediate
value and store the result in Rn.
#xxxx,HHH,Rn
4
2
Add a data register left shifted 1 bit with an unsigned 16-bit
immediate value and store the result in Rn. HHH is accessed
as a signed 16-bit word.
#xxxxxx,HHH,Rn
5
3
Add a data register left shifted 1 bit with a 24-bit immediate
value and store the result in Rn. HHH is accessed as a
signed 16-bit word.
ASLA
Rn,Rn
1
1
Arithmetic shift left AGU register by 1 bit.
ASRA
Rn
1
1
Arithmetic shift right AGU register by 1 bit.
CMPA
Rn,Rn
1
1
24-bit compare between two AGU registers.
CMPA.W
Rn,Rn
1
1
16-bit compare between two AGU registers.
DECA
Rn
1
1
Decrement AGU register by one.
DECA.L
Rn
1
1
Decrement AGU register by two.
DECTSTA
Rn
1
1
Decrement and test AGU register.
ADDA
ADDA.L
4-42
Comments
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-37. AGU Arithmetic and Shifting Instructions (Continued)
Operation
Operands
C
W
Comments
LSRA
Rn
1
1
Logical shift right AGU register by 1 bit.
NEGA
Rn
1
1
Negate AGU register.
SUBA
Rn,Rn
1
1
Subtract the first operand from the second and store the
result in the second operand.
#<1–64>,SP
Subtract a 6-bit unsigned immediate value from the SP and
store in the stack pointer.
SXTA.B
Rn
1
1
Sign extend the value in an AGU register from bit 7.
SXTA.W
Rn
1
1
Sign extend the value in an AGU register from bit 15.
Rn,Rn
1
1
Transfer one AGU register to another.
TSTA.B
Rn
1
1
Test byte portion of an AGU register.
TSTA.L
Rn
1
1
Test long portion of an AGU register.
TSTA.W
Rn
1
1
Test word portion of an AGU register.
TSTDECA.W
Rn
3
1
Test and decrement AGU register.
TFRA
Note: Only operates on the lower 16 bits of the register;
the upper 8 bits are forced to zero.
ZXTA.B
Rn
1
1
Zero extend the value in an AGU register from bit 7.
ZXTA.W
Rn
1
1
Zero extend the value in an AGU register from bit 15.
Table 4-38. Bit-Manipulation Instructions
Operation
BFCHG
Operands
C
W
Comments
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFCHG tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Oterwise it is cleared. Then the operation
inverts all selected bits.
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
Freescale Semiconductor
All registers in DDDDD are permitted except HWS and Y.
Instruction Set Introduction
4-43
Instruction Set Introduction
Table 4-38. Bit-Manipulation Instructions (Continued)
Operation
BFCLR
BFSET
BFTSTH
4-44
Operands
C
W
Comments
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFCLR tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared. Then the operation
clears all selected bits.
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFTSTH tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared.
#<MASK16>,X:(Rn)
2
2
All registers in DDDDD are permitted except HWS and Y.
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
All registers in DDDDD are permitted except HWS and Y.
BFSET tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared. Then the operation sets
all selected bits.
All registers in DDDDD are permitted except HWS and Y.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-38. Bit-Manipulation Instructions (Continued)
Operation
BFTSTL
Operands
C
W
Comments
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFTSTL tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are clear, then the
C bit is set. Otherwise it is cleared.
#<MASK16>,X:(Rn)
2
2
All registers in DDDDD are permitted except HWS and Y.
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
Table 4-39. Branch-on-Bit-Manipulation Instructions
Operation
BRCLR
BRSET
Operands
C
W
Comments
#<MASK8>,DDDDD,<OFFSET7>
7/5
2
#<MASK8>,dd,<OFFSET7>
7/5
2
#<MASK8>,X:(Rn),<OFFSET7>
7/5
2
BRCLR tests all the targeted bits defined by
the immediate mask. If all the targeted bits are
clear, then the carry bit is set and a PC relative
branch occurs. Otherwise it is cleared and no
branch occurs.
#<MASK8>,X:(Rn+xxxx),<OFFSET7>
8/6
3
#<MASK8>,X:(SP–xx),<OFFSET7>
8/6
2
#<MASK8>,X:aa,<OFFSET7>
7/5
2
#<MASK8>,X:<<pp,<OFFSET7>
7/5
2
#<MASK8>,X:xxxx,<OFFSET7>
7/5
3
#<MASK8>,X:xxxxxx,<OFFSET7>
8/6
4
#<MASK8>,DDDDD,<OFFSET7>
7/5
2
#<MASK8>,dd,<OFFSET7>
7/5
2
#<MASK8>,X:(Rn),<OFFSET7>
7/5
2
#<MASK8>,X:(Rn+xxxx),<OFFSET7>
8/6
3
#<MASK8>,X:(SP–xx),<OFFSET7>
8/6
2
#<MASK8>,X:aa,<OFFSET7>
7/5
2
#<MASK8>,X:<<pp,<OFFSET7>
7/5
2
#<MASK8>,X:xxxx,<OFFSET7>
7/5
3
#<MASK8>,X:xxxxxx,<OFFSET7>
8/6
4
Freescale Semiconductor
All registers in DDDDD are permitted except
HWS and Y.
MASK8 specifies a 16-bit immediate value
where either the upper or lower 8 bits contain
all zeros.
BRSET tests all the targeted bits defined by
the immediate mask. If all the targeted bits are
set, then the carry bit is set and a PC relative
branch occurs. Otherwise it is cleared and no
branch occurs.
All registers in DDDDD are permitted except
HWS and Y.
MASK8 specifies a 16-bit immediate value
where either the upper or lower 8 bits contain
all zeros.
Instruction Set Introduction
4-45
Instruction Set Introduction
Table 4-40. Change-of-Flow Instructions
Operation
Bcc
BRA
BRAD
BSR
Operands
C
W
<OFFSET7>
5/3
1
7-bit signed PC-relative offset
<OFFSET18>
5/4
2
18-bit signed PC-relative offset
<OFFSET22>
6/5
3
22-bit signed PC-relative offset
<OFFSET7>
5
1
7-bit signed PC-relative offset
<OFFSET18>
5
2
18-bit signed PC-relative offset
<OFFSET22>
6
3
22-bit signed PC-relative offset
<OFFSET7>
3
1
Delayed branch with 7-bit signed offset;
must fill 2 delay slots (2 program words)
<OFFSET18>
3
2
Delayed branch with 18-bit signed offset;
must fill 2 delay slots (2 program words)
<OFFSET22>
4
3
Delayed branch with 22-bit signed offset;
must fill 2 delay slots (2 program words)
<OFFSET18>
5
2
18-bit signed PC-relative offset
<OFFSET22>
6
3
22-bit signed PC-relative offset
2
1
Delayed return from level 2 interrupt, restoring PC from the
FIRA register and the Y register from the stack in a fast interrupt procedure; must fill 2 delay slots (2 program words)
<ABS19>
5/4
2
19-bit absolute address
<ABS21>
6/5
3
21-bit absolute address
(N)
5
1
Jump to target contained in N register
<ABS19>
4
2
19-bit absolute address
<ABS21>
5
3
21-bit absolute address
<ABS19>
2
2
Delayed jump with 19-bit absolute address;
must fill 2 delay slots (2 program words)
<ABS21>
3
3
Delayed jump with 21-bit absolute address;
must fill 2 delay slots (2 program words)
(RRR)
5
1
Push 21-bit return address and jump to target address contained in RRR register
<ABS19>
4
2
Push 21-bit return address and jump to 19-bit target address
<ABS21>
5
3
Push 21-bit return address and jump to 21-bit target address
8
1
Return from interrupt, restoring 21-bit PC and SR from the
stack
FRTID
Jcc
JMP
JMPD
JSR
RTI
4-46
Comments
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
Table 4-40. Change-of-Flow Instructions (Continued)
Operation
Operands
C
W
Comments
RTID
5
1
Delayed return from interrupt, restoring 21-bit PC and SR from
the stack;
must fill 3 delay slots (3 program words)
RTS
8
1
Return from subroutine, restoring 21-bit PC from the stack
RTSD
5
1
Delayed return from subroutine, restoring 21-bit PC from the
stack;
must fill 3 delay slots (3 program words)
Information on delayed instruction execution is located in Section 9.3.4, “Non-Interruptible Instruction
Sequences,” on page 9-10.
Table 4-41. Looping Instructions
Operation
DO
Operands
C
W
Comments
#<1–63>,<ABS16>
3
2
Load LC register with unsigned value and start hardware DO loop
with 6-bit immediate loop count. Last address is 16-bit absolute.
Executes in 3 cycles when there is a minimum of 2 instruction
words in the loop.
#<1–63>,<ABS16>
5
2
Case of only 1 instruction word in loop body.
#<1–63>,<ABS21>
4
3
Last address is 21-bit absolute address.
Executes in 4 cycles when there is a minimum of 2 instruction
words in the loop.
#<1–63>,<ABS21>
6
3
Case of only 1 instruction word in loop body
DDDDD,<ABS16>
7
2
Load LC register with unsigned value. If LC is not equal to zero,
start hardware DO loop with 16-bit loop count in register. Otherwise skip body of loop (adds 2 additional cycles).
Last address is 16-bit absolute.
Any register is allowed except C2, D2, C0, D0,
C, D, Y, M01, N3, LA, LA2, LC, LC2, SR, OMR, and HWS.
When looping with a value in an accumuator, use A1, B1, C1, or
D1 to avoid saturation when reading the accumulator.
DDDDD,<ABS21>
8
3
Last address is 21-bit absolute address.
Any register is allowed except C2, D2, C0, D0,
C, D, Y, M01, N3, LA, LA2, LC, LC2, SR, OMR, and HWS.
When looping with a value in an accumuator, use A1, B1, C1, or
D1 to avoid saturation when reading the accumulator.
DOSLC
<ABS16>
3
2
If value in LC > 0, execute loop for specified number of times.
Otherwise skip body of loop (adds 3 additional cycles).
Last address is 16-bit absolute.
Minimum of 2 instructions words required in the loop.
<ABS21>
4
3
Last address is 21-bit absolute address.
Minimum of 2 instructions words required in the loop.
Freescale Semiconductor
Instruction Set Introduction
4-47
Instruction Set Introduction
Table 4-41. Looping Instructions (Continued)
Operation
Operands
ENDDO
C
W
1
1
Comments
Remove one value from the hardware stack and update the NL
and LF bits appropriately.
Note:
REP
Does not branch to the end of the loop.
#<0–63>
2
1
Hardware repeat of a 1-word instruction with immediate loop
count.
DDDDD
5
1
Hardware repeat of a 1-word instruction with loop count specified
in register.
If LC is not equal to zero, start hardware REP loop with 16-bit loop
count in register. Otherwise skip body of loop (adds 1 additional
cycle).
Any register is allowed except C2, D2, C0, D0,
C, D, Y, M01, N3, LA, LA2, LC, LC2, SR, OMR, and HWS.
When looping with a value in an accumuator, use A1, B1, C1, or
D1 to avoid saturation when reading the accumulator.
Table 4-42. Control Instructions
Operation
C
W
Comments
ALIGNSP
3
1
Save SP to the stack and align SP for long memory accesses, pointing to an empty location.
DEBUGEV
3
1
Generate a debug event.
DEBUGHLT
3
1
Enter the debug processing state.
ILLEGAL
4
1
Generate an illegal instruction exception; can be used to verify interrupt handlers for illegal instructions.
NOP
1
1
No operation.
STOP
*
1
Enter stop low-power mode.
The number of cycles is dependent upon chip implementation.
1
1
Generate an interrupt at priority level 0, 1, or 2 as specified by the
instruction.
SWI
4
1
Generate an interrupt at the highest priority level (level 3,
non-maskable).
SWILP
1
1
Generate an interrupt at the lowest priority level (lower than level 0).
WAIT
*
1
Enter wait low-power mode.
The number of cycles is dependent upon chip implementation.
SWI
4-48
Operands
#<0–2>
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Instruction Set Summary
4.4.5 Parallel Move Summary Tables
The following tables show the instructions that support move operations that are executed in parallel with
the execution of the primary instruction. Three types of parallel moves are supported: a move of data in
memory to a register, a move of a register value to memory, or two simultaneous moves of data from
memory to a register.
Table 4-43 summarizes the single parallel moves that are legal. Each instruction occupies only 1 program
word and executes in 1 cycle. Data transferred in a parallel move is always treated as a signed 16-bit word.
Table 4-43. Single Parallel Move Instructions
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
MAC
MPY
MACR
MPYR
Y1,X0,F
Y0,X0,F
Y1,Y0,F
Y0,Y0,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
A1,Y0,F
B1,Y1,F
MAC
MPY
MACR
C1,Y0,F
C1,Y1,F
MAC
–C1,Y0,F
–C1,Y1,F
ADD
SUB
CMP
X0,F
Y1,F
Y0,F
C,F
X0
Y1
Y0
X:(Rj)+
X:(Rj)+N
A
B
C
A1
B1
TFR
A,B
B,A
SAT
F,Y0
EOR.L
C,F
ABS
ASL
ASR
CLR
RND
TST
INC.W
DEC.W
NEG
F
SUBL2
A,D,B
X:(R1)+
AD
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.The “AD” destination notation indicates that both the A and D accumulators are written with the same 16-bit
value. Both extension registers are sign extended, and the F0 portion of both accumulators is set to $0000.
Freescale Semiconductor
Instruction Set Introduction
4-49
Instruction Set Introduction
Examples of instructions with a single parallel move appear in Example 4-6.
Example 4-6. Examples of Single Parallel Moves
MAC
MAC
MAC
ASL
ASL
Y1,X0,A
Y1,X0,A
-C1,Y0,A
B
B
X:(R0)+,X0
X0,X:(R0)+
X:(R0)+,C
X:(R0)+,Y1
Y1,X:(R0)+
Table 4-44 summarizes the dual parallel read instructions that are legal. Each instruction occupies only 1
program word and executes in 1 cycle. Data transferred in by each of the reads is always treated as a signed
16-bit word.
Table 4-44. Dual Parallel Read Instructions
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
MAC
MPY
MACR
MPYR
Y1,X0,F
Y1,Y0,F
Y0,X0,F
C1,Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R3)+
X:(R3)+N3
X0
X0,F
Y1,F
Y0,F
X:(R4)+
X:(R4)+N
Y0
ADD
SUB
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
A,B
B,A
TFR
A,B
B,A
CLR
ASL
ASR
F
MOVE.W
1.These instructions are not allowed when the XP bit in the OMR is set (that is, when the instructions are executing from data memory).
4-50
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Register-to-Register Moves
4.5 Register-to-Register Moves
As the instruction set summary shows, several different instructions are available for performing
register-to-register moves. Figure 4-1 summarizes these instructions to aid in choosing the correct
instruction.
Data Registers
TFR (36)
A2
A1
A0
MOVE.W (16)
B2
B1
B0
SAT (32)
C2
C1
C0
SXT.L (32)
D2
D1
D0
SXT.B (8)
Y1
ZXT.B (8)
Y0
ASL16 (36)
X0
ASR16 (36)
LSR16 (36)
MOVE.W (16)
MOVE.W (16)
MOVEU.W (16)
MOVE.L (32)
MOVE.L (24)
Pointer Registers
R0
R1
R2
TFRA* (24)
R3
MOVEU.W (16)
R4
R5
N
SP
* TFRA recommended for AGU register transfers
Figure 4-1. Moving Data in the Register Files
Freescale Semiconductor
Instruction Set Introduction
4-51
Instruction Set Introduction
4-52
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 5
Data Arithmetic Logic Unit
This chapter describes the architecture and operation of the data arithmetic logic unit (ALU).
Multiplication, arithmetic, logical, and shifting operations are performed in this block. (Note that addition
can also be performed in the address generation unit, and that the bit-manipulation unit can also perform
logical operations.)
The data ALU can perform the following operations with a throughput of 1 cycle per instruction, except
where noted:
•
Multiplication (with or without rounding)
•
Multiplication with negated product (with or without rounding)
•
Multiplication and accumulation (with or without rounding)
•
Multiplication and accumulation with negated product (with or without rounding)
•
Multi-precision multiplication support
•
Addition and subtraction
•
Increments and decrements (for 8-, 16-, 32-, and 36-bit operands)
•
Test and comparison (for 8-, 16-, 32-, and 36-bit operands)
•
Logical operations (AND, OR, and EOR)
•
One’s-complement and two’s-complement negation
•
Arithmetic and logical shifts
•
Rotates
•
Rounding
•
Absolute values
•
Sign extension and zero extension
•
Saturation (limiting) on data ALU and move operations
•
Conditional register moves
•
Division iteration
•
Normalization iterations (execute in 4 clock cycles)
Multiple buses within the data ALU allow complex arithmetic operations (such as a multiply-accumulate)
to execute in parallel with up to two memory transfers in a single execution cycle.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-1
Data Arithmetic Logic Unit
5.1 Data ALU Overview and Architecture
The major components of the data ALU are:
•
Three 16-bit data registers (X0, Y0, and Y1).
•
Four 36-bit accumulator registers (A, B, C, and D).
•
A single-cycle multiply-accumulator (MAC) unit.
•
A single-bit accumulator shifter.
•
An arithmetic and logical multi-bit shifter.
•
A MAC output limiter.
•
A data limiter.
A programming model of the data ALU unit is shown in Figure 5-1, and a block diagram is shown in
Figure 5-2 on page 5-3. The blocks and registers within the data ALU are explained in the following
sections.
35 32 31
16 15
0
A
A2
A1
A0
B
B2
B1
B0
C
C2
C1
C0
D
D2
D1
D0
15
Y
0
Y1
Y0
X0
Figure 5-1. Data ALU Programming Model
5-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data ALU Overview and Architecture
Data Registers
* For second access on dual parallel read
XDB2*
35 32 31
A2
(accesses X0 and C only)
CDBR
CDBW
Limiter
16 15
0
A1
A0
B2
B1
B0
C2
C1
C0
D2
D1
D0
Y1
Y0
X0
Optional Inverter
Arithmetic/Logical
Shifter
Shifter/MUX
Latch
MUX
36-Bit Accumulator
Shifter
Rounding Constant
OMR’s SA Bit
MAC Output Limiter
EXT:MSP:LSP
Condition Code
Generation
Condition Codes to Status Register
Figure 5-2. Data ALU Block Diagram
5.1.1 Data Registers (X0, Y1, Y0)
There are three independent 16-bit registers—X0, Y1, and Y0—that serve as data registers for operations
in the data ALU. The 16-bit Y1 register and the 16-bit Y0 register can be concatenated together to form a
32-bit register called Y, which is shown in Figure 5-3 on page 5-4. Y1 forms the most significant word and
Y0 forms the least significant word.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-3
Data Arithmetic Logic Unit
31
16 15
Y
0
Y1
Y0
MSP
LSP
32-Bit Y Register
Figure 5-3. The 32-Bit Y Register—Composed of Y1 Concatenated with Y0
The data registers are used as source or destination operands for most data ALU operations. With the use
of parallel move instructions (see Section 3.3.5, “Parallel Moves,” on page 3-11), these registers can serve
as sources for data ALU operations while new operands are loaded into them, in parallel, from memory.
This process is demonstrated in Example 5-1.
Example 5-1. X0 Register Used in Operation and Loaded in Parallel
ADD.W X0,A
X:(R0)+,X0
; X0 used and simultaneously loaded
The Y1, Y0, and X0 registers can be read or written as a byte or word operand. The Y register is read or
written as a long operand. All of the registers can be read or written using a parallel move. Only the X0
register can be written by the secondary read in a dual read instruction.
5.1.2 Accumulator Registers (A, B, C, D)
The data ALU contains four, independent, 36-bit accumulator registers that serve as the source or
destination for operations in the data ALU.
Each 36-bit data ALU accumulator register is composed of three different portions:
•
4-bit extension register, FF2 (where FF2 represents A2, B2, C2, or D2)
•
16-bit most significant product (MSP), FF1 (where FF1 represents A1, B1, C1, or D1)
•
16-bit least significant product (LSP), FF0 (where FF0 represents A0, B0, C0, or D0)
The “FF” notation is used throughout this chapter and the rest of the manual in references to the
accumulators. In this notation, FF refers to the entire accumulator (bits 35–0), FF2 refers only to the 4-bit
extension portion (bits 35–32), FF1 is the 16-bit most significant portion (bits 31–16), and FF0 is the 16-bit
least significant portion (bits 15–0). The various parts of an accumulator and the corresponding “FF”
notation are shown in Figure 5-4. Note that there is not actually an “FF” accumulator anywhere in the chip.
35
FF
32 31
FF2
Extension (FF2)
16 15
FF1
0
FF0
MSP (FF1)
LSP (FF0)
Long Portion of Accumulator (FF10)
Entire Accumulator (FF)
Figure 5-4. Different Components of an Accumulator (Using “FF” Notation)
5-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Data ALU Overview and Architecture
As Figure 5-4 on page 5-4 shows, it is also possible to directly address the 32-bit long-word portion of the
accumulator, which is referred to as FF10 in this notation. FF10 represents the concatenation of the FF1
and FF0 portions and is useful for manipulating 32-bit quantities.
The accumulators are used as source or destination operands for most data ALU operations. With the use
of parallel move instructions (see Section 3.3.5, “Parallel Moves,” on page 3-11), these registers can serve
as sources for data ALU operation while new operands are loaded into them, in parallel, from memory.
This process is demonstrated in Example 5-2.
Example 5-2. Accumulator A Used in Operation and Stored in Parallel
ADD.W X0,A
A,X:(R0)+
; A used and simultaneously stored
Each register can be read or written as a byte, word, or long operand. In a parallel move instruction, an
accumulator register is specified only as a whole accumulator and not in portions. Only the C register can
be written by the secondary read in a dual read instruction.
Section 5.2, “Accessing the Accumulator Registers,” discusses methods for accessing the accumulators
and strategies for using them properly.
NOTE:
The C2, C0, D2, and D0 portions of the C and D accumulators are
generally not directly accessible through the instruction set, with the
exception of certain operations. See Section 5.2.2, “Accessing Portions of
an Accumulator,” for ways to access these registers.
5.1.3 Multiply-Accumulator (MAC) and Logic Unit
The multiply-accumulator (MAC) and logic unit is the main arithmetic processing unit in the data ALU.
This block performs multiplications, additions, subtractions, logical operations, and other arithmetic
operations. It accepts up to three input operands and outputs one 36-bit result.
The MAC unit is pipelined to maintain a throughput of one instruction per cycle. The MAC pipeline has
two stages, multiplication and arithmetic/logical. Multiplication and MAC operations take 2 cycles to flow
through the two pipeline stages, whereas arithmetic and logical operations are completed in a single cycle.
More information on the two-stage execution of the MAC unit appears in Section 10.2.2, “Data ALU
Execution Stages,” on page 10-4.
The inputs of the MAC and logic unit can come from the seven data ALU registers (A1, B1, C1, D1, X0,
Y0, and Y1), can come from memory, or can be immediate data. Byte, word, and long operands are all
supported. Optional saturation and rounding are supported to ensure correct operation when 36-bit results
are written to memory. See Section 5.9, “Rounding,” for a more detailed discussion.
Arithmetic operations in the MAC unit occur independently and in parallel with memory accesses on the
core data buses. This capability allows a parallel move instruction to update an accumulator in the same
instruction in which the accumulator is used as the source for an ALU operation.
5.1.4 Single-Bit Accumulator Shifter
The accumulator shifter is an asynchronous parallel shifter with a 36-bit input and a 36-bit output. The
accumulator shifter is used to perform single-bit shifts of entire accumulators (as with the ASL and ASR
instructions), or to pre-shift values before they are passed on to the MAC unit (as occurs with the LSRAC
instruction).
Freescale Semiconductor
Data Arithmetic Logic Unit
5-5
Data Arithmetic Logic Unit
5.1.5 Arithmetic and Logical Shifter
An arithmetic and logical shifter block performs shifting of data ALU registers by an immediate value or
by a value specified in a register. The unit is pipelined to maintain a throughput of one instruction per cycle
for 16-bit shifting (one instruction per two cycles for 32-bit shifting). The pipeline has two stages. Shifting
is performed in the first stage, and the second stage can add the result of the first stage to an accumulator in
the ALU unit. Shifting operations take two cycles to flow through the two pipeline stages (three cycles for
32-bit shifts). More information on the two-stage execution of the shifter unit appears in Section 10.2.2,
“Data ALU Execution Stages,” on page 10-4.
5.1.6 Data Limiter and MAC Output Limiter
DSC algorithms can calculate values larger than the data precision of the machine when processing real
data streams. Normally a processor simply overflows such a result, but this treatment can create problems
for processing real-time signals. To eliminate the problems associated with overflow and underflow, the
DSP56800E provides the optional saturation of results using two limiters: the data limiter and the MAC
output limiter. The operation of the two limiter units is discussed in Section 5.8, “Saturation and Data
Limiting.”
5.2 Accessing the Accumulator Registers
The DSP56800E architecture provides four 36-bit accumulator registers for arithmetic operations. To
simplify the development of algorithms for signal processing and control, the DSP56800E provides three
methods for accessing the accumulators:
•
As an entire 36-bit register (FF)
•
As a 32-bit long register for store operations (FF10)
•
As individual component registers (FF2, FF1, or FF0)
Accessing an entire accumulator (A, B, C, or D) is particularly useful for DSC tasks because it preserves
the full precision of multiplication and other ALU operations. Using the full accumulator also provides
limiting (or saturation) capability when storing the result of a computation would cause overflow; see
Section 5.8.1, “Data Limiter.”
Accessing 32-bit long values (A10, B10, C10, or D10) is important for control tasks and general-purpose
computing. It allows long variables to be written to memory and stored to other registers without
saturation.
The ability to access individual portions of an accumulator (FF2, FF1, or FF0) provides a great deal of
flexibility when systems and control algorithms are implemented. Saturation is always disabled when
portions of an accumulator are manipulated, allowing for the accurate manipulation of integer values. This
access method also allows for accumulators to be saved and restored without limiting, preserving the full
precision of a mathematical result. See Section 5.2.6, “Saving and Restoring Accumulators,” for more
information.
Note that while the individual accumulator register portions are normally accessible, C2, C0, D2, and D0
are exceptions. Refer to Section 5.2.2, “Accessing Portions of an Accumulator,” for details on how to
access these portions.
Table 5-1 on page 5-7 summarizes the various possible accesses. These are described in more detail in the
following sections.
5-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Accessing the Accumulator Registers
Table 5-1. Accessing the Accumulator Registers
Register
Reading an Accumulator Register
Writing an Accumulator Register
A
B
C
D
Using a MOVE.W instruction:
If the extension bits are not in use, the 16-bit
contents of the FF1 portion of the accumulator
are read.
If the extension bits are in use, a 16-bit “limited”
value is substituted. See Section 5.8.1, “Data
Limiter.”
Using a MOVE.W instruction:
The 16-bit value is written to the FF1 portion of the
accumulator. The extension portion, FF2, is filled
with sign extension; the FF0 portion is set to zero.
When used in an arithmetic operation:
All 36 bits are used without limiting.
Using a MOVE.B instruction:
The 8-bit value is written into the lower 8 bits of
the FF1 portion of the register. The upper 8 bits of
the FF1 portion and the extension portion, FF2,
are sign extended (zero extended on MOVEU.B).
The FF0 portion is set to zero.
Using a MOVE.L instruction:
All 32 bits of the CDBR bus are written to the FF1
and FF0 portions of the register, FF1:FF0.
The FF2 register is written with sign extension.
A10
B10
C10
D10
Using a MOVE.L instruction:
The 32 bits in the FF1 and FF0 portions of the
accumulator are read.
Saturation logic is bypassed on MOVE.L.
Not available as a destination. Long-word values
must be written to the entire accumulator.
A2
B2
Using a MOVE.W instruction:
The 4-bit register, sign extended to 16 bits, is
read. (See Figure 5-8 on page 5-11.)
Using a MOVE.W instruction:
The 4 LSBs of the 16-bit value are written into the
register. The upper 12 bits are ignored. The corresponding FF1 and FF0 portions are not modified.
(See Figure 5-7 on page 5-10.)
A1
B1
C1
D1
Using a MOVE.W instruction:
The 16-bit FF1 portion is read.
Using a MOVE.W instruction:
The 16-bit value is written into the FF1 register.
The corresponding FF2 and FF0 portions are not
modified.
Using a MOVE.B instruction:
The lower 8 bits of FF1 are read.
When used in an arithmetic operation:
The FF1 register is used as a 16-bit source
operand for an arithmetic operation.
FF1 is also used for unsigned moves
(MOVEU.B, MOVEU.W) and with byte pointer
operations (MOVE.BP, MOVEU.BP).
A0
B0
Using a MOVE.W instruction:
The 16-bit FF0 register is read.
Using a MOVE.W instruction:
The 16-bit value is written into the FF0 register.
The corresponding FF2 and FF1 portions are not
modified.
Note: In all cases where MOVE.W is supported, the MOVEU.W instruction, parallel moves, and bit-manipulation
operations are also supported.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-7
Data Arithmetic Logic Unit
5.2.1 Accessing an Entire Accumulator
The accumulator registers serve as the source or destination for most data ALU operations. The result of an
ALU or multiplication operation is typically a full 36-bit value that, when written to an accumulator,
affects the entire register. Inputs for most arithmetic operations are also full-precision 36-bit accumulator
values.
The entire accumulator register can also be accessed with the explicit execution of a MOVE instruction.
Contents from the 32-bit CDBR bus can be written to all accumulators (A, B, C, or D) with sign extension
propagated to the 4-bit extension register (A2, B2, C2, or D2). When the contents of the 36-bit accumulator
need to be limited, the SAT instruction can be used to saturate the value in the 36-bit accumulator, limiting
with the full-scale positive or negative 32-bit values ($7FFF:FFFF or $8000:0000).
5.2.1.1 Writing an Accumulator with a Small Operand
Automatic sign or zero extension of the 36-bit accumulators is provided when the FF accumulator is
written with a smaller size operand. The extension can occur when FF is written from the CDBR
(MOVE.B, MOVEU.B, MOVE.W, or MOVE.L instruction) or with the results of certain data ALU
operations (for example, ADD.L, SUB.L, or TFR from a 16-bit register to a 36-bit accumulator). If a word
operand is to be written to an accumulator register (FF), the FF1 portion of the accumulator is written with
the word operand, the FF0 portion is zeroed, and the FF2 portion receives sign extension.
Figure 5-5 shows some examples of writing word values to an accumulator. Note that all three portions of
the accumulator are modified by these instructions.
Writing a Positive Value into 36-Bit Accumulator: MOVE.W #$1234,B
Before Execution
B2
B
X
After Execution
B1
X
X
B0
X
35 32 31
X
X
X
B2
X
X
16 15
B
0
0
B1
1
2
B0
3
35 32 31
4
0
0
0
0
16 15
0
Writing a Negative Value into 36-Bit Accumulator: MOVE.W #$A987,B
Before Execution
B2
B
X
After Execution
B1
X
35 32 31
X
B0
X
X
X
16 15
X
B2
X
X
B
0
F
B1
A
9
B0
8
35 32 31
7
0
16 15
0
0
0
0
Figure 5-5. Writing the Accumulator as a Whole
A move instruction that moves one accumulator to another, or a MOVE.L instruction with an immediate
value, behaves similarly. This result does not occur for the TFR instruction; no sign extension is performed
when TFR transfers a smaller register to an accumulator.
When an unsigned value is moved into an accumulator, the extension (FF2) portion of the accumulator
must be cleared because the most significant bit might be set. Automatic sign extension causes this bit to
be propagated into the extension register, making the value negative. Unsigned loads of words or long
words to an accumulator are performed using the technique in Example 5-3 on page 5-9.
5-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Accessing the Accumulator Registers
Example 5-3. Unsigned Load of a Long Word to an Accumulator
MOVE.L X:(R0),B
CLR.W B2
See Section 5.2, “Accessing the Accumulator Registers,” for a discussion of when it is appropriate to
access an accumulator by its individual portions and when it is appropriate to access an entire accumulator.
NOTE:
If the extension bits of an accumulator contain only sign extension (the E
bit in the status register is not set), saturation is unnecessary, and a read of
an entire accumulator is identical to a read of just the FF1 portion.
5.2.1.2 Using the Extension Registers
The extension registers (FF2) offer protection against 32-bit overflow. When the result of an accumulation
crosses the MSB of MSP (bit 31 of FF), the extension in use bit of the status register (E) is set. Up to 15
overflows or underflows are possible using the accumulator extension bits, after which the sign is lost
beyond the MSB of the extension register. When this loss occurs, the overflow bit (V) in the status register
is set. The extension register allows overflow during intermediate calculations without losing important
information. This capability is particularly useful during the execution of DSC algorithms, where
intermediate calculations might overflow.
The extension in use bit is used to determine when to saturate the value of an accumulator when it is
written to memory or when it is transferred to any data ALU register. If saturation occurs, the content of
the original accumulator is not affected (unless the same accumulator is specified as both source and
destination); only the value transferred is limited to a full-scale positive or negative 16-bit value ($7FFF or
$8000). This same logic applies to the SAT instruction.
When limiting occurs, the L flag in the status register is set. Saturation and limiting are explained in more
detail in Section 5.8, “Saturation and Data Limiting.”
NOTE:
Limiting is performed only when the entire 36-bit accumulator register
(FF) is specified as the source for a data move or is transferred to another
register. It is not performed when FF2, FF1, or FF0 is specified.
5.2.2 Accessing Portions of an Accumulator
The instruction set provides for loading and storing one portion of an accumulator register without
affecting the other two portions. When an instruction uses the FF1 or FF0 notation instead of F, the
instruction only operates on the specified 16-bit portion without modifying the other two portions. When
an instruction specifies FF2, the instruction operates only on the 4-bit accumulator extension register
without modifying the FF1 or FF0 portions of the accumulator. Refer to Table 5-1 on page 5-7 for a
summary of ways to access the accumulator registers.
Figure 5-6 on page 5-10 shows some examples of writing values to portions of the accumulator. Note that
only one of the three portions of the accumulator is modified by each of these instructions—the other two
portions remain unmodified.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-9
Data Arithmetic Logic Unit
Writing the FF2 Portion: MOVE.W #$ABCD,A2
Before Execution
A2
A
X
After Execution
A1
X
X
A0
X
35 32 31
X
X
X
A2
X
X
16 15
A
0
D
A1
X
X
A0
X
35 32 31
X
X
X
X
X
16 15
0
Writing the FF1 Portion: MOVE.W #$1234,A1
Before Execution
A2
A
X
After Execution
A1
X
X
A0
X
35 32 31
X
X
X
A2
X
X
16 15
A
0
X
A1
1
2
A0
3
35 32 31
4
X
X
X
X
16 15
0
Writing the FF0 Portion: MOVE.W #$A987,A0
Before Execution
A2
A
X
After Execution
A1
X
X
35 32 31
A0
X
X
X
X
A2
X
16 15
X
A
0
X
A1
X
X
A0
X
35 32 31
X
A
9
16 15
8
7
0
Figure 5-6. Writing the Accumulator by Portions
Limiting does not occur for move instructions that specify one portion of an accumulator as the source
operand.
When FF2 is written, it receives the low-order portion of the word; the high-order portion is not used. See
Figure 5-7. When FF2 is read, the register contents occupy the low-order portion (bits 3–0) of the word;
the high-order portion (bits 15–4) is sign extended. See Figure 5-8 on page 5-11.
15
4 3
0
CDBR Bus Contents
LSB of
Not Used
15
Word
4 3
0
Register FF2 Used
as a Destination
No Bits Present
FF2
Register FF2
Figure 5-7. Writing the Accumulator Extension Registers (FF2)
5-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Accessing the Accumulator Registers
15
Register FF2
4 3
FF2
No Bits Present
Used as a Source
0
Register FF2
LSB of
Word
15
4 3
0
Sign Extension
Contents
of FF2
of FF2
CDBW Bus Contents
Figure 5-8. Reading the Accumulator Extension Registers (FF2)
Although the FF1 portion of every accumulator is accessible by all instructions, the FF2 and FF0 portions
are only accessible for the A and B registers. The C2, C0, D2, and D0 accumulator portions are only
accessible through a limited set of instructions:
•
MOVE.W #xxxx,<register>
•
BFCHG, BFCLR, BFSET, ANDC, ORC, EORC, NOTC
•
BFTSTH, BFTSTL
•
BRSET, BRCLR
•
Push register to stack (C2 and D2 only)
•
Pop register from stack (C2 and D2 only)
There are no other ways to read or write these accumulator portions directly. To read or write the values of
C2 and D2, use the code in Example 5-4 and Example 5-5 (or similar code).
Example 5-4. Reading the Contents of the C2 Register
; First technique, with sign extension
ASR16 C,X0
; Shift C2 into X0 with sign extension
MOVE.W X0,R0
; Write C2 signed contents to final destination
; Second technique, no sign extension
LSR16 C,A
; Shift C2 into A1 with no sign extension
MOVE.W A1,R0
; Write C2 unsigned contents to final destination
Example 5-5. Writing a Value into the C2 Register
; First technique
MOVE.W R2,C1
ASL16 C
; Write value first to C1
; Shift the C1 register into C2
; Second technique
MOVE.W R2,A1
; Write value first to A1
ASL16 A,C
; Shift the A1 register into C2
; Third technique (may saturate if SA = 1)
MOVE.W R3,A2
; Write value first to A2
TFR
A,C
; Transfer value from A to C accumulator
Freescale Semiconductor
Data Arithmetic Logic Unit
5-11
Data Arithmetic Logic Unit
5.2.3 Reading and Writing Integer Data to an Accumulator
General integer and control processing typically uses 16-bit data. When an integer is loaded to an
accumulator, the 36 bits of the accumulator should reflect the 16-bit data correctly. During integer
processing, all accumulator loads of 16-bit data should clear the least significant portion of the
accumulator and sign extend the extension portion. Such loading is accomplished using the instruction
demonstrated in Example 5-6.
Example 5-6. Loading an Accumulator with an Integer Word
MOVE.W X:(R0),A
; A2 receives sign extension
; A1 receives the 16-bit data
; A0 receives the value $0000
In general, the A1 register should not be used when an accumulator is loaded with an integer. Using the
entire accumulator, as in Example 5-6, is almost always preferable. One exception to this rule is discussed
in Section 5.2.6, “Saving and Restoring Accumulators.”
The entire accumulator should also be used when long integers are loaded into the accumulators, as shown
in Example 5-7.
Example 5-7. Loading an Accumulator with a Long Integer
MOVE.L X:(R0),A
; A2 receives sign extension
; A1 receives the upper 16 of the 32 bits
; A0 receives the lower 16 of the 32 bits
NOTE:
It is not possible to use the A10 register when a long value is loaded into
an accumulator.
General integer and control processing does not use saturation or limiting. There is often no overflow
protection when the result of an integer calculation is read. Typically, the accumulators are read with
saturation disabled, as demonstrated in Example 5-8.
Example 5-8. Reading an Integer Value from an Accumulator
MOVE.W A1,X:Variable_1
MOVE.L A10,X:Long_Variable_1
; Word move with saturation disabled
; Long word move without saturation
Note the use of the A1 and A10 registers instead of the entire accumulator, A. Using this notation ensures
that saturation is disabled.
5.2.4 Reading 16-Bit Results of DSC Algorithms
A DSC algorithm can use the full 36-bit precision of an accumulator while performing DSC calculations
such as digital filtering or matrix multiplications. However, the 36-bit result must often be written to a
16-bit memory location or D/A converter. Because DSC algorithms process digital signals, it is important
that saturation is enabled when a 36-bit accumulator value is converted to a 16-bit value so that signals that
overflow 16 bits are clipped to the maximum positive or negative value appropriately. Saturation is
ensured when the entire accumulator (FF) is specified as the source operand, as shown in Example 5-9.
Example 5-9. Reading a Word from an Accumulator with Saturation
MOVE.W A,X:D_to_A_data
; Saturation is enabled
Note the use of the A accumulator instead of the A1 register. Using the A accumulator enables saturation.
5-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Accessing the Accumulator Registers
There is no instruction for reading a long value from an accumulator with saturation enabled. If this
function is required, the SAT instruction can be used, as shown in Example 5-10.
Example 5-10. Reading a Long Value from an Accumulator with Limiting
SAT
A
MOVE.L A10,X:D_to_A_data
; Limit the value in the A accumulator
; Saturation is no longer required
5.2.5 Converting a 36-Bit Accumulator to a 16-Bit Value
There are three useful techniques for converting the 36-bit contents of an accumulator to a 16-bit value,
which can then be stored to memory or used for further computation. This conversion is useful for
processing word-sized operands (16 bits) because it guarantees that an accumulator contains correct sign
extension and that the least significant 16 bits are all zeros. The three techniques appear in Example 5-11.
Example 5-11. Converting a 36-Bit Accumulator to a 16-Bit Value
;Converting with no limiting
MOVE.W A1,A ;Sign extend A2, A0 set to $0000
MOVE.W C1,B ;Sign extend B2, B0 set to $0000
;Extracting the A0 portion (no limiting)
ASL16 A
;Sign extend A2, write A1, clear A0
ASL16 A,D
;Sign extend D2, write D1, clear D0
;Converting with limiting enabled
MOVE.W A,A
;Sign extend A2, limit if required
MOVE.W A,C
;Sign extend C2, limit if required
In the last technique, where limiting is enabled, limiting only occurs when the extension register is in use.
Refer to Section 8.2.2, “Status Register,” on page 8-7. When the extension register is in use, the extension
in use (E) bit of the status register is set.
5.2.6 Saving and Restoring Accumulators
There are times when an accumulator value must be saved to the stack, such as in interrupt-handling
routines. To be saved and restored properly, the accumulator must be saved with saturation disabled. The
MOVE.W A,X:(SP)+ instruction should never be used when a value is being saved to the stack, because
this instruction operates with saturation enabled and can inadvertently store the value $7FFF or $8000 if
the extension register is in use. The solution is to save the individual portions of the accumulator, as
demonstrated in Example 5-12.
Example 5-12. Saving and Restoring an Accumulator—Word Accesses
; Saving the A accumulator to the stack
ADDA #1,SP
; Point to first empty location
MOVE.W A2,X:(SP)+
; Save extension register
MOVE.W A1,X:(SP)+
; Save A1 register
MOVE.W A0,X:(SP)
; Save A0 register
; Restoring the A accumulator from the stack
MOVE.W X:(SP)-,A0
; Restore A0 register
MOVE.W X:(SP)-,A1
; Restore A1 register
MOVE.W X:(SP)-,A2
; Restore extension register
A faster way of saving and restoring accumulators is to access the stack 32 bits at a time, as shown in
Example 5-13 on page 5-14.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-13
Data Arithmetic Logic Unit
Example 5-13. Saving and Restoring an Accumulator—Long Accesses
; Saving the A accumulator to the Stack
ADDA #2,SP
; Point to first empty location
MOVE.L A2,X:(SP)+
; Save extension register
MOVE.L A10,X:(SP)
; Save A1 and A0 registers
; Restoring the A accumulator from the Stack
MOVE.L X:(SP)-,A
; Restore A1 and A0 (changes A2)
MOVE.L X:(SP)-,A2
; Restore extension register
In order for the accumulator to be pushed on the stack 32 bits at a time, the stack pointer must be aligned to
an odd word address. See Section 3.5.3, “Accessing Long-Word Values Using Word Pointers,” on
page 3-19 for more information.
5.2.7 Bit-Manipulation Operations on Accumulators
The DSP56800E bit-manipulation instructions operate in a read-modify-write sequence: the value to be
manipulated is read into a temporary register, modified according to the instruction, and written back to its
original location. The “read” portion of this sequence is performed as if a MOVE.W instruction had been
executed, and thus may cause saturation to occur if an entire accumulator register is specified. In order for
bit-manipulation operations to execute correctly, saturation must be disabled. For this reason,
bit-manipulation instructions should always be performed on the FF1 portion of a register (A1, for
example) instead of the entire register, as demonstrated in Example 5-14.
Example 5-14. Bit Manipulation on a DSP56800E Accumulator
; BFSET using the A register
BFSET #$0F00,A
; Reads A1 with saturation enabled - can limit
; Sets bits 11 through 8 and stores back to A1
; A2 is sign extended and A0 is cleared
; BFSET using the A1 register
BFSET #$0F00,A1
; Reads A1 with saturation disabled
; Sets bits 11 through 8 and stores back to A1
; Note: A2 and A0 unmodified
5.3 Fractional and Integer Arithmetic
Fractional arithmetic is typically required for computation-intensive algorithms such as digital filters,
speech coders, vector and array processing, digital control, and other signal processing tasks. In this mode,
data is interpreted as fractional values, and computations are performed accordingly. When calculations
are performed in this mode, saturation is often used to prevent a problem that occurs without saturation: an
output signal that is generated from a result where a computation overflows without saturation can be
severely distorted (see Figure 5-27 on page 5-40). Saturation can be selectively enabled and disabled so
that intermediate calculations are performed without limiting and so that only the final results are limited.
Integer arithmetic is typically used in controller code, array indexing and address computations, peripheral
setup and handling, bit manipulation, bit-exact algorithms, and other general-purpose tasks. Typically,
saturation is not used when integers are processed, but it is available if desired.
5-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
5.3.1 DSP56800E Data Types
The DSP56800E architecture supports byte (8-bit), word (16-bit), and long-word (32-bit) integer data
types. It also supports word, long-word, and accumulator (36-bit) fractional data types.
Regardless of size, the four basic data types supported by the DSP56800E core are:
•
Signed integer.
•
Unsigned integer.
•
Signed fractional.
•
Unsigned fractional.
One of these four types is used in each data ALU operation. The complete list of data types and their
ranges appears in Table 5-2.
Table 5-2. Data Types and Range of Values
Data Type
Minimum Value
Maximum Value
Integer
Unsigned byte
0
255
–128
127
0
65,535
–32,768
32,767
0
4,294,967,295
–2,147,483,648
2,147,483,647
Signed byte
Unsigned word
Signed word
Unsigned long
Signed long
Fractional1
Signed word
–1.0
0.999 969 482 4
Signed long word
–1.0
0.999 999 999 5
Signed 36-bit accumulator
–16.0
15.999 999 999 5
1.All fractional values are rounded to 10 decimal digits of accuracy.
For more information on the DSP56800E data types, refer to Section 3.2, “DSP56800E Data Types,” on
page 3-5.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-15
Data Arithmetic Logic Unit
5.3.2 Addition and Subtraction
Addition, subtraction, and comparison operations are performed identically for both fractional and integer
data values. The data ALU does not distinguish between the data types for these operations.
To perform integer arithmetic operations with word-sized data, the MOVE.W instruction loads the data
into the FF1 portion of the accumulator as shown in Figure 5-9. FF2 contains sign extension and FF0 is
cleared. Note that the decimal (or binary) point lines up correctly for integer data in the two accumulators.
Integer Addition of 2 Words: 32 + 64 = 96
After Execution
Before Execution
A
B
$0
$0020
$0000
A2
A1
A0
$0
$0040
$0000
B2
B1
B0
MOVE.W
#32,A
MOVE.W
#64,B
ADD
B,A
MOVE.W
A1,X:RESULT
;
;
;
;
;
;
;
A
$0
$0060
$0000
A2
A1
A0
Load integer value “32” ($20) into A Accumulator
(Sign extends A2 and clears A0)
Load integer value “64” ($40) into B Accumulator
(Sign extends B2 and clears B0)
Perform Integer Word Addition
(32 + 64 = $20 + $40 = $60 = 96)
Save Result (without saturating) to Memory
Figure 5-9. Integer Word Addition
Fractional word arithmetic is performed in a similar manner. The MOVE.W instruction loads the data into
the FF1 portion of the accumulator as shown in Figure 5-10. FF2 contains sign extension and FF0 is
cleared. Again, the decimal (or binary) point lines up correctly for fractional data in the two accumulators.
Fractional Addition: 0.5 + 0.25 = 0.75
After Execution
Before Execution
A
B
$0
$4000
$0000
A2
A1
A0
$0
$2000
$0000
B2
B1
B0
MOVE.W
#0.5,A
MOVE.W
#0.25,B
ADD
B,A
MOVE.W
A,X:RESULT
;
;
;
;
;
;
;
A
$0
$6000
$0000
A2
A1
A0
Load fraction value “0.5” ($4000) into A
(Sign extends A2 and clears A0)
Load fraction value “0.25” ($2000) into B
(Sign extends B2 and clears B0)
Perform Fractional Word Addition
(0.5 + 0.25 = $4000 + $2000 = $6000 = 0.75)
Save Result (limiting enabled) to Memory
Figure 5-10. Fractional Word Addition
5-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
When a word-sized integer is added to a long-sized integer, the word value must first be converted to a
long value, as shown in Figure 5-11.
Integer Addition of a Long and a Word: 32 (long) + 64 (word) = 96 (long)
After Execution
Before Execution
A
B
$0
$0000
$0020
A2
A1
A0
$0
$0040
$0000
B2
B1
B0
MOVE.L
#32,A
;
;
#64,B
;
;
B
;
B,A
;
;
A10,X:RESULT ;
MOVE.W
ASR16
ADD
MOVE.L
A
B
$0
$0000
$0060
A2
A1
A0
$0
$0000
$0040
B2
B1
B0
Load integer long “32” ($20) into A Accumulator
(Sign extends A2 and A1)
Load integer word “64” ($40) into B Accumulator
(Sign extends B2 and clears B0)
Convert word value in B Accumulator to a long
Perform Integer Word Addition
(32 + 64 = $20 + $40 = $60 = 96)
Save Result (limiting disabled) to Memory
Figure 5-11. Adding a Word Integer to a Long-Word Integer
When a word-sized fraction is added to a long-sized fraction as shown in Figure 5-12, no conversion is
necessary because their binary points are the same.
Fractional Addition of a Long and a Word: 0.5 (long) + 0.25 (word) = 0.75 (long)
After Execution
Before Execution
A
B
MOVE.L
MOVE.W
ADD
MOVE.L
$0
$4000
$0000
A2
A1
A0
$0
$2000
$0000
B2
B1
B0
X:(R0),A
;
;
#0.25,B
;
;
;
B,A
;
;
A10,X:RESULT ;
A
B
$0
$6000
$0000
A2
A1
A0
$0
$2000
0000
B2
B1
B0
Load fraction long “0.5” ($4000:0000) into A
(Sign extends A2)
Load fraction word “0.25” ($2000) into B
(Sign extends B2 and clears B0)
(Note: Same format as a fractional long)
Perform Fractional Long Addition
(0.5 + 0.25 = $0:6000:0000 = 0.75)
Save Result (limiting disabled) to Memory
Figure 5-12. Adding a Word Fractional to a Long-Word Fractional
If limiting is desired before the long value is written to memory, it is necessary to use the SAT A,A
instruction immediately before the MOVE.L.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-17
Data Arithmetic Logic Unit
5.3.3 Multiplication
The multiplication operation is not the same for integer and fractional arithmetic. The result of a fractional
multiplication differs from the result of an integer multiplication. The difference amounts to a 1-bit shift of
the final result, as illustrated in Figure 5-13. Any binary multiplication of two N-bit signed numbers gives
a signed result that is 2N – 1 bits in length. This (2N – 1)-bit result must then be properly placed in a field
of 2N bits to fit correctly into the on-chip registers. For correct fractional multiplication, an extra zero bit is
inserted in the LSB to give a 2N-bit result. For correct integer multiplication, an extra sign bit is inserted in
the MSB to give a 2N-bit result.
Signed Multiplication: N × N = 2N – 1 Bits
Integer
S
Fractional
S
S
S
X
X
Signed Multiplier
S
S
MSP
LSP
N–1
N
Signed Multiplier
S
MSP
LSP
N
N–1
Sign Extension
2N Bits
0
Zero Fill
2N Bits
Figure 5-13. Comparison of Integer and Fractional Multiplication
The MPY, MAC, MPYR, and MACR instructions perform fractional multiplication and fractional
multiply-accumulation. The IMPY.W, IMPY.L, and IMAC.L instructions perform integer multiplication.
These types of multiplication are explained in more detail in the following sections.
5.3.3.1 Fractional Multiplication
Figure 5-14 on page 5-19 shows the multiplication of two 16-bit, signed, fractional operands. The
multiplication results in an intermediate 32-bit, signed, fractional result with the LSB always cleared. This
intermediate result is then stored in one of the 36-bit accumulators, with sign extension placed in the
extension register. If rounding is specified (using the MPYR instruction), the intermediate results is
rounded to 16 bits before being stored in the destination accumulator, and the LSP is cleared.
5-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
Input Operand 1
Input Operand 2
16 Bits
Signed Fractional
Input Operands
16 Bits
s
s
X
Signed Multiplier
Signed 31-Bit
Intermediate
Multiplier Result
s s
31 Bits
Signed Fractional
MPY Result
EXT
MSP
LSP
0
36 Bits
Figure 5-14. Fractional Multiplication (MPY)
5.3.3.2 Integer Multiplication
There are two techniques for performing integer multiplication on the DSC core:
•
Using the IMPY.W instruction to generate a 16-bit result in the FF1 portion of an accumulator
•
Using the IMPY.L and IMAC.L instructions to generate a 36-bit full-precision result
Each technique offers advantages for different types of computations.
Integer processing code usually requires only a 16-bit result, since greater precision is rarely needed. The
word-size integer multiplication instruction, IMPY.W, provides this capability, generating a 16-bit
unrounded result. Figure 5-15 on page 5-20 shows the multiply operation for integer arithmetic with a
word-sized result. The multiplication of two 16-bit, signed, integer operands using the IMPY.W instruction
gives a 16-bit, signed integer result that is placed in the FF1 portion of the accumulator. The corresponding
extension register (FF2) is filled with sign extension, and the FF0 portion remains unchanged.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-19
Data Arithmetic Logic Unit
Input Operand 1
Input Operand 2
16 Bits
Signed Integer
Input Operands
16 Bits
s
s
X
Signed Multiplier
16 Bits
Signed 31-Bit
Intermediate
Multiplier Result
Signed Integer
IMPY.W Result
s
31 Bits
MSP
(Sign Extension)
(Unchanged)
16 Bits
Figure 5-15. Integer Multiplication with Word-Sized Result (IMPY.W)
At other times, when it is necessary to maintain the full 32-bit precision of an integer multiplication, use
the IMPY.L instruction. Figure 5-16 shows an integer multiplication with a long-word result. The 32-bit
long integer result is placed into the FF1 and FF0 portions of an accumulator, with sign extension placed in
the extension register (FF2).
Input Operand 1
Input Operand 2
16 Bits
Signed Integer
Input Operands
16 Bits
s
s
X
Signed Multiplier
Signed 31-Bit
Intermediate
Multiplier Result
s s
31 Bits
Signed Integer
IMPY.L Result
EXT
MSP
LSP
36 Bits
Figure 5-16. Integer Multiplication with Long-Word-Sized Result (IMPY.L)
5.3.3.3 Operand Re-Ordering for Multiplication Instructions
The source operands for the three-operand multiplication and multiply-accumulate instructions must be
specified in a particular order so that they are dispatched to the appropriate units in the data ALU. The
5-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
DSP56800E assembler automatically rearranges the source operands for the following operations if they
are not specified in the required order:
MAC S1,S2,D
MAC –S1,S2,D
IMAC.L S1,S2,D
MACR S1,S2,D
MACR –S1,S2,D
IMPY.L S1,S2,D
MPY S1,S2,D
MPY –S1,S2,D
IMPY.W S1,S2,D
MPYR S1,S2,D
MPYR –S1,S2,D
This re-ordering by the assembler has no impact on the execution of the instruction. Note, however, that
the instruction dis-assembles as the re-ordered version. For example:
MPY
-X0,Y1,A
; X0 specified as first source operand
This instruction specifies the two source operands in the wrong order (the X0 register cannot be specified
as the first operand). The assembler replaces this instruction with the following:
MPY
-Y1,X0,A
; Y1 specified as first source operand
This instruction performs the same function, but with the operands in the proper order. Note that the
instruction always dis-assembles with the second ordering of operands.
5.3.4 Division
Fractional and integer division of both positive and signed values is supported using the DIV instruction.
The DIV instruction performs a single division iteration, calculating 1 bit of the result with each execution.
The dividend (numerator) is a 32-bit fractional or 31-bit integer value, and the divisor (denominator) is a
16-bit fractional or integer value. A full division requires that the DIV instruction be executed 16 times.
Algorithms for performing division can vary, depending on the values being divided and whether or not
the remainder after integer division must also be calculated. To formulate the correct approach, consider
the following key questions:
•
Are both operands always guaranteed to be positive?
•
Are operands fractional or integer?
•
Is the quotient all that is needed, or is the remainder needed as well?
•
Will the calculated quotient fit in 16 bits in integer division?
•
Are the operands signed or unsigned?
•
How many bits of precision are in the dividend?
•
What about overflow in fractional and integer division?
•
Will there be “integer division” effects?
Once you answer these questions, select the appropriate division algorithm. The most general division
algorithms are the fractional and integer algorithms for four-quadrant division,1 which generate both a
quotient and a remainder. These algorithms require the most time to complete and use the most registers.
Simpler, quicker algorithms can be used when positive numbers are divided or when the remainder is not
required. Note that none of the algorithms that are presented here apply to extended-precision division,
which requires more than 16 quotient bits.
1. Four-quadrant division is so called because it generates correct results for any combination of positive or negative dividends and divisors.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-21
Data Arithmetic Logic Unit
5.3.4.1 General-Purpose Four-Quadrant Division
This general-purpose algorithm generates both a correct quotient and a correct remainder when dividing
any combination of positive or negative, two’s-complement, signed values. Because this algorithm handles
the most general case, it is the slowest and uses the most resources. Example 5-15 presents one algorithm
for division with fractional numbers and another algorithm for the division of integer numbers.
Example 5-15. Signed Division with Remainder
; Four-Quadrant Division of Fractional, Signed Data (B1:B0 / X0)
; Generates Signed quotient and remainder
; Setup
MOVE.W B1,A
; Save sign bit of dividend (B1) in MSB of A1
MOVEU.WB1,N
; Save sign bit of dividend (B1) in MSB of N
ABS
B
; Force dividend positive
EOR
X0,Y1
; Save sign bit of quotient in N bit of SR
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instr
; Division
REP
16
DIV
X0,B
; Correct quotient
TFR
B,A
BGE
QDONE
; If correct result is positive, then done
NEG
B
; Else negate to get correct negative result
QDONE
MOVE.W A0,Y1
; Y1 <- True quotient
MOVE.W X0,A
; A <- Signed divisor
ABS
A
; A <- Absolute value of divisor
ADD
B,A
; A1 <- Restored remainder
BRCLR #$8000,N,DONE
MOVE.W #0,A0
NEG
A
DONE
; (At this point, the correctly signed quotient
; is in Y1 and the correct remainder is in A1)
; Four-Quadrant Division of Integer, Signed Data (B1:B0 / X0)
; Generates Signed quotient and remainder
; Setup
ASL
B
; Shift of dividend required for integer
; division
MOVE.W B1,A
; Save sign bit of dividend (B1) in MSB of A1
MOVEU.WB1,N
; Save sign bit of dividend (B1) in MSB of N
ABS
B
; Force dividend positive
EOR
X0,Y1
; Save sign bit of quotient in N bit of SR
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instr
;Division
REP
16
DIV
X0,B
; Correct quotient
TFR
B,A
BGE
QDONE
; If correct result is positive, then done
NEG
B
; Else negate to get correct negative result
QDONE
MOVE.W A0,Y1
; Y1 <- True quotient
MOVE.W X0,A
; A <- Signed divisor
ABS
A
; A <- Absolute Value of divisor
ADD
B,A
; A1 <- Restored remainder
BRCLR #$8000,N,DONE
MOVE.W #0,A0
NEG
A
ASR
B
; Shift required for correct integer remainder
DONE
; (At this point, signed quotient in Y1, correct
; remainder in A1)
5-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
5.3.4.2 Positive Dividend and Divisor with Remainder
If both the dividend and divisor are positive, signed, two’s-complement numbers, a more efficient
algorithm can replace the general-purpose four-quadrant approach. Consider a simple positive division
with a remainder, such as the following:
64 ÷ 9 = 7 (remainder 1)
This operation can be calculated correctly with the code presented in Example 5-16. The algorithms in this
code are the fastest and require the least amount of program memory. The example presents different
algorithms for the division of fractional and integer numbers. Both algorithms generate the correct positive
quotient and positive remainder.
Example 5-16. Unsigned Division with Remainder
; Division of Positive Fractional Data (B1:B0 / X0)
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instruction
REP
16
DIV
X0,B
; Form positive quotient in B0
ADD
X0,B
; Restore remainder in B1
; (At this point, the positive quotient is in
; B0 and the positive remainder is in B1)
; Division of Positive Integer Data (B1:B0 / X0)
ASL
B
; Shift of dividend required for integer
; division
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instruction
REP
16
DIV
X0,B
; Form positive quotient in B0
MOVE.W B0,Y1
; Save quotient in Y1
; (At this point, the positive quotient is in
; B0 but the remainder is not yet correct)
ADD
X0,B
; Restore remainder in B1
ASR
B
; Required for correct integer remainder
; (At this point, the correct positive
; remainder is in B1)
5.3.4.3 Signed Dividend and Divisor with No Remainder
An algorithm that is slightly more complex but still more efficient than the general-purpose algorithm can
be used for signed values when a correct remainder is not required.
The algorithms in Example 5-17 on page 5-24 are faster than the general-purpose algorithms because they
generate the quotient only; they do not generate a correct remainder. The example presents different
algorithms for the division of fractional and integer numbers.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-23
Data Arithmetic Logic Unit
Example 5-17. Signed DIvision Without Remainder
; Four-Quadrant Division of Signed Fractional Data (B1:B0 / X0)
; Generates signed quotient only, no remainder
; Setup
MOVE.W B,Y1
; Save Sign Bit of dividend (B1) in MSB of Y1
ABS
B
; Force dividend positive
EOR
X0,Y1
; Save sign bit of quotient in N bit of SR
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instr
; Division
REP
16
DIV
X0,B
; Form positive quotient in B0
; Correct quotient
BGE
DONE
; If correct result is positive, then done
NEG
B
; Else negate to get correct negative result
DONE
; (At this point, the correctly signed
; quotient is in B0 but the remainder is not
; correct)
; Four-Quadrant Division of Signed Integer Data (B1:B0 / X0)
; Generates signed quotient only, no remainder
; Setup
ASL
B
; Shift of dividend required for integer
; division
MOVE.W B,Y1
; Save Sign Bit of dividend (B1) in MSB of Y1
ABS
B
; Force dividend positive
EOR
X0,Y1
; Save sign bit of quotient in N bit of SR
BFCLR #$0001,SR
; Clear carry bit: required for 1st DIV instr
; Division
REP
16
DIV
X0,B
; Form positive quotient in B0
; Correct quotient
BGE
DONE
; If correct result is positive, then done
NEG
B
; Else negate to get correct negative result
DONE
; (At this point, the correctly signed
; quotient is in B0 but the remainder is not
; correct)
5.3.4.4 Division Overflow
Both integer and fractional division are subject to division overflow. Overflow occurs when the correct
value of the quotient does not fit into the destination available to store it. For the division of fractional
numbers, the result must be a 16-bit, signed, fractional value that satisfies the following equation:
–1.0 ≤ quotient < +1.0 – 2–15
When the magnitude of the dividend is larger than the magnitude of the divisor, this relation can never be
satisfied; the result is always larger in magnitude than 1.0. The dividend should be scaled to avoid this
condition.
Integer division can also overflow. Correct execution without overflow occurs only when the result of the
division fits within the range of a signed 16-bit word:
–2–15 ≤ quotient ≤ [215 – 1]
The numerator should be scaled if necessary to ensure this condition.
5-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Fractional and Integer Arithmetic
5.3.5 Logical Operations
The logic unit in the data ALU can perform 16- and 32-bit logical operations. All logical operations are
performed on the raw bits that are contained in the operands, regardless of whether they represent integer
or fractional values. Typically, logical operations are only performed on integer values, but the
DSP56800E supports logical operations on fractional values as well.
When logical operations are performed on 16-bit values, they operate on the FF1 portion of an accumulator
register or on any of the 16-bit data registers (X0, Y0, and Y1). Logical operations on 32-bit values are
performed on the FF1:FF0 portion of the accumulators and can also use the 32-bit Y register. Figure 5-17
shows examples of 16- and 32-bit logical operations.
16-Bit Logical Operation: AND.W
#$F,A
Before Execution
A2
A
1
After Execution
A1
2
3
A0
4
35 32 31
5
6
7
A2
8
9
16 15
A
0
1
A1
0
1
2
3
A0
4
5
6
6
7
8
9
16 15
0
After Execution
A1
35 32 31
5
#$F,A
Before Execution
A
A0
0
35 32 31
32-Bit Logical Operation: AND.L
A2
0
7
A2
8
16 15
9
A
0
1
A1
0
0
A0
0
35 32 31
0
0
16 15
0
0
9
0
Figure 5-17. 16- and 32-Bit Logical Operations
Logical AND, OR, and EOR operations are supported for both 16- and 32-bit operands. A logical NOT
operation is also supported, but only for 16-bit operands. See Chapter 4, “Instruction Set Introduction,”
and the appropriate sections in Appendix A, “Instruction Set Details,” for more information on the logical
operation instructions.
5.3.6 Shifting Operations
A variety of shifting operations can be done on both integer and fractional data values. For both types of
data, an arithmetic left shift of 1 bit corresponds to a multiplication by two. An arithmetic right shift of
1 bit corresponds to a signed division by two, and a logical right shift of 1 bit corresponds to an unsigned
division by two.
5.3.6.1 Shifting 16-Bit Words
The shifter performs single-cycle arithmetic or logical shifts of 0 to 15 bits on 16-bit word values.
Figure 5-18 on page 5-26 shows both right and left shifting of a 16-bit word.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-25
Data Arithmetic Logic Unit
EXT
A
F
$AAAA
$4
$AAAA
16
4
16
4
Barrel Shifting
Barrel Shifting
Unit
Unit
MSP
F
$5
A
LSP
A
A
35 32 31
0
0
0
16 15
EXT
0
A
0
0
MSP
5
5
LSP
4
35 32 31
Example: Right Shifting (ASRR.W)
0
0
0
0
16 15
0
0
Example: Left Shifting (ASLL.W)
Figure 5-18. Arithmetic Shifts on 16-Bit Words
At the completion of a 16-bit logical or arithmetic shift, the extension register is loaded with sign extension
and the LSP is cleared. The extension bits are never shifted into the MSP of an accumulator, nor are bits in
the MSP ever shifted into the extension.
Note that sign extension is always performed for 16-bit shifts. In the unusual case in which a negative
value is shifted by zero and its destination is an accumulator, the extension register of the destination is
loaded with $F instead of $0.
5.3.6.2 Shifting 32-Bit Long Words
The shifter can also perform arithmetic or logical shifts of 0 to 31 bits on 32-bit data. If the number of bits
to be shifted is specified using a data ALU register and is positive, the shifting is performed in the direction
indicated by the mnemonic (for example, an ASRR.L instruction shifts right). If the number of bits to shift
is specified by a register and is a negative value, the shifting is performed in the opposite direction by the
absolute value of the number of bits to be shifted (for example, an ASRR.L instruction shifts left).
Figure 5-19 shows both right and left shifting of a 32-bit long word.
$AAAACCCC
$4
32
EXT
A
F
$5555CCCC
4
F
A
32
4
Shifting
Shifting
Unit
Unit
MSP
35 32 31
$4
A
LSP
A
A
C
C
16 15
Example: Right Shifting (ASRR.L)
EXT
C
0
A
0
MSP
5
35 32 31
5
5
LSP
C
C
C
C
16 15
0
0
Example: Left Shifting (ASLL.L)
Figure 5-19. Arithmetic Shifts on 32-Bit Long Words
At the completion of a 32-bit logical shift, the extension register is always cleared. At the end of an
arithmetic shift, the extension register is sign extended. The extension bits are never shifted into the MSP
of an accumulator, nor are bits in the MSP ever shifted into the extension.
5-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Unsigned Arithmetic Operations
5.3.6.3 Shifting Accumulators by 16 Bits
Three instructions—ASL16, ASR16, and LSR16—shift an entire 36-bit accumulator by 16 bits in 1 cycle.
LSR16 and ASR16 logically or arithmetically shift a 36-bit accumulator 16 bits to the right, and are useful
for converting 16-bit values to 32-bit long values that are unsigned and signed, respectively. When it is
necessary to convert a 16-bit value to a 32-bit integer, the FF1 portion must be shifted into the FF0 portion,
and the FF2 portion must be shifted into the 4 LSBs of the FF1 portion. In this manner, the original 16-bit
value is represented as a 32-bit integer. The ASL16 instruction shifts a 36-bit accumulator 16 bits to the
left, filling the FF0 portion with $0000 and the extension register with what were previously the 4 LSBs of
the original FF1 portion.
5.3.6.4 Shifting with Accumulation
The ASRAC and LSRAC instructions are unique in that they arithmetically or logically right shift a 16-bit
value into a 32-bit field and add the result to the previous value of the accumulator. For these two
instructions, the least significant bits of the MSP are shifted into the most significant bits of the LSP.
5.4 Unsigned Arithmetic Operations
The DSP56800E can perform both unsigned and signed arithmetic operations. The addition, subtraction,
multiplication, and comparison instructions work for both signed and unsigned values, but the condition
code computations are different.
5.4.1 Condition Codes for Unsigned Operations
Unsigned arithmetic operations such as addition, subtraction, comparison, and logical operations are
performed with the same instructions, and in the same manner, as for signed computations. The difference
between signed and unsigned operations involves how the data is interpreted (Section 3.2.1, “Data
Formats,” on page 3-6) and which status bits are affected when comparing signed and unsigned numbers.
The difference in the way condition codes are calculated is most evident with any of the conditional jump
and branch instructions, such as Bcc and Jcc. These instructions perform an operation based on the state of
the condition codes, which may be set differently depending on whether a signed or unsigned calculation
has been performed to generate the value tested by the instruction.
Specifically, the following conditions should be used for signed values:
•
GE—greater than or equal to
•
LE—less than or equal to
•
GT—greater than
•
LT—less than
These conditions should be used instead for unsigned values:
•
HS (high or same)—unsigned greater than or equal to
•
LS (low or same)—unsigned less than or equal to
•
HI (high)—unsigned greater than
•
LO (low)—unsigned less than
Note that the HS condition is identical to carry clear (CC) and that LO is identical to carry set (CS).
Freescale Semiconductor
Data Arithmetic Logic Unit
5-27
Data Arithmetic Logic Unit
Accumulator extension registers can also interfere with the correct calculation of condition codes for
unsigned numbers when an arithmetic operation generates a 36-bit result. The TST and CMP instructions,
among others, exhibit this problem.
On the DSP56800, the recommended solution was to set the CM bit in the OMR register before using any
of the unsigned jump and branch conditions (HS, LS, HI, and LO) after a TST or CMP instruction. For
DSP56800E code, use of the CM bit is not generally recommended. Instead, instructions that exactly
match the size of the data should be used:
•
TST.B and CMP.B for bytes
•
TST.W and CMP.W for words
•
TST.L and CMP.L for long words
Using these instructions guarantees that the extension registers are not considered when condition codes
are calculated.
5.4.2 Unsigned Single-Precision Multiplication
Unsigned multiplications are supported with the IMPYUU instruction, which accepts two 16-bit
multiplicands from the lowest portion of the accumulators (FF0). This instruction is illustrated in
Example 5-18.
Example 5-18. Multiplication of 2 Unsigned Words
MOVE.W X:(R0),A
MOVE.W X:(SP-2),B
; Load 1 word from memory
; Load 1 word from memory
LSR16 A
LSR16 B
; Place unsigned value in FF0 portion
; Place unsigned value in FF0 portion
IMPYUU A0,B0,D
; Multiply 2 unsigned words
The IMACUS and IMPYSU instructions are provided for multiplying one signed value and one unsigned
value. However, be careful with these instructions, because one of the 16-bit multiplicands is in the upper
portion (FF1) of an accumulator, and the other is in the lower portion (FF0). See the entries for these
instructions in Appendix A, “Instruction Set Details,” for more information on the placement of operands.
Fractional unsigned multiplications are supported with the MPYSU and MACSU instructions. Again, be
careful, because one of the 16-bit multiplicands is in the upper portion (FF1) of an accumulator, and the
other is in the lower portion (FF0).
5-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Extended- and Multi-Precision Operations
5.5 Extended- and Multi-Precision Operations
Some algorithms require calculations that exceed the range or precision of the 16- and 32-bit operations
that the DSP56800E architecture supports. To assist in implementing these algorithms, the DSP56800E
provides several instructions targeted toward extended-precision and multi-precision calculations.
5.5.1 Extended-Precision Addition and Subtraction
Two instructions, ADC and SBC, assist in performing extended-precision addition and subtraction.
Example 5-19 illustrates the use of the ADC instruction in 64-bit addition. Two 64-bit operands in memory
are summed, 32 bits at a time, with the carry out of the low-order addition added into the high-order
portion. The final sum is stored in both the A and B registers.
Example 5-19. 64-Bit Addition
X:$103:X:$102:X:$101:X:$100 + X:$203:X:$202:X:$201:X:$200 = A2:A1:A0:B1:B0
MOVE.L X:$100,B
MOVE.L X:$200,Y
ADD
Y,B
MOVE.L X:$102,A
MOVE.L X:$202,Y
ADC
Y,A
;
;
;
;
;
;
Get Operand1 (Lower 32
Get Operand2 (Lower 32
First 32-bit addition,
Get Operand1 (Upper 32
Get Operand2 (Upper 32
Second 32-bit addition
bits, sign ext)
bits)
bits)
bits)
Subtraction is carried out in a similar manner. As illustrated in Example 5-20, the low-order 32-bit
subtraction is performed first, with any borrow being reflected in the carry bit in the status register. The
high-order subtraction is then performed, with the borrow subtracted to achieve the correct result.
Example 5-20. 64-Bit Subtraction
X:$103:X:$102:X:$101:X:$100 – X:$203:X:$202:X:$201:X:$200 = A2:A1:A0:B1:B0
MOVE.L X:$100,B
MOVE.L X:$200,Y
SUB
Y,B
MOVE.L X:$102,A
MOVE.L X:$202,Y
SBC
Y,B
;
;
;
;
;
;
Get Operand1 (Lower 32 bits, sign ext.)
Get Operand2 (Lower 32 bits)
First 32-bit subtraction
Get Operand1 (Upper 32 bits)
Get Operand2 (Upper 32 bits)
Second 32-bit subtraction
5.5.2 Multi-Precision Fractional Multiplication
Two instructions are provided to assist with multi-precision multiplications: MPYSU and MACSU. When
these instructions are used, the multiplier accepts one signed two’s-complement operand and one unsigned
two’s-complement operand.
Figure 5-20 on page 5-30 shows the process for multiplying a 16-bit value with a 32-bit value, resulting in
a 36-bit product. The 16-bit value is multiplied by each of the 16-bit halves of the larger value, and the
results are summed, with the second product offset by 16 bits so the products align properly.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-29
Data Arithmetic Logic Unit
32 Bits
Y1
Y0
16 Bits
X0
×
Signed × Unsigned
X0 × Y0
Signed × Signed
X0 × Y1
+
Sign Ext.
A2
A1
A0
36 Bits
Figure 5-20. Single-Precision-Times-Double-Precision Signed Multiplication
The key to making the multiplication work is the use of the MPYSU instruction, as shown in the code in
Example 5-21. Treating the lower half of the 32-bit input value as unsigned ensures that the correct value is
generated for the later addition.
Example 5-21. Fractional Single-Precision Times Double-Precision—Both Signed
(4 Cycles, 4 Instruction Words)
MPYSU X0,Y0,A
ASR16 A
MAC
X0,Y1,A
;
;
;
;
Single-Precision times Lower Portion
16-bit Arithmetic Right Shift
Single-Precision times Upper Portion
and added to Previous
Extended-precision 32-bit multiplication works similarly. Figure 5-21 on page 5-31 shows two 32-bit
values being multiplied to generate a 64-bit result. The code for this figure appears in Example 5-22 on
page 5-32.
5-30
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Extended- and Multi-Precision Operations
32 Bits
OP1UPR
OP1LWR
32 Bits
OP2UPR
OP2LWR
×
Unsigned × Unsigned
OP2LWR × OP1LWR
Unsigned × Signed
OP2LWR × OP1UPR
Signed × Unsigned
OP2UPR × OP1LWR
Signed × Signed
OP2UPR × OP1UPR
+
RES2UPR
RES2LWR
RES1UPR
RES1LWR
64 Bits
Figure 5-21. Double-Precision-Times-Double-Precision Signed Multiplication
Freescale Semiconductor
Data Arithmetic Logic Unit
5-31
Data Arithmetic Logic Unit
Example 5-22. Multiplying Two Fractional Double-Precision Values
X:OP1UPR:X:OP1LWR × X:OP2UPR:X:OP2LWR
(Both 32-Bit Operands Are Signed)
; Unsigned x Unsigned Multiplication, save lower 16 bits of final result
MOVE.W
X:OP1UPR,A
; Get first operand from memory
MOVE.W
X:OP1LWR,A0
; Could use a MOVE.L to move 32-bit value to A
MOVE.W
X:OP2UPR,B
; Get first operand from memory
MOVE.W
X:OP2LWR,B0
; Could use a MOVE.L to move 32-bit value to B
IMPYUU
A0,B0,D
; Perform lower portion of multiplication
LSR16
D,C
; Isolate upper 16 bits for accumulation
; LSP of D for RES1LWR
; Signed x Unsigned Multiplication with Accumulation
IMPYSU
A1,B0,Y
; Perform signed multiplication with upper 16 bits
ADD
Y,C
; Accumulate result
; Unsigned x Signed Multiplication with Accumulation
MOVE.L
#0,Y
IMACUS
A0,B1,Y
; Perform signed multiplication with upper 16 bits
ADD
Y,C
; Accumulate result
; Lower 16 bits Correspond to Lower 32 bits of Final Result
ASL16
C,Y1
; Save lower 16 bits of result
MOVE.W
Y1,D1
; D has lower 32 bits of result
; MSP of D for RES1UPR
; Upper 16 bits Correspond to Upper 32 bits of Final Result
ASR16
C
; Isolate upper 16 bits for accumulation
IMAC.L
A1,B1,C
; Perform upper portion of multiplication
; Correction for Fractional Result (C => RES2UPR:RES2LWR, D => RES1UPR:RES1LWR)
SXT.L
D
; Propagate bit 31 to EXT of D
ASL
D
; Corresponds to lower 32 bits of Final Fractional
; Result
ROL.L
C
; Corresponds to upper 32 bits of Final Fractional
; Result
; Storing 64-bit Fractional Result in Memory
MOVE.L D10,X:RES1
; X:RES1UPR:RES1LWR = Lower 32 bits of Fractional
; Result
MOVE.L C10,X:RES2
; X:RES2UPR:RES2LWR = Upper 32 bits of Fractional
; Result
; ====> C2 may not be correct after the result is generated ...
This type of multiplication can also be performed as a 32 × 32 → 64-bit integer multiplication with a final
left shift of the result. Multi-precision integer multiplication is described in Section 5.5.3, “Multi-Precision
Integer Multiplication.”
5.5.3 Multi-Precision Integer Multiplication
Four provided instructions assist with multi-precision integer multiplications. When these instructions are
used, the multiplier accepts signed two’s-complement operands and unsigned two’s-complement operands.
Each instruction specifies not only which source operand is signed or unsigned, but also the location of the
16-bit operand (FF1 or FF0 portion of an accumulator):
•
IMACUU—multiply-accumulate with two unsigned operands
(first 16-bit operand located in FF0 portion, second in FF1)
•
IMACUS—multiply-accumulate with one unsigned and one signed operand
(unsigned 16-bit operand located in FF0 portion, signed in FF1)
5-32
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Extended- and Multi-Precision Operations
•
IMPYSU—multiplication with one signed and one unsigned operand
(signed 16-bit operand located in FF1 portion, unsigned in FF0)
•
IMPYUU—multiplication with two unsigned operands (2 cases)
(each unsigned 16-bit operand located in the FF0 portion)
(first 16-bit operand located in FF1 portion, second in FF0)
The following sections demonstrate the use of these instructions in multi-precision integer multiplications.
5.5.3.1 Signed 32-Bit × Signed 32-Bit with 32-Bit Result
Figure 5-22 and Example 5-23 demonstrate a signed multiplication of two 32-bit long values that generates
a 32-bit long integer result.
32 Bits
A1
A0
32 Bits
B1
×
B0
Unsigned × Unsigned
A0 × B0
Signed × Unsigned
A1 × B0
Unsigned × Signed
A0 × B1
+
C2
C1
C0
32 Bits
Figure 5-22. 32-Bit × 32-Bit –> 32-Bit Signed Integer Multiplication
Example 5-23. Multiplying Two Signed Long Integers
C1:C0 = A1:A0 × B1:B0
(Both 32-Bit Operands Are Signed)
;Signed x Signed 32-Bit Integer
IMPYSU A1,B0,Y
IMACUS A0,B1,Y
IMPYUU A0,B0,C
ADD
Y0,C
Multiplication
; Y1:Y0 = signed A1 x unsigned B0
; Y1:Y0 = unsigned A0 x signed B1 + Y1:Y0
; C2:C1:C0 = unsigned A0 x unsigned B0
; Combine Results: final 32-bit result in C
This example, which saves only the lower 32 bits of the result, does not require the A1 × B1 product,
which only affects the upper 32 bits of the result. Also note that C2 in the final result is modified and does
not contain valid data.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-33
Data Arithmetic Logic Unit
5.5.3.2 Unsigned 32-Bit × Unsigned 32-Bit with 32-Bit Result
Figure 5-23 and Example 5-24 demonstrate an unsigned multiplication of two 32-bit long values that
generates a 32-bit long integer result.
32 Bits
A1
A0
32 Bits
B1
×
B0
Unsigned × Unsigned
A0 × B0
Unsigned × Unsigned
A1 × B0
Unsigned × Unsigned
A0 × B1
+
C2
C1
C0
32 Bits
Figure 5-23. 32-Bit × 32-Bit –> 32-Bit Unsigned Integer Multiplication
Example 5-24. Multiplying Two Unsigned Long Integers
C1:C0 = A1:A0 × B1:B0
(Both 32-Bit Operands Are Unsigned)
;Unsigned x Unsigned 32-Bit Integer Multiplication
IMPYUU A1,B0,Y
; Y1:Y0 = signed A1 x unsigned B0
IMACUU A0,B1,Y
; Y1:Y0 = unsigned A0 x signed B1 + Y1:Y0
IMPYUU A0,B0,C
; C2:C1:C0 = unsigned A0 x unsigned B0
ADD
Y0,C
; Combine Results: final 32-bit result in C
This example, which saves only the lower 32 bits of the result, does not require the A1 × B1 product,
which only affects the upper 32 bits of the result. Also note that C2 in the final result is modified and does
not contain valid data.
5.5.3.3 Signed 32-Bit × Signed 32-Bit with 64-Bit Result
Figure 5-24 on page 5-35 and Example 5-25 on page 5-35 demonstrate a signed multiplication of two
32-bit long values that generates a 64-bit full-precision integer result.
5-34
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Extended- and Multi-Precision Operations
32 Bits
A1
A0
32 Bits
B1
×
B0
Unsigned × Unsigned
A0 × B0
Signed × Unsigned
A1 × B0
Unsigned × Signed
A0 × B1
Signed × Signed
A0 × B1
+
C2
C1
C0
D1
D0
64 Bits
Figure 5-24. 32-Bit × 32-Bit –> 64-Bit Signed Integer Multiplication
Example 5-25. Multiplying Two Signed Long Integers
D2:D1:D0:C1:C0 = A1:A0 × B1:B0
(Both 32-Bit Operands Are Signed)
;Signed x Signed 32-Bit Integer
IMPYUU A0,B0,D
LSR16 D,C
IMPYSU A1,B0,Y
ADD
Y,C
ASL16 X0,Y
IMACUS A0,B1,Y
ADD
Y,C
ASL16 C0,Y1
MOVE.W Y1,D1
ASR16 C,C
IMAC.L A,B,C
Multiplication with 64-Bit Result
; D2:D1:D0 = unsigned A0 x unsigned B0
; Align upper word of first product in C
; Y1:Y0 = signed A1 x unsigned B0
;
; Clears the 32-bit Y register
; Y1:Y0 = unsigned A0 x signed B1 + Y1:Y0
;
; Copy next 16 bits of result to D1
;
; C2:C1:C0 now contain upper result
5.5.3.4 Other Applications of Multi-Precision Integer Multiplication
In addition to the examples in Section 5.5.3.1, “Signed 32-Bit × Signed 32-Bit with 32-Bit Result,”
through Section 5.5.3.3, “Signed 32-Bit × Signed 32-Bit with 64-Bit Result,” the multi-precision integer
multiplication instructions can be applied in other cases, such as the case of a signed 32-bit times an
unsigned 32-bit. The case of a signed 16-bit times a signed 32-bit with a 32-bit result is shown in
Example 5-26 on page 5-36.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-35
Data Arithmetic Logic Unit
Example 5-26. Multiplying Signed 16-Bit Word with Signed 32-Bit Long
C1:C0 = A1 × B1:B0
(Both Operands Are Signed)
;Signed 16-Bit x Signed 32-Bit Integer Multiplication
IMPYSU A1,B0,Y
; Y1:Y0 = signed A1 x unsigned B0
TFR
Y,C
IMPY.L A,B,Y
; Y1:Y0 = signed A1 x signed B1
ADD
Y0,C
; Combine Results: final 32-bit result in C
5.6 Normalizing
For many algorithms, maximum precision in calculations is required to ensure proper results. For example,
when very small fractional values are worked with, there may not be enough binary digits in an
accumulator to accurately reflect a value. The normalizing capabilities provided by the DSP56800E
architecture can help correct this problem.
Normalizing involves scaling a value to a known magnitude. On the DSP56800E, a normalized value is
one that has no significant digits to the left of the binary point. Thus, in an accumulator register, a
normalized value has 1 sign bit and 31 significant digits. A value can be normalized, the original
magnitude can be saved, calculations can be performed, and the result can be scaled back to its original
magnitude.
5.6.1 Normalized Values
On the DSP56800E architecture, a value is considered normalized if there are no significant digits to the
left of the binary point. Bits to the left of the binary point should contain only the sign and sign extension.
Figure 5-25 shows both non-normalized and normalized values in an accumulator.
Before Normalization
A2
A
F
After Normalization
A1
F
35 32 31
F
A0
E
4
6
16 15
C
A2
3
1
A
0
F
A1
9
1
A0
B
35 32 31
0
C
16 15
4
0
0
0
Figure 5-25. Normalizing a Small Negative Value
The first value in Figure 5-25 is not normalized: the first significant bit in the value is bit 21, and all bits to
the left are merely the sign and sign extension. The second value in Figure 5-25 shows the same value
normalized. The value has been left shifted 10 bits, eliminating the sign-extension bits and placing the sign
in bit 31 and the most significant bit in bit 30.
Figure 5-26 on page 5-37 shows a second value before and after normalization. In this example, the value
has been right shifted 3 bits to place the most significant bit to the right of the binary point.
5-36
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Normalizing
Before Normalization
A2
A
2
After Normalization
A1
7
A0
C
C
35 32 31
3
4
0
A2
0
0
16 15
A
0
0
A1
4
35 32 31
F
A0
9
8
6
16 15
8
0
0
0
Figure 5-26. Normalizing a Large Positive Value
In both Figure 5-25 on page 5-36 and Figure 5-26, the normalized values are aligned so that the most
significant bit is placed in bit 30. On the DSP56800E architecture, this alignment ensures that positive
values p lie in the range 0.5 < p < 1.0 and that negative values n lie in the range –1.0 < n < –0.5. The
amount by which the values were shifted can be used to scale the normalized values back to their original
magnitudes.
5.6.2 Normalizing Methods
There are two methods for normalizing a value in an accumulator. One, using the NORM instruction, is
more flexible but slow. The other method executes much more quickly, but is limited in the values it can
normalize.
The NORM instruction can be used to normalize a full 36-bit accumulator. Each time NORM is executed,
the accumulator to be normalized is shifted 1 bit right or left, as necessary, and a second register is
incremented. NORM is executed repeatedly until the accumulator value is fully normalized. Example 5-27
shows the general method.
Example 5-27. Normalizing with the NORM Instruction
TST
REP
NORM
A
#31
R0,A
; establish condition codes for NORM
; do 31 normalization steps
; execute a normalization step
The NORM instruction uses the E, U, and Z bits in the status register to determine how a value should be
shifted, so a TST instruction on the accumulator that is to be normalized must be executed before NORM
to ensure that the condition codes are set properly. At the end of the sequence in Example 5-27, the A
accumulator is normalized, and the R0 register holds the number of shifts required to normalize A.
Unfortunately, it is not possible to determine in advance how many shifts will be required to normalize a
value. Because up to 31 shifts might be required, NORM must be executed 31 times to ensure that the
value is fully normalized. In Example 5-27, a REP instruction is used to execute NORM for the proper
number of times. Although it wastes time to execute NORM more times than is necessary, NORM has no
effect on already normalized values, so there are no adverse side effects.
There is a second method for normalizing an accumulator that is less flexible but much faster. The CLB
instruction is used to determine the number of leading zeros or ones in a value, and a simple shift
instruction normalizes the accumulator. Example 5-28 shows this method.
Example 5-28. Normalizing with a Shift Instruction
CLB
A,X0
ASLL.L X0,A
; place # of leading bits - 1 into X
; shift A left to normalize
This method is clearly more efficient, requiring only two instructions (the NORM technique requires 33
instructions to be executed). However, the CLB instruction only counts leading bits in the 32-bit MSP:LSP
portion of the accumulator. Because the extension portion of the accumulator is ignored by CLB, fractional
Freescale Semiconductor
Data Arithmetic Logic Unit
5-37
Data Arithmetic Logic Unit
values that are larger than one cannot be normalized. For most applications, this limitation should not be a
problem. However, if it is necessary to consider the extension register when a value is normalized, the
NORM technique must be used.
Regardless of the method that is used to normalize an accumulator, the second register (R0 and X0 in
Example 5-27 and Example 5-28 on page 5-37, respectively) holds the amount by which the accumulator
was scaled. This value can be used later to scale the normalized accumulator back to its original
magnitude.
5.7 Condition Code Calculation
The results of calculations are reflected in the condition code flag bits. To understand how the value of the
condition code bits is calculated after an operation, consider a number of factors:
•
The size of the operands, as specified by the instruction
•
The operation’s destination: accumulator, 16-bit register, or memory location
•
Whether the instruction operates on the whole accumulator or only on a portion
•
The current condition code mode
•
Whether or not the MAC output limiter is enabled
This section discusses how the condition code mode and data sizes affect the condition codes. A detailed
discussion of condition code calculation appears in Appendix B, “Condition Code Calculation.”
5.7.1 Condition Code Modes
In earlier generations of the DSP56800E architecture, two condition code modes were available: 36-bit
mode, where the extension portion of the accumulator was considered when condition codes were
calculated, and 32-bit mode, where the the extension registers were ignored. Setting the CM bit in the
operating mode register (OMR) meant that 32-bit mode was selected. This mode was useful for integer and
control code because the extension registers are not typically used in those algorithms.
Although both condition code modes are supported on the DSP56800E, using 32-bit mode is not generally
recommended, nor is it necessary. The DSP56800E instruction set supports test and compare instructions
for byte, word, long, and 36-bit values, so the exact data size can be specified at all times depending on the
needs of the program. Thirty-two-bit condition code mode should only be used when exact compatibility
with existing DSP56800 program code is required.
5.7.2 Condition Codes and Data Sizes
The DSP56800E properly calculates condition codes for all supported data types. The calculation depends
on the size and type of the data that is being manipulated. Consider the compare instruction, for example.
The DSP56800E instruction set supports four different versions of the compare instruction:
•
CMP.B and CMP.BP—compare two byte values
•
CMP.W—compare two word values
•
CMP.L—compare the lowest 32 bits of an accumulator with the lowest 32 bits of a second
accumulator or with a 16-bit source
•
CMP—compare an entire 36-bit accumulator with a second 36-bit accumulator or with a 16-bit
source
5-38
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Saturation and Data Limiting
In the CMP.B, CMP.BP, and CMP.W instructions, condition codes are based on 8- or 16-bit results, with
corresponding 8- or 16-bit source operands. The CMP.L and CMP instructions generate condition codes on
32- and 36-bit results, respectively, but one of the two operands can be a 16-bit word. In each case,
condition codes are calculated based on the size that is specified in or implied by the instruction opcode.
5.8 Saturation and Data Limiting
DSC algorithms can generate values that are larger than the data precision of the machine when real data
streams are processed. Normally a processor simply overflows its result when this generation occurs, but
overflow creates problems for processing real-time signals. The solution is saturation, or data limiting,
which guarantees that values are always within a given range.
Saturation is especially important when data is run through a digital filter whose output goes to a
digital-to-analog converter (DAC), since saturation “clips” the output data instead of allowing arithmetic
overflow. Without saturation, the output data could incorrectly switch from a large positive number to a
large negative value, which would almost certainly cause unwanted results.
As an alternative to overflow, the DSP56800E provides optional saturation of results through two limiters
that are within the data ALU. The data limiter saturates values when moving data out of an accumulator
with a move instruction or parallel move. The MAC output limiter limits the output of the data ALU’s
MAC unit.
5.8.1 Data Limiter
The data limiter protects against overflow by selectively limiting when an accumulator register is read as a
source operand in a move instruction. Test logic in the extension portion of each accumulator register
detects overflows so that the limiter can substitute one of two constants to minimize errors that are due to
overflow. This process is called “saturation arithmetic.” When limiting occurs, a flag is set and latched in
the status register. The value of the accumulator is not changed.
When a MOVE.W instruction specifies an accumulator (FF) as a source, and when the contents of the
selected source accumulator can be represented in the destination operand size without overflow (that is,
the accumulator extension register is not in use), the data limiter does not saturate and the register contents
are stored unmodified. If a MOVE.W instruction is used and the contents of the selected source
accumulator cannot be represented in the destination operand size without overflow, the data limiter places
a “limited” data value in the destination that has maximum magnitude and the same sign as the source
accumulator. Table 5-3 summarizes these scenarios. The value in the accumulator is not changed.
Table 5-3. Data Limiter Saturation
Extension Bits in Use in Selected
Accumulator?
MSB of FF2
No
(Don’t care)
Same as input—unmodified MSP
Yes
0
$7FFF—maximum positive value
Yes
1
$8000—maximum negative value
Output of Limiter onto the CDBW Bus
Although the following examples all involve fractional data and arithmetic, saturation is equally applicable
to integer arithmetic.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-39
Data Arithmetic Logic Unit
Figure 5-27 graphically demonstrates the advantages of saturation arithmetic. In this example, the A
accumulator contains the following 36-bit value to be read to a 16-bit destination:
0000 1.000 0000 0000 0000 0000 0000 0000 0000 (in binary)
(+1.0 in fractional decimal, $0 8000 0000 in hexadecimal)
If this accumulator is read with a MOVE.W A1,X0 instruction, which disables limiting, the 16-bit X0
register contains the following value after the move instruction, assuming signed fractional arithmetic:
1.000 0000 0000 0000 (–1.0 fractional decimal, $8000 in hexadecimal)
This result is clearly in error because the value –1.0 in the X0 register greatly differs from the value of +1.0
in the source accumulator. In this case, overflow has occurred. To minimize the error due to overflow, it is
preferable to write the maximum (“limited”) value that the destination can assume. In this example, the
limited value would be:
0.111 1111 1111 1111 (+ 0.999969 fractional decimal, $7FFF in hexadecimal)
This value is clearly closer than –1.0 is to the original value, +1.0, and thus introduces less error.
Without Limiting—MOVE.W
A1,X0
With Limiting—MOVE.W
A,X0
35
0
0 . . . 0 1 0 0 . . . . . . . . . . 0 0 0 0 . . . . . . . . . . . 0 0 A = +1.0
35
0
0 . . . 0 1 0 0 . . . . . . . . . . 0 0 0 0 . . . . . . . . . . . 0 0 A = +1.0
3
3
0 15
0 15
100..........00
15
0
0 15
X0 = –1.0
011..........11
IERRORI = 2.0
0
0 15
15
0
0
X0 = +0.999969
IERRORI = .000031
Limiting automatically occurs when the 36-bit operands A, B, C, or D are read with a MOVE.W instruction. Note that
the contents of the original accumulator are NOT changed.
Figure 5-27. Example of Saturation Arithmetic
Example 5-29 is a simple illustration of positive saturation.
Example 5-29. Demonstrating the Data Limiter—Positive Saturation
5-40
MOVE.W #$7FFC,A
; Initialize A = $0:7FFC:0000
INC.W A
MOVE.W A,X:(R0)+
INC.W A
MOVE.W A,X:(R0)+
INC.W A
MOVE.W A,X:(R0)+
;
;
;
;
;
;
A = $0:7FFD:0000
Write $7FFD to memory (limiter enabled)
A = $0:7FFE:0000
Write $7FFE to memory (limiter enabled)
A = $0:7FFF:0000
Write $7FFF to memory (limiter enabled)
INC.W A
MOVE.W A,X:(R0)+
INC.W A
MOVE.W A,X:(R0)+
INC.W A
MOVE.W A,X:(R0)+
;
;
;
;
;
;
A = $0:8000:0000 <===
Write $7FFF to memory
A = $0:8001:0000
Write $7FFF to memory
A = $0:8002:0000
Write $7FFF to memory
MOVE.W A1,X:(R0)+
; Write $8002 to memory (limiter disabled)
Overflows 16 bits!
(limiter saturates)
(limiter saturates)
(limiter saturates)
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Saturation and Data Limiting
Once the accumulator increments to $8000 in Example 5-29, the positive result can no longer be written to
a 16-bit memory location without overflow. So, instead of writing an overflowed value to memory, the
data limiter writes $7FFF, the maximum positive value that can be represented by a signed, 16-bit word.
Note that the data limiter affects only the value written to memory; it does not affect the accumulator. In
the final instruction of the example, the limiter is disabled because the register is specified as A1.
Example 5-30 is a simple illustration of negative saturation.
Example 5-30. Demonstrating the Data Limiter—Negative Saturation
MOVE.W #$8003,A
; Initialize A = $F:8003:0000
DEC.W A
MOVE.W A,X:(R0)+
DEC.W A
MOVE.W A,X:(R0)+
DEC.W A
MOVE.W A,X:(R0)+
;
;
;
;
;
;
A = $F:8002:0000
Write $8002 to memory (limiter enabled)
A = $F:8001:0000
Write $8001 to memory (limiter enabled)
A = $F:8000:0000
Write $8000 to memory (limiter enabled)
DEC.W A
MOVE.W A,X:(R0)+
DEC.W A
MOVE.W A,X:(R0)+
DEC.W A
MOVE.W A,X:(R0)+
;
;
;
;
;
;
A = $F:7FFF:0000 <===
Write $8000 to memory
A = $F:7FFE:0000
Write $8000 to memory
A = $F:7FFD:0000
Write $8000 to memory
MOVE.W A1,X:(R0)+
; Write $7FFD to memory (limiter disabled)
Overflows 16 bits!
(limiter saturates)
(limiter saturates)
(limiter saturates)
Once the accumulator decrements to $7FFF in Example 5-30, the negative result can no longer fit into a
16-bit memory location without overflow. So, instead of writing an overflowed value to memory, the data
limiter writes the most negative 16-bit number, $8000. Limiting is bypassed when individual portions of
the accumulator, rather than the entire accumulator, are read (as in the last line of the example).
5.8.2 MAC Output Limiter
The MAC output limiter optionally saturates or limits results that are calculated by data ALU arithmetic
operations such as multiplication, addition, incrementing, rounding, and so on.
The MAC output limiter can be enabled by setting the SA bit in the operating mode register (see
Section 8.2.1.3, “Saturation (SA)—Bit 4,” on page 8-6). It is also used when the SAT instruction is
executed, which saturates the value of the source accumulator and stores the result in a data ALU register.
NOTE:
When the SA bit in the OMR is modified, a delay of 2 instruction cycles is
necessary before the new saturation mode becomes active.
Consider the simple example in Example 5-31 on page 5-42.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-41
Data Arithmetic Logic Unit
Example 5-31. Demonstrating the MAC Output Limiter
BFSET #$0010,OMR
MOVE.W #$7FFC,A
NOP
; Set SA bit-—enables MAC Output Limiter
; Initialize A = $0:7FFC:0000
INC.W A
INC.W A
INC.W A
; A = $0:7FFD:0000
; A = $0:7FFE:0000
; A = $0:7FFF:0000
INC.W A
INC.W A
ADD.W #9,A
; A = $0:7FFF:FFFF <=== Saturates to 16 bits!
; A = $0:7FFF:FFFF <=== Saturates to 16 bits!
; A = $0:7FFF:FFFF <=== Saturates to 16 bits!
Once the accumulator increments to $7FFF in Example 5-31, the saturation logic in the MAC output
limiter prevents it from growing larger because it can no longer fit into a 16-bit memory location without
overflow. So, an overflowed value is not written to back to the A accumulator; the value of the most
positive 32-bit number, $7FFF:FFFF, is written instead.
The saturation logic operates by checking 3 bits of the 36-bit result out of the MAC unit—EXT[3],
EXT[0], and MSP[15]. As shown in Table 5-4, when the SA bit is set, these 3 bits determine whether
saturation is performed on the MAC unit’s output and whether to saturate to the maximum positive value
($7FFF:FFFF) or to the maximum negative value ($8000:0000).
Table 5-4. MAC Unit Outputs with Saturation Enabled
EXT[3]
EXT[0]
MSP[15]
Result Stored in Accumulator
0
0
0
Result as calculated, with no limiting
0
0
1
$0:7FFF:FFFF
0
1
0
$0:7FFF:FFFF
0
1
1
$0:7FFF:FFFF
1
0
0
$F:8000:0000
1
0
1
$F:8000:0000
1
1
0
$F:8000:0000
1
1
1
Result as calculated, with no limiting
The MAC output limiter affects not only the results calculated by the instruction, but condition code
computation as well. See Section B.1.2, “MAC Output Limiter,” on page B-3 for more information.
5.8.3 Instructions Not Affected by the MAC Output Limiter
The MAC output limiter is always disabled (even if the SA bit is set) when the following instructions are
executed:
5-42
•
ASLL.W, ASRR.W, LSRR.W
•
ASLL.L, ASRR.L, LSRR.L
•
ASL16, ASR16, LSR16, ASRAC,
LSRAC
•
IMPYSU, IMACUS, IMPYUU,
IMACUU
•
IMAC.L, IMPY.L, IMPY.W
•
MPYSU, MACSU
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Rounding
•
AND.W, OR.W, EOR.W
•
AND.L, OR.L, EOR.L
•
LSL.W, LSR.W, ROL.W, ROR.W,
ROL.L, ROR.L
•
SXT.B, ZXT.B, SXT.L
•
NOT.W, CLB, SUBL
•
ADC, DIV, SBC
•
ADD.B, ADD.BP, SUB.B, SUB.BP
•
DEC.BP, INC.BP, NEG.BP
•
TST, TST.B, TST.BP, TST.W, TST.L
•
CMP.B, CMP.BP, CMP.L
The CMP.W instruction is not affected by the MAC output limiter except when the first operand is not a
register (that is, it is a memory location or an immediate value) and the second operand is X0, Y0, or Y1.
In this particular case, the calculation of the U bit might be affected if saturation occurs. No other condition
code bits are affected.
Note also that if the MAC output limiter is enabled, saturation may occur when a value is transferred from
one accumulator to another with the TFR instruction. To move a 32-bit value from one accumulator to
another without limiting when the MAC output limiter is enabled, use the SXT.L instruction.
The MAC output limiter only affects operations performed in the data ALU. It has no effect on instructions
executed in other functional blocks, such as the AGU or program controller.
5.9 Rounding
The DSP56800E architecture provides three instructions that can perform rounding—RND, MACR, and
MPYR. The RND instruction simply rounds a value in the accumulator register that is specified by the
instruction, whereas the MPYR or MACR instructions perform a regular MPY or MAC operation and then
round the result. Each rounding instruction rounds the result to a single-precision value so that the value
can be stored in memory or in a 16-bit register. (Note that saturation can still occur when a rounded result
is moved to a 16-bit destination). In addition, for instructions where the destination is one of the four
accumulators, the FF0 portion of the destination accumulator (A0, B0, C0, or D0) is cleared.
The DSC core implements two types of rounding: convergent rounding and two’s-complement rounding.
In the DSP56800E, the rounding point is between bits 16 and 15 of a 36-bit value. In the A accumulator,
this point is between the A1 register’s LSB and the A0 register’s MSB. The usual rounding method rounds
up any value above one-half (that is, LSP > $8000) and rounds down any value below one-half (that is,
LSP < $8000).
The question arises as to which way the number one-half (LSP equals $8000) should be rounded. If it is
always rounded one way, the results are eventually biased in that direction. Convergent rounding solves
the problem of this boundary case by rounding down if the number is even (bit 16 equals zero) and
rounding up if the number is odd (bit 16 equals one). In contrast, two’s-complement rounding always
rounds this number up. The type of rounding is selected by the rounding bit (R) of the OMR.
NOTE:
When the rounding bit is modified, there is a delay of 2 instruction cycles
before the new rounding mode becomes active.
Freescale Semiconductor
Data Arithmetic Logic Unit
5-43
Data Arithmetic Logic Unit
5.9.1 Convergent Rounding
Convergent rounding, also called “round to the nearest even number,” is the default rounding mode. For
most values, this mode and two’s-complement rounding round identically. They only differ when the least
significant 16 bits of the final result before rounding are exactly $8000. In this case, convergent rounding
rounds down the value if the number is even (bit 16 equals zero) and rounds up the value if it is odd (bit 16
equals one).
The algorithm for convergent rounding is as follows:
1. Add the value $0:0000:8000 to the accumulator (for the RND instruction) or to the final
result without rounding (for the MACR instruction).
2. If the 16 LSBs of the result at this point are $0000, then clear bit 16 of the result.
3. If the SA bit in the OMR is set and the accumulator extension is in use:
— Saturate to $0:7FFF:0000 if positive.
— Saturate to $F:8000:0000 if negative.
4. Clear the LSP of the result before writing to a destination accumulator.
Figure 5-28 on page 5-45 shows the four possible cases for convergent rounding a number in one of the
four accumulators.
5-44
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Rounding
Case I: If A0 < $8000 (1/2), then round down (add nothing)
Before Rounding
After Rounding
0
A2
A1
A0
XX..XX XXX...XXX0100 011XXX....XXX
35
32 31
16 15
0
A2
A1
A0*
XX..XX XXX...XXX0100 000.........000
35
32 31
16 15
0
Case II: If A0 > $8000 (1/2), then round up (add 1 to A1)
Before Rounding
After Rounding
1
A2
A1
A0
XX..XX XXX...XXX0100 1110XX....XXX
35
32 31
16 15
0
A2
A1
A0*
XX..XX XXX...XXX0101 000.........000
35
32 31
16 15
0
Case III: If A0 = $8000 (1/2), and the LSB of A1 = 0 (even), then round down (add nothing)
Before Rounding
After Rounding
0
A2
A1
A0
XX..XX XXX...XXX0100 1000........000
35
32 31
16 15
0
A2
A1
A0*
XX..XX XXX...XXX0100 000.........000
35
32 31
16 15
0
Case IV: If A0 = $8000 (1/2), and the LSB = 1 (odd), then round up (add 1 to A1)
Before Rounding
After Rounding
1
A2
A1
A0
XX..XX XXX...XXX0101 1000........000
35
32 31
16 15
0
A2
A1
A0*
XX..XX XXX...XXX0110 000.........000
35
32 31
16 15
0
*A0 is always clear; performed during RND, MPYR, MACR
Figure 5-28. Convergent Rounding
Freescale Semiconductor
Data Arithmetic Logic Unit
5-45
Data Arithmetic Logic Unit
5.9.2 Two’s-Complement Rounding
When this type of rounding is selected through setting the rounding bit in the OMR, then, during a
rounding operation, one is added to the bit to the right of the rounding point (bit 15 of A0) before the bit
truncation. Figure 5-29 shows the two possible cases.
Case I: A0 < 0.5 ($8000), then round down (add nothing)
Before Rounding
After Rounding
0
A2
A1
A0
A2
X X . . X X X X X . . . X X X 0 1 0 0 0 1 1 0 X. . . . . . . X X X
35
32 31
16 15
0
A1
A0*
XX..XX XXX...XXX0100 000.........000
35
32 31
16 15
0
Case II: A0 >= 0.5 ($8000), then round up (add 1 to A1)
Before Rounding
After Rounding
1
A2
A1
A0
A2
XX..XX XXX...XXX0101 1110X......XXX
35
32 31
16 15
0
A1
A0*
XX..XX XXX...XXX0101 000.........000
35
32 31
16 15
0
*A0 is always clear; performed during RND, MPYR, MACR
AA0050
Figure 5-29. Two’s-Complement Rounding
The algorithm for two’s-complement rounding is as follows:
1. Add the value $0:0000:8000 to the accumulator (for the RND instruction) or to the final
result without rounding (for the MACR instruction).
2. If the SA bit in the OMR is set and the extension is in use:
— Saturate to $0:7FFF:0000 if positive.
— Saturate to $F:8000:0000 if negative.
3. Clear the LSP of the result before writing to a destination accumulator.
5.9.3 Rounding Examples
Example 5-32 shows program code that demonstrates two’s-complement rounding, and Example 5-33
demonstrates convergent rounding.
Example 5-32. Example Code for Two’s-Complement Rounding
MOVE.L #VALUE,A
BFSET #$0020,OMR
NOP
NOP
RND
A
5-46
;
;
;
;
;
Load A Accumulator
Set the R bit for two’s-complement rounding
(2 cycles required for R bit to be valid)
(2 cycles required for R bit to be valid)
Round A accumulator
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Rounding
Example 5-33. Example Code for Convergent Rounding
MOVE.L #VALUE,A
BFCLR #$0020,OMR
NOP
NOP
RND
A
;
;
;
;
;
Load A Accumulator
Clear the R bit for convergent rounding
(2 cycles required for R bit to be valid)
(2 cycles required for R bit to be valid)
Round A accumulator
Table 5-5 shows four sets of results when four different values are substituted for the placeholder
“#VALUE” in Example 5-32 and Example 5-33 on page 5-47. The two algorithms give different results in
one of the four cases.
Table 5-5. Rounding Results for Different Values
Value to be
Rounded
Convergent
Rounding
Result
Two’s-Complement
Rounding Result
$1234:0397
$1234:0000
$1234:0000
Simple case: both round down to same
value.
$1234:C397
$1235:0000
$1235:0000
Simple case: both round up to same value.
$1234:8000
$1234:0000
$1235:0000
Boundary case: LSP of value is $8000 and
MSP is even. In this case, the algorithms
generate different results!
$1235:8000
$1236:0000
$1236:0000
Boundary case: LSP of value is $8000 and
MSP is odd. In this case, both have the
same result.
Freescale Semiconductor
Comments
Data Arithmetic Logic Unit
5-47
Data Arithmetic Logic Unit
5-48
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 6
Address Generation Unit
The address generation unit (AGU) performs all address calculation and generation for the DSP56800E
core. The AGU calculates effective addresses for instruction operands and directly executes the address
arithmetic instructions.
Support is built into the AGU for applications that require both 24- and 16-bit pointers. Byte, word, and
long-word data memory accesses are also available for use by applications. Extensive pointer arithmetic
operations are provided for even greater flexibility.
6.1 AGU Architecture
The address generation unit (AGU) consists of the registers and logic used to calculate the effective
address of data operands in memory. It supports both linear and modulo arithmetic calculations. All AGU
operations are performed in parallel with other chip functions to minimize address-generation overhead.
The major components of the address generation unit are:
•
A 24-bit primary address arithmetic unit.
•
A 24-bit secondary address adder unit.
•
Two single-bit shifters for byte addressing.
The AGU contains two arithmetic units—a primary address arithmetic unit for complex address
calculations, and a secondary address adder for simple calculations. The primary address arithmetic unit
supports both linear and modulo address arithmetic, simplifying the implementation of some useful data
structures.
The two arithmetic units can update up to two 24-bit addresses every instruction cycle: one for primary
memory accesses using XAB1 or PAB, and one for secondary memory accesses performed on XAB2.
AGU operations are performed on internal AGU buses, so bus transfers occur in parallel with AGU
calculations.
Figure 6-1 on page 6-2 presents a block diagram of the AGU on the DSP56800E core. The DSP56800EX
core contains additional shadow registers not reflected in this diagram.
Freescale Semiconductor
Address Generation Unit
6-1
Address Generation Unit
CDBR
CDBW
15
Modifier
Registers
0
Pointer Registers
23
0
15
R0
M01
Secondary
Offset
Register
0
N3
R1
To
R3
R2
Primary
Arithmetic
Unit
R3
Secondary
Adder
R4
R5
N
SP
Short or Long
Immediate Data
R3 Only
pass, <<1
pass, >>1
Byte Select
PAB
XAB1
XAB2
Figure 6-1. Address Generation Unit Block Diagram (DSP56800E Core)
Figure 6-2 illustrates a dual parallel read instruction, which uses 1 program word and executes in
1 instruction cycle. The primary operand is addressed with XAB1, and the second operand is addressed
with XAB2. The data memory, in turn, places its data on the core data bus for reads (CDBR) and on the
second data bus (XDB2), respectively. See Section 3.3.5, “Parallel Moves,” on page 3-11 for more
discussion of parallel memory moves.
MOVE.W
X:(R4)+N,Y0
X:(R3)+N3,X0
Primary Read
(Uses XAB1 and CDBR)
Secondary Read
(Uses XAB2 and XDB2)
Figure 6-2. Dual Parallel Read Instruction
The AGU can directly address 224 (16,777,217) locations in data memory and 221 (2,097,152) locations in
program memory. All three buses can generate addresses to on-chip or off-chip memory.
6.1.1 Primary Address Arithmetic Unit
The primary address arithmetic unit is used when AGU arithmetic instructions are performed and when
complex operand effective addresses are calculated, as in indexing and post-updating. Byte, word, and
long-word accesses are supported.
6-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
AGU Programming Model
Calculations in the primary address arithmetic unit can be performed using either linear arithmetic, for
general-purpose computing, or modulo arithmetic, for circular buffers and other useful data structures. The
contents of the modifier register, M01, specify the type of arithmetic to be performed for the R0 and R1
address registers. All other address registers—R2–R5, N, and SP—always operate with linear arithmetic.
Modulo arithmetic is described in detail in Section 6.8, “Linear and Modulo Address Arithmetic.”
6.1.2 Secondary Address Adder Unit
The secondary address adder unit is used for address update calculations on the R3 register, which is used
for the secondary read in dual memory read instructions (see Figure 6-2 on page 6-2). The adder unit can
increment, decrement, or add the contents of the N3 register to R3. This unit performs only linear
arithmetic; modulo arithmetic is not supported.
6.1.3 Single-Bit Shifting Units
Two single-bit shifters are present to support byte addressing. More information on byte addressing, and
on the shift operations that are performed on byte addresses, can be found in Section 3.5, “Memory Access
and Pointers,” on page 3-17.
6.2 AGU Programming Model
The AGU programming model, which Figure 6-3 on page 6-4 illustrates, consists of 14 programmable
registers:
•
Six 24-bit address registers (R0–R5)
•
A 24-bit stack pointer register (SP)
•
A 24-bit offset register (N, which may also be used as an address register)
•
A 16-bit offset register (N3)
•
A 16-bit modifier register (M01)
•
Four shadow registers (shadows of R0, R1, N, and M01) on the DSP56800E and DSP56800EX
cores, and five additional shadow registers (shadows of R2, R3, R4, R5, and N3) on the
DSP56800EX core
The eight 24-bit registers can be used as pointers in the register-indirect addressing modes. The N register
can also be used as an index or offset by the six address pointer registers. Modulo arithmetic on the R0 and
R1 pointer registers is enabled with the M01 register. The shadowed registers provide extra pointer
registers for interrupt routines or for system-control software.
Although all of the address pointer registers and the SP are available for most addressing modes, there are
some addressing modes that only work with a specific address pointer register. These special cases appear
in Table 6-1 on page 6-6.
Freescale Semiconductor
Address Generation Unit
6-3
Address Generation Unit
23
Pointer Registers
Secondary Offset Register
15
0
0
R0
N3
R1
Modifier Registers
R2
15
R3
0
M01
R4
R5
N
SP
Figure 6-3. Address Generation Unit Programming Model
NOTE:
Pipeline dependencies might be encountered when the AGU registers are
modified. Refer to Section 10.4.2, “AGU Pipeline Dependencies,” on
page 10-28 for more information.
6.2.1 Address Registers (R0–R5, N)
The address register file consists of six 24-bit registers, R0–R5, which are typically used as pointers to
memory. The offset register, N, can also be used as an address register. The address registers can directly
drive the core’s three address buses, minimizing access time to internal and external data and program
memory.
The address registers can be used to access byte, word, and long values in data memory, and they can be
used as byte or word pointers (see Section 3.5.1, “Word and Byte Pointers,” on page 3-17). Any address
register can be used for accessing either on-chip or off-chip data memory, including the R3 register when it
is used in the secondary access of a dual read instruction. Only the R0–R3 registers can be used to access
on-chip or off-chip program memory.
6.2.2 Stack Pointer Register (SP)
The stack pointer register (SP) is a 24-bit register that is used to access the software stack. The stack
pointer register can be used to access byte, word, and long values in data memory. It is always used as a
word pointer (see Section 3.5.1, “Word and Byte Pointers,” on page 3-17).
The SP register can be used by a program to access data on the software stack, or it can be used implicitly
by instructions that store information on the stack as part of their regular operation. These instructions
include jumps to subroutines and interrupt handlers, which push the current program counter and status
register on the stack.
This register is not initialized to a known value after reset. Applications need to explicitly establish the
base of the stack after reset, taking care that the stack area does not overlap any other data area. Note that
the software stack grows upward when values are pushed onto it.
6-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
AGU Programming Model
6.2.3 Offset Register (N)
The N register is one of the most powerful registers in the AGU. In addition to functioning as an address
pointer similar to the R0–R5 registers, it can also be used for indexed and post-update addressing modes.
When the N register is used as an offset for post-updating, its value is truncated to 16 bits and then sign
extended to 24 bits before being passed to the primary arithmetic unit for post-updating. When the N
register is used as an offset for accessing long memory locations, its value is shifted to the left by 1 bit
before it is passed to the primary arithmetic unit for calculating the effective address. Thus, in this case, the
N offset is a long offset.
6.2.4 Secondary Read Offset Register (N3)
The secondary read offset register (N3) is a 16-bit register that is used for post-updating the R3 pointer
register in dual read instructions, which read two values from data memory. The N3 register is sign
extended to 24 bits and passed to the secondary address adder unit for post-updating the R3 pointer
register.
6.2.5 Modifier Register (M01)
The modifier register (M01) specifies whether linear or modulo arithmetic is used when a new address is
calculated. This modifier register is automatically read when the R0 or R1 address register is used in an
address calculation. This register has no effect on address calculations done with the R2–R5, N, or SP
registers.
During processor reset this register is set to $FFFF, which enables linear arithmetic for the R0 and R1
registers. Programming the modifier register is discussed in Section 6.8.3, “Configuring Modulo
Arithmetic.”
NOTE:
The M01 register should never be used for general-purpose storage
because its value affects calculations with the R0 and R1 pointers.
6.2.6 Shadow Registers
The DSP56800E provides four shadow registers corresponding to the R0, R1, N, and M01 address
registers. The DSP56800EX core provides the same four registers as well as five additional shadow
registers corresponding to the R2, R3, R4, R5, and N3 address registers.
The shadow registers are not directly accessible, but become available when their contents are swapped
with the contents of the corresponding AGU core registers. This swapping is accomplished through
executing the SWAP SHADOWS instruction. The contents of the four registers are exchanged with their
shadowed counterparts. When the original values of the registers are required, executing the
SWAP SHADOWS instruction a second time restores the original values.
NOTE:
The shadow register corresponding to M01 is not initialized by the core at
reset. It must be explicitly programmed by the user.
Freescale Semiconductor
Address Generation Unit
6-5
Address Generation Unit
Using shadow registers as dedicated address registers during fast interrupt processing can greatly reduce
the considerable overhead incurred by saving and restoring registers when exception handlers are entered
and exited. Fast interrupts are described in Section 9.3.2.2, “Fast Interrupt Processing,” on page 9-6. The
SWAP instruction enables the shadow registers to be used to minimize the overhead during normal
interrupt processing.
6.3 Using Address Registers
The DSP56800E AGU provides several address registers that can be used as pointers for accessing
memory. Not all of the registers work identically, however. Depending on the register, there are additional
capabilities or restrictions of use. For example, the R3 register is the only register that is available for the
secondary read in instructions that perform two data memory moves. Table 6-1 summarizes the
capabilities of each address register.
The type of address arithmetic to be performed, linear or modulo, is not encoded in the instruction, but is
specified by the address modifier register (M01). See Section 6.8, “Linear and Modulo Address
Arithmetic,” for a discussion of the arithmetic types. Table 6-1 indicates whether or not modulo arithmetic
is supported for a given register.
Table 6-1. Capabilities of the Address Pointer Registers
Pointer
Register
R0
R1
R2
6-6
Addressing
Modes
Allowed
Modulo
Allowed?
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
Yes
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
Yes
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
No
Capabilities and Notes
Counter for the NORM instruction.
Pointer for single parallel move and for primary access in dual parallel reads.
Pointer for P: memory moves.
Optional source register for Tcc transfer.
Supports legacy addressing modes (Rj+N) and (Rj+xxxx).
Shadowed for use with fast interrupt processing.
Refer to Section 6.8.4, “Base Pointer and Offset Values in Modulo Instructions,” on page 6-26 for interpretation of base pointer and offset in update by
index addressing mode.
Pointer for single parallel move and for primary access in dual parallel reads.
Pointer for P: memory moves.
Optional destination register for Tcc transfer.
Supports legacy addressing modes (Rj+N) and (Rj+xxxx).
Shadowed for use with fast interrupt processing.
Refer to Section 6.8.4, “Base Pointer and Offset Values in Modulo Instructions,” on page 6-26 for interpretation of base pointer and offset in update by
index addressing mode.
Pointer for single parallel move.
Pointer for P: memory moves.
Supports legacy addressing modes (Rj+N) and (Rj+xxxx).
Shadowed for use with fast interrupt processing on the DSP56800EX core.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Byte and Word Addresses
Table 6-1. Capabilities of the Address Pointer Registers (Continued)
Pointer
Register
Addressing
Modes
Allowed
Modulo
Allowed?
Capabilities and Notes
R3
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(R3)+N3
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
No
Pointer for single parallel move and for secondary access in dual parallel
reads.
May be post-updated with N3 register.
Pointer for P: memory moves.
Supports legacy addressing modes (Rj+N) and (Rj+xxxx).
Shadowed for use with fast interrupt processing on the DSP56800EX core.
R4
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
No
Pointer for primary access in dual read instructions.
Shadowed for use with fast interrupt processing on the DSP56800EX core.
R5
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
No
Shadowed for use with fast interrupt processing on the DSP56800EX core.
N
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(RRR+x)
(Rn+xxxx)
(Rn+xxxxxx)
No
Available not only as a pointer register, but also as indexing and post-update
register.
Shadowed for use with fast interrupt processing.
SP
(Rn)
(Rn)+
(Rn)–
(Rn)+N
(Rn+N)
(SP–x)
(SP–xx)
(Rn+xxxx)
(Rn+xxxxxx)
No
Supports 1-word indexed addressing with 6-bit offset for word moves.
Used implicitly by the JSR, RTS, RTSD, RTI, RTID and FRTID instructions.
SP is always used as a word pointer to properly support stack operations.
Supports legacy addressing mode (SP+N).
6.4 Byte and Word Addresses
As discussed in Section 3.5.1, “Word and Byte Pointers,” on page 3-17, the DSP56800E supports two
types of addresses for data memory accesses: word and byte. Depending on the type of address used, the
memory map is interpreted somewhat differently. Figure 6-4 on page 6-8 shows the differences between
the memory maps.
Freescale Semiconductor
Address Generation Unit
6-7
Address Generation Unit
Word
Addresses
Byte
Addresses
X Memory
15
0
X Memory
0 7
7
0
$2003
$88
$77
$4006
$88
$77
$2002
$66
$55
$4004
$66
$55
$2001
$44
$33
$4002
$44
$33
$2000
$22
$11
$4000
$22
$11
Identical Memory
Locations
Figure 6-4. Word vs. Byte Addresses
When word addresses are used, each unique address refers to a different 16-bit word in memory. As shown
in Figure 6-4, locations X:$2000 and X:$2001 refer to adjacent 16-bit words. Byte addresses are used to
locate individual bytes in memory. Addresses X:$4000 and X:$4001 refer to 2 bytes contained in the same
word (the word at X:$2000, using word addressing). Note that data is stored in memory with the least
significant byte occupying the lowest memory location. This is often referred to as “little-endian” byte
ordering.
NOTE:
Byte addresses can not be used for accessing program memory. Program
memory accesses are always performed with word addresses.
Byte and word addresses are distinguished by the instruction that uses them. For most instructions,
including those that explicitly perform a word or long-word access, address register values are interpreted
as word addresses. Address register values are interpreted as byte addresses only when instructions with
the “.BP” extension are used.
6.5 Word Pointer Memory Accesses
Instructions that use address registers as word pointers can access bytes, words, and longs from data
memory. Table 6-2 on page 6-9 shows the word address in data memory that is accessed for the different
addressing modes and data types when word pointers are used. For byte accesses, the LSB of the offset
before the right shift selects the upper or lower byte. For the post-update addressing modes, the address in
Rn is used for the memory access and then is post-updated using the arithmetic shown in Table 6-2.
All immediate offsets and absolute addresses for long-word moves must be even values because long
words must be located on an even word address boundary. When the assembler encounters these
instructions, it divides the absolute address and offset values by two before generating the opcode (no
information is lost, since the low-order bit is guaranteed to be zero). When the instruction is executed, the
AGU left shifts the absolute value 1 bit to generate the correct word address or offset.
NOTE:
The values “xx,” “xxxx,” and “xxxxxx” that appear in Table 6-2 on
page 6-9 for long word accesses are the values that are actually encoded by
the assembler, which have been divided by two during assembly. The table
describes what the hardware does after the instruction has been encoded by
the assembler.
6-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Word Pointer Memory Accesses
Table 6-2. Hardware Implementation of Addressing Mode Arithmetic—
Word Pointers to Data Memory
Address for
Byte Access
Address for
Word Access
Address for
Long Access
No update
X:(Rn)
—
Rn
Rn
Post-increment
X:(Rn)+
—
Rn+1
Rn+2
Post-increment occurs
after access.
Post-decrement
X:(Rn)–
—
Rn–1
Rn–2
Post-decrement occurs
after access.
Post-update by offset N
X:(Rn)+N
—
Rn+N
—
The lower 16 bits of N
are sign extended to 24
bits and added to Rn.
Indexed by offset N
X:(Rn+N)
—
Rn+N
Rn+(N<<1)
Indexed by 3-bit offset
X:(RRR+x)
RRR+(x>>1)
Rn+x
—
Offset x from 0 to 7.
Indexed by 6-bit offset
X:(SP–xx)
—
SP–xx
SP–(xx<<1)
6-bit one extended;
SP pointer only.
Indexed by 3-bit offset
X:(SP–x)
SP–(x>>1)
—
—
3-bit one extended.
Indexed by 16-bit offset
X:(Rn+xxxx)
Rn+(xxxx>>1)
Rn+xxxx
Rn+(xxxx<<1)
Signed 16-bit offset.
Indexed by 24-bit offset
X:(Rn+xxxxxx)
Rn+(xxxxxx>>1)
Rn+xxxxxx
Rn+(xxxxxx<<1)
Signed 24-bit offset.
6-bit absolute short
X:aa
—
0000xx
—
6-bit peripheral short
X:<<pp
—
00FFxx1
—
16-bit absolute address2
X:xxxx
—
00xxxx
(00xxxx<<1)
24-bit absolute address2
X:xxxxxx
—
xxxxxx
(xxxxxx<<1)
Addressing Mode
Comments
1.The upper 18 bits are hard-wired to a specific area of memory, which varies depending on the specific
implementation of the chip.
2.The X:xxxx and X:xxxxxx addressing modes are allowed for byte accesses when they are used as the destination address in a byte memory to memory move instruction. In this case, the source address is specified
with a word pointer, and the destination is an absolute byte address.
Freescale Semiconductor
Address Generation Unit
6-9
Address Generation Unit
6.5.1 Accessing Bytes
Word pointers can be used to access bytes in memory with the MOVE.B and MOVEU.B instructions.
Because word pointers typically select an entire 16-bit word at once, the particular byte to access within
the word is determined by the offset that is specified in the instruction. Even offset values (or an offset of
zero) select the lower byte in a word, while odd offsets select the upper byte.
Example 6-1 demonstrates accessing byte values in memory using the MOVE.B instruction. Note that,
even though word pointers are being used, the offset values are all specified in bytes.
Example 6-1. Accessing Bytes with the MOVE.B Instruction
; Load the R0, SP Address Pointers
MOVEU.W#$2000,R0
;
;
MOVEU.W#$4000,SP
;
;
load R0 pointer with the value $2000
(can be either a byte or word pointer)
load the stack pointer (SP) with $4000
(SP must always be a word pointer)
; MOVE.B -- R0 used as a word pointer, offset is a byte offset
MOVE.B x:(r0+0),x0 ; word address = $2000, selects
MOVE.B x:(r0+1),x0 ; word address = $2000, selects
MOVE.B x:(r0+2),x0 ; word address = $2001, selects
MOVE.B x:(r0+3),x0 ; word address = $2001, selects
MOVE.B x:(r0+4),x0 ; word address = $2002, selects
; MOVE.B -- SP always used as a
MOVE.B x:(sp),x0
MOVE.B x:(sp-1),x0
MOVE.B x:(sp-2),x0
MOVE.B x:(sp-3),x0
MOVE.B x:(sp-4),x0
lower
upper
lower
upper
lower
byte
byte
byte
byte
byte
word pointer, offset is a byte offset
; word address = $4000, selects lower
; word address = $3fff, selects upper
; word address = $3fff, selects lower
; word address = $3ffe, selects upper
; word address = $3ffe, selects lower
byte
byte
byte
byte
byte
6.5.2 Accessing Long Words
Long words are always accessed with word pointers. When a long-word value is read or written to
memory, two adjacent 16-bit word values are accessed: the word specified in the pointer, and the word that
immediately follows in memory. (An exception is when the SP register is used to access long-word values;
see Section 3.5.3, “Accessing Long-Word Values Using Word Pointers,” on page 3-19 for more
information.)
Example 6-2 on page 6-11 demonstrates several long-word accesses. Note the arithmetic performed by the
AGU in calculating the long-word address, specifically the use of the N offset register.
6-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Word Pointer Memory Accesses
Example 6-2. Addressing Mode Examples for Long Memory Accesses
;Initialize Registers
MOVEU.W#$1000,R2
TFRA R2,R3
MOVEU.W#4,N
; initialize base address
; make a copy of R2
; initialize register index value
;First Example -- Indexing with Displacement
MOVE.L X:(R2+4),A
; Accesses X:$1005:X:$1004
;Second Example -- Indexing with Offset Register N (N = 4)
MOVE.L X:(R3+N),A
; Accesses X:$1009:X:$1008
;Third Example -- Calculating the New Address (similar to first example)
ADDA N,R2
; Calculated Address = $1004
MOVE.L X:(R2),A
; Accesses X:$1005:X:$1004
;Fourth Example -- Calculating the New Address (similar to second example)
ADDA.L N,R3
; Calculated Address = $1008
MOVE.L X:(R3),A
; Accesses X:$1009:X:$1008
In the second and fourth examples, the N register value is treated as a long-word offset. When the address
is calculated for the memory access, the R2 and R3 registers are offset by 4 long words (8 words), since the
long-word versions of MOVE and ADDA are used. The resulting address in each case is $1008. Where
word offsets are used, in the other two examples, the address is $1004.
6.5.3 Accessing Data Structures
Data structures and unions (such as those used in the C and C++ programming languages) typically contain
a mixture of data types. Because it is not possible to access word or long-word variables with a byte
pointer, word pointers should always be used when structure elements are accessed. Byte values in the
structure can still be accessed with the MOVE.B and MOVEU.B instructions, which use word pointers.
Consider an example structure in data memory. The structure contains byte, word, and long-word variables
and has its base address, a word pointer, stored in R3. Structure elements are accessed with offsets from
this base through using the (R3+x) and (R3+xxxx) addressing modes.
The code in Example 6-3 shows the initialization of a data structure and code used to access the elements.
Each of the four accumulators are loaded with a different structure variable.
Example 6-3. Accessing Elements in a Data Structure
STRUCT1
CODESTART
ORG
DCB
DCB
DCL
DC
x:$7000
$BB,$AA
$DD,$CC
$12345678
$FFFF
ORG
P:
MOVE.L #STRUCT1,R3
MOVE.B x:(R3+1),A
MOVEU.Bx:(R3+2),B
MOVE.W x:(R3+4),C
MOVE.L x:(R3+2),d
Freescale Semiconductor
;
;
;
;
;
Data Structure named “STRUCT1”
four chars: 1st is $AA, 2nd is $BB
3rd is $CC, 4th is $DD
1 long containing $12345678
1 word containing $FFFF
;
;
;
;
;
;
(instructions located in program memory)
set up base to data structure
read with offset of 1 byte from R3
read with offset of 2 bytes from R3
read with offset of 4 words from R3
read with offset of 2 words from R3
Address Generation Unit
6-11
Address Generation Unit
After the code in Example 6-3 on page 6-11 is executed, the accumulators hold the following values:
After Execution
Before Execution
A
X
X
X
A
$F
$FFBB
$0000
B
X
X
X
B
$0
$00CC
$0000
C
X
X
X
C
$F
$FFFF
$0000
D
X
X
X
D
$0
$1234
$5678
Note that the last instruction in Example 6-3, which loads the long-word variable into D, specifies an offset
value of two. This value is specified because constant offsets for both word and long-word memory
accesses are always specified in words. The operation performed by the MOVE.L X:(R3+2),D instruction
is shown in Figure 6-5.
After Execution
Before Execution
D
X
X
X
D
$0
$1234
X Memory
15
$5678
X Memory
15
0
0
$FFFF
$7004
$FFFF
$7003
$1234
$7003
$1234
$7002
$5678
$7002
$5678
$7004
$7001
$DD
$CC
$7001
$DD
$CC
$7000
$BB
$AA
$7000
$BB
$AA
R3
$7000
R3
$7000
N
$9876
M01
$FFFF
Word
Long
4 Bytes
+
N
$9876
M01
$FFFF
<< 1
Short Immediate Value
from the Instruction Word
Figure 6-5. Executing the MOVE.L X:(R3+2),D Instruction
Note that, for instructions that move bytes, the offset is specified in the number of bytes, whereas, for word
and long instructions, the offset is specified in the number of words. Also note that accesses to bytes in the
data structure in Example 6-3 on page 6-11 require the MOVE.B and MOVEU.B instructions instead of
MOVE.BP and MOVEU.BP. This requirement exists because the R3 register is used as a word pointer.
6-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Byte Pointer Memory Accesses
6.5.4 Accessing Program Memory
Program memory accesses are always performed with word pointers. The general rules for word pointer
accesses, as discussed in Section 6.5, “Word Pointer Memory Accesses,” through Section 6.5.3,
“Accessing Data Structures,” apply to program memory accesses. However, many fewer addressing modes
are supported. The addressing modes that can be used when program memory is accessed appear in
Table 6-3.
Table 6-3. Addressing Mode Arithmetic—Program Memory
Addressing Mode
Address for Word Access
Comments
Post-increment
P:(Rj)+
Rn+1
Word accesses only
Post-update by offset N
P:(Rj)+N
Rn+N
Word accesses only
6.6 Byte Pointer Memory Accesses
Instructions that use address registers as byte pointers can only access bytes from data memory. An address
register value is interpreted as a byte pointer when an instruction with a “.BP” extension is used, such as
MOVE.BP or CLR.BP.
Table 6-4 on page 6-14 shows the byte address that is accessed for the different byte pointer addressing
modes. The address of the word that is accessed in memory is the byte address from the table, right shifted
1 bit; the LSB of the byte address in the table selects the upper or lower byte. Note that the X:xxxx and
X:xxxxxx addressing modes specify an absolute byte address, with the upper n – 1 bits specifying the
correct word in memory and the LSB selecting the upper or lower byte.
NOTE:
Bytes can not be accessed in the top half of data memory using byte
pointers. Bytes can still be accessed in the complete data memory space
using word pointers; but if byte pointers are used, only the lower half of
data memory can be accessed.
Freescale Semiconductor
Address Generation Unit
6-13
Address Generation Unit
Table 6-4. Addressing Mode Arithmetic—Byte Pointers to Data Memory
Addressing Mode
No update
X:(RRR)
Address for Byte Access
Comments
RRR
Not allowed for SP register
Post-increment
X:(RRR)+
RRR+1
Not allowed for SP register
Post-decrement
X:(RRR)–
RRR–1
Not allowed for SP register
Post-update by offset N
X:(RRR)+N
Indexed by offset N
X:(RRR+N)
—
RRR+N
Indexed by 3-bit offset
X:(RRR+x)
—
Indexed by 6-bit offset
X:(SP–xx)
—
Indexed by 3-bit offset
X:(SP–x)
—
Indexed by 16-bit offset
X:(RRR+xxxx)
RRR+xxxx
Indexed by 24-bit offset
X:(RRR+xxxxxx)
RRR+xxxxxx
6-bit absolute short
X:aa
—
6-bit peripheral short
X:pp
—
16-bit absolute address
X:xxxx
00xxxx
24-bit absolute address
X:xxxxxx
xxxxxx
Not allowed for SP register
Must use MOVE.B or MOVEU.B with word
pointer
Must use MOVE.B or MOVEU.B with word
pointer
Zero-extended 16-bit offset; not allowed for SP
register
Not allowed for SP register
6.6.1 Byte Pointers vs. Word Pointers
Both the MOVE.B and MOVE.BP instructions (and their unsigned counterparts) can be used to access
bytes in memory. The difference between them is how the address register operand is interpreted. When
the MOVE.B instruction is used, the address register operand is treated as a word pointer. When
MOVE.BP is used, the address register operand is treated as a byte pointer. Note that word pointers have
full visibility of the complete 32Mbyte data memory space, but when byte pointers are used, only the lower
half of data memory can be accessed.
Although it is possible to access bytes in memory with either type of pointer, there are times when using a
byte pointer makes more sense than using a word pointer, and at other times the opposite is true. Word
pointers can be used to access a data element of any size, so they should be used when mixed data is
6-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Byte Pointer Memory Accesses
accessed (such as occurs in data structures). However, post-updating word pointers always occurs in word
addresses, so using a word pointer in a post-update addressing mode to access a byte array would only
access every other byte. Using byte pointers fixes this problem.
Byte pointers are only used if an instruction contains the “.BP” suffix. Otherwise, the pointer is always
interpreted as a word pointer. The offsets for all instructions that are accessing bytes from memory are
always byte offsets, regardless of whether an instruction uses a pointer as a byte or word pointer.
Example 6-4 demonstrates the difference between the MOVE.BP and MOVE.B instructions using
numerical values. For each instruction in Example 6-4, the comment shows the word address where the
access occurs as well as the byte that is selected (upper or lower byte of the word).
Example 6-4. Comparison of MOVE.BP and MOVE.B Instructions
; Load the R0, SP Address Pointers
MOVEU.W#$2000,R0
;
;
MOVEU.W#$4000,SP
;
;
load R0 pointer with the value $2000
(can be either a byte or word pointer)
load the stack pointer (SP) with $4000
(SP must always be a word pointer)
; MOVE.BP -- R0 used as a byte pointer,
MOVE.BPx:(r0+0),x0 ; word
MOVE.BPx:(r0+1),x0 ; word
MOVE.BPx:(r0+2),x0 ; word
MOVE.BPx:(r0+3),x0 ; word
MOVE.BPx:(r0+4),x0 ; word
MOVE.BPx:$2005,x0
offset is
address =
address =
address =
address =
address =
a byte
$1000,
$1000,
$1001,
$1001,
$1002,
offset
selects
selects
selects
selects
selects
byte
byte
byte
byte
byte
; word address = $1002, selects upper byte
; MOVE.B -- R0 used as a word pointer, offset is a byte offset
MOVE.B x:(r0+0),x0 ; word address = $2000, selects
MOVE.B x:(r0+1),x0 ; word address = $2000, selects
MOVE.B x:(r0+2),x0 ; word address = $2001, selects
MOVE.B x:(r0+3),x0 ; word address = $2001, selects
MOVE.B x:(r0+4),x0 ; word address = $2002, selects
; MOVE.B -- SP always used as a
MOVE.B x:(sp),x0
MOVE.B x:(sp-1),x0
MOVE.B x:(sp-2),x0
MOVE.B x:(sp-3),x0
MOVE.B x:(sp-4),x0
lower
upper
lower
upper
lower
lower
upper
lower
upper
lower
byte
byte
byte
byte
byte
word pointer, offset is a byte offset
; word address = $4000, selects lower
; word address = $3fff, selects upper
; word address = $3fff, selects lower
; word address = $3ffe, selects upper
; word address = $3ffe, selects lower
byte
byte
byte
byte
byte
In Example 6-4, the address pointer R0 is loaded with the value $2000. Locations near the word address
$2000 are accessed when R0 is interpreted as a word pointer (when MOVE.B is used). Locations near the
word address $1000 are accessed when MOVE.BP is used, which causes R0 to be interpreted as a byte
pointer.
6.6.2 Byte Arrays
Byte arrays are a common data structure in many applications; they are often used to store string values.
The DSP56800E instruction set makes it easy to access and manipulate byte arrays through the use of byte
pointers.
The code in Example 6-5 on page 6-16 shows an eight-element byte array being initialized and also shows
accesses to the array. The base of the array is loaded first as a byte pointer via the assembler’s lb()
function. The first two move instructions access the fifth and eighth array elements, respectively. The base
of the array is then reloaded, and the last two move instructions demonstrate sequential accesses to byte
elements.
Freescale Semiconductor
Address Generation Unit
6-15
Address Generation Unit
Example 6-5. Accessing Elements in an Array of Bytes
ORG
DCB
DCB
DCB
DCB
ARRAY1
CODESTART
X:$3000
$22,$11
$44,$33
$66,$55
$88,$77
;
;
;
;
;
Array of Bytes named “ARRAY1”
1st is $11, 2nd is $22
3rd is $33, 4th is $44
5th is $55, 6th is $66
7th is $77, 8th is $88
ORG
P:
; (instructions located in program memory)
[email protected](ARRAY1),R1; set up byte pointer to base of array
MOVE.BPX:(R1+4),A ; read with offset of 4 bytes from R1 (byte pointer)
MOVEU.BPX:(R1+7),B ; read with offset of 7 bytes from R1 (byte pointer)
[email protected](ARRAY1),R1; set up byte pointer to base of array
MOVE.BPX:(R1)+,C
; read first array element and advance pointer
MOVE.BPX:(R1)+,D
; read second array element and advance pointer
After the code in Example 6-5 has been executed, the values in the accumulator registers are:
Before Execution
After Execution
A
X
X
X
A
$0
$0055
$0000
B
X
X
X
B
$0
$0088
$0000
C
X
X
X
C
$0
$0011
$0000
D
X
X
X
D
$0
$0022
$0000
Recall that constant offset values are always specified in bytes when byte accesses are performed.
Figure 6-6 on page 6-17 demonstrates the AGU arithmetic that is performed when the instruction
MOVE.B X:(R1+7),B is executed. Because R1 is a byte pointer and an offset of 7 bytes has been
specified, the eighth element in the array is read.
6-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Byte Pointer Memory Accesses
After Execution
Before Execution
B
X
X
X
B
$0
$0088
X Memory
Word
Addresses
15
$0000
X Memory
0
15
0
Word Address: $3003
Byte Select: 1 (upper)
$3003
$88
$77
$3003
$88
$77
$3002
$66
$55
$3002
$66
$55
$3001
$44
$33
$3000
$22
$11
$3001
$44
$33
$3000
$22
$11
>>1
LSB
$6007
R1
$6000
R1
$6000
+
Byte Address: $6007
N
$9876
N
$9876
M01
$FFFF
M01
$FFFF
Short Immediate Value
from the Instruction Word
Figure 6-6. Executing the MOVEU.BP X:(R1+7),B Instruction
As Figure 6-6 shows, the byte address $6007 is accessed to load the B accumulator. Note that because this
address is a byte address, the byte is actually retrieved from the upper half of the word that is located at the
address $3000.
Freescale Semiconductor
Address Generation Unit
6-17
Address Generation Unit
6.7 AGU Arithmetic Instructions
In addition to the address arithmetic performed by the various addressing modes, the AGU supports a
number of powerful instructions for directly manipulating address registers. The AGU arithmetic
instructions enable more complex address calculations. These instructions make no distinction between
word and byte pointers, calculating results the same way for both.
Table 6-5 summarizes the AGU arithmetic instructions. For more detailed information, refer to the
appropriate entry in Appendix A, “Instruction Set Details.”
Table 6-5. AGU Address Arithmetic Instructions
Instruction
Address Calculation
Comments
ADDA Rm,Rn
Rn = Rn+Rm
ADDA.L Rm,Rn
Rn = Rn+(Rm<<1)
ADDA Rm,Rn,N
N = Rn+Rm
ADDA.L Rm,Rn,N
N = Rn+(Rm<<1)
ADDA #x,Rn
Rn = #x+Rn
#x is a 4-bit unsigned value.
ADDA #x,Rn,N
N = #x+Rn
#x is a 4-bit unsigned value.
ADDA #xxxx,Rm,Rn
Rn = #xxxx+Rm
#xxxx is a signed 17-bit value.
ADDA.L #xxxx,Rm,Rn
Rn = #xxxx+(Rm<<1)
#xxxx is an unsigned 16-bit value.
ADDA #xxxx,HHH,Rn
Rn = #xxxx+HHH
HHH—data ALU register that is treated as a
signed 16-bit value.
#xxxx is an unsigned 16-bit value.
ADDA.L #xxxx,HHH,Rn
Rn = #xxxx+(HHH<<1)
HHH—data ALU register that is treated as a
signed 16-bit value.
#xxxx is an unsigned 16-bit value.
ADDA #xxxxxx,Rm,Rn
Rn = #xxxxxx+Rm
#xxxxxx is a signed 24-bit value.
ADDA.L #xxxxxx,Rm,Rn
Rn = #xxxxxx+(Rm<<1)
#xxxxxx is a signed 24-bit value.
ADDA #xxxxxx,HHH,Rn
Rn = #xxxxxx+HHH
HHH—data ALU register that is treated as a
signed 16-bit value.
#xxxx is an unsigned 16-bit value.
ADDA.L #xxxxxx,HHH,Rn
Rn = #xxxxxx+(HHH<<1)
HHH—data ALU register that is treated as a
signed 16-bit value.
#xxxx is an unsigned 16-bit value.
ASLA Rm,Rn
Rn = (Rm<<1)
ASRA Rn
Rn = (Rn>>1)
Arithmetic right shift.
CMPA Rm,Rn
Rn–Rm
The result is not stored, but the condition codes
are set based on the 24-bit result.
CMPA.W Rm,Rn
Rn–Rm
The result is not stored, but the condition codes
are set based on the lowest 16 bits of the result.
DECTSTA Rn
Rn = Rn–1
Decrement by one and then set the condition
codes.
6-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
AGU Arithmetic Instructions
Table 6-5. AGU Address Arithmetic Instructions (Continued)
Instruction
Address Calculation
Comments
DECA Rn
Rn = Rn–1
Decrement by one.
DECA.L Rn
Rn = Rn–2
Decrement by two.
LSRA Rn
Rn = (Rn>>1)
Logical right shift.
NEGA Rn
Rn = –(Rn)
Negate register value
SUBA Rm,Rn
Rn = Rn–Rm
SUBA #xx,SP
SP = SP–#xx
#x is a 6-bit unsigned value.
SXTA.B Rn
Rn = sign_extend(Rn,7)
Sign extend the upper 16 bits of a register using
the value of bit 7 for sign extension.
SXTA.W Rn
Rn = sign_extend(Rn,15)
Sign extend the upper 8 bits of a register using
the value of bit 15 for sign extension.
TFRA Rm,Rn
Rn = Rm
Transfer one 24-bit register to another.
TSTA.B Rn
(Rn & 0x0000FF)–0
Test byte—the result is not stored anywhere, but
the condition codes are set based on the lower 8
bits of the result.
TSTA.W Rn
(Rn & 0x00FFFF)–0
Test word—the result is not stored, but the condition codes are set based on the lower 16 bits
of the result.
TSTA.L Rn
Rn–0
Test long—the result is not stored, but the condition codes are set based on the result.
TSTDECA.W Rn
Rn = Rn–1
Test the lower 16 bits of the value in the Rn register, set the condition codes, and then decrement the register.
ZXTA.B Rn
Rn = Rn & 0x0000FF
Zero extend a byte value.
ZXTA.W Rn
Rn = Rn & 0x00FFFF
Zero extend a word value.
Section 6.8.5.3, “Modulo Addressing for AGU Arithmetic Instructions,” lists the AGU arithmetic
instructions that can be affected by modulo arithmetic.
Freescale Semiconductor
Address Generation Unit
6-19
Address Generation Unit
6.8 Linear and Modulo Address Arithmetic
When an arithmetic operation is performed in the address generation unit, two modes of address
computation can be used: linear or modulo arithmetic. Linear arithmetic is required for general purpose
address computation and is found on all microprocessors. Modulo arithmetic allows the creation of special
data structures in memory. Data is manipulated by updating address registers (pointers) rather than moving
large blocks of data.
Many DSC and standard control algorithms require the use of specialized data structures, such as circular
buffers, FIFOs, and stacks. Using these structures allows data to be manipulated simply by updating
address register pointers, rather than by moving large blocks of data. The DSP56800E architecture
provides support for these algorithms by implementing modulo arithmetic in the address generation unit.
Modulo arithmetic is enabled for the R0 and R1 registers through programming the modifier register
(M01). Modulo arithmetic is not available for the R2–R5, N, and SP registers. Memory accesses using the
R2-R5, N, and SP pointers are always performed with linear arithmetic.
6.8.1 Linear Address Arithmetic
The alternative to modulo address arithmetic is linear arithmetic, as found on general-purpose
microprocessors. It is performed using 24-bit two’s-complement addition and subtraction. The 24-bit
offset register N, or immediate data (+1, –1, or a displacement value), is used in the address calculations.
Addresses are normally considered unsigned; offsets are considered signed.
Linear arithmetic is performed on the R2–R5, N, and SP registers at all times. Linear arithmetic is enabled
for the R0 and R1 registers through setting the modifier register (M01) to $FFFF. The M01 register is set to
$FFFF on reset. The shadow register for M01 is not initialized on reset, and must be manually set
according to the address arithmetic selection when shadow registers are swapped.
6.8.2 Understanding Modulo Arithmetic
To understand modulo address arithmetic, consider a circular buffer. A circular buffer is a block of
sequential memory locations with a special property: a pointer into the buffer is limited to the buffer’s
address range. When a buffer pointer is incremented such that it would point past the end of the buffer, the
pointer is “wrapped” back to the beginning of the buffer. Similarly, decrementing a pointer that is located
at the beginning of the buffer wraps the pointer to the end. This behavior is achieved by performing
modulo arithmetic when the buffer pointers are incremented or decremented. See Figure 6-7.
6-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
Upper Boundary: Lower Boundary + M01
Address
Pointer
Circular
Buffer
M01 = Size of Modulo Region Minus One
Lower Boundary: k LSBs Are All Zeros
Address of Lower Boundary:
23
Base Address
k k–1 ...
1 0
0 0 0 0 0
Figure 6-7. Circular Buffer
The modulo arithmetic unit in the AGU simplifies the use of a circular buffer by handling the address
pointer wrapping for you. After a buffer is established in memory, programming the M01 register enables
the R0 and R1 address pointers to wrap in the buffer area.
Modulo arithmetic is enabled through programming the M01 register with a value that is one less than the
size of the circular buffer. See Section 6.8.3, “Configuring Modulo Arithmetic,” for exact details on
programming the M01 register. Once modulo arithmetic is enabled, updates to the R0 or R1 register using
one of the post-increment or post-decrement addressing modes are performed with modulo arithmetic, and
the pointers wrap correctly in the circular buffer.
The address range within which the address pointers will wrap is determined by the value that is placed in
the M01 register and by the address that is contained within one of the pointer registers. Due to the design
of the modulo arithmetic unit, the address range is not arbitrary, but limited based on the value placed in
M01. The lower bound of the range is calculated by taking the size of the buffer, rounding it up to the next
higher power of two, and then rounding the address contained in the R0 or R1 pointer down to the nearest
multiple of that value.
For example: for a buffer size of M, the smallest value of k is calculated such that 2k > M. This value is the
buffer size rounded up to the next higher power of two. For a value M of 37, 2k would be 64. The lower
boundary of the range in which the pointer registers will wrap is the value in the R0 or R1 register with the
low-order k bits all set to zero, effectively rounding the value down to the nearest multiple of 2k (64 in this
case). This example is shown in Figure 6-8.
Freescale Semiconductor
Address Generation Unit
6-21
Address Generation Unit
Memory
$00B0
(Unavailable
Addresses)
Lower Bound + Size – 1 = Upper Bound
Upper Boundary: $00A4
$009F
Initial R0 Pointer Value
Circular
Buffer
Lower Boundary: $0080
Lower Bound Relative to R0
Figure 6-8. 37-Location Circular Buffer
When modulo arithmetic is performed on the buffer pointer register, only the low-order k bits are
modified; the upper 24 – k bits are held constant, fixing the address range of the buffer. The algorithm used
to update the pointer register (R0 in this case) is as follows:
R0[23:k] = R0[23:k]
R0[k–1:0] = (R0[k–1:0] + offset) MOD (M01 + 1)
Note that this algorithm can result in some memory addresses being inaccessible using modulo addressing.
If the size of the buffer is not an even power of two, there is a range of addresses between M and 2k – 1 (37
and 63 in the preceding example) that are not addressable. Section 6.8.9.3, “Memory Locations Not
Accessible Using Modulo Arithmetic,” discusses this issue in greater detail.
6.8.3 Configuring Modulo Arithmetic
As noted in Section 6.8.2, “Understanding Modulo Arithmetic,” modulo arithmetic is enabled through
programming the address modifier register, M01. This single register enables modulo arithmetic for both
the R0 and R1 registers. However, in order for modulo arithmetic to be enabled for the R1 register, it must
be enabled for the R0 register as well. When both pointers use modulo arithmetic, the sizes of both buffers
are the same. The pointers can refer to the same or different buffers as desired.
6.8.3.1 Configuring for Byte and Word Accesses
Modulo arithmetic affects not only the arithmetic used in calculating effective addresses for move
instructions, but it also affects the AGU arithmetic instructions. Table 6-6 shows how the M01 register is
correctly programmed for instructions that perform byte or word memory accesses as well as for the AGU
arithmetic instructions.
For byte memory accesses:
•
Modulo arithmetic is performed on byte addresses.
•
M01 = (size of the buffer in bytes) – 1.
6-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
For word memory accesses:
•
Modulo arithmetic is performed on word addresses.
•
M01 = (size of the buffer in words) – 1.
Table 6-6. Programming the M01 Register—Byte and Word Accesses
16-Bit M01 Register Contents
Address Arithmetic Performed
Pointer Registers Affected
$0000
(Reserved)
—
$0001
Modulo 2
R0 pointer only
$0002
Modulo 3
R0 pointer only
...
...
...
$3FFE
Modulo 16383
R0 pointer only
$3FFF
Modulo 16384
R0 pointer only
$4000
(Reserved)
—
...
...
...
$7FFF
(Reserved)
—
$8000
(Reserved)
—
$8001
Modulo 2
R0 and R1 pointers
$8002
Modulo 3
R0 and R1 pointers
...
...
...
$BFFE
Modulo 16383
R0 and R1 pointers
$BFFF
Modulo 16384
R0 and R1 pointers
$C000
(Reserved)
—
...
...
...
$FFFE
(Reserved)
—
$FFFF
Linear Arithmetic
R0 and R1 pointers
NOTE:
The reserved sets of modifier values ($0000, $4000–$8000, and
$C000–$FFFE) must not be used. The behavior of the modulo arithmetic
unit is undefined for these values and might result in erratic program
execution.
6.8.3.2 Configuring for Long Word Accesses
The modifier register must be programmed a little differently when long-word data is to be accessed. Since
each long-word location in the modulo buffer uses up two word memory locations, the size of the modulo
buffer in words must always be an even number, which means that M01 will always be programmed with
an odd value.
Freescale Semiconductor
Address Generation Unit
6-23
Address Generation Unit
For long-word memory accesses:
•
Modulo arithmetic is performed on word addresses.
•
M01 = 2 × (size of the buffer in long words) – 1
Table 6-7 on page 6-25 shows how the M01 register is correctly programmed for long memory accesses.
Note that all valid entries in this table are odd values, which results from the fact that 2 words are allocated
for each long value in the modulo buffer.
For example, to create a circular buffer with four 32-bit locations, calculate M01 as follows:
M01 = ( 2 × 4 ) – 1
= 8–1
= 7
The four 32-bit locations would require 8 words of data memory, so the M01 register is programmed with
the value “$0007.”
6-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
Table 6-7. Programming the M01 Register—Long-Word Accesses
16-Bit M01 Register Contents
Address Arithmetic Performed
Pointer Registers Affected
$0000
(Reserved)
—
$0001
Modulo 2
R0 pointer only
$0002
(Not available)
R0 pointer only
$0003
Modulo 4
R0 pointer only
$0004
(Not available)
R0 pointer only
...
...
...
$3FFC
(Not available)
R0 pointer only
$3FFD
Modulo 16382
R0 pointer only
$3FFE
(Not available)
R0 pointer only
$3FFF
Modulo 16384
R0 pointer only
$4000
(Reserved)
—
...
...
...
$7FFF
(Reserved)
—
$8000
(Reserved)
—
$8001
Modulo 2
R0 and R1 pointers
$8002
(Not available)
R0 and R1 pointers
$8003
Modulo 4
R0 and R1 pointers
$8004
(Not available)
R0 and R1 pointers
...
...
...
$BFFC
(Not available)
R0 and R1 pointers
$BFFD
Modulo 16382
R0 and R1 pointers
$BFFE
(Not available)
R0 and R1 pointers
$BFFF
Modulo 16384
R0 and R1 pointers
$C000
(Reserved)
—
...
...
...
$FFFE
(Reserved)
—
$FFFF
Linear Arithmetic
R0 and R1 pointers
NOTE:
The reserved sets of modifier values ($0000, $4000–$8000,
$C000–$FFFE, and all even values) must not be used. The behavior of the
Freescale Semiconductor
Address Generation Unit
6-25
Address Generation Unit
modulo arithmetic unit is undefined for these values and might result in
erratic program execution.
The high-order 2 bits of the M01 register determine the arithmetic mode for R0 and R1. A value of 00 for
M01[15:14] selects modulo arithmetic for R0. A value of 10 for M01[15:14] selects modulo arithmetic for
both R0 and R1. A value of 11 disables modulo arithmetic. The remaining 14 bits of M01 hold the size of
the buffer minus one.
6.8.4 Base Pointer and Offset Values in Modulo Instructions
For all instructions supporting modulo arithmetic (see Section 6.8.5, “Supported Memory Access
Instructions,” on page 6-29), there is always a “base pointer” and an “offset value” or “update value”. The
base pointer specifies an AGU register or absolute address which points to a location in the modulo buffer.
The offset (update) value is an immediate offset or AGU register which specifies the amount used as an
offset or an update to the pointer, and the size of the offsets are subject to the restriction in Section 6.8.9.2,
“Restrictions on the Offset Register,” on page 6-34.
For example, in the X:(Rn+N) addressing mode, the base pointer is Rn and the offset value is N. In the
X:(Rn)+N addressing mode, the base pointer is Rn and the update value is N.
6.8.4.1 Operand Placement Table
Table 6-8 shows which operand is used as a base pointer and which is used as offset value for the
addressing modes (X: notation) or instructions listed below.
This table only applies to instructions where:
•
modulo arithmetic is enabled, and
•
R0 (or R1) are used as source registers in the addressing mode or instruction.
If either of these conditions is not true, then Table 6-8 can be ignored.
Table 6-8. Base Pointer and Offset/Update for DSP56800E Instructions
Addressing Mode
or Instruction
Base Pointer
Offset Value
(Update Value)
Comments
X:(Rn)
Rn
(no offset)
—
X:(Rn)+
Rn
+1
—
X:(Rn)-
Rn
-1
—
X:(Rn)+N
Rn
N
—
X:(Rn+N)
Rn
N
—
RRR
x
—
X:(Rn+>xxxx)
Rn
>xxxx
—
X:(Rn+>xxxx)
>xxxx
Rn
Alternate use for this
addressing mode. Rn
must be positive for
correct modulo operation.
Rn
>>xxxxxx
—
X:(RRR+x)
X:(Rn+>>xxxxxx)
6-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
Table 6-8. Base Pointer and Offset/Update for DSP56800E Instructions
Addressing Mode
or Instruction
Base Pointer
Offset Value
(Update Value)
>>xxxxxx
Rn
Alternate use for this
addressing mode. Rn
must be positive for
correct modulo operation.
ADDA Rx,Ry
Ry
Rx
—
ADDA Rx,Ry,N
Ry
Rx
—
ADDA #x,Rx
Rx
#x
—
X:(Rn+>>xxxxxx)
ADDA #x,Rx,N
Comments
Rx
#x
—
ADDA #>xxxx,Rx,Ry
#>xxxx
Rx
—
ADDA #>xxxx,Rx,Ry
Rx
#>xxxx
ADDA #>>xxxxxx,Rx,Ry
#>>xxxxxx
Rx
ADDA #>>xxxxxx,Rx,Ry
Rx
#>>xxxxxx
ADDA.L Rx,Ry
Ry
Rx
—
ADDA.L Rx,Ry,N
See Section 6.8.4.3
for the case where
the immediate value
is negative.
—
See Section 6.8.4.3
for the case where
the immediate value
is negative.
Ry
Rx
—
#>xxxx
Rx
—
#>>xxxxxx
Rx
—
DECA Rx
Rx
-1
—
DECA.L Rx
Rx
-2
—
DECTSTA Rx
Rx
-1
—
SUBA Rx,Ry
Ry
Rx
—
TSTDECA.W Rx
Rx
-1
—
ADDA.L #>xxxx,Rx,Ry
ADDA.L #>>xxxxxx,Rx,Ry
The following four instructions will not perform modulo arithmetic because R0 and R1 are not source
operands for the instruction. As a result, there are no restrictions on which operand is used as pointer and
which is used as offset.
•
ADDA
#>xxxx,HHH,Rx
•
ADDA
#>>xxxxxx,HHH,Rx
•
ADDA.L #>xxxx,HHH,Rx
•
ADDA.L #>>xxxxxx,HHH,Rx
6.8.4.2 Example of Incorrect Modulo Operation
Using the above table, we can see that the example below incorrectly uses the modulo addressing
mode because the pointer and offset are not mapped to the correct operands.
Freescale Semiconductor
Address Generation Unit
6-27
Address Generation Unit
Example 6-6. Invalid Use of the Modulo Addressing Mode
; Part 1 - Initialization
MOVEU.W#$5-1,M01
MOVEU.W#$008000,N
MOVEU.W#-2,R0
;
;
;
;
;
Modulo Enabled, buffer size = 5
Base Pointer for modulo buffer
NOTE: placed in N, NOT Rn
Offset Value used in addressing mode
NOTE: placed in Rn, NOT N
; Part 2 - INCORRECT - pointer/offset placement violates rules in Table 6-8
MOVE.W X:(R0+N),X0 ; Performs incorrect arithmetic
;
- base pointer in N
;
- offset value in R0
The solution to the above example would be to place $008000 into R0 and #-2 into N. Then the instruction
works correctly.
6.8.4.3 Special Case - ADDA Instructions in Modulo Arithmetic
It is possible to use the ADDA instruction to add or subtract immediate offsets from a pointer when
modulo arithmetic is enabled.
6.8.4.3.1
Case 1. Adding a Positive Immediate Offset to a Pointer
In the case where a positive value is to be added to a pointer, the ADDA instruction can be used. If the
immediate offset satisfies the size restriction in Section 6.8.4.4, then simply use the instruction as shown in
the example below:
Example 6-7. Adding Positive Offset to a Modulo Pointer
BUFF_SIZE
EQU
5
MOVEU.W#$BUFF_SIZE-1,M01
MOVE.L #$008000,R0
ADDA #3,R0
; Modulo Enabled, buffer size = 5
; Base Pointer for modulo buffer
; Update base pointer using positive value
6.8.4.3.2 Case 2. Adding a Negative Immediate Offset to a Pointer
In the case where a negative value is to be added to a pointer, this can also be accomplished using the
ADDA instruction. If the immediate offset satisfies the size restriction in Section 6.8.4.4, then modulo
operation works correctly if the following formula is used:
Offset = Buffer_Size - Desired_Offset
Example 6-8. Adding “–2” to a Modulo Pointer
BUFF_SIZE
EQU
5
MOVEU.W#BUFF_SIZE-1,M01
MOVE.L #$008000,R0
ADDA #(BUFF_SIZE-2),R0
; Modulo Enabled, buffer size = 5
; Base Pointer for modulo buffer
; Update base pointer by -2
6.8.4.4 Restrictions on the Offset Values
Modulo addressing will work correctly with the post-update addressing mode, (Rn)+N, as long as it
satisfies the following condition:
•
6-28
If an offset N is used in the address calculations, the 16-bit absolute value |N| must be less than or
equal to M01 + 1 for proper modulo addressing. This is because only a single modulo wraparound
is detected.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
Modulo addressing also requires that any immediate values or AGU registers (see Section 6.8.4, “Base
Pointer and Offset Values in Modulo Instructions,” on page 6-26) used as offset values are subject to this
same constraint. On Example 6-9, the correct usage of offset values is demonstrated.
Example 6-9. Correct Usage - Offset Values Satisfying Restriction
BUFF_SIZE
EQU
64
; Buffer Size
; Initialization
MOVEU.W#BUFF_SIZE-1,M01; Modulo Enabled, buffer size = 64
MOVE.L #$008000,R0 ; Base Pointer for modulo buffer
MOVE.W #50,N
; Offset register - Note: offset <= BUFF_SIZE
TFRA N,R4
; Offset register - another copy
; Modulo Arithmetic works correctly for the following instructions:
MOVE.W X:(R0+N),X0 ; offset in N
MOVE.W X:(R0)+N,X0 ; offset in N
MOVE.W X:(R0+50),X0 ; offset is 50
MOVE.W X:(R0-50),X0 ; offset is -50
ADDA R4,R0
; offset in R4
SUBA N,R0
; offset in N
6.8.5 Supported Memory Access Instructions
Depending on the size of the memory values that are being accessed when modulo arithmetic is enabled,
different addressing modes and instructions are supported.
6.8.5.1 Modulo Addressing for Word Memory Accesses
The DSP56800E core’s address generation unit supports modulo arithmetic for the following
address-register-indirect modes when Rn is R0 or R1:
(Rn)
(Rn)+
(Rn)–
(Rn+N)
(Rn)+N
(Rn+x)
(Rn+xxxx)
(Rn+xxxxxx)
Modulo arithmetic can also be programmed for both the R0 and the R1 pointers, as shown in Section 6.8.3,
“Configuring Modulo Arithmetic.”
6.8.5.2 Modulo Addressing for Byte and Long Memory Accesses
Modulo arithmetic is also supported for both byte and long memory accesses. When byte pointers are used,
the following addressing modes support modulo address arithmetic, where Rn is R0 or R1:
(Rn)
(Rn+N)
(Rn)+
(Rn)–
(Rn+xxxx)
(Rn+xxxxxx)
The addressing modes that support modulo arithmetic for byte accesses when word pointers are used are
more limited:
(Rn+x)
Freescale Semiconductor
(Rn+xxxx)
Address Generation Unit
(Rn+xxxxxx)
6-29
Address Generation Unit
Finally, when modulo arithmetic is used while accessing long-word values, any of the following
addressing modes can be used:
(Rn)
(Rn+N)
(Rn)+
(Rn)–
(Rn+xxxx)
(Rn+xxxxxx)
Be careful to configure the M01 register properly based on the type of data that is being accessed when
modulo arithmetic has been enabled. See Section 6.8.3, “Configuring Modulo Arithmetic,” for more
information.
6.8.5.3 Modulo Addressing for AGU Arithmetic Instructions
The DSP56800E address generation unit also supports using modulo address arithmetic with some AGU
instructions. The supported instructions are the following:
ADDA*
ADDA.L
SUBA
DECA
DECA.L
DECTSTA
TSTDECA.W
For those supported AGU instructions that have more than one operand, modulo arithmetic will be used if
any of the source operands is a register for which modulo arithmetic has been enabled.
NOTE:
Refer to Section 6.8.4.3, “Special Case - ADDA Instructions in Modulo
Arithmetic,” on page 6-28 for special considerations on the ADDA
instruction.
6.8.6 Simple Circular Buffer Example
Suppose a five-location circular buffer is needed for an application. The application locates this buffer at
X:$800 in memory.1 In order for the AGU to be configured correctly to manage this circular buffer, the
following two pieces of information are needed:
•
The size of the buffer: 5 words
•
The location of the buffer: X:$0800–X:$0804
Modulo addressing is enabled for the R0 pointer through writing the size minus one ($0004) to M01[13:0]
and writing 00 to M01[15:14]. See Figure 6-9.
1. This location is arbitrary—any location in data memory would suffice.
6-30
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
$0804
Circular
Buffer
R0
M01 Register = Size – 1 = 5 – 1 = $0004
$0800
Figure 6-9. Simple Five-Location Circular Buffer
The location of the buffer in memory is determined by the value of the R0 pointer when it is used to access
memory. The size of the memory buffer (five in this case) is rounded up to the nearest power of two, which
is eight. The value in R0 is then rounded down to the nearest multiple of eight. For the base address to be
X:$0800, the initial value of R0 must be in the range X:$0800–X:$0804. Note that the initial value of R0
does not have to be X:$0800 to establish this address as the lower bound of the buffer. However, it is often
convenient to set R0 to the beginning of the buffer. The source code in Example 6-10 shows the
initialization of the example buffer.
Example 6-10. Initializing the Circular Buffer
MOVEU.W#(5-1),M01
MOVEU.W#$0800,R0
;
;
;
;
;
Initialize the buffer for five locations
R0 can be initialized to any location
within the buffer. For simplicity, R0
is initialized to the value of the lower
boundary
The buffer is used simply through being accessed with MOVE instructions. The effect of modulo address
arithmetic becomes apparent when the buffer is accessed multiple times, as in Example 6-11.
Example 6-11. Accessing the Circular Buffer
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
MOVE.W X:(R0)+,X0
;
;
;
;
;
;
;
First time accesses location $0800
and bumps the pointer to location $0801
Second accesses at location $0801
Third accesses at location $0802
Fourth accesses at location $0803
Fifth accesses at location $0804
and bumps the pointer to location $0800
; Sixth accesses at location $0800 <=== NOTE
; Seventh accesses at location $0801
; and so forth...
For the first several memory accesses, the buffer pointer is incremented as expected, from $0800 to $0801,
$0802, and so forth. When the pointer reaches the top of the buffer, rather than incrementing from $0804 to
$0805, the pointer value “wraps” back to $0800.
The behavior is similar when the buffer pointer register is incremented by a value greater than one.
Consider the source code in Example 6-12 on page 6-32, where R0 is post-incremented by three rather
than one. The pointer register correctly “wraps” from $0803 to $0801—the pointer does not have to land
exactly on the upper or lower bound of the buffer for the modulo arithmetic to wrap the value properly.
Freescale Semiconductor
Address Generation Unit
6-31
Address Generation Unit
Example 6-12. Accessing the Circular Buffer with Post-Update by Three
MOVEU.W#$0800,R0
MOVEU.W#3,N
NOP
NOP
MOVE.W X:(R0)+N,X0
MOVE.W X:(R0)+N,X0
MOVE.W X:(R0)+N,X0
MOVE.W X:(R0)+N,X0
; Initialize the pointer to $0800
; Initialize “bump value” to 3
;
;
;
;
First time accesses location $0800
and bumps the pointer to location $0803
Second accesses at location $0803
and wraps the pointer around to $0801
; Third accesses at location $0801
; and bumps the pointer to location $0804
; Fourth accesses at ...
In addition, the pointer register does not need to be incremented. Instructions that post-decrement the
buffer pointer also work correctly. Executing the instruction MOVE.W X:(R0)-,X0 when the value of R0
is $0800 will correctly set R0 to $0804.
6.8.7 Setting Up a Modulo Buffer
The following steps detail the process of setting up and using the 37-location circular buffer that is shown
in Figure 6-8 on page 6-22.
1. Determine the value for the M01 register.
— Select the size of the desired buffer; it can be no larger than 16,384 locations. If modulo
arithmetic is to be enabled only for the R0 address register, the result is the following:
M01 = # locations – 1 = 37 – 1 = 36 = $0024
— If modulo arithmetic is to be enabled for both the R0 and R1 address registers, be sure to set the
high-order bit of M01. In this case:
M01 = # locations – 1 + $8000 = 37 – 1 + 32768 = 32804 = $8024
2. Find the nearest power of two that is greater than or equal to the circular buffer size. In this
example, the value would be 2k ≥ 37, which gives a value of k = 6.
3. From k, derive the characteristics of the lower boundary of the circular buffer. Since the k
number of least significant bits of the address of the lower boundary must all be zeros, then
the buffer base address must be some multiple of 2k. In this case, k = 6, so the base address
is some multiple of 26 = 64.
4. Locate the circular buffer in memory.
— The location of the circular buffer in memory is determined by the upper (24 – k) bits of the
address pointer register that is used in a modulo arithmetic operation. For example, if there is
an open area of memory from locations 111 to 189 ($006F to $00BD), then the addresses of the
lower and upper boundaries of the circular buffer will fit in this open area for J = 2:
Lower boundary = (J × 64) = (2 × 64) = 128 = $0080
Upper boundary = (J × 64) + 36 = (2 × 64) + 36 = 164 = $00A4
— The exact area of memory in which a circular buffer is prepared is specified by picking a value
for the address pointer register, R0 or R1, whose value is inclusively between the desired lower
and upper boundaries of the circular buffer. Thus, selecting a value of 139 ($008B) for R0
would locate the circular buffer between locations 128 and 164 ($0080 to $00A4) in memory
since the upper 18 (from a total of 24 – k) bits of the address indicate that the lower boundary
is 128 ($0080).
6-32
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Linear and Modulo Address Arithmetic
In summary, the size and exact location of the circular buffer is defined once a value is assigned to
the M01 register and to the address pointer register (R0 or R1) that will be used in a modulo
arithmetic calculation.
5. Determine the upper boundary of the circular buffer:
upper boundary = lower boundary + number of locations – 1.
6. Select a value for the offset register if it is used in modulo operations.
— If the offset register is used in a modulo arithmetic calculation, it must be selected as follows:
|N| ≤ M01 + 1
|N| refers to the absolute value of the contents of the offset register.
— The special case where N is a multiple of the block size, 2k, is discussed in Section 6.8.8,
“Wrapping to a Different Bank.”
7. Perform the modulo arithmetic calculation.
— Once the appropriate registers are set up, the modulo arithmetic operation occurs when an
instruction is executed that uses any of the addressing modes in Section 6.8.5, “Supported
Memory Access Instructions,” with the R0 (or R1, if enabled) register.
— If the result of the arithmetic calculation would exceed the upper or lower bound, wrapping
around is correctly performed.
6.8.8 Wrapping to a Different Bank
Normally, when the absolute value of the offset register N, (|N|) used when performing modulo arithmetic
is less than or equal to M01, the primary address arithmetic unit automatically wraps the address pointer
around by the required amount. However, if |N| is greater than M01, the result is data-dependent and
unpredictable except for the special case where N is a multiple of the block size, 2k: N = L × (2k), where L
is a positive integer. In this special case, the pointer Rn is updated using linear arithmetic to the same
relative address that is L blocks forward in memory, as shown in Figure 6-10.
2k
M
(Rn) ± N MOD M01
where N = 2k (L = 1)
2k
M
Figure 6-10. Linear Addressing with a Modulo Modifier
Freescale Semiconductor
Address Generation Unit
6-33
Address Generation Unit
Note that this case requires that the offset N must be a positive two’s-complement integer. This technique
is useful in sequentially processing multiple tables (for example, implementing a bank of parallel IIR
filters) or N-dimensional arrays. The primary address arithmetic unit will automatically wrap around the
address pointer by the required amount.
6.8.9 Side Effects of Modulo Arithmetic
Due to the way modulo arithmetic is implemented by the DSP56800E, there are some potential side effects
that must be noted. Specifically, there are some restrictions and limitations that relate to the fact that the
base address of a buffer must be a power of two, and that the modulo arithmetic unit can only detect a
single wraparound.
6.8.9.1 When a Pointer Lies Outside a Modulo Buffer
If a pointer is outside the valid modulo buffer range, and an operation occurs that causes R0 or R1 to be
updated, the contents of the pointer are still updated using modulo arithmetic. This can result in the pointer
register being updated with an unexpected value, resulting in unusual behavior. Care should be taken to
ensure that the R0 and R1 pointers always point into a valid modulo buffer when modulo address
arithmetic is enabled.
For example, a MOVE.W B,X:(R0)+N instruction (where R0 = 6, M01 =5, and N = 0) would apparently
leave R0 unchanged since N is zero. However, since R0 is outside the boundary, the address calculation is
R0 + N - (M01 + 1) for the new contents of R0 and sets it to 0.
6.8.9.2 Restrictions on the Offset Register
The modulo arithmetic unit in the AGU is only capable of detecting a single wraparound of an address
pointer. As a result, if the post-update addressing mode—(Rn)+N—is used, be careful in selecting the
value of N. The 16-bit absolute value |N| must be less than or equal to M01 + 1 for proper modulo
addressing. Values of |N| that are larger than the size of the buffer may result in the Rn address value
wrapping twice, which the AGU cannot detect.
6.8.9.3 Memory Locations Not Accessible Using Modulo Arithmetic
When the size of a modulo buffer is not a power of two, there is a range of memory locations immediately
after the buffer that are not accessible with modulo addressing. Lower boundaries for modulo buffers
always begin on an address where the lowest k bits are zeros—that is, a power of two. This requirement
means that for buffers that are not an exact power of two, there are locations above the upper boundary that
are not accessible through modulo addressing.
In Figure 6-8 on page 6-22, for example, the buffer size is 37, which is not a power of two. The smallest
power of two that is greater than 37 is 64. Thus, there are 64 – 37 = 27 memory locations that are not
accessible with modulo addressing. These 27 locations are between the upper boundary + 1 = $00A5 and
the next power-of-two boundary address – 1 = $00C0 – 1 = $00BF.
These locations are still accessible when modulo arithmetic is not performed. Using linear addressing
(with the R2–R5 pointers), absolute addresses, or the no-update addressing mode makes these locations
available.
6-34
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 7
Bit-Manipulation Unit
The bit-manipulation unit performs bitfield operations on data memory and registers within the core. It is
capable of testing, setting, clearing, or inverting any bits that are specified in a 16-bit mask. This unit also
performs test-and-set operations, which test and update a value in a single atomic, non-interruptible
operation. Test-and-set instructions are especially useful for implementing semaphores and other key
system-programming operations.
The bit-manipulation unit can perform the following operations:
•
Testing selected bits in a 16-bit word:
— BFTSTH: Test a selected set of bits for all ones
— BFTSTL: Test a selected set of bits for all zeros
•
Testing selected bits in the upper or lower byte of a word and branching accordingly:
— BRSET: Branch if a selected set of bits is all ones
— BRCLR: Branch if a selected set of bits is all zeros
•
Testing and modifying bits in a 16-bit word:
— BFSET: Test and then set a selected set of bits
— BFCLR: Test and then clear a selected set of bits
— BFCHG: Test and then invert a selected set of bits
— BFSC: Test and then set/clear bitfield (DSP56800EX core only)
The bit-manipulation unit is connected to the major data buses within the core, enabling it to manipulate
data ALU registers, AGU registers, and peripheral registers as well as locations in memory. There is no
need to transfer data to dedicated bit-manipulation unit registers; in fact, the bit-manipulation unit does not
have any registers. This design greatly improves program and compiler efficiency.
NOTE:
The bitfield operations cannot be performed on program memory
locations, the Y register, or the HWS register.
This chapter describes the architecture and operation of the bit-manipulation unit. It also covers the use of
the ANDC, EORC, ORC, and NOTC instructions for performing logical operations with immediate data.
A variety of programming techniques for using the bit-manipulation instructions more effectively is also
presented.
Freescale Semiconductor
Bit-Manipulation Unit
7-1
Bit-Manipulation Unit
7.1 Bit-Manipulation Unit Overview and
Architecture
The bit-manipulation unit contains the following:
•
8-bit mask shifting unit
•
16-bit masking unit
•
16-bit testing unit
•
16-bit logic unit
A block diagram of the bit-manipulation unit appears in Figure 7-1.
DSC Core
Registers
Data Memory
Locations
Peripheral
Registers
CDBR
IP-BUS
Interface
CDBW
PDB
Optional 8-Bit Mask Shift
16-Bit Masking Unit
Test with 16-Bit Mask
To Carry Bit
in the
Status Register
16-Bit Logic Unit
Bit-Manipulation Unit
Figure 7-1. Bit-Manipulation Unit Block Diagram
The blocks within the bit-manipulation unit are explained in the following sections.
7.1.1 8-Bit Mask Shift Unit
The 8-bit mask shift unit performs two dedicated functions:
•
Right shifting an 8-bit immediate mask from the upper byte of a word to the lower byte of a word,
zeroing the upper 8 bits of the mask
•
Passing the upper 8 bits of the immediate mask to the 16-bit masking unit, zeroing out the lower
8 bits of the mask
This shifter is used when the BRCLR and BRSET instructions are executed. These instructions test only
the upper or lower byte of a word. See Example 7-1 on page 7-3.
7-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Bit-Manipulation Unit Overview and Architecture
Example 7-1. Examples of Byte Masks in BRSET and BRCLR Instructions
BRCLR #$0081,X0,LABEL1; Immediate Mask in lower byte
BRSET #$81,X:$3,LABEL1; Immediate Mask in lower byte
BRCLR #$8100,X0,LABEL1; Immediate Mask in upper byte
The other bit-manipulation instructions (BFTSTH, BFTSTL, BFCHG, BFCLR, and BFSET) work with a
full 16-bit mask, so no shifting is required.
This unit can optionally be bypassed, passing through a 16-bit mask directly to the 16-bit masking unit.
7.1.2 16-Bit Masking Logic
The 16-bit masking logic selects which of the bits in a 16-bit word will be operated on by the
bit-manipulation unit. Bits that are set to one in the mask are tested when the bit-manipulation operation is
performed. Bits that are set to zero in the mask are ignored.
Example 7-2 demonstrates an instruction that specifies a bit mask. The 4 bits that are set to one, bits 7–4,
are selected by the 16-bit masking unit, and only these 4 bits are tested and then cleared by the
bit-manipulation unit. The result of the test is stored in the status register’s carry bit. All other bits in the
X0 register (bits 15–8 and bits 3–0) are ignored and not modified by this instruction.
Example 7-2. Using a Mask to Operate on Bits 7–4
BFCLR #$00F0,X0
; Immediate Mask = $00F0
Note that bit masks are always specified with the use of an immediate value. The DSP56800E instruction
set does not support mask values in a register.
7.1.3 16-Bit Testing Logic
The 16-bit testing logic tests all bits that are specified in the immediate mask value. It is capable of
determining if the selected bits are either all ones or all zeros. The result of the test is then recorded in the
status register’s carry bit. Based on the instruction used, the testing logic performs the following:
For the BFTSTH, BRSET, BFCHG, BFCLR, and BFSET instructions:
•
Tests the selected bits for ones
•
Sets the C bit if all tested bits are one
•
Clears the C bit if not all tested bits are ones
For the BFTSTL and BRCLR instructions:
•
Tests the selected bits for zeros
•
Sets the C bit if all tested bits are zero
•
Clears the C bit if not all tested bits are zeros
These testing steps are performed before any modifications are made to the operand (by the BFCHG,
BFCLR, and BFSET instructions). Only the carry bit in the status register is affected.
Freescale Semiconductor
Bit-Manipulation Unit
7-3
Bit-Manipulation Unit
7.1.4 16-Bit Logic Unit
The 16-bit logic unit performs any modifications to the operand value before it is written back to the
original register or memory location. This unit performs the following operations for the following
instructions:
•
BFCHG—Inverts the bits selected by the 16-bit mask
•
BFCLR—Clears the bits selected by the 16-bit mask
•
BFSET—Sets the bits selected by the 16-bit mask
Any bit that is not selected by the 16-bit mask is not modified.
7.2 Bit-Manipulation Unit Operation
There are three different types of operations performed by the bit-manipulation unit. A description of each
operation appears in its own subsection.
7.2.1 Testing Bits
The bit-manipulation unit can test a set of bits within an operand. This testing operation is performed by
the following instructions:
•
BFTSTH
•
BFTSTL
The basic operations performed are:
1. Read the 16-bit operand from memory or from a register.
2. Create a 16-bit mask directly from the instruction itself. In most cases, the instruction
directly provides the 16-bit mask, but for the BRSET and BRCLR instructions, a 16-bit
mask is reduced to 8 bits, where either the upper or lower eight bits are zeros.
3. Use the mask to select the desired bits within the 16-bit operand that was already read from
an on-chip register or memory location.
4. Test all of the selected bits within this value. Check for whether all selected bits are zeros
or ones, as described in Section 7.1.3, “16-Bit Testing Logic.”
5. Write the result of this test to the C bit in the status register (SR).
Example 7-3 presents an example of an instruction that performs this operation.
Example 7-3. Testing Bits in an Operand
BFTSTL #$000F,X:$C000
; Test lower 4 bits of memory location
7.2.2 Conditional Branching
The bit-manipulation unit can test a set of bits in an operand and execute a conditional branch based on the
result of the test. This operation is performed by the following instructions:
7-4
•
BRCLR
•
BRSET
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ANDC, EORC, ORC, and NOTC
The basic operations performed are:
1. Perform steps 1 through 5 in Section 7.2.1, “Testing Bits.”
2. Branch to the specified target address if the result of the test performed is True. Otherwise,
continue program execution with the next sequential instruction.
Example 7-4 presents an example of an instruction that performs this operation.
Example 7-4. Branching on Bits in an Operand
BRSET #$8000,X:(R0),LABEL4
; Branch to LABEL4 if MSB set in X:(Rn)
7.2.3 Modifying Selected Bits
The bit-manipulation unit can perform logical operations on selected bits in an operand. The instructions
that perform these operations, in addition to performing the testing that is described in Section 7.1.3,
“16-Bit Testing Logic,” process selected bits in the original 16-bit source using the 16-bit logic unit and
write the results back to their original source. This operation is performed by the following instructions:
•
BFCHG
•
BFCLR
•
BFSET
•
BFSC (DSP56800EX core only)
The basic operations performed are:
1. Perform steps 1 through 5 in Section 7.2.1, “Testing Bits.”
2. Invert, clear, or set all bits selected by the 16-bit mask.
3. Write this modified result back to the 16-bit source operand.
Note that these three steps are a non-interruptible sequence because they are implemented within a single
bit-manipulation instruction.
Example 7-5 presents an example of an instruction that performs this operation.
Example 7-5. Clearing Bits in an Operand
BFCLR #$FF00,X:(R0)
; Clear upper byte of memory location
7.3 ANDC, EORC, ORC, and NOTC
With the use of the following four operations, the bit-manipulation unit gives the DSP56800E core the
capability to perform logical operations with immediate data:
•
ANDC—logically AND a 16-bit immediate value with an operand
•
EORC—logically exclusive OR a 16-bit immediate value with an operand
•
ORC—logically OR a 16-bit immediate value with an operand
•
NOTC—take the logical one’s-complement of a 16-bit destination
The operations ANDC, EORC, ORC, and NOTC are not instructions; they are aliases to the
bit-manipulation instructions that are identified in the preceding list. See Section 4.2.1, “The ANDC,
EORC, ORC, and NOTC Aliases,” on page 4-11 for additional information.
Freescale Semiconductor
Bit-Manipulation Unit
7-5
Bit-Manipulation Unit
7.4 Other Bit-Manipulation Capabilities
The bit-manipulation unit is supplemented by the capabilities found within the DSP56800E’s data ALU
unit. The data ALU instructions complement the capabilities of the bit-manipulation unit. Together these
two units provide very powerful bit-manipulation capabilities for efficient control processing.
The bit-manipulation capabilities within the data ALU unit include:
•
16- or 32-bit bi-directional logical and arithmetic shifting.
•
Single-bit arithmetic and logical shifts.
•
Single-bit 16- and 32-bit rotate instructions.
•
16- or 32-bit logical operations.
•
Incrementing and decrementing of memory locations.
In all but the last case, operations are performed directly on the registers within the data ALU unit. Refer to
Chapter 5, “Data Arithmetic Logic Unit,” for more details.
7.5 Programming Considerations
In order to use the bit-manipulation unit effectively, some considerations must be kept in mind when
writing code that uses it. The following sections describe the recommended approach to take, and a variety
of programming techniques that can be employed, when using the bit-manipulation unit.
7.5.1 Bit-Manipulation Operations on Registers
There are some potential side effects to consider when performing bit-manipulation operations on AGU
registers or the accumulators:
When bit-manipulation operations (BFCHG, BFCLR, or BFSET) are performed on 24-bit AGU registers,
the upper 8 bits of the register are set to zero.
Take special care when performing a bitfield operation on one of the data ALU accumulator registers.
Saturation may occur when an accumulator is accessed by the bit-manipulation unit. See Section 5.2.7,
“Bit-Manipulation Operations on Accumulators,” on page 5-14 for more information.
7.5.2 Bit-Manipulation Operations on Byte Values
The bit-manipulation instructions are designed to manipulate 16-bit quantities. It is possible, however, to
perform bit-manipulations on byte values by carefully selecting the 16-bit mask.
In general, the 8-bit mask to be used should be placed in the upper or lower byte of the 16-bit mask, and the
other byte in the word should be set to zero. This ensures that only bits in the appropriate byte are affected.
Note, however, that the ANDC instruction alias inverts the mask, so the byte mask should be padded with
ones instead of zeros.
Note that these operations still access and store 16-bit quantities. The mask is simply set so that only 1 byte
is operated on. This arrangement might have potentially adverse side effects when memory-mapped
peripheral registers are operated on.
7-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Programming Considerations
7.5.2.1 Absolute Addresses
For absolute addresses, the following rules apply:
•
The address used in the bit-manipulation instruction is the byte address, logically right shifted 1 bit.
•
For even byte addresses, the 8-bit mask is placed in the lower 8 bits of the 16-bit mask, and the upper
8 bits of the mask are zeroed.
•
For odd byte addresses, the 8-bit mask is placed in the upper 8 bits of the 16-bit mask, and the lower
8 bits of the mask are zeroed.
Two examples appear in Example 7-6.
Example 7-6. Logical Operations on Bytes in Memory
; AND the value $1F with a byte in data memory
; ===> 8-bit mask in lower byte of 16-bit mask, for lower byte at word address
ANDC #$FF1F,X:$1000
; Bit Operation
; (8-bit mask placed in lower byte)
; OR the value $F8 with a byte in data memory
; ===> 8-bit mask in upper byte of 16-bit mask, for upper byte at word address
ORC
#$F800,X:$1000
; Bit Operation
; (8-bit mask placed in upper byte)
Similar techniques can be used for performing bit operations on bytes with other addressing modes, such
as (Rn+xxxx).
7.5.2.2 Word Pointers with Byte Offsets
A technique that is similar to the one described in Section 7.5.2.1, “Absolute Addresses,” can be used for
manipulating a byte referenced through a word pointer with a byte offset. In this case, the technique that is
outlined in Section 7.5.3, “Using Complex Addressing Modes,” is used for synthesizing an address.
For addresses with byte offsets, the following rules apply:
•
The base address is stored in an Rn register as a word pointer.
•
The offset that is added to the pointer is the offset value in bytes, arithmetically right shifted 1 bit.
•
For even byte addresses, the 8-bit mask is placed in the lower 8 bits of the 16-bit mask, and the upper
8 bits of the mask are zeroed.
•
For odd byte addresses, the 8-bit mask is placed in the upper 8 bits of the 16-bit mask, and the lower
8 bits of the mask are zeroed.
Two examples appear in Example 7-7 on page 7-8.
Freescale Semiconductor
Bit-Manipulation Unit
7-7
Bit-Manipulation Unit
Example 7-7. Logical Operations on Bytes Using Word Pointers
;
;
;
;
AND the value $1F with the byte in data memory
(that is, the lower byte at word address X:$1001)
===> Word Pointer = $1000, byte offset = 2
===> 8-bit mask in lower byte of 16-bit mask
ADDA #1,Rn,N
; N = Rn + (byte offset >> 1)
ANDC #$FF1F,X:(N)
; Bit Operation
; (8-bit mask placed in lower byte)
;
;
;
;
AND the value $F8 with the byte in data memory
(that is, the upper byte at word address X:$1001)
===> Word Pointer = $1000, byte offset = 3
===> 8-bit mask in upper byte of 16-bit mask
ADDA #1,Rn,N
; N = Rn + (byte offset >> 1)
ANDC #$F8FF,X:(N)
; Bit Operation
; (8-bit mask placed in upper byte)
Similar techniques can be used for performing bit operations on bytes with other addressing modes.
7.5.3 Using Complex Addressing Modes
It is possible to create bit-manipulation operations with more complex addressing modes. AGU arithmetic
can be performed to emulate the desired addressing mode, with the resulting address stored in the N
register. Then the bit-manipulation operation is performed with the X:(N) addressing mode. Example 7-8
shows code that emulates more complex addressing modes.
Example 7-8. Bit-Manipulation Operations Using Complex Addressing Modes
; BFSET #MASK,X:(Rn+xxxx) Operation — performed in two instructions
ADDA #xxxx,Rn,N
; N = (Rn+xxxx)
BFSET #MASK,X:(N)
; Perform operation with synthesized address
; BFCLR #MASK,X:(Rn+Rm) Operation — performed in two instructions
ADDA Rm,Rn,N
; N = (Rn+Rm)
BFCLR #MASK,X:(N)
; Perform operation with synthesized address
7.5.4 Synthetic Conditional Branch and Jump Operations
The flexible instruction set of the DSP56800E architecture allows new bit-manipulation operations to be
synthesized with the use of existing DSP56800E instructions. This section presents some of these useful
operations that are not directly supported by the DSP56800E instruction set but that can be efficiently
synthesized by the user. Table 7-1 lists operations that can be synthesized in this manner.
Table 7-1. Operations Synthesized Using DSP56800E Instructions
Operation
7-8
Description
JRCLR
Jumps if all selected bits in bitfield clear
JRSET
Jumps if all selected bits in bitfield set
BR1CLR
Branches if at least 1 selected bit in bitfield is clear
BR1SET
Branches if at least 1 selected bit in bitfield is set
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Programming Considerations
Table 7-1. Operations Synthesized Using DSP56800E Instructions (Continued)
Operation
Description
JR1CLR
Jumps if at least 1 selected bit in bitfield is clear
JR1SET
Jumps if at least 1 selected bit in bitfield is set
Several operations for jumping and branching can be emulated, depending on the selected bits in a bitfield,
overflows, or other condition codes.
NOTE:
None of these operations are actual DSP56800E instructions; they are
macros that can be created from existing instructions.
7.5.4.1 JRSET and JRCLR Operations
The JRSET and JRCLR operations are very similar to the BRSET and BRCLR instructions. Like BRSET
and BRCLR, they perform a bitfield test and branch based on the result. However, the BRSET and BRCLR
instructions only allow branches to locations that are up to 64 locations away from the current instruction,
and they can only test an 8-bit bitfield. The JRSET and JRCLR operations allow jumps to anywhere in the
program address space and can specify a 16-bit mask.
Example 7-9. JRSET and JRCLR Operations
; JRSET Operation — performed in two DSP56800E instructions
BFTSTH #MASK,X:<ea>
; 16-bit mask allowed
JCS
LABEL9
; 19- or 21-bit jump address allowed
; JRCLR Operation — performed in two DSP56800E instructions
BFTSTL #MASK,X:<ea>
; 16-bit mask allowed
JCS
LABEL9
; 19- or 21-bit jump address allowed
JRSET and JRCLR use the BFTSTH and BFTSTL instructions to perform the bitfield test. Thus, they can
use the same addressing modes as those bit-manipulation instructions.
7.5.4.2 BR1SET and BR1CLR Operations
The BRSET and BRCLR instructions are very useful, since they branch to a different address based on a
bitfield comparison. However, the design of these instructions is such that all the bits in the mask must
match the value being tested, or the branch is not taken. In some cases, it would be more useful to branch if
at least 1 bit in the mask matched. The BR1SET and BR1CLR operations provide just that functionality.
See Example 7-10.
Example 7-10. BR1SET and BR1CLR Operations
; BR1SET Operation — performed in two DSP56800E instructions
BFTSTL #MASK,X:<ea>
; 16-bit mask allowed
BCC
LABEL10
; 7-, 18-, 22-bit signed PC-relative offset
; allowed
; BR1CLR Operation — performed in two DSP56800E instructions
BFTSTH #MASK,X:<ea>
; 16-bit mask allowed
BCC
LABEL10
; 7-, 18-, 22-bit signed PC-relative offset
; allowed
Freescale Semiconductor
Bit-Manipulation Unit
7-9
Bit-Manipulation Unit
In addition to having the ability to branch based on a single bit, the BR1SET and BR1CLR operations can
also specify a 16-bit mask, as compared to an 8-bit mask for BRSET and BRCLR. These operations allow
the same addressing modes as the BFTSTH and BFTSTL instructions.
7.5.4.3 JR1SET and JR1CLR Operations
The JR1SET and JR1CLR operations function almost identically to the BR1SET and BR1CLR operations
that are described in Section 7.5.4.2, “BR1SET and BR1CLR Operations.” The JR1SET and JR1CLR
operations differ from the BR1SET and BR1CLR operations in that the former pair uses absolute
addressing. See Example 7-11.
Example 7-11. JR1SET and JR1CLR Operations
; JR1SET Operation — performed in two DSP56800E instructions
BFTSTL #MASK,X:<ea> ; 16-bit mask allowed
JCC
LABEL11
; 19- and 21-bit jump to absolute address allowed
; JR1CLR Operation — performed in two DSP56800E instructions
BFTSTH #MASK,X:<ea> ; 16-bit mask allowed
JCC
LABEL11
; 19- and 21-bit jump to absolute address allowed
The JR1SET and JR1CLR operations specify a 16-bit mask and a 19-bit target address, allowing jumps to
anywhere in the program address space. These operations allow the same addressing modes as the
BFTSTH and BFTSTL instructions.
7-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 8
Program Controller
The program controller is perhaps the most important unit in the DSC core. It fetches and decodes
instructions, coordinates the other core units in executing the instructions, and directs program flow,
including exception processing. It also contains dedicated circuitry to accelerate looping operations.
This chapter describes the program controller’s function, including details on stack handling and
no-overhead hardware looping. The different processing states, including reset and exception processing,
are covered in Chapter 9, “Processing States.” For more in-depth information on the execution pipeline,
see Chapter 10, “Instruction Pipeline.”
8.1 Program Controller Architecture
A block diagram of the program controller is given in Figure 8-1 on page 8-2. As the figure shows, the
following major blocks are located within the program controller:
•
Instruction latch and decoder
•
Program counter (PC)
•
Hardware stack
•
Looping control unit
•
Interrupt control unit
The blocks and registers within the program controller are explained in the following sections.
Freescale Semiconductor
Program Controller
8-1
Program Controller
PAB
CDBR
CDBW
PDB
21-Bit Incrementer
Program Counter
20
15
0
0
PC
Instruction Latch
Hardware Stack
23
0
HWS0
Instruction Decoder
LF NL
HWS1
Control Signals
20 Fast Interrupt Return Address 0
FIRA
23
Loop Address
15
LA
Interrupt Controller
(Located Outside the DSC Core)
LA2
IPR
Loop Counter
0
0
LC
Interrupt
Control
LC2
Interrupt
Arbitration
Interrupt Request
|1,|0 Bits from SR
Looping Control
Operating Mode and Status Register
(OMR, SR)
15
MODA, MODB Signals
0
Control Bits to DSC Core
OMR
SR
Condition Codes from
Data ALU or AGU
Interrupt
Priority
Update
Status and Control Bits
to DSC Core
12
0
FISR
Fast Interrupt Status Register
Figure 8-1. Program Controller Block Diagram
8.1.1 Instruction Latch and Decoder
The instruction latch is a 16-bit internal register that is used to hold instruction opcodes that are fetched
from memory. The instruction decoder uses the contents of the instruction latch to control and synchronize
the other execution units in performing the specified operation.
8-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Architecture
8.1.2 Program Counter
The program counter (PC) is a 21-bit register that contains the address of the next item that is to be fetched
from program memory. The PC can point to instructions, data operands, or addresses of operands. Under
normal operation, all references to this register are implicit; no instruction can manipulate it directly.
The program counter value is split between two locations in the core. The lowest 16 bits are stored in the
PC register, while the top 5 bits are located in the upper word of the status register (SR). See
Section 8.2.2.10, “Program Counter Extension (P0–P4)—Bits 10–14,” for more information.
8.1.3 Looping Control Unit
The looping control unit controls the hardware-accelerated looping capability in the core. With the REP,
DO, and DOSLC instructions, program loops can be executed with very little overhead, resulting in
substantial time savings. For more information on the hardware looping capabilities that are included in the
core, see Section 8.5, “Hardware Looping.”
8.1.4 Hardware Stack
The hardware stack is a 2-deep, 24-bit-wide, last-in-first-out (LIFO) stack that is used to enable the nesting
of hardware loops. It stores the address of the first instruction in a loop, so execution of an outer hardware
loop can continue when an inner hardware loop has completed.
When the stack limit is exceeded, the oldest loop information (top-of-loop address and LF bit) is lost, and a
non-maskable hardware stack overflow interrupt occurs. There is no interrupt on hardware stack
underflow.
The hardware stack can be manipulated under program control using the hardware stack register (HWS),
which is discussed in Section 8.2.7, “Hardware Stack Register.”
8.1.5 Interrupt Control Unit
The interrupt control unit coordinates interrupt and exception processing in the core. It is assisted in this
task by the interrupt controller (located outside the core), which performs interrupt arbitration and
indicates when an enabled interrupt request is pending. See Section 8.1.6, “Interrupt Controller.” Interrupt
arbitration and the exception processing state are discussed in Section 9.3, “Exception Processing State,”
on page 9-2.
8.1.6 Interrupt Controller
The interrupt controller is responsible for arbitrating all interrupt requests from the core and on-chip
resources. It typically arbitrates among all available interrupt requests, and then it checks the priority of the
highest request against the interrupt mask bits for the DSC core (I1 and I0 in the SR). If the requesting
interrupt has higher priority than the current priority level of the DSC core, then the unit generates a single
enabled interrupt request signal to the interrupt control unit within the core.
NOTE:
The interrupt controller is not part of the DSC core, but it is included on
any chip that is based on the DSP56800E or DSP56800EX core.
Freescale Semiconductor
Program Controller
8-3
Program Controller
8.2 Program Controller Programming Model
The programming model for the program controller consists of seven user-accessible registers and two
special registers for fast interrupt processing:
•
Status register (SR)
•
Operating mode register (OMR)
•
Hardware stack register (HWS)
•
Two loop address registers (LA and LA2)
•
Two loop count registers (LC and LC2)
•
Fast interrupt return address register (FIRA)
•
Fast interrupt status register (FISR)
Figure 8-2 depicts the registers graphically.
20
Program Counter
0
PC
Loop Address
23
Operating Mode and Status
Register
(OMR, SR)
15
0
OMR
0
SR
LA
LA2
Hardware Stack
23
Fast Interrupt Status
Register
12
0
0
FISR
HWS0
HWS1
20
Fast Interrupt Return Address
15
0
FIRA
Loop Counter
0
LC
LC2
Figure 8-2. Program Controller Programming Model
8.2.1 Operating Mode Register
The operating mode register (OMR) is a 16-bit register that controls the current operating mode of the
processor. It is used to configure the memory map and the operation of the data ALU, and it reflects the
status of these and other units in the core. The operating mode register’s format is described in the
following register display and in Table 8-1 on page 8-5.
8-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Programming Model
OMR
Operating Mode Register
BIT 15
14
8
7
6
5
4
3
NL
CM
XP
SD
R
SA
TYPE
rw
rw
rw
rw
rw
RESET
0
0
0
0
0
0
13
0
12
0
11
0
10
9
0
0
2
1
BIT 0
EX
MB
MA
rw
rw
rw
rw
0
—
—
—
0
Table 8-1. OMR Bit Descriptions
Name
NL
Bit 15
Description
Settings
Nested Looping—Indicates whether a nested
hardware DO loop is active or whether HWS has
been written to at least two times without being
read
0 = No nested DO loop active.
1 = Nested DO loop active.
Reserved
Bits 14–9
Reserved
These bits are reserved and always read
zero.
CM
Bit 8
Condition Code Mode—Selects whether 36-bit
or 32-bit values are used for condition codes
0 = 36-bit values are used.
1 = 32-bit values are used.
XP
Bit 7
X or P Memory Select—Determines the memory space from which instructions are fetched
0 = Fetched from P (program) memory.
1 = Fetched from X (data) memory.
SD
Bit 6
Stop Delay—Selects length of wake-up time
from stop mode
Dependent on individual chip’s implementation.
R
Bit 5
Rounding—Selects the rounding method
0 = Convergent rounding.
1 = Two’s-complement rounding.
SA
Bit 4
Saturation—Enables automatic saturation in
the data ALU
0 = Saturation disabled.
1 = Saturation enabled.
EX
Bit 3
External X Memory Select—Forces all data
memory access to be in external memory
0 = Internal data memory accesses.
1 = Data memory accesses are external.
Note:
See Section 8.4, “Hardware Stack.”
This bit is dependent on the individual chip's
implementation.
Reserved
Bit 2
Reserved
This bit is reserved and always reads zero.
MB and MA
Bits 1–0
Operating Mode—Selects the memory map
and operating mode
This bit is dependent on the individual chip’s
implementation.
NOTE:
When a bit of the OMR is changed by an instruction, a delay of
2 instruction cycles is necessary before the new mode comes into effect.
When individual bits in the OMR are modified, the BFCLR, BFCHG, or
BFSET instructions should be used instead of a MOVE instruction to
prevent the accidental modification of other bits.
Freescale Semiconductor
Program Controller
8-5
Program Controller
8.2.1.1 Operating Mode (MA and MB)—Bits 0–1
The operating mode (MB and MA) bits are used to select the operating mode and memory map. Their
initial values after reset are typically established by external mode select pins. After the chip leaves the
reset state, MB and MA can be changed under program control. Consult the specific DSC device’s
reference manual for more information about how these bits are established on reset and about their
specific effect on operation.
8.2.1.2 External X Memory (EX)—Bit 3
The external X memory (EX) bit can be used to configure the location of data memory. Typically, a
DSP56800E– or DSP56800EX–based device has some quantity of on-chip data memory, which can be
supplemented by external data memory as needed. The EX bit can be used by a chip to select whether both
on-chip and external memories are used or whether all data memory accesses are sent to external memory.
The exact effect of the EX bit depends on the architecture of a given device. Consult the appropriate
device’s user’s manual for more information on the EX bit.
8.2.1.3 Saturation (SA)—Bit 4
The saturation (SA) bit enables automatic saturation in the data ALU on 32-bit arithmetic results.
Normally, saturation occurs only when an accumulator is written to memory. When the SA bit is set,
saturation is performed on the results of all basic arithmetic operations, such as multiplication or addition,
before they are stored in an accumulator. This automatic saturation is useful for bit-exact DSC algorithms
that do not recognize or cannot take advantage of the extension registers that are available with each
accumulator. Automatic saturation is discussed in detail in Section 5.8.2, “MAC Output Limiter,” on
page 5-41. This bit is cleared by processor reset.
8.2.1.4 Rounding (R)—Bit 5
The rounding (R) bit selects the type of rounding that is used when RND, MACR, and other instructions
that round values are executed. When set, two’s-complement rounding (always round up) is used. When
cleared, convergent rounding is selected. The two rounding modes are discussed in Section 5.9,
“Rounding,” on page 5-43. This bit is cleared by processor reset.
8.2.1.5 Stop Delay (SD)—Bit 6
The stop delay (SD) bit selects the amount of time it takes to wake up from stop mode. When the bit is set,
the processor exits quickly from stop mode; when the bit is cleared, a delay is inserted before the processor
exits stop mode. A long wake-up time can be useful to allow a crystal oscillator to settle before resuming
instruction execution. The exact length of the delay depends on the particular DSC device that is being
used. Consult the device’s user’s manual for more information. This bit is cleared by processor reset.
8.2.1.6 X or P Memory (XP)—Bit 7
The X or P memory (XP) bit is used to select the memory space—program or data—from which
instructions are fetched. In most cases, this bit is cleared and instructions are fetched from program
memory. On devices that support execution from both memory spaces, this bit can be set so that
instructions are fetched from data memory. Refer to Section 8.6, “Executing Programs from Data
Memory,” for more information on executing programs from data memory. This bit is cleared by processor
reset.
8-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Programming Model
8.2.1.7 Condition Code Mode (CM)—Bit 8
The condition code mode (CM) bit selects whether condition codes are calculated with 36-bit or 32-bit data
ALU results. When this bit is set, the C, N, V, and Z condition codes are calculated based on 32-bit results.
When this bit is cleared, these condition codes are generated based on 36-bit results. See Section B.1.3,
“Condition Code Mode,” on page B-3 for a more detailed description of the effect of the CM bit on the
condition codes. This bit is cleared by processor reset.
In general, programs should not set the CM bit unless it is required for compatibility with the DSP56800
architecture. The DSP56800E and DSP56800EX instruction set contains test and compare instructions for
byte, word, long-word, and 36-bit values in the accumulators, obviating the need for the CM bit
functionality.
NOTE:
The CM bit on the DSP56800E and DSP56800EX architecture is identical
in function to the DSP56800’s CC bit. The bit has been renamed for the
DSP56800E and DSP56800EX in the interest of clarity.
8.2.1.8 Nested Looping (NL)—Bit 15
The nested looping (NL) bit reflects the status of hardware DO loops and the hardware stack. If this bit is
set, then the program is currently executing a DO loop that is nested inside another DO loop. If this bit is
clear, a nested DO loop is not being executed. This bit is used by the looping hardware to correctly save
and restore the contents of the hardware stack. REP looping does not affect this bit.
The NL bit is also affected by any direct accesses to the hardware stack register. See Section 8.4,
“Hardware Stack,” for a more detailed discussion. The NL bit is cleared on processor reset.
8.2.2 Status Register
The status register (SR) is a 16-bit register that consists of an 8-bit mode register (MR) and an 8-bit
condition code register (CCR). MR occupies the high-order 8 bits of the SR; CCR occupies the low-order
8 bits.
The mode register reflects and defines the operating state of the DSC core, including the current interrupt
priority level. The condition code register reflects various properties of the values that result from
instruction execution.
Freescale Semiconductor
Program Controller
8-7
Program Controller
SR
Status Register
BIT 15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
BIT 0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
TYPE
rw
r
r
r
r
r
rw
rw
rw
rw
rw
rw
rw
rw
rw
rw
RESET
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
Table 8-2. SR Bit Descriptions
Name
LF
Bit 15
Description
Settings
Loop Flag—Indicates whether a program loop is
active or whether HWS has been written to at
least once without being read
0 = No DO loop active.
1 = DO loop active.
P4–P0
Bits 14–10
Program Counter Extension—Bits 20–16 of
the program counter
Dependent on execution.
I1–I0
Bits 9–8
Interrupt Mask—Masks or enables the four
interrupt levels
00
01
10
11
SZ
Bit 7
Size—Indicates growth beyond a certain point in
the size of an accumulator value
0 = Accumulator value is small.
1 = Accumulator value is large.
L
Bit 6
Limit—Indicates whether data limiting has been
performed since this bit was last cleared
0 = No limiting performed.
1 = Limiting has been performed.
E
Bit 5
Extension in Use—Indicates whether an accumulator extension register is in use
0 = Extension not in use.
1 = Extension in use.
U
Bit 4
Unnormalized—Shows whether a result value
is normalized or not
0 = Normalized.
1 = Not normalized.
N
Bit 3
Negative—Indicates whether result of last operation was negative or positive
0 = Result was positive.
1 = Result was negative.
Z
Bit 2
Zero—Indicates whether result of last operation
was zero or not
0 = Result was non-zero.
1 = Result was zero.
V
Bit 1
Overflow—Indicates whether result of last operation overflowed its destination
0 = Result did not overflow.
1 = Result overflowed destination.
C
Bit 0
Carry—Set if a carry out or borrow was generated in addition or subtraction
0 = No carry occurred during operation.
1 = Carry out occurred during operation.
Note:
See Section 8.4, “Hardware Stack.”
= Allow all interrupts.
= Mask level 0.
= Mask levels 0 and 1.
= Mask levels 0, 1, and 2.
Bits in the CCR portion of the status register are affected by data ALU operations, AGU arithmetic
instructions, bit-manipulation instructions, and so forth. Bits in the MR are affected by processor reset,
exception processing, flow control instructions, and many others. During processor reset, all CCR bits are
cleared, the interrupt mask bits in the MR are both set, and the LF bit is cleared. The program extension bit
values depend on the value of the reset vector.
8-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Programming Model
A description of each of the bits in the status register appears in the following subsections. The
descriptions that are given for the CCR bits are the standard definitions, but these bits may be set or cleared
slightly differently depending on the instruction that is being executed. More information on the condition
code bits is found in Section 5.7, “Condition Code Calculation,” on page 5-38 and in Appendix B,
“Condition Code Calculation.”
NOTE:
When individual bits in the SR are modified, the BFCLR, BFCHG, or
BFSET instructions should be used instead of a MOVE instruction to
prevent the accidental modification of other bits.
8.2.2.1 Carry (C)—Bit 0
The carry (C) bit is used to reflect a variety of conditions. It is set under the following circumstances:
•
If an addition operation results in a carry out of the MSB of the result
•
If a borrow was necessary when a subtraction operation was performed
•
When all bits specified by the mask are set (or cleared, depending on the instruction) in their
corresponding operand for bit-manipulation instructions
•
When the last bit that is to be shifted or rotated out of the MSB or LSB of an operand in a shift or
rotate operation is a one
When not set under one of these conditions, this bit is always cleared.
8.2.2.2 Overflow (V)—Bit 1
The overflow (V) bit is set if the result of an arithmetic operation overflows (is too large to fit in) the size
of the specified destination. If overflow does not occur, this bit is always cleared.
8.2.2.3 Zero (Z)—Bit 2
The zero (Z) bit is set if the result of an operation is equal to zero. If the result is non-zero, this bit is
cleared.
8.2.2.4 Negative (N)—Bit 3
The negative (N) bit is set if the result of an operation is negative. A value is considered negative if the
MSB is set; otherwise it is considered positive. If the MSB of the result is not set, this bit is cleared.
8.2.2.5 Unnormalized (U)—Bit 4
The unnormalized (U) bit is set if the value resulting from an operation is not normalized. A value is
considered normalized if all bits to the right of the binary point are significant. For an accumulator result,
this condition means that bits 31 and 30 of the result should be different. Thus, the U bit is computed as
follows:
U = (Bit 31 ⊕ Bit 30)
Normalized values have the property that, for a positive number p, the relation 0.5 < p < 1.0 is satisfied; for
a negative value n, the relation is –1.0 < n < –0.5.
This bit is not affected by the OMR’s CM bit.
Freescale Semiconductor
Program Controller
8-9
Program Controller
8.2.2.6 Extension in Use (E)—Bit 5
The extension in use (E) bit is cleared if the high-order 5 bits (bits 35–31) of a 36-bit result are the same
(00000 or 11111). Otherwise, this bit is set.
When the high-order 5 bits all contain the same value, the extension portion of an accumulator (bits 35–32)
just holds sign extension and can be ignored. When they are not all the same, the bits in the extension
register are significant and must be considered when additional computations are performed or when the
accumulator is written to memory.
This bit is not affected by the OMR’s CM bit.
8.2.2.7 Limit (L)—Bit 6
The limit (L) bit is a latching bit (sticky bit) that is set if the overflow bit is set or if the data limiters
perform a limiting operation. It is not affected otherwise. The L bit is cleared only by a processor reset or
by an instruction that specifically clears it.
8.2.2.8 Size (SZ)—Bit 7
The size (SZ) bit is a latching bit (sticky bit) that indicates that word growth is occurring in an algorithm.
The bit is set when a 36-bit accumulator is moved to data memory and bits 30 and 29 of the source
accumulator are not the same. The setting of the SZ bit occurs via the following computation:
SZ = SZ | (Bit 30 ⊕ Bit 29)
This bit is especially useful for attaining maximum accuracy when a block-floating-point fast Fourier
transform (FFT) is performed. See the application note Implementation of Fast Fourier Transforms on
Freescale’s Digital Signal Processors (document order number APR4/D) for information on
implementing FFT algorithms on the DSC core.
The SZ bit is cleared only by a processor reset or by an instruction that specifically clears it.
8.2.2.9 Interrupt Mask (I0–I1)—Bits 8–9
The interrupt mask (I1 and I0) bits set the interrupt priority level (IPL) that is needed for an interrupt
source to interrupt the processor. The current priority level of the processor may be changed under
software control. Both interrupt mask bits are set to one during processor reset. Table 8-3 shows the
exceptions that are permitted and masked for the various settings of I1 and I0.
Table 8-3. Interrupt Mask Bits Settings
I1
I0
Exceptions Permitted
Exceptions Masked
0
0
IPL 0, 1, 2, 3, LP
None
0
1
IPL 1, 2, 3
IPL 0
1
0
IPL 2, 3
IPL 0, 1
1
1
IPL 3
IPL 0, 1, 2
Exception processing is explained in detail in Section 9.3, “Exception Processing State,” on page 9-2.
8-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Programming Model
8.2.2.10 Program Counter Extension (P0–P4)—Bits 10–14
The program extension (P4–P0) bits form bits 20 through 16 of the program counter. P4 corresponds to the
MSB of the 21-bit program address, and P0 corresponds to bit 16. Bits 15–0 of the program counter are
found in the PC register.
The program extension bits are stacked by the JSR and BSR instructions for subroutines and interrupts
because the complete status register is pushed by these instructions. They are restored from the stack when
an RTS, RTSD, RTI, or RTID instruction is executed.
NOTE:
Because these bits represent part of the program counter, they cannot be
directly modified. Instructions that change the value of the status register
do not affect these bits.
NOTE:
The values read (from reading the SR) are not guaranteed to be valid.
8.2.2.11 Loop Flag (LF)—Bit 15
The loop flag (LF) bit is set when a hardware (DO or DOSLC) loop is initiated or when a value is written
under program control to the hardware stack. Reading the hardware stack or terminating a DO or DOSLC
loop causes LF to be set to the value in the OMR’s NL bit. See Section 8.2.1.8, “Nested Looping
(NL)—Bit 15.”
REP looping does not affect this bit. The LF bit is cleared during processor reset.
NOTE:
This bit should never be explicitly cleared by a move or bitfield instruction
when the NL bit in the OMR register is set.
See Section 8.4, “Hardware Stack,” for more information on how accesses to the hardware stack affect the
value in LF.
8.2.3 Loop Count Register
The loop count register (LC) is a special 16-bit counter that specifies the number of times to repeat a
hardware loop (one that is begun with a DO, DOSLC, or REP instruction). When the last instruction in a
hardware program loop is reached, the contents of the loop counter register are tested. If the loop counter is
one, the program loop is terminated. If the loop counter is not one, it is decremented by one and the
program loop is repeated.
The loop count register can be read and written under program control. This capability gives software
programs access to the value of the current loop iteration. The LC register is also updated with the contents
of the LC2 register when a loop is exited. See Section 8.5, “Hardware Looping,” for a full discussion of
hardware looping.
8.2.4 Loop Count Register 2
The loop count register 2 (LC2) is a 16-bit register that is used to save the value that is in LC whenever LC
is modified, as when a nested hardware loop is begun. The contents of LC are copied to LC2 whenever a
DO instruction is executed or when an instruction is executed that explicitly modifies the LC register. This
arrangement ensures that LC is backed up properly when LC is loaded under program control, such as
Freescale Semiconductor
Program Controller
8-11
Program Controller
when LC is loaded with a loop count before DOSLC is executed. When a DO or DOSLC loop terminates,
the value in the LC2 register is copied back into the LC register when the OMR's NL bit is set. See
Section 8.5, “Hardware Looping,” for a full discussion hardware looping.
LC2 may be pushed onto or popped from the software stack under program control. This capability allows
an application to save and restore this register when necessary.
8.2.5 Loop Address Register
The loop address (LA) register holds the location of the last instruction word in a hardware DO loop, and it
is used by the looping hardware to determine when the end of a loop has been reached.
The value in the LA register is set when the DO instruction is executed, and it may also be updated when a
DO loop that is nested in another DO loop is exited, at which point the contents of LA2 are copied to it.
The LA register can be read or written using a MOVE instruction. When the register is read as a 32-bit
long with a MOVE.L instruction, the upper 8 bits of the destination are zero extended. When it is written as
a 32-bit long by a MOVE.L instruction, only the lower 24 bits are stored in LA.
8.2.6 Loop Address Register 2
The loop address 2 register (LA2) is a 24-bit register that is used to save the value of LA when a DO loop
that is nested within another DO loop is executed. When a DO or DOSLC instruction is executed, the
contents of LA are copied to LA2 before the end-of-loop address for the inner loop is stored in LA. When
the nested loop terminates, the value in LA2 is copied back to LA to allow the outer loop to continue. See
Section 8.5, “Hardware Looping,” for more information on nested hardware loops.
LA2 may be read from and written to the stack under program control. This capability allows an
application to save and restore this register when necessary.
8.2.7 Hardware Stack Register
The hardware stack register (HWS) is used to manipulate the program controller’s hardware stack under
program control. Accesses to HWS always read or write the value on the top of the stack; the second stack
location is not directly accessible. Reading from or writing to HWS can affect the LF bit in the status
register and the NL bit in the operating mode register. See Section 8.4, “Hardware Stack,” for more
information.
The HWS register is accessed with standard MOVE instructions. When the register is read as a 32-bit long
by a MOVE.L instruction, the upper 8 bits of the destination register are zero extended. When it is written
as a 32-bit long by a MOVE.L instruction, only the lower 24 bits are stored on the hardware stack.
8.2.8 Fast Interrupt Status Register
The fast interrupt status register (FISR) is a 13-bit register that is used to hold the state of the DSC core
during fast interrupt processing. Critical bits in the status register (SR) and operating mode register
(OMR), as well as the alignment of the stack pointer, are copied into the FISR at the beginning of fast
interrupt processing. The value in the FISR is used to restore the core state when a fast interrupt processing
routine is exited.
8-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Program Controller Programming Model
The FISR holds copies of the status register’s LF, I1, I0, SZ, L, E, U, N, Z, V, and C bits as well as the
operating mode register’s NL bit. The SPL bit holds a copy of the LSB of the stack pointer (SP), which
allows the stack pointer to be restored to its original value after interrupt processing is complete. See
Section 9.3.2.2, “Fast Interrupt Processing,” on page 9-6 for more information on fast interrupt processing
and on the use of the FISR register. This register is not affected by processor reset.
FISR
Fast Interrupt Status Register
BIT 15
14
13
TYPE
12
11
10
9
8
7
6
5
4
3
2
1
BIT 0
SPL
LF
NL
I1
I0
SZ
L
E
U
N
Z
V
C
rw
rw
rw
rw
rw
rw
rw
rw
rw
rw
rw
rw
rw
Table 8-4. FISR Bit Descriptions
Name
Description
Settings
Undefined
Bits 15–13
Undefined
These bits are undefined and should be ignored.
SPL
Bit 12
Stack Pointer LSB—Contains a copy of the
LSB of the SP register
Value in stack pointer on interrupt.
LF
Bit 11
Loop Flag—Contains a copy of the LF bit in the
status register
Value in status register on interrupt.
NL
Bit 10
Nested Looping—Contains a copy of the NL bit
in the operating mode register
Value in operating mode register on interrupt.
I1–I0
Bits 9–8
Interrupt Mask—Contains a copy of the I1 and
I0 bits in the status register
Value in status register on interrupt.
SZ
Bit 7
Size—Contains a copy of the SZ bit in the status
register
Value in status register on interrupt.
L
Bit 6
Limit—Contains a copy of the L bit in the status
register
Value in status register on interrupt.
E
Bit 5
Extension in Use—Contains a copy of the E bit
in the status register
Value in status register on interrupt.
U
Bit 4
Unnormalized—Contains a copy of the U bit in
the status register
Value in status register on interrupt.
N
Bit 3
Negative—Contains a copy of the N bit in the
status register
Value in status register on interrupt.
Z
Bit 2
Zero—Contains a copy of the Z bit in the status
register
Value in status register on interrupt.
V
Bit 1
Overflow—Contains a copy of the V bit in the
status register
Value in status register on interrupt.
C
Bit 0
Carry—Contains a copy of the C bit in the status
register
Value in status register on interrupt.
Freescale Semiconductor
Program Controller
8-13
Program Controller
8.2.9 Fast Interrupt Return Address
The fast interrupt return address (FIRA) is a 21-bit register that holds a copy of the program counter when
fast interrupt processing is initiated. This address is used to return control to the interrupted program when
the fast interrupt service routine is complete.
This register is not affected by processor reset.
8.3 Software Stack
The software stack is a last-in-first-out (LIFO) stack of arbitrary depth that is located in data memory. Any
instruction that accesses data memory can be used to access locations on the stack, although typically
accesses are made using the stack pointer register (SP).
The JSR and BSR instructions use the software stack for saving the program counter and status register
when a subroutine or interrupt service routine is called. The stack can also be used for passing parameters
to subroutines, for creating variables that are local to a subroutine, or for any other temporary-storage
needs.
The stack pointer value is undefined after reset, and it must be set in software before the stack can be used.
The initial value for the stack pointer is the lower boundary of the stack—the software stack on the core
grows up as values are pushed onto it.
NOTE:
Be careful when initializing the stack pointer to set aside enough space for
the stack. If the address space used by the stack overlaps other data areas,
erratic behavior may result. For maximum performance, the software stack
should be located in on-chip memory.
8.3.1 Pushing and Popping Values
Because the stack grows up in memory, and because the SP register always points at the item that is on the
top of the stack, the stack pointer must be pre-incremented when values are pushed on the stack. This
process involves two instructions, as shown in Example 8-1.
Example 8-1. Pushing a Value on the Software Stack
; Placing One value onto the software stack
; Performed in 2 cycles, 2 instruction words
ADDA #2,SP
; Increment the SP (1 cycle, 1 Word)
MOVE.L A10,X:(SP)
; Place value onto the stack
For pushing multiple values to the stack, there is a more efficient technique in terms of both time and
space. Instead of repeating the two-instruction sequence for each value to be stored, implement the push
operations that are shown in Example 8-2.
Example 8-2. Pushing Multiple Values on the Software Stack
; Faster technique for pushing four values onto the software stack
; Finishes in 5 cycles, 5 instruction words
ADDA #2,SP
; Increment the SP (1 cycle, 1 Word)
MOVE.L A10,X:(SP)+
MOVE.L B10,X:(SP)+
MOVE.L R0,X:(SP)+
MOVE.L R1,X:(SP)
; <== No post-increment SP on last MOVE
8-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Software Stack
Popping values from the software stack is fairly straightforward. With the use of the post-decrement
addressing mode, values can be popped from the stack in a single instruction. To pop the four values that
are saved on the stack in Example 8-2 on page 8-14, the code in Example 8-3 can be executed.
Example 8-3. Popping Values from the Software Stack
; Popping four values from the software stack
; Finishes in 4 cycles, 4 instruction words
MOVE.L X:(SP)-,R1
MOVE.L X:(SP)-,R0
MOVE.L X:(SP)-,B10
MOVE.L X:(SP)-,A10
; SP left pointing at previous top of stack
8.3.2 Subroutines
The JSR and BSR instructions are used to call subroutines. When a JSR or BSR is executed, the return
address (the value in the program counter) is pushed onto the stack. Because the high-order 5 bits of the
program counter are contained in the status register, the return address is saved by pushing both the PC and
the SR, in that order, onto the stack. Figure 8-3 shows the software stack after a JSR has been executed.
Data Memory
SP
Status Register (Contains P4–P0)
Return Address (16 LSBs)
Figure 8-3. Effects of the JSR Instruction on the Stack
The RTS and RTSD instructions pop the PC and SR off the stack when a subroutine is exited. Only the
P4–P0 bits are actually updated in the SR; the remaining bits are discarded.
8.3.3 Interrupt Service Routines
Entries in the DSC core interrupt and exception vector table frequently consist of a JSR instruction, with a
service routine target address as its argument. When an exception occurs, the program counter is moved to
the address of the appropriate entry in the vector table. If there is a JSR instruction at that location, it is
fetched and executed in the same way that a JSR would normally be executed.
The JSR instruction stacks the program counter (the return address from the interrupted program) and
status register, as shown in Figure 8-3. When the interrupt service routine is complete, an RTI or RTID
instruction is executed. Like the RTS and RTSD instructions, these instructions pop the program counter
and status register from the stack. Unlike RTS and RTSD, the RTI and RTID instructions do not discard
the contents of the stored status register, but use them to restore the status bits in SR. This restoration
ensures that the processor state is not changed by the actions of the interrupt service routine.
Freescale Semiconductor
Program Controller
8-15
Program Controller
Note that if the fast interrupt processing method is used to handle an interrupt, the process is quite
different, and it does not involve a JSR to an interrupt service routine. For more information on both types
of interrupt processing, see Section 9.3.2, “Interrupt and Exception Processing,” on page 9-4.
8.3.4 Parameter Passing and Local Variables
The software stack supports structured programming techniques, such as parameter passing to subroutines
and local variables. These techniques can be used for both assembly language programming as well as
high-level language compilers.
Parameters can be passed to a subroutine by placing these variables on the software stack immediately
before a JSR to the subroutine is performed. Placing these variables on the stack is referred to as building a
“stack frame.” These passed parameters can then be accessed in the called subroutine with the use of
SP-relative addressing modes. This process is demonstrated in Example 8-4.
Variables that are local to a subroutine can also be conveniently allocated on the stack. Stack locations that
are above the status register and return address can be set aside for local variables by incrementing the
stack pointer the required number of words. Local variables can then be accessed relative to the stack
pointer, as subroutine parameters are. Example 8-4 also illustrates the creation and use of local variables
on the stack.
Example 8-4. Subroutine Call with Passed Parameters
ADDA #1,SP
MOVE.W X:$35,X0
MOVE.W X0,X:(SP)+
MOVE.W X:$21,X0
MOVE.W X0,X:(SP)
JSR
ROUTINE1
SUBA #2,SP
;
;
;
;
;
;
;
;
(pre-increment before pushing two variables)
Pointer variable to be passed to subroutine
(push onto stack)
2nd variable to be passed to subroutine
(push onto stack)
*** Execute Subroutine ***
Remove the two passed parameters from
stack when done
ROUTINE1
;
;
ADDA #4,SP
; Allocate room for local variables
(instructions)
MOVEU.WX:(SP-7),R0 ; Get pointer variable
MOVE.W X:(SP-6),B
; Get 2nd variable
MOVE.W X:(R0),X0
; Get data pointed to by pointer variable
ADD
X0,B
MOVE.W B,X:(SP-6)
; Store sum in 2nd variable
(other instructions...)
SUBA #4,SP
RTS
The stack frame created by the code in Example 8-4 is shown in Figure 8-4 on page 8-17.
8-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Hardware Stack
X Data Memory
SP
4th Local Variable
3rd Local Variable
2nd Local Variable
1st Local Variable
Status Register
Return Address
2nd Passed Parameter
1st Passed Parameter
Figure 8-4. Example Stack Frame
Before a subroutine is exited, be careful to de-allocate space that is reserved on the stack for local
variables. The stack pointer should be decremented so that it points to the saved status register before the
RTS instruction is executed, so the correct return address is popped from the stack.
8.4 Hardware Stack
The hardware stack is a last-in-first-out (LIFO) stack that consists of two 24-bit internal registers.
Although there are two locations on the stack, the stack is always accessed through the hardware stack
register (HWS). Reads or writes to the HWS access or modify the top location in the stack.
The hardware stack is updated when a hardware DO loop is entered or exited. Executing a DO or DOSLC
instruction (or a write to HWS) pushes the address of the first instruction in the loop onto the stack. When
the loop terminates, the address is popped off the stack. The hardware stack can also be manipulated under
program control with the use of standard MOVE instructions.
When a value is written to HWS, either through a MOVE instruction or by the DO and DOSLC
instructions saving the looping state, the following occur:
1. The SR’s LF bit is copied to the OMR’s NL bit, overwriting the previous NL value.
2. The value in the first HWS location (HWS0) is copied to the second (HWS1), overwriting
the previous value.
3. The LF bit in the status register is set.
4. The appropriate value is written to the top hardware stack register.
Reading a value from HWS does the following:
1. Copies the OMR’s NL bit to the SR’s LF bit, overwriting the previous LF value
2. Copies the value in the second hardware stack register to the first, or top, register
3. Clears the OMR’s NL bit
Freescale Semiconductor
Program Controller
8-17
Program Controller
The state of the NL and LF bits can be used to determine the status of program looping and thus of the
hardware stack, as shown in Table 8-5. To ensure the integrity of the hardware stack values, make certain
that a program never puts the processor in the illegal state that this table specifies. Avoid this illegal state
by ensuring that the LF bit is never explicitly cleared when the NL bit is set.
Table 8-5. Hardware Stack Status
NL
LF
DO Loop Status
# Words of Hardware Stack
0
0
No DO loops active
0
0
1
Single DO loop active
1
1
0
(Illegal)
—
1
1
Two DO loops active
2
If both the NL and LF bits are set (that is, two DO loops are active) and a DO or DOSLC instruction (or a
write to HWS) is executed, a hardware stack overflow interrupt occurs because there is no more space on
the hardware stack to support a third DO loop. There is no interrupt on hardware stack underflow.
8.5 Hardware Looping
Loops are one of the most common software constructs, especially in DSC algorithms. In order to speed up
these critical algorithms, the core includes special hardware to accelerate loops. Two types of
hardware-accelerated loops are supported: fast repetition of a single instruction a specified number of
times, using the REP instruction; and more traditional multi-instruction loops, using the DO and DOSLC
instructions.
8.5.1 Repeat (REP) Looping
Repeat looping, using the REP instruction, executes a single 1-word instruction a number of times. The
number of times the instruction should be repeated is specified by the parameter to the REP instruction,
which is either a 6-bit immediate or 16-bit register value. The instruction that is to be repeated is the one
that immediately follows REP.
Example 8-5 demonstrates repeat looping on the move instruction. In this example, 64 words are cleared in
data memory, 2 words at a time.
Example 8-5. Repeat Loop Example
MOVE.W #0,A
REP
#32
MOVE.L A10,X:(R0)+
; Clear the A Accumulator
; Set up hardware repeat of the following instruction
; Clear 2 words in memory
The instruction that is to be repeated (MOVE.L in this case) is fetched only once from program memory.
Until the repeat loop is complete, the program counter is frozen and interrupts are disabled. If a repeat loop
must be interruptible, a DO loop should be used instead. See Section 8.5.2, “DO Looping.”
The repeat count that is specified in the REP instruction must be a positive value. If the count specified is
zero, the instruction following REP is skipped, and execution continues with the subsequent instruction.
8-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Hardware Looping
The REP instruction can only be used to repeatedly execute single-word instructions. Repeat looping
cannot be used on:
•
An instruction that is more than 1 program word in length.
•
An instruction that accesses program memory.
•
A REP or ENDDO instruction.
•
Any instruction that changes program flow.
•
A SWI, SWI #x, SWILP, DEBUGEV, DEBUGHLT, WAIT, or STOP instruction.
•
A Tcc, SWAP SHADOWS, or ALIGNSP instruction.
8.5.2 DO Looping
The DO instruction performs hardware looping on a single instruction or a block of instructions. DO loops
can be nested up to two deep, accelerating more complex algorithms. Unlike REP loops, loops initiated
with DO are interruptible.
Hardware DO looping (DO or DOSLC) executes a block of instructions for the number of specified times.
For a DO instruction, the loop count is specified with a 6-bit unsigned value or 16-bit register value. The
DOSLC instruction works identically to DO, but assumes that the loop count has already been placed in
the LC register.
Example 8-6 demonstrates hardware DO looping on a block of two instructions. This example copies a
block of forty 32-bit memory locations from one area of memory to another.
Example 8-6. DO Loop Example
DO
#40,END_CPY
MOVE.L X:(R0)+,A
MOVE.L A10,X:(R1)+
; Set up hardware DO loop
; Copy a 32-bit memory location
;
END_CPY
When a hardware loop is initiated with a DO or DOSLC instruction, the following events occur:
1. When the DO instruction is executed, the contents of the LC register are copied to the LC2
register, and LC is loaded with the loop count that the instruction specifies. The DOSLC
instruction does not modify the LC and LC2 registers.
2. The old contents of the LA register are copied to the LA2 register, and the LA register is
loaded with the address of the last instruction word in the loop. If a 16-bit address is
specified, the upper 8 bits of LA are cleared.
3. The address of the first instruction in the program loop (top-of-loop address) is pushed onto
the hardware stack. This push sets the LF bit and updates the NL bit, as occurs with any
hardware stack push.
Instructions in the loop are then executed. The address of each instruction is compared to the value in LA
to see if it is the last instruction in the loop. When the end of the loop is reached, the loop count register is
checked to see if the loop should be repeated. If the value in LC is greater than one, LC is decremented and
the loop is re-started from the top. If LC is equal to one, the loop has been executed for the proper number
of times and should be exited.
When a hardware loop ends, the hardware stack is popped (and the popped value is discarded), the LA2
register is copied to LA, the LC2 register is copied to LC, and the NL bit in the operating mode register is
Freescale Semiconductor
Program Controller
8-19
Program Controller
copied to the LF bit. The OMR's NL bit is then cleared. Instruction execution then continues at the address
that immediately follows the end-of-loop address.
One hardware stack location is used for each nested DO or DOSLC loop. Thus, a two-deep hardware stack
allows for a maximum of two nested loops. The REP instruction does not use the hardware stack, so repeat
loops can be nested within DO loops.
8.5.3 Specifying a Loop Count of Zero
If a loop count of zero is specified for the DO instruction, or if a zero or negative loop count is specified for
DOSLC, the instructions in the body of the loop are skipped, and execution continues with the instruction
immediately following the loop body. An example of this process appears in Example 8-7.
Example 8-7. DO Loop Special Case
MOVE.W #0,X0
.
.
.
DO
X0,END_CPY
MOVE.L X:(R0)+,A
MOVE.L A10,X:(R1)+
; Loop count is zero upon entry
; Copy a 32-bit memory location
;
END_CPY
Note that an immediate loop count of zero (for the DO instruction) is not allowed and will be rejected by
the assembler. A loop count of zero can only be specified by using a register that is loaded with zero as the
argument to the DO instruction, or by placing a zero in the LC register and executing DOSLC.
8.5.4 Terminating a DO Loop
A DO loop normally terminates when the body of the loop has been executed for the specified number of
times (the end of the loop has been reached, and LC is one). Alternately, a DO loop terminates if the count
specified is zero. Similarly, if the LC register is zero or negative, a DOSLC loop will also terminate, which
causes the body of the loop to be skipped entirely.
When the inner loop of a nested loop terminates naturally, the LA2 and LC2 registers are copied into the
LA and LC registers, respectively, restoring these two registers with their values for the outer loop. A loop
is determined to be a nested inner loop if the OMR’s NL bit is set. If the NL bit is not set, the LA and LC
registers are not modified when a loop is terminated or skipped.
If it is necessary to terminate a DO loop early, use one of the techniques discussed in Section 8.5.4.1,
“Allowing Current Block to Finish and Then Exiting,” and Section 8.5.4.2, “Immediate Exit from a
Hardware Loop.”
8.5.4.1 Allowing Current Block to Finish and Then Exiting
One method for terminating a DO loop is to modify the loop counter register so that the remainder of the
instructions in the loop are executed, but so that the loop does not return to the top of the loop. This
modification can be accomplished through explicitly setting the value in LC to one:
MOVEU.W #1,LC
Because the loop is allowed to complete, the hardware stack will be popped, and the internal looping state
will be reset correctly.
8-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Hardware Looping
This technique should not be used to terminate a loop that is nested within another loop. A nested DO loop
can be terminated by using the ENDDO instruction (see Section 8.5.4.2, “Immediate Exit from a Hardware
Loop,” for the correct usage of this instruction). Writing a value to LC causes the previous value in LC to
be copied to LC2, thus destroying the outer loop’s count.
NOTE:
There are restrictions on the location of instructions that modify the LC
register with respect to the end of the loop. See the sections concerning DO
and DOSLC in Section A.2, “Instruction Descriptions,” on page A-7.
8.5.4.2 Immediate Exit from a Hardware Loop
When it is necessary to break out of a loop immediately, without executing any more iterations in the loop,
use the ENDDO instruction.
Note that the ENDDO instruction does not cause execution to jump to the end of the loop. ENDDO only
cleans up the hardware stack and the internal loop processing state. A BRA or JMP instruction must be
used to stop the execution of instructions within the body of the loop.
Two examples of code that show how to perform immediate exits appear in Example 8-8.
Example 8-8. Immediate Exit from Hardware Loop
;
;
LABEL
DO
#LoopCount,LABEL
(instructions in loop)
Bcc
EXITLP
;
;
(other instructions in loop (skipped if immediate exit))
BRA
EXITLP
OVER
ENDDO
; additional cycle for BRA for normal loop exit
; 1 additional cycle for ENDDO when exiting
; loop if exit via Bcc
OVER
;
;
------ alternate method ------
;
;
OVER
;
LABEL
DO
#LoopCount,LABEL
(instructions)
Bcc
OVER
; executed each iteration
ENDDO
; executed only for immediate termination
BRA
LABEL
(instructions)
8.5.5 Specifying a Large Immediate Loop Count
The DO instruction allows an immediate value up to 63 to be specified for the loop count. In cases where it
is necessary to specify a value that is larger than 63, the DOSLC instruction should be used. A 16-bit
immediate loop count can be loaded into the LC register before the loop is started. The loop is then
initiated with the DOSLC instruction, which assumes that the count has previously been loaded into LC.
Example 8-9 on page 8-22 demonstrates this technique.
Freescale Semiconductor
Program Controller
8-21
Program Controller
Example 8-9. Using the DOSLC Instruction
MOVEU.W#2048,LC
;
LABEL
NOP
NOP
DOSLC LABEL
(instructions)
;
;
;
;
;
Specify a loop count greater than 63
using the LC register
(delay required due to pipeline)
...
Start loop with count already in LC
Note that a delay of 2 instruction words must be inserted between the instruction that updates LC and the
DOSLC instruction. Each of these words can consist of any instruction, including NOP if no useful
instruction can be placed in the sequence.
8.5.6 Nested Hardware Looping
The DSC core architecture allows one hardware-accelerated DO loop to be nested within another. It is
possible to nest one hardware DO loop within another, or to nest a REP loop within a DO loop or within
two nested DO loops. The following sections describe the nesting of hardware loops.
8.5.6.1 Nesting a REP Loop Within a DO Loop
A hardware repeat loop can be nested within a hardware DO loop without any additional setup or
processing. Example 8-10 demonstrates a repeat loop nested within a DO loop. In this example, the repeat
loop accumulates 8 values and stores the result for 10 different blocks of data.
Example 8-10. Example of a REP Loop Nested Within a DO Loop
MOVE.W X:(R0)+,X0
DO
#10,END_NST
CLR.W A
REP
#8
ADD
X0,A X:(R0)+,X0
MOVE.W A1,X:(R1)+
; (read first value)
; Set up hardware DO loop
; (body of DO loop)
; accumulate eight values
; store result of eight accumulated values
END_NST
Note that the REP instruction does not affect the value of the loop count for the outer DO loop.
8.5.6.2 Nesting a DO Loop Within a DO Loop
Nested looping of DO and DOSLC loops is permitted on the DSC core architecture. The hardware stack,
dual loop count, and dual loop address registers act as a LIFO stack for hardware looping state information.
The loop count, the “top-of-loop” address, and the state of the LF and NL bits are maintained for an outer
loop when a nested hardware loop is executed. Because the hardware stack only contains two locations,
hardware DO and DOSLC loops can only be stacked two deep.
Example 8-11 on page 8-23 demonstrates one hardware loop nested within another.
8-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Executing Programs from Data Memory
Example 8-11. Example of Nested DO Loops
ADDA #1,SP
CLR.W A
; (bump to unoccupied stack location)
DO
DO
INC.W
ASL
;
;
;
;
#4,END_OUTR
#3,END_INNR
A
B
Outer
Inner
(body
(body
loop
loop: saves LC->LC2, LA->LA2
of innermost loop)
of innermost loop)
END_INNR
NOP
; (required by pipeline)
END_OUTR
Note that, due to dependencies in the execution pipeline, the outer and inner loops must not end on the
same instruction. In Example 8-11, a NOP instruction has been placed between the loop end labels to
ensure that they end on different instructions. Any useful instruction could be substituted for the NOP.
8.5.6.3 Nesting a DO Loop Within a Software Loop
If more than two loops need to be nested, one of the loops can always be performed with standard software
looping techniques. Example 8-12 demonstrates a hardware DO loop that is nested in a regular software
loop.
Example 8-12. Example of Nested Looping in Software
MOVEU.W#4,R5
; Load R5 for four outer loop iterations
DO
#4,END_INNR
INC.W A
ASL
B
; Inner DO loop
; (body of innermost loop)
; (body of innermost loop)
DECTSTAR5
BGT
OUTER
; Decrement Outer Loop Counter
; Branch to top of loop
OUTER
END_INNR
As compared to a hardware loop, a software loop involves considerably more looping overhead. Software
loops should only be used when necessary, or in code where execution time is not critical.
8.6 Executing Programs from Data Memory
The core is designed with the ability to execute programs stored in data memory. Although this capability
is not intended for high-throughput DSC applications, it is useful for executing diagnostic and test code on
parts where program memory resides in ROM. Program instructions and interrupt vectors are downloaded
into data memory, where they can be executed later.
When instructions from data memory are executed, the core drives the address of the instruction onto the
XAB2 bus, and the memory places its result on the XDB2 bus. The data on this bus is then internally
transferred to the PDB bus, where the execution units expect to find it. Note that, because the program
address bus (PAB) is only 21 bits wide, only the lower 221 locations in data memory can be accessed in
data-memory execution mode.
Figure 8-5 on page 8-24 shows the memory map in this mode.
Freescale Semiconductor
Program Controller
8-23
Program Controller
16M × 16
$FFFFFF
Data
Memory
Space
$0
Interrupt
Vectors
0
The XP bit in OMR enables this operating mode.
Data Memory
(EX = 0)
$FFFFFF
Locations above $1FFFFF are
not accessible in data-memory
execution mode.
Program
Memory
$1FFFFF
$1FFFFF
External
program
memory
is not
accessible in
this mode.
$00FFFF
$00FF80
On-Chip
Peripherals
External
Data
Internal
program
memory
is not
accessible in
this mode.
$0
On- and off-chip data memory can
hold both data and program
instructions to be executed. The
interrupt vector table can also be
located in these memory spaces.
Memory
On-Chip
Data
Memory
$0
Figure 8-5. Example Data-Memory Execution Mode Memory Map
When reset occurs, the XP bit in the OMR register is cleared. This event places the device back into
normal program-memory execution mode. It is not possible to have the core exit reset and then go straight
into data-memory execution mode.
8-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Executing Programs from Data Memory
8.6.1 Entering Data-Memory Execution Mode
A specific sequence must be followed to switch to executing programs from data memory. To enter
data-memory execution mode, perform the following steps:
1. Download the desired program—including interrupt vectors, interrupt service routines, and
data constants—into data memory.
2. Disable interrupts in the status register (SR).
3. Set the XP bit in the operating mode register (OMR).
4. Jump to the first instruction in data memory.
5. Re-enable interrupts from code in data memory (if desired).
These steps translate into one of two code sequences, which are shown in Example 8-13 and Example 8-14
on page 8-26. Depending on the size of the target address specified in the JMP to instructions in data
memory, a slightly different sequence must be used.
Example 8-13 shows the sequence that must be used when a 19-bit target address is used:
Example 8-13. Entering Data Memory Execution, 19-Bit Target Address
BEGIN_X
EQU
$1000
; Beginning address of program in data memory
ORG
P:
; (indicates code located in program memory)
.
.
.
; Exact Sequence for Steps 3 through 5
BFSET #$0300,SR
; Disable Interrupts
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
BFSET #$0080,OMR
; Enable data memory instruction fetches
NOP
; (wait for mode to switch)
NOP
; (wait for mode to switch)
; NOTE: Must Use Assembler Forcing Operator - Forces 19-bit Address
JMP
>XMEM_TARGET ; Jump to 1st instruction in data memory
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
ORG P:BEGIN_X,X:BEGIN_X; (both must be the same value)
XMEM_TARGET
; Remember to re-enable interrupts
If a 21-bit target address is specified, the code sequence is slightly different. In particular, only a single
NOP instruction must be inserted between the BFSET instruction that sets the XP bit and the JMP
instruction (rather than two), and a different assembler forcing operator is specified in the JMP instruction.
This code sequence is given in Example 8-14 on page 8-26.
Freescale Semiconductor
Program Controller
8-25
Program Controller
Example 8-14. Entering Data Memory Execution, 21-Bit Target Address
BEGIN_X
EQU
$1000
; Beginning address of program in data memory
ORG
P:
; (indicates code located in program memory)
.
.
.
; Exact Sequence for Steps 3 through 5
BFSET #$0300,SR
; Disable Interrupts
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
BFSET #$0080,OMR
; Enable data memory instruction fetches
NOP
; (wait for mode to switch)
; NOTE: Must Use Assembler Forcing Operator -- Forces 21-bit address
JMP
>>XMEM_TARGET; Jump to 1st instruction in data memory
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
ORG P:BEGIN_X,X:BEGIN_X; (both must be the same value)
XMEM_TARGET
; Remember to re-enable interrupts
Choose the location of the first instruction in data memory carefully. The target addresses of the JMP
instructions in Example 8-13 on page 8-25 and Example 8-14, which are located in data memory, must be
known absolute addresses. Labels should not be used unless the technique that is shown in the examples is
employed. This technique defines the target code address as the same absolute address in both program and
data memory, which causes the assembler to generate the correct JMP target address.
NOTE:
The code that is used to enter data-memory execution mode must contain
the exact number of NOP instructions that is shown in Example 8-13 on
page 8-25 or Example 8-14. There can be no jumps or branches to
instructions within this sequence.
8.6.2 Exiting Data-Memory Execution Mode
When executing instructions from data memory is no longer required, and when it is necessary to begin
executing instructions from the program memory space, the following sequence of operations must be
performed:
1. Disable interrupts in the status register.
2. Clear the XP bit in the operating mode register.
3. Jump to the return location in the program memory space.
4. Re-enable interrupts from code that is located in program memory space.
Either the code sequence given in Example 8-15 on page 8-27 or the one in Example 8-16 on page 8-27
must be used for exiting data-memory execution mode. The sequence that is used depends on the size of
the target address specified by the JMP instruction. Because of the nature of this operation, it is very
important that the instruction segment between setting the XP bit (or clearing it) and the JMP instruction
should not be single stepped.
8-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Executing Programs from Data Memory
Example 8-15. Exiting Data-Memory Execution Mode, 19-Bit Target Address
ORG P:BEGIN_X,X:BEGIN_X; (code located in data memory)
.
.
.
; Exact Sequence for Steps 1 through 3
BFSET #$0300,SR
; Disable interrupts
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
BFCLR #$0080,OMR
; Disable data memory instruction fetches
NOP
; (wait for mode to switch)
NOP
; (wait for mode to switch)
; NOTE: Must Use Assembler Forcing Operator -- Forces 19-bit address
JMP
>PMEM_TARGET ; Jump to 1st instruction in program memory
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
ORG
.
.
.
P:
; (indicates code located in prgm mem)
PMEM_TARGET
; Remember to re-enable interrupts
If a 21-bit target address must be specified for the JMP instruction, the code sequence in Example 8-16
must be used.
Example 8-16. Exiting Data-Memory Execution Mode, 21-Bit Target Address
ORG P:BEGIN_X,X:BEGIN_X; (code located in data memory)
.
.
.
; Exact Sequence for Steps 1 through 3
BFSET #$0300,SR
; Disable interrupts
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
NOP
; (wait for interrupts to be disabled)
BFCLR #$0080,OMR
; Disable data memory instruction fetches
NOP
; (wait for mode to switch)
; NOTE: Must Use Assembler Forcing Operator - Forces 21-bit address
JMP
>>PMEM_TARGET; Jump to 1st instruction in program memory
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
NOP
; (fetched but not executed)
ORG
.
.
.
P:
; (indicates code located in program memory)
PMEM_TARGET
; Remember to re-enable interrupts
The rules for determining the target address of the JMP instruction that are discussed in Section 8.6.1,
“Entering Data-Memory Execution Mode,” also apply when exiting data-memory execution.
NOTE:
Freescale Semiconductor
Program Controller
8-27
Program Controller
The code that is used to exit data-memory execution mode must contain
the exact number of NOP instructions that is shown in Example 8-15 or
Example 8-16 on page 8-27. There can be no jumps or branches to
instructions within this sequence.
8.6.3 Interrupts in Data-Memory Execution Mode
Regular interrupt processing is supported in data-memory execution mode. The interrupt vector table and
all interrupt service routines must be copied to data memory because program memory is completely
disabled when data-memory execution mode is active. It is only necessary to provide interrupt vectors and
service routines for interrupts that will actually occur during data memory execution.
During the transition in and out of data-memory execution mode, interrupts must be disabled.
8.6.4 Restrictions on Data-Memory Execution Mode
The following restrictions apply when programs are executed from data memory:
•
Instructions that perform two reads from data memory are not permitted.
•
Instructions that access program memory are not permitted.
•
Interrupts must be disabled when data-memory execution mode is entered or exited.
Instructions that perform one parallel move operation are allowed in this mode.
8-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 9
Processing States
The DSP56800E core has six processing states, and it is always in one of these states. The states reflect the
variety of operating modes that are available to a DSP56800E device, which include low-power and debug
capabilities. The processing states are:
•
Normal—the normal instruction execution state.
•
Reset—the state where the core is forced into a known reset state. The first program instruction is
fetched upon exiting this state.
•
Exception—the interrupt processing state, where the core transfers program control from its current
location to an interrupt service routine using the interrupt vector table.
•
Wait—a low-power state where the core is shut down but the peripherals and interrupts remain
active.
•
Stop—a low-power state where the core, interrupts, and selected peripherals are shut down.
•
Debug—a debugging state where the core is halted and the Enhanced On-Chip Emulation
(Enhanced OnCE) module is enabled and used for debug activity.
These processing states are available when programs are executed normally from program memory and
when instructions are fetched from data memory (see Section 8.6, “Executing Programs from Data
Memory,” on page 8-23). Each of these processing states is considered in the following pages.
9.1 Normal Processing State
The normal processing state is the typical state of the processor, where it performs normal instruction
execution. The core enters the normal processing state after reset, if debugging is not active.
Additional information on the normal processing state can be found in Section 10.2, “Normal Pipeline
Operation,” on page 10-3.
9.2 Reset Processing State
The processor enters the reset processing state when a hardware reset signal is asserted. The core is held in
reset during power up through the assertion of the RESET terminal, making this the first processing state
entered by the DSC.
The reset processing state takes precedence over all other processing states. When the reset terminal to the
core is asserted, the core exits the processing state it was in previously and immediately enters the reset
processing state.
Freescale Semiconductor
Processing States
9-1
Processing States
On devices with a computer operating properly (COP) timer, it is also possible for the COP timer to assert
the RESET signal if the timer reaches zero, forcing the core into the reset processing state.
The DSP56800E core remains in the reset processing state until the cause for reset is de-asserted. When the
reset trigger is deasserted, the following occurs:
1. The internal registers are set to their reset state:
— The modifier register (M01) is set to $FFFF.
— The status register’s (SR) loop flag and condition code bits are cleared.
— The interrupt mask bits (I1 and I0) in the status register are both set to one.
— All bits in the operating mode register (except MA and MB) are cleared.
2. The chip operating mode bits (MA and MB) in the OMR are loaded from external mode
select pins, establishing the operating mode of the chip.
3. The core begins instruction execution at the program memory address that is defined by the
address of the reset vector that is provided to the core. There may be different vector
addresses for different reset sources, such as the RESET signal or the COP and RTI timer.
The reset vector or vectors are specific to a particular DSP56800E–based device. Consult
the appropriate device’s user’s manual for details.
The DSP56800E core enters the normal processing state upon exiting reset. It is also possible for the core
to enter the debug processing state upon exiting reset when system debug is underway. See Section 9.6,
“Debug Processing State.”
9.3 Exception Processing State
In the exception processing state, the DSP56800E core recognizes and processes interrupts and exceptions.
Interrupts and exceptions can be generated by conditions inside the core, such as illegal instructions, or
from external sources, such as an interrupt request signal. When an exception occurs, control is transferred
from the currently executing program to an interrupt service routine. Upon entering the interrupt service
routine, the core exits the exception processing state and enters the normal processing state. When the
interrupt routine is terminated, the interrupted program resumes execution.
In digital signal processing, some common uses of interrupts are to transfer data between the data memory
and a peripheral device or to begin execution of a DSC algorithm upon the reception of a new sample.
Interrupts are also useful for system calls in an operating system and for servicing peripherals. An interrupt
that is enabled can also be used to exit the DSC’s low-power wait processing state.
There are many sources for interrupts on the DSP56800E Family of chips, and some of these sources can
generate more than one interrupt. Interrupt requests can be generated from conditions within the core, from
the on-chip peripherals, or from external pins. The DSP56800E core features a prioritized interrupt vector
scheme to provide faster interrupt servicing. The interrupt priority structure is discussed in Section 9.3.1,
“Interrupt Priority Structure.”
Several types of exceptions are supported: interrupts, which are generated by the core, the debug port,
on-chip peripherals or interrupt request pins, and instruction level exceptions, which are caused by the
execution of an instruction. The DSP56800E supports an unlimited number of exceptions. Core interrupts
and instruction level exceptions have a fixed priority level (there are software interrupt instructions for
requesting an interrupt at each of the five priority levels); peripheral and debug port interrupts may be
programmed to one of three priority levels or be disabled.
The following sections discuss the interrupt priority levels, the ways in which interrupts are processed, and
the various sources for interrupts and exceptions.
9-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Exception Processing State
9.3.1 Interrupt Priority Structure
The DSP56800E architecture supports five interrupt priority levels. Levels LP, 0, 1, and 2, in ascending
priority, are maskable. Level 3 is the highest priority and is non-maskable. Priority levels 0–2 are used for
programmable interrupt sources, such as peripherals and external interrupt requests. The lowest priority
level, LP, can only be generated by the SWILP instruction. Level 3 interrupts are generated by the core.
Table 9-1 shows the different interrupt priority levels.
Table 9-1. Interrupt Priority Level Summary
IPL
Description
Priority
Interrupt Sources
LP
Maskable
Lowest
0
Maskable
.
On-chip peripherals, IRQA and IRQB, SWI #0 instruction
1
Maskable
.
On-chip peripherals, IRQA and IRQB, SWI #1 instruction,
Enhanced OnCE interrupts
2
Maskable
.
On-chip peripherals, IRQA and IRQB, SWI #2 instruction,
Enhanced OnCE interrupts
3
Non-maskable
Highest
Illegal instruction, hardware stack overflow, SWI instruction,
Enhanced OnCE interrupts, misaligned data access
SWILP instruction
When exceptions or interrupts occur simultaneously, higher-priority exceptions take precedence. It is also
possible for a higher-priority exception to interrupt the interrupt handler of a lower-priority exception.
Reset conditions take precedence over all interrupt priorities. If a reset occurs, the chip immediately enters
the reset processing state.
The current core interrupt priority level (CCPL) defines which interrupt priority levels will be accepted and
which will be rejected by the core. Interrupt sources with a priority level that is equal to or greater than the
CCPL are accepted. Interrupt sources with a priority level that is lower than the CCPL are rejected.
Non-maskable interrupts (level 3) are always accepted. The CCPL is determined from the I1 and I0 bits in
the status register. Table 9-2 shows the CCPL values.
Table 9-2. Current Core Interrupt Priority Levels
Exceptions
Accepted
Exceptions
Masked
0
IPL 0, 1, 2, 3
and SWILP
None
1
1
IPL 1, 2, 3
IPL 0 and
SWILP
The interrupt controller accepts all non-maskable
interrupts and any unmasked interrupts that are
programmed at level 1 or 2.
1
0
2
IPL 2, 3
IPL 0, 1
and SWILP
The interrupt controller accepts all non-maskable
interrupts and any unmasked interrupts that are
programmed at level 2.
1
1
3
IPL 3
IPL 0, 1, 2
and SWILP
The interrupt controller only accepts non-maskable
interrupts (level 3).
I1
I0
CCPL
0
0
0
Freescale Semiconductor
Comments
The interrupt controller accepts any unmasked
interrupt, including the SWILP.
Processing States
9-3
Processing States
Every interrupt source has an associated priority level. For some interrupt sources, such as the SWI
instructions and non-maskable interrupts, the interrupt level is pre-assigned. Other interrupt sources, such
as on-chip peripherals, support a programmable priority level. Programmable interrupt sources other than
those in the debug port can be set to one of the maskable priority levels (0, 1, or 2) or be disabled.
Enhanced OnCE interrupt sources can be programmed as level 1, 2, or 3 or as disabled. The CCPL is set to
level 3 on reset.
When an exception or interrupt is recognized and the CCPL is low enough to allow it to be processed, the
CCPL is automatically updated to be one higher than the level of the interrupt (except for the case of
SWILP, which does not update the CCPL, or the case of level 3 interrupts, which leave the priority level at
level 3). This updating prevents interrupts that have the same or a lower priority level from interrupting the
handler for the current interrupt. When the interrupt service routine finishes, the CCPL is set back to its
original value.
To better understand the interrupt priority structure, consider a simple example with nested interrupts.
Assume that the following have already taken place:
1. A serial port on a chip has requested a level 1 interrupt when the core’s CCPL was at level 0.
2. The core has recognized this interrupt and entered the exception processing state. The
CCPL was updated from level 0 to level 2, which is one level higher than the priority of the
recognized interrupt (level 1).
3. Program flow has been transferred to the interrupt handler for the serial port.
Now consider that a second peripheral, a timer with interrupt priority level 0, generates an interrupt.
Although the interrupt request is valid, the interrupt will not be acknowledged and serviced because the
peripheral’s priority level is lower than the core’s CCPL. If the interrupt request can be latched as pending,
the interrupt will be serviced after the current interrupt service routine completes, because the CCPL will
be restored to its original level (level 0). A higher-priority interrupt (at level 2, for instance) would
interrupt the level 1 service routine, and the level 1 routine would resume later after the level 2 handler
completed.
9.3.2 Interrupt and Exception Processing
When an interrupt or exception occurs, the current program is stopped, and control is passed to an interrupt
handling routine. Once the handling routine has completed processing the interrupt, control is returned to
the original program at the point at which it was interrupted. The location of the interrupt handling routine
that is to be executed is determined with the interrupt vector table.
Interrupt vectors are typically located in a block of memory locations in program memory (although
interrupt vectors can be located anywhere in the program memory map, if desired). Each interrupt vector
typically holds a 2- or 3-word JSR instruction, except for fast interrupts, which are covered in
Section 9.3.2.2, “Fast Interrupt Processing.” When an interrupt occurs, the interrupting device provides a
vector number to the core. Program control is then transferred to the address specified by the vector
provided. At this address, the JSR instruction is fetched and executed, transferring control to the interrupt
service routine. Figure 9-1 on page 9-5 shows an example of the vector table.
When the chip is in data-memory execution mode (see Section 8.6, “Executing Programs from Data
Memory,” on page 8-23), the interrupt vector table is located in data memory, not program memory, and
the interrupt vector is fetched appropriately from data memory when entering exception processing.
9-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Exception Processing State
Interrupt Vector
Table
.
.
.
.
.
.
Handler Address
JSR
Vector #23
JSRs to Normal Interrupt
Handler Routines
Handler Address
JSR
Vector #22
.
.
.
Figure 9-1. Interrupt Vector Table
Two types of interrupt processing routines are supported: normal and fast. Normal interrupt processing is
supported for all types of interrupts, but it involves a certain amount of overhead. Fast interrupt processing
requires substantially less overhead, but it is only available for level 2 interrupts. The type of interrupt
processing that will be performed is determined by the opcode that is located in the vector for a given
interrupt and by the priority level of the interrupt source. If the instruction is a JSR, normal interrupt
processing occurs. If it is any other instruction and level 2, fast interrupt processing is used. The case
where the first instruction is not a JSR and the priority level is 0, 1, or 3 is not permitted.
9.3.2.1 Normal Interrupt Processing
Under most circumstances, normal interrupt processing is used to handle an interrupt or exception. When
an interrupt occurs, the following occurs:
1. The currently executing instruction is allowed to complete, and all subsequent instructions
are flushed from the pipeline.
2. The program counter is frozen.
3. The CCPL is raised to be one higher than the level of the current interrupt.
4. The program controller fetches the JSR instruction that is located at the vector for this
interrupt, and then it unfreezes the PC.
5. The JSR instruction is executed, saving the original program counter and status register on
the software stack.
The interrupt routine that is located at the target address of the JSR is then executed. Be careful in the
interrupt handler routine to save any registers that will be used; otherwise, the operation of the interrupted
program may be affected. The status register, however, is automatically saved when an interrupt occurs, so
it does not need to be saved by the handler.
Freescale Semiconductor
Processing States
9-5
Processing States
Interrupt
Vector Table
Interrupt
Subroutine
JSR
Jump Address (LBL)
ii2
Main
Program
PC Resumes
Operation
ii3
ii4
n1
n2
Interrupt
Routine
Explicit
Return From
Interrupt
(RTI)
iin
RTI
Figure 9-2. Control Flow in Normal Interrupt Processing
When interrupt processing is complete, the interrupt routine should be terminated by an RTI or RTID
instruction. These instructions return control to the interrupted program and restore the status register to its
original value.
Normal interrupts can be nested (refer to Section 10.3.3, “Nested Interrupts,” on page 10-11).
9.3.2.2 Fast Interrupt Processing
The default implementation of fast interrupt processing in the DSP56800E core, which is available only for
level 2 interrupts, is performed when the instruction that is located in the appropriate slot in the vector table
is not a JSR. Fast interrupt processing has lower overhead than normal processing and should be used for
all low-latency or time-critical interrupts. Since the interrupt controller is external to the core, chip
implementations of this core can provide an alternate scheme in detecting fast interrupt processing. For
example, the 568xx family of chips has implemented a scheme whereas, the interrupt controller intercepts
the normal vector table processing and inserts the absolute address into the core via the VAB bus. In this
implementation, the IRQ selected for fast interrupt processing and the address of the code for the fast
interrupts are coded in special chip registers. Please refer to the specific chip implementation for complete
description of fast interrupt processing. The description of fast interrupt throughout this manual follows the
default implementation prescribed by the DSP56800E core.
Initially, fast interrupt processing resembles normal interrupt processing: the core performs steps 1–3 in
Section 9.3.2.1, “Normal Interrupt Processing,” for fast interrupt processing as well. When the core
recognizes that fast interrupt processing should be used—by determining that the interrupt is a level 2
interrupt and that the instruction in the vector is not a JSR—fast interrupt processing is initiated. The
following additional steps are performed:
1. The frozen program counter (return address) is copied to the fast interrupt return address
register (FIRA).
2. The status register (with the exception of the P4–P0 bits) and the NL bit in the operating
mode register are copied to the fast interrupt status register (FISR).
3. The stack pointer (SP) is aligned for long-word accesses.
9-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Exception Processing State
4. The Y register is pushed onto the stack, and the stack pointer is advanced to an empty 32-bit
location.
5. The shadowed registers (R0, R1, N, and M01 on the DSP56800E core, or all Rn, N, N3,
and M01 on the DSP56800EX core) are swapped with their shadows.
Execution of the fast interrupt handling routine then continues with the execution of the instruction in the
interrupt’s vector. The code for a fast interrupt routine might be contained entirely in the interrupt vector
table or might reside outside the table at a user-determined location. If it is located in the vector table, note
that the code for the handling routine may overlap the locations of other vectors, rendering them unusable.
It is more practical to have the interrupt vector for a fast interrupt handler to point to a location outside the
main portion of the interrupt vector table, to avoid the overlap problem of a fast interrupt service routine
with more than 2 words.
Interrupt
Vector Table
Main
Program
Fast Interrupt
Subroutine
ii0
PC Resumes
Operation
ii1
ii2
n1
n2
ii3
FRTID
Explicit
Return From
Fast Interrupt
(FRTID)
di0
di1
Figure 9-3. Control Flow in Fast Interrupt Processing
A fast interrupt handling routine is terminated with the FRTID instruction, a delayed return from a fast
interrupt. This instruction performs the following:
1. Swaps the shadowed registers back to their original values
2. Decrements the SP by two
3. Pops the Y register off the stack and restores the stack pointer to its original value
4. Restores the SR and the NL bit in the OMR from the FISR register
5. Sets the PC to the value in the FIRA register, returning control to the interrupted program
Note that fast interrupt handlers, like interrupt handlers that are executed in normal interrupt processing
mode, can be interrupted by a higher-priority interrupt.
The execution of a fast interrupt service routine always conforms to the following rules:
1. The first instruction in the interrupt vector table is the first instruction of the level 2 interrupt
service routine for its associated interrupt source.
2. The following instructions are not allowed in the first four instructions of a fast interrupt
service routine:
– JSR, BSR, RTS, RTSD, RTI, RTID
Freescale Semiconductor
Processing States
9-7
Processing States
– BRA, BRAD, Bcc, JMP, JMPD
– STOP, WAIT, DEBUGHLT
– DEBUGEV when programmed to halt the core
– SWI, SWI #n, SWILP, ALIGNSP
– REP, DO, DOSLC
3. The first 5 instruction words in a fast interrupt service routine cannot contain an instruction
that accesses program memory.
4. The instructions for the level 2 interrupt service routine are located directly in the interrupt
vector table unless a jump or branch transfers control out of the vector table. As a result, a
level 2 interrupt service routine typically occupies more than 2 program words in the
interrupt vector table.
5. To prevent one level 2 fast interrupt from interrupting another, the status register’s I1 and
I0 bits should not be explicitly changed during a fast interrupt service routine. A fast
interrupt handler can still be interrupted by a level 3 interrupt.
Fast interrupts are not nestable because fast interrupts are only available as level 2 interrupts—one level 2
interrupt cannot interrupt another level 2 interrupt.
9.3.3 Interrupt Sources
Interrupt requests on a DSP56800E–based chip are generated by one of three sources: hardware sources
outside the core (peripherals, interrupt request signals), hardware sources within the core (illegal
instructions, data access exceptions, debug port exceptions), and software interrupt instructions.
Each interrupt source has at least one associated interrupt vector—the address to which program flow is
transferred when an interrupt occurs. Interrupt vectors are located in a block of memory called the interrupt
vector table. The interrupt source provides the location of the appropriate vector to the interrupt control
hardware.
Exact information on possible interrupt sources, and the size and location of the vector table, can be found
in the user’s manual for the particular DSP56800E–based device.
9.3.3.1 External Hardware Interrupt Sources
Interrupt and reset sources outside the core are unique to a chip’s particular configuration of peripherals
and so on. Consult the user’s manual for the particular DSP56800E–based device.
9.3.3.2 Hardware Interrupt Sources Within the Core
The hardware interrupt sources within the core include the following:
9-8
•
Illegal instruction interrupts
•
Hardware stack overflow interrupts
•
Misaligned data access interrupts
•
Debugging (Enhanced OnCE) interrupts
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Exception Processing State
9.3.3.2.1 Illegal Instruction Interrupt
The illegal instruction interrupt is a non-maskable level 3 interrupt source. It is generated when the
DSP56800E core identifies an instruction as invalid. The illegal instruction interrupt is serviced
immediately following the attempted execution of an undefined operation code—that is, no other
instructions are executed between the illegal instruction and the first JSR instruction that is fetched from
the interrupt vector table in the exception processing state.
It is not possible to recover from an illegal instruction exception because critical state information is lost
when an invalid instruction is executed. However, handling this interrupt can be used for diagnostic
purposes—to locate the faulty code. The address of the instruction that immediately follows the illegal
instruction is pushed on the stack when the illegal instruction exception handler is entered. This address
can be used to locate the illegal instruction in memory.
The ILLEGAL instruction is a mnemonic for one of the invalid instruction opcodes. It can be used to test
the illegal instruction interrupt service routine.
Note that the illegal instruction exception is not necessarily generated for all invalid opcodes. Opcodes
with addressing modes that are not technically illegal, but that perform no useful work, might not generate
an exception even though these opcodes are not supported and thus are considered illegal.
9.3.3.2.2 Hardware Stack Overflow Interrupt
The hardware stack overflow interrupt is a non-maskable level 3 interrupt source. Encountering the
hardware stack overflow interrupt request means that more than two values have been stacked onto the
hardware stack and that the oldest top-of-loop address has been lost (see Section 8.4, “Hardware Stack,” on
page 8-17). The hardware stack overflow interrupt is non-recoverable and is used primarily for debugging.
The hardware stack overflow refers only to the hardware stack and is not affected by the software stack
operation.
9.3.3.2.3 Misaligned Data Access Interrupt
The misaligned data access interrupt is a non-maskable level 3 interrupt source. It occurs when a 32-bit
long-word value is accessed from data memory and the address that is used to access the data is
misaligned. A long-word value must be accessed from memory using an even word address, except when
SP is used in an indirect addressing mode. In the latter case, the value must be accessed using an odd word
address when it is accessed via the stack pointer register. If the long word is not aligned in this manner, a
misaligned data access interrupt is generated. See Section 3.5.3, “Accessing Long-Word Values Using
Word Pointers,” on page 3-19 for more information on the correct alignment for long-word values in
memory.
9.3.3.2.4 Debugging (Enhanced OnCE) Interrupts
The Enhanced On-Chip Emulation module, which provides integrated debugging support for the
DSP56800E, is capable of generating interrupts. These interrupts provide the Enhanced OnCE module
with the capability of executing instructions. See Chapter 11, “JTAG and Enhanced On-Chip Emulation
(Enhanced OnCE),” for more information on the capabilities of the Enhanced OnCE module.
The Enhanced OnCE interrupts can be disabled or programmed to one of three different priority
levels—level 1 through level 3.
Freescale Semiconductor
Processing States
9-9
Processing States
9.3.3.3 Software Interrupt Instructions
The DSP56800E instruction set contains instructions that trigger an interrupt. Depending on the instruction
that is used, any priority interrupt can be generated. These instructions are commonly used for debugging
purposes or operating system calls.
9.3.3.3.1 SWI Instruction—Level 3
The SWI instruction generates a non-maskable level 3 interrupt request. This request is serviced
immediately following the execution of the SWI instruction; no other instructions are ever executed
between the SWI instruction and the first instruction of the interrupt handler.
SWI’s ability to mask out lower-level interrupts makes it very useful for setting breakpoints in monitor
programs. The instruction can also be used for making a system call in a simple operating system.
9.3.3.3.2 SWI #x Instructions—Levels 0–2
The SWI #0, SWI #1, and SWI #2 instructions are maskable interrupt sources. Executing these instructions
generates an interrupt request at the specified priority level, and each typically has its own vector address.
These instructions execute in 1 clock cycle. If the interrupt requested by the SWI #x instruction is at a
priority level greater than or equal to the CCPL, the interrupt is recognized by the core. A minimum of 3
additional clock cycles are executed before the core forces three NOPs into the pipeline and executes the
first instruction located in the vector table. As a result, up to three instructions immediately after the
SWI #x instruction may be executed before the interrupt is serviced.
If the SWI #x instruction is executed with a priority level that is lower than the CCPL, the request is
latched as pending by the interrupt controller and will be serviced only after the core’s CCPL is lowered to
a level that is less than or equal to the priority of the instruction.
Note that the SWI #2 instruction can also be used for fast interrupt processing.
9.3.3.3.3 SWILP Instruction—Lowest Priority
The operation of the SWILP instruction is very similar to the operation of the maskable SWI instructions.
Executing SWILP generates the lowest-priority interrupt request that is available.
This instruction executes in 1 clock cycle. If the CCPL is at level 0, the interrupt is recognized by the core.
In this case, a minimum of 3 additional clock cycles are executed before the core forces three NOPs into
the pipeline and executes the JSR located in the vector table. As a result, up to three instructions
immediately after the SWILP instruction may be executed before the interrupt is serviced.
If the SWILP instruction is executed when the CCPL is greater than level 0, the request is latched as
pending by the interrupt controller and will be serviced only after the core’s CCPL is lowered to level 0.
Processing SWILP, the lowest-priority interrupt, does not update the CCPL. It is possible for a level 0
interrupt request to interrupt the handler for SWILP.
This instruction is typically executed within other interrupt handlers, where its low priority will not be
recognized until all other interrupt handlers have completed execution. Used in this manner, the SWILP
instruction can schedule code for execution after all of the interrupt handlers have completed execution.
9.3.4 Non-Interruptible Instruction Sequences
In general, interrupts can only occur between the execution of two instructions. However, there are certain
sequences of instructions that are not interruptible. When one of these sequences is executed, interrupts are
effectively disabled until after the last instruction in the sequence. In the following sets of instructions,
interrupts cannot occur between the instructions:
•
9-10
A delayed flow control instruction (such as JMPD) and the instructions in the delay slots
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Wait Processing State
•
A REP instruction and the instruction that is to be repeated
•
A 1-word Bcc instruction and either of the following:
— A multi-word instruction
— A 1-word instruction and the instruction that immediately follows it
•
A multi-word Bcc and the instruction immediately after the Bcc
•
BRSET or BRCLR and either of the following:
— A multi-word instruction
— A 1-word instruction and the instruction that immediately follows it
•
A Jcc instruction and the instruction that is executed immediately after the Jcc
•
A Tcc instruction with an R0,R1 register transfer and the instruction that immediately follows it
•
An ADD.W X:(SP-xx),EEE instruction and the instruction that immediately follows it
•
An SWI at the highest priority level and the instruction that immediately follows it (see following
paragraph on SWI)
•
Any of the last 3 program words in a hardware DO or DOSLC loop during the last iteration of the
hardware loop
Consider the code fragment in Example 9-1. BRSET is an instruction that causes interrupts to be
temporarily disabled, as noted in the preceding list.
Example 9-1. BRSET Non-Interruptible Sequence
NOP
BRSET #34,X0,LABEL
ASL
A
DEC.W X:$3400
MOVE.W Y0,X0
;
;
;
;
;
(interrupt may occur before BRSET)
Begins Non-Interruptible Sequence
===> No interrupt allowed before ASL
===> No interrupt allowed before DEC
(interrupt allowed before MOVE)
ADD
; (interrupt allowed if branch taken)
LABEL
X0,A
If the branch is not taken, interrupts will be disabled until after the DEC.W instruction is executed. Any
interrupts that occur during the time that is taken to execute these three instructions will be deferred until
the end of this sequence. If the branch is taken, interrupts can occur between the BRSET and ADD
instructions.
The SWI instruction is included in this list because of the nature of this instruction. The SWI instruction is
designed so that upon execution, the instruction immediately after the SWI will not be executed; instead,
the processor directly enters the exception processing state. Thus, by design, no interrupts can occur
between the execution of the SWI instruction and the processor’s direct entry into the exception processing
state.
9.4 Wait Processing State
One of the DSP56800E core’s low-power-consumption states is wait mode. This mode is entered by
executing the WAIT instruction. After a delay, the processor enters a state where the internal clock to the
core is disabled and clocks to the memories are typically disabled, but where clocks continue to run to the
on-chip peripherals and to the interrupt controller.
Freescale Semiconductor
Processing States
9-11
Processing States
Wait mode is exited when an interrupt request is sent to the core. The interrupt must be enabled
(unmasked) and must be at a higher priority level than the core’s current interrupt priority level, as defined
by the I1 and I0 bits in the status register. Upon exiting this mode, the program continues execution in the
exception processing state, where it processes the recognized interrupt request. Wait mode is also exited
when the chip is reset, or by certain debug actions in the JTAG/Enhanced OnCE unit.
9.4.1 Wait Mode Timing
The timing for entering and exiting the wait processing state is determined by the architecture of the
particular DSP56800E–based device being used. Consult that device’s user’s manual for exact wait mode
timing information.
9.4.2 Disabling Wait Mode
The DSP56800E core supports the permanent disabling of the wait processing state. If disabled, wait mode
can never be entered, and the WAIT instruction simply executes five NOPs. Upon completing the NOP
cycles, program execution continues with the instruction that immediately follows the WAIT instruction.
Consult the specific DSP56800E–based device’s user’s manual for more information on disabling wait
mode.
9.5 Stop Processing State
The second of the DSP56800E core’s low-power-consumption states is stop mode. In this state the core
consumes the lowest amount of power. This mode is entered by executing the STOP instruction. After a
delay, the internal core clock, the interrupt controller, and any on-chip memories are disabled. The clock is
also disabled to selected peripherals on the chip, but it may continue to run to the PLL block or to a timer
block.
All peripheral and external interrupts are typically cleared on entering the stop state. Hardware stack
overflows that were pending remain pending. The priority levels of the peripherals remain as they were
before the STOP instruction was executed. The on-chip peripherals are held in their respective individual
reset states.
In a typical system architecture, the following events can bring the core out of the stop processing state:
•
An external pin is asserted.
•
The RESET signal is asserted.
•
An on-chip timer reaches zero.
•
Debug actions in the JTAG/Enhanced OnCE unit occur.
Any of these actions will re-activate the oscillator, and, after a clock stabilization delay, clocks to the
processor and peripherals will be re-enabled. The clock stabilization delay period is determined by the stop
delay (SD) bit in the operating mode register (OMR).
If an interrupt is used to wake the processor from stop mode, the first code to be executed on leaving stop
mode is either the interrupt handler for that request or the instruction immediately following the STOP
instruction (see the user’s manual for a particular DSP56800E–based device for more details). Likewise,
the processor will enter the reset processing state if a reset signal was the cause for waking from stop
mode.
9-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Debug Processing State
9.5.1 Stop Mode Timing
The timing for entering and exiting stop mode is determined by the architecture of the particular
DSP56800E–based device being used. Consult the specific device’s user’s manual for more information
on stop mode timing.
9.5.2 Disabling Stop Mode
The DSP56800E core supports the permanent disabling of the stop processing state. If disabled, stop mode
can never be entered, and the STOP instruction simply executes five NOPs. Upon completing the NOP
cycles, program execution continues with the instruction that immediately follows the STOP instruction.
Consult the specific DSP56800E device’s user’s manual for more information on disabling stop mode.
9.6 Debug Processing State
The debug processing state is a state where the core is halted and placed under the control of the Enhanced
OnCE debug port. Serial data is shifted in and out of this port, and it is possible to execute instructions
from this processing state. It is also possible to use the debug port without entering the debug processing
state. This is useful for applications where the core must not be halted.
Freescale Semiconductor
Processing States
9-13
Processing States
9-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 10
Instruction Pipeline
The DSP56800E architecture is built around an eight-stage execution pipeline. The eight stages overlap
instruction fetches, operand fetches, and instruction execution, resulting in higher execution throughput.
The eight stages of the pipeline are shown in Figure 10-1.
Pre-Fetch 1 (P1)
Pre-Fetch 2 (P2)
Instruction Fetch (IF)
Instruction Decode (ID)
Address Generation (AG)
Operand Pre-Fetch 2 (OP2)
Execute and Operand Fetch (EX)
Execute 2 (EX2)
Figure 10-1. DSP56800E Eight-Stage Pipeline
Instructions typically require 7 or 8 clock cycles to be fetched, to be decoded, and to finish execution,
depending on their complexity. Most instructions will complete and be retired (their results written back
and condition codes updated) by the end of the Execute stage of the pipeline. Some more complex
instructions require additional processing and are retired in the Execute 2 stage. AGU arithmetic
instructions complete execution in the Address Generation stage. Although it takes as many as 8 clock
cycles to fill the pipeline and to complete the execution of the first instruction, subsequent instructions
typically complete execution on each clock cycle thereafter.
Although the execution pipeline is composed of many stages, its operation is largely hidden from the user.
Knowledge of the pipeline is useful, however, because certain code sequences can introduce pipeline
dependencies. These dependencies, and the resultant pipeline stalls, can affect overall performance if they
are not addressed. The following sections describe the pipeline in detail, including those circumstances that
can result in pipeline dependencies.
Freescale Semiconductor
Instruction Pipeline
10-1
Instruction Pipeline
10.1 Pipeline Stages
The eight stages of the pipeline, and their abbreviations, are as follows:
1. Pre-Fetch 1 (P1)—The address of the instruction that is to be fetched is driven onto the
program address bus (PAB).
2. Pre-Fetch 2 (P2)—Program memory latches the instruction address and begins program
memory access.
3. Instruction Fetch (IF)—Program memory places the instruction opcode onto the program
data bus (PDB).
4. Instruction Decode (ID)—The instruction latch latches and decodes the opcode. It is at
this point in the pipeline that the instruction is identified.
5. Address Generation (AG)—The address generation unit (AGU) drives data memory
access addresses onto the primary and secondary data address buses (XAB1 and XAB2).
Address and AGU calculations (including transfers done with the TFRA instruction) are
performed in the AGU’s arithmetic units and are stored in the destination AGU register.
6. Operand Pre-Fetch 2 (OP2)—Data memory latches the data address and begins data
memory access.
7. Execute and Operand Fetch (EX)—For a memory read, data memory places its value
onto the primary and secondary data read buses (CDBR and XDB2), and the value or values
are captured in the move’s destination registers. For a memory write operation, data that is
to be written to data memory is placed onto the core data bus for writes (CDBW).
Multiplications and MACs begin in this stage in the data ALU’s arithmetic unit, and the
multiplication result is stored in an intermediate pipeline latch. Multi-bit shifting
instructions (arithmetic and logical) begin in this stage in the data ALU’s arithmetic unit,
and the temporary result is stored in an intermediate pipeline latch. All data ALU
calculations other than those that are previously listed are performed in the data ALU’s
arithmetic unit and are stored in the destination data ALU register, unless they are executed
using Late Execution.
8. Execute 2 (EX2)—Multiplications, MACs, and multi-bit shift instructions complete in this
stage in the data ALU’s arithmetic unit, and the final result is stored in the destination data
ALU register (ASLL.L, ASRR.L, and LSRR.L take an additional cycle since they are
2-cycle instructions). Data ALU calculations other than those that are listed previously are
performed in the data ALU’s arithmetic unit and are stored in the destination data ALU
register when they are executed using Late Execution.
Table 10-1 on page 10-3 shows the relationship between fundamental operations, such as memory
accesses and calculations, and the various pipeline stages. The execution of data ALU operations in the
pipeline is discussed in more detail in Section 10.2.2, “Data ALU Execution Stages.”
10-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Normal Pipeline Operation
Table 10-1. Mapping Fundamental Operations to Pipeline Stages
Operation
Pipeline Stages
Instruction fetch
P1, P2, IF
Data memory access
AG, OP2, EX
AGU calculation
AG
Data ALU calculation—Normal
EX
Data ALU calculation—Late
EX2
Data ALU calculation—multiplication and shifts
EX, EX2
Note that memory accesses take place across three stages of the pipeline: an address is provided in the first
cycle of an access, the memory latches the address on the second cycle, and the memory drives the
corresponding data bus on the third cycle. This requirement applies when accessing both program and data
memory, and when fetching both instructions and operands.
10.2 Normal Pipeline Operation
Normal instruction execution occurs in an eight-stage pipeline, allowing most instructions to be retired at a
rate of one instruction per clock cycle. Certain instructions, however, require more than 1 clock cycle to
complete. These include the following:
•
Instructions longer than 1 instruction word
•
Instructions using an addressing mode that requires more than 1 cycle for the address calculation
•
Data ALU arithmetic instructions with one operand in memory
•
Instructions causing a change of flow
•
Instructions accessing program memory
•
Special instructions:
— Multi-bit shifting instructions that operate on 32-bit values
— TSTDECA.W instruction
— NORM instruction
— ALIGNSP instruction
— REP instruction
10.2.1 General Pipeline Operations
Pipelining allows instruction executions to overlap so that the execution of one pipeline stage for a given
instruction occurs concurrently with the execution of other pipeline stages for other instructions. The
processor fetches only 1 instruction word per clock cycle; if an instruction is more than 1 instruction word
in length, it fetches each additional word with an additional cycle before fetching the next instruction.
Table 10-2 on page 10-4 demonstrates simultaneous execution through the pipelining of the five
instructions that are found in Example 10-1 on page 10-4.
Freescale Semiconductor
Instruction Pipeline
10-3
Instruction Pipeline
Example 10-1. Example Code to Demonstrate Pipeline Flow
MOVE.W X:(R0),A
ADD
A,B
MOVE.W B,C
MOVE.W C1,X:$0C00
INC.W C
;
;
;
;
;
n1:
n2:
n3:
n4:
n5:
1-word,
1-word,
1-word,
2-word,
1-word,
1-cycle
1-cycle
1-cycle
2-cycle
1-cycle
instruction
instruction
instruction
instruction
instruction
The abbreviations n1 and n2 refer to the first and second instructions, respectively, that are executed in the
pipeline. The fourth instruction, n4, contains an instruction extension word (typically an absolute address
or immediate value), which is labeled n4e. As shown in Table 10-2, it takes an additional clock cycle to
fetch and process the extension word.
All instructions are referred to by their n abbreviations before they reach the Instruction Decode stage of
the pipeline. Then, as Table 10-2 demonstrates, the instructions are referred to by name (or by a shortened
version thereof) to reflect that they have been identified.
Table 10-2. Instruction Pipelining
Instruction Cycle
Pipeline Stage
P1 (Pre-Fetch 1)
P2 (Pre-Fetch 2)
IF (Instruction Fetch)
ID (Instruction Decode)
AG (Address Generation)
1
2
3
4
5
6
7
8
9
10
11
•
n1
n2
n3
n4
n4e
n5
•
•
•
•
•
•
n1
n2
n3
n4
n4e
n5
•
•
•
•
•
n1
n2
n3
n4
n4e
n5
•
•
•
•
mov1
add
mov
mov
mov
inc
•
•
•
mov
add
mov
mov
mov
inc
•
•
mov
add
mov
mov
mov
inc
•
mov
add
mov
mov
mov
•
—
—
—
—
—
OP2 (Operand Pre-Fetch 2)
EX (Execute and Operand Fetch)
EX2 (Execute 2)
1.In all of the pipeline tables in this chapter, MOVE instructions are notated as “mov.”
It can be seen that although each instruction takes many clock cycles to complete execution, throughput
remains high due to the pipelining.
10.2.2 Data ALU Execution Stages
Data ALU instructions are executed in the last two stages of the pipeline, Execute and Execute 2. Data
ALU instructions execute in one of four ways:
•
Normal Execution—Arithmetic and logical instructions that begin and complete execution in the
Execute phase.
•
Late Execution—Arithmetic and logical instructions that begin and complete execution in the
Execute 2 phase.
•
Two-Stage Execution—Multiplication, multiply-accumulate, and multi-bit shifting instructions
that begin execution in the Execute phase and complete in the Execute 2 phase. These instructions
place the data ALU into Late mode.
•
Multi-Cycle Execution—Data ALU instructions that execute in more than 1 clock cycle.
10-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Normal Pipeline Operation
Data ALU instructions such as ADD, CMP, TST, and NEG are typically executed by the data ALU using
Normal Execution. When a multiplication or multi-bit shifting instruction is encountered, it is processed
using Two-Stage Execution (still executing in a single cycle), and it places the data ALU into the Late
Execution state. The data ALU then remains in the Late state until a non–data ALU instruction is executed.
The transitions between states are determined as follows:
•
Instructions that are not executed in the data ALU, and multi-cycle data ALU instructions—except
ASLL.L, ASRR.L, and LSRR.L—place the data ALU into the Normal state.
•
Two-stage instructions and the ASLL.L, ASRR.L, and LSRR.L instructions place the data ALU
into the Late state.
•
All other instructions keep the data ALU in its current state.
The complete list of two-stage instructions follows. Each of these instructions uses two pipeline stages and
places the data ALU into the Late Execution state.
•
IMAC.L, IMPY.L, IMPY.W
•
IMACUS, IMACUU, IMPYSU, IMPYUU
•
MAC, MACR, MPY, MPYR
•
MACSU, MPYSU
•
ASLL.W, ASRR.W, LSRR.W
•
ASLL.L, ASRR.L, LSRR.L
•
ASRAC, LSRAC
There are three conditions where the data ALU can cause pipeline dependencies. They occur when:
•
The result of a data ALU instruction that is executed in the Late state is used in the immediately
following instruction as the source register in a move instruction.
•
The result of a data ALU instruction that is executed in the Late state is used in the immediately
following two-stage instruction as the source register to a multiplication or multi-bit shifting
operation. A dependency does not occur if the result is used in an accumulation, arithmetic, or logic
operation on the immediately following instruction.
•
An instruction requiring condition codes, such as Bcc, is executed immediately after a data ALU
instruction is executed in the Late state.
When a data ALU dependency occurs, interlocking hardware on the core automatically stalls the core for
1 cycle to remove the dependency.
Example 10-2 on page 10-6 contains a code sequence demonstrating the behavior of the pipeline with a
variety of different instructions. Note how instructions that are executed using Normal Execution, such as
n2, n3, and n4, complete before the final stage of the pipeline.
Freescale Semiconductor
Instruction Pipeline
10-5
Instruction Pipeline
Example 10-2. Demonstrating the Data ALU Execution Stages
NOP
ADD
X0,A
ASL
A
MOVE.W A,X:(R0)+
;
;
;
;
n1:
n2:
n3:
n4:
Non-data ALU (restores to Normal state)
Normal Execution (Execute phase)
Normal Execution (Execute phase)
Normal Execution (no dependency)
MPY
X0,Y0,B
MOVE.W B,X:(R0)+
; n5: Two-Stage (Execute and Execute 2)
; n6: (dependency occurs--1 stall cycle)
; Non-data ALU (restores to Normal state)
MAC
X0,Y0,A
MAC
X0,Y0,A
SUB
Y1,A
MOVE.W A,X:(R0)+
;
;
;
;
;
ASRR.W #3,A
BNE
LABEL
; n11: Two-Stage (Execute and Execute 2)
; n12: (dependency occurs--1 stall cycle)
; Non-data ALU (restores to Normal state)
n7: Two-Stage (Execute and Execute 2)
n8: Two-Stage (Execute and Execute 2)
n9: Late Execution (Execute 2 phase)
n10: (dependency occurs--1 stall cycle)
Non-data ALU (restores to Normal state)
Table 10-3. Execution of Data ALU Instructions in the Pipeline
Pipeline
Stage
P1
P2
IF
ID
AG
OP2
EX
EX2
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
11
n1
n2
n3
n4
n5
n6
n7
n8
n9
—
n10 n11 n12
n1
n2
n3
n4
n5
n6
n7
n8
—
n9
n10 n11 n12
n2
n3
n4
n5
n6
n7
—
n8
n9
n1
nop add asl mov mpy
—
—
add
asl
mov mpy
—
add
asl
—
add
asl
—
—
12
13
mov mpy
16
17
18
19
20
21
•
—
•
—
•
•
•
•
—
•
—
•
•
•
•
—
n12
—
•
•
•
•
mov asrr
—
bcc
•
•
•
—
bcc
•
•
—
bcc
•
—
bcc
asrr
—
—
mov mac mac sub
—
mov mpy
—
15
n10 n11
mov mac mac sub
—
14
—
—
mov mac mac sub
—
mpy
mov mac mac
—
—
mov asrr
—
—
mov asrr
—
mac mac sub
mov asrr
—
—
Several pipeline effects occur in the code in Example 10-2:
•
No pipeline effect between ASL (n3) and MOVE.W (n4), since the ASL is done in Execute
•
Pipeline stall occurs because the result of MPY (n5) is not available for write to memory (n6) until
the end of cycle #12
•
No pipeline effect between successive MAC and data ALU instructions (n7, n8, and n9)
•
Pipeline stall because result of SUB (n9) is not available for write to memory (n10) until the end of
cycle #17
•
Pipeline stall because result of ASRR.W (n11) is not available for conditional branching (n12) until
cycle #20
10-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
10.3 Pipeline During Interrupt Processing
The instruction pipeline functions slightly differently when processing interrupt requests. Beyond the
standard eight-stage pipeline, additional cycles are required for arbitrating and interrupting the core. On a
typical chip implementation, two extra stages are required. This addition effectively makes the interrupt
pipeline 10 levels deep. The two additional stages are as follows:
•
Interrupt Arbitration (Int Arbitr)
•
Interrupt Request (Int Req)
The Interrupt Arbitration stage is required for arbitrating among all the different possible requesting
sources. If a valid interrupt is found at a high enough priority level after this arbitration is performed, the
program interrupt controller asserts an interrupt request to the core. This assertion occurs during the
Interrupt Request stage.
Note in this example that these 2 additional processing cycles are not real stages in the pipeline. Rather,
they are performed in the interrupt controller, and they do not directly affect the operation of the pipeline.
However, these cycles do affect the overall processing time for an interrupt, so they can be considered
additional pipeline stages for the purpose of calculating interrupt latency. For an exact calculation of
interrupt latency, refer to Section 10.3.8, “Interrupt Latency.”
10.3.1 Standard Interrupt Processing Pipeline
Figure 10-2 on page 10-8 shows the program flow and pipeline during standard interrupt processing.
Freescale Semiconductor
Instruction Pipeline
10-7
Instruction Pipeline
Interrupt
Vector Table
Interrupt
Subroutine
JSR
Jump Address (LBL)
ii2
Main
Program
PC Resumes
Operation
ii3
ii4
n1
n2
Interrupt
Routine
Explicit
Return From
Interrupt
(RTI)
iin
RTI
(a) Instruction Flow
Interrupt Requests Sampled
by the Arbiter
Pipeline
Stage
Instruction Cycle
6
7
8
9
10
11
12
13
14
P1
n1 n2 n3 n4 ii0 ii1
ii1
ii1
ii2
ii3
ii4
ii5
•
•
•
•
•
•
•
n2
•
•
•
P2
n1 n2 n3 n4 ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
•
•
•
•
•
•
•
n2
•
•
IF
n1 n2 n3 n4
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
•
•
•
•
•
•
•
n2
•
ID
n1 — —
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rti
rti
rti
rti
rti
rti
rti
rti
n2
AG
n1 —
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rti
rti
rti
rti
rti
rti
rti
rti
n1
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rti
rti
rti
rti
rti
rti
rti
n1
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rti
rti
rti
rti
rti
rti
n1
—
—
—
—
—
—
—
ii2
ii3
ii4
—
—
—
—
—
Int Arbitr
Int Req
OP2
1
2
3
4
5
15 16 17 18 19 20 21 22 23
i
i
EX
EX2
i = Interrupt Arbitration and Request
ii = Interrupt instruction word
ii0 = First word of JSR instruction
ii1 = Second word of JSR instruction
ii5 = RTI instruction
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-2. Standard Interrupt Processing
10-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
When an interrupt request is asserted, the interrupt controller takes 2 cycles to arbitrate between interrupts
and to send an interrupt request to the core. During this time, the pipeline continues to function normally.
When the core recognizes an interrupt request, as in cycle #5 in Figure 10-2 on page 10-8, the transition to
the exception processing state begins. Any instructions in the pipeline that have not yet been decoded are
replaced with NOPs, and the JSR instruction is fetched from the interrupt vector table.
Upon entering the interrupt service routine after executing the JSR instruction, the core returns to the
normal processing state, and the CCPL has been updated to reflect the new priority level.
When the interrupt handler completes (by executing the RTI instruction), control returns to the interrupted
program. The return address, which is saved on the stack by the JSR, points to instruction n2, since the PC
was frozen as soon as the interrupt was recognized. The PC was not updated to point past n2, even though
instructions n2–n4 had already begun to be fetched.
10.3.2 The RTID Instruction
In the example interrupt processing pipeline that is presented in Figure 10-2 on page 10-8, most of the time
that is needed to execute the (admittedly short) interrupt routine is taken up by the JSR and RTI
instructions. Because the RTI instruction manipulates the software stack and causes execution flow to
change, it takes several cycles to execute. To help reduce the overhead that is required in processing an
interrupt, an alternative to the RTI instruction is provided: the delayed return from interrupt (RTID).
The RTID instruction performs the same function as RTI, but it reduces overhead by executing the
instructions in the 3 subsequent program words before returning control to the interrupt program. These
instruction words, or “delay slots,” must always be filled. If it is not possible to fill all of the delay slots
with useful instructions, then NOP instructions must be placed in the unfilled slots. See Section 4.3,
“Delayed Flow Control Instructions,” on page 4-12 for more information on the RTID instruction.
The interrupt processing pipeline when RTID is used is given in Figure 10-3 on page 10-10. Note the
difference between Figure 10-3 and Figure 10-2 on page 10-8 from cycle #13 onward: the di0–di2
instructions are executed before control returns to instruction n2.
Freescale Semiconductor
Instruction Pipeline
10-9
Instruction Pipeline
Interrupt
Vector Table
Interrupt
Subroutine
Main
Program
PC Resumes
Operation
ii2
ii3
JSR
Jump Address (LBL)
n1
n2
iin
Interrupt
Routine
RTID
di0
Explicit
Return From
Interrupt
(RTID)
di1
di2
(a) Instruction Flow
Interrupt Requests Sampled
by the Arbiter
Pipeline
Stage
Instruction Cycle
6
7
8
9
10
11
12
13
14
15 16 17 18 19 20 21 22 23
P1
n1 n2 n3 n4 ii0 ii1
ii1
ii1
ii2
ii3
ii4
ii5
di0
di1
di2
P2
n1 n2 n3 n4 ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
di0
di1 di2
IF
n1 n2 n3 n4
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
di0 di1 di2
ID
n1 — —
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rtid rtid rtid rtid rtid di0 di1 di2 n2
AG
n1 —
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
rtid rtid rtid rtid rtid di0 di1 di2
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4 rtid rtid rtid rtid rtid di0 di1
n1
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4 rtid rtid rtid rtid rtid di0
n1
—
—
—
—
—
—
—
ii2
ii3
Int Arbitr
Int Req
OP2
1
2
3
4
5
i
i
n1
EX
EX2
•
•
•
•
•
n2
•
•
•
•
•
•
n2
•
•
•
•
•
•
n2
•
ii4
—
—
—
—
—
i = Interrupt Arbitration and Request
ii = Interrupt instruction word
ii0 = First word of JSR instruction
ii1 = Second word of JSR instruction
ii5 = RTID instruction
di = Instruction in RTID delay slot
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-3. Execution of the RTID Instruction
10-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
10.3.3 Nested Interrupts
Interrupts on the DSP56800E architecture can be nested; one exception can interrupt another exception’s
interrupt service routine if it has a higher priority. During initial interrupt processing, interrupts are
disabled. Once the JSR instruction reaches the point in the pipeline where it has begun execution, the core
can safely re-enable interrupts because the return address will be stacked properly before another interrupt
can occur. This re-enabling occurs at cycle #11 in Figure 10-4 on page 10-12. Interrupts are disabled
during cycles #4 through #10.
If a second, higher-priority interrupt request occurs after cycle #4, it is not arbitrated until after interrupts
are re-enabled in cycle #11. This scenario is illustrated in Figure 10-4 as interrupt request i2a. The second
interrupt request interrupts the processing of the first at cycle #13, and it is processed before the interrupt
handler for request i1 resumes.
If the vector table contains a 2-word JSR instruction, no interrupts are allowed between the JSR and the
first instruction in the interrupt service routine (ii2). If the vector table contains a 3-word JSR instruction,
interrupts are permitted between the JSR instruction and the first instruction in the interrupt service routine
(ii2).
10.3.4 SWI and Illegal Instructions During Interrupt
Processing
Another case of interest is where a first interrupt request begins the interrupt pipeline and the instruction at
n1 in Figure 10-4 on page 10-12 is a non-maskable SWI instruction or an illegal instruction. The SWI and
illegal instructions execute in 4 clock cycles. Upon completion of these cycles, the exception that is
serviced will not be the original interrupt request. Instead, the core will service the SWI or illegal
instruction exception that is caused by instruction n1. This condition is true only when the first interrupt
request is at a lower priority level than the exception that is caused by the instruction at n1.
Freescale Semiconductor
Instruction Pipeline
10-11
Instruction Pipeline
Interrupt
Vector Table
Interrupt
Subroutine
JSR
2nd ISR — ii2
Jump Address (LBL)
2nd ISR — ii3
Interrupt
Handler
1st ISR — ii8
PC Resumes
Operation
2nd ISR — ii4
1st ISR — ii9
1st ISR — ii10
Interrupt
Routine
Explicit
Return From
Interrupt
(RTI or RTID)
2nd ISR — iin
2nd ISR — RTI
(a) Instruction Flow
First Interrupt Request
Sampled by the Arbiter
Interrupt Requests Again
Sampled by the Arbiter
Pipeline
Stage
Int Arbitr
Int Req
Instruction Cycle
1
2
3
4
i1
i2
i1
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
i2a
i2
i2a
P1
n1 n2 n3 n4 ii0 ii1
ii1
ii1
ii2
ii3
ii4
ii5
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
ii6
ii7
P2
n1 n2 n3 n4 ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
ii6
IF
n1 n2 n3 n4
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
ii0
ii1
ii1
ii1
ii2
ii3
ii4
ii5
ID
n1 — —
—
jsr
jsr
jsr
jsr
ii2
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
AG
n1 —
—
—
jsr
jsr
jsr
jsr
ii2
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
n1
—
—
—
jsr
jsr
jsr
jsr
ii2
—
—
—
jsr
jsr
jsr
jsr
ii2
n1
—
—
—
jsr
jsr
jsr
jsr
ii2
—
—
—
jsr
jsr
jsr
jsr
n1
—
—
—
—
—
—
—
ii2
—
—
—
—
—
—
OP2
EX
EX2
i = Interrupt Arbitration and Request
ii = Interrupt instruction word
ii0 = First word of JSR instruction
ii1 = Second word of JSR instruction
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-4. Interrupting an Interrupt Handler (Nested Interrupt)
10-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
10.3.5 Fast Interrupt Processing Pipeline
Figure 10-5 shows the program flow, and the corresponding pipeline, during fast interrupt processing.
Within the pipeline, ii0 refers to the first instruction word in the fast interrupt handler, and ii4 refers to the
FRTID instruction. The instructions ii5 and ii6 are the 2 instruction words filling the FRTID’s delay slots.
Interrupt
Vector Table
Main
Program
Fast Interrupt
Subroutine
PC Resumes
Operation
ii0
ii1
ii2
n1
n2
ii3
FRTID
di0
di1
Explicit
Return From
Fast Interrupt
(FRTID)
(a) Instruction Flow
Interrupt Requests Sampled
by the Arbiter
Pipeline
Stage
Int Arbitr
Instruction Cycle
1
2
3
4
5
6
7
8
9 10 11
12
13
14
15
16
17
18
19 20 21
i
Int Req
i
P1
n1 n2 n3 n4 ii0 ii1 ii2 ii3 ii4 ii5
ii6
ii7
n2
n3
•
•
•
•
•
•
•
P2
n1 n2 n3 n4 ii0 ii1 ii2 ii3 ii4
ii5
ii6
ii7
n2
n3
•
•
•
•
•
•
IF
n1 n2 n3 n4 ii0 ii1 ii2 ii3
ii4
di0
di1
ii7
n2
n3
•
•
•
•
•
ID
n1 — — — ii0 ii1 ii2
ii3
ii4
di0
di1
—
n2
n3
•
•
•
•
AG
n1 — — — ii0 ii1
ii2
ii3
frtid
di0
di1
—
n2
n3
•
•
•
n1 — — — ii0
ii1
ii2
ii3
frtid
di0
di1
—
n2
n3
•
•
n1 — — —
ii0
ii1
ii2
ii3
frtid
di0
di1
—
n2
n3
•
n1 — —
—
ii0
ii1
ii2
ii3
frtid
di0
di1
—
n2
n3
OP2
EX
EX2
i = Interrupt Arbitration and Request
ii = Interrupt instruction word
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-5. Fast Interrupt Processing
Freescale Semiconductor
Instruction Pipeline
10-13
Instruction Pipeline
10.3.6 Interrupting a Fast Interrupt Service Routine
Fast interrupt service routines can be interrupted by a level 3 interrupt. However, the first few instructions
in a fast interrupt service routine cannot be interrupted, even if a level 3 interrupt is received. Figure 10-6
on page 10-15 shows the fast interrupt pipeline and the point at which interrupts are re-enabled and
subsequent interrupts can be arbitrated. Even if a level 3 interrupt is received prior to this point in the
pipeline, it is not sampled by the interrupt arbiter until instruction cycle #13 (as shown in the figure), so at
least 7 clock cycles in the fast interrupt routine are executed without being interrupted. Note that the
instructions in the FRTID’s 2 delay slots cannot be interrupted.
10-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
Interrupt
Vector Table
Level 3 Interrupt
Subroutine
JSR
Jump Address
ii2
Level 2
Interrupt
Handler
ii4
ii5
ii3
ii6
ii7
ii4
ii8
Interrupt
Routine
FRTID
dly0
Explicit
Return From
Interrupt
(RTI or RTID)
dly1
PC Resumes
Operation
iin
RTI
(a) Instruction Flow
Level 2 Interrupt Request Sampled
by the Arbiter
Pipeline
Stage
Int Arbitr
Level 3 Interrupt Request
Sampled by the Arbiter
Instruction Cycle
1
2
3
4
5
6
7
8
9 10 11
12
i
Int Req
13
14
15
16
17
18
19 20 21
i
i
i
P1
n1 n2 n3 n4 ii0 ii1 ii2 ii3 ii4 ii5
ii6
ii7
ii8
ii9
ii0
ii1
ii2
•
•
•
•
P2
n1 n2 n3 n4 ii0 ii1 ii2 ii3 ii4
ii5
ii6
ii7
ii8
ii9
ii0
ii1
ii2
•
•
•
IF
n1 n2 n3 n4 ii0 ii1 ii2 ii3
ii4
ii5
ii6
ii7
ii8
ii9
ii0
ii1
ii2
•
•
ID
n1 — — — ii0 ii1 ii2
ii3
ii4
ii5
ii6
—
—
—
ii0
ii1
ii2
•
AG
n1 — — — ii0 ii1
ii2
ii3
ii4
ii5
ii6
—
—
—
ii0
ii1
ii2
n1 — — — ii0
ii1
ii2
ii3
ii4
ii5
ii6
—
—
—
ii0
ii1
n1 — — —
ii0
ii1
ii2
ii3
ii4
ii5
ii6
—
—
—
ii0
n1 — —
—
ii0
ii1
ii2
ii3
ii4
ii5
ii6
—
—
—
OP2
EX
EX2
i = Interrupt Arbitration and Request
ii = Interrupt instruction word
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-6. Interrupting a Fast Interrupt Routine
Freescale Semiconductor
Instruction Pipeline
10-15
Instruction Pipeline
10.3.7 FIRQ Followed by Another Interrupt
Figure 10-7 on page 10-17 shows the fast interrupt pipeline for the case of a short, three-instruction, fast
interrupt service routine where the following occur:
•
A fast interrupt request is received.
•
Simultaneously with this request or a short time after it is received, a second interrupt is received.
The point at which interrupts are re-enabled after the exception processing state is exited is shown in the
interrupt pipeline in Figure 10-7. Interrupt arbitration begins again in cycle #11. Even if a level 3 priority
interrupt is received, it is not sampled by the interrupt arbiter until instruction cycle #11, as the figure
shows. This arrangement allows a minimum of 5 clock cycles in the fast interrupt routine to be executed
without being interrupted.
For this short, 3-word interrupt service routine, the fast interrupt routine completes and control returns to
the main program before the second interrupt request is serviced. All interrupt priority levels are eligible
already by cycle #11 because, by this time, the FRTID instruction has restored the status register to its
original value.
In Figure 10-7 on page 10-17, the second interrupt is level 0, 1, or 2. In this case, the interrupt will be
successfully arbitrated in cycle #12 after the contents of the status register have been restored by the
FRTID instruction in cycle #11. This allows a minimum of 2 instruction cycles from the main program to
be executed before the second interrupt is serviced. Additional cycles will be executed if n2 is more than 2
cycles or if n3 is a multi-cycle instruction.
Consider a second case, slightly different from the one shown in Figure 10-7, in which the second interrupt
is level 3. In this case, the interrupt will be successfully arbitrated in cycle #11; exactly one instruction
from the main program will be executed before the second interrupt is serviced.
10-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
Main
Program
Level 2
Fast Interrupt
FRTID
dly0
dly1
Level 0–2
General Interrupt
n1
n2
n3
JSR
n4
Jump Address
n5
n6
(a) Instruction Flow—Fast Interrupt Routine Followed by Another Interrupt
Level 2 Fast Interrupt Request Sampled
by the Arbiter
Pipe
Stage
Int Arbitr
Second Interrupt Request
Sampled by the Arbiter
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
i
Int Req
i
P1
n1 n2 n3 n4 ii0 ii1 ii2
n3
n2
n3
P2
n1 n2 n3 n4 ii0 ii1
ii2
n3
IF
n1 n2 n3 n4 ii0
ii1
ii2
ID
n1 — — —
frtid
AG
n1 — —
OP2
EX
EX2
11
12
i
i
13
14
15
16
17
ii1
ii2
•
18 19 20
i
i
n4
n5
n6
ii0
n2
n3
n4
n5
n6
ii0
ii1
ii2
•
•
•
n3
n2
n3
n4
n5
n6
ii0
ii1
ii2
•
•
dly0
dly1
—
n2
n3
—
—
—
ii0
ii1
ii2
•
—
frtid
dly0
dly1
—
n2
n3
—
—
—
ii0
ii1
ii2
n1 —
—
—
frtid
dly0
dly1
—
n2
n3
—
—
—
ii0
ii1
n1
—
—
—
frtid
dly0 dly1
—
n2
n3
—
—
—
ii0
n1
—
—
—
—
n2
n3
—
—
—
frtid dly0 dly1
•
•
•
i = Interrupt Arbitration and Request
ii = Interrupt Instruction Word
n = Normal Instruction Word
(b) Interrupt Pipeline—Servicing an Interrupt Immediately After a Fast Interrupt Routine
Figure 10-7. Interrupting After Completing the Fastest Fast Interrupt Routine
Freescale Semiconductor
Instruction Pipeline
10-17
Instruction Pipeline
Figure 10-8 on page 10-19 shows the fast interrupt pipeline for the case of a fast interrupt service routine
where the following occur:
•
Two cycles are executed before the FRTID instruction.
•
Simultaneously to this execution or a short time afterwards, a second interrupt at level 3 is received.
Interrupt arbitration begins again in cycle #11. At this point, the level 3 interrupt is successfully arbitrated
and exception processing begins. However, the 2-cycle FRTID instruction (with 2 delay slots), which is
shown in the box in the ID stage of the pipeline, is a non-interruptible sequence. Since interrupts can only
occur when instructions complete execution, the pending level 3 interrupt must wait 1 cycle before
continuing into the exception processing state. This wait is indicated by the jagged arrow in Figure 10-8 on
page 10-19.
If 3 cycles were executed before the FRTID instruction (a case that is not shown in the figure), exception
processing would be delayed 2 cycles instead of the 1 cycle shown in the figure.
10-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
Level 2
Fast Interrupt
Main
Program
ii0
ii1
FRTID
dly0
dly1
n1
n2
Level 3
General Interrupt
n3
n4
JSR
n5
Jump Address
n6
(a) Instruction Flow—Fast Interrupt Routine Followed by Another Interrupt
Second Interrupt Request
(Level 3) Sampled by the Arbiter
Level 2 Fast Interrupt Request Sampled
by the Arbiter
Pipe
Stage
Int Arbitr
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
i
11
12
(Wait 1 Cycle)
13
14
15
16
17
18
ii1
ii2
•
•
•
19 20
i
Int Req
i
P1
n1 n2 n3 n4 ii0 ii1 ii2
ii3
ii4
ii5
n2
n3
ii0
P2
n1 n2 n3 n4 ii0 ii1
ii2
ii3
II4
II5
n2
n3
ii0
ii1
ii2
•
•
•
•
IF
n1 n2 n3 n4 ii0
ii1
ii2
ii3
II4
II5
n2
n3
ii0
ii1
ii2
•
•
•
ID
n1 — — —
ii0
ii1
frtid
dly0
dly1
—
—
—
—
ii0
ii1
ii2
•
AG
n1 — —
—
ii0
ii1
frtid
dly0 dly1
—
—
—
—
ii0
ii1
ii2
n1 —
—
—
ii0
ii1
—
—
—
—
ii0
ii1
n1
—
—
—
ii0
ii1
—
—
—
—
ii0
n1
—
—
—
ii0
—
—
—
—
i
OP2
EX
EX2
frtid dly0 dly1
frtid dly0 dly1
ii1
frtid dly0 dly1
•
•
i = Interrupt Arbitration and Request
ii = Interrupt Instruction Word
n = Normal Instruction Word
(b) Interrupt Pipeline—Servicing an Interrupt Immediately After a Fast Interrupt Routine
Figure 10-8. Interruption by Level 3 Interrupt During FRTID Execution
Freescale Semiconductor
Instruction Pipeline
10-19
Instruction Pipeline
Figure 10-9 on page 10-21 shows the fast interrupt pipeline for the case of a short fast interrupt service
routine where the following occur:
•
A fast interrupt request is received.
•
Simultaneously with this request or a short time after it is received, a second interrupt is received.
In this case, the instructions in the FRTID’s are multi-cycle instructions such that the 2 delay slots execute
in 4 cycles.
For the fast interrupt service routine in this example, control does not return to the main program but
instead immediately enters the second interrupt. This is true anytime the instructions in the FRTID’s delay
slots execute in 4 or more cycles.
If the second interrupt is level 0, 1, or 2, successful arbitration occurs in cycle #12 because the FRTID
instruction must first restore the status register. If the second interrupt request is level 3, arbitration begins
1 cycle earlier in cycle #11. The level 3 interrupt completes successful arbitration 1 cycle earlier. The
exception processing state, however, can only be entered upon the completion of an instruction. Since the
second cycle of the FRTID instruction executes after the completion of the instructions in the delay slots,
the exception processing state is entered at the same time, regardless of the priority level of the second
interrupt.
Consider another scenario that is not shown in Figure 10-9: If the instructions in the FRTID’s 2 delay slots
execute in 3 clock cycles, then 1 instruction from the main program, n2, will be executed before the second
interrupt is entered.
10-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
Main
Program
Level 2
Fast Interrupt
FRTID
dly0
dly1
Level 0–3
General Interrupt
n1
n2
JSR
n3
Jump Address
n4
n5
n6
(a) Instruction Flow—Fast Interrupt Routine Followed by Another Interrupt
Level 2 Fast Interrupt Request Sampled
by the Arbiter
Pipe
Stage
Int Arbitr
Second Interrupt Request
Sampled by the Arbiter
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
i
Int Req
i
P1
n1 n2 n3 n4 ii0 ii1 ii2
ii3
n2
n3
P2
n1 n2 n3 n4 ii0 ii1
ii2
ii3
IF
n1 n2 n3 n4 ii0
ii1
ii2
ID
n1 — — —
frtid
AG
n1 — —
—
OP2
EX
EX2
11
12
i
i
13
14
15
16
17
ii1
ii2
•
i
i
n4
n5
n6
ii0
n2
n3
n4
n5
n6
ii0
ii1
ii3
n2
n3
n4
n5
n6
ii0
dly0
dly0
dly1
dly1
—
—
—
frtid
dly0
dly0
dly1 dly1
—
—
n1 —
—
—
frtid
dly0
dly0 dly1 dly1
n1
—
—
—
frtid
dly0 dly0 dly1 dly1
—
n1
—
—
—
18 19 20
•
•
•
ii2
•
•
•
ii1
ii2
•
•
—
ii0
ii1
ii2
•
—
—
ii0
ii1
ii2
—
—
—
ii0
ii1
—
—
—
—
ii0
—
—
—
—
frtid dly0 dly0 dly1 dly1
i = Interrupt Arbitration and Request
ii = Interrupt Instruction Word
n = Normal Instruction Word
(b) Interrupt Pipeline—Servicing an Interrupt Immediately After a Fast Interrupt Routine
Figure 10-9. Second Interrupt Case with 4 Cycles Executed in FRTID Delay Slots
Freescale Semiconductor
Instruction Pipeline
10-21
Instruction Pipeline
10.3.8 Interrupt Latency
Interrupt latency is the time between when an interrupt request first appears and when the first instruction
in an interrupt service routine is actually executed. The interrupt can only take place on instruction
boundaries (which are subject to the non-interruptible sequences that are described in Section 9.3.4,
“Non-Interruptible Instruction Sequences,” on page 9-10). The length of execution of an instruction can
affect interrupt latency.
For purposes of calculation, interrupt latency is defined here as the time between when the interrupt
controller first arbitrates among the interrupt sources and when the first instruction in an interrupt handler
is latched into the instruction latch and is ready to be executed. This first instruction is defined as the
instruction that is executed immediately after the JSR from the interrupt vector table. See Figure 10-10.
Interrupt Request Sampled
by the Arbiter
Pipeline
Stage
First Instruction in Handler Reaches
Instruction Decode
Instruction Cycle
7
8
9
10
11
12
13
14
Int Req
i
P1
n1 n2 n3 n4 ii0 ii1 ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
•
•
•
•
•
•
P2
n1 n2 n3 n4 ii0 ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
•
•
•
•
•
Int Arbitr
1
2
3
4
5
6
15 16 17 18 19 20 21 22
i
IF
n1 n2 n3 n4 ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
•
•
•
•
ID
n1 — — —
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
•
•
•
•
•
•
AG
n1 — —
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
•
•
•
•
•
OP2
n1 —
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
•
•
•
•
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3 ii4
•
•
•
•
•
n1
—
—
—
jsr
jsr
jsr
jsr
ii2 ii3 ii4
•
•
•
•
EX
n1
EX2
Figure 10-10. Interrupt Latency Calculation
10.3.8.1 Interrupt Latency
Interrupt latency is calculated as follows:
Latency = Execution time of instruction n1
+ 4 clock cycles (1 for arbitration and 3 NOPs)
+ the number of clock cycles to execute the JSR (4 or 5 cycles)
+ wait states when the JSR instruction pushes the PC and SR to the stack
+ wait states due to program fetches of n3, n4, and ii0–ii3
(or ii0–ii4 if the JSR instruction executes in 5 cycles)
The largest execution time for instruction n1 is 8 clock cycles (when n1 is an RTI or RTS instruction). See
Section 10.3.8.3, “Cases That Increase Interrupt Latency.”
10-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
10.3.8.2 Re-Enabling Interrupt Arbitration
The time when interrupt arbitration is allowed to resume is calculated as follows:
Re-enable = Execution time of instruction n1
+ 4 clock cycles (1 for arbitration and 3 NOPs forced into pipeline)
+ 3 clock cycles (first 3 cycles executing the JSR instruction)
+ wait states when the JSR instruction pushes the PC and SR to the stack
+ wait states due to program fetches of n3, n4, and ii0–ii2
10.3.8.3 Cases That Increase Interrupt Latency
Some special cases increase interrupt latency. Section 9.3.4, “Non-Interruptible Instruction Sequences,” on
page 9-10 documents instruction sequences that are not interruptible. Such sequences increase latency.
Figure 10-11 demonstrates such a case. When the instruction n1 is a 1-word conditional branch instruction,
and when the condition evaluates to false, the two instructions immediately following the Bcc, n2 and n3,
are non-interruptible.
Interrupt Requests
Sampled by the Arbiter
Pipeline
Stage
Int Arbitr
First Instruction
Reaches Decode
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19 20 21 22
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
•
•
i
Int Req
i
P1
n1 n2 n3 n4
n5
n6
n7
n8
ii0
P2
n1 n2 n3
n4
n5
n6
n7
n8
ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
•
IF
n1 n2
n3
n4
n5
n6
n7
n8
ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
ID
bcc bcc bcc n2
n3
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
•
•
AG
bcc bcc bcc
n2
n3
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
•
bcc bcc bcc
n2
n3
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
bcc bcc bcc
n2
n3
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
bcc bcc bcc
n2
n3
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
OP2
EX
EX2
Figure 10-11. Interrupt Latency Calculation—Non-Interruptible Instructions
The STOP instruction places the core into the stop processing state, where interrupts are not recognized.
The WAIT instruction places the core into the wait processing state. An enabled interrupt brings the core
out of this low-power state.
The REP instruction and the instruction that it repeats are not interruptible. Instead, these two instructions
are treated as a single 2-word instruction, regardless of the number of times that the second instruction is
repeated. Instruction fetches are suspended and are re-activated only after the repeat loop is finished (see
Figure 10-12 on page 10-24). During the execution of n2 in Figure 10-12, no interrupts will be serviced.
When the loop finally completes, instruction fetches are re-initiated and pending interrupts can be serviced.
Freescale Semiconductor
Instruction Pipeline
10-23
Instruction Pipeline
Interrupt
Synchronized and
Recognized
as Pending
Main
Program
n2
n2
n2
n2
Instruction n2 Replaced per
the REP Instruction
Repeat
4 Times
n1 (REP #4)
n2
n3
n4
n5
Interrupts
Re-Enabled
n6
JSR
Jump Address
Process Interrupt:
Fetch JSR Instruction from
the Interrupt Vector Table
(a) Instruction Flow
Interrupt Requests Sampled
by the Arbiter
Pipeline
Stage
Int Arbitr
Instruction Cycle
1
P2
IF
ID
AG
OP2
EX
EX2
3
4
5
6
7
8
9
i%
i%
i%
i%
i%
i%
i%
i
10
11
12
13
14
15
16
17
18
ii0
ii1
ii1
ii1
ii2
ii0
ii1
ii1
ii1
ii3
•
•
•
ii2
ii3
•
•
ii0
ii1
ii1
ii1
ii2
ii3
•
—
—
jsr
jsr
jsr
jsr
ii2
ii3
i
Int Req
P1
2
rep
n2
rep
n2
rep
n2
rep
rep
n2
n2
n2
n2
—
rep
rep
n2
n2
n2
n2
—
—
—
jsr
jsr
jsr
jsr
ii2
rep
rep
n2
n2
n2
n2
—
—
—
jsr
jsr
jsr
jsr
rep
rep
n2
n2
n2
n2
—
—
—
jsr
jsr
jsr
rep
rep
n2
n2
n2
n2
—
—
—
jsr
jsr
i = Interrupt Arbitration and Request
i% = Interrupt Request rejected by core and remains pending
ii = Interrupt instruction word
n = Normal instruction word
(b) Interrupt Pipeline
Figure 10-12. Interrupt Latency and the REP Instruction
10.3.8.4 Delay When Enabling Interrupts via CCPL
Another case of interest is the time from the enabling of an interrupt by updating the CCPL in the status
register until the time when the interrupt controller first arbitrates with the newly modified CCPL and an
already pending interrupt is serviced.
10-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline During Interrupt Processing
This case is demonstrated in Figure 10-13. The following notation is used in the figure:
•
n1 is a 1-cycle instruction that modifies the SR register.
•
p0 and p1 are the 2 instruction cycles that are executed immediately before instruction n1. They can
be a single multi-cycle instruction or two single-cycle instructions.
•
ii0 is the first word that is fetched from the interrupt vector table for the interrupt that is serviced.
In Figure 10-13, ii0 is the first word of the JSR instruction.
The single-cycle instruction n1 in this example writes to the status register, lowering the CCPL. The actual
write to the CCPL occurs at the end of cycle #7 (the Execute 2 stage is not used by instruction n1). In cycle
#8, the program interrupt controller arbitrates the already pending interrupts, but now with a lower CCPL.
An interrupt is now recognized as valid, and interrupt processing begins.
Write to SR Changes the
CCPL, Enabling Interrupts
EE
Pipeline
Stage
Arbitrates with NEW CCPL, and
Pending Interrupt Is Serviced
Instruction Cycle
1
2
3
4
5
6
7
Int Arbitr
8
9
10
11
12
13
14
15
16
17
18
19
20
21
i
Int Req
i
P1
n1 n2 n3 n4 n5 n6
n7
n8
n9
ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
•
P2
p0 n1 n2 n3 n4 n5
n6
n7
n8
n9
ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
•
IF
p1 p0 n1 n2 n3 n4
n5
n6
n7
n8
n9
ii0
ii1
ii1
ii1
ii2
ii3
ii4
•
•
•
ID
p1 p0 n1 n2 n3
n4
n5
n6
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
•
AG
p1 p0 n1 n2
n3
n4
n5
n6
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
•
p1 p0 n1
n2
n3
n4
n5
n6
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
ii4
p1 p0
n1
n2
n3
n4
n5
n6
—
—
—
jsr
jsr
jsr
jsr
ii2
ii3
p1
p0
—
n2
n3
n4
n5
n6
—
—
—
—
—
—
—
ii2
OP2
EX
EX2
Figure 10-13. Delay When Updating the CCPL
The exact calculation of the time to recognize and process a pending interrupt after modifying the CCPL is
measured from the decode of instruction n1, which modifies CCPL (the beginning of cycle #4 in
Figure 10-13), to the first decode cycle of the first instruction that is fetched from the vector table after a
pending interrupt is recognized (beginning of cycle #13):
Delay =Execution time of instruction n1
+ 3 clock cycles for n1 to reach the end of the Execute phase
+ 1 clock cycle for arbitration with updated CCPL
+ remaining execution time of “Instruction at Int Req” (see following discussion)
+ 3 clock cycles for NOPs forced into pipeline
+ any pipeline core stalls due to data memory dependencies or wait states for p0 and p1
+ wait states due to program fetches of n2 through n5
– 1 clock cycle if n1 is a 2-cycle instruction that writes an immediate to the SR
In the preceding equation, the “Instruction at Int Req” is defined as the instruction in the Instruction
Decode stage of the pipeline when the pending interrupt is at the Int Req stage of the pipeline. In this
example, the instruction is n6. The “remaining execution time of ‘Instruction at Int Req’” is the number of
cycles from the time that the interrupt request reaches the Int Req stage for the pipeline to the time when
this instruction completes the pipeline’s decode stage.
Freescale Semiconductor
Instruction Pipeline
10-25
Instruction Pipeline
In the example in Figure 10-13 on page 10-25, the “remaining execution time” is 1 cycle. If n6 is a 2-cycle
instruction with its first decode cycle in cycle #9, the remaining execution time is 2 cycles. If a 2-word,
2-cycle instruction is contained in n5 and n6, the remaining execution time is 1 because there is only 1
remaining instruction cycle once the Int Req takes place.
The preceding timing calculation also applies when pending interrupts are already waiting and interrupts
are enabled by instruction n1.
10.4 Pipeline Dependencies and Interlocks
The pipeline is normally transparent to the user. However, there are certain instruction sequences that can
cause the pipeline to stall, affecting program execution. Most of these pipeline dependencies and resulting
interlocks occur because the result of an operation occurring very deep in the pipeline is used by the
immediately following instructions that are in earlier stages in the pipeline. Dependencies and interlocks
can also occur when there is contention for an internal resource, such as the status register (SR).
There are three methods for handling pipeline dependencies:
1. Hardware interlocking—the DSC automatically stalls the pipeline 1 or more cycles
2. Handling by development tools—the assembler automatically inserts NOP instructions
3. Instruction sequence restrictions—the instruction sequence is not allowed
In the first case, dependencies are detected in hardware, and the pipeline automatically stalls for the
required number of cycles. In the second case, the DSC does not stall the pipeline; rather, the assembler
issues a warning and inserts the appropriate number of NOP instructions between the dependent
instructions. In the third case, the assembler generates an error, and the sequence must be re-coded.
10.4.1 Data ALU Pipeline Dependencies
There are some cases within the data ALU unit where the nature of the pipeline can result in interlocks and
stalls, affecting the execution of a sequence of instructions. Data ALU dependencies fall into three
different categories:
•
Interlocks due to two-stage data ALU execution
•
Dependencies with OMR bits taking effect
•
Dependencies on reading status bits in the SR
In most cases, the pipeline will automatically stall when one of these dependencies occurs. In some
instances, NOP instructions are automatically inserted between instructions by the assembler to correct the
dependency.
One common dependency occurs when results that are calculated in the Execute 2 stage of the pipeline are
used as input operands in an immediately following two-stage instruction. Example 10-3 and Table 10-4
on page 10-27 illustrate this type of dependency.
10-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline Dependencies and Interlocks
Example 10-3. Data ALU Operand Dependencies
NOP
ADD
X0,A
SUB
A,B
MPY
B1,C1,D
MAC
X0,Y0,D
MPY
D1,X0,C
AND.W Y0,C
ASLL.W #3,C
ASLA R0
MPY
C1,D1,C
;
;
;
;
;
;
;
;
;
;
n1: Non-data ALU (restores to Normal state)
n2: Normal Execution (Execute phase)
n3: Normal Execution (Execute phase)
n4: Two-Stage (Execute and Execute 2)
n5: Two-Stage (Execute and Execute 2)
n6: Two-Stage (Execute and Execute 2)
n7: Late Execution (Execute 2 phase)
n8: Two-Stage (Execute and Execute 2)
n9: Non-data ALU (restores to Normal state)
n10: Two-Stage (Execute and Execute 2)
Operand dependencies occur in the example between n5 and n6 and between n7 and n8. Instruction n9
removes a potential dependency by resetting the pipeline to the Normal state. Note that no operand
dependency exists with the D register between n4 and n5 because it is used only in accumulation, not
multiplication. Note also that n7 completes in Execute 2, since the pipeline is forced Late by n6.
Table 10-4. Data ALU Operand Dependency Pipeline
Pipeline
Stage
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
P1
n1 n2 n3
n4
n5
n6
n7
n8
n9
—
n10
•
—
•
•
•
•
•
•
•
•
P2
n1 n2
n3
n4
n5
n6
n7
n8
—
n9
n10
—
•
•
•
•
•
•
•
•
IF
n1
n2
n3
n4
n5
n6
n7
—
n8
n9
—
n10
•
mpy and
—
asll asla mpy
Int Arbitr
Int Req
ID
nop add sub mpy mac
AG
—
OP2
—
add sub mpy mac
—
EX
—
add sub mpy mac
—
EX2
mpy and
—
add sub mpy mac
—
—
—
—
mpy and
—
mpy mac
mpy
—
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
mpy
•
•
•
•
asll asla mpy
—
asll
—
—
—
asll
—
mpy
•
•
•
—
asll
—
mpy
•
•
mpy and
It should be noted that there are no pipeline effects when the data ALU executes instructions using Late
Execution as long as the following instruction neither writes the results to memory nor depends on the
condition codes that are generated.
This situation is demonstrated in Example 10-4. As the associated pipeline in Table 10-5 on page 10-28
shows, there are no pipeline dependencies. Note that n2 and n3 in this example complete in the Execute 2
stage because the pipeline is placed in the Late state by n1.
Example 10-4. Case with No Data ALU Pipeline Dependencies
MAC
X0,Y0,A
SUB
Y1,A
ASL
A
TFRA R2,R1
MOVE.W A,X:(R0)+
ADD
X0,A
Freescale Semiconductor
;
;
;
;
;
;
n1:
n2:
n3:
n4:
n5:
n6:
performed in Execute and Execute 2
Late Execution (Execute 2 phase)
Late Execution (Execute 2 phase)
Non-data ALU (restores to Normal state)
(no dependency)
Normal Execution (Execute phase)
Instruction Pipeline
10-27
Instruction Pipeline
Table 10-5. Data ALU Pipeline with No Dependencies
Instruction Cycle
Pipeline Stage
P1
P2
IF
1
2
3
4
5
6
7
8
9
10
11
•
•
•
mac
sub
asl
tfra
mov
add
•
•
•
•
•
•
•
•
mac
sub
asl
tfra
mov
add
•
•
•
•
•
•
•
sub
asl
tfra
mov
add
•
•
•
•
•
•
mac
sub
asl
tfra
mov
add
•
•
•
•
•
mac
sub
asl
tfra
mov
add
•
•
•
•
mac
sub
asl
—
mov
add
•
•
•
mac
ID
AG
OP2
EX
mac
EX2
—
—
—
mov
add
•
•
mac
sub
asl
—
—
—
•
10.4.2 AGU Pipeline Dependencies
Dependencies that are similar to those presented for the data ALU can occur with the address generation
unit, affecting the execution of a sequence of instructions. Many pipeline dependencies are caused by the
fact that addresses are issued early in the pipeline (AG stage), while registers are written deeper within the
pipe (EX stage).
The most frequently occurring dependencies take place when an AGU register (R0–R5, N, or SP) is
modified using a move or bit-manipulation instruction. A dependency occurs if the same register is used
within the next 2 immediately following instruction cycles and if it is:
•
used as a pointer in an addressing mode.
•
used as an offset in an addressing mode.
•
used as an operand in an AGU calculation.
•
used in a TFRA instruction.
When these conditions occur, a hardware interlock occurs and the DSC automatically stalls the pipeline
1 or 2 cycles. This AGU dependency is demonstrated in Example 10-5.
Example 10-5. Pipeline Dependency with AGU Registers
MOVE.L A10,R0
MOVE.W X:(R0),X0
ADD
X0,B
; n1: Write AGU pointer register
; n2: Use same register as an address
; n3: Use value in x0 just read from memory
A pipeline interlock occurs between n1 and n2 because the address for the MOVE.W instruction (n2) is
formed at the Address Generation stage of the pipeline, which would normally occur at cycle #6 for n2 in
Table 10-6 on page 10-29. The MOVE.L instruction (n1), however, updates the R0 register very deep in
the pipeline—at cycle #7. Because the R0 register is available for use in cycle #8, interlocking hardware on
the core automatically stalls the core for 2 cycles.
10-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline Dependencies and Interlocks
Table 10-6. AGU Write Dependency Pipeline
Pipeline
Stage
P1
P2
IF
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
11
•
•
•
n1
n2
n3
n4
n5
—
—
•
•
•
•
•
•
•
n1
n2
n3
n4
—
—
n5
•
•
•
•
•
•
n1
n2
n3
—
—
n4
n5
•
•
•
•
•
mov.l
—
—
mov.w
add
n4
n5
•
•
•
•
mov.l
—
—
mov.w
add
n4
n5
•
•
•
mov.l
—
—
mov.w
add
n4
n5
•
•
mov.l
—
—
mov.w
add
n4
n5
•
—
—
—
—
—
n4
n5
ID
AG
OP2
EX
EX2
If a dependency is caused by a modification of the N3 or M01 registers by a move or bit-manipulation
instruction, or if a bit-manipulation operation is performed on the N register, the DSC does not
automatically stall the pipeline. Instead, the development tools automatically insert the appropriate number
of NOP instructions to ensure that the program executes as intended.
There are some special cases where there are no AGU dependencies. There is no dependency when
immediate values are written to the address pointer registers—R0–R5, N, and SP. Similarly, there are no
dependencies when a register is loaded with a TFRA instruction. Example 10-6 and Table 10-7 on
page 10-30 illustrate this case.
Example 10-6. Case Without AGU Pipeline Dependencies
MOVEU.W#$4,R0
MOVE.W X:(R0),A
; n1: Write AGU pointer register with immediate
; n2: Use same register to access memory
MOVE.W #3,R1
ADDA R0,R1
MOVE.W X:(R1)-,B
; n3: Write AGU pointer register with immediate
; n4: Use same register in AGU calculation
; n5: Use same register to access memory
TFRA R1,R2
MOVE.W X:(R2),C
; n6: Copy one AGU pointer register to another
; n7: Use same register to access memory
Freescale Semiconductor
Instruction Pipeline
10-29
Instruction Pipeline
Table 10-7. AGU Pipeline With No Dependencies
Pipeline
Stage
P1
P2
IF
Instruction Cycle
1
2
3
4
5
6
7
8
9
10
11
•
•
•
n1
n2
n3
n4
n5
n6
n7
•
•
•
•
•
•
•
n1
n2
n3
n4
n5
n6
n7
•
•
•
•
•
•
n1
n2
n3
n4
n5
n6
n7
•
•
•
•
•
movu
mov
mov
add
mov
tfra
mov
•
•
•
•
movu
mov
mov
add
mov
tfra
mov
•
•
•
movu
mov
mov
add
mov
tfra
mov
•
•
movu
mov
mov
add
mov
tfra
mov
•
—
—
—
—
—
—
—
ID
AG
OP2
EX
EX2
10.4.3 Instructions with Inherent Stalls
There is an infrequently used class of move instructions that introduce stalls into the pipeline due to
pipeline effects. The assembler will issue a warning when any of these instructions are encountered.
The pipeline automatically inserts 2 stall cycles when move instructions that satisfy all of the following
characteristics are executed:
•
The instruction is a move from a register to data memory.
•
The source of the move is an AGU register (R0–R5, N, or SP).
•
The AGU register that is used for the effective address is the same AGU register that is used as the
source of the move instruction.
•
The addressing mode is one of the three post-update addressing modes:
— Post-increment
— Post-decrement
— Post-update by offset register
The inserted stall cycles effectively make these instructions 3-cycle instructions. The stalls are inserted so
that the register is updated by the addressing mode after being used as the source register in the move
instruction. Example 10-7 shows three instructions that fall into this category.
Example 10-7. MOVE Instructions That Introduce Stalls
MOVE.W R1,X:(R1)+
; R1 stored with R1 post-update
MOVE.W N,X:(N)-
; N stored with N post-update
MOVE.W R5,X:(R5)+N
; R5 stored with R5 post-update
This type of dependency occurs whenever an address pointer register is used as the source in a store
instruction, while, within the same instruction, the same pointer is being updated (modified) by an
addressing mode. There is no dependency if the register is used as a destination in the move instruction.
Example 10-8 on page 10-31 shows this case and other instances where there is no dependency.
10-30
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Pipeline Dependencies and Interlocks
Example 10-8. Instructions with No Stalls
MOVE.W R2,X:(R1)+
; R2 stored with R1 post-update
MOVEU.WX:(N)-,N
; N loaded with N post-update
MOVE.W R5,X:(R5+N)
; R5 stored with no R5 post-update
10.4.3.1 Dependencies with Hardware Looping
There are a few dependencies that occur when one is working with the DO, DOSLC, and REP hardware
looping mechanisms. In particular, a dependency occurs when the LC register is loaded prior to executing
one of the hardware looping instructions. Due to the architecture of the instruction pipeline, none of the
hardware looping instructions can be executed immediately after a value is placed in the LC register.
Example 10-9 shows a code sequence that has such a dependency.
Example 10-9. Dependency with Load of LC and Start of Hardware Loop
MOVEU.WR0,LC
DOSLC LABEL
MOVE.W X:(R3)+,X0
ADD
X0,B
;
;
;
;
n1: Write to LC immediately followed by:
n2: 3-cycle, 2-word DOSLC loop
n3
n4
LABEL
In the code sequence in Example 10-9, the value that is loaded into LC in the first instruction is not
available when it is needed by the DOSLC instruction: 2 more cycles are required before it is available in
the right place in the pipeline.
The solution to this problem is to insert instructions that require at least 2 cycles to execute between the
load of LC and the DOSLC instruction. If instructions are not inserted to correct this problem, the
assembler will insert as many NOP instructions as necessary to ensure that the code executes correctly.
Freescale Semiconductor
Instruction Pipeline
10-31
Instruction Pipeline
10-32
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Chapter 11
JTAG and Enhanced On-Chip
Emulation (Enhanced OnCE)
The DSP56800E Family includes extensive integrated support for application software development and
real-time debugging. Two modules, the Enhanced On-Chip Emulation module (Enhanced OnCE) and the
core test access port (TAP, commonly called the JTAG port), work together to provide these capabilities.
Both are accessed through a common JTAG/Enhanced OnCE interface. Using these modules allows the
user to insert the DSC chip into a target system while retaining debug control. This capability is especially
important for devices without an external bus, since it eliminates the need for a costly cable to bring out the
footprint of the chip, as is required by a traditional emulator system.
The DSP56800E Enhanced OnCE module is a Freescale-designed module that is used to develop and
debug application software used with the chip. This module allows non-intrusive interaction with the DSC
and is accessible either through the pins of the JTAG interface or by software program control of the
DSP56800E core. Among the many features of the Enhanced OnCE module is the support for data
communication between the DSC chip and the host software development and debug systems in real-time
program execution. Other features allow for hardware breakpoints, the monitoring and tracking of program
execution, and the ability to examine and modify the contents of registers, memory, and on-chip
peripherals, all in a special debug environment. No user-accessible resources need to be sacrificed to
perform debugging operations.
The DSP56800E JTAG port is used to provide an interface for the Enhanced OnCE module to the DSC
JTAG pins. This TAP controller is designed to be incorporated into a chip multi–JTAG TAP Linking
Module (JTAG TLM) system. The JTAG TLM is a dedicated, user-accessible, test access port (TAP)
system that is compatible with the IEEE Standard 1149.1a-1993, IEEE Standard Test Access Port and
Boundary-Scan Architecture.
This chapter presents an overview of the capabilities of the JTAG and Enhanced OnCE modules. Because
their operation is dependent upon the architecture of a specific DSP56800E device, the exact
implementation is necessarily device dependent.
11.1 Enhanced OnCE Module
The Enhanced OnCE module provides emulation and debug capability directly on the chip, eliminating the
need for expensive and complicated stand-alone in-circuit emulators (ICEs). The Enhanced OnCE module
permits full-speed, non-intrusive emulation on a user’s target system. This section describes the Enhanced
OnCE emulation environment for use in debugging real-time embedded applications.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-1
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
Because emulation capabilities are tied to the particular implementation of a DSP56800E–based device,
the user’s manual for the appropriate device should be consulted for complete details on implementation
and supported functions.
11.1.1 Enhanced OnCE Module Capabilities
The capabilities of the Enhanced OnCE module include the following:
•
Examine or modify the contents of any core or memory-mapped peripheral register
•
Examine and modify program or data memory
•
Step at full speed on one or more instructions
•
Save a programmable change-of-flow instruction capture to the trace buffer
•
Display the contents of the real-time instruction trace buffer
•
Allow the transfer of data between the core and external host in real-time program execution by
using peripheral-mapped transmit and receive registers
•
Access Enhanced OnCE registers and programming model by either the DSP56800E software or
the debugging system through the JTAG port
•
Provide status of Enhanced OnCE events in a status register or on an output pin from the core
•
Count a variety of events including clock cycles and instructions executed
•
Enter debug mode in any of the following ways:
— microprocessor instruction
— the actions of the Enhanced OnCE module
— the core JTAG port
— a special debug request input pin to the core
•
Interrupt or break into debug mode on program memory addresses (fetch, read, write, or read and
write access)
•
Interrupt or break into debug mode on accesses to data memory or on-chip peripheral registers
(read, write, or read and write access) and for byte, word, or long data type accesses
•
Save or restore the current state of the chip’s pipeline
•
Display the contents of the real-time instruction trace buffer
•
Return to normal user mode from debug mode
These capabilities will be explained in more detail in the following sections. Additional debugging and
emulation capabilities may be provided on particular DSP56800E-based devices. Consult the user’s
manual for the particular device for more information.
11.2 Enhanced OnCE System Level View
A system level view of the Enhanced OnCE module resources is shown in Figure 11-1. Although the
Enhanced OnCE module is currently contained in the DSP56800E core, they are conceptually shown
separate in this picture for a simpler understanding of the debug port capabilities.
In this conceptual diagram, the DSP56800E core contains the core’s execution units, core register files, etc.
It is this block that executes DSP56800E instructions. The Enhanced OnCE module can be viewed as a
11-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE System Level View
separate module which acts concurrently with the DSP56800E core. This module contains its own
programming model, simple Enhanced OnCE instructions, and its own units:
•
Enhanced OnCE Control Unit, which contains:
— Enhanced OnCE Control
— Step Counter
— Realtime Data Transfer Unit
•
Breakpoint Unit
•
Trace Buffer
Program
Memory
Data
Memory
DSP56800E
IP-Bus
Bridge
System Buses
Core
PIC
EOnCE
Control
Breakpoint
Unit
Trace
Buffer
JTAG
JTAG
Pins
IP-BUS
Figure 11-1. DSP56800E On-Chip System with Debug Port
After being properly initialized and programmed for breakpoint triggering and associated actions, the
EOnCE module operates in parallel with the DSP56800E core. As the DSP56800E core is executing
instructions, the Enhanced OnCE module can do the following:
•
Receive new Enhanced OnCE commands
•
Read / Write Enhanced OnCE registers through the JTAG interface
(can also be accessed through the DSP56800E core’s system buses)
•
Monitor DSP56800E buses for breakpoint conditions
•
Capture DSP56800E program addresses when appropriate in the Trace Buffer
•
Generate any of several different Enhanced OnCE interrupt requests
•
Halt the DSP56800E core upon a certain debug event so it enters the Debug processing state
If the DSP56800E core has been halted by entering the Debug processing state, the Enhanced OnCE module
is still capable of receiving new commands as well as reading or writing any of the Enhanced OnCE
registers.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-3
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
NOTE:
The Enhanced OnCE blocks shown in Figure 11-1 (EOnCE Control,
Breakpoint Unit, and Trace Buffer) are actually located inside the
DSP56800E core. The figure is only conceptual and was drawn this way
to better demonstrate how these individual blocks are used in a
DSP56800E system.
11.3 Accessing the Enhanced OnCE Module
Resources in the Enhanced OnCE module can be accessed either through the JTAG port or under software
program control from the DSC core. These two methods allow debugging activity to be controlled either
by a host development system or by a program that is executing on the DSP56800E device. The two
methods are discussed below.
11.3.1 External Interaction via JTAG
Development and debugging systems can control Enhanced OnCE debugging actions by communicating
with the Enhanced OnCE via the JTAG port. All of the Enhanced OnCE resources are available serially
through the normal JTAG access protocol.
When interacting via JTAG, the DSP56800E JTAG and Enhanced OnCE modules are tightly coupled. The
interface for both modules is handled by the JTAG port, which communicates with the host software
development and debug systems. Figure 11-2 shows a block diagram of the JTAG/Enhanced OnCE
modules and the JTAG terminals used in the external interface.
The JTAG acts as an external interface controller for the Enhanced OnCE, transparently passing all
communication between the Enhanced OnCE and the host development system. The JTAG port enables
interaction with the debug capabilities provided by the Enhanced OnCE, and its external serial interface is
used by the Enhanced OnCE module for sending and receiving debugging commands and data.
A special JTAG instruction is executed to enable communication with the Enhanced OnCE module. While
Enhanced OnCE communication is active, the JTAG module transparently transfers all data that is
received on the JTAG port to the Enhanced OnCE module.
The JTAG port can also act as a completely independent module. When it is disabled, it has no impact on
the function of the core.
11-4
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Accessing the Enhanced OnCE Module
Enhanced OnCE
Enhanced OnCE
Command
Status & Control
JTAG
CORE_TDI
CORE_TDO
TMS
TCK
TLM_RES_B
Enhanced OnCE
Instruction
Queue
Test
Access
Port
Controller
PAB
Step Logic
CORETAP_EN
CORE_TLM_SEL
Step Counter
TX/RX Logic
Transmit Register
Receive Register
Trace Buffer
(Eight stages)
PAB
XAB1
Breakpoint Logic
PAB
Event Counter
CDBR / CDBW
Figure 11-2. JTAG/Enhanced OnCE Interface Block Diagram
11.3.2 Core Access to the Enhanced OnCE Module
The core can also access the Enhanced OnCE module directly executing DSP56800E instructions which
access the Enhanced OnCE module as memory mapped registers. This technique operates independent of
the JTAG port.
Access to the Enhanced OnCE module from the DSC core is enabled through a set of memory-mapped
registers. All of the Enhanced OnCE resources are available through the memory mapped registers,
allowing access to the port via normal instruction execution. When accessed in this manner, there is no
need to access the port via JTAG.
Core access provides the ability to initialize the Enhanced OnCE module, use its resources, and monitor its
actions under program control. It also allows data to be uploaded or downloaded between the core and each
of the four Enhanced OnCE submodules. Both polled and interrupt driven communication between the
core and the Enhanced OnCE module is supported where appropriate.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-5
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
An unlocking sequence must first be executed by the core to gain access to the Enhanced OnCE module.
This prevents accidental access to the Enhanced OnCE resources. Core access to the Enhanced OnCE
module can optionally be disabled via the JTAG port to prevent programs from affecting the Enhanced
OnCE module’s operation.
11.3.3 Other Supported Interactions
The DSP56800E supports two instructions, DEBUGEV and DEBUGHLT, that will trigger actions in the
Enhanced OnCE module when executed by the DSP56800E core. The DEBUGEV instruction causes a
debugging event to be generated, similar to the generation of a breakpoint trigger. The DEBUGHLT
instruction is used to halt the core, placing it in the Debug processing state, where state information can be
easily read and modified.
11.4 Enhanced OnCE and the Processing States
The DSP56800E core supports six different processing states (see Table 11-1).
Table 11-1. Processing States
State
Description
Normal
The state of the core where instructions are normally executed.
Reset
The state where the core is forced into a known reset state. The first program instruction
is fetched upon exiting this state.
Exception
The state of interrupt processing, where the core transfers program control from its current location to an interrupt service routine using the interrupt vector table.
Wait
A low power state where the core is shut down but the peripherals and the interrupt
machine remain active.
Stop
A low power state where the core, the interrupt machine, and most (if not all) of the
peripherals are shut down.
Debug
The state where the core is halted and all registers in the Enhanced On-Chip Emulation
(EOnCE) port of the processor are accessible for program debug.
11.4.1 Using the Debug Processing State
The Debug processing state is a state where the core is halted, breakpoints and other resources can be
initialized and setup for debugging, and on-chip registers and memory locations can be examined and
modified. The chip is often placed in the Debug processing state to initialize the Enhanced OnCE module
for a debug system. It is also possible for the core to enter the Debug processing state immediately upon
exiting reset to setup a debug session before the core begins executing instructions.
Any of the following can place the core in the Debug processing state:
•
Hardware reset with JTAG DEBUG_REQUEST in the JTAG Instruction Register (IR)
•
JTAG DEBUG_REQUEST placed in the JTAG IR during
— STOP mode
— WAIT mode
— wait states
•
11-6
Pulling the core debug_req_b pin low for three peripheral clock cycles
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
•
Execution of the DEBUGHLT instruction while the EOnCE is powered up
•
Step Counter expires while configured for debug request
•
Trace Buffer is full and configured for debug request
•
Breakpoint Unit Triggers occurs when programed for debug request
11.4.2 Debugging and the Other Processing States
It is not necessary, however, to place the core in the Debug processing state to initialize the module. An
alternative technique is to first setup the desired Enhanced OnCE resources and then to enable these
resources. This can either be done through the JTAG port or through Core access via setup routines located
in an application, typically executed in the Normal processing state.
The Enhanced OnCE module also has the capability to generate interrupt requests in response to difference
debug events, each with its own dedicated interrupt vectors in the DSP56800E interrupt vector table. The
Enhanced OnCE exception trap is available to the user so that when a debug event is detected, an interrupt
can be generated and the program can initiate the appropriate handler routine. This allows the core to
perform many different actions in response to Debug events without halting the core. Instead, the event is
serviced by executing a dedicated interrupt service routine.
NOTE:
Care must be taken when the core is in the Stop processing states. In this
state, all core clocks are disabled and it is not possible to access the
Enhanced OnCE module. The JTAG interface provides the means of
polling the device status (sampled in the capture-IR state). The core JTAG
TAP will bring the core out of Stop or Wait modes when
DEBUG_REQUEST is decoded in the TAP IR. A small amount of
additional power above the minimum possible will be expended by the
core TAP logic if the core TAP is utilized during Stop mode.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-7
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.4.3 Enhanced OnCE Module Architecture
The Enhanced OnCE module is composed of several submodules, each of which performs a different task:
•
Command, status, control, instruction execution
•
Breakpoint unit
•
Step counter
•
Change-of-flow trace buffer
•
Enhanced OnCE transmit and receive registers
Together, these submodules provide a full-featured emulation and debug environment. External
communication with the Enhanced OnCE module is handled via the JTAG port, although it operates
independently. The operations of the Enhanced OnCE module can occur independently of the main
DSP56800E core logic, requiring no core resources. Alternatively, DSP56800E software can directly
program, control, and communicate with the Enhanced OnCE module.
11.4.3.1 Command, Status, and Control
The command, status, and control portion of the Enhanced OnCE module handles the processing of
emulation and debugging commands from a host development system. Communication with the external
host system are provided by the JTAG port module and passed transparently through to this logic, which is
responsible for coordinating all emulation and debugging activity. This enables emulation and debug
processing to occur independently of the main DSP56800E processor core instructions in a non-intrusive
fashion. The Enhanced OnCE module can also enable the core to enter debug mode.
Status bits can be examined to determine which source caused the processor was halted. Additional bits are
provided to report the condition of the Trace Buffer.
11-8
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
11.4.3.2 Breakpoint Unit
Traditionally, processors have set a breakpoint in program memory by replacing the instruction at the
breakpoint address with an illegal instruction that causes a breakpoint exception. This technique is limiting
in that breakpoints can only be set in RAM at the beginning of an opcode and not on an operand. In
addition, breakpoints can never be set on data memory locations.
The DSP56800E Enhanced OnCE breakpoint unit provides a breakpoint unit with hardware trigger
generation blocks containing address comparators for setting breakpoints on program or data memory
accesses. Breakpoints can be set on program ROM as well as program RAM locations.
The DSP56800E Enhanced OnCE breakpoint unit includes two trigger modules, a 16-bit counter, and
combining logic to trigger breakpoints from a substantially wider variety of conditions than traditional
processors. These conditions include accessing a particular memory location or value, the occurrence of a
particular number of events, or a combination of these conditions. In response to a breakpoint trigger, the
breakpoint unit can generate an interrupt, control trace buffer or counter operation, or halt the core.
Figure 11-3 is a diagram of the breakpoint unit.
XAB1
[0:PAB]
[0:PAB] CDBR/CDBW
Read/Write
Trigger 1
Fetch
Trigger 2
32-bit
Mask
DEBUGEV
instruction
Combining
Logic
Overflow or
saturation
Select
Action
Breakpoint
Interrupt
Start
Trace
Buffer
16-Bit Counter
Output
Action
Halt
Trace
Buffer
Figure 11-3. Breakpoint Unit Block Diagram
The Breakpoint Unit capabilities will be demonstrated in detail in Section 11.4.4, “Effectively Using the
Debug Port,” on page 11-13.
11.4.3.2.1 Trigger Blocks
The first trigger block, shown in Figure 11-4, can be programmed for program fetches, reads, writes or
memory accesses. It can also be programmed for data memory reads, writes, or accesses. Triggering is also
possible for on-chip peripheral register accesses, since these registers are implemented as
data-memory-mapped registers.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-9
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
XAB1
Read/Write
PAB
Memory Address
Multiplexer
Fetch
JTAG
Breakpoint 1
Address Register
24
Core Peripheral Bus
24
Comparator
Match 1
Figure 11-4. Trigger 1 Logic
The second trigger block, shown in Figure 11-5, can be programmed for program fetches, or data memory
reads, writes, or accesses on 8, 16, or 32-bit data. It is also possible to mask bits in the second trigger block
to only examine desired bit fields.
CDRB/CDBW
PAB
Read/Write
Fetch
Memory Address
Multiplexer
Breakpoint 2
Address Register
24
JTAG
Core Peripheral Bus
24
Breakpoint 2
Mask Register
Comparator
JTAG
Core Peripheral Bus
Mask
Optional Inverter
Match 2
Figure 11-5. Trigger 2 Logic
11.4.3.2.2 16-bit Counter
The breakpoint unit contains a 16-bit counter which can be programmed to act in one of two different
modes. In triggering mode, the counter is used to count occurrences of a desired trigger condition. In
capture mode, the counter can instead independently count clock cycles or instructions executed between
two points of interest.
In capture mode, the breakpoint counter can also be cascaded with the step counter to create a 40-bit
counter for longer time measurements.
11-10
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
11.4.3.2.3 Combining Logic
The breakpoint unit combining logic supports combinations of breakpoints. This allows for the execution
of OR and AND operations as well as the sequencing of more than one breakpoint.
11.4.3.3 Step Counter
This submodule also provides the capability for full-speed instruction stepping. A 24-bit instruction step
counter provides for up to 16,777,216 instructions to be executed at full speed before the processor core is
interrupted (or halted) and enters the Debug processing state. This capability allows the user to single step
through a program or to execute whole functions at a time.
This counter can be used very effectively in combination with the Breakpoint Unit capabilities for more
complex debugging scenarios. This will be demonstrated in detail in Section 11.4.4, “Effectively Using the
Debug Port,” on page 11-13.
11.4.3.4 Change-of-Flow Trace Buffer
To ease debugging activity and to help keep track of program flow, a read-only buffer is provided that
tracks the change-of-program-flow execution history of an application. It can store the address of the most
recent change-of-flow instruction as well as the addresses of the previous seven change-of-flow
instructions. The trace buffer is intended to provide a snapshot of the recent execution history of the
DSP56800E processor core. This buffer is capable of capturing any combination of the following
execution flow events:
•
Interrupts—captures the address of the interrupt vector and the target address of returns
•
Subroutines—captures the target address of JSR and BSR instructions
•
Conditional branches, whether taken or not, forward or backward—captures the target addresses for
the Bcc, Jcc, BRSET, and BRCLR instructions
Sequential program flow can be assumed to have occurred between the recorded instructions, so it is
possible for the user to reconstruct the program execution flow extending back quite a number of
instructions. To complete the execution history, a circular pointer is used to indicate the location of the
buffer that holds the address of the most recent change-of-flow instruction. The pointer is then
decremented while reading the eight buffer locations to obtain a sequential trace of these instructions back
in time.
The Enhanced OnCE module provides flexible control over the trace buffer. Starting and stopping capture
into the buffer is programmable, so capture only occurs when it is needed. Once the eight-position buffer is
filled, there are several programmable options for what action the Enhanced OnCE module takes:
•
No action—Buffer continues to capture change of flows.
•
Halt buffer—Buffer capture is stopped.
•
Enter debug—Buffer capture is stopped and core enters debug mode.
•
Interrupt—Buffer capture is stopped and an interrupt occurs.
11.4.3.5 Realtime Data Transfer Unit
The Realtime Data Transfer Unit enables the user to transmit data from the DSP56800E processor core to
the external host through the JTAG port, and enables the core to receive data from the external host, in
real-time program execution.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-11
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
32
1
32-bit
RX Data Shifter
Serial Input
1
32-bit
TX Data Shifter
Serial Output
32
TXRX
Status Register
TX Data
Interrupt
RX Data
Interrupt
Core
Data Buses
Figure 11-6. Realtime Data Transfer Unit
The 32-bit transmit and receive registers are memory mapped in the core’s data memory. The core writes
to the transmit register and reads the receive register in parallel via the DSP56800E instruction set, and the
host writes to the receive register and reads the transmit register serially through the JTAG interface.
Communication between these registers and the core can be either polled or interrupt driven. Status bits
indicate when the transmit or the receive portion need servicing. Similarly, interrupts can be enabled
separately for the transmit and receive portions, signalling to the core that the Realtime Data Transfer Unit
should be serviced.
11-12
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
11.4.4 Effectively Using the Debug Port
Different features in the above blocks of the Enhanced OnCE module can be used together and
programmed in different manners for handling complex as well as simpler debugging problems. This
section demonstrates how to best program the above modules and what triggering is available. It also
shows what actions are allowed once a particular debug event or set of events has occurred.
11.4.4.1 Using the Step Counter
The 24-bit step counter can be used in the two manners presented below. If not needed for either of these,
it can be used to create a 40-bit Capture Counter as shown in Section 11.4.4.3, “Capture Counter,” on page
11-20.
11.4.4.1.1 Usage upon Exiting the Debug Processing State
In its simplest usage, the Step Counter can be used for full speed execution of a programmable number of
clock cycles before performing an action. In this case, the Breakpoint Unit still generate a Breakpoint Unit
Trigger for everything except halting the core and entering the Debug processing state. This is the
configuration used, for example, when single stepping.
Exit Debug State
24-Bit
Step Counter
Select
Action
Halt Trace
Buffer
Halt
Core
Figure 11-7. Step Counter — Started upon Exiting Debug State
In another simple usage, the Step Counter can be used for full speed execution of a programmable number
of clock cycles before performing an action. In this case, the Breakpoint Unit can now generate a
Breakpoint Unit Trigger for halting the core upon this trigger and entering the Debug processing state.
Exit Debug State
24-Bit
Step Counter
Breakpoint Unit Trigger
Select
Action
Halt
Core
Figure 11-8. Step Counter — Started upon Exiting Debug State with Breakpoint Active
11.4.4.1.2 Step Counter Actions
Table 11-7 lists the possible actions when using the Step Counter.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-13
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.4.4.1.3 Other Step Counter Configurations
The Step Counter. can also be configured to work with the Breakpoint Unit, covered in Section 11.4.4.2.3,
“Combining the Breakpoint Unit with the Step Counter,” on page 11-19, as well as the Capture Counter as
discussed in Section 11.4.4.3.3, “Using the Capture Counter with the Step Counter,” on page 11-23.
Table 11-2. Step Counter Operation
Start Step Counter
Case SC-1
Exit Debug State
Trigger for Step Counter Action
Step Counter reaches zero
Case SC-2
Action Performed
Enter Debug state
Halt Trace Buffer Capture when
Step Counter reaches zero.
Case SC-3
Step Counter reaches zero OR
Breakpoint Unit Trigger arrives
Enter Debug state
11.4.4.2 Using the Breakpoint Unit
The Breakpoint Unit is used to generate trigger(s) for any one of the following:
•
Traditional breakpointing
•
Start and/or Stop triggers for Trace Buffer Capture
•
Start and/or Stop triggers for measuring cycles executed in the Capture Counter
This section covers the first two uses. Triggers for the Capture Counter will be covered in Section 11.4.4.3,
“Capture Counter,” on page 11-20.
The breakpoint triggering capabilities can be examined using the block diagram in Figure 11-9. This unit is
capable of generating two triggers. There are also inputs for DEBUGEV instruction execution as well as an
overflow condition within the core. These four different inputs are then combined in the Combining Logic
to get the final Breakpoint Unit Trigger, which can then be used to perform one of several different actions
or can also be passed to a different block such as the step counter.
11-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
XAB1
[0:PAB]
[0:PAB] CDBR/CDBW
Read/Write
Fetch
Trigger 1
Trigger 2
32-bit
Mask
DEBUGEV
instruction
Combining
Logic
Overflow or
saturation
Select
Action
Breakpoint
Interrupt
Start
Trace
Buffer
16-Bit Counter
Output
Action
Halt
Trace
Buffer
Figure 11-9. Breakpoint Unit Block Diagram
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-15
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.4.4.2.1 Listing the Breakpoint Unit Triggers Available
The full set of breakpoint triggers which can be created by this unit is shown in Table 11-4 and Table 11-5,
where Table 11-4 contains most of the unit’s triggering capability and is combined with the capabilities of
Table 11-5 to get the final Breakpoint Unit Trigger generated from the unit.
The notation for these tables is explained below:
Table 11-3. Notation used in Breakpoint Unit Triggering
Notation
Description
PAB-1
Trigger 1 configured to look for match on the PAB bus. On 1st occurrence of a
match, the trigger is asserted.
PAB-1*
Trigger 1 configured to look for match on the PAB bus. On Nth occurrence of a
match, the trigger is asserted, where N is the programmed 16-bit counter value.
XAB1
Trigger 1 configured to look for match on the XAB1 bus. On 1st occurrence of a
match, the trigger is asserted.
XAB1*
Trigger 1 configured to look for match on the XAB1 bus. On Nth occurrence of a
match, the trigger is asserted, where N is the programmed 16-bit counter value.
PAB-2
Trigger 2 configured to look for match on the PAB bus. On 1st occurrence of a
match, the trigger is asserted.
PAB-2*
Trigger 2 configured to look for match on the PAB bus. On Nth occurrence of a
match, the trigger is asserted, where N is the programmed 16-bit counter value.
CDB — Data Value
Trigger 2 configured to look for an 8-bit, 16-bit, or 32-bit match on a data value
on the CDB bus. In addition, any bits in the value can be masked to look at only
a portion of the data value. On 1st occurrence of a match, the trigger is asserted.
Fetch
The trigger is only asserted on instruction fetches from program memory. It is
not asserted if data is accessed from the program memory.
Access
The trigger is only asserted on data accesses from memory. It is not asserted for
instruction fetches from the memory.
F/R/W/A
The trigger is asserted on any access to the memory — instruction fetch, data
read, write, or access.
R/W/A
The trigger is asserted on any data access to the memory — data read, write, or
access.
(expression)*
The trigger is asserted on the Nth occurrence of detecting the expression. This
is used when breakpoints are ORed or ANDed together.
expr1 OR expr2
11-16
The trigger is asserted when “expr1” occurs OR when “expr2” occurs. The
occurrence of either asserts the trigger.
expr1 AND expr2
The trigger is asserted when “expr1” occurs at the same time as when “expr2”
occurs. Both must occurrence for the trigger to be asserted. This is particularly
useful for examining a data value at a particular location in data memory.
expr1 ==> expr2
“expr1” must first occur, followed by “expr2”. When this occurs, the condition
becomes true and the trigger is asserted.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
Table 11-4. First Part of Breakpoint Unit Trigger(s)— 16-bit Counter Available for Triggering
First Breakpoint
Trigger
Op
Second Breakpoint
Trigger
Comments
Single Triggers
PAB-1* — F/R/W/A
(none)
Nth occurrence of F/R/W/A on PAB bus, Trigger 1
XAB1* — R/W/A
(none)
Nth occurrence of R/W/A on XAB1 bus, Trigger 1
ORed Triggers
PAB-1 — Fetch
OR
PAB-2* — Fetch
1st Fetch on PAB Trig1, or Nth Fetch Trig2
(PAB-1 — Fetch
OR
PAB-2 — Fetch)*
Nth occur, (1st F on PAB Trig1, or 1st F Trig2)
PAB-1* — Access
OR
PAB-2 — Fetch
Nth Access on PAB Trig1, or 1st Fetch Trig2
PAB-1 — Access
OR
PAB-2* — Fetch
1st Access on PAB Trig1, or 1st Fetch Trig2
(PAB-1 — Access
OR
PAB-2 — Fetch)*
Nth occur, (1st A on PAB Trig1, or 1st F Trig2)
PAB-2* — Fetch
OR
XAB1 — Access
Nth F on PAB Trig2, or 1st A on XAB1
PAB-2 — Fetch
OR
XAB1* — Access
1st F on PAB Trig2, or Nth A on XAB1
CDB — Data Value)*
Nth occur, (1st R/W/A XAB1 Trig1 and CDB Trig2)
ANDed Triggers
(XAB1 — R/W/A
AND
Sequenced Triggers
PAB-1* — Fetch
==>
PAB-2 — Fetch
Nth F on PAB Trig1 followed by 1st F PAB Trig2
PAB-2 — Fetch
==>
PAB-1* — Fetch
1st F on PAB Trig2 followed by Nth F PAB Trig1
PAB-1* — Access
==>
PAB-2 — Fetch
Nth A on PAB Trig1 followed by 1st F PAB Trig2
PAB-1 — Access
==>
PAB-2* — Fetch
1st A on PAB Trig1 followed by Nth F PAB Trig2
PAB-2* — Fetch
==>
PAB-1 — Access
Nth F on PAB Trig2 followed by 1st A PAB Trig1
PAB-2 — Fetch
==>
PAB-1* — Access
1st F on PAB Trig2 followed by Nth A PAB Trig1
XAB1* — R/W/A
==>
PAB-2 — Fetch
Nth R/W/A XAB1 Trig1 followed 1st F PAB Trig2
XAB1 — R/W/A
==>
PAB-2* — Fetch
1st R/W/A XAB1 Trig1 followed Nth F PAB Trig2
PAB-2* — Fetch
==>
XAB1 — R/W/A
Nth F PAB Trig2 followed 1st R/W/A XAB1 Trig1
PAB-2 — Fetch
==>
XAB1* — R/W/A
1st F PAB Trig2 followed Nth R/W/A XAB1 Trig1
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-17
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
Table 11-4. First Part of Breakpoint Unit Trigger(s)— 16-bit Counter Available for Triggering
First Breakpoint
Trigger
Op
Second Breakpoint
Trigger
Comments
Generation of Two Triggers — Start and Stop
PAB-1 — Fetch
=> Start Trace Buffer
—
PAB-2 — Fetch
=> Stop Trace Buffer
Start Trace Buffer on 1st Fetch on PAB Trig1 and
Stop Trace on 1st Fetch on PAB Trigger 2
PAB-1 — Access
=> Start Trace Buffer
—
PAB-2 — Fetch
=> Stop Trace Buffer
Start Trace Buffer on 1st Access on PAB Trig1
and Stop Trace on 1st Fetch on PAB Trigger 2
PAB-2 — Fetch
=> Start Trace Buffer
—
PAB-1 — Access
=> Stop Trace Buffer
Start Trace Buffer on 1st Fetch on PAB Trig2 and
Stop Trace on 1st Access on PAB Trigger 1
The final Breakpoint Unit trigger will then be one of the following:
Table 11-5. Breakpoint Unit Trigger — 16-bit Counter Available for Triggering
Breakpoint Unit Trigger
Case 1
(First Part of Breakpoint trigger) OR Enabled DEBUGEV OR Enabled Limiting
Case 2
(First Part of Breakpoint trigger) => (Enabled DEBUGEV OR Enabled Limiting)
This is true except for the cases where the Breakpoint Unit is used to generate both the Start and Stop
triggers.
11.4.4.2.2 Breakpoint Unit Actions
Once a valid Breakpoint Unit Trigger has occurred, one of the following actions can be performed. Other
actions can be found in further sections which use “Breakpoint Unit Trigger” as a triggering condition.
Table 11-6. Possible Breakpoint Unit Actions
Trigger for Action
Case BK1
Breakpoint Unit Trigger
Action Performed
Enter Debug state
Case BK2
Generate Breakpoint Unit Interrupt Request
Case BK3
Start Trace Buffer Capture
Case BK4
Halt Trace Buffer Capture
Case BK5
Signal Watchpoint
The “Signal Watchpoint” action listed above refers to simply toggling the event terminal, one of the
terminals available as an output of the Enhanced OnCE module.
11-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
11.4.4.2.3 Combining the Breakpoint Unit with the Step Counter
The breakpoint unit can work in conjunction with the 24-bit step counter so that the action is taken a
specified number of clock cycles after the breakpoint condition is detected. This configuration is illustrated
in Figure 11-10.
XAB1
[0:PAB]
[0:PAB] CDBR/CDBW
Read/Write
Fetch
Breakpoint 1
32-bit
Mask
Breakpoint 2
DEBUGEV
instruction
Overflow or
saturation
Combining
Logic
16-Bit Counter
Breakpoint Unit
Trigger
24-Bit
Step Counter
Select
Action
Select
Action
Step Counter Halt Halt Trace
Interrupt
Core
Buffer
Start
Trace
Buffer
Figure 11-10. Triggering the Step Counter with the Breakpoint Unit
11.4.4.2.4 Breakpoint Unit — Step Counter Actions
Table 11-7 lists the possible actions when using the Step Counter, where the Breakpoint Unit can use any
of the configurations in Table 11-4 and Table 11-5.
Table 11-7. Breakpoint Unit — Step Counter Operation
Start Step Counter
Case BKSC1
Breakpoint Unit Trigger
Trigger for Step Counter Action
Step Counter reaches zero
Action Performed
Enter Debug state
Case BKSC2
Generate Step Counter
Interrupt Request
Case BKSC3
Start Trace Buffer Capture when
Breakpoint Unit Trigger arrives.
Halt Trace Buffer Capture when
Step Counter reaches zero.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-19
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.4.4.3 Capture Counter
The Breakpoint Unit can also be configured as a Capture Counter to measure the number of clocks
executed between two different points. The Capture Counter can be configured as 16-bits or 40-bits.
11.4.4.3.1 16-Bit Capture Counter (Non-Cascaded)
In this case, the 16-bit breakpoint counter is configured to count clocks between two different points and is
no longer available for generating breakpoint triggers. The Non-Cascaded configuration (Figure 11-11)
uses the 16-bit counter providing count values up to 216.
Start
Stop
Clocks
Clocks w/o Wait States
MUX
Instructions Executed
16-Bit Breakpoint
Counter
Select
Action
Status
Bits
Counter
Interrupt
Halt
Core
Figure 11-11. Capture Counter — 16-bit Configuration (Non-Cascaded)
The Capture Counter is configured by the user to count any of the three inputs to the MUX above:
•
Clocks executed
•
Clocks executed without Wait States
•
Instructions executed
The counter measures any of these three values between two different points — the counter start trigger
and the counter stop trigger. The triggers supported are shown in Table 11-8.
Table 11-8. Starting and Stopping the Capture Counter — Non-Cascaded
Counter Start
Trigger
11-20
Counter Stop
Trigger
Case CCT1
PAB Trigger 1
PAB Trigger 2
Case CCT2
PAB Trigger 2
PAB Trigger 1
Case CCT3
Breakpoint Unit Trigger
Enter Debug state
Case CCT4
Exit Reset or Debug state
Breakpoint Unit Trigger
Case CCT5
Execute DEBUGEV
Breakpoint Unit Trigger
Case CCT6
Limit occurs
Breakpoint Unit Trigger
Case CCT7
Execute DEBUGEV
or Limit occurs
Breakpoint Unit Trigger
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
Cases CCT1 and CCT2 directly use the first and second triggers of the Breakpoint Unit as Start and Stop
triggers. The remaining cases use the Breakpoint Unit to generate either the Start trigger (case CCT3) or
the Stop trigger (remaining cases). These remaining cases use any of the triggers supported in Table 11-9:
Table 11-9. First Part of Breakpoint Unit Trigger— 16-bit Counter in Capture Mode
First Breakpoint
Trigger
Op
Second Breakpoint
Trigger
Comments
Single Triggers
PAB-1 — F/R/W/A
(none)
1st F/R/W/A on PAB Bus, Trigger 1
XAB1 — R/W/A
(none)
1st R/W/A on XAB1 Bus, Trigger 1
ORed Triggers
PAB-1 — Fetch
OR
PAB-2 — Fetch
1st Fetch on PAB Trig1 or 1st Fetch PAB Trig2
PAB-1 — Access
OR
PAB-2 — Fetch
1st Access on PAB Trig1 or 1st Fetch PAB Trig2
PAB-2 — Fetch
OR
XAB1 — Access
1st F on PAB Trig2 or 1st Access XAB1 Trig1
CDB — Data Value
1st R/W/A on XAB1 Trig1 and CDB Data Val Trig2
ANDed Triggers
XAB1 — R/W/A
AND
Sequenced Triggers
PAB-1 — Fetch
==>
PAB-2 — Fetch
1st F on PAB Trig1 followed by 1st F PAB Trig2
PAB-1 — Access
==>
PAB-2 — Fetch
1st A on PAB Trig1 followed by 1st F PAB Trig2
PAB-2 — Fetch
==>
PAB-1 — Access
1st F on PAB Trig2 followed by 1st A PAB Trig1
XAB1 — R/W/A
==>
PAB-2 — Fetch
1st R/W/A XAB1 Trig1 followed 1st F PAB Trig2
PAB-2 — Fetch
==>
XAB1 — R/W/A
1st F PAB Trig2 followed 1st XAB1 R/W/A Trig2
Generation of Two Triggers — Start and Stop
PAB-1 — Fetch
=> Start Capture Ctr
—
PAB-2 — Fetch
=> Stop Capture Ctr
Start Capture on 1st Fetch on PAB Trig 1 and Stop
Capture on 1st Fetch on PAB Trig2
PAB-1 — Access
=> Start Capture Ctr
—
PAB-2 — Fetch
=> Stop Capture Ctr
Start Capture on 1st Access on PAB Trig 1 and
Stop Capture on 1st Fetch on PAB Trig2
PAB-2 — Fetch
=> Start Capture Ctr
—
PAB-1 — Access
=> Stop Capture Ctr
Start Capture on 1st Fetch on PAB Trig 2 and Stop
Capture on 1st Access on PAB Trig1
Note that the triggers above do not support the ability of triggering on the Nth occurrence because the
16-bit counter is now dedicated to counting operations, and no longer available for breakpoint triggering.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-21
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
The final Breakpoint Unit trigger will then be one of the following:
Table 11-10. Breakpoint Unit Trigger — for 16-bit Capture Counter
Breakpoint Unit Trigger
Case 1
(First Part of Breakpoint trigger) OR Enabled DEBUGEV OR Enabled Limiting
Case 2
(First Part of Breakpoint trigger) => (Enabled DEBUGEV OR Enabled Limiting)
NOTE:
The equation in Table 11-10 above can be used except for the cases
entitled “Generation of Two Triggers — Start and Stop” where the
Breakpoint Unit is used to generate both the Start and Stop triggers.
11.4.4.3.2 Actions for 16-Bit Capture Counter (Non-Cascaded)
Table 11-11 shows the actions which can be performed when the Capture Counter expires. Note the
unusual triggering which can be performed to check that the counter expires before a Stop trigger arrives.
Similarly, the reverse triggering is also supported - trigger only if the Stop trigger arrives before the
counter expires.
Table 11-11. Possible Capture Counter Actions — Non-Cascaded
Trigger for Action
Case CC1
Capture Counter reaches zero before
Counter Stop Trigger occurs
Case CC2
Action Performed
Enter Debug state
Generate EOnCE Interrupt Request
— OR —
Case CC3
Case CC4
Case CC5
Set Capture Counter status bits — CS, CZ
Capture Counter reaches zero
(for cases where no Stop Trigger is configured)
Counter Stop Trigger occurs before
Capture Counter reaches zero
Signal Watchpoint
Enter Debug state
Case CC6
Generate EOnCE Interrupt Request
Case CC7
Signal Watchpoint
11-22
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
11.4.4.3.3 Using the Capture Counter with the Step Counter
The Capture Counter can also work in conjunction with the 24-bit step counter so that the action is taken a
specified number of clock cycles after a Capture Counter trigger is generated. This configuration is
illustrated in Figure 11-12.
Start
Stop
Capture Counter
Trigger
Clocks
Clocks w/o Wait States
Instructions Executed
MUX
24-Bit
Step Counter
16-Bit Breakpoint
Counter
Select
Action
Start
Trace
Buffer
Step Counter Halt Halt Trace
Interrupt
Core
Buffer
Figure 11-12. Triggering the Step Counter with the Capture Counter
This configuration also uses the Start-Stop triggers listed in Table 11-8, where the Breakpoint Unit can use
any of the configurations in Table 11-9 and Table 11-10.
11.4.4.3.4 16-bit Capture Counter — Step Counter Actions
Table 11-12 shows the actions which can be performed in this configuration. Note the unusual triggering
which can be performed to check that the Capture Counter expires before a Stop trigger arrives. Similarly,
the reverse triggering is also supported - trigger only if the Stop trigger arrives before the Capture Counter
expires.
Table 11-12. Possible Capture Counter Actions — Non-Cascaded
Trigger for Action
Case CCSC1
Action Performed
Case CCSC2
Capture Counter reaches zero before
Counter Stop Trigger occurs
=> Step Counter reaches zero
Case CCSC3
— OR —
Start Trace Buffer Capture when
Capture Counter reaches zero.
Capture Counter reaches zero
=> Step Counter reaches zero
(for cases where no Stop Trigger is configured)
Halt Trace Buffer Capture when
Step Counter reaches zero.
Case CCSC4
Case CCSC5
Case CCSC6
Counter Stop Trigger occurs before
Capture Counter reaches zero
=> Step Counter reaches zero
Enter Debug state
Generate Step Counter Interrupt Request
Enter Debug state
Generate Step Counter Interrupt Request
Start Trace Buffer Capture when
Capture Counter reaches zero.
Halt Trace Buffer Capture when
Step Counter reaches zero.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-23
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.4.4.3.5 40-Bit Capture Counter (Cascaded)
If additional counter bits are needed, the Capture Counter can also be cascaded with the 24-bit step counter
to provide 40-bit counting operations. This configuration is illustrated in Figure 11-12.
Start
Stop
Clocks
Clocks w/o Wait States
Instructions Executed
MUX
16-Bit Breakpoint
Counter
24-Bit Breakpoint
Counter
Select
Action
Status
Bits
Counter
Interrupt
Halt
Core
Figure 11-13. Capture Counter — 40-bit Configuration (Cascaded)
This configuration also uses the Start-Stop triggers listed in Table 11-8, where the Breakpoint Unit can use
any of the configurations in Table 11-9 and Table 11-10.
11.4.4.3.6 Actions for 40-Bit Capture Counter (Cascaded)
The actions supported by this configuration are the same as those listed in Table 11-11.
11.4.4.4 Programmable Trace Buffer
The Trace Buffer is used to the change-of-flows selected by the user. Separate control bits are available for
the following five cases, allowing any combination of these to be selected by the user:
•
Interrupts—captures the address of the interrupt vector and target address of RTI and FRTID
•
Subroutines—captures target address of JSR and BSR instructions
•
Change-of-Flow Not Taken —captures target address of Bcc, Jcc, BRSET, BRCLR instructions
•
Change-of-Flow Case 0 —captures target address of Jcc or forward branches of Bcc, BRSET,
BRCLR instructions
•
Change-of-Flow Case 1 —captures the target address of backward branches of Bcc, BRSET,
BRCLR instructions
11-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Enhanced OnCE and the Processing States
PAB
Conditional Branches and Jumps
Interrupts, Subroutines
Address
Selection
Start Capture
8-Location
Trace buffer
Trace Buffer Full
Stop Capture
Select
Action
Status
Bits
Halt Trace Buffer Halt
Capture Interrupt
Core
Figure 11-14. Programmable Trace Buffer
Several different options are available for starting and/or stopping Trace Buffer capture (Table 11-13). In
addition, Trace Buffer capture can also be programmed to stop once it has filled (Table 11-14). The
Breakpoint Unit Trigger can use any of the configurations in Table 11-9 and Table 11-10.
Table 11-13. Starting and Stopping Trace Buffer Capture
Start
Trigger
Stop
Trigger
Case 1
PAB Trigger 1
PAB Trigger 2
Case 2
Breakpoint Unit Trigger
—
Case 3
Exit Debug state
Breakpoint Unit Trigger
The Trace Buffer can be programmed to perform any of the actions listed in Table 11-14 when the Trace
Buffer is full:
Table 11-14. Possible Actions on Trace Buffer Full
Action Performed
Case TBF1
No Action Performed — Trace Buffer continues to capture
new addresses, overwriting the old addresses as needed.
Case TBF2
Buffer Capture Halted — TBH is asserted
Case TBF3
Buffer Capture Halted — Enter Debug state
Case TBF4
Buffer Capture Halted — Generate Trace Buffer Interrupt Request
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-25
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
The Trace Buffer can also be configured to Start and Stop capture as shown in Section 11.4.4.2.4,
“Breakpoint Unit — Step Counter Actions,” on page 11-19 and Section 11.4.4.3.3, “Using the Capture
Counter with the Step Counter,” on page 11-23.
11.4.5 Example Breakpoint Scenarios
The following are examples of the variety of conditions that can trigger a breakpoint or step counter action.
•
Fetch, read, write, or read or write of specific program address
Example: PAB == $000080.
•
Read, write, or read or write of specific data address
Example: XAB1 == $0C0000.
•
The nth occurrence of an instruction
Example: 500 occurrences of PAB == $008794.
•
Either of two instructions
Example: PAB == $3792 || PAB == $7E45
•
A sequence of two instructions
Example: PAB == $3792 → PAB == $7E45
•
The nth occurrence of an instruction followed by another instruction
Example: 1037 occurrences of PAB == $394 → PAB == 7E45
•
Write a specific value to a data address
Example: XAB1 == $00FFE7 && CDBW == $AAAA
•
Read value from data address
Example: XAB1 == $00FFE7 && CDBR == $5555
•
Read a data value other than the one specified from a particular data address
Example: XAB1 == $00FFE7 && CDBR != $AAAA
•
Read or write a particular set of bits from/to a data address
Example: XAB1 == $00FFE7 && CDBW[2:0] == 011b
•
Either of two program addresses or a DEBUGEV instruction followed by n instructions
Example: PAB == $3792 || PAB == $7E45 || DEBUGEV → 4000 instructions
•
A sequence of two program addresses followed by a DEBUGEV instruction followed by n
instructions
Example: PAB == $3792 → PAB == $7E45 → DEBUGEV → 4000 instructions
•
The nth occurrence of an instruction followed by another instruction followed by an overflow
condition followed by m instructions
Example: 900 occurrences of PAB == $3792 → PAB == $7E45 → OV → 9 instructions
•
A particular bit pattern not occurring at a specific data address followed by n instructions
Example: XAB1 == $00FFE7 && CDB[14:12] != 011b → 20,000 instructions
•
The nth occurrence of the above condition followed by m instructions
Example: 400 occurrences of (XAB1 == $00FFE7 && CDB[14:12] != 011b) → 350 instructions
11-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
JTAG Port
11.5 JTAG Port
The DSP56800E core Joint Test Action Group (JTAG) test access port (TAP) provides the interface for the
Enhanced OnCE module to the DSC JTAG pins. This TAP controller is designed to be incorporated into a
chip multi–JTAG TAP Linking Module (JTAG TLM) system. The JTAG TLM is a dedicated,
user-accessible, test access port (TAP) system that is compatible with the IEEE Standard 1149.1a-1993,
IEEE Standard Test Access Port and Boundary-Scan Architecture. Problems associated with testing
high-density circuit boards have led to the development of this standard under the sponsorship of the Test
Technology Committee of IEEE and the JTAG. If the core TAP is not incorporated into a JTAG TLM
system it will not be compliant with the IEEE 1149.1a-1993 standard, but the TAP will still serve as an
interface to the core Enhanced OnCE module. Specific details on the implementation of the JTAG port for
a given DSP56800E–based device are provided in the user’s manual for that device.
11.5.1 JTAG Capabilities
The DSP56800E JTAG port has the following capabilities:
•
Provides queried identification information for the DSP56800E core (manufacturer, technology
process, part, and version numbers)
•
Provides a means of accessing the Enhanced OnCE module controller and circuits to control a target
system
•
Provides a means of entering the debug mode of operation
•
Bypasses the TAP through a single-bit register in the Shift-DR-Scan path
The following sections provide an overview of the port’s architecture and commands.
11.5.2 JTAG Port Architecture
The JTAG port consists of the following components:
•
Serial communication interface
•
Command decoder and interpreter
•
DSP56800E identification register
The serial interface provides the communication link between the core and the host development or debug
system. All JTAG data is sent over this interface. Enhanced OnCE commands and data from the host
system can also sent over this interface if accessed via JTAG. It is implemented as a serial interface to
occupy as few external pins on the device as possible. For a full description of the interface signals, consult
the user’s manual for the specific device.
Commands sent to the JTAG module are decoded and processed by the command decoder. Commands for
the JTAG port are completely independent from the DSP56800E instruction set, and they are executed in
parallel by the JTAG logic.
The JTAG module contains the DSP56800E identification register, which provides a unique ID for each
revision of the DSP56800E core. This register enables a development system to determine the
manufacturer, process technology, part, and revision numbers of the DSP56800E core via the JTAG port.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-27
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.5.2.1 JTAG Terminal Description
As described in the IEEE 1149.1a-1993 specification, a JTAG TAP requires a minimum of 4 pins to
support TDI, TDO, TCK, and TMS signals. TDI and TDO are the serial input and output, respectively.
TCK is the serial clock input and TMS is an input used to selectively step through the JTAG state machine.
A fifth pin TRST is an optional asynchronous reset pin for the chip JTAG TLM system (refer to the
particular chip users manual to see if this pin is available).
These pins for the core JTAG port are CORE_TDI, CORE_TDO, TCK, TMS. The core pin functions are
described in Table 11-15. The core JTAG TAP also uses the TLM_RESET_B pin to provide an
asynchronous reset of the core JTAG port from the chip JTAG TLM. If TRST is present on a chip the core
TLM_RESET_B pin will always be asserted whenever TRST is asserted.
Table 11-15. JTAG Pin Descriptions
Pin Name
Pin Description
CORE_TDI
Test Data Input—This input pin to the core provides a serial input data
stream to the core TAP and the EOnCE module. It is sampled on the rising
edge of TCK.
CORE_TDO
Test Data Output—This output pin provides a serial output data stream
from the core TAP and the EOnCE module. It is driven in the Shift-IR and
Shift-DR controller states of the core TAP state machine.
TCK
Test Clock Input—This input pin provides the clock to synchronize the test
logic and shift serial data to and from the core EOnCE/JTAG port. When
accessing the EOnCE module through the JTAG TAP, the maximum
frequency for TCK is 1/4 the maximum frequency specified for the Hawk
Version 2 core.
TMS
Test Mode Select Input—This input pin is used to sequence the core JTAG
TAP controller’s state machine. It is sampled on the rising edge of TCK.
TLM_RESET_B
Test Reset—This input pin, comes from the chip TLM and provides an
asynchronous reset signal to the JTAG TAP controller,.
CORE_TAP_EN
Core TAP Enable—This input, comes from the chip TLM module and gates
the input TMS signal to force the TAP controller to the Run-Test/Idle state
when the enable signal is deasserted (logic 0). When the enable signal is
asserted, the TAP controller will follow the transitions and state of the input
pin TMS signal.
CORE_TLM_SEL
Core TLM Selects—This output from the core JTAG TAP selects the chip
TLM register for the data register to be scanned.
The core JTAG TAP must be enabled (CORE_TAP_EN asserted) before the core JTAG state machine will
follow the transitions and state of the TMS pin. The core TAP will only leave the Run-Test/Idle state to
enter the DR or IR states while the CORE_TAP_EN pin is asserted, and will return to Run-Test/Idle when
the pin is deasserted in the Update-DR state.
11-28
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
JTAG Port
11.5.2.2 Core JTAG Programming Model
Figure 11-15 shows the programming models for the core JTAG registers. There are 2 read/write registers
in the JTAG port: the IR, and the core Bypass Register. A third register, the Core Identification Register, is
read only.
3
2
1
0
INSTRUCTION
Core JTAG B3 B2 B1 B0
Instruction
Register
Reset = $2
ID—(IR = $2)
Core Identification
Register
Reset = Core ID
Read
BYPASS—(IR = $F)
Core JTAG Bypass
Register
Reset = $0
Read/Write
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
0
Figure 11-15.
JTAG Port Programming Model
11.5.2.3 Core JTAG Port Block Diagram
A block diagram of the JTAG port is shown in Figure 11-16.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-29
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
To EOnCE Port
Instruction Register
CORE_TDI
Decode
CORE_TDO
Core ID Register
Core Bypass Register
TMS
TCK
TLM_RESET_B
CORE_TAP_EN
CORE_TLM_SEL
From EOnCE Port
TAP
Controller
Figure 11-16. Core JTAG Block Diagram
The TAP controller provides access to the IR through the core JTAG port. The other core JTAG registers
must be individually selected by the IR.
11.5.2.4 Core TAP Controller
The TAP controller is a sixteen state synchronous finite state machine, used to sequence the core JTAG
port through its valid operations:
•
Serially shift in or out a core JTAG instruction
•
Update (and decode) the core JTAG Instruction Register
•
Serially output the core ID code
•
Serially shift in or out and update the EOnCE registers.
NOTE:
The core JTAG port oversees the shifting of data into and out of the
EOnCE port through the CORE_TDI and CORE_TDO pins, respectively.
The shifting, in this case, is guided by the same tap controller used when
shifting core JTAG Instruction Register (IR) information.
The TAP controller is shown in Figure 11-17. The TAP controller will asynchronously be reset to the
Test-Logic-Reset state upon assertion low of tlm_res_b pin. When the tlm_res_b signal is deasserted and
the core_tap_en pin is asserted, the TAP controller responds to changes of the TMS and TCK signals.
Transitions from one state to another occur on the rising edge of TCK. The value shown adjacent to each
state transition in this figure represents the signal present at TMS at the time of a rising edge of TCK.
11-30
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
JTAG Port
When the core_tap_en pin is deasserted the TAP controller returns to the Run-Test/Idle state at the next
rising edge of TCK and remains there until the TAP is re-enabled to follow the transitions and state of the
TMS signal, by core_tap_en pin assertion.
Test-Logic-Reset
1
0
Run-Test/Idle
1
Select-DR-Scan
0
1
Select-IR-Scan
0
0
1
1
Capture-DR
Capture-IR
0
0
Shift-DR
Shift-IR
0
1
0
1
1
1
Exit1-DR
Exit1-IR
0
0
Pause-IR
Pause-DR
0
1
0
Exit2-DR
0
1
0
Exit2-IR
1
1
Update-DR
1
1
0
Update-IR
1
0
Figure 11-17. TAP Controller State Diagram
There are two paths through the 16-state machine. The Shift-IR_Scan path is used to capture and load core
JTAG instructions into the core JTAG IR. The Shift-DR_Scan path captures and loads data into the other
core JTAG registers. The core TAP controller executes the last instruction decoded until a new instruction
is entered at the Update-IR state or until the Test-Logic-Reset state is entered. When using the core JTAG
port to access EOnCE module registers, accesses are first enabled by shifting the ENABLE_EOnCE
instruction into the core JTAG IR. After this is selected, the EOnCE module registers and commands are
read and written through the core JTAG pins using the Shift-DR_Scan path. Asserting the tlm_reset_b pin
low asynchronously forces the core JTAG state machine into the Test-Logic-Reset state.
Freescale SemiconductorJTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11-31
JTAG and Enhanced On-Chip Emulation (Enhanced OnCE)
11.5.3 JTAG Port Restriction — STOP Processing State
The core features a low-power stop mode, that is invoked by the Hawk V2 core executing a STOP
instruction. Since all Hawk V2 core clocks are disabled during Stop mode, the JTAG interface provides the
means of polling the device status (sampled in the capture-IR state). The core JTAG TAP will bring the
core out of Stop or Wait modes when DEBUG_REQUEST is decoded in the TAP IR. A small amount of
additional power above the minimum possible will be expended by the core TAP logic if the core TAP is
utilized during Stop mode.
11-32
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Appendix A
Instruction Set Details
This appendix contains detailed information about each instruction of the DSC core instruction set.
Section A.1, “Notation,” explains most of the notation that is used in Section A.2, “Instruction
Descriptions,” which shows the syntax of all allowed instructions and summarizes addressing modes,
condition codes, and instruction timing. Section A.5, “Instruction Opcode Encoding,” provides additional
details about the notation for opcode encoding.
For more detailed information on condition codes, see Appendix B, “Condition Code Calculation.”
A.1 Notation
Each instruction description abbreviates operands using the notation that is contained in the following
tables. Table A-1 on page A-2 defines the register notation that is used in general read and write
operations.
Freescale Semiconductor
Instruction Set Details
A-1
Table A-1. Register Fields for General-Purpose Writes and Reads
Register Field
Registers in this
Field
HHH
(source)
A1, B1, C1, D1
X0, Y0, Y1
Comments
Seven data ALU registers—four 16-bit MSP portions of the accumulators and three 16-bit data registers that are used as source registers.
Note the usage of A1, B1, C1, and D1.
This field is identical to the FFF1 field.
HHH
(destination)
A, B, C, D
Y
X0, Y0, Y1
Seven data ALU registers—four 16-bit MSP portions of the accumulators and three 16-bit data registers that are used as destination registers. Note the usage of A, B, C, and D. Writing word data to the 32-bit Y
register clears the Y0 portion.
HHH.L
(source)
A10, B10, C10, D10
Y
Five data ALU registers—four 32-bit MSP:LSP portions of the accumulators and one 32-bit Y data register (Y1:Y0) that is used as a source
register.
Used for long memory accesses.
HHH.L
(destination)
A, B, C, D
Y
Five data ALU registers—four 32-bit MSP:LSP portions of the accumulators and one 32-bit Y data register (Y1:Y0) that is used as a destination register.
Used for long memory accesses.
HHHH
(source)
A1, B1, C1, D1
X0, Y0, Y1
R0–R5, N
Seven data ALU and seven AGU registers that are used as source registers. Note the usage of A1, B1, C1, and D1.
HHHH
(destination)
A, B, C, D
Y
X0, Y0, Y1
R0–R5, N
Seven data ALU and seven AGU registers that are used as destination
registers. Note the usage of A, B, C, and D. Writing word data to the
32-bit Y register clears the Y0 portion.
HHHH.L
(source)
A10, B10, C10, D10
Y
R0–R5, N
Five data ALU and seven AGU registers that are used as source registers.
Used for long memory accesses. Also see dddd.L.
HHHH.L
(destination)
A, B, C, D
Y
R0–R5, N
Five data ALU and seven AGU registers that are used as destination
registers.
Used for long memory accesses. Also see dddd.L.
Table A-2 on page A-3 shows the registers that are available for use as pointers in address-register-indirect
addressing modes. The most common fields that are used in this table are Rn and RRR. This table also
shows the notation that is used for AGU registers in AGU arithmetic operations.
A-2
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Table A-2. Address Generation Unit (AGU) Registers
Register Field
Registers in this
Field
Comments
Rn
R0–R5
N
SP
Eight AGU registers that are available as pointers for addressing and
address calculations
RRR
(or SSS)
R0–R5
N
Seven AGU registers that are available as sources and destinations
for move instructions
Rj
R0, R1, R2, R3
N3
N3
M01
M01
Address modifier register
FIRA
FIRA
Fast interrupt return register
Four pointer registers that are available as pointers for addressing
One index register that is available only for the second access in dual
parallel read instructions
Table A-3 shows the register set that is available for use in data ALU arithmetic operations. The most
common field that is used in this table is FFF.
Table A-3. Data ALU Registers
Register
Field
Registers in this
Field
Comments
FFF
A, B, C, D
Y
X0, Y0, Y1
Eight data ALU registers—four 36-bit accumulators, one 32-bit long register Y, and three 16-bit data registers that are accessible during data ALU
operations.
FFF1
A1, B1, C1, D1
X0, Y0, Y1
Seven data ALU registers—four 16-bit MSP portions of the accumulators
and three 16-bit data registers that are accessible during data ALU operations.
This field is identical to the HHH (source) field. It is very similar to FFF, but
it indicates that the MSP portion of the accumulator is in use. Note the
usage of A1, B1, C1, and D1.
EEE
A, B, C, D
X0, Y0, Y1
Seven data ALU registers—four accumulators and three 16-bit data registers that are accessible during data ALU operations.
This field is similar to FFF but is missing the 32-bit Y register. Used for
instructions where Y is not a useful operand (use Y1 instead).
fff
A, B, C, D
Y
Four 36-bit accumulators and one 32-bit long register that are accessible
during data ALU operations.
FF
A, B, C, D
Four 36-bit accumulators that are accessible during data ALU operations.
DD
X0, Y0, Y1
Three 16-bit data registers.
F
A, B
Two 36-bit accumulators that are accessible during parallel move instructions and some data ALU operations.
F1
A1, B1
The 16-bit MSP portions of two accumulators that are accessible as source
operands in parallel move instructions.
Freescale Semiconductor
Instruction Set Details
A-3
Table A-4 shows additional register fields that are available for move instructions.
Table A-4. Additional Register Fields for Move Instructions
Register Field
DDDDD
Registers in this
Field
A, A2, A1, A0
B, B2, B1, B0
C, C1
D, D1
Y
Y1, Y0, X0
R0, R1, R2, R3
R4, R5, N, SP
M01, N3
Comments
This field lists the CPU registers. It contains the contents of the HHHHH
and SSSS register fields.
Y is permitted only as a destination, not as a source.
Writing word data to the 32-bit Y register clears the Y0 portion.
Note that the C2, C0, D2, and D0 registers are not available within this
field. See the dd register field in this table for these registers
OMR, SR
LA, LC
HWS
dd
C2, D2, C0, D0
Extension and LS portion of the C and D accumulators.
This register set supplements the DDDDD field.
HHHHH
SSSS
A, A2, A1, A0
B, B2, B1, B0
C, C1
D, D1
Y
Y1, Y0, X0
R0, R1, R2, R3
R4, R5, N, SP
M01, N3
LA, LC, HWS
OMR, SR
dddd.L
A-4
A2, B2, C2, D2
Y0, Y1, X0
SP, M01, N3,
LA, LA2, LC, LC2,
HWS, OMR, SR
This set designates registers that are written with signed values when
they are written with word values.
Y is permitted only as a destination, not as a source.
The registers in this field and SSSS combine to make the DDDDD register field.
This set designates registers that are written with unsigned values when
they are written with word values.
The registers in this field and in HHHHH combine to make the DDDDD
register field.
Miscellaneous set of registers that can be placed onto or removed from
the stack 32 bits at a time.
This list supplements the registers in the HHHH.L field, which also can
access the stack via the MOVE.L instruction.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Table A-5 provides an alphabetical overview of the fields and refers to the additional section and tables
that contain the precise encoding values.
Table A-5. Opcode Encoding Fields
Encoding Field
Description
Location
AAA
Top 3 address bits for branch
Section A.5.7
AA
Top 2 address bits for branch
Section A.5.7
6-bit positive offset for X:(R2+xx) addressing mode
Section A.5.7
AAAAAA
aaa
aaaaaa
Aaaaaaa
Data ALU register (excluding Y)
Table A-7
6-bit negative offset for X:(SP–xx) addressing mode
Section A.5.7
7-bit signed offset for branch instructions
Section A.5.7
bb
Accumulator
Table A-7
bbb
Data ALU register
Table A-7
BBBBB
5-bit signed integer immediate
Section A.5.7
BBBBBB
6-bit signed integer immediate
Section A.5.7
BBBBBBB
7-bit signed integer immediate
Section A.5.7
ccc
16-bit data ALU register or accumulator portion
Table A-7
CCC
Condition code specifier
Table A-17
CCCC
Condition code specifier
Table A-18
DD
16-bit data ALU register
Table A-7
dddd
Special 32-bit stack push/pop register
Table A-13
ddddd
Full set of DSP56800E registers
Table A-12
DDDDD
Full set of DSP56800E registers
Table A-11
hhhhh
DALU set registers
Table A-11
SSSS
Non-DALU set registers
Table A-11
Data ALU register (excluding Y)
Table A-7
A or B accumulator
Table A-7
FF
Accumulator
Table A-7
fff
Accumulator or Y
Table A-7
FFF
Data ALU register
Table A-7
GGG
Data ALU register
Table A-10
GGG
Parallel move destination register
Table A-14
24-bit AGU pointer register or 16-bit data ALU register
Table A-10
EEE
F
GGGG
Freescale Semiconductor
Instruction Set Details
A-5
Table A-5. Opcode Encoding Fields (Continued)
Encoding Field
Description
Location
hhh
Data ALU register
Table A-13
hhhh
Full set of DSP56800E registers
Table A-13
iii
3-bit offset for X:(Rn+x) and X:(SP–x) addressing modes
Table A-19
iiii
4-bit unsigned integer immediate
JJ
16-bit data ALU register
Table A-9
JJJ
Accumulator or 16-bit data ALU register
Table A-9
Two input registers for three-operand instructions
Table A-8
m
Addressing mode specifier
Table A-16
MM
Addressing mode specifier
Table A-16
nnn
24-bit AGU pointer register or 16-bit data ALU register
Table A-10
NNN
24-bit AGU pointer register
Table A-10
JJJJJ
Ppppppp
Section A.5.7
7-bit absolute address for X:<<pp addressing mode
Section A.5.7
QQ
16-bit data ALU register
Table A-9
qqq
Two input registers for three-operand instructions
Table A-8
QQQ
Two input registers for three-operand instructions
Table A-8
R0–R3 pointer registers
Table A-10
RRR
24-bit AGU pointer register
Table A-10
SSS
24-bit AGU pointer register
Table A-10
RR
U
vvvv
Single bit to indicate lower or upper byte in BRSET and
BRCLR
Dual parallel read destination registers
Section A.5.7
Table A-15
Certain core instructions use symbols in the instruction field to represent operands or addressing modes in
the opcodes. These symbols are listed in Table A-6.
Table A-6. Instruction Field Symbols
Symbol
Meaning
Reference
Q1
Q2
First source register in the QQQ field
Second source register in the QQQ field
Table A-8 on page A-332
Q3
Q4
First source register in the QQ field
Second source register in the QQ field
Table A-8 on page A-332
X:<ea_m>
Addressing mode of ‘m’ field in single parallel move
or the first operand in a dual parallel read
Table A-16 on page A-345
A-6
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Table A-6. Instruction Field Symbols
Symbol
Meaning
Reference
X:<ea_v>
Addressing mode of ‘vvvv’ field in the second operand of a dual parallel read
Table A-15 on page A-344
X:<ea_MM>
Addressing mode of ‘MM’ field for memory access
Table A-16 on page A-345
A.2 Instruction Descriptions
The following section describes each instruction in the instruction set in complete detail. Aspects of each
instruction description are explained in Section A.1, “Notation.”
The “Operation” and “Assembler Syntax” fields appear at the beginning of each description. For
instructions that allow parallel moves, these fields include the parenthetical comment “(parallel move).”
Every description also includes an example. The example discusses the contents of all the registers and
memory locations that are referenced by the opcode and operand portion of the instruction, although it
does not discuss those that are referenced by the parallel move portion of the instruction.
Whenever an instruction uses an accumulator as both a destination operand for a data ALU operation and
as a source for a parallel move operation, the parallel move operation uses the value in the accumulator
prior to the execution of any data ALU operation.
A brief overview of the condition codes that are affected by each instruction is presented in each
instruction’s “Condition Codes Affected” section. For a more thorough discussion of condition code
calculation, refer to Appendix B, “Condition Code Calculation.”
For more information about the notation that is used in the “Instruction Opcode” sections of the instruction
descriptions, see Section A.5, “Instruction Opcode Encoding.”
Freescale Semiconductor
Instruction Set Details
A-7
ABS
Operation:
|D| → D
|D| → D
ABS
Absolute Value
Assembler Syntax:
(one parallel move)
(no parallel move)
ABS
ABS
D
D
(one parallel move)
(no parallel move)
Description: Take the absolute value of the destination operand (D) and store the result in the destination accumulator or 16-bit register. Duplicate destination is not allowed when this instruction is used in conjunction
with a parallel read.
Example:
ABS
A
X:(R0)+,Y0
; take ABS value, move data into Y0,
; update R0
After Execution
Before Execution
F
FFFF
FFF2
0
0000
000E
A2
A1
A0
A2
A1
A0
SR
SR
0301
0311
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $F:FFFF:FFF2. The execution of the
ABS instruction takes the two’s-complement of that value and returns $0:0000:000E.
Note:
When the D operand equals $8:0000:0000 (–16.0 when interpreted as a decimal fraction), the ABS instruction causes an overflow to occur since the result cannot be correctly expressed using the standard
36-bit, fixed-point, two’s-complement data representation. When saturation is enabled (SA = 1 in the
OMR register), data limiting will occur to value $F:8000:000. If saturation is not enabled, the value
will remain unchanged.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
A-8
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extended portion of accumulator result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Set if overflow has occurred in result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ABS
ABS
Absolute Value
Instruction Fields:
Operation
Operands
C
W
ABS
FFF
1
1
Comments
Absolute value.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
ABS2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
ABS
F GGG,X:<ea_m>
ABS
F X:<ea_m>,GGG
0
0
0
0
1
15
0
15
ABS
FFF
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
1
1
12
11
0
1
12
11
0
1
12
11
1
0
8
7
G
G
G
F
8
7
G
G
G
F
8
7
F
F
1
F
Instruction Set Details
4
3
0
1
0
0
4
3
0
1
0
0
4
3
0
0
0
1
0
m
R
R
m
R
R
0
0
1
1
1
A-9
ADC
ADC
Add Long with Carry
Operation:
Assembler Syntax:
S + C + D →D
(no parallel move)
ADC
S,D
(no parallel move)
Description: Add the source operand (S) and the carry bit (C) to the second operand, and store the result in the destination (D). The source operand (register Y) is first sign extended internally to form a 36-bit value
before being added to the destination accumulator. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is typically used in multi-precision addition operations (see Section 5.5.1, “Extended-Precision Addition and Subtraction,” on page 5-29) when it is necessary to add together two numbers that are larger than 32 bits (as in 64-bit or 96-bit addition).
Example:
ADC
Y,A
; add Y and carry to A
Before Execution
After Execution
0
2000
8000
0
4001
0001
A2
A1
A0
A2
A1
A0
2000
8000
2000
8000
Y1
Y0
Y1
Y0
SR
SR
0301
0300
Explanation of Example:
Prior to execution, the 32-bit Y register—which is composed of the Y1 and Y0 registers—contains the
value $2000:8000, and the 36-bit accumulator contains the value $0:2000:8000. In addition, the initial
value of C is set to one. The ADC instruction automatically sign extends the 32-bit Y register to 36 bits
and adds this value to the 36-bit accumulator. The carry bit, C, is added into the LSB of this 36-bit
operation. The 36-bit result is stored back in the A accumulator, and the condition codes are set appropriately. The Y1:Y0 register pair is not affected by this instruction.
Note:
C is set correctly for multi-precision arithmetic, using long-word operands only when the extension
register of the destination accumulator (FF2) contains only sign extension information (bits 31 through
35 are identical in the destination accumulator).
Condition Codes Affected:
MR
L
E
U
N
Z
V
C
A-10
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extended portion of accumulator result is in use
Set according to the standard definition of the U bit
Set if bit 35 of accumulator result is set
Set if accumulator result is zero; cleared otherwise
Set if overflow has occurred in accumulator result
Set if a carry (or borrow) occurs from bit 35 of accumulator result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADC
ADC
Add Long with Carry
Instruction Fields:
Operation
Operands
C
W
Comments
ADC
Y,F
1
1
Add with carry (set C bit also)
12
11
1
0
Instruction Opcodes:
15
ADC
Y,F
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
1
1
1
0
Instruction Set Details
8
7
0
F
0
0
4
3
0
0
0
1
1
1
A-11
ADD
Operation:
S+D→
S+D→
S+D→
ADD
Add
Assembler Syntax:
D
D
D
(no parallel move)
(one parallel move)
(two parallel reads)
ADD
ADD
ADD
S,D
S,D
S,D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Add the source register to the destination register and store the result in the destination (D). If the destination is a 36-bit accumulator, 16-bit source registers are first sign extended internally and concatenated with 16 zero bits to form a 36-bit operand (the Y register is only sign extended). When the destination is X0, Y0, or Y1, 16-bit addition is performed. In this case, if the source operand is one of the
four accumulators, the FF1 portion (properly sign extended) is used in the 16-bit addition; the FF2 and
FF0 portions are ignored. Similarly, if the destination is the Y register, the FF2 portion is ignored.
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Example:
ADD
X0,A
X:(R2)+N,X0 ; 16-bit addition, load X0, update R2
Before Execution
After Execution
0
0058
1234
0
005A
1234
A2
A1
A0
A2
A1
A0
X0
0002
X0
3456
R2
002001
R2
002000
N
FFFFFF
N
FFFFFF
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $0002, and the 36-bit A accumulator contains the value $0:0058:1234. The ADD instruction automatically appends the 16-bit value in the X0
register with 16 LS zeros, sign extends the resulting 32-bit long word to 36 bits, and adds the result to
the 36-bit A accumulator. A new word is read into the X0 register and address register R2 is updated
by –1.
Note:
A-12
The carry bit (C) in the CCR is set correctly using word or long-word source operands if the extension
register of the destination accumulator contains sign extension from bit 31 of the destination accumulator. C is always set correctly using accumulator source operands.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD
ADD
Add
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extended portion of the accumulator result is in use
Set if the result is unnormalized
Set if the high-order bit of the result is set
Set if the result equals zero
Set if overflow has occurred in the result
Set if a carry occurs from the high-order bit of the result
Instruction Fields:
Operation
Operands
C
W
ADD
FFF,FFF
1
1
Comments
36-bit add two registers.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
ADD2
X0,F
Y1,F
Y0,F
C,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A,B
B,A
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-13
ADD
ADD
Add
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
ADD2
X0,F
Y1,F
Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
A,B
B,A
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
A-14
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD
ADD
Add
Instruction Opcodes:
15
ADD
FFF,FFF
ADD
C,F GGG,X:<ea_m>
0
1
1
0
0
15
0
15
ADD
C,F X:<ea_m>,GGG
0
0
1
15
ADD
DD,F GGG,X:<ea_m>
ADD
DD,F X:<ea_m>,GGG
ADD
DD,F X:<ea_m>,reg1
X:<ea_v>,reg2
0
0
0
0
1
1
1
15
0
15
0
15
ADD
~F,F GGG,X:<ea_m>
ADD
~F,F X:<ea_m>,GGG
ADD
~F,F X:<ea_m>,reg1
X:<ea_v>,reg2
0
0
0
0
1
1
1
15
0
15
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
12
11
1
1
12
11
0
0
12
11
0
0
12
11
0
0
12
11
0
0
12
11
0
0
12
11
0
0
12
11
0
0
12
11
0
0
8
7
F
0
F
F
8
7
G
G
G
F
8
7
G
F
8
7
F
G
G
G
G
G
8
7
G
G
G
F
8
7
0
v
v
F
8
7
F
G
G
G
8
7
G
G
G
F
8
7
0
v
v
F
Instruction Set Details
4
3
0
b
b
b
4
3
1
1
0
0
4
3
0
0
4
3
0
1
1
J
J
J
4
3
J
J
J
0
4
3
v
J
J
0
4
3
0
0
0
0
4
3
0
0
0
0
4
3
v
1
0
0
0
0
0
0
m
R
R
0
0
m
R
R
0
m
R
R
m
R
R
m
0
v
0
0
0
m
R
R
m
R
R
m
0
v
0
0
A-15
ADD.B
Operation:
S+D→
ADD.B
Add Byte (Word Pointer)
Assembler Syntax:
D
(no parallel move)
ADD.B
S,D
(no parallel move)
Description: Add a 9-bit signed immediate integer to the 8-bit portion of the destination register, and store the result
in the destination (D). The value is internally sign extended to 20 bits before the operation. If the destination is a 16-bit register, it is first correctly sign extended before the 20-bit addition is performed.
The immediate integer is used to represent 8-bit unsigned values from 0 to 255 as well as the signed
range: –128 to 127. The condition codes are calculated based on the 8-bit result, with the exception of
the E and U bits, which are calculated based on the 20-bit result. The result is not affected by the state
of the saturation bit (SA).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Example:
ADD.B
#$55,A
; add hex 55 to A accumulator
Before Execution
After Execution
0
3122
1234
0
3177
1234
A2
A1
A0
A2
A1
A0
SR
SR
0300
0310
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:3122:1234. The ADD.B instruction
automatically sign extends the immediate value to 20 bits and then adds the result to the A2:A1 portion
of the A accumulator. The 8-bit result ($77) is stored back into the low-order 8 bits of A1.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if bit 7 of the result is set
Set if the result equals zero
Set if overflow has occurred in the result
Set if a carry occurs from bit 7 of the result
Instruction Fields:
Operation
Operands
C
W
Comments
ADD.B
#xxx,EEE
2
2
Add 9-bit signed immediate
12
11
0
0
Instruction Opcodes:
15
ADD.B #xxx,EEE
0
1
0
1
E
8
7
E
E
1
0
4
3
0
0
0
0
1
0
iiiiiiiiiiiiiiii
Timing:
2 oscillator clock cycle
Memory:
2 program word
A-16
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD.BP
Operation:
S+D→
ADD.BP
Add Byte (Byte Pointer)
Assembler Syntax:
D
(no parallel move)
ADD.BP
S,D
(no parallel move)
Description: Add a byte stored in memory to the 8-bit portion of the destination register, and store the result in the
destination (D). The value is internally sign extended to 20 bits before the operation. If the destination
is a 16-bit register, it is first correctly sign extended before the 20-bit addition is performed. The condition codes are calculated based on the 8-bit result, with the exception of the E and U bits, which are
calculated based on the 20-bit result. Absolute addresses are expressed as byte addresses. The result is
not affected by the state of the saturation bit (SA).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Example:
ADD.BP X:$4000,A
; add byte at word address $2000
; to A accumulator
After Execution
Before Execution
0
3122
1234
0
3177
1234
A2
A1
A0
A2
A1
A0
(word address) X:$2000
FF55
X:$2000
FF55
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:3122:1234. The ADD.BP instruction automatically sign extends the memory byte to 20 bits and then adds the result to the A2:A1 portion of the A accumulator. The 8-bit result ($77) is stored back into the low-order 8 bits of A1.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if bit 7 of the result is set
Set if the result equals zero
Set if overflow has occurred in the result
Set if a carry occurs from bit 7 of the result
Freescale Semiconductor
Instruction Set Details
A-17
ADD.BP
ADD.BP
Add Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
Comments
ADD.BP
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
Add memory byte to register; address is expressed as
byte address
Instruction Opcodes:
15
ADD.BP X:xxxx,EEE
0
1
0
12
11
0
0
1
E
8
7
E
E
1
0
4
3
0
0
4
3
0
1
1
0
AAAAAAAAAAAAAAAA
15
ADD.BP X:xxxxxx,EEE
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
1
E
E
E
1
0
0
0
1
1
0
AAAAAAAAAAAAAAAA
Timing:
2–3 oscillator clock cycles
Memory:
2–3 program words
A-18
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD.L
Operation:
S+D→
ADD.L
Add Long
Assembler Syntax:
D
(no parallel move)
ADD.L
S,D
(no parallel move)
Description: Add a long-word value in memory or a 16-bit signed immediate value to the second operand, and store
the result in the destination (D). Source values are internally sign extended to 36 bits before the addition. Condition codes are calculated based on the 32-bit result, with the exception of the E and U bits,
which are calculated based on the 36-bit result for accumulator destinations. Absolute addresses pointing to long elements must always be even aligned (that is, pointing to the lowest 16 bits).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Example:
ADD.L
X:$4000,A
; add long value at word address $4001:4000
; to A accumulator
After Execution
Before Execution
0
6666
1111
0
8888
2222
A2
A1
A0
A2
A1
A0
X:$4001
2222
X:$4001
2222
X:$4000
1111
X:$4000
1111
SR
0300
SR
032A
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:6666:1111. The ADD.L instruction
automatically sign extends the long value at address X:$4001:4000 to 36 bits and adds the result to the
A accumulator. The 32-bit result ($8888:2222) is stored back into the accumulator.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extended portion of the 36-bit result is in use
Set if the 36-bit result is unnormalized
Set if bit 31 of the result is set
Set if bits 31–0 of the result are zero
Set if overflow has occurred in the result
Set if a carry occurs from bit 31 of the result
Freescale Semiconductor
Instruction Set Details
A-19
ADD.L
ADD.L
Add Long
Instruction Fields:
Operation
Operands
C
W
Comments
ADD.L
X:xxxx,fff
2
2
X:xxxxxx,fff
3
3
#xxxx,fff
2
2
Add a 16-bit immediate value sign extended to 32 bits
to a data register
12
11
0
0
Add memory long to register
Instruction Opcodes:
15
ADD.L #xxxx,fff
0
1
0
1
f
8
7
f
f
4
3
0
1
0
0
4
3
1
0
0
0
0
0
1
1
1
1
1
iiiiiiiiiiiiiiii
15
ADD.L X:xxxx,fff
0
1
0
12
11
0
0
1
f
8
7
f
f
0
AAAAAAAAAAAAAAAA
15
ADD.L X:xxxxxx,fff
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
0
0
1
f
f
f
1
0
0
0
1
1
1
AAAAAAAAAAAAAAAA
Timing:
2–3 oscillator clock cycles
Memory:
2–3 program words
A-20
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD.W
Operation:
S+D→
ADD.W
Add Word
Assembler Syntax:
D
(no parallel move)
ADD.W
S,D
(no parallel move)
Description: Add the source operand to the second operand (register or memory), and store the result in the destination (D). The source operand (except for a short immediate operand) is first sign extended internally
to form a 20-bit value; this value is concatenated with 16 zero bits to form a 36-bit value when the destination is one of the four accumulators. A short immediate (0–31) source operand is zero extended
before the addition. The addition is then performed as a 20-bit operation. Condition codes are calculated based on the size of the destination.
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Example:
ADD.W
#3,A
; add decimal 3 to A
Before Execution
After Execution
0
0058
1234
0
005B
1234
A2
A1
A0
A2
A1
A0
SR
SR
0300
0310
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0058:1234. The ADD.W instruction
automatically sign extends the immediate value to 20 bits and adds the result to accumulator A. The
result is stored back in A.
Condition Codes Affected:
MR
L
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set if overflow has occurred in the result
Set if the extended portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if the high-order bit of the result is set
Set if the result equals zero (accumulator bits 35–0 or bits 15–0 of a 16-bit register)
Set if overflow has occurred in the result
Set if a carry occurs from the high-order bit of the result
Freescale Semiconductor
Instruction Set Details
A-21
ADD.W
ADD.W
Add Word
Instruction Fields:
A-22
Operation
Operands
C
W
ADD.W
X:(Rn),EEE
2
1
X:(Rn+xxxx),EEE
3
2
X:(SP–xx),EEE
3
1
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
EEE,X:(SP–xx)
4
2
Comments
Add memory word to register
EEE,X:xxxx
3
2
Add register to memory word, storing the result back to
memory
#<0–31>,EEE
1
1
Add an immediate integer 0–31 (zero extended)
#xxxx,EEE
2
2
Add a signed 16-bit immediate
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADD.W
ADD.W
Add Word
Instruction Opcodes:
15
ADD.W #<0–31>,EEE
0
1
0
15
ADD.W #xxxx,EEE
0
1
0
12
11
0
0
12
11
0
0
1
E
1
E
8
7
E
E
8
7
E
E
0
1
0
0
4
3
B
B
4
3
0
0
4
3
a
a
4
3
1
0
0
B
B
B
0
0
0
0
iiiiiiiiiiiiiiii
15
ADD.W EEE,X:(SP–xx)
0
1
0
12
11
0
0
0
E
8
7
E
E
1
a
0
a
a
a
$E702
15
ADD.W EEE,X:xxxx
0
1
1
12
11
1
0
1
E
8
7
E
E
1
0
0
1
1
1
1
R
R
AAAAAAAAAAAAAAAA
15
ADD.W X:(Rn),EEE
0
1
0
15
ADD.W X:(Rn+xxxx),EEE
0
1
0
12
11
0
0
12
11
0
0
1
E
1
E
8
7
E
E
8
7
E
E
1
1
0
0
4
3
1
R
4
3
1
R
0
0
0
R
R
a
a
a
AAAAAAAAAAAAAAAA
15
ADD.W X:(SP–xx),EEE
0
1
0
15
ADD.W X:xxxx,EEE
0
1
0
12
11
0
0
12
11
0
0
0
E
1
E
8
7
E
E
8
7
E
E
1
1
a
0
4
3
a
a
4
3
0
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
15
ADD.W X:xxxxxx,EEE
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
0
0
1
E
E
E
1
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–4 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-23
ADDA
ADDA
Add AGU Register
Operation:
Assembler Syntax:
S+D→D
S1 + S2 → D
(no parallel move)
(no parallel move)
ADDA
ADDA
S,D
(no parallel move)
S1,S2,D (no parallel move)
Description: Add an AGU register or immediate value to an AGU register or a data ALU register, and store the result in the second AGU register, a separate address pointer register, or the N register. The addition is
performed using 24-bit two’s-complement arithmetic. Immediate values that are less than 24 bits in
length are either sign extended or zero extended to 24 bits before the addition takes place. Refer to
Section 6.8.4.3 on page 6-28 when using “ADDA #<immediate_value>,Rn” in Modulo Addressing.
Example:
ADDA
#$254,R0,R1
; add hex 254 to R0 and store the result in R1
Before Execution
After Execution
R0
005000
R0
005000
R1
17C624
R1
005254
Explanation of Example:
The address pointer register R0 initially contains $005000, while R1 initially contains $17C624. When
the ADDA #$254,R0,R1 instruction is executed, the immediate hexadecimal value 254 is added to
the value in R0, and the result is stored in address register R1.
Condition Codes Affected:
The condition codes are not affected by this instruction
Instruction Fields:
Operation
Operands
C
W
Comments
ADDA
Rn,Rn
1
1
Add first operand to the second and store the result in the second operand.
Rn,Rn,N
1
1
Add first operand to the second and store result in the N register.
#<0–15>,Rn
1
1
Add unsigned 4-bit value to Rn.
#<0–15>,Rn,N
1
1
Add an unsigned 4-bit value to an AGU register and store
result in the N register.
#xxxx,Rn,Rn
2
2
Add first register with a signed 17-bit immediate value and
store the result in Rn.
#xxxx,Rn
2
2
An alternate syntax for the preceding instruction if the second
source and the destination are the same.
#xxxxxx,Rn,Rn
3
3
Add first register with a 24-bit immediate value and store the
result in Rn.
#xxxxxx,Rn
3
3
An alternate syntax for the preceding instruction if the second
source and the destination are the same.
#xxxx,HHH,Rn
4
2
Add a data register with an unsigned 16-bit value and store
the result in Rn. HHH is accessed as a signed 16-bit word.
#xxxxxx,HHH,Rn
5
3
Add a data register with a 24-bit immediate value and store
the result in Rn. HHH is accessed as a signed 16-bit word.
A-24
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADDA
ADDA
Add AGU Register
Instruction Opcodes:
15
ADDA
#<0–15>,Rn
1
0
0
15
ADDA
#<0–15>,Rn,N
1
0
0
15
ADDA
#xxxx,HHH,Rn
1
0
0
12
11
0
i
12
11
0
i
12
11
0
0
i
i
i
i
1
1
8
7
i
0
8
7
i
0
8
7
0
h
1
1
0
1
1
1
4
3
1
R
4
3
1
R
4
3
h
R
4
3
n
R
4
3
0
0
R
R
0
1
R
R
0
h
R
R
AAAAAAAAAAAAAAAA
15
ADDA
#xxxx,Rn,Rn
1
0
0
12
11
0
V
0
1
8
7
0
n
0
1
0
n
R
R
AAAAAAAAAAAAAAAA
15
ADDA
#xxxxxx,HHH,Rn
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
1
1
0
h
0
1
h
R
h
R
R
AAAAAAAAAAAAAAAA
15
ADDA
#xxxxxx,Rn,Rn
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
0
0
1
0
n
0
1
n
R
n
R
R
4
3
n
R
4
3
n
R
AAAAAAAAAAAAAAAA
15
ADDA
Rn,Rn
1
0
0
15
ADDA
Rn,Rn,N
1
0
0
12
11
0
1
12
11
0
1
0
1
0
0
8
7
1
n
8
7
1
n
0
0
1
1
Timing:
1–5 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Note:
Refer to Section 6.8.4.3 on page 6-28 when ADDA is used in Modulo Arithmetic.
Freescale Semiconductor
Instruction Set Details
0
n
R
R
0
n
R
R
A-25
ADDA.L
Add to Left-Shifted AGU Register
Operation:
Assembler Syntax:
(S << 1) + D → D
(no parallel move)
S1 + (S2 << 1) → D (no parallel move)
ADDA.L
ADDA.L
ADDA.L
S,D
(no parallel move)
S1,S2,D (no parallel move)
Description: Left shift one of the source operands by one (S or S2), and add it either to the destination or to the other
source operand (S1). Store the result in the destination AGU register (D).
Usage:
The ADDA.L instruction is most useful for accessing arrays of long words in memory. The address of
an element in the array is calculated by adding the base address to the index value multiplied by 2
(since long words occupy 2 words in memory). The ADDA.L instruction can accomplish this in one
step.
Example:
ADDA.L #$4000,R0,R1
Before Execution
; add $4000 to left-shifted R0 and store the
; result in R1
After Execution
R0
000044
R0
000044
R1
000624
R1
004088
Explanation of Example:
The address pointer register R0 initially contains $000044, while R1 initially contains $000624. When
the ADDA.L #$4000,R0,R1 instruction is executed, R0 is internally shifted 1 bit to the left, resulting in the intermediate value $000088. The immediate value $4000 is then added to the shifted value,
and the result ($004088) is stored in address register R1.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-26
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ADDA.L
ADDA.L
Add to Left-Shifted AGU Register
Instruction Fields:
Operation
Operands
C
W
Comments
ADDA.L
Rn,Rn
1
1
Add first operand, left shifted 1 bit, to the second, and store the
result in the second operand
Rn,Rn,N
1
1
Add first operand, left shifted 1 bit, to the second, and store result
in the N register
#xxxx,Rn,Rn
2
2
Add first register, left shifted 1 bit, with an unsigned 16-bit immediate value, and store the result in Rn
#xxxx,Rn
2
2
An alternate syntax for the preceding instruction if the second
source and the destination are the same
#xxxxxx,Rn,Rn
3
3
Add first register, left shifted 1 bit, with a 24-bit immediate value,
and store the result in Rn
#xxxxxx,Rn
3
3
An alternate syntax for the preceding instruction if the second
source and the destination are the same
#xxxx,HHH,Rn
4
2
Add data register, left shifted 1 bit, with unsigned 16-bit immediate value, store result in Rn; HHH is accessed as 16-bit signed
#xxxxxx,HHH,Rn
5
3
Add data register, left shifted 1 bit, with a 24-bit immediate value,
and store the result in Rn; HHH is accessed as 16-bit signed
Instruction Opcodes:
15
ADDA.L #xxxx,HHH,Rn
1
0
0
12
11
0
0
1
1
8
7
1
h
0
1
4
3
h
R
4
3
n
R
4
3
0
h
R
R
AAAAAAAAAAAAAAAA
15
ADDA.L #xxxx,Rn,Rn
1
0
0
12
11
0
0
0
1
8
7
1
n
0
1
0
n
R
R
AAAAAAAAAAAAAAAA
15
ADDA.L #xxxxxx,HHH,Rn
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
1
1
1
h
0
1
h
R
h
R
R
AAAAAAAAAAAAAAAA
15
ADDA.L #xxxxxx,Rn,Rn
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
0
0
1
1
n
0
1
n
R
n
R
R
4
3
n
R
4
3
n
R
AAAAAAAAAAAAAAAA
15
ADDA.L Rn,Rn
1
0
0
15
ADDA.L Rn,Rn,N
Timing:
1–5 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
1
0
0
12
11
0
1
12
11
0
1
0
1
0
0
Instruction Set Details
8
7
0
n
8
7
0
n
0
0
1
1
0
n
R
R
0
n
R
R
A-27
ALIGNSP
ALIGNSP
Align Stack Pointer
Operation:
Assembler Syntax:
If SP is odd:
ALIGNSP(no parallel move)
SP + 2 → SP
else if SP is even:
SP + 3 → SP
SP → X:(SP)
SP + 2 → SP
Description: The ALIGNSP instruction aligns the stack pointer register (SP) correctly for a long-word value to be
pushed onto the stack. The SP should point to the (odd) upper word address of the long word in order
for it to be pushed and popped properly. The ALIGNSP instruction guarantees that the SP points to an
odd word address and that at least 2 words are available to receive the long-word value. The value of
the SP previous to the alignment adjustment is placed on the stack (as a long word) so the stack can be
restored to its original state.
Usage:
ALIGNSP should be used to align the stack prior to pushing a long-word value.
Example:
ALIGNSP
MOVE.L Y,X:(SP)+
; align the stack for a long word
; push long word on stack
Before Execution
After Execution
SP
X:$1007
X:$1006
X:$1005
‘Y1’
‘Y0’
0000
1001
5499
0000
X:$1004
X:$1003
X:$1001
X:$1000
5499
0000
X:$1002
X:$1001
X:$1000
SP
Explanation of Example:
The SP register initially has a value of $001001. Since the initial value of SP is odd, it is only incremented by two, the original value is pushed onto the stack, and SP is updated. After ALIGNSP is executed, the SP has a new value of $001005. The MOVE.L instruction adds two to the SP (for the
post-increment) after pushing register Y onto the stack, setting the final SP value to $001007.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
ALIGNSP
C
W
Comments
3
1
Save SP to the stack and align SP for long memory accesses, pointing
to an empty location
Instruction Opcodes:
15
ALIGNSP
1
Timing:
3 oscillator clock cycles
Memory:
1 program word
A-28
1
1
12
11
0
0
1
1
8
7
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
4
3
0
0
0
1
0
0
Freescale Semiconductor
AND.L
AND.L
AND Long
Operation:
Assembler Syntax:
S•D→D
(no parallel move)
where • denotes the logical AND operator
AND.L
S,D
(no parallel move)
Description: Perform a logical AND operation on the source operand and the destination operand, and store the result in the destination. This instruction is a 32-bit operation. If the destination is a 36-bit accumulator,
the AND operation is performed on the source and bits 31–0 of the accumulator. The remaining bits
of the destination accumulator are not affected. If the source is a 16-bit register, it is first internally
concatenated with 16 zero bits to form a 32-bit operand. If the source is an immediate 5-bit constant,
it is first zero extended to form a 32-bit operand. When the destination is an accumulator, bits 35–32
remain unchanged. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used for the logical AND of two registers or of a register and a small immediate
value. The ANDC instruction is appropriate for performing an AND operation on a 16-bit immediate
value and a register or memory location.
Example:
AND.L
Y,A
; logically AND Y with A10
After Execution
Before Execution
6
1234
5678
6
1200
0078
A2
A1
A0
A2
A1
A0
7F00
00FF
7F00
00FF
Y1
Y0
Y1
Y0
SR
SR
0302
0300
Explanation of Example:
Prior to execution, the 32-bit Y register contains the value $7F00:00FF, and the 36-bit A accumulator
contains the value $6:1234:5678. The AND.L Y,A instruction performs a logical AND operation on
the 32-bit value in the Y register and on bits 31–0 of the A accumulator (A10), and it stores the 36-bit
result in the A accumulator. Bits 35–32 in the A2 register are not affected by this instruction.
Condition Codes Affected:
MR
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 31 of accumulator or register result is set
— Set if bits 31–0 of accumulator or register result are zero
— Always cleared
Freescale Semiconductor
Instruction Set Details
A-29
AND.L
AND.L
AND Long
Instruction Fields:
Operation
Operands
C
W
Comments
AND.L
#<0–31>,fff
1
1
AND with a zero-extended 5-bit positive immediate
integer (0–31)
FFF,fff
1
1
32-bit logical AND
Instruction Opcodes:
15
AND.L #<0–31>,fff
0
1
0
1
1
15
AND.L FFF,fff
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-30
12
11
0
0
12
11
1
1
1
f
0
f
8
7
f
f
8
7
f
f
4
3
B
1
1
B
4
3
b
b
b
1
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
1
0
0
0
Freescale Semiconductor
AND.W
AND.W
AND Word
Operation:
Assembler Syntax:
S•D→D
S • D[31:16] → D[31:16]
(no parallel move)
(no parallel move)
AND.W
AND.W
S,D
S,D
(no parallel move)
(no parallel move)
where • denotes the logical AND operator
Description: Perform a logical AND operation on the source operand (S) and the destination operand (D), and store
the result in the destination. This instruction is a 16-bit operation. If the destination is a 36-bit accumulator, the operation is performed on the source and bits 31–16 of the accumulator. The remaining bits
of the destination accumulator are not affected. If the source is an immediate 5-bit constant, it is first
zero extended to form a 32-bit operand. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used for the logical AND of two registers or of a register and a small immediate
value. The ANDC instruction is appropriate for performing an AND operation on a 16-bit immediate
value and a register or memory location.
Example:
AND.W
X0,A
; logically AND X0 with A1
Before Execution
After Execution
6
1234
5678
6
1200
5678
A2
A1
A0
A2
A1
A0
X0
7F00
X0
7F00
SR
SR
030F
0301
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $7F00, and the 36-bit A accumulator contains the value $6:1234:5678. The AND.W X0,A instruction performs a logical AND operation on the
16-bit value in the X0 register and on bits 31–16 of the A accumulator (A1), and it stores the 36-bit
result in the A accumulator. Bits 35–32 in the A2 register and bits 15–0 in the A0 register are not affected by this instruction.
Condition Codes Affected:
MR
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 31 of accumulator result or MSB of register result is set
— Set if bits 31–16 of accumulator result or all bits of register result are zero
— Always cleared
Freescale Semiconductor
Instruction Set Details
A-31
AND.W
AND.W
AND Word
Instruction Fields:
Operation
Operands
C
W
Comments
AND.W
#<0–31>,EEE
1
1
AND with a zero-extended 5-bit positive immediate integer (0–31)
EEE,EEE
1
1
16-bit logical AND
Instruction Opcodes:
15
AND.W #<0–31>,EEE
0
1
0
1
1
15
AND.W EEE,EEE
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-32
12
11
1
0
12
11
1
1
8
7
E
1
E
E
8
7
0
E
E
E
4
3
B
1
1
B
4
3
a
a
a
1
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
0
0
0
0
Freescale Semiconductor
ANDC
ANDC
Logical AND Immediate
Operation:
Assembler Syntax:
#xxxx • D → D
(no parallel move)
#xxxx • X:<ea> → X:<ea> (no parallel move)
where • denotes the logical AND operator
ANDC
ANDC
#iiii,D (no parallel move)
#iiii,X:<ea>(no parallel move)
Implementation Note:
This instruction is implemented by the assembler as an alias to the BFCLR instruction, with the 16-bit
immediate value inverted (one’s-complement) and used as the bit mask. It will dis-assemble as a
BFCLR instruction.
Description: Perform a logical AND operation on a 16-bit immediate data value with the destination operand, and
store the results back into the destination. C is also modified as described in “Condition Codes Affected.” This instruction performs a read-modify-write operation on the destination and requires two destination accesses.
Example:
ANDC
#$0055,X:$5000
; AND with immediate data
Before Execution
After Execution
X:$5000
FFFF
X:$5000
0055
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $FFFF. Execution of the
instruction performs a logical AND operation on the 16-bit value in X:$5000 (that is, $FFFF) and the
mask value $0055 and stores the result in X:$5000. The C bit is set because all of the bits selected by
the inverted value of the mask are set.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
For this destination only, the C bit is not updated as is done for all other destination operands.
All SR bits except bits 14–10 are updated with values from the bitfield unit.
Bits 14–10 of the mask operand must be set.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the one’s-complement of the mask are set
Cleared if at least 1 bit specified by the one’s-complement of the mask is not set
Note:
If all bits in the mask are set, the instruction executes two NOPs and sets the C bit.
Instruction Fields:
Refer to the section on the BFCLR instruction for legal operand and timing information.
Freescale Semiconductor
Instruction Set Details
A-33
ASL
ASL
Arithmetic Shift Left
Operation:
Assembler Syntax:
(see following figure)
ASL
ASL
ASL
D
D
D
(no parallel move)
(one parallel move)
(two parallel reads)
:
0
C
D2
D1
D0
Description: Arithmetically shift the destination operand (D) 1 bit to the left, and store the result in the destination.
The MSB of the destination prior to the execution of the instruction is shifted into C, and a zero is shifted into the LSB of the destination. If the destination is the Y register, the MSB is bit 31. A duplicate
destination is not allowed when ASL is used in conjunction with a parallel read. For arithmetic shifts
left on 16-bit registers, refer to ASL.W.
Usage:
This instruction can be used to cast a long to an integer value.
Example:
ASL
A
X:(R3)+N,Y0; shift A left by 1, update R3 and Y0
Before Execution
After Execution
A
0111
0222
4
0222
0444
A2
A1
A0
A2
A1
A0
SR
SR
0300
0373
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $A:0111:0222. Execution of the ASL
instruction shifts the 36-bit value in the A accumulator 1 bit to the left and stores the result back in the
A accumulator. C is set by the operation because bit 35 of A was set prior to the execution of the instruction. The V bit of CCR (bit 1) is also set because bit 35 of A has changed during the execution of
the instruction. The U bit of CCR (bit 4) is set because the result is not normalized, the E bit of CCR
(bit 5) is set because the extension portion of the result is in use, and the L bit of CCR (bit 6) is set
because an overflow has occurred. A new value for register Y0 is read and address register R3 is updated by the contents on index register N.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
A-34
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of SZ (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of accumulator result is in use
Set according to the standard definition of the U bit
Set if bit 35 of accumulator result is set
Set if accumulator result equals zero
Set if bit 35 of accumulator result is changed due to left shift
Set if bit 35 of accumulator was set prior to the execution of the instruction
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ASL
ASL
Arithmetic Shift Left
Instruction Fields:
Operation
Operands
C
W
ASL
fff
1
1
Comments
Arithmetic shift left entire register by 1 bit.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
ASL2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-35
ASL
ASL
Arithmetic Shift Left
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
ASL2
F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
ASL
F GGG,X:<ea_m>
0
0
0
15
ASL
F X:<ea_m>,GGG
ASL
F X:<ea_m>,reg1
X:<ea_v>,reg2
ASL
fff
0
0
1
1
1
1
1
15
0
15
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-36
12
11
1
0
12
11
1
0
12
11
1
0
12
11
1
0
G
G
8
7
G
F
8
7
F
G
G
G
8
7
1
v
v
F
8
7
0
f
f
f
0
1
4
3
1
0
4
3
0
0
1
1
4
3
v
1
1
0
4
3
1
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
m
R
R
0
m
R
R
m
0
v
0
1
1
0
0
Freescale Semiconductor
ASL.W
ASL.W
Arithmetic Shift Left
Operation:
Assembler Syntax:
(see following figure)
ASL.W
D
(no parallel move)
:
15
0
C
0
Description: Arithmetically shift the destination operand (D) 1 bit to the left, and store the result in the destination
register. The MSB, bit 15 of the destination prior to the execution of the instruction, is shifted into C,
and a zero is shifted into the LSB of the destination. This instruction is used only when the destination
is X0, Y0, or Y1 register. For the purpose of calculating condition code, the 16-bit register is first sign
extended and concatenated to 16 zero bits to form a 36-bit operand. For arithmetic shifts left on the Y
register or accumulator, refer to ASL.
Example:
ASL.W
Y0
; shift Y0 left by 1
Before Execution
After Execution
2000
C000
2000
8000
Y1
Y0
Y1
Y0
SR
SR
0300
0309
Explanation of Example:
Prior to execution, the 16-bit Y0 register contains the value $C000. Execution of the ASL.W instruction shifts the 16-bit value in Y0 by 1 bit to the left and stores the result back in Y0. C is set by the
operation because bit 15 of Y0 was set prior to the execution of the instruction. The N bit is set because
the MSB of the result is set.
Condition Codes Affected:
MR
L
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extension portion of the result is in use
Set according to the standard definition of the U bit
Set if bit 15 of result is set
Set if the result equals zero
Set if bit 15 of result is changed due to left shift
Set if bit 15 of was set prior to the execution of the instruction
Freescale Semiconductor
Instruction Set Details
A-37
ASL.W
ASL.W
Arithmetic Shift Left
Instruction Fields:
Operation
Operands
C
W
ASL.W
DD
1
1
Comments
Arithmetic shift left entire register by 1 bit
Instruction Opcodes:
15
ASL.W DD
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-38
1
1
12
11
1
0
0
1
8
7
D
D
1
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
0
0
0
0
1
1
Freescale Semiconductor
ASL16
Operation:
S << 16 →
ASL16
Arithmetic Shift Left 16 Bits
Assembler Syntax:
D
(no parallel move)
ASL16
S,D
(no parallel move)
Description: Arithmetically shift the source operand to the left by 16 bits, and store the result in the destination (D).
This operation effectively places the LSP of the source register into the MSP of the destination register.
The low-order 16 bits of the destination are always set to zero. Bits are shifted into the extension register (FF2) if the destination is an accumulator. When the destination operand is a 16-bit register, the
LSP of an accumulator or Y register is written to it. When both the source and destination are 16-bit
registers, the destination is cleared. The result is not affected by the state of the saturation bit (SA).
Example:
ASL16
Y,A
; shift Y left 16 bits, store in A
After Execution
Before Execution
0
3456
3456
0
7FFF
0000
A2
A1
A0
A2
A1
A0
0000
7FFF
0000
7FFF
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the Y register contains the value to be shifted ($0000:7FFF). The contents of the
destination register are not important prior to execution because they have no effect on the calculated
value. The ASL16 instruction arithmetically shifts the value $0000:7FFF by 16 bits to the left and places the result in the destination register A.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
Comments
ASL16
FFF,FFF
1
1
Arithmetic shift left the first operand by 16 bits, placing
result in the destination operand
FFF
1
1
An alternate syntax for the preceding instruction if the
source and the destination are the same
Instruction Opcodes:
15
ASL16 FFF,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
1
1
F
Instruction Set Details
8
7
F
F
b
b
4
3
b
0
0
1
0
1
A-39
ASLA
Operation:
S << 1 →
ASLA
1-Bit Left Shift AGU Register
Assembler Syntax:
D
(no parallel move)
ASLA
S,D
(no parallel move)
Description: Arithmetically shift the source address register 1 bit to the left, and store the result in the destination
register.
Example:
ASLA
R1,R0
; shift R1 left 1 bit and store in R0
Before Execution
After Execution
R0
00B360
R0
008888
R1
004444
R1
004444
Explanation of Example:
Prior to execution, the R1 register contains the value $004444, and the R0 register contains $00B360.
Execution of the ASLA instruction shifts the value in R1 by 1 bit to the left and stores the result
($008888) in the R0 register.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
Comments
ASLA
Rn,Rn
1
1
Arithmetic shift left AGU register by 1 bit
Rn
1
1
An alternate syntax for the preceding instruction if the
source and the destination are the same
Instruction Opcodes:
15
ASLA
Rn,Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-40
0
0
12
11
0
0
0
0
8
7
0
n
0
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
n
R
0
n
R
R
Freescale Semiconductor
ASLL.L
ASLL.L
Multi-Bit Arithmetic Left Shift Long
Operation:
Assembler Syntax:
If S[15] = 0 or S is not a register,
D << S → D
(no parallel move)
Else
D >> –S → D
(no parallel move)
ASLL.L
S,D
(no parallel move)
ASLL.L
S,D
(no parallel move)
Description: Arithmetically shift the second operand to the left by the value contained in the 5 lowest bits of the first
operand (or by an immediate integer). Store the result back in the destination (D) with zeros shifted
into the LSB. The shift count can be a 5-bit positive immediate integer or the value contained in X0,
Y0, Y1, or the MSP of an accumulator. For 36- and 32-bit destinations, the MSP:LSP are shifted with
sign extension from bit 31 (the FF2 portion is ignored). If the shift count in a register is negative (bit
15 is set), the direction of the shift is reversed, maintaining sign integrity. The result is not affected by
the state of the saturation bit (SA).
Example:
ASLL.L Y0,A
; shift A left by amount in Y0 and store in A
After Execution
Before Execution
0
0123
4567
0
1234
5670
A2
A1
A0
A2
A1
A0
2000
0024
2000
0024
Y1
Y0
Y1
Y0
SR
SR
0300
0300
Explanation of Example:
Prior to execution, the A accumulator contains the value to be shifted ($0123:4567), and the Y0 register contains the amount by which to shift ($04). The ASLL.L instruction arithmetically shifts the value
$0123:4567 by 4 bits to the left and places the result in the destination register A. Since the destination
is an accumulator, the extension word (A2) is filled with sign extension.
Condition Codes Affected:
MR
N
Z
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if MSB of result is set
— Set if result equals zero
Freescale Semiconductor
Instruction Set Details
A-41
ASLL.L
ASLL.L
Multi-Bit Arithmetic Left Shift Long
Instruction Fields:
Operation
Operands
C
W
Comments
ASLL.L
#<0–31>,fff
2
1
Arithmetic shift left by a 5-bit positive immediate integer
EEE,FFF
2
1
Bi-directional arithmetic shift of destination by value in
the first operand: positive –> left shift
Instruction Opcodes:
15
ASLL.L #<0–31>,fff
0
1
0
1
1
15
ASLL.L EEE,FFF
0
Timing:
2 oscillator clock cycles
Memory:
1 program word
A-42
12
11
0
0
12
11
1
1
8
7
f
1
f
f
8
7
1
F
F
F
4
3
B
0
1
B
4
3
a
a
a
1
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
1
1
0
0
Freescale Semiconductor
ASLL.W
ASLL.W
Multi-Bit Arithmetic Left Shift Word
Operation:
Assembler Syntax:
S1 << S2 → D
D << S → D
(no parallel move)
(no parallel move)
ASLL.W
ASLL.W
S1,S2,D
S,D
(no parallel move)
(no parallel move)
Description: This instruction can have two or three operands. It arithmetically shifts the source operand S1 or D to
the left by the value contained in the lowest 4 bits of either S2 or S, respectively (or by an immediate
integer), and stores the result in the destination (D) with zeros shifted into the LSB. The shift count can
be a 4-bit positive integer, a value in a 16-bit register, or the MSP of an accumulator. For 36- and 32-bit
destinations, only the MSP is shifted and the LSP is cleared, with sign extension from bit 31 (the FF2
portion is ignored). The result is not affected by the state of the saturation bit (SA).
Example:
ASLL.W Y1,X0,A
; shift Y1 left by amount in X0 and store in A
Before Execution
After Execution
0
3456
3456
F
AAA0
0000
A2
A1
A0
A2
A1
A0
AAAA
8000
AAAA
8000
Y1
Y0
Y1
Y0
X0
SR
0014
X0
0014
0300
SR
0308
Explanation of Example:
Prior to execution, the Y1 register contains the value to be shifted ($AAAA), and the least significant
4 bits of the X0 register contain the amount by which to shift ($4). The contents of the destination register are not important prior to execution because they have no effect on the calculated value. The
ASLL.W instruction arithmetically shifts the value $AAAA by 4 bits to the left and places the result
in the destination register A. Since the destination is an accumulator, the extension word (A2) is filled
with sign extension, and the LSP (A0) is cleared.
Condition Codes Affected:
MR
N
Z
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if MSB of result is set
— Set if result equals zero
If the CM bit is set, N is cleared. When the destination is a 16-bit register, condition codes are based
on the 16-bit result.
Freescale Semiconductor
Instruction Set Details
A-43
ASLL.W
ASLL.W
Multi-Bit Arithmetic Left Shift Word
Instruction Fields:
Operation
Operands
C
W
Comments
ASLL.W
#<0–15>,FFF
1
1
Arithmetic shift left by a 4-bit positive immediate integer
EEE,FFF
1
1
Arithmetic shift left destination by value specified in 4
LSBs of the first operand
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
1
1
Arithmetic shift left the first operand by value specified in
4 LSBs of the second operand; place result in FFF
Instruction Opcodes:
15
ASLL.W #<0–15>,FFF
0
1
0
1
1
1
1
15
ASLL.W EEE,FFF
0
15
ASLL.W Q1,Q2,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-44
0
12
11
1
1
12
11
1
1
12
11
1
0
8
7
F
1
F
F
8
7
1
F
F
F
8
7
0
F
F
F
4
3
B
1
1
1
4
3
a
a
a
1
4
3
Q
Q
Q
1
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
0
1
0
1
1
0
0
0
Freescale Semiconductor
ASR
Arithmetic Shift Right
Operation:
Assembler Syntax:
(see following figure)
ASR
ASR
ASR
ASR
D
D
D
(no parallel move)
(one parallel move)
(two parallel reads)
C
MSB
D2
D1
(parallel move)
D0
Description: Arithmetically shift the destination operand (D) 1 bit to the right and store the result in the destination
accumulator. The LSB of the destination prior to the execution of the instruction is shifted into C, and
the MSB of the destination is held constant. When the destination register is Y or a 16-bit register, the
MSB is bit 31 or bit 15, respectively. A duplicate destination is not allowed when ASR is used in conjunction with a parallel read.
Example:
ASR
B
X:(R3)+,Y0; divide B by 2, load Y0, and update R3
Before Execution
After Execution
8
AAAA
AAAA
C
5555
5555
B2
B1
B0
B2
B1
B0
SR
SR
0300
0328
Explanation of Example:
Prior to execution, the 36-bit B accumulator contains the value $8:AAAA:AAAA. Execution of the
ASR instruction shifts the 36-bit value in the B accumulator 1 bit to the right and stores the result back
in the B accumulator. C is cleared by the operation because bit 0 of A was cleared prior to the execution
of the instruction. The N bit of CCR (bit 3) is set because bit 35 of the result in A is set. The E bit of
CCR (bit 5) is set because the extension portion of B is used by the result.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if data limiting has occurred during parallel move
Set if the extension portion of result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Always cleared
Set if bit 0 of source operand was set prior to the execution of the instruction
Condition code results depend on the size of the destination operand.
Freescale Semiconductor
Instruction Set Details
A-45
ASR
ASR
Arithmetic Shift Right
Instruction Fields:
Operation
Operands
C
W
ASR
FFF
1
1
Comments
Arithmetic shift right entire register by 1 bit.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
ASR2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
A-46
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ASR
ASR
Arithmetic Shift Right
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
ASR2
F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
ASR
F GGG,X:<ea_m>
0
0
0
15
ASR
F X:<ea_m>,GGG
0
0
1
15
ASR
F
X:<ea_m>,reg1
X:<ea_v>,reg2
0
1
1
15
ASR
FFF
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
1
1
12
11
1
1
12
11
1
1
12
11
1
0
12
11
1
0
G
G
1
0
G
G
v
F
Instruction Set Details
8
7
G
F
8
7
G
F
8
7
v
F
8
7
F
F
0
0
v
1
1
1
1
1
4
3
1
0
4
3
1
0
4
3
0
0
4
3
0
1
0
m
R
R
0
m
R
R
0
m
0
v
0
0
1
1
A-47
ASR16
ASR16
Arithmetic Shift Right 16 Bits
Operation:
Assembler Syntax:
S >> 16 →
D
(no parallel move)
ASR16
S,D
(no parallel move)
Description: Arithmetically shift the source operand to the right by 16 bits, and store the result in the destination
(D), sign extending to the left. This operation effectively places the MSP of the source register into the
LSP of the destination register, propagating the sign bit through the MSP (and the extension register
for accumulator destinations). If the source is an accumulator, both the extension register and MSP are
shifted. When the destination operand is a 16-bit register, the sign information is written to it. For example, if the source is an accumulator, the 4 bits of the EXT are written to the lower 4 bits of the destination register with sign extension. If the source is a 16-bit register or the Y register, the msb (sign
bit) is written with sign extension. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction can be used to cast an integer to a long value.
Example 1:
ASR16
Y,A
; shift long in Y right by 16 bits and place in A
Before Execution
After Execution
0
3456
3456
F
FFFF
A1A2
A2
A1
A0
A2
A1
A0
A1A2
A3A4
A1A2
A3A4
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the Y register contains the value to be shifted ($A1A2:A3A4). The contents of the
destination register are not important prior to execution because they have no effect on the calculated
value. The ASR16 instruction arithmetically shifts the value $A1A2:A3A4 by 16 bits to the right, sign
extends to a full 36 bits, and places the result in the destination register A.
Example 2:
ASR16
Y,X0
; shift sign bit in Y right by 16 bits and sign extend
After Execution
Before Execution
A3A2
A1A0
A3A2
A1A0
Y1
Y0
Y1
Y0
0000
FFFF
X0
X0
Explanation of Example:
Prior to execution, the Y register contains the value to be shifted ($A3A2:A1A0). The contents of the
destination register are not important prior to execution because they have no effect on the calculated
value. Since the destination is a 16-bit register, the ASR16 instruction arithmetically shifts the value
of the sign bit by 16 bits to the right with sign extension, and places the result in the destination register
X0.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-48
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ASR16
ASR16
Arithmetic Shift Right 16 Bits
Instruction Fields:
Operation
Operands
C
W
Comments
ASR16
FFF,FFF
1
1
Arithmetic shift right the first operand by 16 bits, placing
result in the destination operand.
FFF
1
1
An alternate syntax for the above instruction if the
source and the destination are the same.
Instruction Opcodes:
15
ASR16 FFF,FFF
Timing:
0
1
1
12
11
1
1
1
F
8
7
F
F
b
b
4
3
b
0
0
1
1
0
1 oscillator clock cycle
1 program word
Memory:
1 program word
Freescale Semiconductor
Instruction Set Details
A-49
ASRA
ASRA
1-Bit Arithmetic Shift Right AGU Register
Operation:
Assembler Syntax:
D >> 1 → D
(no parallel move)
ASRA
D
(no parallel move)
Description: Arithmetically shift the address register operand 1 bit to the right, and store the result back in the register.
Example:
ASRA
R0
; arithmetically shift R0 to the right 1 bit
Before Execution
R0
After Execution
R0
80B360
C059B0
Explanation of Example:
Prior to execution, the R0 register contains $80B360. Execution of the ASRA instruction shifts the value in the R0 register 1 bit to the right and stores the result ($C059B0) back in R0.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
ASRA
Rn
1
1
Comments
Arithmetic shift right AGU register by 1 bit
Instruction Opcodes:
15
ASRA
Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-50
0
0
12
11
0
0
1
0
8
7
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
1
R
0
0
R
R
Freescale Semiconductor
ASRAC
Arithmetic Right Shift with Accumulate
Operation:
Assembler Syntax:
(S1 >> S2) + D → D (no parallel move)
ASRAC
S1,S2,D
ASRAC
(no parallel move)
Description: Arithmetically shift the first 16-bit source operand (S1) to the right by the value contained in the lowest
4 bits of the second source operand (S2), and accumulate the result with the value in the destination
(D). Operand S1 is internally sign extended and concatenated with 16 zero bits to form a 36-bit value
before the shift operation. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is typically used for multi-precision arithmetic right shifts.
Example:
ASRAC
Y1,X0,A
; arithmetic right shift Y1 by 4 and
; accumulate in A
Before Execution
After Execution
0
0000
0099
F
FC00
3099
A2
A1
A0
A2
A1
A0
C003
8000
C003
8000
Y1
Y0
Y1
Y0
X0
00F4
X0
00F4
SR
0300
SR
0308
Explanation of Example:
Prior to execution, the Y1 register contains the value that is to be shifted ($C003), the X0 register contains the amount by which to shift ($4), and the destination accumulator contains $0:0000:0099. The
ASRAC instruction arithmetically shifts the value $C003 by 4 bits to the right and accumulates this
result with the value that is already in the destination register A.
Condition Codes Affected:
MR
N
Z
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 35 of accumulator result is set
— Set if accumulator result equals zero
If the SA bit is set, the N bit is equal to bit 31 of the result.
If the SA bit is clear, the N bit is equal to bit 35 of the result.
Freescale Semiconductor
Instruction Set Details
A-51
ASRAC
ASRAC
Arithmetic Right Shift with Accumulate
Instruction Fields:
Operation
Operands
C
W
ASRAC
Y1,X0,FF
Y0,X0,FF
Y1,Y0,FF
Y0,Y0,FF
A1,Y0,FF
B1,Y1,FF
C1,Y0,FF
C1,Y1,FF
1
1
Comments
Arithmetic word shift with accumulation
Instruction Opcodes:
15
ASRAC Q1,Q2,FF
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-52
0
1
1
12
11
1
0
0
0
8
7
F
F
Q
DSP56800E and DSP56800EX Core Reference Manual
Q
4
3
Q
0
0
1
1
0
Freescale Semiconductor
ASRR.L
ASRR.L
Multi-Bit Arithmetic Right Shift Long
Operation:
Assembler Syntax:
If S[15] = 0 or S is not a register,
D >> S → D
(no parallel move)
Else
D << –S → D
(no parallel move)
ASRR.L
S,D
(no parallel move)
ASRR.L
S,D
(no parallel move)
Description: Arithmetically shift the second operand to the right by the value contained in the 5 lowest bits of the
first operand (or by an immediate integer), and store the result back in the destination (D). The shift
count can be a 5-bit positive immediate integer or the value contained in X0, Y0, Y1, or the MSP of
an accumulator. For 36- and 32-bit destinations, the MSP:LSP are shifted, with sign extension from bit
31 (the FF2 portion is ignored). If the shift count in a register is negative (bit 15 is set), the direction
of the shift is reversed, maintaining sign integrity. The result is not affected by the state of the saturation bit (SA).
Example:
ASRR.L Y0,A
; shift A right by the amount in Y0 and
; store result in A
After Execution
Before Execution
0
0123
4567
0
1234
5670
A2
A1
A0
A2
A1
A0
2000
FFFC
2000
FFFC
Y1
Y0
Y1
Y0
SR
SR
0300
0300
Explanation of Example:
Prior to execution, the A accumulator contains the value that is to be shifted ($0123:4567), and the Y0
register contains the amount by which to shift ($FFFC). Since the count is a negative number, the shift
is reversed—that is, the value will be shifted left. The ASRR.L instruction arithmetically shifts the value $0123:4567 by 4 bits to the left and places the result in the destination register A.
Condition Codes Affected:
MR
N
Z
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if the MSB of the result is set
— Set if the result equals zero
Condition code results depend on the size of the destination operand.
Freescale Semiconductor
Instruction Set Details
A-53
ASRR.L
ASRR.L
Multi-Bit Arithmetic Right Shift Long
Instruction Fields:
Operation
Operands
C
W
Comments
ASRR.L
#<0–31>,fff
2
1
Arithmetic shift right by a 5-bit positive immediate integer
EEE,FFF
2
1
Bi-directional arithmetic shift of destination by value in the
first operand: positive –> right shift
Instruction Opcodes:
15
ASRR.L #<0–31>,fff
0
1
0
1
1
15
ASRR.L EEE,FFF
0
Timing:
2 oscillator clock cycles
Memory:
1 program word
A-54
12
11
0
1
12
11
1
1
8
7
f
1
f
f
8
7
1
F
F
F
4
3
B
1
1
B
4
3
a
a
a
1
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
1
0
0
0
Freescale Semiconductor
ASRR.W
Multi-Bit Arithmetic Right Shift Word
Operation:
ASRR.W
Assembler Syntax:
S1 >> S2 → D
D >> S → D
(no parallel move)
(no parallel move)
ASRR.W
ASRR.W
S1,S2,D
S,D
(no parallel move)
(no parallel move)
Description: This instruction can have two or three operands. Arithmetically shift either the source operand S1 or
D to the right by the value contained in the lowest 4 bits of either S2 or S, respectively (or by an immediate integer), and store the result in the destination (D). The shift count can be a 4-bit positive integer, a value in a 16-bit register, or the MSP of an accumulator. For 36- and 32-bit destinations, only
the MSP is shifted and the LSP is cleared, with sign extension from bit 31 (the FF2 portion is ignored).
The result is not affected by the state of the saturation bit (SA).
Example 1:
ASRR.W
Y1,Y0,A
; arithmetic right shift of 16-bit Y1 by
; least 4 bits of Y0
After Execution
Before Execution
0
1234
5678
F
D555
0000
A2
A1
A0
A2
A1
A0
AAAA
FFF1
AAAA
FFF1
Y1
Y0
Y1
Y0
SR
SR
0300
0308
Explanation of Example:
Prior to execution, the Y1 register contains the value that is to be shifted ($AAAA), and the Y0 register
contains the number by which to shift (least 4 bits of $FFF1 = 1). The contents of the destination register are not important prior to execution because they have no effect on the calculated value. The
ASRR.W instruction arithmetically shifts the value $AAAA by 1 bit to the right and places the result
in the destination register A with sign extension (the LSP is cleared).
Example 2:
ASRR.W
Y1,A
; arithmetic right shift of 16-bit A1 by
; least 4 bits of Y1
After Execution
Before Execution
0
AAAA
4567
F
D555
0000
A2
A1
A0
A2
A1
A0
0001
000F
0001
000F
Y1
Y0
Y1
Y0
SR
0300
SR
0308
Explanation of Example:
Prior to execution, A1 contains the value that is to be shifted ($AAAA), and the Y1 register contains
the amount by which to shift ($1). The ASRR.W instruction arithmetically shifts the sign extended value $AAAA by 1 bit to the right and places the result in the destination register A (the LSP is cleared).
Freescale Semiconductor
Instruction Set Details
A-55
ASRR.W
ASRR.W
Multi-Bit Arithmetic Right Shift Word
Condition Codes Affected:
MR
N
Z
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if MSB of result is set
— Set if result equals zero
Instruction Fields:
Operation
Operands
C
W
Comments
ASRR.W
#<0–15>,FFF
1
1
Arithmetic shift right by a 4-bit positive immediate integer
EEE,FFF
1
1
Arithmetic shift right the destination by value specified in
4 LSBs of the first operand
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
1
1
Arithmetic shift right of the first operand by value specified in 4 LSBs of the second operand; place result in FFF
Instruction Opcodes:
15
ASRR.W #<0–15>,FFF
0
1
0
1
1
1
1
15
ASRR.W EEE,FFF
0
15
ASRR.W Q1,Q2,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-56
0
12
11
1
1
12
11
1
1
12
11
1
0
8
7
1
F
F
F
8
7
1
F
F
F
8
7
0
F
F
F
4
3
1
1
0
B
4
3
a
a
a
1
4
3
Q
Q
Q
0
DSP56800E and DSP56800EX Core Reference Manual
0
B
B
B
0
0
0
0
1
0
0
0
Freescale Semiconductor
Bcc
Bcc
Branch Conditionally
Operation:
Assembler Syntax:
If (cc), then PC + <OFFSET> → PC
else PC + 1 → PC
Bcc
Bcc
Bcc
<OFFSET7>
<OFFSET18>
<OFFSET22>
Description: If the specified condition is true, program execution continues at the location PC + displacement. The
PC contains the address of the next instruction. If the specified condition is false, the PC is incremented, and program execution continues sequentially. The offset can be 7, 18, or 22 bits; 7- and 18-bit
offsets are sign extended to 21 bits.
The term “cc” specifies the following:
“cc” Mnemonic
Condition
CC (HS*) — carry clear (higher or same)
C=0
CS (LO*) — carry set (lower)
C=1
EQ
— equal
Z=1
GE
— greater than or equal
N⊕V=0
GT
— greater than
Z + (N ⊕ V) = 0
HI*
— higher
C•Z=1
LE
— less than or equal
Z + (N ⊕ V) = 1
LS*
— lower or same
C+Z=1
LT
— less than
N⊕V=1
NE
— not equal
Z=0
NN
— not normalized
Z + (U • E) = 0
NR
— normalized
Z + (U • E) = 1
* Only available when CM bit set in the OMR
Xdenotes the logical complement of X
+denotes the logical OR operator
•denotes the logical AND operator
⊕denotes the logical exclusive OR operator
Example:
BNE
INC.W
INC.W
<LABEL
A
A
ADD
B,A
; branch to LABEL if Z condition is zero
LABEL
See Table 3-14 on page 3-27 for usage of forcing operator “<LABEL.”
Explanation of Example:
In this example, if the Z bit is zero when the BNE instruction is executed, program execution skips the
two INC.W instructions and continues with the ADD instruction. If the specified condition is not true,
no branch is taken, the program counter is incremented by one, and program execution continues with
the first INC.W instruction. The Bcc instruction uses a PC-relative offset of two for this example.
Freescale Semiconductor
Instruction Set Details
A-57
Bcc
Bcc
Branch Conditionally
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Condition Codes Affected:
The condition codes are tested but not modified by this instruction.
Instruction Fields:
Operation
Operands
C1
W
Bcc
<OFFSET7>
5 or 3
1
7-bit signed offset
<OFFSET18>
5 or 4
2
18-bit signed offset
<OFFSET22>
6 or 5
3
22-bit signed offset
Comments
1.The clock-cycle count depends on whether the branch is taken. The first value applies if the branch is taken,
and the second applies if it is not.
Instruction Opcodes:
15
Bcc
<OFFSET7>
1
0
1
15
Bcc
<OFFSET18>
1
1
1
12
11
0
C
12
11
0
0
C
C
C
C
8
7
C
0
8
7
C
0
A
1
a
1
4
3
a
a
4
3
0
1
4
3
0
a
a
a
0
C
A
A
AAAAAAAAAAAAAAAA
15
Bcc
<OFFSET22>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
C
C
C
0
1
1
0
1
C
0
0
AAAAAAAAAAAAAAAA
Timing:
3–6 oscillator clock cycles
Memory:
1–3 program word(s)
A-58
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFCHG
BFCHG
Test Bitfield and Change
Operation:
Assembler Syntax:
(<bitfield> of destination) → (<bitfield> of destination)BFCHG
BFCHG
#iiii,X:<ea> (no parallel move)
#iiii,D (no parallel move)
Description: Test all selected bits of the destination operand. If all selected bits are set, C is set; otherwise, C is
cleared. Then complement the selected bits, and store the result in the destination. A 16-bit immediate
value is used to specify which bits are tested and changed. Those bits that are set in the immediate value
are the same bits that are tested and changed in the destination; those bits that are cleared in the immediate value are ignored in the destination. This instruction performs a read-modify-write operation on
the destination memory location or register and requires two destination accesses.
Usage:
This instruction is very useful in performing I/O and flag bit manipulation.
Example:
BFCHG
#$0310,X:$5000
; test and change bits 4, 8, and 9
; in a data memory location
Before Execution
After Execution
X:$5000
0010
X:$5000
0300
SR
0301
SR
0300
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $0010. Execution of the
BFCHG instruction tests the state of bits 4, 8, and 9 in X:$5000, does not set C (because all of the selected bits were not set), and then complements the bits.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
For this destination only, the C bit is not updated as is done for all other destination operands.
All SR bits except bits 14–10 are updated with values from the bitfield unit.
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
Freescale Semiconductor
Instruction Set Details
A-59
BFCHG
BFCHG
Test Bitfield and Change
Instruction Fields:
Operation
Operands
C
W
Comments
BFCHG
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFCHG tests all targeted bits defined by the 16-bit immediate mask. If all targeted bits are set, then the C bit is set.
Otherwise it is cleared. Then the instruction inverts all
selected bits.
A-60
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
All registers in DDDDD are permitted except HWS and Y.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFCHG
BFCHG
Test Bitfield and Change
Instruction Opcodes:
15
BFCHG #<MASK16>,DDDDD
1
0
0
12
11
0
0
1
0
8
7
1
0
4
3
d
1
0
d
4
3
1
0
0
R
4
3
1
0
0
R
4
3
a
a
4
3
p
0
d
d
d
0
R
R
1
R
R
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:(Rn)
1
0
0
12
11
0
0
1
0
8
7
0
0
0
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:(Rn+xxxx)
1
0
0
12
11
0
0
1
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:(SP–xx)
1
0
1
12
11
0
0
1
0
8
7
0
1
1
a
0
a
a
a
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:<<pp
1
0
1
12
11
0
0
1
0
8
7
1
1
1
p
p
4
3
0
p
p
p
4
3
1
0
1
0
4
3
0
p
p
p
p
p
p
1
0
0
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:aa
1
0
1
12
11
0
0
1
0
8
7
1
1
0
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:xxxx
1
0
0
12
11
0
0
1
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
1
0
0
0
1
0
1
0
1
0
0
4
3
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCHG #<MASK16>,dd
1
0
0
12
11
0
0
1
0
8
7
0
0
1
0
0
0
d
d
iiiiiiiiiiiiiiii
Timing:
2–4 oscillator clock cycles
Memory:
2–4 program words
Freescale Semiconductor
Instruction Set Details
A-61
BFCLR
BFCLR
Test Bitfield and Clear
Operation:
Assembler Syntax:
0 →(<bitfield> of destination) (no parallel move)
BFCLR
BFCLR
#iiii,X:<ea>
#iiii,D
(no parallel move)
(no parallel move)
Description: Test all selected bits of the destination operand. If all selected bits are set, C is set; otherwise, C is
cleared. Then clear the selected bits, and store the result in the destination. A 16-bit immediate value
is used to specify which bits are tested and cleared. Those bits that are set in the immediate value are
the same bits that are tested and cleared in the destination; those bits that are cleared in the immediate
value are ignored in the destination. This instruction performs a read-modify-write operation on the
destination memory location or register and requires two destination accesses.
Usage:
This instruction is very useful in performing I/O and flag bit manipulation.
Example:
BFCLR
#$0310,X:$5000
; test and clear bits 4, 8, and 9 in
; an on-chip peripheral register
Before Execution
After Execution
X:$5000
7FF5
X:$5000
7CE5
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $7FF5. Execution of the
BFCLR instruction tests the state of bits 4, 8, and 9 in X:5000, sets the C bit (because all the selected
bits were set), and then clears the selected bits.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
For this destination only, the C bit is not updated as is done for all other destination operands.
All SR bits except bits 14–10 are updated with values from the bitfield unit.
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
A-62
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFCLR
BFCLR
Test Bitfield and Clear
Instruction Fields:
Operation
Operands
C
W
Comments
BFCLR
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFCLR tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared. Then the instruction
clears all selected bits.
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
Freescale Semiconductor
All registers in DDDDD are permitted except HWS and Y.
Instruction Set Details
A-63
BFCLR
BFCLR
Test Bitfield and Clear
15
BFCLR #<MASK16>,DDDDD
1
0
0
12
11
0
0
0
0
8
7
1
0
4
3
d
1
0
d
4
3
1
0
0
R
4
3
1
0
0
R
4
3
a
a
4
3
p
0
d
d
d
0
R
R
1
R
R
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:(Rn)
1
0
0
12
11
0
0
0
0
8
7
0
0
0
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:(Rn+xxxx)
1
0
0
12
11
0
0
0
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:(SP–xx)
1
0
1
12
11
0
0
0
0
8
7
0
1
1
a
0
a
a
a
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:<<pp
1
0
1
12
11
0
0
0
0
8
7
1
1
1
p
p
4
3
0
p
p
p
4
3
1
0
1
0
4
3
0
p
p
p
p
p
p
1
0
0
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:aa
1
0
1
12
11
0
0
0
0
8
7
1
1
0
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:xxxx
1
0
0
12
11
0
0
0
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
0
0
0
0
1
0
1
0
1
0
0
0
d
d
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFCLR #<MASK16>,dd
1
0
0
12
11
0
0
0
0
8
7
0
0
1
0
4
3
1
0
0
iiiiiiiiiiiiiiii
Timing:
2–4 oscillator clock cycles
Memory:
2–4 program words
A-64
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFSET
BFSET
Test Bitfield and Set
Operation:
Assembler Syntax:
1 → (<bitfield> of destination) (no parallel move)
BFSET
BFSET
#iiii,X:<ea>
#iiii,D
(no parallel move)
(no parallel move)
Description: Test all selected bits of the destination operand. If all selected bits are set, C is set; otherwise, C is
cleared. Then set the selected bits, and store the result in the destination memory. A 16-bit immediate
value is used to specify which bits are tested and set. Those bits that are set in the immediate value are
the same bits that are tested and set in the destination; those bits that are cleared in the immediate value
are ignored in the destination. This instruction performs a read-modify-write operation on the destination memory location or register and requires two destination accesses.
Usage:
This instruction is very useful in performing I/O and flag bit manipulation.
Example:
BFSET
#$CC00,X:$5000
; set bits in peripheral register
Before Execution
After Execution
X:$5000
3300
X:$5000
FF00
SR
0301
SR
0300
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $3300. Execution of the
instruction tests the state of bits 10, 11, 14, and 15 in X:$5000, clears the C bit (because none of the
selected bits was set), and then sets the selected bits.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
For this destination only, the C bit is not updated as is done for all other destination operands.
All SR bits except bits 14–10 are updated with values from the bitfield unit.
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
Freescale Semiconductor
Instruction Set Details
A-65
BFSET
BFSET
Test Bitfield and Set
Instruction Fields:
Operation
Operands
C
W
Comments
BFSET
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFSET tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared. Then the instruction sets
all selected bits.
A-66
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
All registers in DDDDD are permitted except HWS and Y.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFSET
BFSET
Test Bitfield and Set
Instruction Opcodes:
15
BFSET #<MASK16>,DDDDD
1
0
0
12
11
0
0
0
1
8
7
1
0
4
3
d
1
0
d
4
3
1
0
0
R
4
3
1
0
0
R
4
3
a
a
4
3
p
0
d
d
d
0
R
R
1
R
R
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:(Rn)
1
0
0
12
11
0
0
0
1
8
7
0
0
0
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:(Rn+xxxx)
1
0
0
12
11
0
0
0
1
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:(SP–xx)
1
0
1
12
11
0
0
0
1
8
7
0
1
1
a
0
a
a
a
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:<<pp
1
0
1
12
11
0
0
0
1
8
7
1
1
1
p
p
4
3
0
p
p
p
4
3
1
0
1
0
4
3
0
p
p
p
p
p
p
1
0
0
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:aa
1
0
1
12
11
0
0
0
1
8
7
1
1
0
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:xxxx
1
0
0
12
11
0
0
0
1
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
0
1
0
0
1
0
1
0
1
0
0
4
3
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFSET #<MASK16>,dd
1
0
0
12
11
0
0
0
1
8
7
0
0
1
0
0
0
d
d
iiiiiiiiiiiiiiii
Timing:
2–4 oscillator clock cycles
Memory:
2–4 program words
Freescale Semiconductor
Instruction Set Details
A-67
BFTSTH
BFTSTH
Test Bitfield High
Operation:
Assembler Syntax:
Test <bitfield> of destination for ones(no parallel move)
BFTSTH#iiii,X:<ea>(no parallel move)
BFTSTH#iiii,D (no parallel move)
Description: Test all selected bits of the destination operand. If all selected bits are set, C is set; otherwise, C is
cleared. A 16-bit immediate value is used to specify which bits are tested. Those bits that are set in the
immediate value are the same bits that are tested in the destination; those bits that are cleared in the
immediate value are ignored in the destination. This instruction performs two destination accesses.
Usage:
This instruction is very useful for testing I/O and flag bits.
Example:
BFTSTH
#$0310,X:5000
; test high bits 4, 8, and 9 in
; an on-chip peripheral register
Before Execution
After Execution
X:$5000
0FF0
X:$5000
0FF0
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$FFE2 contains the value $0FF0. Execution of the
instruction tests the state of bits 4, 8, and 9 in X:$FFE2 and sets the C bit (because all the selected bits
were set).
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
A-68
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFTSTH
BFTSTH
Test Bitfield High
Instruction Fields:
Operation
Operands
C
W
Comments
BFTSTH
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFTSTH tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are set, then the C
bit is set. Otherwise it is cleared.
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
Freescale Semiconductor
All registers in DDDDD are permitted except HWS and Y.
Instruction Set Details
A-69
BFTSTH
BFTSTH
Test Bitfield High
Instruction Opcodes:
15
BFTSTH #<MASK16>DDDDD
1
0
0
12
11
0
1
1
0
8
7
1
0
4
3
d
1
0
d
4
3
1
0
0
R
4
3
1
0
0
R
4
3
a
a
4
3
p
p
4
3
p
p
4
3
1
0
0
d
d
d
0
R
R
1
R
R
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:(Rn)
1
0
0
12
11
0
1
1
0
8
7
0
0
0
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:(Rn+xxxx)
1
0
0
12
11
0
1
1
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:(SP–xx)
1
0
1
12
11
0
1
1
0
8
7
0
1
1
a
0
a
a
a
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:<<pp
1
0
1
12
11
0
1
1
0
8
7
1
1
1
p
0
p
p
p
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:aa
1
0
1
12
11
0
1
1
0
8
7
1
1
0
p
0
p
p
p
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:xxxx
1
0
0
12
11
0
1
1
0
8
7
0
0
1
0
0
1
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
1
1
0
0
0
1
0
1
0
1
0
0
0
d
d
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTH #<MASK16>,dd
1
0
0
12
11
0
1
1
0
8
7
0
0
1
0
4
3
1
0
0
iiiiiiiiiiiiiiii
Timing:
2–4 oscillator clock cycles
Memory:
2–4 program words
A-70
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFTSTL
BFTSTL
Test Bitfield Low
Operation:
Assembler Syntax:
Test <bitfield> of destination for zeros (no parallel move)
BFTSTL#iiii,X:<ea>(no parallel move)
BFTSTL#iiii,D (no parallel move)
Description: Test all selected bits in the destination operand. If all selected bits are clear, C is set; otherwise, C is
cleared. A 16-bit immediate value is used to specify which bits are tested. Those bits that are set in the
immediate value are the same bits that are tested in the destination; those bits that are cleared in the
immediate value are ignored in the destination. This instruction performs two destination accesses.
Usage:
This instruction is very useful for testing I/O and flag bits.
Example:
BFTSTL #$0310,X:$5000
; test low bits 4, 8, and 9 in
; an on-chip peripheral register
Before Execution
After Execution
X:$5000
0CC0
X:$5000
0CC0
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $0CC0. Execution of the
instruction tests the state of bits 4, 8, and 9 in X:$5000 and sets the C bit (because all the selected bits
were cleared).
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
Freescale Semiconductor
Instruction Set Details
A-71
BFTSTL
BFTSTL
Test Bitfield Low
Instruction Fields:
Operation
Operands
C
W
Comments
BFTSTL
#<MASK16>,DDDDD
2
2
#<MASK16>,dd
2
2
BFTSTL tests all the targeted bits defined by the 16-bit
immediate mask. If all the targeted bits are clear, then the
C bit is set. Otherwise it is cleared.
A-72
#<MASK16>,X:(Rn)
2
2
#<MASK16>,X:(Rn+xxxx)
3
3
#<MASK16>,X:(SP–xx)
3
2
#<MASK16>,X:aa
2
2
#<MASK16>,X:<<pp
2
2
#<MASK16>,X:xxxx
3
3
#<MASK16>,X:xxxxxx
4
4
All registers in DDDDD are permitted except HWS and Y.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BFTSTL
BFTSTL
Test Bitfield Low
Instruction Opcodes:
15
BFTSTL #<MASK16>DDDDD
1
0
0
12
11
0
1
0
0
8
7
1
0
4
3
d
1
0
d
4
3
1
0
0
R
4
3
1
0
0
R
4
3
a
a
4
3
p
p
4
3
p
0
d
d
d
0
R
R
1
R
R
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:(Rn)
1
0
0
12
11
0
1
0
0
8
7
0
0
0
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:(Rn+xxxx)
1
0
0
12
11
0
1
0
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:(SP–xx)
1
0
1
12
11
0
1
0
0
8
7
0
1
1
a
0
a
a
a
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:<<pp
1
0
1
12
11
0
1
0
0
8
7
1
1
1
p
0
p
p
p
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:aa
1
0
1
12
11
0
1
0
0
8
7
1
1
0
p
p
4
3
1
0
1
0
0
p
p
p
1
0
0
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:xxxx
1
0
0
12
11
0
1
0
0
8
7
0
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
1
0
0
0
0
1
0
1
0
1
0
0
0
d
d
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
BFTSTL #<MASK16>,dd
1
0
0
12
11
0
1
0
0
8
7
0
0
1
0
4
3
1
0
0
iiiiiiiiiiiiiiii
Timing:
2–4 oscillator clock cycles
Memory:
2–4 program words
Freescale Semiconductor
Instruction Set Details
A-73
BRA
BRA
Branch
Operation:
Assembler Syntax:
PC + <OFFSET> → PC
BRA
BRA
BRA
<OFFSET7>
<OFFSET18>
<OFFSET22>
Description: Branch to the location in program memory at PC + displacement. The PC contains the address of the
next instruction. The displacement is a 7-bit, 18-bit, or 22-bit signed value that is sign extended to form
the PC-relative offset.
Example:
BRA
INC.W
INC.W
LABEL
A
A
; jump to instruction at “LABEL”
; these two instructions are skipped
ADD
B,A
; execution resumes here
LABEL
Explanation of Example:
In this example, program execution skips the two INC.W instructions and continues with the ADD instruction. The BRA instruction uses a PC-relative offset of two for this example.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
A BRA instruction used within a DO loop cannot begin at the LA or LA – 1 within that DO loop.
A BRA instruction cannot be repeated using the REP instruction.
Instruction Fields:
Operation
Operands
C
W
Comments
BRA
<OFFSET7>
5
1
7-bit signed offset
<OFFSET18>
5
2
18-bit signed offset
<OFFSET22>
6
3
22-bit signed offset
Instruction Opcodes:
15
BRA
<OFFSET7>
1
0
1
15
BRA
<OFFSET18>
1
1
1
12
11
0
1
12
11
0
0
0
0
0
0
8
7
1
0
8
7
1
0
A
1
a
1
4
3
a
a
4
3
0
1
4
3
0
a
a
a
0
1
A
A
AAAAAAAAAAAAAAAA
15
BRA
<OFFSET22>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
0
0
1
0
1
1
0
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
5–6 oscillator clock cycles
Memory:
1–3 program word(s)
A-74
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BRAD
BRAD
Delayed Branch
Operation:
Assembler Syntax:
Execute instructions in next 2 words
PC + <OFFSET> → PC
BRAD
BRAD
BRAD
<OFFSET7>
<OFFSET18>
<OFFSET22>
Description: Branch to the location in program memory at PC + displacement, but first execute the instruction or
instructions in the following 2 program words. The PC contains the address of the next instruction. The
displacement is a 7-bit, 18-bit, or 22-bit signed value that is sign extended to form the PC-relative offset.
Example:
BRAD
INC.W
INC.W
...
LABEL
A
A
ADD
B,A
; delayed branch to “LABEL”
; these two increments are executed
; before the branch!
LABEL
Explanation of Example:
In this example, the program executes the two INC.W instructions that follow the BRAD instruction,
and then it continues with the ADD instruction that follows LABEL.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
A BRAD instruction used within a DO loop cannot begin at the LA or LA – 1 within that DO loop.
A BRAD instruction cannot be repeated using the REP instruction.
Refer to Section 4.3.2, “Delayed Instruction Restrictions,” on page 4-14.
Freescale Semiconductor
Instruction Set Details
A-75
BRAD
BRAD
Delayed Branch
Instruction Fields:
Operation
Operands
C
W
Comments
BRAD
<OFFSET7>
3
1
Delayed branch with 7-bit signed offset; must fill 2 delay slots
<OFFSET18>
3
2
Delayed branch with 18-bit signed offset; must fill 2 delay slots
<OFFSET22>
4
3
Delayed branch with 22-bit signed offset; must fill 2 delay slots
Instruction Opcodes:
15
BRAD
<OFFSET7>
BRAD
<OFFSET18>
1
0
1
1
1
15
1
12
11
0
1
12
11
0
0
8
7
0
0
1
1
8
7
0
1
1
0
4
3
a
A
a
a
4
3
1
1
0
1
4
3
0
a
a
a
1
A
A
0
AAAAAAAAAAAAAAAA
15
BRAD
<OFFSET22>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
0
1
1
0
1
1
0
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
1–3 program word(s)
A-76
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BRCLR
BRCLR
Branch if Bits Clear
Operation:
Assembler Syntax:
Branch if <bitfield> of destination is all zeros (no parallel move)
BRCLR #<MASK8>,X:<ea>,AA
BRCLR #<MASK8>,D,AA
Description: Test all selected bits of the destination operand. If all the selected bits are clear, C is set, and program
execution continues at the location in program memory at PC + displacement. Otherwise, C is cleared,
and execution continues with the next sequential instruction. A 16-bit immediate value is used to specify which bits are tested. Those bits that are set in the immediate value are the same bits that are tested
in the destination; those bits that are cleared in the immediate value are ignored in the destination.
Usage:
This instruction is useful in performing I/O flag polling.
Example:
BRCLR
#$0068,X:$5000,LABEL
INC.W
INC.W
A
A
ADD
B,A
; next two instructions
; are bypassed
LABEL
Before Execution
After Execution
X:$5000
FF00
X:$5000
FF00
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $FF00. Execution of the
BRCLR instruction tests the state of bits 3, 5, and 6 in X:$5000 and sets the C bit (because all the mask
bits were clear). Since C is set, program execution is then transferred to the address offset from the
current program counter by the displacement that is specified in the instruction.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
— Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
If all bits in the mask are cleared, C is set, and the branch is taken.
Freescale Semiconductor
Instruction Set Details
A-77
BRCLR
BRCLR
Branch if Bits Clear
Instruction Fields:
Operation
Operands
C1
W
Comments
BRCLR
#<MASK8>,DDDDD,AA
7/5
2
BRCLR tests all the targeted bits defined by
the immediate mask. If all the targeted bits
are clear, then the carry bit is set and a
PC-relative branch occurs. Otherwise it is
cleared and no branch occurs.
#<MASK8>,dd,AA
7/5
2
#<MASK8>,X:(Rn),AA
7/5
2
#<MASK8>,X:(Rn+xxxx),AA
8/6
3
#<MASK8>,X:(SP–xx),AA
8/6
2
#<MASK8>,X:aa,AA
7/5
2
#<MASK8>,X:<<pp,AA
7/5
2
#<MASK8>,X:xxxx,AA
7/5
3
#<MASK8>,X:xxxxxx,AA
8/6
4
All registers in DDDDD are permitted except
HWS and Y.
MASK8 specifies a 16-bit immediate value,
where either the upper or lower 8 bits contain
all zeros. AA specifies a 7-bit PC-relative offset.
1.The first cycle count refers to the case when the condition is true and the branch is taken. The
second cycle count refers to the case when the condition is false and the branch is not taken.
A-78
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BRCLR
BRCLR
Branch if Bits Clear
Instruction Opcodes:
15
BRCLR #<MASK8>,DDDDD,AA
1
0
0
12
11
0
1
0
1
8
7
1
0
1
4
3
0
d
d
4
3
0
0
R
4
3
0
0
R
4
3
a
a
4
3
p
p
p
4
3
p
p
p
4
3
1
0
0
d
d
d
0
R
R
1
R
R
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:(Rn),AA
1
0
0
12
11
0
1
0
1
8
7
0
0
1
0
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:(Rn+xxxx),AA
1
0
0
12
11
0
1
0
1
8
7
0
0
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:(SP–xx),AA
1
0
1
12
11
0
1
0
1
8
7
0
1
1
a
0
a
a
a
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:<<pp,AA
1
0
1
12
11
0
1
0
1
8
7
1
1
1
0
p
p
p
p
p
p
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:aa,AA
1
0
1
12
11
0
1
0
1
8
7
1
1
0
0
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:xxxx,AA
1
0
0
12
11
0
1
0
1
8
7
0
0
1
0
0
1
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,X:xxxxxx,AA
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
0
1
0
1
0
0
1
0
1
0
1
0
0
4
3
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRCLR #<MASK8>,dd,AA
1
0
0
12
11
0
1
0
1
8
7
0
0
1
0
0
0
d
d
iiiiiiiiUAaaaaaa
Timing:
5–8 oscillator clock cycles
Memory:
2–4 program words
Freescale Semiconductor
Instruction Set Details
A-79
BRSET
BRSET
Branch if Bits Set
Operation:
Assembler Syntax:
Branch if <bitfield> of destination is all ones (no parallel move)
BRSET #<MASK8>,X:<ea>,AA
BRSET #<MASK8>,D,AA
Description: Test all selected bits of the destination operand. If all the selected bits are set, C is set, and program
execution continues at the location in program memory at PC + displacement. Otherwise, C is cleared,
and execution continues with the next sequential instruction. A 16-bit immediate value is used to specify which bits are tested. Those bits that are set in the immediate value are the same bits that are tested
in the destination; those bits that are cleared in the immediate value are ignored in the destination.
Usage:
This instruction is useful in performing I/O flag polling.
Example:
BRSET
#$0500,X:$5000,LABEL
INC.W
INC.W
A
A
ADD
B,A
; next two instructions
; are bypassed
LABEL
Before Execution
After Execution
X:$5000
0FF0
X:$5000
0FF0
SR
0300
SR
0301
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $0FF0. Execution of the
BRSET instruction tests the state of bits 8 and 10 in X:$5000 and sets the C bit (because all the mask
bits were set). Since C is set, program execution is then transferred to the address offset from the current program counter by the displacement that is specified in the instruction.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
— Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
A-80
If all bits in the mask are cleared, C is set and the branch is taken.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BRSET
BRSET
Branch if Bits Set
Instruction Fields:
Operation
Operands
C1
W
Comments
BRSET
#<MASK8>,DDDDD,AA
7/5
2
BRSET tests all the targeted bits defined by the immediate mask. If all the targeted bits are set, then the carry
bit is set and a PC-relative branch occurs. Otherwise it
is cleared and no branch occurs.
#<MASK8>,dd,AA
7/5
2
#<MASK8>,X:(Rn),AA
7/5
2
#<MASK8>,X:(Rn+xxxx),AA
8/6
3
#<MASK8>,X:(SP–xx),AA
8/6
2
#<MASK8>,X:aa,AA
7/5
2
#<MASK8>,X:<<pp,AA
7/5
2
#<MASK8>,X:xxxx,AA
7/5
3
#<MASK8>,X:xxxxxx,AA
8/6
4
All registers in DDDDD are permitted except HWS and
Y.
MASK8 specifies a 16-bit immediate value, where
either the upper or lower 8 bits contain all zeros. AA
specifies a 7-bit PC-relative offset.
1.The first cycle count refers to the case when the condition is true and the branch is taken. The second cycle
count refers to the case when the condition is false and the branch is not taken.
Freescale Semiconductor
Instruction Set Details
A-81
BRSET
BRSET
Branch if Bits Set
Instruction Opcodes:
15
BRSET #<MASK8>,DDDDD,AA
1
0
0
12
11
0
1
1
1
8
7
1
0
1
4
3
0
d
d
4
3
0
0
R
4
3
0
0
R
4
3
a
a
4
3
p
p
p
4
3
p
p
p
4
3
0
1
0
4
3
0
d
d
d
0
R
R
1
R
R
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:(Rn),AA
1
0
0
12
11
0
1
1
1
8
7
0
0
1
0
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:(Rn+xxxx),AA
1
0
0
12
11
0
1
1
1
8
7
0
0
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:(SP–xx),AA
1
0
1
12
11
0
1
1
1
8
7
0
1
1
a
0
a
a
a
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:<<pp,AA
1
0
1
12
11
0
1
1
1
8
7
1
1
1
0
p
p
p
p
p
p
1
0
0
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:aa,AA
1
0
1
12
11
0
1
1
1
8
7
1
1
0
0
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:xxxx,AA
1
0
0
12
11
0
1
1
1
8
7
0
0
1
0
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,X:xxxxxx,AA
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
1
1
1
0
0
1
0
1
0
1
0
0
0
d
d
AAAAAAAAAAAAAAAA
iiiiiiiiUAaaaaaa
15
BRSET #<MASK8>,dd,AA
1
0
0
12
11
0
1
1
1
8
7
0
0
1
0
4
3
1
0
0
iiiiiiiiUAaaaaaa
Timing:
5–8 oscillator clock cycles
Memory:
2–4 program words
A-82
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
BSR
Branch to Subroutine
Operation:
Assembler Syntax:
SP + 1
→ SP
PC
→ X:(SP)
SP + 1
→ SP
SR
→ X:(SP)
PC + <OFFSET> → PC
BSR
BSR
<OFFSET18> or <OFFSET22>
Description: Place the PC and SR on the software stack and branch to the location in program memory at PC + displacement. The PC contains the address of the next instruction. The displacement is an 18-bit or 22-bit
signed value that is sign extended to form the PC-relative offset.
Example:
BSR
LABEL
; branch to PC-relative address “LABEL”
Explanation of Example:
In this example, program execution is transferred to the subroutine at the PC-relative address that is
represented by LABEL. The relative offset that is given by the label can be an 18- or 22-bit signed
value.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Instruction Fields:
Operation
Operands
C
W
Comments
BSR
<OFFSET18>
5
2
18-bit signed offset
<OFFSET22>
6
3
22-bit signed offset
Instruction Opcodes:
15
BSR
<OFFSET18>
1
1
1
12
11
0
0
0
1
8
7
0
0
1
1
4
3
0
1
4
3
0
1
A
A
AAAAAAAAAAAAAAAA
15
BSR
<OFFSET22>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
0
1
0
0
1
1
0
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
5–6 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-83
CLB
CLB
Count Leading Bits
Operation:
Assembler Syntax:
If S[MSB] = 0
(# of leading zeros – 1) in S → D
else
(# of leading ones – 1) in S → D
CLB
S,D
(no parallel move)
Description: Count the number of leading bits in the source operand, and place that number minus one in the destination. The bits to count are based on the high-order bit of the source operand: if the high-order bit is
zero, the number of zeros in the source operand (minus one) is placed in the destination. If the source
register is an accumulator, the extension portion is ignored, and only the bits in the FF10 portion are
counted. The result is not affected by the state of the saturation bit (SA). This instruction is used in
conjunction with the ASLL.L instruction to normalize a number.
Example:
CLB
A,X0
; count leading bits in A, placing
; result minus one in X0
Before Execution
After Execution
0
D7B2
4836
0
D7B2
4836
A2
A1
A0
A2
A1
A0
X0
7FFF
X0
0001
SR
030F
SR
0301
Explanation of Example:
The A register initially contains the value $F:D7B2:4836, and the X0 register contains $AAAA. After
the CLB A,X0 instruction is executed, the value $0001 is placed in X0, since there are two leading
ones in the value contained in A10. In order to normalize A, this instruction may be followed by the
operation ASLL.L X0,A (the resulting normalized number would be $F:AF64:906C).
Condition Codes Affected:
MR
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if the high-order bit of the result is set
— Set if the result is zero
— Always cleared
Instruction Fields:
A-84
Operation
Operands
C
W
CLB
FFF,EEE
1
1
Comments
Count leading bits (minus one); designed to operate
with the ASLL and ASRR instructions
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CLB
CLB
Count Leading Bits
Instruction Opcodes:
15
CLB
FFF,EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
1
0
E
Instruction Set Details
8
7
E
E
b
b
4
3
b
1
0
0
1
1
A-85
CLR
CLR
Clear Accumulator
Operation:
0→D
0→D
0→D
Assembler Syntax:
(no parallel move)
(one parallel move)
(two parallel reads)
CLR
CLR
CLR
D
D
D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Set the A or B accumulator to zero. Data limiting may occur during a parallel write.
Example:
CLR
A
A,X:(R0)+; save A into memory before clearing it
After Execution
Before Execution
2
3456
789A
0
0000
0000
A2
A1
A0
A2
A1
A0
SR
SR
032F
03D5
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $2:3456:789A. Execution of the
CLR A instruction sets the A accumulator to zero, and the saturation value $7FFF is written to memory.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
*
*
*
*
*
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if data limiting has occurred during parallel move
Always cleared
Always set
Always cleared
Always set
Always cleared
This instruction operates only on the A and B accumulator registers. The CLR.W instruction should
be used to clear any of the other registers (including A and B if desired).
Instruction Fields:
A-86
Operation
Operands
C
W
CLR
F
1
1
Comments
Clear 36-bit accumulator and set condition codes.
Also see CLR.W.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CLR
CLR
Clear Accumulator
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
CLR2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-87
CLR
CLR
Clear Accumulator
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
CLR2
F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
CLR
F
0
CLR
F GGG,X:<ea_m>
CLR
F X:<ea_m>,GGG
CLR
F X:<ea_m>,reg1
X:<ea_v>,reg2
1
1
0
0
0
1
1
1
15
0
15
0
15
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-88
0
12
11
1
0
12
11
0
1
12
11
0
1
12
11
1
0
8
7
F
1
1
1
8
7
G
G
G
F
8
7
G
G
G
F
8
7
1
v
v
F
4
3
0
1
0
0
4
3
0
1
1
0
4
3
0
1
1
0
4
3
v
0
1
0
DSP56800E and DSP56800EX Core Reference Manual
0
1
1
1
m
R
R
m
R
R
m
0
v
0
0
0
Freescale Semiconductor
CLR.B
Operation:
0→D
CLR.B
Clear Byte (Word Pointer)
Assembler Syntax:
(no parallel move)
CLR.B
D
(no parallel move)
Description: Set a byte in memory to zero. Addresses are expressed as word pointers.
Example:
CLR.B
X:(SP-1)
; clear a byte in the stack
Before Execution
After Execution
X:$4443
3333
X:$4443
3333
X:$4442
2222
X:$4442
0022
SP
SP
004443
004443
Explanation of Example:
The contents of the upper byte from stack address $004442 are cleared.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
CLR.B
X:(SP)
1
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
Comments
Clear a byte in memory using appropriate addressing
mode
Instruction Opcodes:
15
CLR.B
X:(Rn+xxxx)
1
1
0
12
11
1
1
0
0
8
7
1
1
1
1
4
3
0
R
4
3
0
0
R
R
AAAAAAAAAAAAAAAA
15
CLR.B
X:(Rn+xxxxxx)
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
1
0
0
1
1
1
1
0
R
0
R
R
4
3
1
1
AAAAAAAAAAAAAAAA
15
CLR.B
X:(SP)
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
1
1
0
12
11
1
1
0
0
Instruction Set Details
8
7
1
1
0
1
0
1
1
1
A-89
CLR.BP
Operation:
0→D
CLR.BP
Clear Byte (Byte Pointer)
Assembler Syntax:
(no parallel move)
CLR.BP
D
(no parallel move)
Description: Set a byte in memory to zero. An absolute address is expressed as a byte address.
Example:
CLR.BP X:$3065
; set byte at (byte) address $3065 to zero
Before Execution
Byte
Addresses
$3068
$3066
$3064
$3062
After Execution
X Memory
7
0
07
88
66
44
77
55
33
22
11
Byte
Addresses
7
$3068
$3066
$3064
$3062
X Memory
0
07
88
66
00
77
55
33
22
11
Explanation of Example:
The byte value in X memory at byte address $3065 is cleared. Note that this address is equivalent to
the upper byte of word address $1832.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-90
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CLR.BP
CLR.BP
Clear Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
CLR.BP
X:(RRR)
1
1
X:(RRR)+
1
1
X:(RRR)–
1
1
X:(RRR+N)
2
1
X:(RRR+xxxx)
2
2
X:(RRR+xxxxxx)
3
3
X:xxxx
2
2
X:xxxxxx
3
3
Comments
Clear a byte in memory
Instruction Opcodes:
15
CLR.BP X:(RRR+xxxx)
1
1
0
12
11
1
1
0
0
8
7
1
1
1
1
4
3
0
N
0
1
N
N
AAAAAAAAAAAAAAAA
15
CLR.BP X:(RRR+xxxxxx)
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
0
0
1
1
1
1
0
N
1
N
N
M
N
N
1
0
1
AAAAAAAAAAAAAAAA
15
CLR.BP X:<ea_MM>
1
1
0
1
0
15
CLR.BP X:xxxx
1
12
11
1
1
12
11
1
1
8
7
0
0
1
1
8
7
0
0
1
1
4
3
0
1
M
N
4
3
1
1
1
1
0
0
AAAAAAAAAAAAAAAA
15
CLR.BP X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
0
0
1
1
1
1
1
1
1
0
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-91
CLR.L
Operation:
0→D
CLR.L
Clear Long
Assembler Syntax:
(no parallel move)
CLR.L
D
(no parallel move)
Description: Set a long word in memory to zero. The destination address of the long word that is to be cleared must
be an even word pointer value, and it indicates the address of the lower half of the long word.
Example:
CLR.L
X:$3000
; set long word at address $3000 to zero
Before Execution
After Execution
Word
Addresses
Word
Addresses
X Memory
15
$3002
$3001
$3000
$2FFF
0
4444
3333
2222
X Memory
15
$3002
$3001
$3000
$2FFF
1111
0
4444
0000
0000
1111
Explanation of Example:
The long-word value in X memory at the address $3000 is cleared.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
A-92
Operation
Operands
C
W
CLR.L
X:(Rn)
1
1
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:xxxx
2
2
X:xxxxxx
3
3
Comments
Clear a long in memory
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CLR.L
CLR.L
Clear Long
Instruction Opcodes:
15
CLR.L X:(Rn+xxxx)
1
1
0
12
11
1
1
1
1
8
7
1
0
1
1
4
3
0
R
0
0
R
R
AAAAAAAAAAAAAAAA
15
CLR.L X:(Rn+xxxxxx)
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
1
1
1
0
1
1
0
R
0
R
R
M
R
R
1
0
1
AAAAAAAAAAAAAAAA
15
CLR.L X:<ea_MM>
1
1
0
1
0
15
CLR.L X:xxxx
1
12
11
1
1
12
11
1
1
8
7
1
1
1
0
8
7
1
1
1
0
4
3
0
1
M
R
4
3
1
1
1
1
0
0
AAAAAAAAAAAAAAAA
15
CLR.L X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
1
1
1
0
1
1
1
1
1
0
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-93
CLR.W
Operation:
0→D
CLR.W
Clear Word
Assembler Syntax:
(no parallel move)
CLR.W
D
(no parallel move)
Description: Set a word in memory or in an ALU register to zero. If an accumulator register or an AGU address
register is specified, the entire register is cleared.
Example:
CLR.W
X:$3000
; set word at (word) address $3000 to zero
Before Execution
After Execution
Word
Addresses
Word
Addresses
X Memory
15
$3002
$3001
$3000
$2FFF
0
4444
3333
2222
X Memory
15
$3002
$3001
$3000
$2FFF
1111
0
4444
3333
0000
1111
Explanation of Example:
The word value in X memory at the address $3000 is cleared.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Note:
This instruction should be used instead of the CLR instruction for clearing accumulator registers in all
new programs.
Instruction Fields:
A-94
Operation
Operands
C
W
Comments
CLR.W
DDDDD
1
1
Clear a register. The instruction clears an entire accumulator when FF is specified, and it clears an entire
AGU register when Rn is specified.
X:(Rn)
1
1
Clear a word in memory.
X:(Rn)+
1
1
X:(Rn)–
1
1
X:(Rn+N)
2
1
X:(Rn)+N
1
1
X:(Rn+xxxx)
2
2
X:(Rn+xxxxxx)
3
3
X:aa
1
1
X:<<pp
1
1
X:xxxx
2
2
X:xxxxxx
3
3
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CLR.W
CLR.W
Clear Word
Instruction Opcodes:
15
CLR.W
DDDDD
1
0
0
15
CLR.W X:(Rn)+N
1
1
0
15
CLR.W X:(Rn+xxxx)
1
1
0
12
11
0
D
12
11
1
1
12
11
1
1
D
1
D
1
1
1
8
7
D
D
8
7
1
0
8
7
1
0
0
1
1
0
0
0
4
3
0
1
4
3
1
R
4
3
0
R
4
3
0
1
1
1
0
1
R
R
0
0
R
R
AAAAAAAAAAAAAAAA
15
CLR.W X:(Rn+xxxxxx)
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
1
1
1
1
0
1
0
0
R
0
R
R
M
R
R
AAAAAAAAAAAAAAAA
15
CLR.W X:<ea_MM>
1
1
0
15
CLR.W X:<<pp
1
1
0
15
CLR.W X:aa
1
1
0
15
CLR.W X:xxxx
1
1
0
12
11
1
1
12
11
0
1
12
11
0
1
12
11
1
1
1
1
1
1
1
1
1
1
8
7
1
0
8
7
1
0
8
7
1
0
8
7
1
0
0
1
0
1
0
p
p
1
4
3
M
R
4
3
p
p
4
3
p
p
4
3
1
1
4
3
0
0
p
p
p
0
p
p
p
0
1
0
0
AAAAAAAAAAAAAAAA
15
CLR.W X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
1
1
1
1
0
1
1
1
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-95
CMP
Operation:
D–S
D–S
CMP
Compare
Assembler Syntax:
(one parallel move)
(no parallel move)
CMP
CMP
S,D
S,D
(one parallel move)
(no parallel move)
Description: Subtract the first operand from the second operand and update the CCR without storing the result. If
the second operand is a 36-bit accumulator, 16-bit source registers are first sign extended internally
and concatenated with 16 zero bits to form a 36-bit operand. When the second operand is X0, Y0, or
Y1, 16-bit subtraction is performed. In this case, if the first operand is one of the four accumulators;
the FF1 portion (properly sign extended) is used in the 16-bit subtraction (the FF2 and FF0 portions
are ignored).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Note:
In order for the carry bit (C) to be set correctly as a result of the subtraction, the operands must be properly sign extended. The destination can be improperly sign extended by writing the FF1 portion explicitly prior to executing the compare, so that FF2 might not represent the correct sign extension. This
note particularly applies to the case in which the source is extended to compare 16-bit operands, such
as X0 with A1.
Example:
CMP
Y0,A
X0,X:(R1)+N
; compare Y0 and A, save X0, update R1
Before Execution
After Execution
0
0020
0000
0
0020
0000
A2
A1
A0
A2
A1
A0
2000
0024
2000
0024
Y1
Y0
Y1
Y0
SR
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000, and the 16-bit Y0 register contains the value $0024. Execution of the CMP Y0,A instruction automatically appends the
16-bit value in the Y0 register with 16 LS zeros, sign extends the resulting 32-bit long word to 36 bits,
subtracts the result from the 36-bit A accumulator, and updates the CCR (leaving the A accumulator
unchanged).
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
A-96
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of the result is in use
Set if result is not normalized
Set if bit 35 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 35 of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CMP
CMP
Compare
Instruction Fields:
Operation
Operands
C
W
CMP
EEE,EEE
1
1
36-bit compare two accumulators or data registers.
X:(Rn),FF
2
1
Compare memory word with 36 bit accumulator.
X:(Rn+xxxx),FF
3
2
X:(SP–xx),FF
3
1
X:xxxx,FF
2
2
X:xxxxxx,FF
3
3
Comments
Also see CMP.W.
Note: Condition codes are set based on a 36-bit
result. See CMP.W for condition codes on 16 bits.
#<0–31>,FF
1
1
Compare accumulator with an immediate integer 0–31.
#xxxx,FF
2
2
Compare accumulator with a signed 16-bit immediate.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
CMP2
X0,F
Y1,F
Y0,F
C,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A,B
B,A
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-97
CMP
CMP
Compare
Instruction Opcodes:
15
CMP
#<0–31>,FF
CMP
#xxxx,FF
0
1
0
1
0
15
0
12
11
1
1
12
11
1
1
8
7
F
1
0
F
8
7
1
0
F
F
4
3
B
0
0
B
4
3
1
0
0
0
4
3
0
0
4
3
0
0
4
3
0
0
B
B
B
0
0
0
0
iiiiiiiiiiiiiiii
15
CMP
C,F GGG,X:<ea_m>
0
0
0
15
CMP
C,F X:<ea_m>,GGG
0
0
1
15
CMP
DD,F GGG,X:<ea_m>
CMP
DD,F X:<ea_m>,GGG
CMP
EEE,EEE
0
0
0
0
1
1
1
15
0
15
0
15
CMP
X:(Rn),FF
0
1
0
15
CMP
X:(Rn+xxxx),FF
0
1
0
12
11
1
1
12
11
1
1
12
11
1
1
12
11
1
1
12
11
1
1
12
11
1
1
12
11
1
1
G
G
G
G
8
7
G
F
8
7
G
F
8
7
F
G
G
G
8
7
G
G
G
F
8
7
0
E
E
E
8
7
F
F
8
7
F
F
1
0
1
0
1
1
1
1
J
J
J
4
3
J
J
J
0
4
3
a
a
a
0
4
3
1
R
4
3
1
R
1
1
0
0
0
m
R
R
0
m
R
R
0
m
R
R
m
R
R
1
0
0
0
0
0
1
R
R
0
0
R
R
a
a
a
AAAAAAAAAAAAAAAA
15
CMP
X:(SP–xx),FF
0
1
0
15
CMP
X:xxxx,FF
0
1
0
12
11
1
1
12
11
1
1
0
0
1
0
8
7
F
F
8
7
F
F
1
1
a
0
4
3
a
a
4
3
0
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
15
CMP
X:xxxxxx,FF
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
1
1
1
0
F
F
1
0
0
0
1
0
0
4
3
0
0
4
3
0
0
AAAAAAAAAAAAAAAA
15
CMP
~F,F GGG,X:<ea_m>
0
0
0
15
CMP
~F,F X:<ea_m>,GGG
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-98
0
0
1
12
11
1
1
12
11
1
1
G
G
G
G
8
7
G
F
8
7
G
F
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
0
0
m
R
R
0
m
R
R
Freescale Semiconductor
CMP.B
Operation:
D–S
CMP.B
Compare Byte
Assembler Syntax:
(no parallel move)
CMP.B
S,D
(no parallel move)
Description: Compare 8-bit portions of two registers or a register and an immediate value. The two operands are
subtracted to perform the comparison, and the CCR is updated accordingly. The result of the subtraction operation is not stored. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Note:
This instruction subtracts 8-bit operands. When a register is specified, the low-order 8 bits of the register is used for the comparison, unless the register is an accumulator, in which case the low-order
8 bits of the FF1 portion are used. Both registers and immediate values are sign extended internally to
20 bits before comparison.
Example:
CMP.B
#$24,A
; compare value in A accumulator to hex 24
Before Execution
After Execution
0
0020
0000
0
0020
0000
A2
A1
A0
A2
A1
A0
SR
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000. Execution of the CMP.B
instruction automatically sign extends the immediate value to 20 bits, sign extends the low-order 8 bits
of A1, and subtracts the immediate from the accumulator. The CCR is updated based on the result of
the 8-bit comparison; the A accumulator is unchanged.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is not normalized
Set if bit 7 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 7 of the result
Freescale Semiconductor
Instruction Set Details
A-99
CMP.B
CMP.B
Compare Byte
Instruction Fields:
Operation
Operands
C
W
Comments
CMP.B
EEE,EEE
1
1
Compare the 8-bit byte portions of two data registers
#<0–31>,EEE
1
1
Compare the byte portion of a data register with an
immediate integer 0–31
#xxx,EEE
2
2
Compare with a 9-bit signed immediate integer
12
11
1
0
12
11
1
1
Instruction Opcodes:
15
CMP.B #<0–31>,EEE
0
1
0
15
CMP.B #xxx,EEE
0
1
0
1
1
E
E
8
7
E
E
8
7
E
E
0
1
4
3
B
B
4
3
0
1
0
0
4
3
a
a
a
0
0
B
B
B
0
0
1
0
1
0
1
iiiiiiiiiiiiiiii
15
CMP.B EEE,EEE
0
Timing:
1–2 oscillator clock cycle(s)
Memory:
1–2 program word(s)
A-100
1
1
12
11
1
1
0
E
8
7
E
E
DSP56800E and DSP56800EX Core Reference Manual
0
Freescale Semiconductor
CMP.BP
Operation:
D–S
CMP.BP
Compare Byte (Byte Pointer)
Assembler Syntax:
(no parallel move)
CMP.BP
S,D
(no parallel move)
Description: Compare a byte in memory with the 8-bit portion of a register. The two operands are subtracted to perform the comparison, and the CCR is updated accordingly. The result of the subtraction operation is
not stored. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Note:
This instruction subtracts 8-bit operands. The low-order 8 bits of the register is used for the comparison, unless the register is an accumulator, in which case the low-order 8 bits of the FF1 portion are
used. Both the register and the byte located in memory are sign extended internally to 20 bits before
the comparison.
Example:
CMP.BP X:$3065,A
; compare byte at X:$3065 and A
Before Execution
After Execution
0
0020
0000
0
0020
0000
A2
A1
A0
A2
A1
A0
X Memory
7
0
07
Byte
Addresses
7
$3068
$3066
$3064
$3062
X Memory
0
07
88
66
44
77
55
33
22
11
SR
Byte
Addresses
$3068
$3066
$3064
$3062
88
66
44
77
55
33
22
11
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000, and location $3065 in
data memory contains $44. Execution of the CMP.BP instruction automatically sign extends the memory byte and low-order 8 bits of A1 to 20 bits, and then it subtracts the memory value from the accumulator. The CCR is updated based on the result of the 8-bit comparison; the A accumulator is unchanged. Note that this address is equivalent to the upper byte of word address $1832.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is not normalized
Set if bit 7 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 7 of the result
Freescale Semiconductor
Instruction Set Details
A-101
CMP.BP
CMP.BP
Compare Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
CMP.BP
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
Comments
Compare memory byte with register
Instruction Opcodes:
15
CMP.BP X:xxxx,EEE
0
1
0
12
11
1
1
1
E
8
7
E
E
1
0
4
3
0
0
4
3
0
1
1
0
AAAAAAAAAAAAAAAA
15
CMP.BP X:xxxxxx,EEE
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
1
1
1
E
E
E
1
0
0
0
1
1
0
AAAAAAAAAAAAAAAA
Timing:
2–3 oscillator clock cycles
Memory:
2–3 program words
A-102
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CMP.L
Operation:
D–S
CMP.L
Compare Long
Assembler Syntax:
(no parallel move)
CMP.L
S,D
(no parallel move)
Description: Compare 32-bit portions of two registers, a register and a long word in memory, or a register and a
16-bit immediate value (sign extended to 32 bits). The two operands are subtracted to perform the comparison, and the CCR is updated accordingly. The result of the subtraction operation is not stored. The
result is not affected by the state of the saturation bit (SA).
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Note:
This instruction subtracts 32-bit operands. All values are sign extended internally to 36 bits before the
comparison.
Example:
CMP.L
Y,A
; 32-bit compare of Y and A
Before Execution
After Execution
0
0020
0000
0
0020
0000
A2
A1
A0
A2
A1
A0
0024
0000
0024
0000
Y1
Y0
Y1
Y0
SR
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000. Execution of the
CMP.L Y,A instruction automatically sign extends both operands to 36 bits and then subtracts the Y
register from the accumulator. The CCR is updated based on the result of the 32-bit comparison; both
registers are unchanged.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 36-bit result is in use
Set if the 36-bit result is not normalized
Set if bit 31 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 31 of the result
Freescale Semiconductor
Instruction Set Details
A-103
CMP.L
CMP.L
Compare Long
Instruction Fields:
Operation
Operands
C
W
Comments
CMP.L
FFF,FFF
1
1
Compare the 32-bit long portions of two data registers
or accumulators
X:xxxx,fff
2
2
Compare memory long with a data register
X:xxxxxx,fff
3
3
#xxxx,fff
2
2
Compare a 16-bit immediate value, sign extended to
32 bits, with a data register
12
11
1
1
Instruction Opcodes:
15
CMP.L #xxxx,fff
0
1
0
1
f
8
7
f
f
1
0
4
3
0
0
4
3
0
0
0
1
1
iiiiiiiiiiiiiiii
15
CMP.L FFF,FFF
0
1
1
1
0
15
CMP.L X:xxxx,fff
0
12
11
1
1
12
11
1
1
0
F
1
f
8
7
F
F
8
7
f
f
b
b
b
4
3
1
0
0
0
0
1
1
1
1
1
1
0
AAAAAAAAAAAAAAAA
15
CMP.L X:xxxxxx,fff
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
1
1
1
f
f
f
1
0
0
0
1
1
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-104
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CMP.W
Operation:
D–S
CMP.W
Compare Word
Assembler Syntax:
(no parallel move)
CMP.W
S,D
(no parallel move)
Description: Compare two 16-bit operands. The operands are subtracted, and the CCR is updated based on the result. The result of the subtraction operation is not stored.
Usage:
This instruction can be used for both integer and fractional two’s-complement data.
Note:
This instruction subtracts 16-bit operands. When an accumulator is used as one of the operands, the
FF1 portion is compared. Registers and 16-bit immediate values are sign extended internally to 20 bits
before the subtraction is performed. Five-bit immediate values are zero extended to 20 bits. The CCR
is updated based on the 16-bit result, with the exception of the U and E bits, which are based on the
20-bit result.
Example:
CMP.W
Y0,A
; compare Y0 and A
Before Execution
After Execution
0
0020
0000
0
0020
0000
A2
A1
A0
A2
A1
A0
2000
0024
2000
0024
Y1
Y0
Y1
Y0
SR
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000, and the 16-bit Y0 register contains the value $0024. Execution of the CMP.W Y0,A instruction automatically sign extends
the 16-bit value in Y0 to 20 bits and subtracts the result from the FF2:FF1 portion of the A accumulator. The CCR is updated based on the result of the subtraction. Neither the Y0 nor the A registers are
changed.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is not normalized
Set if bit 15 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 15 of the result
Freescale Semiconductor
Instruction Set Details
A-105
CMP.W
CMP.W
Compare Word
Instruction Fields:
Operation
Operands
C
W
Comments
CMP.W
EEE,EEE
1
1
Compare the 16-bit word portions of two data registers
or accumulators
X:(Rn),EEE
2
1
X:(Rn+xxxx),EEE
3
2
Compare memory word with a data register or the word
portion of an accumulator
X:(SP–xx),EEE
3
1
X:xxxx,EEE
2
2
X:xxxxxx,EEE
3
3
#<0–31>,EEE
1
1
Compare the word portion of a data register with an
immediate integer 0–31
#xxxx,EEE
2
2
Compare the word portion of a data register with a
signed 16-bit immediate
A-106
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CMP.W
CMP.W
Compare Word
Instruction Opcodes:
15
CMP.W #<0–31>,DD
0
1
0
15
CMP.W #<0–31>,FF
0
1
0
15
CMP.W #xxxx,DD
0
1
0
12
11
1
1
12
11
0
1
12
11
1
1
1
1
1
1
0
1
8
7
D
D
8
7
F
F
8
7
D
D
0
0
1
0
0
0
4
3
B
B
4
3
B
B
4
3
0
0
4
3
0
0
4
3
a
0
4
3
R
0
B
B
B
0
B
B
B
0
0
0
0
iiiiiiiiiiiiiiii
15
CMP.W #xxxx,FF
0
1
0
12
11
0
1
1
0
8
7
F
F
1
0
0
0
0
0
iiiiiiiiiiiiiiii
15
CMP.W EEE,EEE
0
1
1
15
CMP.W X:(Rn),DD
0
1
0
1
0
1
0
15
CMP.W X:(Rn),FF
0
15
CMP.W X:(Rn+xxxx),DD
0
12
11
1
1
12
11
1
1
12
11
0
1
12
11
1
1
0
E
8
7
E
E
8
7
D
1
1
D
8
7
1
0
F
F
8
7
1
1
D
D
a
a
1
0
1
4
3
1
0
1
R
4
3
1
0
1
R
0
1
1
0
0
1
R
R
1
R
R
0
R
R
0
R
R
0
0
AAAAAAAAAAAAAAAA
15
CMP.W X:(Rn+xxxx),FF
0
1
0
12
11
0
1
1
0
8
7
F
F
1
0
4
3
1
R
4
3
a
0
AAAAAAAAAAAAAAAA
15
CMP.W X:(SP–xx),DD
0
1
0
1
0
1
0
15
CMP.W X:(SP–xx),FF
0
15
CMP.W X:xxxx,DD
0
12
11
1
1
12
11
0
1
12
11
1
1
8
7
D
0
1
D
8
7
0
0
F
F
8
7
1
1
D
D
1
a
a
4
3
1
a
a
a
4
3
1
0
0
0
0
a
a
a
a
a
a
1
0
0
1
0
0
0
0
AAAAAAAAAAAAAAAA
15
CMP.W X:xxxx,FF
0
1
0
12
11
0
1
1
0
8
7
F
F
1
0
4
3
0
0
0
AAAAAAAAAAAAAAAA
Freescale Semiconductor
Instruction Set Details
A-107
CMP.W
CMP.W
Compare Word
Instruction Opcodes:(continued)
15
CMP.W X:xxxxxx,FF
12
11
8
7
4
3
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
0
F
F
1
0
0
0
1
0
0
4
3
AAAAAAAAAAAAAAAA
15
CMP.W X:xxxxxx,DD
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
1
1
1
1
D
D
1
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-108
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
CMPA
Operation:
D–S
CMPA
Compare AGU Registers
Assembler Syntax:
(no parallel move)
CMPA
S,D
(no parallel move)
Description: Compare two AGU address registers by subtracting the source from the destination, and update the
CCR based on the result of the subtraction. The result of the subtraction operation is not stored.
Example:
CMPA
R0,R1
; compare R0 and R1
Before Execution
After Execution
R0
082473
R0
082473
R1
002473
R1
002473
SR
SR
0300
0309
Explanation of Example:
Prior to execution, the R0 register contains the value $082473, R1 contains the value $002473, and the
status register (SR) contains $0300. Execution of the CMPA R0,R1 instruction subtracts R0 from R1
and updates the CCR, leaving the registers unchanged.
Condition Codes Affected:
MR
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
Set if bit 23 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a borrow occurs from bit 23 of the result
Instruction Fields:
Operation
Operands
C
W
CMPA
Rn,Rn
1
1
Comments
24-bit compare between two AGU registers
Instruction Opcodes:
15
CMPA
Rn,Rn
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
1
0
0
12
11
0
1
1
1
Instruction Set Details
8
7
1
n
0
1
4
3
n
R
0
n
R
R
A-109
CMPA.W
Operation:
D–S
CMPA.W
Compare AGU Registers (Word)
Assembler Syntax:
(no parallel move)
CMPA.W
S,D
(no parallel move)
Description: Compare the low-order 16 bits of two AGU address registers by subtracting the source from the destination, and update the CCR based on the result of the subtraction. The result of the subtraction operation is not stored.
Usage:
This instruction is provided for compatibility with the DSP56800 CMPA instruction, and it should be
used when only 16-bit address comparisons are required.
Example:
CMPA.W R0,R1
; compare R0 and R1
After Execution
Before Execution
R0
082473
R0
082473
R1
002473
R1
002473
SR
SR
0300
0304
Explanation of Example:
Prior to execution, the R0 register contains the value $082473, R1 contains the value $002473, and the
status register (SR) contains $0300. Execution of the CMPA.W R0,R1 instruction subtracts the
low-order 16 bits of R0 from the low-order 16 bits of R1 and updates the CCR, leaving the registers
unchanged. In this case, both address registers are considered equal.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
N
Z
V
C
—
—
—
—
Set if bit 15 of the result is set
Set if result equals zero
Set if overflow has occurred in result
Set if a borrow occurs from bit 15 of the subtraction
Instruction Fields:
Operation
Operands
C
W
CMPA.W
Rn,Rn
1
1
Comments
16-bit compare between two AGU registers
Instruction Opcodes:
15
CMPA.W Rn,Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-110
0
0
12
11
0
1
1
1
8
7
0
n
0
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
n
R
0
n
R
R
Freescale Semiconductor
DEBUGEV
DEBUGEV
Generate Debug Event
Operation:
Assembler Syntax:
Generate a debugging event
DEBUGEV
Description: Generate a debugging event in the Enhanced OnCE module. For more information on the Enhanced
OnCE port and hardware debugging support, see the manual for the appropriate DSC device.
Note:
This instruction is equivalent to the DSP56800 DEBUG instruction. Programs that are being ported
from the DSP56800 should use this instruction in place of the DEBUG instruction to remain compatible with the DSP56800 behavior.
Condition Codes Affected:
No condition codes are affected.
Instruction Fields:
Operation
Operands
DEBUGEV
C
W
3
1
Comments
Generate a debug event
Instruction Opcodes:
15
DEBUGEV
1
Timing:
3 oscillator clock cycles
Memory:
1 program word
Freescale Semiconductor
1
1
12
11
0
0
1
1
Instruction Set Details
8
7
1
0
0
0
4
3
0
0
0
0
1
1
A-111
DEBUGHLT
DEBUGHLT
Enter Debug Mode
Operation:
Assembler Syntax:
Enter the debug processing state
DEBUGHLT
Description: Enter the debug processing state and wait for Enhanced OnCE port commands, if this state is enabled
in the Enhanced OnCE unit. If this state is not enabled, then the processor simply executes two NOPs
and continues program execution. For more information on the Enhanced OnCE port and hardware debugging support, see the manual for the appropriate DSC device.
Note:
This instruction is not compatible with the DSP56800 DEBUG instruction. Please see the DEBUGEV
instruction for information on DSP56800–compatible debugging.
Condition Codes Affected:
No condition codes are affected.
Instruction Fields:
Operation
Operands
DEBUGHLT
C
W
3
1
Comments
Enter debug processing state
Instruction Opcodes:
15
DEBUGHLT
1
Timing:
3 oscillator clock cycles
Memory:
1 program word
A-112
1
1
12
11
0
0
1
1
8
7
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
4
3
0
0
0
0
0
1
Freescale Semiconductor
DEC.BP
DEC.BP
Decrement Byte (Byte Pointer)
Operation:
Assembler Syntax:
D–1→D
(no parallel move)
DEC.BP
D
(no parallel move)
Description: Decrement a byte value in memory. The value is internally sign extended to 20 bits before being decremented. The low-order 8 bits of the result are stored back to memory. The condition codes are calculated based on the 8-bit result, with the exception of the E and U bits, which are calculated based on
the 20-bit result. Absolute addresses are expressed as byte addresses. The result is not affected by the
state of the saturation bit (SA).
Usage:
This instruction is typically used when integer data is processed.
Example:
DEC.BP X:$3065
; decrement the byte at (byte) address $3065
Before Execution
After Execution
Byte
Addresses
Byte
Addresses
X Memory
0 7
7
0
$3068
$3066
$3064
$3062
88
66
00
77
55
33
22
11
SR
X Memory
0
07
7
$3068
$3066
$3064
$3062
SR
0300
88
66
FF
77
55
33
22
11
0319
Explanation of Example:
Prior to execution, the value at byte address X:$3065 is $00. Execution of the DEC.BP instruction decrements this value by one and generates the result, $FF, with a borrow (the carry bit is set). The result
is negative since bit 7 is set. Note that this address is equivalent to the upper byte of word address
$1832.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if bit 7 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 7 of the result
Freescale Semiconductor
Instruction Set Details
A-113
DEC.BP
DEC.BP
Decrement Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
DEC.BP
X:xxxx
3
2
X:xxxxxx
4
3
Comments
Decrement byte in memory
Instruction Opcodes:
15
DEC.BP X:xxxx
0
1
0
12
11
0
1
1
1
8
7
0
0
1
0
4
3
0
0
4
3
0
1
1
0
AAAAAAAAAAAAAAAA
15
DEC.BP X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
0
0
1
0
0
0
1
1
0
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
2–3 program words
A-114
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DEC.L
DEC.L
Decrement Long
Operation:
Assembler Syntax:
D–1→D
(no parallel move)
DEC.L
D
(no parallel move)
Description: Decrement a long-word value in a register or memory. When an operand located in memory is operated
on, the low-order 32 bits of the result are stored back to memory. The condition codes are calculated
based on the 32-bit result. Absolute addresses pointing to long elements must always be even aligned
(that is, pointing to the lowest 16 bits).
Usage:
This instruction is typically used when integer data is processed.
Example:
DEC.L
X:$2000
; decrement value in location: $2001:2000 by 1
Before Execution
X Memory
After Execution
X Memory
$2001
$2000
$1FFF
1000
0000
8000
$2001
$2000
$1FFF
0FFF
FFFF
8000
SR
0301
0300
SR
0310
Explanation of Example:
Prior to execution, the 32-bit value at location $2001:2000 is $1000:0000. Execution of the DEC.L instruction subtracts this value by one and generates $0FFF:FFFF. The CCR is updated based on the result of the subtraction.
Condition Codes Affected:
MR
E
U
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the result is in use
Set if the 32-bit result is unnormalized
Set if bit 31 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 31 of the result
Freescale Semiconductor
Instruction Set Details
A-115
DEC.L
DEC.L
Decrement Long
Instruction Fields:
Operation
Operands
C
W
Comments
DEC.L
fff
1
1
Decrement long
X:xxxx
3
2
Decrement long in memory
X:xxxxxx
4
3
Instruction Opcodes:
15
DEC.L X:xxxx
0
1
0
12
11
0
1
1
1
8
7
0
0
1
0
4
3
0
0
4
3
0
1
1
1
AAAAAAAAAAAAAAAA
15
DEC.L X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
0
0
1
0
0
0
1
1
1
0
1
1
AAAAAAAAAAAAAAAA
15
DEC.L fff
0
Timing:
1–4 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-116
1
1
12
11
1
0
0
f
8
7
f
f
0
DSP56800E and DSP56800EX Core Reference Manual
0
4
3
1
1
0
Freescale Semiconductor
DEC.W
Operation:
D–1→D
D–1→D
DEC.W
Decrement Word
Assembler Syntax:
(one parallel move)
(no parallel move)
DEC.W
DEC.W
D
D
(one parallel move)
(no parallel move)
Description: Decrement a 16-bit destination by one. If the destination is an accumulator, only the EXT and MSP
portions of the accumulator are used and the LSP remains unchanged. The condition codes are calculated based on the 16-bit result (or on the 20-bit result for accumulators).
Usage:
This instruction is typically used when integer data is processed.
Example:
DEC.W
A
X:(R2)+,X0
; Decr the 20 MSBs of A, update R2,X0
A After Execution
A Before Execution
0
0001
0033
0
0000
0033
A2
A1
A0
A2
A1
A0
SR
SR
0300
0314
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0001:0033. Execution of the
DEC.W instruction decrements by one the upper 20 bits of the A accumulator and sets the zero bit in
the CCR. A new value is read in parallel and stored in register X0; the address register R2 is post-incremented.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of the result is in use
Set if result is unnormalized
Set if bit MSB of the result is set
Set if the result is zero (20 MSB for accumulator destinations)
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 15 of the result (bit 35 for accumulators)
When the destination is one of the four accumulators, condition code calculations follow the rules for
20-bit arithmetic; otherwise, the rules for 16-bit arithmetic apply.
Freescale Semiconductor
Instruction Set Details
A-117
DEC.W
DEC.W
Decrement Word
Instruction Fields:
Operation
Operands
C
W
Comments
DEC.W
EEE
1
1
Decrement word.
X:(Rn)
3
1
X:(Rn+xxxx)
4
2
Decrement word in memory using appropriate addressing mode.
X:(SP–xx)
4
1
X:xxxx
3
2
X:xxxxxx
4
3
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
DEC.W2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
A-118
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DEC.W
DEC.W
Decrement Word
Instruction Opcodes:
15
DEC.W EEE
0
1
1
15
DEC.W F GGG,X:<ea_m>
0
0
0
15
DEC.W F X:<ea_m>,GGG
0
0
1
1
0
1
0
15
DEC.W X:(Rn)
0
15
DEC.W X:(Rn+xxxx)
0
12
11
1
0
12
11
0
0
12
11
0
0
12
11
0
1
12
11
0
1
0
E
G
G
8
7
E
E
8
7
G
F
8
7
F
G
G
G
8
7
1
1
0
0
8
7
1
1
0
0
0
0
0
0
4
3
0
1
4
3
1
0
4
3
0
0
0
1
4
3
1
0
1
R
4
3
1
0
1
R
0
0
1
1
0
m
R
R
0
m
R
R
1
R
R
0
R
R
a
a
a
1
0
0
0
0
AAAAAAAAAAAAAAAA
15
DEC.W X:(SP–xx)
0
1
0
1
0
15
DEC.W X:xxxx
0
12
11
0
1
12
11
0
1
8
7
0
1
0
0
8
7
1
1
0
0
4
3
1
a
a
a
4
3
1
0
0
0
0
0
AAAAAAAAAAAAAAAA
15
DEC.W X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
0
1
1
1
0
0
1
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–4 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-119
DECA
DECA
Decrement AGU Register
Operation:
Assembler Syntax:
D–1→D
(no parallel move)
DECA
D
(no parallel move)
Description: Decrement a value in an AGU pointer register. The full 24-bit value of the pointer register is used when
decrementing.
Usage:
This instruction can be used to step backwards through a memory buffer.
Example:
DECA
R0
; decrement R0
Before Execution
R0
After Execution
R0
002222
002221
Explanation of Example:
Prior to execution, the R0 register contains $002222. Execution of the DECA R0 instruction causes
the value in R0 to be reduced by one, and the result ($002221) is stored back in R0.
Condition Codes Affected:
The condition codes are not modified by this instruction.
Instruction Fields:
Operation
Operands
C
W
DECA
Rn
1
1
Comments
Decrement AGU register by one
Instruction Opcodes:
15
DECA
Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-120
0
0
12
11
0
0
1
0
8
7
0
1
0
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
1
R
0
0
R
R
Freescale Semiconductor
DECA.L
DECA.L
Decrement Long in AGU Register
Operation:
Assembler Syntax:
D–2→D
(no parallel move)
DECA.L
D
(no parallel move)
Description: Decrement a value in an AGU pointer register by two. The full 24-bit value of the pointer register is
used when decrementing.
Usage:
This instruction is used to step backwards through a memory buffer that is composed of long-word values. Since each long word consists of 2 words, this instruction can be used to step through a buffer by
every other word.
Example:
DECA.L R0
; decrement R0 by 2
Before Execution
R0
After Execution
R0
002222
002220
Explanation of Example:
Prior to execution, the R0 register contains $002222. Execution of the DECA.L R0 instruction causes
the value in the R0 to be reduced by two, and the result ($002220) is stored back in R0.
Condition Codes Affected:
The condition codes are not modified by this instruction.
Instruction Fields:
Operation
Operands
C
W
DECA.L
Rn
1
1
Comments
Decrement AGU register by two
Instruction Opcodes:
15
DECA.L Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
0
12
11
0
0
1
0
Instruction Set Details
8
7
0
1
0
1
4
3
1
R
0
1
R
R
A-121
DECTSTA Decrement and Test AGU Register DECTSTA
Operation:
Assembler Syntax:
D–1→D
D–0
(no parallel move)
DECTSTA
D
(no parallel move)
Description: Decrement a value in an AGU pointer register and then compare the result to zero, updating the condition codes based on the comparison. The full 24-bit value of the pointer register is used when decrementing.
Usage:
This instruction can be used to step backwards through a memory buffer, testing to see that the pointer
is still valid after each step.
Example:
DECTSTA R0
; decrement R0 and then compare to 0
Before Execution
R0
After Execution
R0
002222
SR
002221
SR
0308
0300
Explanation of Example:
Prior to execution, the R0 register contains $002222. Execution of the DECTSTA R0 instruction causes the value in R0 to be reduced by one, and the result ($002221) is stored back in R0. The updated
value in R0 is then compared with zero, and the CCR is updated accordingly.
Condition Codes Affected:
MR
N
Z
V
C
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
Set if bit 23 of the result is set
Set if all bits in the result are zero
Set if overflow has occurred in result
Set if a borrow occurs from bit 23 of the result
Instruction Fields:
Operation
Operands
C
W
DECTSTA
Rn
1
1
Comments
Decrement and test AGU register
Instruction Opcodes:
15
DECTSTA Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-122
0
0
12
11
0
0
1
0
8
7
0
1
0
DSP56800E and DSP56800EX Core Reference Manual
1
4
3
0
R
0
1
R
R
Freescale Semiconductor
DIV
DIV
Divide Iteration
Operation:
Assembler Syntax:
(see following figure)
DIV
S,D
(no parallel move)
If D[35] ⊕ S[15] = 1
Then
D2
D1
C;
D1 + S
D1
C;
D1 – S
D1
D0
Else
D2
D1
D0
Description: This instruction is a divide iteration that is used to calculate 1 bit of the result of a division. After the
correct number of iterations, this instruction will divide the destination operand (D)—dividend or numerator—by the source operand (S)—divisor or denominator—and store the result in the destination
accumulator. The 32-bit dividend must be a positive value that is correctly sign extended to 36 bits and
that is stored in the full 36-bit destination accumulator. The 16-bit divisor is a signed value and is
stored in the source operand. (The division of signed numbers is handled using the techniques documented in Section 5.3.4, “Division,” on page 5-21.) This instruction can be used for both integer and
fractional division. Each DIV iteration calculates 1 quotient bit using a non-restoring division algorithm (see the description that follows). After the execution of the first DIV instruction, the destination
operand holds both the partial remainder and the formed quotient. The partial remainder occupies the
high-order portion of the destination accumulator and is a signed fraction. The formed quotient occupies the low-order portion of the destination accumulator (A0 or B0, C0, or D0) and is a positive fraction. One bit of the formed quotient is shifted into the LSB of the destination accumulator at the start
of each DIV iteration. The formed quotient is the true quotient if the true quotient is positive. If the true
quotient is negative, the formed quotient must be negated. For fractional division, valid results are obtained only when |D| < |S|. This condition ensures that the magnitude of the quotient is less than one
(that is, it is fractional) and precludes division by zero.
The DIV instruction calculates 1 quotient bit based on the divisor and the previous partial remainder.
To produce an N-bit quotient, the DIV instruction is executed N times, where N is the number of bits
of precision that is desired in the quotient (1 < N < 16). Thus, for a full-precision (16-bit) quotient,
16 DIV iterations are required. In general, executing the DIV instruction N times produces an N-bit
quotient and a 32-bit remainder, which has (32 – N) bits of precision and whose N MSBs are zeros.
The partial remainder is not a true remainder and must be corrected (due to the non-restoring nature of
the division algorithm) before it may be used. Therefore, once the divide is complete, it is necessary
to reverse the last DIV operation and restore the remainder to obtain the true remainder. The result is
not affected by the state of the saturation bit (SA).
The DIV instruction uses a non-restoring division algorithm that consists of the following operations:
1. Compare the source and destination operand sign bits. An exclusive OR operation is performed on
bit 35 of the destination operand and bit 15 of the source operand.
2. Shift the partial remainder and the quotient. The 36-bit destination accumulator is shifted 1 bit to the
left. C is moved into the LSB (bit 0) of the accumulator.
Freescale Semiconductor
Instruction Set Details
A-123
DIV
DIV
Divide Iteration
3. Calculate the next quotient bit and the new partial remainder. The 16-bit source operand (signed
divisor) is either added to or subtracted from the MSP of the destination accumulator (FF1 portion),
and the result is stored back into the MSP of the destination accumulator. If the result of the exclusive
OR operation in the first step was one (that is, the sign bits were different), the source operand S is
added to the accumulator. If the result of the exclusive OR operation was zero (that is, the sign bits
were the same), the source operand S is subtracted from the accumulator. Due to the automatic sign
extension of the 16-bit signed divisor, the addition or subtraction operation correctly sets the C bit
with the next quotient bit.
Usage:
The DIV iteration instruction can be used in one of several different division algorithms, depending on
the needs of an application. Section 5.3.4, “Division,” on page 5-21 shows the correct usage of this instruction for fractional and integer division routines, discusses in detail issues related to division, and
provides several examples. The division routine is greatly simplified if both operands are positive, or
if it is not necessary also to calculate a remainder.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
L
V
C
— Set if overflow bit V is set
— Set if the MSB of the destination operand (bit 35 for an accumulator,
bit 35 after sign extension for the Y register) is changed as a result
of the instruction’s left shift operation; otherwise, V is cleared
— Set if MSB of the result is zero (bit 35 for an accumulator,
bit 35 after sign extension for the Y register)
Example:
DIV
Y0,A
; divide A by Y0
Before Execution
After Execution
0
0702
0000
0
0E00
0001
A2
A1
A0
A2
A1
A0
2000
0004
2000
0004
Y1
Y0
Y1
Y0
SR
0301
SR
0301
Explanation of Example:
This example shows only a single iteration of the division instruction. Please refer to Section 5.3.4,
“Division,” on page 5-21 for a complete description of a division algorithm.
A-124
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DIV
DIV
Divide Iteration
Instruction Fields:
Operation
Operands
C
W
Comments
DIV
FFF1,fff
1
1
Divide iteration
12
11
1
1
Instruction Opcodes:
15
DIV
FFF1,fff
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
0
f
Instruction Set Details
8
7
f
f
c
c
4
3
c
1
0
1
1
1
A-125
DO
DO
Start Hardware DO Loop
Operation:
Assembler Syntax:
HWS0 → HWS1;
LC → LC2
LA → LA2
LF → NL
PC → HWS0
S → LC
D → LA
1 → LF
DO
S,D
Operation When Loop Completes (End-of-Loop Processing):
If NL == 1
LC2 → LC, LA2 → LA
HWS1 → HWS0
NL → LF
0 → NL
Description: Begin a hardware DO loop that is to be repeated for the number of times specified in the instruction’s
source operand, and whose range of execution is terminated by the destination operand. The source
operand specifies the loop count and can be either an immediate 6-bit unsigned value or an on-chip
register value, and the destination operand is a 16- or 21-bit absolute address. No overhead other than
the execution of the DO instruction is required to set up this loop. When a DO loop is executed, the
instructions are actually fetched each time through the loop. Therefore, a DO loop can be interrupted.
The DO instruction performs hardware looping on a single instruction or a block of instructions. DO
loops can be nested up to two deep, accelerating more complex algorithms.
Example 1:
DO
MOVE.L
MOVE.L
#40,END_CPY ; Set up hardware DO loop
X:(R0)+,A
; Copy a 32-bit memory location
A10,X:(R1)+ ;
END_CPY
Explanation of Example:
This example copies a block of forty 32-bit memory locations from one area of memory to another.
When a hardware DO loop is initiated, the following events occur:
1.
When the DO instruction is executed, the contents of the LC register are copied to the LC2
register, and LC is loaded with the loop count that the instruction specifies.
2.
The old contents of the LA register are copied to the LA2 register, and the LA register is
loaded with the address of the last instruction word in the loop. If a 16-bit address is
specified, the upper 8 bits of LA are cleared.
3.
The address of the first instruction in the program loop (top-of-loop address) is pushed onto
the hardware stack. This push sets the LF bit and updates the NL bit, as occurs with any
hardware stack push.
Instructions in the loop are then executed. The address of each instruction is compared to the value in
LA to see if it is the last instruction in the loop. When the end of the loop is reached, the loop count
register is checked to see if the loop should be repeated. If the value in LC is greater than one, LC is
decremented and the loop is re-started from the top. If LC is equal to one, the loop has been executed
for the proper number of times and should be exited.
When a hardware loop ends, the hardware stack is popped (and the popped value is discarded), the LA2
register is copied to LA, the LC2 register is copied to LC, and the NL bit in the operating mode register
is copied to the LF bit. Instruction execution then continues at the address that immediately follows the
end-of-loop address.
A-126
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DO
Start Hardware DO Loop
DO
Explanation of Example:(continued)
One hardware stack location is used for each nested DO or DOSLC loop. Thus, a two-deep hardware
stack allows for a maximum of two nested loops. The REP instruction does not use the hardware stack,
so repeat loops can be nested within DO loops.
Example 2:
MOVE.W
.
.
.
DO
MOVE.L
MOVE.L
#0,X0
X0,END_CPY ; Loop count is zero upon entry
X:(R0)+,A
; Copy a 32-bit memory location
A10,X:(R1)+ ;
END_CPY
Explanation of Example:
A loop count of zero is specified, so the instructions in the body of the loop are skipped, and execution
continues with the instruction immediately following the loop body.
Note that an immediate loop count of zero for the DO instruction is not allowed and will be rejected
by the assembler. A loop count of zero can only be specified by using a register that is loaded with zero
as the argument to the DO instruction, or by placing a zero in the LC register and executing DOSLC.
A DO loop normally terminates when the body of the loop has been executed for the specified number
of times (the end of the loop has been reached, and LC is one). Alternately, a DO loop terminates if the
count specified is zero, which causes the body of the loop to be skipped entirely.
When the inner loop of a nested loop terminates naturally, the LA2 and LC2 registers are copied into
the LA and LC registers, respectively, restoring these two registers with their values for the outer loop.
A loop is determined to be a nested inner loop if the OMR’s NL bit is set. If the NL bit is not set, the
LA and LC registers are not modified when a loop is terminated or skipped.
If it is necessary to terminate a DO loop early, use one of the techniques discussed in Section 8.5.4.1,
“Allowing Current Block to Finish and Then Exiting,” on page 8-20 and Section 8.5.6.2, “Nesting a
DO Loop Within a DO Loop,” on page 8-22.
During the end-of-loop processing, the NL bit is written into the LF, and the NL bit is cleared. The
contents of the second HWS location (HWS1) are written into the first HWS location (HWS0). Instruction fetches now continue at the address of the instruction that follows the last instruction in the DO
loop.
DO loops can also be nested as shown in Section 8.5.6, “Nested Hardware Looping,” on page 8-22.
When DO loops are nested, the end-of-loop addresses must also be nested and are not allowed to be
equal. The assembler generates an error message when DO loops are improperly nested.
Note:
The assembler calculates the end-of-loop address that is to be loaded into LA by subtracting one from
the absolute address specified in the destination operand. This process occurs to accommodate the case
in which the last instruction in the DO loop is a multiple-word instruction. Thus, the end-of-loop absolute address in the source code must represent the address of the instruction after the last instruction
in the loop.
The LF is cleared by a hardware reset.
Note:
Any data dependencies due to pipelining also apply to the pair of instructions formed by the last instruction in the DO loop and the first instruction of the DO loop.
Freescale Semiconductor
Instruction Set Details
A-127
DO
DO
Start Hardware DO Loop
Example 3:
DO
MOVE.W
REP
ASL
MOVE.W
END
#cnt1,END
X:(R0),A
#cnt2
A
A,X:(R0)+
:
; begin DO loop
;
;
;
;
nested REP loop
repeat this instruction
last instruction in DO loop
(outside DO loop)
Explanation of Example:
This example illustrates a DO loop with a REP loop nested within the DO loop. In this example, “cnt1”
values are fetched from memory; each value is left shifted by “cnt2” counts and is stored back in memory. The DO loop executes “cnt1” times while the ASL instruction inside the REP loop executes for a
number of times equal to “cnt1” × “cnt2.” The END label is located at the first instruction past the end
of the DO loop, as mentioned previously.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
*
*
*
*
*
I1
I0
SZ
L
E
U
N
Z
V
C
LF —
L —
Set when a DO loop is in progress
Set if data limiting occurred
Restrictions: Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
A-128
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DO
DO
Start Hardware DO Loop
Instruction Fields:
Operation
Operands
C
W
DO
#<1–63>,<ABS16>
3
2
#<1–63>,<ABS21>
4
3
Note:
#<1–63>,<ABS16>
5
2
#<1–63>,<ABS21>
6
3
DDDDD,<ABS16>
7
2
DDDDD,<ABS21>
8
3
Comments
At least 2 instruction words in the loop (t = 0 in the opcode field).
Only 1 instruction word in the loop (t = 1 in the opcode field).
If LC value is zero, body of loop is skipped (adds 2 instruction
cycles).
When looping with a value in an accumulator, use A1, B1, C1, or
D1 to avoid saturation when reading the accumulator.
Any DDDDD register is allowed except C2, D2, C0, D0,
C, D, Y, M01, N3, LA, LA2, LC, LC2, SR, OMR, and HWS.
The immediate value of zero is not allowed.
Instruction Opcodes:
15
DO
#<1–63>,<ABS16>
1
1
1
12
11
0
1
0
0
8
7
t
0
0
B
4
3
B
B
4
3
0
B
B
B
AAAAAAAAAAAAAAAA
15
DO
#<1–63>,<ABS21>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
1
0
0
t
0
0
B
B
B
B
B
B
d
d
d
AAAAAAAAAAAAAAAA
15
DO
DDDDD,<ABS16>
1
1
1
12
11
0
1
0
1
8
7
1
0
0
0
4
3
d
d
0
AAAAAAAAAAAAAAAA
15
DO
DDDDD,<ABS21>
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
0
1
0
1
1
0
0
0
d
d
d
d
d
AAAAAAAAAAAAAAAA
Timing:
3–8 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-129
DOSLC
DOSLC
DO Loop with Value in LC
Operation:
Assembler Syntax:
HWS0 → HWS1;
LA → LA2
LF → NL
PC → HWS0
D → LA
1 → LF
DOSLC
D
Operation When Loop Completes (End-of-Loop Processing):
If NL == 1
LC2 → LC, LA2 → LA
HWS1 → HWS0
NL → LF
0 → NL
Description: Begin a hardware DO loop that is to be repeated for the number of times specified in the loop counter
(LC) register. The value of LC must be loaded prior to executing this instruction. If the value in LC is
zero or negative, the instructions in the body of the loop are skipped. The destination operand D can
be a 16- or 21-bit absolute address. See the section on the DO instruction for more information on hardware looping.
Example:
MOVEU.W
...
DOSLC
MOVE.W
NEG
MOVE.W
END
#count,LC
; load LC register
END
X:(R0),A
A
A,X:(R0)+
:
; begin DO loop with value in LC
; negate value from buffer
; last instruction in DO loop
; (outside DO loop)
Explanation of Example:
This example illustrates a DO loop with a pre-existing value for LC. For a number of words in the buffer equal to “count,” the loop reads word values from a buffer in memory, negates them, and writes the
values back. The END label is located at the first instruction past the end of the DO loop.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
*
*
*
*
*
I1
I0
SZ
L
E
U
N
Z
V
C
LF —
Set when a DO loop is in progress
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
A-130
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
DOSLC
DOSLC
DO Loop with Value in LC
Instruction Fields:
Operation
Operands
C
W
DOSLC
<ABS16>
3
2
<ABS21>
4
3
Comments
If LC ≤ 0, the body of the loop is skipped, adding 3 additional
cycles.
A minimum of 2 instruction words is required in the loop. The
assembler will generate an error if the loop body is less than 2
words.
Instruction Opcodes:
15
DOSLC <ABS16>
1
1
1
12
11
0
0
1
1
8
7
1
0
0
0
4
3
1
1
4
3
0
0
0
1
AAAAAAAAAAAAAAAA
15
DOSLC <ABS21>
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
1
1
1
0
0
0
1
1
0
0
1
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-131
ENDDO
ENDDO
End Current DO Loop
Operation:
Assembler Syntax:
If NL == 1
LC2 → LC, LA2 → LA
HWS1 → HWS0
NL → LF
0 → NL
ENDDO
Description: Terminate the current hardware DO loop immediately. Normally, a hardware DO loop is terminated
when the last instruction of the loop is executed and the current LC equals one, but this instruction can
terminate a loop before normal completion. If the value of the current DO LC is needed, it must be read
before the execution of the ENDDO instruction. Initially, the LF is restored from the NL bit, and the
top-of-loop address is purged from the HWS. The contents of the second HWS location are written into
the first HWS location, and the NL bit is cleared.
Example:
DO
Y0,ENDLP
; execute loop ending at ENDLP for (Y0)
:
MOVE.W
CMP
JNE
ENDDO
LC,A
Y1,A
CONTINU
;
;
;
;
get current value of loop counter (LC)
compare loop counter with value in Y1
go to ONWARD if LC not equal to Y1
LC equal to Y1, restore all DO regis-
ENDLP
:
:
#$1234,X0
;
;
;
;
go to NEXT
LC not equal to Y1, continue DO loop
(last instruction in DO loop)
(first instruction AFTER DO loop)
times
ters
JMP
CONTINU
ENDLP
MOVE.W
Explanation of Example:
This example illustrates the use of the ENDDO instruction to terminate the current DO loop. The value
of the LC is compared with the value in the Y1 register to determine if execution of the DO loop should
continue. The ENDDO instruction updates certain program controller registers but does not automatically jump past the end of the DO loop. Thus, if this action is desired, a JMP or BRA instruction (such
as JMP NEXT) must be included after the ENDDO instruction to transfer program control to the first
instruction past the end of the DO loop.
Note:
The ENDDO instruction updates the program controller registers appropriately but does not automatically jump past the end of the loop. This must be done explicitly by the programmer if it is desired.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-132
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
ENDDO
ENDDO
End Current DO Loop
Instruction Fields:
Operation
Operands
ENDDO
C
W
Comments
1
1
Remove one value from the hardware stack and update the NL and
LF bits appropriately
Note:
Does not branch to the end of the loop
Instruction Opcodes:
15
ENDDO
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
0
12
11
0
1
1
1
Instruction Set Details
8
7
1
0
0
0
4
3
1
1
0
0
1
1
A-133
EOR.L
EOR.L
Logical Exclusive OR Long
Operation:
Assembler Syntax:
S ⊕ D→ D (no parallel move)
S ⊕ D → D (one parallel move)
EOR.L
EOR.L
FFF,fff (no parallel move)
C,F
(one parallel move)
where ⊕ denotes the logical exclusive OR operator
Description: Perform a logical exclusive OR operation on the source operand with the destination operand, and store
the result in the destination. This instruction is a 32-bit operation. If the destination is a 36-bit accumulator, the exclusive OR operation is performed on the source with bits 31–0 of the accumulator. The
remaining bits of the destination accumulator are not affected. If the source is a 16-bit register, the
EOR.L operation is performed on the source and bits 31–16 of the destination. The other bits of the
destination remain unchanged. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used for the logical exclusive OR of two registers. If an exclusive OR of a 16-bit
immediate value with a register or memory location is desired, the EORC instruction is appropriate.
Example:
EOR.L
Y,B
;Exclusive OR of Y with B10
After Execution
Before Execution
5
5555
CC89
5
AA55
3389
B2
B1
B0
B2
B1
B0
FF00
FF00
FF00
FF00
Y1
Y0
Y1
Y0
SR
SR
030F
0309
Explanation of Example:
Prior to execution, the 32-bit Y register contains the value $FF00:FF00, and the 36-bit B accumulator
contains the value $5:5555:CC89. The EOR.L Y,B instruction performs a logical exclusive OR operation on the 32-bit value in the Y register with bits 31–0 of the B accumulator (B10) and stores the
36-bit result in the B accumulator. The the extension portion (B2) is not affected by the operation.
Condition Codes Affected:
MR
N
Z
V
A-134
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 31 of accumulator result or the MSB of the register result is set
— Set if bits 31–0 of accumulator result or all bits of the register result are zero
— Always cleared
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
EOR.L
EOR.L
Logical Exclusive OR Long
Instruction Fields:
Operation
Operands
C
W
EOR.L
FFF,fff
1
1
Comments
32-bit exclusive OR (XOR).
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
EOR.L2
C,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
EOR.L C,F GGG,X:<ea_m>
0
0
0
0
1
15
EOR.L C,F X:<ea_m>,GGG
0
15
EOR.L FFF,fff
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
1
12
11
1
1
12
11
1
1
8
7
G
G
G
F
8
7
G
G
G
F
8
7
f
f
0
f
Instruction Set Details
4
3
0
1
0
0
4
3
0
1
0
0
4
3
b
1
b
b
0
m
R
R
m
R
R
0
0
1
1
0
A-135
EOR.W
EOR.W
Logical Exclusive OR Word
Operation:
Assembler Syntax:
S ⊕ D → D (no parallel move)
S ⊕ D[31:16] → D[31:16] (no parallel move)
EOR.W
EOR.W
S,D
S,D
(no parallel move)
(no parallel move)
where ⊕ denotes the logical exclusive OR operator
Description: Perform a logical exclusive OR operation on the source operand (S) with the destination operand (D),
and store the result in the destination. This instruction is a 16-bit operation. If the destination is a 36-bit
accumulator, the exclusive OR operation is performed on the source with bits 31–16 of the accumulator. The remaining bits of the destination accumulator are not affected. The result is not affected by the
state of the saturation bit (SA).
Usage:
This instruction is used for the logical exclusive OR of two registers. If an exclusive OR of a 16-bit
immediate value with a register or memory location is desired, the EORC instruction is appropriate.
Example:
EOR.W
Y1,B
;Exclusive OR of Y1 with B1
After Execution
Before Execution
5
5555
6789
5
AA55
6789
B2
B1
B0
B2
B1
B0
FF00
8000
FF00
8000
Y1
Y0
Y1
Y0
SR
SR
030F
0309
Explanation of Example:
Prior to execution, the 16-bit Y1 register contains the value $FF00, and the 36-bit B accumulator contains the value $5:5555:6789. The EOR.W Y1,B instruction performs a logical exclusive OR operation on the 16-bit value in the Y1 register with bits 31–16 of the B accumulator (B1) and stores the
36-bit result in the B accumulator. The lower word of the accumulator (B0) and the extension byte (B2)
are not affected by the operation.
Condition Codes Affected:
MR
N
Z
V
A-136
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 31 of accumulator result or MSB of register result is set
— Set if bits 31–16 of accumulator result or all bits of register result are zero
— Always cleared
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
EOR.W
EOR.W
Logical Exclusive OR Word
Instruction Fields:
Operation
Operands
C
W
EOR.W
EEE,EEE
1
1
Comments
16-bit exclusive OR (XOR)
Instruction Opcodes:
15
EOR.W EEE,EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
1
0
E
Instruction Set Details
8
7
E
E
a
a
4
3
a
1
0
0
1
0
A-137
EORC
EORC
Logical Exclusive OR Immediate
Operation:
Assembler Syntax:
#xxxx ⊕ X:<ea> → X:<ea>(no parallel move)
#xxxx ⊕ D → D(no parallel move)
EORC
EORC
#iiii,X:<ea>
#iiii,D
(no parallel move)
(no parallel move)
where ⊕ denotes the logical exclusive OR operator
Implementation Note:
This instruction is implemented by the assembler as an alias to the BFCHG instruction, and it uses the
16-bit immediate value as the bit mask. This instruction will dis-assemble as a BFCHG instruction.
Description: Perform a logical exclusive OR operation on a 16-bit immediate data value with the destination operand (D), and store the results back into the destination. C is also modified as described in “Condition
Codes Affected.” This instruction performs a read-modify-write operation on the destination and requires two destination accesses.
Example:
EORC
#$0FF0,X:$5000; Exclusive OR with immediate data
Before Execution
After Execution
X:$5000
5555
X:$5000
5AA5
SR
0300
SR
0300
Explanation of Example:
Prior to execution, the 16-bit X memory location X:$5000 contains the value $5555. Execution of the
instruction tests the state of bits 4–11 in X:$5000, does not set C (because all of the selected bits were
not set), and then complements the bits.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
For destination operand SR:
All SR bits except bits 14–10 are updated with values from the bitfield unit.
Bits 14–10 of the mask operand must be cleared.
For other destination operands:
L — Set if data limiting occurred during 36-bit source move
C — Set if all bits specified by the mask are set
Cleared if at least 1 bit specified by the mask is not set
Note:
If all bits in the mask are cleared, the instruction executes two NOPs and sets the C bit.
Instruction Fields:
Refer to the section on the BFCHG instruction for legal operand and timing information.
A-138
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
FRTID
FRTID
Delayed Return from Fast Interrupt
Operation:
Assembler Syntax:
Swap shadow registers, then
return from fast interrupt service routine
FRTID
Description: Refer to Section 9.3.2.2, “Fast Interrupt Processing,” on page 9-6.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
All bits are set according to the value removed from the stack
Restrictions:
Refer to Section 4.3.2, “Delayed Instruction Restrictions,” on page 4-14.
Instruction Fields:
Operation
Operands
FRTID
C
W
2
1
Comments
Delayed return from interrupt, restoring 21-bit PC and SR from
the stack; must fill 2 word slots
Instruction Opcodes:
15
FRTID
1
Timing:
2 oscillator clock cycles
Memory:
1 program word
Freescale Semiconductor
1
1
12
11
0
0
1
1
Instruction Set Details
8
7
1
0
0
0
4
3
1
1
0
0
1
0
A-139
ILLEGAL
ILLEGAL
Illegal Instruction Interrupt
Operation:
Assembler Syntax:
Begin illegal instruction exception routine
ILLEGAL
(no parallel move)
Description: Normal instruction execution is suspended, and illegal instruction exception processing is initiated.
The interrupt priority level bits (I1 and I0) are set to 11 in the status register. The purpose of the illegal
interrupt is to force the DSC into an illegal instruction exception for test purposes. Executing an
ILLEGAL instruction is a fatal error; the exception routine should indicate this condition and cause the
system to be re-started.
If the ILLEGAL instruction is in a DO loop at the LA and the instruction at the LA – 1 is being interrupted, then LC will be decremented twice. This situation is due to the same mechanism that causes
LC to be decremented twice if JSR, REP, and so on are located at the LA.
Since REP is uninterruptable, the result of repeating an ILLEGAL instruction is that the interrupt is
not taken until after the REP completes. After servicing the interrupt, program control returns to the
address of the second word that follows the ILLEGAL instruction. Of course, the ILLEGAL interrupt
service routine should abort further processing, and the processor should be re-initialized.
Usage:
The ILLEGAL instruction provides a means for testing the interrupt service routine that is executed
when an illegal instruction is encountered. This capability allows a user to verify that the interrupt service routine can correctly recover from an illegal instruction and re-start the application. The
ILLEGAL instruction is not used in normal programming.
Example:
ILLEGAL
Explanation of Example: See the description.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
ILLEGAL
C
W
Comments
4
1
Execute the illegal instruction exception. This instruction is made available so that code can be written to test and verify interrupt handlers for
illegal instructions.
Instruction Opcodes:
15
ILLEGAL
1
Timing:
4 oscillator clock cycles
Memory:
1 program word
A-140
1
1
12
11
0
0
1
1
8
7
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
4
3
0
1
0
0
1
0
Freescale Semiconductor
IMAC.L
IMAC.L
Integer Multiply with Accumulate Long
Operation:
Assembler Syntax:
D + (S1 × S2) → D
(no parallel move)
IMAC.L
S1,S2,D
(no parallel move)
Description: Multiply the two signed 16-bit source operands, and add the 32-bit integer product to the destination
(D). Both source operands must be located in the FF1 portion of an accumulator. The destination for
this instruction can be an accumulator or the Y register. If an accumulator is used as the destination,
the product is first sign extended from bit 31 and a 36-bit addition is then performed. The result is not
affected by the state of the saturation bit (SA).
Example:
IMAC.L A1,B1,Y
After Execution
Before Execution
0
0002
FFFF
0
0002
FFFF
A2
A1
A0
A2
A1
A0
0
0004
1234
0
0004
1234
B2
B1
B0
B2
B1
B0
0000
0002
0000
000A
Y1
Y0
Y1
Y0
SR
SR
0300
0310
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:0002:FFFF, the B accumulator contains
$0:0004:1234, and the 32-bit Y register contains $0000:0002. Execution of the IMAC.L instruction
multiplies the 16-bit signed value in A1 by the 16-bit signed value in B1, adds the resulting sign-extended product to the 32-bit Y register, and stores the 32-bit signed result ($0000:000A) into Y.
Condition Codes Affected:
MR
L
E
U
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extension portion of the result is in use
Set if the result is unnormalized
Set if bit 35 (or 31) of the result is set
Set if the result is zero
Set if overflow has occurred in result
Condition codes are calculated based on the 36-bit result if the destination is an accumulator, and on
the 32-bit result if the destination is the Y register.
Freescale Semiconductor
Instruction Set Details
A-141
IMAC.L
IMAC.L
Integer Multiply with Accumulate Long
Instruction Fields:
Operation
Operands
C
W
IMAC.L
FFF1,FFF1,fff
1
1
Comments
Integer 16 × 16 multiply-accumulate with 32-bit result
Instruction Opcodes:
15
IMAC.L FFF1,FFF1,fff
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-142
0
1
1
12
11
0
1
1
f
8
7
f
f
J
DSP56800E and DSP56800EX Core Reference Manual
J
4
3
J
J
0
J
0
0
Freescale Semiconductor
IMACUS
IMACUS
Integer MAC Unsigned and Signed
Operation:
Assembler Syntax:
D + (S1 × S2) → D
(S1 unsigned; S2 signed)
IMACUS
S1,S2,D
(no parallel move)
Description: Multiply one unsigned 16-bit source operand by one signed 16-bit operand, and add the 32-bit integer
product to the destination (D). The order of the registers is important. The first source register (S1)
must contain the unsigned value, and the second source (S2) must contain the signed value to produce
the correct integer multiplication. The destination for this instruction is always the Y register. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used to perform extended-precision multiplication calculations. It provides a method for calculating one of the intermediate values that is needed when a 32-bit × 32-bit multiplication
is performed, for example. See Section 5.5.3, “Multi-Precision Integer Multiplication,” on page 5-32
for an example that uses the IMACUS instruction.
Example:
IMACUS A0,B1,Y
; multiply unsigned A0 and signed B1,; add to Y
After Execution
Before Execution
0
FFFF
0002
0
FFFF
0002
A2
A1
A0
A2
A1
A0
0
FFFE
1234
0
FFFE
1234
B2
B1
B0
B2
B1
B0
0000
0004
0000
0000
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:FFFF:0002, the B accumulator contains
$0:FFFE:1234, and the 32-bit Y register contains $0000:0004. Execution of the IMACUS instruction
multiplies the 16-bit unsigned value in A0 by the 16-bit signed value in B1, adds the resulting 32-bit
product to the 32-bit Y register, and stores the result ($0000:0000) into Y.
Condition Codes Affected:
The condition codes are not modified by this instruction.
Instruction Fields:
Operation
IMACUS
Operands
A0,A1,Y
A0,B1,Y
A0,C1,Y
A0,D1,Y
B0,C1,Y
B0,D1,Y
C0,C1,Y
C0,D1,Y
C
W
1
1
Comments
Integer 16 × 16 multiply-accumulate:
F0 (unsigned) × F1 (signed)
Instruction Opcodes:
15
IMACUS q1.l,q2.h,Y
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
0
1
Instruction Set Details
8
7
0
1
q
q
4
3
q
0
0
1
1
1
A-143
IMACUU
IMACUU
Integer MAC Two Unsigned Values
Operation:
Assembler Syntax:
D + (S1 × S2) → D
(S1 unsigned; S2 unsigned)
IMACUU
S1,S2,D
(no parallel move)
Description: Multiply the two unsigned 16-bit source operands (S1 and S2), and add the 32-bit integer product to
the destination (D). The destination for this instruction is always the Y register. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used to perform extended-precision multiplication calculations. It provides a method for calculating one of the intermediate values that is needed when a 32-bit × 32-bit multiplication
is performed, for example. See Section 5.5.3, “Multi-Precision Integer Multiplication,” on page 5-32
for an example that uses the IMACUU instruction.
Example:
IMACUU A0,B1,Y
; multiply unsigned in A0 and B1, add to Y
After Execution
Before Execution
0
FFFF
0002
0
FFFF
0002
A2
A1
A0
A2
A1
A0
0
FFFE
1234
0
FFFE
1234
B2
B1
B0
B2
B1
B0
0000
0004
0002
0000
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:FFFF:0002, the B accumulator contains
$0:FFFE:1234, and the 32-bit Y register contains $0000:0004. Execution of the IMACUU instruction
multiplies the 16-bit unsigned value in A0 by the 16-bit unsigned value in B1, adds the resulting 32-bit
product to the 32-bit Y register, and stores the 32-bit unsigned result ($0002:0000) into Y.
Condition Codes Affected:
The condition codes are not modified by this instruction.
Instruction Fields:
Operation
Operands
C
W
IMACUU
A0,A1,Y
A0,B1,Y
A0,C1,Y
A0,D1,Y
B0,C1,Y
B0,D1,Y
C0,C1,Y
C0,D1,Y
1
1
Comments
Integer 16 × 16 multiply-accumulate:
F0 (unsigned) × F1 (unsigned)
Instruction Opcodes:
15
IMACUU q1.l,q2.h,Y
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-144
0
1
1
12
11
1
0
0
1
8
7
1
1
q
DSP56800E and DSP56800EX Core Reference Manual
q
4
3
q
0
0
1
1
1
Freescale Semiconductor
IMPY.L
IMPY.L
Integer Multiply Long
Operation:
Assembler Syntax:
S1 × S2 → D
(no parallel move)
IMPY.L
S1,S2,D
(no parallel move)
Description: Multiply the two signed 16-bit source operands, and place the 32-bit product in the destination (D).
Both source operands must be located in the FF1 portion of an accumulator or in X0, Y0, or Y1. The
destination for this instruction can be an accumulator or the Y register. If an accumulator is used for
the destination, the result is sign extended from bit 31 into the extension portion (FF2) of the accumulator. The result is not affected by the state of the saturation bit (SA).
Example:
IMPY.L A1,B1,Y
; integer mult with 32-bit result
After Execution
Before Execution
0
0002
FFFF
0
0002
FFFF
A2
A1
A0
A2
A1
A0
0
FFFE
1234
0
FFFE
1234
B2
B1
B0
B2
B1
B0
0001
37A2
FFFF
FFFC
Y1
Y0
Y1
Y0
SR
SR
0300
0318
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:0002:FFFF, the B accumulator contains
$0:FFFE:1234, and the 32-bit Y register contains $0001:37A2. Execution of the IMPY.L instruction
multiplies the 16-bit (signed) positive value in A1 by the (signed) negative 16-bit value in B1, and
stores the (signed) 32-bit negative result ($FFFF:FFFC) into Y. The negative bit is set to indicate the
sign of the result.
Condition Codes Affected:
MR
L
E
U
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extension portion of the result is in use
Set if the result is unnormalized
Set if bit 35 (or 31) of the result is set
Set if the result is zero
Set if overflow has occurred in result
Condition codes are calculated based on the 36-bit result if the destination is an accumulator, and on
the 32-bit result if the destination is the Y register.
Freescale Semiconductor
Instruction Set Details
A-145
IMPY.L
IMPY.L
Integer Multiply Long
Instruction Fields:
Operation
Operands
C
W
IMPY.L
FFF1,FFF1,fff
1
1
Comments
Integer 16 × 16 multiply with 32-bit result
Instruction Opcodes:
15
IMPY.L FFF1,FFF1,fff
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-146
0
1
1
12
11
0
1
1
f
8
7
f
f
J
DSP56800E and DSP56800EX Core Reference Manual
J
4
3
J
J
0
J
0
1
Freescale Semiconductor
IMPY.W
IMPY.W
Integer Multiply Word
Operation:
Assembler Syntax:
S1 × S2 → D
(no parallel move)
IMPY.W
S1,S2,D
(no parallel move)
Description: Perform an integer multiplication on the two 16-bit, signed, integer source operands (S1 and S2), and
store the lowest 16 bits of the integer product in the destination (D). If the destination is an accumulator, the product is stored in the MSP with sign extension while the LSP remains unchanged. The order
of the first two operands is not important. The V bit is set if the calculated integer product does not fit
into 16 bits. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is useful in general computing when it is necessary to multiply two integers and the
nature of the computation can guarantee that the result fits in a 16-bit destination. In this case, it is better to place the result in the MSP (FF1 portion) of an accumulator because more instructions have access to this portion than to the other portions of the accumulator.
Example:
IMPY.W A1,Y0,A
; integer 16-bit multiplication
After Execution
Before Execution
4
0002
1234
F
FFFC
1234
A2
A1
A0
A2
A1
A0
2000
FFFE
2000
FFFE
Y1
Y0
Y1
Y0
SR
SR
0300
0308
Explanation of Example:
Prior to execution, the A accumulator contains the value $4:0002:1234, and the data ALU register Y0
contains the 16-bit (signed) negative integer value $FFFE. Execution of the IMPY.W instruction integer multiplies the (signed) positive value in A1 and the (signed) negative value in Y0, and stores the
(signed) negative result ($FFFC) in A1. A0 remains unchanged, and A2 is sign extended. The negative
bit is set to indicate the sign of the result.
Condition Codes Affected:
MR
L
N
Z
V
Note:
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
Set if overflow has occurred in the 16-bit result
Set if bit 15 of the result is set
Set if the 16-bit result or 20 MSBs of a destination accumulator equal zero
Set if overflow occurs in the 16-bit result
A 31-bit integer product is calculated for this instruction, while the lowest 16 bits are stored in the destination register. When SA or CM are set, the N bit is set to the value in bit 30 of the internally computed result. When SA and CM are zero, the N bit is set to the value in bit 15 of the result.
Freescale Semiconductor
Instruction Set Details
A-147
IMPY.W
IMPY.W
Integer Multiply Word
Instruction Fields:
Operation
Operands
C
W
IMPY.W
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
1
1
Comments
Integer 16 × 16 multiply with 16-bit result.
When the destination is the Y register or an accumulator, the LSP portion is unchanged by the instruction.
Note: Assembler also accepts the first two operands
when they are specified in the opposite order.
Instruction Opcodes:
15
IMPY.W Q1,Q2,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-148
0
1
1
12
11
1
0
0
F
8
7
F
F
Q
DSP56800E and DSP56800EX Core Reference Manual
Q
4
3
Q
1
0
0
1
0
Freescale Semiconductor
IMPYSU
IMPYSU
Integer Multiply Signed and Unsigned
Operation:
Assembler Syntax:
S1 × S2 → D
(S1 signed; S2 unsigned)
IMPYSU
S1,S2,D
(no parallel move)
Description: Multiply one signed 16-bit source operand by one unsigned 16-bit operand, and place the 32-bit integer
product in the destination (D). The order of the registers is important. The first source register (S1)
must contain the signed value, and the second source (S2) must contain the unsigned value to produce
the correct integer multiplication. The destination for this instruction is always the Y register. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used to perform extended-precision multiplication calculations. It provides a method for calculating one of the intermediate values that is needed when a 32-bit × 32-bit multiplication
is performed, for example. See Section 5.5.3, “Multi-Precision Integer Multiplication,” on page 5-32
for an example that uses the IMPYSU instruction.
Example:
IMPYSU A1,B0,Y
; multiply signed A1 to unsigned B0, store in Y
After Execution
Before Execution
0
FFFE
1234
0
FFFE
1234
A2
A1
A0
A2
A1
A0
0
0000
0002
0
0000
0002
B2
B1
B0
B2
B1
B0
1234
5678
FFFF
FFFC
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:FFFE:1234, the B accumulator contains
$0:0000:0002, and the 32-bit Y register contains $1234:5678. Execution of the IMPYSU instruction
multiplies the 16-bit (signed) negative value in A1 by the 16-bit (unsigned) positive value in B0 and
stores the (signed) negative result ($FFFF:FFFC) into Y.
Condition Codes Affected:
The condition codes are not modified by this instruction.
Instruction Fields:
Operation
IMPYSU
Operands
A1,A0,Y
A1,B0,Y
A1,C0,Y
A1,D0,Y
B1,C0,Y
B1,D0,Y
C1,C0,Y
C1,D0,Y
C
W
1
1
Comments
Integer 16 × 16 multiply:
F1 (signed) × F0 (unsigned)
Instruction Opcodes:
15
IMPYSU q1.h,q2.l,Y
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
0
1
Instruction Set Details
8
7
0
0
q
q
4
3
q
0
0
1
1
1
A-149
IMPYUU
Unsigned Integer Multiply
Operation:
IMPYUU
Assembler Syntax:
S1 × S2 → D
(S1 unsigned; S2 unsigned)
IMPYUU
S1,S2,D
(no parallel move)
Description: Multiply the two unsigned 16-bit source operands (S1 and S2), and place the 32-bit product in the destination (D). If the destination is an accumulator, the 32-bit product is stored in the MSP:LSP with zeros propagated in the extension portion (FF2) of the accumulator. The result is not affected by the state
of the saturation bit (SA).
Usage:
This instruction is used to perform extended-precision multiplication calculations. It provides a method for calculating one of the intermediate values that is needed when a 32-bit × 32-bit multiplication
is performed, for example. See Section 5.5.3, “Multi-Precision Integer Multiplication,” on page 5-32
for an example that uses the IMPYUU instruction.
Example:
IMPYUU A1,B0,Y
; multiply two unsigned integers, store in Y
After Execution
Before Execution
0
FFFE
1234
0
FFFE
1234
A2
A1
A0
A2
A1
A0
0
0000
0002
0
0000
0002
B2
B1
B0
B2
B1
B0
1234
5678
0001
FFFC
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the A accumulator contains the value $0:FFFE:1234, the B accumulator contains
$0:0000:0002, and the 32-bit Y register contains $1234:5678. Execution of the IMPYUU instruction
multiplies the 16-bit (positive) unsigned value in A1 by the 16-bit unsigned value in B0 and stores the
unsigned result ($0001:FFFC) into Y.
Condition Codes Affected:
The condition codes are not modified by this instruction.
A-150
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
IMPYUU
IMPYUU
Unsigned Integer Multiply
Instruction Fields:
Operation
Operands
C
W
Comments
IMPYUU
A1,A0,Y
A1,B0,Y
A1,C0,Y
A1,D0,Y
B1,C0,Y
B1,D0,Y
C1,C0,Y
C1,D0,Y
1
1
Integer 16 × 16 multiply:
F1 (unsigned) × F0 (unsigned)
A0,A0,FF
A0,B0,FF
A0,C0,FF
A0,D0,FF
B0,C0,FF
B0,D0,FF
C0,C0,FF
C0,D0,FF
1
1
Integer 16 × 16 multiply:
F0 (unsigned) × F0 (unsigned)
Instruction Opcodes:
15
IMPYUU q1.l,q2.l,FF
0
1
1
15
IMPYUU q1.h,q2.l,Y
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
12
11
1
0
0
0
0
1
Instruction Set Details
8
7
F
F
8
7
1
0
q
q
q
q
4
3
q
0
4
3
q
0
0
1
1
1
0
1
1
1
A-151
INC.BP
INC.BP
Increment Byte (Byte Pointer)
Operation:
Assembler Syntax:
D+1→D
(no parallel move)
INC.BP
D
(no parallel move)
Description: Increment a byte value in memory. The value is internally sign extended to 20 bits before being incremented. The low-order 8 bits of the result are stored back to memory. The condition codes are calculated based on the 8-bit result, with the exception of the E and U bits, which are calculated based on
the 20-bit result. Absolute addresses are expressed as byte addresses. The result is not affected by the
state of the saturation bit (SA).
Usage:
This instruction is typically used when integer data is processed.
Example:
INC.BP X:$3065
; increment the byte at (byte) address $3065
Before Execution
After Execution
Byte
Addresses
Byte
Addresses
X Memory
07
7
0
$3068
$3066
$3064
$3062
88
66
00
77
55
33
22
11
SR
X Memory
07
0
7
$3068
$3066
$3064
$3062
SR
0300
88
66
01
77
55
33
22
11
0310
Explanation of Example:
Prior to execution, the value at byte address X:$3065 is $00. Execution of the INC.BP instruction increments this value by one and generates the result $01. Note that this address is equivalent to the upper
byte of word address $1832.
Condition Codes Affected:
MR
E
U
N
Z
V
C
A-152
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if bit 7 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a carry occurs from bit 7 of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
INC.BP
INC.BP
Increment Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
INC.BP
X:xxxx
3
2
X:xxxxxx
4
3
Comments
Increment byte in memory
Instruction Opcodes:
15
INC.BP X:xxxx
0
1
0
12
11
0
1
1
1
8
7
0
1
1
0
4
3
0
0
4
3
0
1
1
0
AAAAAAAAAAAAAAAA
15
INC.BP X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
0
1
1
0
0
0
1
1
0
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-153
INC.L
INC.L
Increment Long
Operation:
Assembler Syntax:
D+1→D
(no parallel move)
INC.L
D
(no parallel move)
Description: Increment a long-word value in a register or memory. When an operand located in memory is operated
on, the low-order 32 bits of the result are stored back to memory. The condition codes are calculated
based on the 32-bit result. Absolute addresses pointing to long elements must always be even aligned
(that is, pointing to the lowest 16 bits).
Usage:
This instruction is typically used when integer data is processed.
Example:
INC.L
A
; increment value in A by one
After Execution
Before Execution
0
0020
0000
0
0020
0001
A2
A1
A0
A2
A1
A0
SR
SR
0300
0310
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:0020:0000. Execution of the INC.L
instruction adds one to the A accumulator. The CCR is updated based on the result of the addition.
Condition Codes Affected:
MR
E
U
N
Z
V
C
A-154
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 36-bit result is in use
Set if the 36-bit result is unnormalized
Set if bit 31 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a carry occurs from bit 31 of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
INC.L
INC.L
Increment Long
Instruction Fields:
Operation
Operands
C
W
Comments
INC.L
fff
1
1
Increment long
X:xxxx
3
2
Increment long in memory
X:xxxxxx
4
3
Instruction Opcodes:
15
INC.L X:xxxx
0
1
0
12
11
0
1
1
1
8
7
0
1
1
0
4
3
0
0
4
3
0
1
1
1
AAAAAAAAAAAAAAAA
15
INC.L X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
0
1
1
0
0
0
1
1
1
0
1
1
AAAAAAAAAAAAAAAA
15
INC.L fff
0
Timing:
1–4 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
1
1
12
11
1
0
0
f
Instruction Set Details
8
7
f
f
0
0
4
3
1
0
0
A-155
INC.W
INC.W
Increment Word
Operation:
Assembler Syntax:
D+1→D
D+1→D
(one parallel move)
(no parallel move)
INC.W
INC.W
D
D
(one parallel move)
(no parallel move)
Description: Increment a 16-bit destination by one. If the destination is an accumulator, only the EXT and MSP portions of the accumulator are used and the LSP remain unchanged. The condition codes are calculated
based on the 16-bit result (or on the 20-bit result for accumulators).
Usage:
This instruction is typically used when integer data is processed.
Example:
INC.W
A
X:(R0)+,X0
; Increment the 20 MSBs of A and
; update X0 and R0
Before Execution
After Execution
0
FFFF
0033
1
0000
0033
A2
A1
A0
A2
A1
A0
SR
SR
0300
0330
Explanation of Example:
Prior to execution, the 36-bit A accumulator contains the value $0:FFFF:0033. Execution of the
INC.W instruction increments by one the upper 20 bits of the A accumulator and sets the E and U bits
in the CCR. A new value is read in parallel and stored in register X0; the address register R0 is post-incremented.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
Note:
A-156
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of the result is in use
Set if result is unnormalized
Set if MSB of the result is set
Set if the result is zero (20 MSB for accumulator destinations)
Set if overflow has occurred in result
Set if a carry (or borrow) occurs from bit 15 of the result (bit 35 for accumulators)
When the destination is one of the four accumulators, condition code calculations follow the rules for
20-bit arithmetic; otherwise, the rules for 16-bit arithmetic apply.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
INC.W
INC.W
Increment Word
Instruction Fields:
Operation
Operands
C
W
Comments
INC.W
EEE
1
1
Increment word.
X:(Rn)
3
1
X:(Rn+xxxx)
4
2
Increment word in memory using appropriate addressing mode.
X:(SP–xx)
4
1
X:xxxx
3
2
X:xxxxxx
4
3
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
INC.W2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-157
INC.W
INC.W
Increment Word
Instruction Opcodes:
15
INC.W EEE
0
1
1
15
INC.W F GGG,X:<ea_m>
0
0
0
15
INC.W F X:<ea_m>,GGG
0
0
1
1
0
1
0
15
INC.W X:(Rn)
0
15
INC.W X:(Rn+xxxx)
0
12
11
1
0
12
11
0
0
12
11
0
0
12
11
0
1
12
11
0
1
0
E
G
G
8
7
E
E
8
7
G
F
8
7
F
G
G
G
8
7
1
1
0
1
8
7
1
1
0
1
0
0
0
1
4
3
0
0
4
3
1
0
4
3
0
0
1
1
4
3
1
0
1
R
4
3
1
0
1
R
0
0
1
1
0
m
R
R
0
m
R
R
1
R
R
0
R
R
a
a
a
1
0
0
0
0
AAAAAAAAAAAAAAAA
15
INC.W X:(SP–xx)
0
1
0
1
0
15
INC.W X:xxxx
0
12
11
0
1
12
11
0
1
8
7
0
1
0
1
8
7
1
1
0
1
4
3
1
a
a
a
4
3
1
0
0
0
0
0
AAAAAAAAAAAAAAAA
15
INC.W X:xxxxxx
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
0
1
0
0
1
1
1
0
1
1
0
0
0
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–4 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-158
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
Jcc
Jcc
Jump Conditionally
Operation:
Assembler Syntax:
If (cc), then S
→ PC
else
PC + 1 → PC
Jcc
S {<ABS19> or <ABS21>}
Description: If the specified condition is true, program execution continues at the effective address specified in the
instruction. If the specified condition is false, the PC is incremented and program execution continues
sequentially. The effective address is a 19- or 21-bit absolute address.
The term “cc” specifies the following:
“cc” Mnemonic
Condition
CC (HS*)— carry clear (higher or same)
C=0
CS (LO*)— carry set (lower)
C=1
EQ— equal
Z=1
GE— greater than or equal
N⊕V=0
GT— greater than
Z + (N ⊕ V) = 0
LE— less than or equal
Z + (N ⊕ V) = 1
LT— less than
N⊕V=1
NE— not equal
Z=0
NN— not normalized
Z + (U • E) = 0
NR— normalized
Z + (U • E) = 1
* Only available when CM bit is set in the OMR
Xdenotes the logical complement of X
+denotes the logical OR operator
•denotes the logical AND operator
⊕denotes the logical exclusive OR operator
Example:
JCS
INC.W
INC.W
LABEL ; jump to LABEL if carry bit is set
A
A
ADD
B,A
LABEL
Explanation of Example:
In this example, if C is one when the JCS instruction is executed, program execution skips the two
INC.W instructions and continues with the ADD instruction. If the specified condition is not true, no
jump is taken, the program counter is incremented by one, and program execution continues with the
first INC.W instruction. The Jcc instruction uses a 19-bit absolute address for this example.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Condition Codes Affected:
The condition codes are tested but not modified by this instruction.
Freescale Semiconductor
Instruction Set Details
A-159
Jcc
Jcc
Jump Conditionally
Instruction Fields:
Operation
Operands
C1
W
Jcc
<ABS19>
5 or 4
2
19-bit absolute address
<ABS21>
6 or 5
3
21-bit absolute address
Comments
1.The clock-cycle count depends on whether the jump is taken. The first value applies if the jump is taken, and
the second applies if it is not.
Instruction Opcodes:
15
Jcc
<ABS21>
12
11
8
7
4
3
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
C
C
C
0
1
0
1
0
C
0
0
C
A
A
AAAAAAAAAAAAAAAA
15
Jcc
<ABS19>
1
1
1
12
11
0
0
C
C
8
7
C
0
1
0
4
3
1
A
0
AAAAAAAAAAAAAAAA
Timing:
4–6 oscillator clock cycles
Memory:
2–3 program words
A-160
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
JMP
JMP
Unconditional Jump
Operation:
Assembler Syntax:
S → PC
JMP
S {(N) or <ABS19> or <ABS21>}
Description: Jump to program memory at the location given by the instruction’s effective address, which can be the
value in the N register or a 19- or 21- bit absolute address.
Example:
JMP
LABEL
Explanation of Example:
In this example, program execution is transferred to the address represented by LABEL. The DSC core
supports up to 21-bit program addresses.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Instruction Fields:
Operation
Operands
C
W
Comments
JMP
(N)
5
1
Jump to target contained in N register
<ABS19>
4
2
19-bit absolute address
<ABS21>
5
3
21-bit absolute address
Instruction Opcodes:
15
JMP
<ABS21>
12
11
8
7
4
3
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
0
0
0
0
1
0
1
0
1
0
1
0
0
1
1
1
1
A
A
AAAAAAAAAAAAAAAA
15
JMP
(N)
1
JMP
<ABS19>
1
1
1
1
15
1
12
11
0
0
12
11
0
0
8
7
1
1
1
0
8
7
0
0
1
0
4
3
0
0
1
1
4
3
1
0
1
A
0
0
AAAAAAAAAAAAAAAA
Timing:
4–5 oscillator clock cycles
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-161
JMPD
JMPD
Delayed Unconditional Jump
Operation:
Assembler Syntax:
Execute instructions in next 2 words
S→ PC
JMPD
S {<ABS19> or <ABS21>}
Description: Jump to program memory at the location that is given by the instruction’s effective address, but execute
the following 2 words of instructions before completing the jump. That is, execute the next two 1-word
instructions or the next single 2-word instruction following the JMPD instruction before jumping to
the destination address.
Example:
JMPD
ADD.W
NOP
...
LABEL
#1,X0
; delayed JMP to label
; first delay slot
; second delay slot (unused)
LABEL
; JMP target address
Explanation of Example:
In this example, program execution is transferred to the address represented by LABEL after the two
1-word instructions following the JMPD instruction are executed.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Refer to Section 4.3.2, “Delayed Instruction Restrictions,” on page 4-14.
Instruction Fields:
Operation
Operands
C
W
Comments
JMPD
<ABS19>
2
2
Delayed jump with 19-bit absolute address; must fill 2 delay slots
<ABS21>
3
3
Delayed jump with 21-bit absolute address; must fill 2 delay slots
Instruction Opcodes:
15
JMPD
<ABS21>
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
0
0
0
1
1
0
1
0
1
0
1
0
0
4
3
1
A
AAAAAAAAAAAAAAAA
15
JMPD
<ABS19>
1
1
1
12
11
0
0
0
1
8
7
1
0
1
0
0
1
A
A
AAAAAAAAAAAAAAAA
Timing:
2–3 oscillator clock cycles
Memory:
2–3 program words
A-162
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
JSR
Operation:
Assembler Syntax:
→ SP
→ X:(SP)
→ SP
→ X:(SP)
→ PC
SP + 1
PC
SP + 1
SR
S
JSR
Jump to Subroutine
JSR
S {(RRR) or <ABS19> or <ABS21>}
Description: Jump to subroutine in program memory located at the effective address specified by the operand. The
operand can be a 19- or 21-bit absolute address or a register.
Example:
JSR
LABEL
; jump to absolute address indicated by “LABEL”
Explanation of Example:
In this example, program execution is transferred to the subroutine at the address that is represented by
LABEL. The DSC core supports program addresses up to 21 bits wide.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Restrictions:
Refer to Section 10.4, “Pipeline Dependencies and Interlocks,” on page 10-26.
Instruction Fields:
Operation
Operands
C
W
Comments
JSR
(RRR)
5
1
Push 21-bit return address and jump to target address contained
in the RRR register
<ABS19>
4
2
Push 21-bit return address and jump to 19-bit target address
<ABS21>
5
3
Push 21-bit return address and jump to 21-bit target address
Instruction Opcodes:
15
JSR
<ABS21>
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
0
0
0
1
0
0
1
0
1
0
1
0
0
4
3
1
N
4
3
1
A
AAAAAAAAAAAAAAAA
15
JSR
(RRR)
1
1
1
15
JSR
<ABS19>
1
1
1
12
11
0
0
12
11
0
0
1
0
1
1
8
7
0
0
8
7
0
0
0
1
0
0
0
1
N
N
0
1
A
A
AAAAAAAAAAAAAAAA
Timing:
4–5 oscillator clock cycles
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-163
LSL.W
LSL.W
Logical Shift Left Word
Operation:
Assembler Syntax:
(see following figure)
LSL.W
C
D
Unch.
(no parallel move)
0
Unchanged
D2
D1
D0
Description: Logically shift 16 bits of the destination operand (D) by 1 bit to the left, and store the result in the destination. If the destination is a 36-bit accumulator, the result is stored in the MSP of the accumulator
(FF1 portion), and the remaining portions of the accumulator are not modified. The MSB of the destination (bit 31 if the destination is a 36-bit accumulator) prior to the execution of the instruction is shifted into C, and zero is shifted into the LSB of D1 (bit 16 if the destination is a 36-bit accumulator). The
result is not affected by the state of the saturation bit (SA).
Example:
LSL.W
B
; multiply B1 by 2
Before Execution
After Execution
6
C555
00AA
6
8AAA
00AA
B2
B1
B0
B2
B1
B0
SR
SR
0302
0309
Explanation of Example:
Prior to execution, the 36-bit B accumulator contains the value $6:C555:00AA. Execution of the
LSL.W instruction shifts the 16-bit value in the B1 register by 1 bit to the left and stores the result back
in the B1 register. The C bit is set because bit 31 of B1 was set prior to the execution of the instruction.
The N bit is also set because bit 31 of accumulator B is set. The overflow bit V is always cleared.
Condition Codes Affected:
MR
N
Z
V
C
A-164
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
Set if bit 31 of an accumulator result or bit 15 of a 16-bit register result is set
Set if the MSP of result or all bits of a 16-register result are zero
Always cleared
Set if bit 31 of accumulator or bit 15 of a 16-bit register was set prior to the execution
of the instruction
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
LSL.W
LSL.W
Logical Shift Left Word
Instruction Fields:
Operation
Operands
C
W
LSL.W
EEE
1
1
Comments
1-bit logical shift left word
Instruction Opcodes:
15
LSL.W EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
0
E
Instruction Set Details
8
7
E
E
1
1
4
3
1
0
0
0
1
1
A-165
LSR.W
LSR.W
Logical Shift Right Word
Operation:
Assembler Syntax:
(see following figure)
LSR.W
D
(no parallel move)
0
Unch.
C
Unchanged
D2
D1
D0
Description: Logically shift 16 bits of the destination operand (D) by 1 bit to the right, and store the result in the
destination. If the destination is a 36-bit accumulator, the result is stored in the MSP of the accumulator
(FF1 portion), and the remaining portions of the accumulator are not modified. The LSB of the destination (bit 16 if the destination is a 36-bit accumulator) prior to the execution of the instruction is shifted into C, and zero is shifted into the MSB of D1 (bit 31 if the destination is a 36-bit accumulator). The
result is not affected by the state of the saturation bit (SA).
Example:
LSR.W
B
; divide B1 by 2 (B1 considered unsigned)
After Execution
Before Execution
F
0001
00AA
F
0000
00AA
B2
B1
B0
B2
B1
B0
SR
SR
0302
0305
Explanation of Example:
Prior to execution, the 36-bit B accumulator contains the value $F:0001:00AA. Execution of the
LSR.W instruction shifts the 16-bit value in the B1 register by 1 bit to the right and stores the result
back in the B1 register. C is set by the operation because bit 0 of B1 was set prior to the execution of
the instruction. The Z bit of CCR (bit 2) is also set because the result in B1 is zero. The overflow bit
(V) is always cleared.
Condition Codes Affected:
MR
N
Z
V
C
A-166
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
Always cleared
Set if the MSP of result or all bits of a 16-register result are zero
Always cleared
Set if bit 31 of accumulator or bit 15 of a 16-bit register was set prior to the execution
of the instruction
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
LSR.W
LSR.W
Logical Shift Right Word
Instruction Fields:
Operation
Operands
C
W
LSR.W
EEE
1
1
Comments
1-bit logical shift right word
Instruction Opcodes:
15
LSR.W EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
0
E
Instruction Set Details
8
7
E
E
1
1
4
3
1
1
0
0
1
1
A-167
LSR16
Operation:
S >> 16 →
LSR16
Logical Shift Right 16 Bits
Assembler Syntax:
D
(no parallel move)
LSR16
S,D
(no parallel move)
Description: Logically shift the source operand to the right by 16 bits, and store the result in the destination (D),
zero extending to the left. This operation effectively places the MSP of the source register into the LSP
of the destination register, propagating zero bits through the MSP and the extension register (for accumulator destinations). If the source is an accumulator, both the extension register and MSP are shifted.
When the destination operand is a 16-bit register, the MSP of an accumulator or Y register is written
to it. If both the source and destination are 16-bit registers, the destination is cleared. The result is not
affected by the state of the saturation bit (SA).
Usage:
This instruction can be used to cast an unsigned integer to a long value.
Example:
LSR16
Y,A;
; shift MSP of Y into A0
Before Execution
After Execution
0
3456
3456
0
0000
A1A2
A2
A1
A0
A2
A1
A0
A1A2
A3A4
A1A2
A3A4
Y1
Y0
Y1
Y0
Explanation of Example:
Prior to execution, the Y register contains the value to be shifted ($A1A2:A3A4). The contents of the
destination register are not important prior to execution because they have no effect on the calculated
value. The LSR16 instruction logically shifts the value $A1A2:A3A4 by 16 bits to the right, zero extends to a full 36 bits, and places the result in the destination register A.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
Comments
LSR16
FFF,FFF
1
1
Logical shift right the first operand by 16 bits, placing
result in the destination operand (new bits zeroed)
FFF
1
1
An alternate syntax for the preceding instruction if the
source and the destination are the same
Instruction Opcodes:
15
LSR16 FFF,FFF
0
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-168
1
1
12
11
1
1
1
F
8
7
F
F
b
DSP56800E and DSP56800EX Core Reference Manual
b
4
3
b
0
0
1
1
1
Freescale Semiconductor
LSRA
LSRA
Logical Shift Right AGU Register
Operation:
Assembler Syntax:
D >> 1 → D
(no parallel move)
LSRA
D
(no parallel move)
Description: Logically shift the address register operand 1 bit to the right, and store the result back in the register.
Example:
LSRA
R0
; logically shift R0 to the right 1 bit
Before Execution
R0
After Execution
R0
A0A0A0
505050
Explanation of Example:
Prior to execution, the R0 register contains $A0A0A0. Execution of the LSRA R0 instruction shifts
the value in the R0 register 1 bit to the right, and stores the result ($505050) back in R0.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Operands
C
W
LSRA
Rn
1
1
Comments
Logical shift right AGU register by 1 bit
Instruction Opcodes:
15
LSRA
Rn
1
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
0
12
11
0
0
1
0
Instruction Set Details
8
7
1
0
0
1
4
3
1
R
0
1
R
R
A-169
LSRAC
LSRAC
Logical Shift Right with Accumulate
Operation:
Assembler Syntax:
(S1 >> S2) + D →D (no parallel move)
LSRAC
S1,S2,D
(no parallel move)
Description: Logically shift the first 16-bit source operand (S1) to the right by the value contained in the lowest
4 bits of the second source operand (S2), and accumulate the result with the value in the destination
(D). Operand S1 is internally zero extended and concatenated with 16 zero bits to form a 36-bit value
before the shift operation. The result is not affected by the state of the saturation bit (SA).
Usage:
This instruction is used for multi-precision logical right shifts.
Example:
LSRAC
Y1,X0,A
; logical right shift Y1 by 4 and
; accumulate in A
Before Execution
After Execution
0
0000
0099
0
0C00
3099
A2
A1
A0
A2
A1
A0
C003
8000
C003
8000
Y1
Y0
Y1
Y0
X0
00F4
X0
00F4
SR
0300
SR
0300
Explanation of Example:
Prior to execution, the Y1 register contains the value to be shifted ($C003), the lowest 4 bits of the X0
register contain the amount by which to shift ($4), and the destination accumulator contains
$0:0000:0099. The LSRAC instruction logically shifts the value $C003 by 4 bits to the right and accumulates this result with the value that is already in accumulator A.
Condition Codes Affected:
MR
N
Z
Note:
A-170
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if bit 35 of accumulator result is set
— Set if accumulator result equals zero
If the SA bit is set, the N bit is equal to bit 31 of the result; if SA is cleared, N is equal to bit 35 of the
result.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
LSRAC
LSRAC
Logical Shift Right with Accumulate
Instruction Fields:
Operation
Operands
C
W
LSRAC
Y1,X0,FF
Y0,X0,FF
Y1,Y0,FF
Y0,Y0,FF
A1,Y0,FF
B1,Y1,FF
C1,Y0,FF
C1,Y1,FF
1
1
Comments
Logical word shift right with accumulation
Instruction Opcodes:
15
LSRAC Q1,Q2,FF
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
1
0
Instruction Set Details
8
7
F
F
Q
Q
4
3
Q
0
0
1
1
0
A-171
LSRR.L
LSRR.L
Multi-Bit Logical Right Shift Long
Operation:
Assembler Syntax:
If S[15] = 0 or S is not a register,
D >> S → D
(no parallel move)
Else
D << –S → D
(no parallel move)
LSRR.L
S,D
(no parallel move)
LSRR.L
S,D
(no parallel move)
Description: Logically shift the second operand to the right by the value contained in the 5 lowest bits of the first
operand (or by an immediate integer), and store the result back in the destination (D). The shift count
can be a 5-bit positive immediate integer or the value contained in X0, Y0, Y1, or the MSP of an accumulator. For 36- and 32-bit destinations, the MSP:LSP are shifted, with zero extension from bit 31
(the FF2 portion is ignored). If the shift count in a register is negative (bit 15 is set), the direction of
the shift is reversed. The result is not affected by the state of the saturation bit (SA).
Example:
LSRR.L Y1,A
; left shift 32-bit A10 by Y1
After Execution
Before Execution
F
F123
3456
0
0000
F123
A2
A1
A0
A2
A1
A0
0010
8000
0010
8000
Y1
Y0
Y1
Y0
SR
SR
0300
0300
Explanation of Example:
Prior to execution, the A accumulator contains the value to be shifted, $F:F123:3456, and the Y1 register contains the amount by which to shift ($10 = 16). The LSRR.L instruction logically shifts the destination accumulator 16 bits to the right and places the result back in A.
Condition Codes Affected:
MR
N
Z
Note:
A-172
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if the MSB of the result is set
— Set if the result equals zero
Condition code results are set according to the size of the destination operand.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
LSRR.L
LSRR.L
Multi-Bit Logical Right Shift Long
Instruction Fields:
Operation
Operands
C
W
Comments
LSRR.L
#<0–31>,fff
2
1
Logical shift right by a 5-bit positive immediate integer
EEE,FFF
2
1
Bi-directional logical shift destination by value in the first
operand: positive –> right shift
Instruction Opcodes:
15
LSRR.L #<0–31>,fff
0
1
0
1
1
15
LSRR.L EEE,FFF
Timing:
2 oscillator clock cycles
Memory:
1 program word
Freescale Semiconductor
0
12
11
0
1
12
11
1
1
8
7
f
1
f
f
8
7
1
F
F
F
Instruction Set Details
4
3
B
0
1
B
4
3
a
a
a
1
0
B
B
B
1
0
1
0
A-173
LSRR.W
Multi-Bit Logical Right Shift Word
Operation:
LSRR.W
Assembler Syntax:
D >> S → D
S1 >> S2 → D
(no parallel move)
(no parallel move)
LSRR.W
LSRR.W
S,D
S1,S2,D
(no parallel move)
(no parallel move)
Description: This instruction can have two or three operands. Logically shift the source operand S1 or D to the right
by the value contained in the lowest 4 bits of either S2 or S, respectively (or by an immediate integer),
and store the result in the destination (D). The shift count can be a 4-bit positive integer, a value in a
16-bit register, or the MSP of an accumulator. For 36- and 32-bit destinations, only the MSP is shifted
and the LSP is cleared, with zero extension from bit 31 (the FF2 portion is ignored). The result is not
affected by the state of the saturation bit (SA).
Example 1:
LSRR.W Y1,Y0,A
; logical right shift of 16-bit Y1 by
; least 4 bits of Y0
After Execution
Before Execution
0
3456
3456
0
5555
0000
A2
A1
A0
A2
A1
A0
AAAA
FFF1
AAAA
FFF1
Y1
Y0
Y1
Y0
SR
SR
0300
0300
Explanation of Example:
Prior to execution, the Y1 register contains the value to be shifted ($AAAA), and the Y0 register contains the amount by which to shift (least 4 bits of $FFF1 = 1). The contents of the destination register
are not important prior to execution because they have no effect on the calculated value. The LSRR.W
instruction logically shifts the value $AAAA by 1 bit to the right and places the result in the destination
register A (the LSP is cleared).
Example 2:
LSRR.W Y1,A
; logical right shift of 16-bit A1 by
; least 4 bits of Y1
After Execution
Before Execution
F
AAAA
4567
0
5555
0000
A2
A1
A0
A2
A1
A0
0001
000F
0001
000F
Y1
Y0
Y1
Y0
SR
0300
SR
0300
Explanation of Example:
Prior to execution, A1 contains the value that is to be shifted ($AAAA), and the Y1 register contains
the amount by which to shift ($1). The LSRR.W instruction logically shifts the zero-extended value
$AAAA by 1 bit to the right and places the result in the destination register A (the LSP is cleared).
A-174
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
LSRR.W
LSRR.W
Multi-Bit Logical Right Shift Word
Condition Codes Affected:
MR
N
Z
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
— Set if MSB of result is set
— Set if accumulator result equals zero
Instruction Fields:
Operation
Operands
C
W
Comments
LSRR.W
#<0–15>,FFF
1
1
Logical shift right by a 4-bit positive immediate integer
(sign extends into FF2)
EEE,FFF
1
1
Logical shift right destination by value specified in 4
LSBs of the first operand (sign extends into FF2)
Y1,X0,FFF
Y0,X0,FFF
Y1,Y0,FFF
Y0,Y0,FFF
A1,Y0,FFF
B1,Y1,FFF
C1,Y0,FFF
C1,Y1,FFF
1
1
Logical shift right the first operand by value specified in
4 LSBs of the second operand; places result in FFF,
sign extends into FF2
Instruction Opcodes:
15
LSRR.W #<0–15>,FFF
0
1
0
1
1
1
1
15
LSRR.W EEE,FFF
0
15
LSRR.W Q1,Q2,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
12
11
1
1
12
11
1
1
12
11
1
0
8
7
F
1
F
F
8
7
1
F
F
F
8
7
1
F
F
F
Instruction Set Details
4
3
B
0
1
0
4
3
a
a
a
1
4
3
Q
Q
Q
0
0
B
B
B
0
0
1
0
1
0
0
0
A-175
MAC
MAC
Multiply-Accumulate
Operation:
Assembler Syntax:
D + (S1 × S2) → D (no parallel move)
D + (S1 × S2) → D (one parallel move)
D + (S1 × S2) → D (two parallel reads)
MAC
MAC
MAC
(+)S1,S2,D
(+)S1,S2,D
S1,S2,D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Multiply the two signed 16-bit source operands, and add or subtract the 32-bit fractional product to or
from the destination (D). Both source operands must be located in the FF1 portion of an accumulator
or in X0, Y0, or Y1. The fractional product is first sign extended before the 36-bit addition (or subtraction) is performed. If the destination is one of the 16-bit registers, it is first sign extended internally and
concatenated with 16 zero bits to form a 36-bit operand before the operation to the fractional product;
the high-order 16 bits of the result are then stored.
Usage:
This instruction is used for the multiplication and accumulation of fractional data or integer data when
a full 32-bit product is required (see Section 5.3.3, “Multiplication,” on page 5-18). When the destination is a 16-bit register, this instruction is useful only for fractional data.
Example:
MAC
Y0,X0,A
X:(R0)+,Y0
X:(R3)+,X0
Before Execution
; fractional MAC, two reads
After Execution
0
0000
8000
0
000A
8000
A2
A1
A0
A2
A1
A0
FF00
0200
FF00
0300
Y1
Y0
Y1
Y0
X0
0280
X0
0288
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $0280 (or fractional value 0.019531250),
the 16-bit Y0 register contains the value $0200 (or fractional value 0.015625), and the 36-bit A accumulator contains the value $0:0000:8000 (or fractional value 0.000015259). Execution of the MAC
instruction multiplies the 16-bit signed value in the X0 register by the 16-bit signed value in Y0 (yielding the fractional product result of $000A:0000 = 0.000305176), adds the resulting 32-bit product to
the 36-bit A accumulator, and stores the result ($0:000A:8000 = 0.00320435) back into the A accumulator. In parallel, X0 and Y0 are updated with new values that are fetched from the data memory, and
the two address registers (R0 and R3) are post-incremented by one.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
A-176
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of accumulator result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if accumulator result equals zero
Set if overflow has occurred in accumulator result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MAC
MAC
Multiply-Accumulate
Instruction Fields:
Operation
Operands
C
W
Comments
MAC
(±)FFF1,FFF1,FFF
1
1
Fractional multiply-accumulate; multiplication
result optionally negated before accumulation.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
MAC2
Y1,X0,F
Y0,X0,F
Y1,Y0,F
Y0,Y0,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A1,Y0,F
B1,Y1,F
C1,Y0,F
C1,Y1,F
–C1,Y0,F
–C1,Y1,F
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
MAC2
Y1,X0,F
Y1,Y0,F
Y0,X0,F
C1,Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-177
MAC
MAC
Multiply-Accumulate
Instruction Opcodes:
15
MAC
–C1,Q2,F GGG,X:<ea_m>
0
0
0
15
MAC
–C1,Q2,F X:<ea_m>,GGG
0
0
1
15
MAC
FFF1,FFF1,FFF
MAC
Q1,Q2,F GGG,X:<ea_m>
MAC
Q1,Q2,F X:<ea_m>,GGG
0
1
1
0
0
0
1
15
0
15
0
15
MAC
Q3,Q4,F X:<ea_m>,reg1
X:<ea_v>,reg2
0
1
1
15
MAC
–FFF1,FFF1,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-178
0
1
1
12
11
1
0
12
11
1
0
12
11
0
1
12
11
0
1
12
11
0
1
12
11
0
0
12
11
0
1
G
G
G
G
8
7
G
F
8
7
G
F
8
7
F
0
F
F
8
7
G
G
G
F
8
7
G
G
G
F
8
7
v
F
8
7
F
F
1
1
v
F
Q
Q
Q
Q
4
3
Q
1
4
3
Q
1
4
3
J
J
J
J
4
3
Q
Q
Q
1
4
3
Q
Q
Q
1
4
3
Q
1
4
3
J
J
v
J
DSP56800E and DSP56800EX Core Reference Manual
Q
J
0
m
R
R
0
m
R
R
0
J
0
0
m
R
R
m
R
R
0
0
0
m
0
v
0
J
1
0
Freescale Semiconductor
MACR
MACR
Multiply-Accumulate and Round
Operation:
Assembler Syntax:
D + (S1 × S2) + r → D (no parallel move)
D + (S1 × S2) + r → D (one parallel move)
D + (S1 × S2) + r → D (two parallel reads)
MACR
MACR
MACR
(+)S1,S2,D
S1,S2,D
S1,S2,D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Multiply the two signed 16-bit source operands, add or subtract the 32-bit fractional product to or from
the third operand, and round and store the result in the destination (D). Both source operands must be
located in the FF1 portion of an accumulator or in X0, Y0, or Y1. The fractional product is first sign
extended before the 36-bit addition is performed, followed by the rounding operation. If the destination
is one of the 16-bit registers, it is first sign extended internally and concatenated with 16 zero bits to
form a 36-bit operand before being added to the fractional product. The addition is then followed by
the rounding operation, and the high-order 16 bits of the result are then stored. This instruction uses
the rounding technique that is selected by the R bit in the OMR. When the R bit is cleared (default
mode), convergent rounding is selected; when the R bit is set, two’s-complement rounding is selected.
Refer to Section 5.9, “Rounding,” on page 5-43 for more information about the rounding modes. Note
that the rounding operation always zeros the LSP of the result if the destination (D) is an accumulator
or the Y register.
Usage:
This instruction is used for the multiplication, accumulation, and rounding of fractional data.
Example:
MACR
Y0,X0,A
X:(R0)+,Y0
Before Execution
X:(R3)+,X0 ; multiply-accumulate
; fractional with rounding
After Execution
0
0000
8000
0
000A
0000
A2
A1
A0
A2
A1
A0
FF00
0200
FF00
0300
Y1
Y0
Y1
Y0
X0
0280
X0
0288
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $0280 (or fractional value 0.019531250),
the 16-bit Y0 register contains the value $0200 (or fractional value 0.015625), and the 36-bit A accumulator contains the value $0:0000:8000 (or fractional value 0.000015259). Execution of the MACR
instruction multiplies the 16-bit signed value in the X0 register by the 16-bit signed value in Y0 (yielding the fractional product result of $000A:0000 = 0.000305176), adds the resulting 32-bit product to
the 36-bit A accumulator ($0:000A:8000 = 0.00320435), rounds the result, and stores the rounded result ($0:000A:0000 = 0.000305176) back into the A accumulator. In parallel, X0 and Y0 are updated
with new values that are fetched from the data memory, and the two address registers (R0 and R3) are
post-incremented by one. In this example, the default rounding technique (convergent rounding) is performed (bit R in the OMR is cleared). If two’s-complement rounding is utilized (R bit is set), the result
in accumulator A is $0:000B:0000 = 0.000335693.
Freescale Semiconductor
Instruction Set Details
A-179
MACR
MACR
Multiply-Accumulate and Round
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ bit (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extended portion of accumulator result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Set if overflow has occurred in result
Instruction Fields:
Operation
Operands
C
W
Comments
MACR
(±)FFF1,FFF1,FFF
1
1
Fractional MAC with round; multiplication result
optionally negated before addition.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
MACR2
Y1,X0,F
Y0,X0,F
Y1,Y0,F
Y0,Y0,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A1,Y0,F
B1,Y1,F
C1,Y0,F
C1,Y1,F
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
A-180
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MACR
MACR
Multiply-Accumulate and Round
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
MACR2
Y1,X0,F
Y1,Y0,F
Y0,X0,F
C1,Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
MACR
FFF1,FFF1,FFF
MACR
Q1,Q2,F GGG,X:<ea_m>
MACR
Q1,Q2,F X:<ea_m>,GGG
MACR
Q3,Q4,F X:<ea_m>,reg1
X:<ea_v>,reg2
0
1
1
0
0
0
1
1
1
15
0
15
0
15
0
15
MACR –FFF1,FFF1,FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
0
1
12
11
1
1
12
11
1
1
12
11
1
0
12
11
0
1
8
7
F
0
F
F
8
7
G
G
G
F
8
7
G
G
G
F
8
7
1
v
v
F
8
7
F
F
1
F
Instruction Set Details
4
3
J
J
J
J
4
3
Q
Q
Q
1
4
3
Q
Q
Q
1
4
3
v
Q
Q
1
4
3
J
J
J
J
0
J
1
0
m
R
R
m
R
R
m
0
v
0
0
0
0
J
1
1
A-181
MACSU
Multiply-Accumulate Signed × Unsigned
Operation:
MACSU
Assembler Syntax:
D + (S1 × S2) → D
(S1 signed, S2 unsigned)
MACSU
S1,S2,D
(no parallel move)
Description: Multiply one signed 16-bit source operand by one unsigned 16-bit operand, and add the 32-bit fractional product to the destination (D). The order of the registers is important. The first source register
(S1) must contain the signed value, and the second source (S2) must contain the unsigned value to produce correct fractional results. The fractional product is first sign extended before the 36-bit addition
is performed. If the destination is one of the 16-bit registers, only the high-order 16 bits of the fractional
result are stored. The result is not affected by the state of the saturation bit (SA). Note that for 16-bit
destinations, the sign bit may be lost for large fractional magnitudes.
Usage:
In addition to single-precision multiplication of a signed-times-unsigned value and accumulation, this
instruction is used for multi-precision multiplications, as shown in Section 5.5, “Extended- and
Multi-Precision Operations,” on page 5-29.
Example:
MACSU
Y1,B1,A
; multiply signed Y1 to unsigned B1 and
; accumulate in A
Before Execution
After Execution
0
0000
0020
F
FFFF
FFF0
A2
A1
A0
A2
A1
A0
0
0002
3456
0
0002
3456
B2
B1
B0
B2
B1
B0
FFF4
8000
FFF4
8000
Y1
Y0
Y1
Y0
SR
SR
0300
0318
Explanation of Example:
Prior to execution, the 16-bit Y1 register contains the (signed) negative value $FFF4, and the 16-bit
B1 register contains the (unsigned) positive value $0002. Execution of the MACSU instruction multiplies the 16-bit signed value in the Y1 register by the 16-bit unsigned value in B1 (yielding the fractional product result of $FFFF:FFD0), then adds the sign extended result to the A accumulator, and
stores the signed result ($F:FFFF:FFF0) back into the A accumulator.
Condition Codes Affected:
MR
L
E
U
N
Z
V
A-182
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extended portion of the result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Set if overflow has occurred in result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MACSU
MACSU
Multiply-Accumulate Signed × Unsigned
Instruction Fields:
Operation
Operands
C
W
MACSU
X0,Y1,EEE
X0,Y0,EEE
Y0,Y1,EEE
Y0,Y0,EEE
Y0,A1,EEE
Y1,B1,EEE
Y0,C1,EEE
Y1,C1,EEE
1
1
Comments
16 × 16 => 32-bit unsigned/signed fractional MAC.
The first operand is treated as signed and the second as
unsigned.
Instruction Opcodes:
15
MACSU Q2,Q1,EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
1
E
Instruction Set Details
8
7
E
E
Q
Q
4
3
Q
1
0
1
1
0
A-183
MOVE.B
Operation:
S→D
MOVE.B
Move Byte (Word Pointer)
Assembler Syntax:
(no parallel move)
MOVE.B
S,D
(no parallel move)
Description: Move an 8-bit value from a register to memory or from memory to a register. Register-indirect memory
locations are specified with word pointers, offsets are specified as byte offsets, and absolute addresses
are specified as byte addresses. Register operands are affected as follows:
– If the source operand is a 16-bit register, the lower 8 bits are moved.
– If the destination operand is a 16-bit register, the lower 8 bits are written and the upper 8 bits are
filled with sign extension.
– If the source operand is an accumulator, the lower 8 bits of FF1 are moved.
– If the destination operand is an accumulator, the lower 8 bits of FF1 are written, FF2 and the upper
8 bits of FF1 are filled with sign extension, and FF0 is zero filled.
– If the destination operand is the Y register, the lower 8 bits of Y1 are written, the upper 8 bits of Y1
are filled with sign extension, and Y0 is zero filled.
Example 1:
MOVE.B
X:(R0+$21),A
Before Execution
; move byte from memory into A
After Execution
0
6677
8899
F
FF90
0000
A2
A1
A0
A2
A1
A0
X:$4454
R0
9060
004444
X:$4454
R0
9060
004444
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $0:6677:8899. After execution of the MOVE.B X:(R0+$21),A instruction, the FF1 portion of A is updated with the value in
memory that is pointed to by the word pointer R0, with a byte offset of $21, which results in the upper
byte of the word memory X:$4454. The results is sign extended through bit 35 of A. The FF0 portion
of A is filled with zero. The content of the A accumulator becomes $F:FF90:0000.
Example 2:
MOVE.B
X:(R0+$20),X:$2223 ; move byte from memory into memory
Before Execution
After Execution
X:$4454
9060
X:$4454
9060
X:$1111
3333
X:$1111
6033
R0
004444
R0
004444
Explanation of Example:
Prior to execution, the word location X:$1111 contains the value $3333. After execution of the
MOVE.B X:(R0+$20),X:$2223 instruction, the lower byte of the word memory location pointed
to by (R0+$20), which is location X:$4454, is written to the upper byte of the word memory location
X:$1111, which is specified as X:$2223 in byte address. The value at X:$1111 becomes $6033.
A-184
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.B
MOVE.B
Move Byte (Word Pointer)
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVE.B
X:(Rn+xxxx)
HHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHH
3
3
24-bit offset
X:(SP)
HHH
1
1
Pointer is SP
HHH
X:(RRR+x)
2
1
x: offset ranging from 0 to 7
HHH
X:(Rn+xxxx)
2
2
Signed 16-bit offset
HHH
X:(Rn+xxxxxx)
3
3
24-bit offset
HHH
X:(SP–x)
2
1
x: offset ranging from 1 to 8
HHH
X:(SP)
1
1
Pointer is SP
X:(Rn+xxxx)
X:xxxx
3
3
Signed 16-bit offset
Notes: • Each absolute address operand is specified as a byte address. In this address, all bits except the LSB
select the appropriate word location in memory, and the LSB selects the upper or lower byte of that word.
• Pointer Rn is a word pointer.
• Offsets x, xxxx, and xxxxxx are byte offsets.
Instruction Opcodes:
15
MOVE.B HHH,X:(RRR+x)
1
0
0
0
0
15
MOVE.B HHH,X:(SP–x)
1
15
MOVE.B HHH,X:(Rn+xxxx)
1
1
0
12
11
1
1
12
11
1
1
12
11
1
0
8
7
h
h
h
0
8
7
h
h
h
0
8
7
h
1
h
h
0
i
0
i
1
1
4
3
i
N
4
3
i
1
4
3
0
R
4
3
0
i
N
N
i
1
1
0
0
0
R
R
AAAAAAAAAAAAAAAA
15
MOVE.B HHH,X:(Rn+xxxxxx)
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
h
h
h
1
1
1
0
R
0
R
R
1
1
1
0
R
R
AAAAAAAAAAAAAAAA
15
MOVE.B HHH,X:(SP)
1
1
0
1
1
15
MOVE.B X:(Rn+xxxx),HHH
1
12
11
1
0
12
11
1
0
8
7
h
h
h
1
8
7
h
h
h
1
4
3
0
1
1
1
4
3
1
1
0
R
0
0
AAAAAAAAAAAAAAAA
Freescale Semiconductor
Instruction Set Details
A-185
MOVE.B
MOVE.B
Move Byte (Word Pointer)
Instruction Opcodes:(continued)
15
MOVE.B X:(Rn+xxxx),X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
1
1
1
4
3
0
R
0
0
R
R
0
R
R
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.B X:(Rn+xxxx),Y
1
1
0
12
11
1
0
1
1
8
7
0
1
1
1
4
3
0
R
0
AAAAAAAAAAAAAAAA
15
MOVE.B X:(Rn+xxxxxx),HHH
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
1
0
h
h
h
1
1
1
0
R
0
R
R
4
3
AAAAAAAAAAAAAAAA
15
MOVE.B X:(Rn+xxxxxx),Y
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
1
1
0
1
1
1
0
R
0
R
R
1
1
1
1
1
1
AAAAAAAAAAAAAAAA
15
MOVE.B X:(SP),HHH
1
1
1
1
0
15
MOVE.B X:(SP),Y
1
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-186
12
11
1
0
12
11
1
0
8
7
h
h
h
1
8
7
1
1
0
1
4
3
0
1
1
1
4
3
0
1
1
1
DSP56800E and DSP56800EX Core Reference Manual
0
0
Freescale Semiconductor
MOVE.BP
Operation:
S→D
MOVE.BP
Move Byte (Byte Pointer)
Assembler Syntax:
(no parallel move)
MOVE.BP
S,D
(no parallel move)
Description: Move an 8-bit value from a register to memory, from memory to a register, or between two memory
locations. Register-indirect memory locations are specified with byte pointers, offsets are specified as
byte offsets, and absolute addresses are specified as byte addresses. Register operands are affected as
follows:
– If the source operand is a 16-bit register, the lower 8 bits are moved.
– If the destination operand is a 16-bit register, the lower 8 bits are written and the upper 8 bits are
filled with sign extension.
– If the source operand is an accumulator, the lower 8 bits of FF1 are moved.
– If the destination operand is an accumulator, the lower 8 bits of FF1 are written, FF2 and the upper
8 bits of FF1 are filled with sign extension, and FF0 is zero filled.
– If the destination operand is the Y register, the lower 8 bits of Y1 are written, the upper 8 bits of Y1
are filled with sign extension, and Y0 is zero filled.
Example 1:
MOVE.BP X:(R0)+,A; move byte into A, update R0
Before Execution
After Execution
0
6677
8888
F
FF96
0000
A2
A1
A0
A2
A1
A0
X:$2222
R0
X:$2222
6996
R0
004444
6996
004445
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $0:6677:8888. After execution of the MOVE.BP X:(R0)+,A instruction, the lower 8 bits of A1 are updated with the value in
memory pointed to by the byte pointer R0, the result is sign extended through bit 35 of A, and the FF0
portion is filled with zero. The value in A becomes $F:FF96:0000. The R0 pointer is then incremented
by one.
Example 2:
MOVE.BP X0,X:(R0+$21) ; move byte into data memory location
Before Execution
After Execution
X0
77AA
X0
77AA
X:$2232
9060
X:$2232
AA60
R0
004444
R0
004444
Explanation of Example:
Prior to the memory move, the word memory location X:$2232 contains the value $9060. After execution of the MOVE.BP X0,X:(R0+$21) instruction, the lower 8 bits of X0 are written to the upper
byte of the word memory location X:$2232. This memory location is the result of the effective address
(R0+$21) in bytes.
Freescale Semiconductor
Instruction Set Details
A-187
MOVE.BP
MOVE.BP
Move Byte (Byte Pointer)
Condition Codes Affected:
The condition codes are not affected by this instruction.
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVE.BP
X:(RRR)
X:(RRR)+
X:(RRR)–
HHH
1
1
Move signed byte from memory
X:(RRR+N)
HHH
2
1
Address = Rn+N
X:(RRR+xxxx)
HHH
2
2
Unsigned 16-bit offset
X:(RRR+xxxxxx)
HHH
3
3
24-bit offset
X:xxxx
HHH
2
2
Unsigned 16-bit address
X:xxxxxx
HHH
3
3
24-bit address
HHH
X:(RRR)
X:(RRR)+
X:(RRR)–
1
1
Move signed byte to memory
HHH
X:(RRR+N)
2
1
Address = Rn+N
HHH
X:(RRR+xxxx)
2
2
Unsigned 16-bit offset
HHH
X:(RRR+xxxxxx)
3
3
24-bit offset
HHH
X:xxxx
2
2
Unsigned 16-bit address
HHH
X:xxxxxx
3
3
24-bit address
X:(RRR)
X:(RRR)+
X:(RRR)–
X:xxxx
2
2
Move byte from one memory location to
another; RRR used as a byte pointer
X:(RRR+N)
X:xxxx
3
2
RRR used as a byte pointer
X:(RRR+xxxx)
X:xxxx
3
3
Unsigned 16-bit offset; RRR used as a byte
pointer
X:xxxx
X:xxxx
3
3
16-bit absolute address
Notes: • Each absolute address operand is specified as a byte address. In this address, all bits except the LSB
select the appropriate word location in memory, and the LSB selects the upper or lower byte of that word.
• Pointer RRR is a byte pointer.
• Offsets xxxx and xxxxxx are byte offsets.
A-188
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.BP
MOVE.BP
Move Byte (Byte Pointer)
Instruction Opcodes:
15
MOVE.BP HHH,X:(RRR+xxxx)
1
1
0
12
11
1
0
h
h
8
7
h
1
1
1
4
3
0
N
4
3
0
1
N
N
AAAAAAAAAAAAAAAA
15
MOVE.BP HHH,X:(RRR+xxxxxx)
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
h
h
h
1
1
1
0
N
1
N
N
4
3
M
N
4
3
1
1
4
3
AAAAAAAAAAAAAAAA
15
MOVE.BP HHH,X:<ea_MM>
1
1
0
15
MOVE.BP HHH,X:xxxx
1
1
0
12
11
1
0
12
11
1
0
h
h
h
h
8
7
h
1
8
7
h
1
0
1
1
1
0
M
N
N
0
1
0
1
AAAAAAAAAAAAAAAA
15
MOVE.BP HHH,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
h
h
h
1
1
1
1
1
1
0
1
1
N
N
1
N
N
AAAAAAAAAAAAAAAA
15
MOVE.BP X:(RRR+xxxx),HHH
1
1
1
12
11
1
0
h
h
8
7
h
1
1
1
4
3
0
N
0
AAAAAAAAAAAAAAAA
15
MOVE.BP X:(RRR+xxxx),X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
1
1
1
4
3
0
N
4
3
0
N
0
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.BP X:(RRR+xxxx),Y
1
1
0
12
11
1
0
1
1
8
7
0
1
1
1
0
1
N
N
AAAAAAAAAAAAAAAA
15
MOVE.BP X:(RRR+xxxxxx),HHH
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
1
0
h
h
h
1
1
1
0
N
1
N
N
AAAAAAAAAAAAAAAA
15
MOVE.BP X:(RRR+xxxxxx),Y
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
0
1
1
0
1
1
1
0
N
1
N
N
4
3
M
N
AAAAAAAAAAAAAAAA
15
MOVE.BP X:<ea_MM>,HHH
Freescale Semiconductor
1
1
1
12
11
1
0
h
h
Instruction Set Details
8
7
h
1
0
1
0
M
N
N
A-189
MOVE.BP
MOVE.BP
Move Byte (Byte Pointer)
Instruction Opcodes:(continued)
15
MOVE.BP X:<ea_MM>,X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
1
0
1
4
3
M
N
0
M
N
N
M
N
N
AAAAAAAAAAAAAAAA
15
MOVE.BP X:<ea_MM>,Y
1
1
0
15
MOVE.BP X:xxxx,HHH
1
1
1
12
11
1
0
12
11
1
0
1
1
h
h
8
7
0
1
8
7
h
1
0
1
1
1
4
3
M
N
4
3
1
1
4
3
1
1
0
0
1
0
1
AAAAAAAAAAAAAAAA
15
MOVE.BP X:xxxx,X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
1
1
1
0
1
0
1
1
0
1
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.BP X:xxxx,Y
1
1
0
12
11
1
0
1
1
8
7
0
1
1
1
4
3
1
1
4
3
0
AAAAAAAAAAAAAAAA
15
MOVE.BP X:xxxxxx,HHH
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
0
h
h
h
1
1
1
1
1
1
0
1
4
3
AAAAAAAAAAAAAAAA
15
MOVE.BP X:xxxxxx,Y
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
1
1
0
1
1
1
1
1
1
0
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycles
Memory:
1–3 program words
A-190
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.L
Operation:
S→D
MOVE.L
Move Long
Assembler Syntax:
(no parallel move)
MOVE.L
S,D
(no parallel move)
Description: Move a 32-bit value between a register and memory, between two memory locations, or between two
registers, or load a memory or a register with an immediate value. Register-indirect memory locations
are specified with word pointers, offsets of constants are specified as word offsets, an offset of the N
register is specified as a long offset, and absolute addresses are specified as word addresses. The address of the data memory location to be accessed must be an even word address.
When a memory location is accessed using MOVE.L, two consecutive locations—an even address location and the next higher odd address location—are accessed for both reading and writing. If pointer
RRR is used, it points to the even-address location. If pointer SP is used, it points to the odd-address
location.
Register operands are affected as follows:
– When an accumulator is a source, the value of the FF1 and FF0 portions are loaded into the
destination.
– When any other register (32-bit or smaller) is a source, the value of the entire register is loaded into
the destination.
– When an accumulator is a destination, the FF1 and FF0 portions are written with the source value
and sign extended through bit 35.
– When an RRR register is a destination, the entire register is filled with the source value and sign
extended if the source is a sign immediate data (otherwise zero extended).
– When any other register (32-bit or smaller) is a destination, the entire register is filled with the
source value.
Example 1:
MOVE.L
X:(SP-$2),Y
; move long word from stack into Y
Before Execution
After Execution
1234
5678
3333
2222
Y1
Y0
Y1
Y0
X:$1113
3333
X:$1113
3333
X:$1112
2222
X:$1112
2222
SP
001115
SP
001115
Explanation of Example:
Prior to the memory move, the Y register contains the value $1234:5678. After execution of the
MOVE.L X:(SP),Y instruction, Y is updated with the value in memory (on the stack) that is pointed
to by the SP register, with a long-word offset of two. Y becomes $3333:2222, and SP remains unchanged. Note that since this value is a reference to a long word on the stack, an odd word address is
specified and points to the upper word of the long. The base address of the long word retrieved is
$001112.
Freescale Semiconductor
Instruction Set Details
A-191
MOVE.L
MOVE.L
Move Long
Example 2:
MOVE.L
A10,X:(R3+$1000)
Before Execution
; move long word to stack
After Execution
F
8765
4321
F
8765
4321
A2
A1
A0
A2
A1
A0
X:$5445
2222
X:$5445
8765
X:$5444
3333
X:$5444
4321
R3
004444
R3
004444
Explanation of Example:
Prior to the memory move, the memory locations X:$5444 and X:$5445 contain the values
$3333:2222, respectively. After execution of the MOVE.L instruction, location X:$5444 is updated
with the value in the FF0 portion of accumulator A, and X:$5445 is updated with the value in the FF1
portion of accumulator A. These memory locations are pointed to by (R3+$1000). The final result is
that X:$5444 contains $4321 and X:$5445 contains $8765.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-192
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.L
MOVE.L
Move Long
Instruction Fields:
Operation
Source
Destination
MOVE.L
X:(Rn)
X:(Rn)+
X:(Rn)–
HHHH.L
1
Comments
1 Move signed 32-bit long word to or from memory.
X:(SP)–
dddd.L
1
1 Pop 32 bits from stack (does not modify bits 14–10 in SR).
X:(Rn+N)
HHHH.L
2
1 Address = Rn+N.
X:(Rn+xxxx)
HHHH.L
2
2 Signed 16-bit offset.
X:(Rn+xxxxxx)
HHHH.L
3
3 24-bit offset.
X:(SP–xx)
HHHH.L
2
1 Unsigned 6-bit offset left shifted 1 bit.
X:xxxx
HHHH.L
2
2 Unsigned 16-bit address.
X:xxxxxx
HHHH.L
3
3 24-bit address.
HHHH.L
X:(Rn)
X:(Rn)+
X:(Rn)–
1
1
Move signed 32-bit long word to memory.
Note that Rn includes SP.
dddd.L
X:(SP)+
1
1
Push 32 bits onto stack.
SP not permitted in dddd.L.
HHHH.L
X:(Rn+N)
2
1
Address = Rn+N.
HHHH.L
X:(Rn+xxxx)
2
2
Signed 16-bit offset.
X:(Rn+xxxxxx) 3
3
24-bit offset.
1
Unsigned 6-bit offset left shifted 1 bit.
HHHH.L
HHHH.L
Notes: •
•
•
•
•
•
•
•
•
C W
X:(SP–xx)
2
HHHH.L
X:xxxx
2
2
Unsigned 16-bit address.
HHHH.L
X:xxxxxx
3
3
24-bit address.
X:(SP–xx)
X:xxxx
3
2 Move long from one memory location to another.
X:(Rn)
X:(Rn)+
X:(Rn)–
X:xxxx
2
2
X:(Rn+N)
X:xxxx
3
2
X:(Rn+xxxx)
X:xxxx
3
3 Signed 16-bit offset.
X:xxxx
X:xxxx
3
3 16-bit absolute address.
#xxxx
X:xxxx
3
3 Sign extend 16-bit value and move to 32-bit memory location.
#xxxxxxxx
X:xxxx
4
4 Move to 32-bit memory location.
The absolute address operand X:xxxx is specified as a word address.
Pointer Rn is a word pointer.
Offsets xx, xxxx, and xxxxxx are word offsets.
N offsets are long offsets.
RRR pointers must be even.
SP pointers must be odd.
Immediate offsets must be even.
Absolute addresses must be even.
Offset N can be even or odd since it is a long offset that will be shifted left by 1.
Freescale Semiconductor
Instruction Set Details
A-193
MOVE.L
Instruction Fields:
MOVE.L
Move Long
(continued)
Operation
Source
Destination
MOVE.L
#<–16,15>
HHH.L
1
1 4-bit integer data, sign extended to 36 bits.
#xxxx
HHHH.L
2
2 Sign extend 16-bit immediate data to 36 bits when moving to
an accumulator; sign extend to 24 bits when moving to an AGU
register.
Use MOVEU.W to move unsigned 16-bit immediate data to the
AGU.
#xxxxxxxx
HHH.L
3
3 Move signed 32-bit immediate data to a 32-bit accumulator.
#xxxxxx
RRR
3
3 Move unsigned 24-bit immediate value to AGU register.
HHH.L
RRR
1
1
RRR
HHH.L
1
1 Move pointer register to data ALU register; zero extend the
24-bit value contained in the RRR register.
Notes: •
•
•
•
•
•
•
•
•
A-194
C W
Comments
The absolute address operand X:xxxx is specified as a word address.
Pointer Rn is a word pointer.
Offsets xx, xxxx, and xxxxxx are word offsets.
N offsets are long offsets.
RRR pointers must be even.
SP pointers must be odd.
Immediate offsets must be even.
Absolute addresses must be even.
Offset N can be even or odd since it is a long offset that will be shifted left by 1.
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.L
MOVE.L
Move Long
Instruction Opcodes:
15
MOVE.L #<–16,15>,HHH.L
1
1
1
15
MOVE.L #xxxx,HHH.L
1
1
1
12
11
0
1
12
11
0
0
1
1
1
0
8
7
h
0
8
7
0
0
h
0
4
3
B
B
4
3
0
0
4
3
0
1
4
3
0
0
4
3
0
1
1
4
3
0
1
0
4
3
1
0
4
3
0
h
0
0
B
B
B
0
h
h
h
iiiiiiiiiiiiiiii
15
MOVE.L #xxxx,RRR
1
1
1
12
11
0
0
1
0
8
7
0
0
0
0
0
S
S
S
iiiiiiiiiiiiiiii
15
MOVE.L #xxxx,X:xxxx
1
1
1
12
11
0
0
1
0
8
7
0
0
0
0
0
1
1
0
iiiiiiiiiiiiiiii
AAAAAAAAAAAAAAAA
15
MOVE.L #xxxxxx,RRR
1
1
1
12
11
0
0
1
0
8
7
0
0
0
0
S
S
S
h
h
h
iiiiiiiiiiiiiiii.lwr
iiiiiiiiiiiiiiii.upr
15
MOVE.L #xxxxxxxx,HHH.L
1
1
1
12
11
0
0
1
0
8
7
0
0
0
0
iiiiiiiiiiiiiiii.lwr
iiiiiiiiiiiiiiii.upr
15
MOVE.L #xxxxxxxx,X:xxxx
1
1
1
12
11
0
0
1
0
8
7
0
0
0
0
0
1
1
0
iiiiiiiiiiiiiiii.lwr
iiiiiiiiiiiiiiii.upr
AAAAAAAAAAAAAAAA
15
MOVE.L HHH.L,RRR
1
1
1
1
0
15
MOVE.L HHHH.L,X:(Rn+xxxx)
1
12
11
0
0
12
11
1
h
8
7
0
h
h
h
8
7
h
h
h
0
0
1
0
4
3
1
1
0
R
0
S
S
S
0
R
R
0
AAAAAAAAAAAAAAAA
15
MOVE.L HHHH.L,X:(Rn+xxxxxx)
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
h
h
h
h
0
1
1
0
R
0
R
R
AAAAAAAAAAAAAAAA
Freescale Semiconductor
Instruction Set Details
A-195
MOVE.L
MOVE.L
Move Long
Instruction Opcodes:(continued)
15
MOVE.L HHHH.L,X:(SP–xx)
1
0
0
15
MOVE.L HHHH.L,X:<ea_MM>
1
1
0
1
0
15
MOVE.L HHHH.L,X:xxxx
1
12
11
1
h
12
11
1
h
12
11
1
h
h
h
8
7
h
1
8
7
0
h
h
h
8
7
h
h
h
0
1
a
4
3
a
a
4
3
R
0
1
M
4
3
1
1
1
1
4
3
0
a
a
a
0
M
R
R
1
0
1
0
AAAAAAAAAAAAAAAA
15
MOVE.L HHHH.L,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
h
h
h
h
0
1
1
1
1
1
0
1
1
0
1
AAAAAAAAAAAAAAAA
15
MOVE.L LA2,X:(SP)+
1
1
1
15
MOVE.L RRR,HHH.L
1
1
1
1
1
15
MOVE.L X:(Rn+xxxx),HHHH.L
1
12
11
0
0
12
11
0
0
12
11
1
h
1
1
8
7
1
0
8
7
0
h
h
h
8
7
h
h
h
0
0
0
4
3
1
0
4
3
0
1
1
0
4
3
1
1
0
R
0
0
S
S
S
0
R
R
0
R
R
0
AAAAAAAAAAAAAAAA
15
MOVE.L X:(Rn+xxxx),X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
0
1
1
4
3
0
R
4
3
0
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.L X:(Rn+xxxxxx),HHHH.L
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
h
h
h
h
0
1
1
0
R
0
R
R
4
3
a
a
4
3
a
a
4
3
1
d
AAAAAAAAAAAAAAAA
15
MOVE.L X:(SP–xx),HHHH.L
1
0
1
15
MOVE.L X:(SP–xx),X:xxxx
1
0
1
12
11
1
h
12
11
1
0
h
1
h
1
8
7
h
1
8
7
0
1
1
1
a
a
0
a
a
a
0
a
a
a
AAAAAAAAAAAAAAAA
15
MOVE.L X:(SP)–,dddd.L
A-196
1
1
1
12
11
0
0
1
0
8
7
1
0
0
DSP56800E and DSP56800EX Core Reference Manual
0
0
d
d
d
Freescale Semiconductor
MOVE.L
MOVE.L
Move Long
Instruction Opcodes:(continued)
15
MOVE.L X:<ea_MM>,HHHH.L
1
1
1
15
MOVE.L X:<ea_MM>,X:xxxx
1
1
1
12
11
1
h
12
11
1
0
h
h
1
1
8
7
h
0
8
7
0
0
0
0
1
1
4
3
M
R
4
3
M
R
0
M
R
R
0
M
R
R
1
0
1
1
0
1
AAAAAAAAAAAAAAAA
15
MOVE.L X:xxxx,HHHH.L
1
1
1
12
11
1
h
h
h
8
7
h
0
1
1
4
3
1
1
0
AAAAAAAAAAAAAAAA
15
MOVE.L X:xxxx,X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
0
1
1
4
3
1
1
4
3
0
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.L X:xxxxxx,HHHH.L
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
h
h
h
h
0
1
1
1
1
1
0
1
d
d
d
AAAAAAAAAAAAAAAA
15
MOVE.L dddd.L,X:(SP)+
1
1
1
15
MOVE.L X:(SP)–,LA2
Timing:
1–4 oscillator clock cycles
Memory:
1–4 program words
Freescale Semiconductor
1
1
1
12
11
0
0
12
11
0
0
1
1
0
1
Instruction Set Details
8
7
1
0
8
7
1
0
0
0
0
0
4
3
0
d
4
3
1
0
0
0
1
0
0
A-197
MOVE.W
Operation:
S→D
S→D
MOVE.W
Move Word
Assembler Syntax:
(two parallel reads)
(no parallel move)
MOVE.W
MOVE.W
S,D
S,D
(two parallel reads)
(no parallel move)
Description: Move a 16-bit value from a register to memory, from memory to a register, from one memory location
to another, or from one register to another register, or load a register or a memory location with an immediate value. All memory locations are specified with word pointers, offsets are specified as word
offsets, and absolute addresses are specified as word addresses.
Operands are affected as follows:
–
–
–
–
–
When a 16-bit or FF2 register is a source, the entire register is loaded into the destination.
When a 24-bit register is a source, the lower 16 bits are loaded into the destination.
When a full accumulator is a source, the value of the FF1 portion is loaded into the destination.
When a 16-bit or FF2 register is a destination, the entire register is filled with the source value.
When a 24-bit register is a destination, the lower 16 bits are written and signed extended
appropriately. Refer to MOVEU.W for unsigned word initialization of AGU registers.
– When a full accumulator is a destination, the FF1 portion is written with the source value and sign
extended through bit 35; the FF0 portion is zero filled. Sign extension is also performed when the
source operand is an immediate value that is smaller than 16 bits.
– When the Y register is a destination, the Y1 portion is written and Y0 is zero filled.
– When the N register is used for post-update (for example, X:(Rn)+N), the value of N is truncated
to 16 bits and sign extended to 24 bits before it is added to Rn.
Example 1:
MOVE.W
X:(R0+$20),A
Before Execution
; move word from memory into A
After Execution
0
6677
8888
F
9060
0000
A2
A1
A0
A2
A1
A0
X:$4464
R0
9060
004444
X:$4464
R0
9060
004444
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $0:6677:8888. After execution of the MOVE.W X:(R0+$20),A instruction, the FF1 portion of A is updated with the value in
memory that is pointed to by the word pointer R0, with a word offset of $20 (word location $004464).
The FF2 portion of A is sign extended, and the FF0 portion of A is zero filled. The value in the A accumulator becomes $F:9060:0000; R0 is unchanged.
A-198
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.W
MOVE.W
Move Word
Example 2:
MOVE.W
X:(R0)+N,A1
; move word from memory into A1
Before Execution
After Execution
0
6677
8888
0
9060
8888
A2
A1
A0
A2
A1
A0
X:$4444
X:$4444
9060
9060
R0
004444
R0
FFC444
N
018000
N
018000
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $0:6677:8888. After execution of the MOVE.W X:(R0)+N,A1 instruction, the FF1 portion of A is updated with the value in
memory that is pointed to by the R0 register, word location $004444. The FF2 and FF0 portions of A
are unchanged. The value in the A accumulator becomes $0:9060:8888. R0 is post-updated to
$FFC444 as a result of 16-bit truncation and sign extension in N before the addition (R0)+N.
Condition Codes Affected:
MR
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
SZ — Set according to the standard definition after moving an accumulator value to memory
L — Set if data limiting occurred during the move of an accumulator value to a memory
Freescale Semiconductor
Instruction Set Details
A-199
MOVE.W
MOVE.W
Move Word
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVE.W
X:(Rn)
X:(Rn)+
X:(Rn)–
HHHHH
1
1
Move signed 16-bit integer word from memory
A-200
X:(Rn+N)
HHHHH
2
1
Address = Rn+N
X:(Rn)+N
HHHHH
1
1
Post-update Rn register
X:(Rn+x)
HHH
2
1
x: offset ranging from 0 to 7
X:(Rn+xxxx)
HHHHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHHHH
3
3
24-bit offset
X:(SP–xx)
HHH
2
1
Unsigned 6-bit offset
X:xxxx
HHHHH
2
2
Unsigned 16-bit address
X:xxxxxx
HHHHH
3
3
24-bit address
X:<<pp
X0, Y1, Y0
A, B, C, A1, B1
1
1
6-bit peripheral address
X:aa
X0, Y1, Y0
A, B, C, A1, B1
1
1
6-bit absolute short address
DDDDD
X:(Rn)
X:(Rn)+
X:(Rn)–
1
1
Move signed 16-bit integer word to memory
DDDDD
X:(Rn+N)
2
1
Address = Rn+N
DDDDD
X:(Rn)+N
1
1
Post-update Rn register
HHH
X:(Rn+x)
2
1
x: offset ranging from 0 to 7
DDDDD
X:(Rn+xxxx)
2
2
Signed 16-bit offset
DDDDD
X:(Rn+xxxxxx)
3
3
24-bit offset
HHHH
X:(SP–xx)
2
1
Unsigned 6-bit offset
DDDDD
X:xxxx
2
2
Unsigned 16-bit address
DDDDD
X:xxxxxx
3
3
24-bit address
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
X:<<pp
1
1
6-bit peripheral address
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
X:aa
1
1
6-bit absolute short address
X:(Rn+x)
X:xxxx
3
2
Move word from one memory location to
another; x: offset ranging from 0 to 7
X:(SP–xx)
X:xxxx
3
2
X:(Rn)
X:(Rn)+
X:(Rn)–
X:xxxx
2
2
X:(Rn+N)
X:xxxx
3
2
X:(Rn)+N
X:xxxx
2
2
X:(Rn+xxxx)
X:xxxx
3
3
Signed 16-bit offset
X:xxxx
X:xxxx
3
3
16-bit absolute address
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.W
Instruction Fields:
MOVE.W
Move Word
(continued)
Operation
Source
Destination
C
W
MOVE.W
#<–64,63>
HHHH
1
1
Signed 7-bit integer data (data is put in the
lowest 7 bits of the word portion of any accumulator, and the LSP portion is set to zero)
X:xxxx
2
2
Signed 7-bit integer data (data put in the low
portion of the word)
HHHHH
2
2
Signed 16-bit immediate data
Move to C2, D2, C0, D0 registers
#xxxx
dd
2
2
X:(Rn)
2
2
X:(Rn+xxxx)
3
3
X:(SP–xx)
2
2
X:<<pp
2
2
Move 16-bit immediate data to the last 64
locations of X data memory—peripheral registers
X:aa
2
2
Move 16-bit immediate data to the first 64
locations of X data memory
X:xxxx
3
3
X:xxxxxx
4
4
HHHHH
1
1
Move signed word to register
DDDDD
Note:
Comments
HHH
RRR
1
1
Move signed word to register
P:(Rj)+
P:(Rj)+N
X0, Y1, Y0
A, B, C, A1, B1
5
1
Read signed word from program memory. Not
allowed when the XP bit in the OMR is set
X0, Y1, Y0
A, B, C, A1, B1
R0–R5, N
P:(Rj)+
P:(Rj)+N
5
1
Write word to program memory. Not allowed
when the XP bit in the OMR is set.
•The absolute address operand X:xxxx is specified as a word address.
• Pointer Rn is a word pointer.
• Offsets x, xx, xxxx, and xxxxxx are word offsets.
Parallel Dual Reads:
First Memory Read
Second Memory Read
Operation1
MOVE.W2
Source 1
Destination 1
Source 2
Destination 2
X:(R0)
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-201
MOVE.W
MOVE.W
Move Word
Instruction Opcodes:
15
MOVE.W
#xxxx,Y
1
0
0
12
11
0
0
1
1
8
7
1
0
4
3
0
1
0
0
4
3
0
0
d
d
4
3
1
R
4
3
R
0
1
1
0
d
d
d
iiiiiiiiiiiiiiii
15
MOVE.W
DDDDD,Y
1
0
0
15
MOVE.W
X:(Rn)+N,Y
1
1
0
15
MOVE.W
X:(Rn+x),Y
MOVE.W
X:(Rn+xxxx),Y
1
0
0
1
0
15
1
12
11
0
0
12
11
1
0
12
11
1
0
12
11
1
0
1
1
1
1
8
7
0
0
8
7
0
0
8
7
0
1
1
0
8
7
1
1
0
0
1
0
0
i
i
4
3
1
0
0
R
4
3
0
0
1
R
R
0
i
R
R
0
R
R
0
AAAAAAAAAAAAAAAA
15
MOVE.W
X:(Rn+xxxxxx),Y
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
1
1
0
0
1
0
0
R
0
R
R
a
a
a
AAAAAAAAAAAAAAAA
15
MOVE.W
X:(SP–xx),Y
1
0
0
15
MOVE.W
X:<ea_MM>,Y
1
1
0
15
MOVE.W
X:xxxx,Y
1
1
0
12
11
1
0
12
11
1
0
12
11
1
0
1
1
1
1
1
1
8
7
0
0
8
7
0
0
8
7
0
0
1
0
1
a
0
1
4
3
a
a
4
3
M
R
4
3
1
1
4
3
0
0
M
R
R
0
1
0
0
AAAAAAAAAAAAAAAA
15
MOVE.W
X:xxxxxx,Y
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
0
1
1
0
0
1
1
1
1
1
0
0
B
B
B
B
B
B
B
B
B
AAAAAAAAAAAAAAAA
15
MOVE.W #<–64,63>,HHHH
1
1
1
1
1
15
MOVE.W #<–64,63>,X:xxxx
1
12
11
0
h
12
11
0
0
8
7
h
h
h
1
8
7
1
1
0
1
4
3
B
B
B
B
4
3
B
B
B
B
0
0
AAAAAAAAAAAAAAAA
15
MOVE.W #<–64,63>,Y
A-202
1
1
1
12
11
0
1
1
1
8
7
1
1
B
DSP56800E and DSP56800EX Core Reference Manual
B
4
3
B
B
0
Freescale Semiconductor
MOVE.W
MOVE.W
Move Word
Instruction Opcodes:(continued)
15
MOVE.W #xxxx,X:(Rn)
1
0
0
12
11
0
0
1
1
8
7
0
0
4
3
R
1
0
0
4
3
1
0
0
R
0
0
R
R
1
R
R
a
a
a
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,X:(Rn+xxxx)
1
0
0
12
11
0
0
1
1
8
7
0
0
0
iiiiiiiiiiiiiiii
AAAAAAAAAAAAAAAA
15
MOVE.W #xxxx,X:(SP–xx)
1
0
1
12
11
0
0
1
1
8
7
0
1
1
a
4
3
a
a
4
3
p
0
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,X:<<pp
1
0
1
12
11
0
0
1
1
8
7
1
1
1
p
p
4
3
0
p
p
p
4
3
1
0
4
3
0
p
p
p
p
p
p
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,X:aa
1
0
1
12
11
0
0
1
1
8
7
1
1
0
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,X:xxxx
1
0
0
12
11
0
0
1
1
8
7
0
0
1
0
0
1
0
0
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
0
0
0
0
1
1
0
0
1
0
1
0
1
0
0
4
3
1
0
4
3
1
R
4
3
0
R
AAAAAAAAAAAAAAAA
iiiiiiiiiiiiiiii
15
MOVE.W #xxxx,dd
1
0
0
12
11
0
0
1
1
8
7
0
0
1
0
0
0
d
d
iiiiiiiiiiiiiiii
15
MOVE.W DDDDD,X:(Rn)+N
1
1
0
15
MOVE.W DDDDD,X:(Rn+xxxx)
1
1
0
12
11
1
D
12
11
1
D
D
D
D
D
8
7
D
D
8
7
D
D
1
1
0
0
0
1
R
R
0
0
R
R
AAAAAAAAAAAAAAAA
Freescale Semiconductor
Instruction Set Details
A-203
MOVE.W
MOVE.W
Move Word
Instruction Opcodes:(continued)
15
MOVE.W DDDDD,X:(Rn+xxxxxx)
12
11
8
7
4
3
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
D
D
D
D
D
1
0
0
R
0
R
R
M
R
R
AAAAAAAAAAAAAAAA
15
MOVE.W DDDDD,X:<ea_MM>
1
1
0
15
MOVE.W DDDDD,X:xxxx
1
1
0
12
11
1
D
12
11
1
D
D
D
D
D
8
7
D
D
8
7
D
D
0
1
0
1
4
3
M
R
4
3
1
1
4
3
0
0
1
0
0
AAAAAAAAAAAAAAAA
15
MOVE.W DDDDD,X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
D
D
D
D
D
1
1
1
1
1
0
0
m
R
R
p
p
p
AAAAAAAAAAAAAAAA
15
MOVE.W GGGG,P:<ea_m>
1
0
0
1
0
15
MOVE.W GGGG,X:<<pp
1
15
MOVE.W GGGG,X:aa
1
1
0
1
1
15
MOVE.W HHH,RRR
1
15
MOVE.W HHH,X:(Rn+x)
1
0
0
15
MOVE.W HHHH,X:(SP–xx)
1
0
0
15
MOVE.W P:<ea_m>,GGG
1
0
0
1
1
15
MOVE.W X:(Rn)+N,X:xxxx
1
12
11
0
G
12
11
0
G
12
11
0
G
12
11
0
0
12
11
1
0
12
11
1
h
12
11
0
0
12
11
1
0
8
7
G
G
G
0
8
7
G
G
G
0
8
7
0
G
G
G
8
7
h
h
h
0
8
7
h
0
8
7
h
0
8
7
0
h
h
h
h
G
G
G
8
7
1
1
0
0
4
3
1
1
0
0
4
3
1
p
p
p
4
3
p
0
p
p
4
3
0
1
0
1
4
3
i
R
4
3
a
a
4
3
1
0
1
i
a
1
1
0
4
3
1
0
1
R
0
0
0
p
p
p
S
S
S
0
0
i
R
R
0
a
a
a
0
m
R
R
1
R
R
0
AAAAAAAAAAAAAAAA
A-204
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVE.W
MOVE.W
Move Word
Instruction Opcodes:(continued)
15
MOVE.W X:(Rn+x),X:xxxx
1
0
1
12
11
1
0
1
1
8
7
0
0
0
i
4
3
i
R
0
i
R
R
0
R
R
a
a
a
M
R
R
m
0
v
p
p
p
AAAAAAAAAAAAAAAA
15
MOVE.W X:(Rn+xxxx),X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
0
1
0
4
3
0
R
0
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
15
MOVE.W X:(SP–xx),X:xxxx
1
0
1
12
11
1
0
1
1
8
7
0
0
1
a
4
3
a
a
0
AAAAAAAAAAAAAAAA
15
MOVE.W X:<ea_MM>,X:xxxx
1
1
1
12
11
1
0
1
1
8
7
0
0
0
0
4
3
M
R
0
AAAAAAAAAAAAAAAA
15
MOVE.W X:<ea_m>,reg1
X:<ea_v>,reg2
0
1
1
1
0
15
MOVE.W X:<<pp,GGG
1
15
MOVE.W X:aa,GGG
1
1
0
15
MOVE.W X:xxxx,X:xxxx
1
1
1
12
11
0
0
12
11
0
0
12
11
0
0
12
11
1
0
8
7
1
v
v
0
8
7
G
G
G
1
8
7
G
1
8
7
0
0
G
1
G
1
4
3
v
0
0
0
4
3
1
p
p
p
4
3
p
p
4
3
1
1
0
1
p
1
0
0
0
p
p
p
0
1
0
0
AAAAAAAAAAAAAAAA.s
AAAAAAAAAAAAAAAA.d
Freescale Semiconductor
Instruction Set Details
A-205
MOVE.W
MOVE.W
Move Word
Instruction Opcodes:(continued)
Note:
These instructions only allow data ALU registers as destinations except register Y.
15
MOVE.W X:(Rn+x),HHH
1
0
1
15
MOVE.W #xxxx,HHHHH
1
0
0
12
11
1
0
12
11
0
0
h
h
1
1
8
7
h
0
8
7
1
0
0
1
i
0
4
3
i
R
4
3
d
d
4
3
d
d
4
3
a
0
i
R
R
0
d
d
d
iiiiiiiiiiiiiiii
15
MOVE.W DDDDD,HHHHH
1
0
0
15
MOVE.W X:(SP–xx),HHH
1
0
1
1
1
1
1
15
MOVE.W X:(Rn)+N,HHHHH
1
15
MOVE.W X:(Rn+xxxx),HHHHH
1
12
11
0
D
12
11
1
0
12
11
1
d
12
11
1
d
D
D
8
7
D
D
8
7
0
h
h
h
8
7
d
d
d
d
8
7
d
d
d
d
0
0
1
a
a
4
3
1
0
1
R
4
3
1
0
0
R
4
3
0
d
d
d
0
a
a
a
1
R
R
0
R
R
0
0
AAAAAAAAAAAAAAAA
15
MOVE.W X:(Rn+xxxxxx),HHHHH
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
d
d
d
d
d
1
0
0
R
0
R
R
M
R
R
AAAAAAAAAAAAAAAA
15
MOVE.W X:<ea_MM>,HHHHH
1
1
1
15
MOVE.W X:xxxx,HHHHH
1
1
1
12
11
1
d
12
11
1
d
d
d
d
d
8
7
d
d
8
7
d
d
0
1
0
1
4
3
M
R
4
3
1
1
4
3
0
0
1
0
0
AAAAAAAAAAAAAAAA
15
MOVE.W X:xxxxxx,HHHHH
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
d
d
d
d
d
1
1
1
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–5 oscillator clock cycle(s)
Memory:
1–4 program word(s)
A-206
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVEU.B
Operation:
S→D
MOVEU.B
Move Unsigned Byte
Assembler Syntax:
(no parallel move)
MOVEU.B
S,D
(no parallel move)
Description: Move an 8-bit value from memory to a register or from one memory location to another. The source
operand cannot be a register. Register-indirect memory locations are specified with word pointers, offsets are specified as byte offsets, and absolute addresses are specified as byte addresses. Register operands are affected as follows:
– If the destination operand is a 16-bit register, the lower 8 bits are written and the upper 8 bits are
filled with zero extension.
– If the destination operand is the Y register, the lower 8 bits of Y1 are written, and the upper 8 bits
of Y1 and all of Y0 are filled with zero.
– If the destination operand is an accumulator, the lower 8 bits of FF1 are written, the upper 8 bits of
FF1 an FF2 are filled with zero extension, and FF0 is zero filled.
Example 1:
MOVEU.B X:(R0+$21),A
Before Execution
; move byte from memory into A
After Execution
F
CCDD
2233
0
0090
0000
A2
A1
A0
A2
A1
A0
X:$4454
R0
X:$4454
9060
R0
004444
9060
004444
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $F:CCDD:2233. After execution of the MOVEU.B X:(R0+$21),A instruction, the low-order 8 bits of A1 are updated with the
value in memory that is pointed to by the word pointer R0, with a byte offset of $21, and the value is
zero extended through bit 35. The FF0 portion of A is zero filled. The value in the A accumulator becomes $0:0090:0000; R0 is unaffected.
Example 2:
MOVEU.B X:(SP),X0; move byte from memory into X0
Before Execution
After Execution
X0
X:$4443
SP
X0
DDEE
6996
004443
X:$4443
SP
0096
6996
004443
Explanation of Example:
Prior to the memory move, the accumulator register X0 contains the value $DDEE. After execution of
the MOVEU.B X:(SP),X0 instruction, the low-order 8 bits of X0 are updated with the value in
memory that is pointed to by the word pointer SP, and the value is zero extended through bit 15. The
X0 register becomes $0096; SP is unaffected.
Condition Codes Affected:
The condition codes are not affected by this instruction.
Freescale Semiconductor
Instruction Set Details
A-207
MOVEU.B
MOVEU.B
Move Unsigned Byte
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVEU.B
X:(RRR+x)
HHH
2
1
x: offset ranging from 0 to 7
X:(Rn+xxxx)
HHH
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
HHH
3
3
24-bit offset
X:(SP–x)
HHH
2
1
x: offset ranging from 1 to 8
X:(SP)
HHH
1
1
Pointer is SP
X:(RRR+x)
X:xxxx
3
2
x: offset ranging from 0 to 7
X:(SP)
X:xxxx
2
2
Signed 16-bit offset
X:(SP–x)
X:xxxx
3
2
x: offset ranging from 1 to 8
Notes: • Each absolute address operand is specified as a byte address. In this address, all bits except the LSB
select the appropriate word location in memory, and the LSB selects the upper or lower byte of that word.
• Pointer Rn is a word pointer.
• Offsets x, xxxx, and xxxxxx are byte offsets
Instruction Opcodes:
15
MOVEU.B X:(RRR+x),HHH
1
0
1
0
1
15
MOVEU.B X:(RRR+x),X:xxxx
1
12
11
1
1
12
11
1
1
8
7
h
h
h
0
8
7
1
1
0
0
0
i
0
i
4
3
i
N
4
3
i
N
0
i
N
N
i
N
N
i
N
N
0
R
R
0
R
R
0
AAAAAAAAAAAAAAAA
15
MOVEU.B X:(RRR+x),Y
1
0
0
1
1
15
MOVEU.B X:(Rn+xxxx),HHH
1
12
11
1
1
12
11
1
1
8
7
1
1
0
0
8
7
h
h
h
1
4
3
0
i
i
N
4
3
1
1
0
R
0
0
AAAAAAAAAAAAAAAA
15
MOVEU.B X:(Rn+xxxx),Y
1
1
0
12
11
1
1
0
1
8
7
0
1
1
1
4
3
0
R
0
AAAAAAAAAAAAAAAA
15
MOVEU.B X:(Rn+xxxxxx),HHH
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
1
1
h
h
h
1
1
1
0
R
0
R
R
4
3
AAAAAAAAAAAAAAAA
15
MOVEU.B X:(Rn+xxxxxx),Y
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
0
1
1
0
1
0
1
1
1
0
R
0
R
R
AAAAAAAAAAAAAAAA
A-208
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVEU.B
MOVEU.B
Move Unsigned Byte
Instruction Opcodes:(continued)
15
MOVEU.B X:(SP),HHH
1
1
1
15
MOVEU.B X:(SP),Y
1
1
0
15
MOVEU.B X:(SP–x),HHH
1
0
1
15
MOVEU.B X:(SP–x),X:xxxx
1
0
1
12
11
1
1
12
11
1
1
12
11
1
1
12
11
1
1
h
0
h
1
h
1
h
1
8
7
h
1
8
7
0
1
8
7
h
0
8
7
0
0
0
0
0
0
1
1
i
i
4
3
1
1
4
3
1
1
4
3
i
1
4
3
i
1
4
3
i
1
4
3
1
1
0
1
1
1
0
1
1
1
0
i
1
1
0
i
1
1
AAAAAAAAAAAAAAAA
15
MOVEU.B X:(SP–x),Y
1
0
0
15
MOVEU.B X:(SP),X:xxxx
1
1
1
12
11
1
1
12
11
1
0
1
1
1
1
8
7
0
0
8
7
0
1
0
0
i
1
0
i
1
1
0
1
1
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-209
MOVEU.BP
MOVEU.BP
Move Unsigned Byte
(Byte Pointer)
Operation:
S→D
Assembler Syntax:
(no parallel move)
MOVEU.BP
S,D
(no parallel move)
Description: Move an 8-bit value from memory to a register or between two memory locations. Register-indirect
memory locations are specified with byte pointers, offsets are specified as byte offsets, and absolute
addresses are specified as byte addresses. Register operands are affected as follows:
– If the destination operand is a 16-bit register, the lower 8 bits are written and the upper 8 bits are
filled with zero extension.
– If the destination operand is the Y register, the lower 8 bits of Y1 are written. The upper 8 bits of
Y1 and all of Y0 are filled with zero.
– If the destination operand is a full accumulator, the lower 8 bits of FF1 are written. FF2 and the
upper 8 bits of FF1 are filled with zero extension, and FF0 is zero filled.
Example 1:
MOVEU.BP X:(R0)+,A
; move byte into A, update R0
Before Execution
After Execution
F
CCDD
2233
0
0099
0000
A2
A1
A0
A2
A1
A0
X:$2222
R0
X:$2222
5599
R0
004444
5599
004445
Explanation of Example:
Prior to the memory move, the accumulator register A contains the value $F:CCDD:2233. After execution of the MOVEU.BP X:(R0)+,A instruction, the FF1 portion of A is updated with the value in
memory that is pointed to by the byte pointer R0, and it is zero extended. The FF0 portion of A is zero
filled, resulting in the value $0:0099:0000. The R0 pointer is then incremented by one.
Example 2:
MOVEU.BP X:(R0+$21),Y0
Before Execution
; move byte into Y0
After Execution
00FF
1111
00FF
0088
Y1
Y0
Y1
Y0
X:$2232
R0
8899
004444
X:$2232
R0
8899
004444
Explanation of Example:
Prior to the memory move, the register Y0 contains the value $1111. After execution of the
MOVEU.BP X:(R0+$21),A instruction, the lower 8-bit portion of Y0 is updated with the value in
memory that is pointed to by the byte pointer R0, with an offset of $21 bytes, and is zero extended. The
Y0 register becomes $0088. The R0 pointer is unchanged.
Condition Codes Affected:
The condition codes are not affected by this instruction.
A-210
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVEU.BP
Move Unsigned Byte
MOVEU.BP
(Byte Pointer)
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVEU.BP
X:(RRR)
X:(RRR)+
X:(RRR)–
HHH
1
1
Move unsigned byte from memory
X:(RRR+N)
HHH
2
1
Address = Rn+N
X:(RRR+xxxx)
HHH
2
2
Unsigned 16-bit offset
X:(RRR+xxxxxx)
HHH
3
3
24-bit offset
X:xxxx
HHH
2
2
Unsigned 16-bit address
X:xxxxxx
HHH
3
3
24-bit address
Notes: • Each absolute address operand is specified as a byte address. In this address, all bits except the LSB
select the appropriate word location in memory, and the LSB selects the upper or lower byte of that word.
• Pointer Rn is a byte pointer.
• Offsets xxxx and xxxxxx are byte offsets
Freescale Semiconductor
Instruction Set Details
A-211
MOVEU.BP
MOVEU.BP
Move Unsigned Byte
(Byte Pointer)
Instruction Opcodes:
Note:
All MOVEU.BP instructions only allow memory locations as source operands.
15
MOVEU.BP X:(RRR+xxxx),HHH
1
1
1
12
11
1
1
h
h
8
7
h
1
1
1
4
3
0
N
4
3
0
N
0
1
N
N
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:(RRR+xxxx),Y
1
1
0
12
11
1
1
0
1
8
7
0
1
1
1
0
1
N
N
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:(RRR+xxxxxx),HHH
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
1
1
1
h
h
h
1
1
1
0
N
1
N
N
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:(RRR+xxxxxx),Y
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
0
1
0
1
1
1
0
N
1
N
N
M
N
N
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:<ea_MM>,HHH
1
1
1
15
MOVEU.BP X:<ea_MM>,Y
1
1
0
15
MOVEU.BP X:xxxx,HHH
1
1
1
12
11
1
1
12
11
1
1
12
11
1
1
h
h
0
1
h
h
8
7
h
1
8
7
0
1
8
7
h
1
0
0
1
1
1
1
4
3
M
N
4
3
M
N
4
3
1
1
4
3
1
1
4
3
0
0
M
N
N
0
1
0
1
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:xxxx,Y
1
1
0
12
11
1
1
0
1
8
7
0
1
1
1
0
1
0
1
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:xxxxxx,HHH
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
1
h
h
h
1
1
1
1
1
1
0
1
AAAAAAAAAAAAAAAA
15
MOVEU.BP X:xxxxxx,Y
12
11
8
7
4
3
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
1
0
1
1
0
1
0
1
1
1
1
1
1
0
1
AAAAAAAAAAAAAAAA
Timing:
1–3 oscillator clock cycle(s)
Memory:
1–3 program word(s)
A-212
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVEU.W
Operation:
S→D
MOVEU.W
Move Unsigned Word
Assembler Syntax:
(no parallel move)
MOVEU.W
S,D
(no parallel move)
Description: Move an unsigned 16-bit value from memory or a register to a register, or load a register with an immediate value. All memory locations are specified with word pointers, offsets are specified as word
offsets, and absolute addresses are specified as word addresses.
Operands are affected as follows:
– When a 16-bit register is a destination, the entire register is filled with the source value.
– When a 24-bit register is a destination, the lower 16 bits are written and the upper 8 bits are filled
with zero.
– When the N register is used for post-update (for example, X:(Rn)+N), the value of N is truncated
to 16 bits and then sign extended to 24 bits before it is added to Rn.
Note:
Only AGU registers are allowed as destination operands for this instruction.
Example 1:
MOVEU.W X:(R0+$21),R3 ; move word from memory to R3
Before Execution
X:$4465
After Execution
X:$4465
9060
9060
R0
004444
R0
004444
R3
CCDD22
R3
009060
Explanation of Example:
Prior to the memory move, the AGU register R3 contains the value $CCDD22. After execution of the
MOVEU.W X:(R0+$21),R3 instruction, the lower 16-bit portion of R3 is updated with the value in
memory that is pointed to by the R0 register, with a word offset of $21 (word location $004465). The
value is zero extended through bit 23, and the register R3 becomes $009060; R0 is unchanged.
Example 2:
MOVEU.W A,R0
; move word from a register to an AGU register
Before Execution
After Execution
F
CCDD
2233
F
CCDD
2233
A2
A1
A0
A2
A1
A0
R0
654321
R0
00CCDD
Explanation of Example:
Prior to the memory move, the AGU register R0 contains the value $654321. After execution of the
MOVEU.W A,R0 instruction, the lower 16-bit portion of R0 is updated with the value in the FF1 portion of the accumulator A. The value is zero extended through bit 23, and the register R0 becomes
$00CCDD.
Condition Codes Affected:
The condition codes are not affected by this instruction unless SR is specified as the destination.
Freescale Semiconductor
Instruction Set Details
A-213
MOVEU.W
Move Unsigned Word
MOVEU.W
Instruction Fields:
Operation
Source
Destination
C
W
Comments
MOVEU.W
X:(Rn)
X:(Rn)+
X:(Rn)–
SSSS
1
1
Move signed 16-bit integer word from memory
X:(Rn+N)
SSSS
2
1
Address = Rn+N
X:(Rn)+N
SSSS
1
1
Post-update Rn register
X:(Rn+xxxx)
SSSS
2
2
Signed 16-bit offset
X:(Rn+xxxxxx)
SSSS
3
3
24-bit offset
X:(SP–xx)
SSS
2
1
Unsigned 6-bit offset
X:xxxx
SSSS
2
2
Unsigned 16-bit address
X:xxxxxx
SSSS
3
3
24-bit address
X:<<pp
SSS
1
1
6-bit peripheral address
X:aa
SSS
1
1
6-bit absolute short address
#xxxx
SSSS
2
2
Unsigned 16-bit immediate data
DDDDD
SSSS
1
1
Move unsigned word to register
P:(Rj)+
P:(Rj)+N
SSS
5
1
Read unsigned word from program memory.
Not allowed when the XP bit in the OMR is set.
Notes: • The absolute address operand X:xxxx is specified as a word address.
• Pointer Rn is a word pointer.
• Offsets x, xxxx, and xxxxxx are word offsets.
A-214
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MOVEU.W
MOVEU.W
Move Unsigned Word
Instruction Opcodes:
Note:
All MOVEU.W instructions only allow AGU registers as destinations.
15
MOVEU.W P:<ea_m>,SSS
1
0
0
1
0
1
0
15
MOVEU.W X:<<pp,SSS
1
15
MOVEU.W X:aa,SSS
1
15
MOVEU.W #xxxx,SSSS
1
0
0
12
11
0
1
12
11
0
1
12
11
0
1
12
11
0
0
8
7
S
S
S
0
8
7
S
S
S
1
8
7
S
S
S
1
8
7
1
0
1
1
4
3
1
1
0
1
4
3
1
p
p
p
4
3
0
p
p
p
4
3
d
d
4
3
d
d
4
3
a
a
4
3
R
1
0
0
m
R
R
p
p
p
p
p
p
0
0
0
d
d
d
iiiiiiiiiiiiiiii
15
MOVEU.W DDDDD,SSSS
1
0
0
15
MOVEU.W X:(SP–xx),SSS
1
0
1
15
MOVEU.W X:(Rn)+N,SSSS
1
1
1
1
1
15
MOVEU.W X:(Rn+xxxx),SSSS
1
12
11
0
D
12
11
1
h
12
11
1
D
12
11
1
D
D
h
D
h
8
7
D
D
8
7
h
0
8
7
D
D
D
D
8
7
D
D
D
D
0
1
0
a
1
0
1
4
3
1
0
0
R
0
d
d
d
0
a
a
a
0
1
R
R
0
R
R
M
R
R
1
0
0
0
AAAAAAAAAAAAAAAA
15
MOVEU.W X:<ea_MM>,SSSS
1
1
1
1
1
15
MOVEU.W X:xxxx,SSSS
1
12
11
1
D
12
11
1
D
8
7
D
D
D
D
8
7
D
D
D
D
4
3
0
0
M
R
4
3
1
1
1
1
4
3
0
0
AAAAAAAAAAAAAAAA
15
MOVEU.W X:(Rn+xxxxxx),SSSS
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
D
D
D
D
D
1
0
0
R
0
R
R
4
3
AAAAAAAAAAAAAAAA
15
MOVEU.W X:xxxxxx,SSSS
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
1
1
1
1
D
D
D
D
D
1
1
1
1
1
0
0
AAAAAAAAAAAAAAAA
Timing:
1–5 oscillator clock cycle(s)
Memory:
1–3 program word(s)
Freescale Semiconductor
Instruction Set Details
A-215
MPY
MPY
Signed Multiply
Operation:
Assembler Syntax:
+ S1 × S2 → D
S1 × S2 → D
S1 × S2 → D
(no parallel move)
(one parallel move)
(two parallel reads)
MPY
MPY
MPY
(+)S1,S2,D
S1,S2,D
S1,S2,D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Multiply the two signed 16-bit source operands, and place the 32-bit fractional product in the destination (D). Both source operands must be located in the FF1 portion of an accumulator or in X0, Y0, or
Y1. If an accumulator is used as the destination, the result is sign extended into the extension portion
(FF2) of the accumulator. If the destination is one of the 16-bit registers, only the higher 16 bits of the
fractional product are stored.
Usage:
This instruction is used for multiplication of fractional data or integer data when a full 32-bit product
is required (see Section 5.3.3, “Multiplication,” on page 5-18). When the destination is a 16-bit register, this instruction is useful only for fractional data.
Example:
MPY
Y0,X0,A
X:(R0)+,Y0
X:(R3)+,X0
Before Execution
; multiply X0 by Y0
After Execution
0
1000
0000
0
000A
8000
A2
A1
A0
A2
A1
A0
FF00
0200
FF00
0300
Y1
Y0
Y1
Y0
X0
02A0
X0
0288
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $02A0 (or fractional value 0.020507813),
the 16-bit Y0 register contains the value $0200 (or fractional value 0.015625). The contents of the destination register are not important prior to execution because they have no effect on the calculated value. Execution of the MPY instruction multiplies the 16-bit signed value in the X0 register by the 16-bit
signed value in Y0 (yielding the fractional product result of $000A:8000 = 0.000320435) and stores
the result back into the A accumulator. In parallel, X0 and Y0 are updated with new values that are
fetched from the data memory, and the two address registers (R0 and R3) are post-incremented by one.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
A-216
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ (parallel move)
Set if limiting (parallel move) has occurred
Set if the extended portion of the result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Always cleared
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MPY
MPY
Signed Multiply
Instruction Fields:
Operation
Operands
C
W
Comments
MPY
FFF1,FFF1,FFF
1
1
Fractional multiply.
–Y1,X0,FFF
–Y0,X0,FFF
–Y1,Y0,FFF
–Y0,Y0,FFF
–A1,Y0,FFF
–B1,Y1,FFF
–C1,Y0,FFF
–C1,Y1,FFF
1
1
Fractional multiply where one operand is
negated before multiplication.
Note: Assembler also accepts first two
operands when they are specified in opposite
order.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
MPY2
Y1,X0,F
Y0,X0,F
Y1,Y0,F
Y0,Y0,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A1,Y0,F
B1,Y1,F
C1,Y0,F
C1,Y1,F
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Freescale Semiconductor
Instruction Set Details
A-217
MPY
MPY
Signed Multiply
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
MPY2
Y1,X0,F
Y1,Y0,F
Y0,X0,F
C1,Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
MPY
FFF1,FFF1,FFF
0
1
1
15
MPY
Q1,Q2,F GGG,X:<ea_m>
0
0
0
15
MPY
Q1,Q2,F X:<ea_m>,GGG
0
0
1
15
MPY
Q3,Q4,F X:<ea_m>,reg1
X:<ea_v>,reg2
MPY
–Q1,Q2,FFF
0
1
1
1
1
15
Timing:
1 oscillator clock cycle
Memory:
1 program word
A-218
0
12
11
0
1
12
11
0
0
12
11
0
0
12
11
0
0
12
11
1
0
0
G
G
F
G
G
8
7
F
F
8
7
G
F
8
7
G
F
8
7
F
0
v
v
8
7
1
F
F
F
J
Q
Q
J
Q
Q
4
3
J
J
4
3
Q
1
4
3
Q
1
4
3
1
v
Q
Q
4
3
Q
Q
Q
0
DSP56800E and DSP56800EX Core Reference Manual
0
J
0
1
0
m
R
R
0
m
R
R
0
m
0
v
0
1
1
0
Freescale Semiconductor
MPYR
MPYR
Signed Multiply and Round
Operation:
Assembler Syntax:
+ S1 × S2 + r → D
S1 × S2 + r → D
S1 × S2 + r → D
(no parallel move)
(one parallel move)
(two parallel reads)
MPYR
MPYR
MPYR
(+)S1,S2,D
S1,S2,D
S1,S2,D
(no parallel move)
(one parallel move)
(two parallel reads)
Description: Multiply the two signed 16-bit source operands, round the 32-bit fractional product, and place the result in the destination (D). Both source operands must be located in the FF1 portion of an accumulator
or in X0, Y0, or Y1. The fractional product is sign extended before the rounding operation, and the
result is then stored in the destination. If the destination is one of the 16-bit registers, only the high-order 16 bits of the rounded fractional result are stored. This instruction uses the rounding technique that
is selected by the R bit in the OMR. When the R bit is cleared (default mode), convergent rounding is
selected; when the R bit is set, two’s-complement rounding is selected. Refer to Section 5.9, “Rounding,” on page 5-43 for more information about the rounding modes. Note that the rounding operation
will always zero the LSP of the result if the destination (D) is an accumulator or the Y register.
Usage:
This instruction is used for the multiplication and rounding of fractional data.
Example:
MPYR
Y0,X0,A
X:(R0)+,Y0
Before Execution
X:(R3)+,X0
; multiply fractional
; signed and round
After Execution
0
1000
1234
0
000A
0000
A2
A1
A0
A2
A1
A0
FF00
0200
FF00
0300
Y1
Y0
Y1
Y0
X0
02A0
X0
0288
SR
0300
SR
0310
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the value $02A0 (or fractional value 0.020507813),
and the 16-bit Y0 register contains the value $0200 (or fractional value 0.015625). The contents of the
destination register are not important prior to execution because they have no effect on the calculated
value. Execution of the MPYR instruction multiplies the 16-bit signed value in the X0 register by the
16-bit signed value in Y0 (yielding the fractional product result of $000A:8000 = 0.000320435),
rounds the result, and stores the rounded result ($0:000A:0000 = 0.000305176) back into the A accumulator. In parallel, X0 and Y0 are updated with new values that are fetched from the data memory,
and the two address registers (R0 and R3) are post-incremented by one. In this example, the default
rounding technique (convergent rounding) is performed (bit R in the OMR is cleared). If two’s-complement rounding is utilized (R bit is set), the result in accumulator A is $0:000B:0000 = 0.000335693.
Freescale Semiconductor
Instruction Set Details
A-219
MPYR
MPYR
Signed Multiply and Round
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
Set according to the standard definition of the SZ (parallel move)
Set if limiting (parallel move) has occurred
Set if the extended portion of the result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Always cleared
Instruction Fields:
Operation
Operands
C
W
Comments
MPYR
FFF1,FFF1,FFF
1
1
Fractional multiply; result rounded.
–Y1,X0,FFF
–Y0,X0,FFF
–Y1,Y0,FFF
–Y0,Y0,FFF
–A1,Y0,FFF
–B1,Y1,FFF
–C1,Y0,FFF
–C1,Y1,FFF
1
1
Fractional multiply where one operand is
negated before multiplication; result is rounded.
Note: Assembler also accepts first two
operands when they are specified in opposite
order.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
MPYR2
Y1,X0,F
Y0,X0,F
Y1,Y0,F
Y0,Y0,F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
A1,Y0,F
B1,Y1,F
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
A-220
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MPYR
MPYR
Signed Multiply and Round
Parallel Dual Reads:
Data ALU Operation1
First Memory Read
Second Memory Read
Operation
Operands
Source 1
Destination 1
Source 2
Destination 2
MPYR2
Y1,X0,F
Y1,Y0,F
Y0,X0,F
C1,Y0,F
X:(R0)+
X:(R0)+N
X:(R1)+
X:(R1)+N
Y0
Y1
X:(R3)+
X:(R3)–
X0
X:(R4)+
X:(R4)+N
Y0
X:(R3)+
X:(R3)+N3
X0
X:(R0)+
X:(R0)+N
X:(R4)+
X:(R4)+N
Y1
X:(R3)+
X:(R3)+N3
C
1.This instruction is not allowed when the XP bit in the OMR is set (that is, when the instructions are executing
from data memory).
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
MPYR
FFF1,FFF1,FFF
0
1
1
15
MPYR
Q1,Q2,F GGG,X:<ea_m>
0
0
0
15
MPYR
Q1,Q2,F X:<ea_m>,GGG
MPYR
Q3,Q4,F X:<ea_m>,reg1
X:<ea_v>,reg2
MPYR
–Q1,Q2,FFF
0
0
1
1
1
1
1
15
0
15
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
12
11
0
1
12
11
1
0
12
11
1
0
12
11
1
0
12
11
1
0
0
G
F
G
8
7
F
F
8
7
G
F
8
7
F
G
G
G
8
7
0
v
v
F
8
7
1
F
F
F
Instruction Set Details
J
Q
J
Q
4
3
J
J
4
3
Q
1
4
3
1
Q
Q
Q
4
3
v
Q
Q
1
4
3
Q
Q
Q
1
0
J
1
1
0
m
R
R
0
m
R
R
m
0
v
0
1
1
0
0
A-221
MPYSU
MPYSU
Signed × Unsigned Multiply
Operation:
Assembler Syntax:
S1 × S2 → D
(S1 signed; S2 unsigned)
MPYSU
S1,S2,D
(no parallel move)
Description: Multiply one signed 16-bit source operand by one unsigned 16-bit operand, and place the 32-bit fractional product in the destination (D). The order of the registers is important. The first source register
(S1) must contain the signed value, and the second source (S2) must contain the unsigned value to produce correct fractional results. If the destination is one of the 16-bit registers, only the high-order
16 bits of the fractional result are stored. The result is not affected by the state of the saturation bit
(SA). Note that for 16-bit destinations, the sign bit may be lost for large fractional magnitudes.
Usage:
In addition to single-precision multiplication of a signed value times an unsigned value, this instruction
is also used for multi-precision multiplications, as shown in Section 5.5, “Extended- and Multi-Precision Operations,” on page 5-29.
Example:
MPYSU
X0,Y0,A
; multiply signed X0 by unsigned Y0
; and store the result in A
Before Execution
After Execution
0
0000
0000
F
FFFF
FFD0
A2
A1
A0
A2
A1
A0
2000
0002
2000
0002
Y1
Y0
Y1
Y0
X0
FFF4
X0
FFF4
SR
0300
SR
0318
Explanation of Example:
Prior to execution, the 16-bit X0 register contains the (signed) negative value $FFF4, and the 16-bit
Y0 register contains the (unsigned) positive value $0002. The contents of the destination register are
not important prior to execution because they have no effect on the calculated value. Execution of the
MPYSU instruction multiplies the 16-bit signed value in the X0 register by the 16-bit unsigned value
in Y0 (yielding the fractional product result of $FFFF:FFD0) and stores the signed result
($F:FFFF:FFD0) back into the A accumulator.
Condition Codes Affected:
MR
L
E
U
N
Z
V
A-222
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if overflow has occurred in result
Set if the extended portion of the result is in use
Set according to the standard definition of the U bit
Set if MSB of result is set
Set if result equals zero
Set if overflow has occurred in result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
MPYSU
MPYSU
Signed × Unsigned Multiply
Instruction Fields:
Operation
Operands
C
W
Comments
MPYSU
X0,Y1,EEE
X0,Y0,EEE
Y0,Y1,EEE
Y0,Y0,EEE
Y0,A1,EEE
Y1,B1,EEE
Y0,C1,EEE
Y1,C1,EEE
1
1
16 × 16 => 32-bit signed-and-unsigned fractional multiply.
The first operand is treated as signed and the second as
unsigned.
Instruction Opcodes:
15
MPYSU
Q2,Q1,EEE
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
1
0
1
E
Instruction Set Details
8
7
E
E
Q
Q
4
3
Q
1
0
0
1
0
A-223
NEG
Operation:
0–D→
0–D→
NEG
Negate Register
Assembler Syntax:
D
D
(no parallel move)
(one parallel move)
NEG
NEG
D
D
(no parallel move)
(one parallel move)
Description: The destination operand (D) is subtracted from zero, and the two’s-complement result is stored in the
destination (D). If the destination is a 16-bit register, it is first sign extended internally and concatenated with 16 zero bits to form a 36-bit operand (the Y register is only sign extended).
Example:
NEG
B
X0,X:(R3)+
; 0 - B → B, save X0, update R3
Before Execution
After Execution
0
00AA
FF00
F
FF55
0100
B2
B1
B0
B2
B1
B0
SR
SR
0300
0319
Explanation of Example:
Prior to execution, the 36-bit B accumulator contains the value $0:00AA:FF00. The NEG instruction
takes the two’s-complement of the value in the B accumulator and stores the 36-bit result
($F:FF55:0100) back in the B accumulator. The value for X0 is stored in memory and the address register R3 is post-incremented by one. The N bit is set because the result is negative.
Note:
When the D operand equals $8:0000:0000 (–16.0 when interpreted as a decimal fraction), the NEG
instruction causes an overflow to occur since the result cannot be correctly expressed using the standard 36-bit, fixed-point, two’s-complement data representation. When saturation is enabled (the OMR
register’s SA bit is set to one), data limiting will occur to value $F:8000:0000. If saturation is not enabled, the value will remain unchanged.
Condition Codes Affected:
MR
SZ
L
E
U
N
Z
V
C
A-224
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
—
—
Set according to the standard definition of the SZ (parallel move)
Set if limiting (parallel move) or overflow has occurred in result
Set if the extension portion of result is in use
Set according to the standard definition of the U bit
Set if bit MSB of result is set
Set if the result equals zero
Set if overflow has occurred in the result
Set if a borrow is generated from the MSB of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
NEG
NEG
Negate Register
Instruction Fields:
Operation
Operands
C
W
NEG
FFF
1
1
Comments
Two’s-complement negation.
Parallel Moves:
Data ALU Operation
Parallel Memory Move
Operation
Operands
Source
Destination1
NEG2
F
X:(Rj)+
X:(Rj)+N
X0
Y1
Y0
A
B
C
A1
B1
X0
Y1
Y0
A
B
C
A1
B1
X:(Rj)+
X:(Rj)+N
1.The case where the destination of the data ALU operation is the same register as the destination of the parallel read operation is not allowed. Memory writes are allowed in this case.
2.This instruction occupies only 1 program word and executes in 1 cycle for every addressing mode.
Instruction Opcodes:
15
NEG
F GGG,X:<ea_m>
NEG
F X:<ea_m>,GGG
0
0
0
0
1
15
0
15
NEG
FFF
Timing:
1 oscillator clock cycle
Memory:
1 program word
Freescale Semiconductor
0
1
1
12
11
0
1
12
11
0
1
12
11
1
1
8
7
G
G
G
F
8
7
G
G
G
F
8
7
F
F
1
F
Instruction Set Details
4
3
0
0
1
0
4
3
0
0
1
0
4
3
b
1
b
b
0
m
R
R
m
R
R
0
0
1
1
1
A-225
NEG.BP
Operation:
0 – D→ D
NEG.BP
Negate Byte (Byte Pointer)
Assembler Syntax:
(no parallel move)
NEG.BP
D
(no parallel move)
Description: Compute the two’s-complement of a byte value in memory. The value is internally sign extended to
20 bits before being negated. The low-order 8 bits of the result are stored back to memory. The condition codes are calculated based on the 8-bit result, with the exception of the E and U bits, which are
calculated based on the 20-bit result. Absolute addresses are expressed as byte addresses. The result is
not affected by the state of the saturation bit (SA).
Usage:
This instruction is typically used when integer data is processed.
Example:
NEG.BP X:$3065
; negate the byte at (byte) address $3065
Before Execution
After Execution
Byte
Addresses
Byte
Addresses
X Memory
07
7
0
$3068
$3066
$3064
$3062
SR
88
66
44
77
55
33
22
11
7
$3068
$3066
$3064
$3062
SR
030F
X Memory
07
0
88
66
BC
77
55
33
22
11
0319
Explanation of Example:
Prior to execution, the value at byte address X:$3065 is $44. Execution of the NEG.BP instruction computes the two’s-complement of this value and generates the result $BC with a borrow (the carry bit is
set). The result is negative since bit 7 is set. Note that this address is equivalent to the upper byte of
word address $1832.
Condition Codes Affected:
MR
E
U
N
Z
V
C
A-226
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 20-bit result is in use
Set if the 20-bit result is unnormalized
Set if bit 7 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a borrow occurs from bit 7 of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
NEG.BP
NEG.BP
Negate Byte (Byte Pointer)
Instruction Fields:
Operation
Operands
C
W
NEG.BP
X:xxxx
3
2
X:xxxxxx
4
3
Comments
Negate byte in memory
Instruction Opcodes:
15
NEG.BP X:xxxx
0
1
0
12
11
0
1
1
1
8
7
1
0
1
0
4
3
0
0
4
3
0
1
1
0
AAAAAAAAAAAAAAAA
15
NEG.BP X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
1
0
1
0
0
0
1
1
0
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-227
NEG.L
Operation:
0 – D→ D
NEG.L
Negate Long
Assembler Syntax:
(no parallel move)
NEG.L
D
(no parallel move)
Description: Compute the two’s-complement of a long-word value in memory. When an operand located in memory is operated on, the low-order 32 bits of the result are stored back to memory. The condition codes
are calculated based on the 32-bit result. Absolute addresses pointing to long elements must always be
even aligned (that is, pointing to the lowest 16 bits).
Usage:
This instruction is typically used when integer data is processed.
Example:
NEG.L
X:$2000
; negate the long word at address $2001:2000
Before Execution
After Execution
Word
Addresses
Word
Addresses
X Memory
15
0
X Memory
15
0
0000
$2003
0000
$2002
$2001
$2000
0001
00AA
FF00
$2001
$2000
0001
FF55
0100
SR
030F
SR
0319
$2003
$2002
Explanation of Example:
Prior to execution, the 32-bit value at location $2001:2000 is $00AA:FF00. Execution of the
NEG.L instruction computes the two’s-complement of this value and generates $FF55:0100. The CCR
is updated based on the result of the subtraction.
Condition Codes Affected:
MR
E
U
N
Z
V
C
A-228
CCR
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
LF
P4
P3
P2
P1
P0
I1
I0
SZ
L
E
U
N
Z
V
C
—
—
—
—
—
—
Set if the extension portion of the 36-bit result is in use
Set if the 36-bit result is unnormalized
Set if bit 31 of the result is set
Set if the result is zero
Set if overflow has occurred in result
Set if a borrow occurs from bit 31 of the result
DSP56800E and DSP56800EX Core Reference Manual
Freescale Semiconductor
NEG.L
NEG.L
Negate Long
Instruction Fields:
Operation
Operands
C
W
NEG.L
X:xxxx
3
2
X:xxxxxx
4
3
Comments
Negate long in memory
Instruction Opcodes:
15
NEG.L
X:xxxx
0
1
0
12
11
0
1
1
1
8
7
1
0
1
0
4
3
0
0
4
3
0
1
1
1
AAAAAAAAAAAAAAAA
15
NEG.L
X:xxxxxx
12
11
8
7
0
1
1
1
0
0
A
A
A
0
A
1
1
A
A
A
A
0
1
0
0
1
1
1
1
0
1
0
0
0
1
1
1
AAAAAAAAAAAAAAAA
Timing:
3–4 oscillator clock cycles
Memory:
2–3 program words
Freescale Semiconductor
Instruction Set Details
A-229
NEG.W
Operation:
0 – D→ D
NEG.W
Negate Word
Assembler Syntax:
(no parallel move)
NEG.W
D
(no parallel move)
Description: Compute the two’s-complement of a word value in memory. The value is internally sign extended to
20 bits before being subtracted from zero. The low-order 16 bits of the result are stored back to memory. The condition codes are calculated based on the 16-bit result, with the exception of the E and U
bits, which are calculated based on the 20-bit result.
Usage:
This instruction is typically used when integer data is processed.
Example