Assembler

Jian-hua Yeh (葉建華)

真理大學資訊科學系助理教授
[email protected]
 Role of Assembler

Source Object
Assembler Linker
Program Code

Executable
Code

Loader

2
 Chapter 2 -- Outline
• Basic Assembler Functions
• Machine-dependent Assembler Features
• Machine-independent Assembler Features
• Assembler Design Options

3
 Introduction to Assemblers
• Fundamental functions
– translating mnemonic operation codes to their machine
language equivalents
– assigning machine addresses to symbolic labels

• Machine dependency
– different machine instruction formats and codes

4
 Example Program (Fig. 2.1)

• Purpose
– reads records from input device (code F1)
– copies them to output device (code 05)
– at the end of the file, writes EOF on the output device, then
RSUB to the operating system

5
 Example Program (Fig. 2.1)
• Data transfer (RD, WD)
– a buffer is used to store record
– buffering is necessary for different I/O rates
– the end of each record is marked with a null character (0016)
– the end of the file is indicated by a zero-length record

• Subroutines (JSUB, RSUB)

– RDREC, WRREC
– save link register first before nested jump

6
 Assembler Directives
• Pseudo-Instructions
– Not translated into machine instructions
– Providing information to the assembler

• Basic assembler directives

– START
– END
– BYTE
– WORD
– RESB
– RESW

7
 Object Program
• Header
Col. 1 H
Col. 2~7 Program name
Col. 8~13 Starting address (hex)
Col. 14-19 Length of object program in bytes (hex)

• Text
Col.1 T
Col.2~7 Starting address in this record (hex)
Col. 8~9 Length of object code in this record in bytes (hex)
Col. 10~69 Object code (69-10+1)/6=10 instructions

• End
Col.1 E
Col.2~7 Address of first executable instruction (hex)
8
(END program_name)
 Fig. 2.3
H COPY 001000 00107A
T 001000 1E 141033 482039 001036 281030 301015 482061 ...
T 00101E 15 0C1036 482061 081044 4C0000 454F46 000003 000000
T 002039 1E 041030 001030 E0205D 30203F D8205D 281030 …
T 002057 1C 101036 4C0000 F1 001000 041030 E02079 302064 …
T 002073 07 382064 4C0000 05
E 001000

9
 Figure 2.1 (Pseudo code)
Program copy {
save return address;
cloop: call subroutine RDREC to read one record;
if length(record)=0 {
call subroutine WRREC to write EOF;
} else {
call subroutine WRREC to write one record;
goto cloop;
}
load return address
return to caller
}

10
 An Example (Figure 2.1, Cont.)
EOR:
Subroutine RDREC { character x‘00’
clear A, X register to 0;
rloop: read character from input device to A register
if not EOR {
store character into buffer[X];
X++;
if X < maximum length
goto rloop;
}
store X to length(record);
return
}

11
 An Example (Figure 2.1, Cont.)

Subroutine WDREC {
clear X register to 0;
wloop: get character from buffer[X]
write character from X to output device
X++;
if X < length(record)
goto wloop;
return
}

12
 Assembler’s functions

• Convert mnemonic operation codes to their machine

language equivalents
• Convert symbolic operands to their equivalent
machine addresses 9
• Build the machine instructions in the proper format
• Convert the data constants to internal machine
representations
• Write the object program and the assembly listing

13
 Example of Instruction Assemble

STCH BUFFER,X 549039

8 1 15
opcode x address
m
(54)16 1 (001)2 (039)16

• Forward reference

14
 Difficulties: Forward Reference
• Forward reference: reference to a label that is
defined later in the program.

Loc Label Operator Operand

1000 FIRST STL RETADR

1003 CLOOP JSUB RDREC
… … … … …
1012 J CLOOP
… … … … …
1033 RETADR RESW 1

15
 Two Pass Assembler
• Pass 1
– Assign addresses to all statements in the program
– Save the values assigned to all labels for use in Pass 2
– Perform some processing of assembler directives

• Pass 2
– Assemble instructions
– Generate data values defined by BYTE, WORD
– Perform processing of assembler directives not done in Pass 1
– Write the object program and the assembly listing

16
 Two Pass Assembler
• Read from input line
– LABEL, OPCODE, OPERAND

Source
program

Intermediate Object
Pass 1 Pass 2
file codes

OPTAB SYMTAB SYMTAB

17
 Data Structures

• Operation Code Table (OPTAB)

• Symbol Table (SYMTAB)
• Location Counter(LOCCTR)

18
 OPTAB (operation code table)

• Content
– menmonic, machine code (instruction format, length) etc.

• Characteristic
– static table

• Implementation
– array or hash table, easy for search

19
 SYMTAB (symbol table)
• Content COPY 1000
– label name, value, flag, (type, length) etc.FIRST 1000
CLOOP 1003
• Characteristic ENDFIL 1015
EOF 1024
– dynamic table (insert, delete, search)
THREE 102D
• Implementation ZERO 1030
RETADR 1033
– hash table, non-random keys, hashing function
LENGTH 1036
BUFFER 1039
RDREC 2039

20
 Assembler Design
• Machine Dependent Assembler Features
– instruction formats and addressing modes
– program relocation
• Machine Independent Assembler Features
– literals
– symbol-defining statements
– expressions
– program blocks
– control sections and program linking

21
 Machine Dependent Assembler Features

• Instruction Format and Addressing Mode

– PC-relative or Base-relative addressing: op m
– Indirect addressing: op @m
– Immediate addressing: op #c
– Extended format: +op m
– Index addressing: op m,x
– register-to-register instructions
– larger memory -> multi-programming (program allocation)

22
 Translation

• Register translation
– register name (A, X, L, B, S, T, F, PC, SW) and their values
(0,1, 2, 3, 4, 5, 6, 8, 9)
– preloaded in SYMTAB

• Address translation
– Most register-memory instructions use program counter
relative or base relative addressing
– Format 3: 12-bit address field
• base-relative: 0~4095
• pc-relative: -2048~2047
– Format 4: 20-bit address field 23
• pc-relative first
 Relative Addressing Modes

• PC-relative
– e.g. 10 0000 FIRST STL RETADR 17202D
• displacement= RETADR - PC = 30-3 = 2D
– e.g. 40 0017 J CLOOP 3F2FEC
• displacement= CLOOP - PC = 6 - 1A = -14 = FEC

• Base-relative
– base register is under the control of the programmer
– e.g. 12 LDB #LENGTH
– e.g. 13 BASE LENGTH
– e.g. 160 104E STCH BUFFER, X 57C003
24
• displacement= BUFFER - B = 0036 - 0033 = 3
 Address Translation

• Immediate addressing
– e.g. 55 0020 LDA #3 010003
– e.g. 133 103C +LDT #4096 75101000
– e.g. 12 0003 LDB #LENGTH 69202D
• the immediate operand is the symbol LENGTH
• the address of this symbol LENGTH is loaded into register B
• LENGTH=0033=PC+displacement=0006+02D
• if immediate mode is specified, the target address becomes the
operand

25
 Address Translation (Cont.)
• Indirect addressing
– target addressing is computed as usual (PC-relative or BASE-
relative)
– only the n bit is set to 1
e.g. 70 002A J @RETADR 3E2003

26
 Program Relocation

• Example Fig. 2.1

– Absolute program, starting address 1000
e.g. 55 101B LDA THREE 00102D
– Relocate the program to 2000
e.g. 55 101B LDA THREE 00202D
– Each Absolute address should be modified

• Example Fig. 2.5:

– Except for absolute address, the rest of the instructions need
not be modified
• not a memory address (immediate addressing)
• PC-relative, Base-relative 27
– The only parts of the program that require modification at
Example

28
 Relocatable Program
• Modification record
– Col 1 M
– Col 2-7 Starting location of the address field to be
modified, relative to the beginning of the program
– Col 8-9 length of the address field to be modified, in half-
bytes

29
Object Code

30
 Machine-Independent Assembler
Features
• Literals
• Symbol Defining Statement
• Expressions
• Program Blocks
• Control Sections and Program Linking

31
 Literals
• Design idea
– Let programmers to be able to write the value of a constant
operand as a part of the instruction that uses it.
– This avoids having to define the constant elsewhere in the
program and make up a label for it.

• Example
– e.g. 45 001A ENDFIL LDA =C’EOF’ 032010
– 93 LTORG
– 002D * =C’EOF’ 454F46
– e.g. 215 1062 WLOOP TD =X’05’ E32011

32
 Literals vs. Immediate Operands
• Immediate Operands
– The operand value is assembled as part of the machine
instruction
– e.g. 55 0020 LDA #3 010003

• Literals
– The assembler generates the specified value as a constant
at some other memory location
– e.g. 45 001A ENDFIL LDA =C’EOF’ 032010

• Compare (Fig. 2.6)

– e.g. 45 001A ENDFIL LDA EOF 032010
– 80 002D EOF BYTE C’EOF’ 454F46

33
 Literal - Implementation (1/3)
• Literal pools
– Normally literals are placed into a pool at the end of the
program
• see Fig. 2.10 (END statement)
– In some cases, it is desirable to place literals into a pool at
some other location in the object program
• assembler directive LTORG
• reason: keep the literal operand close to the instruction

34
 Literal - Implementation (2/3)
• Duplicate literals
– e.g. 215 1062 WLOOP TD =X’05’
– e.g. 230 106B WD =X’05’
– The assemblers should recognize duplicate literals and store
only one copy of the specified data value
• Comparison of the defining expression
– Same literal name with different value, e.g. LOCCTR=*
• Comparison of the generated data value
– The benefits of using generate data value are usually not great
enough to justify the additional complexity in the assembler

35
 Literal - Implementation (3/3)

• LITTAB
– literal name, the operand value and length, the address
assigned to the operand

• Pass 1
– build LITTAB with literal name, operand value and length,
leaving the address unassigned
– when LTORG statement is encountered, assign an address
to each literal not yet assigned an address

• Pass 2
– search LITTAB for each literal operand encountered
– generate data values using BYTE or WORD statements 36

– generate modification record for literals that represent an

 Symbol-Defining Statements
• Labels on instructions or data areas
– the value of such a label is the address assigned to the
statement

• Defining symbols
– symbol EQU value
– value can be: ‘ constant, ’ other symbol, “ expression
– making the source program easier to understand
– no forward reference

37
 Symbol-Defining Statements
• Example 1
– MAXLEN EQU 4096
– +LDT #MAXLEN
+LDT #4096
• Example 2
– BASE EQU R1
– COUNT EQU R2
– INDEX EQU R3

• Example 3
– MAXLEN EQU BUFEND-BUFFER

38
 ORG (origin)
• ORG
– indirectly assign values to symbols
– reset the location counter to the specified value
– ORG value
– value can be: ‘ constant, ’ other symbol, “ expression
– no forward reference

• Example SYMBOL VALUE FLAGS

STAB
– SYMBOL: 6bytes
(100 entries)
– VALUE: 1word
– FLAGS: 2bytes . . .
. . .
. . .
– LDA VALUE, X
39
 ORG Example
• Using EQU statements
– STAB RESB 1100
– SYMBOL EQU STAB
– VALUE EQU STAB+6
– FLAG EQU STAB+9

• Using ORG statements

– STAB RESB 1100
– ORG STAB
– SYMBOL RESB 6
– VALUE RESW 1
– FLAGS RESB 2
– ORG STAB+1100 40
 Expressions
• Expressions can be classified as absolute expressions or
relative expressions
– MAXLEN EQU BUFEND-BUFFER
– BUFEND and BUFFER both are relative terms, representing
addresses within the program
– However the expression BUFEND-BUFFER represents an absolute
value
• When relative terms are paired with opposite signs, the
dependency on the program starting address is canceled out;
the result is an absolute value

41
 SYMTAB
• None of the relative terms may enter into a multiplication or
division operation
• Errors:
– BUFEND+BUFFER
– 100-BUFFER
– 3*BUFFER
• The type of an expression
– keep track of the types of all symbols defined in the program

Symbol Type Value

RETADR R 30
BUFFER R 36
BUFEND R 1036
MAXLEN A 1000
42
 Example 2.9

SYMTAB Name Value LITTAB

COPY 0 C'EOF' 454F46 3 002D
FIRST 0 X'05' 05 1 1076
CLOOP 6
ENDFIL 1A
RETADR 30
LENGTH 33
BUFFER 36
BUFEND 1036
MAXLEN 1000
RDREC 1036
RLOOP 1040
EXIT 1056
INPUT 105C
WREC 105D
WLOOP 1062 43
 Program Blocks
• Program blocks
– refer to segments of code that are rearranged within a single
object program unit
– USE [blockname]
– At the beginning, statements are assumed to be part of the
unnamed (default) block
– If no USE statements are included, the entire program
belongs to this single block
– Example: Figure 2.11
– Each program block may actually contain several separate
segments of the source program

44
 Program Blocks - Implementation

• Pass 1
– each program block has a separate location counter
– each label is assigned an address that is relative to the start
of the block that contains it
– at the end of Pass 1, the latest value of the location counter
for each block indicates the length of that block
– the assembler can then assign to each block a starting
address in the object program

• Pass 2
– The address of each symbol can be computed by adding the
assigned block starting address and the relative address of
the symbol to that block 45
 Figure 2.12
• Each source line is given a relative address assigned
and a block number

Block name Block number Address Length

(default) 0 0000 0066
CDATA 1 0066 000B
CBLKS 2 0071 1000
• For absolute symbol, there is no block number
– line 107
• Example
– 20 0006 0 LDA LENGTH 032060
– LENGTH=(Block 1)+0003= 0066+0003= 0069
– LOCCTR=(Block 0)+0009= 0009 46
 Program Readability
• Program readability
– No extended format instructions on lines 15, 35, 65
– No needs for base relative addressing (line 13, 14)
– LTORG is used to make sure the literals are placed ahead of
any large data areas (line 253)

• Object code
– It is not necessary to physically rearrange the generated
code in the object program
– see Fig. 2.13, Fig. 2.14

47
 Control Sections and Program Linking

• Control sections
– can be loaded and relocated independently of the others
– are most often used for subroutines or other logical
subdivisions of a program
– secname CSECT
– the programmer can assemble, load, and manipulate each of
these control sections separately
– because of this, there should be some means for linking
control sections together
– example: instruction in one control section may need to refer
to instructions or data located in another section
– Fig. 2.15, 2.16
48
 External Definition and References

• External definition
– EXTDEF name [, name]
– EXTDEF names symbols that are defined in this control section
and may be used by other sections

• External reference
– EXTREF name [,name]
– EXTREF names symbols that are used in this control section and
are defined elsewhere

• Example
– 15 0003 CLOOP +JSUB RDREC 4B100000
– 160 0017 +STCH BUFFER,X 5790000049
– 190 0028 MAXLEN WORD BUFEND-BUFFER 000000
 Implementation

• The assembler must include information in the object program that will
cause the loader to insert proper values where they are required
• Define record
– Col. 1 D
– Col. 2-7 Name of external symbol defined in this control section
– Col. 8-13 Relative address within this control section (hexadeccimal)
– Col.14-73 Repeat information in Col. 2-13 for other external symbols
• Refer record
– Col. 1 D
– Col. 2-7 Name of external symbol referred to in this control section
– Col. 8-73 Name of other external reference symbols
50
 Modification Record
• Modification record
– Col. 1 M
– Col. 2-7 Starting address of the field to be modified (hexiadecimal)
– Col. 8-9 Length of the field to be modified, in half-bytes
(hexadeccimal)
– Col.11-16 External symbol whose value is to be added to or
subtracted from the indicated field
– Note: control section name is automatically an external symbol, i.e.
it is available for use in Modification records.
• Example
– Figure 2.17
– M00000405+RDREC
– M00000705+COPY
51
 External References in Expression

• Earlier definitions
– required all of the relative terms be paired in an expression
(an absolute expression), or that all except one be paired (a
relative expression)

• New restriction
– Both terms in each pair must be relative within the same
control section
– Ex: BUFEND-BUFFER
– Ex: RDREC-COPY

• In general, the assembler cannot determine whether or not the

expression is legal at assembly time. This work will be handled
by a linking loader. 52
 Assembler Design Options
• One-pass assemblers
• Multi-pass assemblers
• Two-pass assembler with overlay structure

53
Two-Pass Assembler with Overlay Structure

• For small memory

– pass 1 and pass 2 are never required at the same time
– three segments
• root: driver program and shared tables and subroutines
• pass 1
• pass 2
– tree structure
– overlay program

54
 One-Pass Assemblers
• Main problem
– forward references
• data items
• labels on instructions

• Solution
– data items: require all such areas be defined before they are
referenced
– labels on instructions: no good solution

55
 One-Pass Assemblers
• Main Problem
– forward reference
• data items
• labels on instructions

• Two types of one-pass assembler

– load-and-go
• produces object code directly in memory for immediate
execution
– the other
• produces usual kind of object code for later execution

56
 Load-and-go Assembler
• Characteristics
– Useful for program development and testing
– Avoids the overhead of writing the object program out and
reading it back
– Both one-pass and two-pass assemblers can be designed
as load-and-go.
– However one-pass also avoids the over head of an
additional pass over the source program
– For a load-and-go assembler, the actual address must be
known at assembly time, we can use an absolute program

57
 Forward Reference in One-pass Assembler

• For any symbol that has not yet been defined

1. omit the address translation
2. insert the symbol into SYMTAB, and mark this symbol
undefined
3. the address that refers to the undefined symbol is added to a
list of forward references associated with the symbol table
entry
4. when the definition for a symbol is encountered, the proper
address for the symbol is then inserted into any instructions
previous generated according to the forward reference list

58
 Load-and-go Assembler (Cont.)
• At the end of the program
– any SYMTAB entries that are still marked with * indicate
undefined symbols
– search SYMTAB for the symbol named in the END
statement and jump to this location to begin execution

• The actual starting address must be specified at

assembly time
• Example
– Figure 2.18, 2.19

59
 Producing Object Code
• When external working-storage devices are not
available or too slow (for the intermediate file
between the two passes
• Solution:
– When definition of a symbol is encountered, the assembler
must generate another Tex record with the correct operand
address
– The loader is used to complete forward references that could
not be handled by the assembler
– The object program records must be kept in their original
order when they are presented to the loader

• Example: Figure 2.20

60
 Multi-Pass Assemblers
• Restriction on EQU and ORG
– no forward reference, since symbols’ value can’t be defined
during the first pass

• Example
– Use link list to keep track of whose value depend on an
undefined symbol

• Figure 2.21

61
 Implementation Examples
• Microsoft MASM Assembler
• Sun Sparc Assembler
• IBM AIX Assembler

62
 Microsoft MASM Assembler
• SEGMENT
– a collection segments, each segment is defined as belonging
to a particular class, CODE, DATA, CONST, STACK
– registers: CS (code), SS (stack), DS (data), ES, FS, GS
– similar to program blocks in SIC

• ASSUME
– e.g. ASSUME ES:DATASEG2

– e.g. MOVE AX, DATASEG2

MOVE ES,AX

– similar to BASE in SIC

63
 Microsoft MASM Assembler
• JUMP with forward reference
– near jump: 2 or 3 bytes
– far jump: 5 bytes
– e.g. JMP TARGET

– Warning: JMP FAR PTR TARGET

– Warning: JMP SHORT TARGET

– Pass 1: reserves 3 bytes for jump instruction

– phase error

• PUBLIC, EXTRN
– similar to EXTDEF, EXTREF in SIC

64
 Sun Sparc Assembler
• Sections
– .TEXT, .DATA, .RODATA, .BSS

• Symbols
– global vs. weak
– similar to the combination of EXTDEF and EXTREF in SIC

• Delayed branches
– delayed slots
– annulled branch instruction

65
 Sun Sparc Assembler
LOOP: .
.
LOOP: .
ADD %L2, %L3, %L4
.
CMP %L0, 10
.
BLE LOOP
ADD %L2, %L3, %L4
.
CMP %L0, 10
. BLE LOOP
. NOP
CMP %L0, 10
BLE LOOP
ADD %L2, %L3, %L4 66

.
 Sun Sparc Assembler

LOOP: ADD %L2, %L3, %L4

.
.
CMP %L0, 10
BLE,A LOOP LOOP: .
. .
.
CMP %L0, 10
BLE,A LOOP
ADD %L2, %L3, %L4
67
 AIX Assembler for PowerPC

• Similar to System/370
• Base relative addressing
– save instruction space, no absolute address
– base register table:
• general purpose registers can be used as base register
– easy for program relocation
• only data whose values are to be actual address needs to be
modified
– e.g. USING LENGTH, 1

– USING BUFFER, 4
68
– Similar to BASE in SIC
 AIX Assembler for PowerPC
• Alignment
– instruction (2)
– data: halfword operand (2), fullword operand (4)
– Slack bytes

• .CSECT
– control sections: RO(read-only data), RW(read-write data),
PR(executable instructions), BS(uninitialized read/write data)
– dummy section

69