Digital PDFs

DEC-08-ARXB-D

March 1970

108 pages

Original

3.1MB

Document:	8K SABR Assembler 197003
Order Number:	DEC-08-ARXB-D
Revision:
Pages:	108
Original Filename:	http://bitsavers.org/pdf/dec/pdp8/fortran/DEC-08-ARXB-D_8K_SABR_Assembler_197003.pdf

OCR Text

Digital Equipment Corporation
Maynard, Massachusetts

PDP-8 Family
Programmer's Reference

Manual

BK SABR
ASSEMBLER

DEC-08-ARXB-D

PDP-8
BK SABR ASSEMBLER
PROGRAMMER'S REFERENCE MANUAL

For additional copies of this document, order No. DEC-08-ARXB-D from Program Library,
Digital Equipment Corporation, Maynard, N"oss. 01754

Price $2 .00

DIGITAL EQUIPMENT CORPORATION o MAYNARD, MASSACHUSETTS

1st Printing Apri I 1969
2nd Printing (Rev) March 1970

Your attention is invited to the last two pages of this
manual. The Reader 1s Comments page, when filled in
and returned, is beneficial to both you and DEC. All
comments received are considered when documenting
subsequent manuals, and when assistance is required,
a knowledgeable DEC representative wi 11 contact you.
The Software Information page offers you a means of
keeping up-to-date with DEC's software.

Documents Referenced {avai Iable from DEC's Program Library):
Introduction to Programming, C-18
8K FORTRAN Programmer's Reference Manual, DEC-08-KFXB-D
Paper Tape System User's Guide, DEC-08-NGCC-D
PDP-8/I Disk Monitor System, DEC-08-SDAB-D

The following are trademarks of Digital Equipment Corporation,
Maynard, Massachusetts:
PDP
FLIP CHIP
DIGITAL

DEC
FOCAL
COMPUTER LAB

CONTENTS
Page
CHAPTER 1
THE SABR LANGUAGE
1.1

Introduction

1-1

1.2

The Character Set

1-2

1.3

Statements

1-2

l .3. l

Labels

1-3

1 .3 .2

Operators

1-3

1 .3 .3

Operands

1-4

1 .3 .4

Comments

1-4

1 .3 .5

Format Effectors

1-5

1.4

1-6

Symbols

1 .4. 1

Permanent Symbo Is

1-6

1 .4.2

User-Defined Symbo Is

1-6

1 .4.3

Equivalent Symbols

1-6

1 .4.4

Incrementing Operands

1-7

1 .4.5

The Symbol Table Listing

1-8

1.5

1-8

Constants

l .5. l

Numeric Constants

1-9

l .5 .2

ASCII Constants

1-9

1.6

1-9

Litera Is
CHAPTER 2
PSEUDO-OPERATORS

2. l

Assembly Control

2-1

2.2

Symbol Definition

2-4

2.3

Data Generating

2-5

2.4

Externa I Subroutine

2-7

2 .4.1

The CALL and ARG Statements

2-8

2.4.2

The ENTRY and DUMMY Statements

2-9

2 .4.3

The RETRN Statement

2-10

2.4.4

Picking up Subprogram Arguments

2-10

iii

CONTENTS (Cont)
Page
CHAPTER 3
THE ASSEMBLED BINARY CODE
3 .1

The Binary Output Tape

3-1

3.2

The Loader Relocation Codes

3-1

3.3

Page Assembly

3-3

3 .3.1

Page Format

3-4

3.3.2

Page Escapes

3-4

3.3.3

Multiple Word Instructions

3-5

3.3.4

Run-Time Linkage Routines

3-5

3.3.5

Skip Instructions

3-7

3.4

Program Addresses

3-8

3.5

The Symbol Table

3-8
CHAPTER 4
SABR OPERATING PROCEDURES

4.1

Loading SABR in a Basic PDP-8 System

4-1

4.2

Loading SABR in a Disk Monitor System

4-1

4.3

Operating SABR

4-2

4.4

Operating Procedure for use as FORTRAN Pass 2

4-3

4.4.1

Method 1

4-4

4.4.2

Method 2

4-4

4.4.3

Method 3

4-5
CHAPTER 5
THE LINKING LOADER

5 .1

Introduction

5-1

5.2

Loading with the Linking Loader

5-1

5.3

Loading Information Options

5-2

5.4

How to Load the Linking Loader

5-3

5.5

Operation of the Linking Loader

5-3

CONTENTS (Cont)

CHAPTER 6
DEMONSTRATION PROGRAM
APPENDIX A
ASCII CHARACTER SET
APPENDIX B
PERMANENT SYMBOL TABLE
APPENDIX C
ERROR MESSAGES
APPENDIX D
FREE PAGE 0 LOCATIONS
APPENDIX E
THE LIBRARY SUBPROGRAMS
APPENDIX F
SAMPLE OF AN ASSEMBLY LISTING
APPENDIX G
OPERATING PROCEDURES
APPENDIX H
DISK LINKING LOADER

PREFACE

This manual contains a detailed description of the 8K SABR Symbolic Assembly
System. The SABR (Symbolic Assembler for Binary Relocatable programs) programming language is similar to that of PAL III with many additional features. It is an
advanced one-pass assembler for use with a PD P-8/I, -8/L, -8, -8/S, or -5 computer with at least 8K (up to 32K) words of core memory and an ASR-33 Teletype;
a high-speed photoelectric paper tape reader and punch is not required, although
it is highly recommended.
DEC offers four symbolic assemblers for use on PDP-8 family computers: as follows:
a.
PAL III Symbolic Assembler, the basic 4K assembly system. It is a twopass assembler with an optional third pass which produces an octal/symbolic
assembly program listing, and is highly recommended for the computer with 4K
words of core memory. It is an excel lent assembly language for the less experienced programmer yet powerfu I enough to satisfy the needs of the advanced
programmer.
b. MACR0-8 Symbolic Assembler, essentially PAL III with the following additional features: user-defined macros, double precision integers, floatingpoint constants, arithmetic and Boolean operators, literals, text facilities,
and automatic off-page I inkage generation. It is recommended for the computer with 4K words of core memory when any of the additional features listed
above are desired.
c. PAL-D Symbolic Assembler, essentially MACR0-8 excluding macros. It
is used only in the PDP-8/I Disk Monitor System and requires 4K words of core
memory.
d; 8K SABR Symbolic Assembler; primarily for experienced programmers to
use with a computer that has 8K to 32K words of core memory. It differs from
the preceding assemblers in its operating procedures, character set, pseudo-ops,
execution of the assembled program, and especially in its assembled output
(relocatable binary code).
It is assumed that the reader is familiar with assembly language programming. For
an elementary approach to this type of programming, we recommend D'EC's publication, Introduction To Programming (specify Order No. C-18), avai Iable from the
Program Library (address on Title page).

vii

CHAPTER l
THE SABR LANGUAGE

1. 1

INTRODUCTiON
SABR (Symbolic Assembler for Binary Reiocatable programs) is an advanced one-pass symbolic

assembler. It translates symbolic programs written in the SABR language into binary relocatable code
acceptable to the computer. SABR programs are core page independent. Therefore, programs may be
written without regard to the 128-word core page of the computer.

SABR automatically generates off-

page and off-field references for direct or indirect statements. It also automatica i iy connects instructions on one page to those that overflow onto the next. The list of available pseudo-ops is extensive,
including external subroutine calling, argument passing, and conditional assembly. SABR offers an optional second pass to produce a side-by-side octal/symbolic listing of the assembled program.
The relocatable binary tapes produced by SABR are loaded into any field of core memory
using the 8K Linking Loader, as are the I ibrary of subprograms. These subprograms may be cal led by any
SABR program.
In addition to being a stand-alone symboiic assembler, SABR aiso acts as the second pass of
the 8K FORTRAN cbmpi ler (see 8K FORTRAN Programmer's Reference Manual, DEC-08-KFXB-D).
SABR requires a PDP-8/1 1 -8/L / -8 1 -8/S, or -5* computer with at !east 8K words of core
memory and an ASR-33 Teletype. A high-speed photoelectric paper tape reader and punch is not necessary, although it is highly recommended.
The assembler system is furnished on four appropriately identified paper tapes. The SABR
Assembler and 8K Linking Loader tapes are punched in binary coded format and are loaded into core
memory using the Binary Loader (see PDP-8/I System User's Guide, DEC-08-NGCB-D). The two
library of subprograms tapes are punched in relocatable binary coded format and are loaded into core
memory using the BK Linking Loader as explained in this manual.
With the exception of a few minor differences (and an entirely different list of pseudo-ops),
the symbolic programming language for SABR is similar tothe PAL III language.

However, the binary

output from SABR is in relocatable binary code and is quite different from the PAL III binary output.
The rest of this chapter describes the SABR language in full.

For a more elementary approach to assem-

bly language programming, we recommend DEC's new Introduction To Programming (specify Order No.
C-18), available from the Program Library (address on Title page).

*The PDP-5 computer requires a PDP-8 extended memory control modification.

1-1

l. 2

THE CHARACTER SET
a.

Alphabetic:

Besides the normal alphabetic characters A, B, C, ... , X, Y, Z, the following
characters are considered to be alphabetic by SABR:
[
]
\
b.

Ieft bracket,
right bracket,
back slash,
up arrow.

Numeric:

0, 1, 2, ... , 8, 9
c.

Special:

.JI.
TT

Comma

delimits a symbolic address label

Slash

indicates start of a comment

Left parenthesis

indicates a literal
(D indicates numeric literal is decimal;
(K indicates numeric literal is octal

Quote

precedes an ASCII constant

Minus sign

negates a constant

Number sign

increases value of preceding symbol by one

RETURN
(carriage return)

terminates a statement

Semicolon

terminates an instruction

LINE FEED

ignored

FORM FEED

ignored

SPACE

separates and delimits items on the statement line

TAB

same as space

RUBOUT

ignored

All other characters are i Ilega I except when used as ASCII constants fo Ilowing a quote ( 11 ) ,
or in comments or text strings.
Legal characters used in ways different from the above and all illegal characters cause the
error message C (Illegal Character) to be printed by SABR.

1.3

STATEMENTS
SABR symbolic programs are written as a sequence of statements, and are usually prepared

on a teletype with the aid of the Symbolic Editor program.

Each statement is written on a sing le Iine

and is terminated by typing the RETURN key (carriage return/I ine feed sequence, abbreviated CR/LF).

1-2

SABR statements are virtually format free, because elements of a statement are not placed in numbered
columns with rigidly controlled spacing between elements, as in punched-card oriented assemblers.
A statement line is composed of one or all of the following elements: label, operator, operand, comment, and/or format effectors. The types of elements in a statement are identified by the
order of appearance in the line and by the separating or delimiting character which follows or precedes
the e Iement.
Statements are written in the general form
label,

operator operand

/comment

SABR interprets and processes the statements, generating one or more machine (binary) instructions or data words during assembly.
An input line may be up to 72

characters iong, including spaces and tabs. Any characters

beyond this limit are ignored.

l. 3. 1

Labels
A label is a symbolic name or location tag created by the programmer to identify the address

of a statement in the program.

Subsequent references to the statement can be made merely by referenc-

ing the label. If present, the label is written first in a statement and is terminated by a comma.
Examples:

0200
0201

0000
1200

SAVE,
ABC,

0
TAD

SAVE

Where SAVE and ABC are the labels, the statements are in location 0200 and 0201, and generate the
instructions 0000 and 1200.

1.3.2

Operators
An operator may be any one of the following items.
a.

A mnemonic memory reference instruction fol lowed by an operand.

b. A mnemonic memory reference instruction fol lowed by an I fol lowed by an operand.
This creates an indirect memory reference instruction.
c. A single mnemonic microinstruction (operate or IOT instruction) or a string of such instructions separated by spaces or tabs. Combinations of microinstructions are formed by inclusive ORing
the octal values of the instructions. Group l operate instructions can be combined with Group l instructions only, and Group 2 with Group 2 only, except for the CLA instruction which may be combined
with either group. IOT instructions may not be combined with operate instructions. (Refer to Appendix B for a summary of all microinstructions.)

1-3

A pseudo-operator (Refer to Chapter 2, Pseudo-Operators) .

Operators are terminated with a space or tab if an operand fol lows, otherwise they may be
terminated with either a semicolon, slash, or carriage return.
Examples:

0200
0201
0202
0203

1320
1550
7004
7620

TAD
SAVE
TAD I
POINTR
RAL
SNL SMA CLA
PAGE

A 11 SAB R operators a re Ii sted in Appendix B.

l .3 .3

Operands
Operands may occur in three ways:

a. Fol lowing a memory reference instruction, and separated from it by a space or tab; the
operand is the address of the data to be accessed by the instruction. This address may be a user-defined
address symbol or a numeric constant. If a symbol is used as the operand, it must be defined somewhere
in the program. Constant addresses must be used with great care because the assembled program wi 11 be
relocatable. If the memory reference instruction is indirect (fol lowed by I) the operand is the address
of the address of the data to be accessed. An opeicmd fo!!o;ving c direct memory reference instruction
may also be a literal.
b.

As the argument of a pseudo-operator.

c. On a line with no operator. In this case, the operand is cal led a parameter. A parameter may be a numeric constant, a literal, or a user-defined address symbol.
Examples:

0200
0201
0202
0203

0200
7460
0315
0176

ABC,

200;-320; M

POINTR,

1000

1576
1375

START I

PGOADR
REORG
TAD I
TAD

1001
1.3.4

1000
POINTR

Comments
A programmer may add notes to a statement fol lowing a slash mark. Such comments do not

affect assembly processing or program execution, but they are useful in the program listing for later
analysis and debugging. Entire Iines of comments may be present in the program.

NOTE
None of the special characters or symbols have significance when they appear in a comment.

1-4

Examples:
/THIS IS A COMMENT LINE.
/THIS TOO. TAD; CALL; # 11 -2 (+ = !
A

i .3 .5

TAD

SAVE

/COMMENT

Format Effectors
Spaces and tabs are the formatting characters, usually used in the body of a symbolic pro-

gram to provide a neat page. They can separate elements of a statement / as between an instruction
and a comment. For example, the lines
GO, TAD TOTAL/MAIN LOOP
DCA I SAVE
TAD BUFPTR
SZA CLA/CHECK FOR END LOOP
JMP GO
are much easier to read when written as:
GO,

TAD
DCA I
TAD
SZA CLA
JMP

TOTAL
SAVE
BUFPTR

/MAIN LOOP

/CHECK FOR END LOOP
GO

The RETURN key (CR/LF) is both a statement and a line terminator. The semicolon may be used to
terminate an instruction without terminating a statement line. This allows the programmer to place
several I ines of coding on a single line. If, for example, he wishes to write a sequence of instructions
to rotate the contents of the accumulator (AC) and link (L) six places to the right, it might look like

RTR
RTR
RTR
But, with the semicolon, the programmer may place all three RTR's on a single line, separating each
RTR with a semicolon and terminating the line with the RETURN key. The above sequence of instructions cou Id then be written
RTR; RTR; RTR (terminated with the RETURN key)
This format is particularly useful when creating a list of data.
Example:

0200
0201
0202
0203

0020
0050
7750
0062

LIST I

20; 50; -30; 62

Null lines may also be used as format effectors. A null line is a line containing only a carriage return,
and possibly spaces or tabs. Such lines appear in the listing simply as blank lines.

1-5

l .4

SYMBOLS
Symbols are composed of legal alphanumeric characters. There are two major types of

symbols, permanent symbols and user-defined symbols, and there are variations within each major type.
A symbol is delimited by a nonalphanumeric character.

1.4.1

Permanent Symbo Is
Permanent symbols are predefined and maintained in SABR's permanent symbol table. They

include all of the basic instructions and pseudo-ops listed in Appendix B. These symbols may be used
without prior definition by the user. The OPDEF and SKPDF pseudo-operators are used to define instruction operators not included in the permanent symbol table.

l .4 .2

User-Defined Symbols
A user-defined symbol is a string of from one to six legal alphanumeric characters delimited

by a nonalphanumeric character. User-defined symbols are composed according to the following rules.
a.

The characters must be legal alphanumerics, which are:
ABCD ... XYZ [ \J t and 0123456789.

The first character must be alphabetic.

c. The symbol should not contain more than six characters. Only the first six characters
of any symbol are meaningful, the remainder, if any, are ignored. Therefore, a symbol such as
INTEGER would be interpreted as INT EGE since the seventh character is ignored, and because the two
symbols GEORG El and GEORGE2 differ only in the seventh character, they would be treated as the
same symbol, GEORGE.
d.

A user-defined symbol cannot be the same as any of the predefined permanent symbols,

and,
e. A user-defined symbol must be defined only once. Subsequent definitions of the same
symbol wil I be ignored and cause SABR to type the error message M (Multiple Definition).
A symbol is defined by appearing as a symbolic address label (Refer to Section l .3. 1) or by
appearing in an ABSYM, COMMN, OPDEF or SKPDF statement (Refer to Chapter 2, Pseudo-Operators).
No more than 64 different user-defined symbois may occur on any one core page.

1 .4.3

Equivalent Symbols
When an address label appears alone on a line, i.e., with no instruction or parameter, the

iabel is assigned the value of the next address assembled.

1-6

Foi example,

TAGl I
TAG2,
TAG3,

TAG 1 and TAG2 are equivalent in that they are assigned the same value. Therefore, a TAD TAG 1
will reference the data at TAG2. TAG3, however, is not equivalent to TAG2. TAG3 would be defined as 1 greater than TAG2.

1 .4.4

Incrementing Operands
Because SABR is a one-pass assembler and also sometimes generates more than one machine

instruction for a single user instruction, operand arithmetic is impossible; i.e., statements of the form
TAD
TAD
JMP

TAG+3
LISTl - LIST2
• +6

are i I legal.
However, in one special case such references are possible. By appending a number sign
(#)to an address operand, the user will reference a location exactly one (1) greater than the location
of the address operand. Thus TAD Loe# is equivalent to the PAL language statement TAD LOe+ 1.
Example:
0200
0201
0202
0203

0020
0030
1200
1201

0400
0401

0200
0201

LOe,
START I

A,
B,

20
30
TAD
TAD
PAGE
LOe
Loe#

LOe
Loe#

/GET 20
/GET 30

NOTE
In assembling# - references, SABR does not attempt to
determine if multiple machine code words are generated
at the symbolic address referenced.

ExampJe:
START I

/Loe IS OFF-PAGE
/USER HOPES TO MODIFY

TAD I
NOP

LOe

TAD
DeA

(7500
/SMA
START#

1-7

The user hopes to change the NOP instruction to an SMA. However, this is not possible
because the TAD I LOC wi 11 be assembled as three machine code words; if ST ART is at 0200, the NOP
will be at 0203. The SMA will be inserted at 0201, thus destroying the second word of the TAD I LOC
execution.
To avoid this error, the user should carefully examine the assembly listing before attempting
to execute a program with # - references •
In the previous example, the proper sequence is:
START,
VAR,

TAD I
NOP

LOC

TAD
DCA

(7500
VAR

The # - sign feature is intended primarily for use in manipulating DUMMY variables, in
picking up subroutine arguments in external subroutines, and returning from external subroutines. Refer
to Section 2 .4 .4 for a fut I explanation of how this is done.

1.4.5

The Symbol Table Listing
Symbols are listed in alphabetic order at the end of the assembly pass (Pass 1) with their

relative addresses beside them.
The fol lowing flags are added to special types of symbols.
ABS
COM
OP
EXT
UNDF

l .5

The address is absolute.
The address is in COMMON.
The symbol is an operator.
The symbol is an external and may or may not be defined. If not
defined, there is no difficulty; it is in another program.
The symbol is not an external symbol and has not been defined in
the program. This is a programmer error. No earlier diagnostic
can be given because it is not known that the symbol is undefined
unti I the end of Pass l .
A location is reserved for the instruction containing the undefined
symbol, but nothing is placed in it.

CONSTANTS
There are two types of constants: numeric and ASCII. These are discussed individually

below. ASCII constants are used only as parameters. Numeric constants may be used as parameters
or as operand addresses .
Example:

0200

1412

TAD I

1-8

Constant operand addresses are treated as absolute addresses, just as a symbol defined by an
ABSYM statement. References to them are not generally relocatable. Therefore, they should be used
only with great care. The primar1 use of constant operand addresses is to ieference locations in page
0. (See Appendix D for a list of free locations in page 0 of each field.) All constant operand addresses
are assumed to be in the field into which the program is loaded by the Linking Loader.
Constants may not be added or subtracted to/from each other or to/from symbols.

1.5.1

Numeric Constants
A numeric constant consists of a singie string of from one to four digits. It may be preceded

by a minus sign(-) to negate the constant. The digit string will be interpreted as either octal or1decimal according to the latest permanent mode setting by an OCTAL or DECIM pseudo-op. Octal mode is
assumed at the beginning of assembly. The digits 8 and 9 must not appear in an octal string.
Examples:

0200
0201

5020
7575

5020
-203

0202

0120

DEC!M

l .5 .2

ASCII Constants
Eight-bit ASCII values may be created as constants by typing the ASCII character imme-

diately following a double quotation mark ( 11 ) . A minus may be used to negate an alpha constant. The
minus sign must precede the quotation mark.
Examples:

0200
0201
0202

0273
7477
0207

_,.A
II

I ' -301
I BELL FOLLOWS

The following characters are illegal as alpha constants: carriage return, line feed, form
feed, rubout .

l .6

LITERALS
The use of I iterals is a special and convenient way of generating constant data in a program.

Literals are normally used by TAD and AND instructions, as in the fol lowing examples:

1-9

0200
0201
0202

0376
1375
1374

0374
0375
0376

0303
7730
0777

AND
TAD
TAD

(777
(-50
("C

A literal is always a numeric or ASCII constant and must be preceded by a left parenthesis.
The value of the literal will be assembled in a table near the end of the core page on which the instruction referencing it is assembled. The instruction itself will be assembled as an appropriate reference to
the location where the numeric value of the literal is assembled. Literals may not be referenced indirectly.
The current numeric conversion mode can be changed on a purely local basis for a literal by
inserting a D for decimal or a K for octal between the left parenthesis and the constant.
Examples:
(D32 becomes 0040 (octal)
(K-32 becomes 7746 (octal)
This usage does not alter the prevailing permanent conversion mode.
A literal may also be used as a parameter (i.e., with no operator). In such a case the
numeric value of the literal is assembled as usual in the literal table near the end of the core page
currently being assembled, and a relocatable pointer to the address of the I iteral is assembled in the
location where the literal parameter appeared.
Example:
0200

0376 01

0376

0020

(20

This feature is intended primarily for use in passing external subroutine arguments with the
ARG pseudo-op (see Section 2 .4. l).

1-10

CHAPTER 2
PSEUDO-OPERA TO RS

2 .1

ASSEMBLY CONTROL
END

Every program or subprogram to be assembled must contain the END
pseudo-op as its last line. If this requirement is not met 1 an error
message (E) is given.

PAUSE

The PAUSE pseudo-op causes assembly to halt. It is designed to
allow the user to break up large source tapes into several smaller
ones. To do this 1 the user need on iy pi ace a PAUSE statement at
the end of each section of this source except the last. Then when
assembly halts at a PAUSE 1 he may remove the source tape just read
from the reader and insert the next one. Assembly may then be
continued by pressing the console CO NTi nue switch.

WARNING
The PAUSE pseudo-op is designed specifically for use at
the end of partia I tapes and should not be used otherwise.

The reason for this is that the reader routine may have read data
from the paper tape into its buffei that is actually beyond the
PAUSE statement. Consequently 1 when CONTinue is pressed
after the PAUSE is found by the line interpreting routine 1 the
entire content of the reader buffer fol lowing the PAUSE is destroyed 1
and the next tape begins reading into a fresh buffer. Thus, if there
is any meaningful data on the tape beyond the PAUSE statement 1 it
wi II be lost.
DEC IM

Initially the numeric conversion mode is set for octal conversion.
However 1 if the user wishes, he may change it to decimal by use of
the DECIM pseudo-op.

OCTAL

If the numeric conversion mode has been set to decimal 1 it may be
changed back to octal by use of the OCTAL pseudo-op.
No matter which conversion mode has been permanently set, it may
always be changed locally for literals by use of the (Dor (K syntax
described earlier.
Examples:

320

0200

0320

0201
0202
0203

0500
0377 01
1000

320
(K320
512

0204

0512

512

START I

DEC IM

OCTAL

2-1

0205
0206

0376 01
0320

(0512
320
END

0376
0377

1000
0320

LAP

The assembler is initia I ly set for automatic generation of jumps
to the next core page when the page being assembled fi I ls up
(Page Escapes), or when PAGE or REORG pseudo-ops are encountered. This feature may be suppressed by use of the LAP
(Leave Automatic Paging) pseudo-op.

EAP

If the user has previously suppressed the automatic paging
feature, it may be restored to operation by use of the EAP
(Enter Automatic Paging) pseudo-op.

PAGE

The PAGE pseudo-op causes the current core page to be assembled
as is. Assembly of succeeding instructions wi II begin on the next
core page. No argument is required.

REORG

The REORG pseudo-op is similar to the PAGE pseudo-op, except
that a numerical argument specifying the relative location within
the subprogram where assembly of succeeding instructions is to
begin must be given. A REORG below 200 may not be given. A
REC RG shou Id a Iways bs to ths fl iSt addisss of a core page . If a
REORG address is not the first address of a page, it wi 11 be converted to the first address of the page it is on.
Examples:

CPAGE

0200

7200

0400

7040

1000

7041

START,

CLA
PAGE
CMA
REORG
CIA

1000

The CPAGE pseudo-op followed by a numerical argument N specifies
that the following N words of code* must be kept together in a single
unit and not be split up by page escapes and literal tables. If the N
words of code wi 11 not fit on the current page of code, the current
page is assembled as if a PAGE pseudo-op had been encountered. The
N words of code wi !! then be assembled as a unit on the next core page.

NOTE
N must be less than or equal 200 (octal) in nonautomatic
paging mode or less than or equal 176 octal in automatic
paging mode •

*Normally data. However, if these N words are instructions (for example, a JMS with
arguments), it is the user 1 s responsibility to count extra machine instructions which must
be inserted by SAB R.

2-2

Example:
START I

CLA
LAP
CPAGE 200
NAME 1
NAME2

/INHIBIT PAGE ESCAPE
/CLOSES THE
/CURRENT PAGE
/&ASSEMBLES THE
/NAMES ON THE
/NEXT PAGE.

ihe conditional pseudo-op, iF, is used with the foi iowing syntax:
IF

NAME, 7

The action of the pseudo-op, so given, is to first determine
whether the symbol NAME has been previously defined. If
NAME is defined, the pseudo-op has no effect. If NAME is
not defined, the next seven symbolic instructions (not counting
null lines and comment lines) wi II be treated as comments
and not assembled.
Example:
/ABSYM NAME
176
IF NAME, 2
/THE NEXT LINE TO BE
CLL RTL
/ASSEMBLED WILL BE
RAL
/"DCA LOC".
/IF THE SLASH BEFORE 11 ABSYM NAME 176 11 IS
/REMOVED, THE 11 CLL RTL" AND 11 RAL 11 WILL
/BE ASSEMBLED.
DCA
LOC
Norma fly the symbol referenced by an IF statement shouid be
either an undefined symbol or a symbol defined by an ABSYM
statement. If this is done, the situation mentioned below
cannot occur.

WARNING
In a situation such as the fol lowing, a specia I
restriction applies.
/EXAMPLE:
NAME, 0

IF NAME, 3

2-3

The restriction is that if the line NAME, 0 happens to occur
on the same core page of instructions as the IF statement, then,
even though it is before the IF statement, NAME wi 11 not have
been previously defined when the IF statement is encountered,
and on the first pass (though not in the listing pass) the three
lines after the IF statement wi 11 not be assembled. The reason
for this is that location tags cannot be defined unti I the page
on which they occur is assembled as a unit.

2 .2

SYMBOL DEFINITION
ABSYM

An absolute core address may be named using the ABSYM
pseudo-op. This address must be in the same core field as the
subprogram in which it is defined. The most common use of
this pseudo-op is to name page zero addresses not used by the
operating system. These addresses are listed in Appendix D.

OPDEF
SK PDF

Operation codes not already included in the symbol table
may be defined by use of the OPDEF or SKPDF pseudo-ops.
Non-skip instructions must be defined with the OPDEF pseudoop and skip-type instructions must be defined with the SKPDF
pseudo-op.

Examples of ABSYl·.1, OPDEF end SKPDF syntax:
ABSYM
ABSYM
OPDEF
SK PDF
SK PDF

TEM

AX
DTRA
DTSF

SMZ

177
10
6761
6771
7540

/PAGE ZERO ADDRESSES
/A NON-SKIP INSTR.
/SKIP-TYPE INSTRUCTIONS

NOTE
ABSYM, OPDEF and SKPDF definitions must be
made before they are used in the program.

COM MN

The COMMN pseudo-op is used to name locations in field l as
externa Is so that they may be referenced by any program. If any
COMMN statements are used, they must occur at the beginning
of the source, before everything else including the ENTRY statement. COMMON storage is always in field l and is allocated
from location 0200 upwards. Since the top page of field l is reserved, no more than 3840 words of COMMON storage may be
10
defined.
A COMMN statement normally takes a symbolic address label,
since storage is being allocated. However, COMMON storage
may be allocated without an address label.

2-4

A COMMN statement always takes a numerical argument which
specifies how many words of COMMON storage to be allocated;
however, aO argument is allowed. A COMMNstatement
with 0 argument allocates no COMMON storage; it merely defines the given location symbol at the next free COMMON
location.
The syntax of the COMMN statement is shown below.
Example:
A,
B,

c,
D,

COMMN
COMMN
COM MN
COMMN
COMMN
ENTRY

10
300
0
10
SUB RUT

In this example 20 words of COMMON storage are allocated from
0200 to 0217, and A is defined at location 0200. Then, 10 words
are a 11 ocated from 0220 to 0227, and B is defined at 0220. Notice
that if A is actually a 30 word array, this example equates B(l) with
A(21).
The example continues by allocating COMMON storage from 0230 to
0527 with no name being assigned to this block. Then 10 words are
allocated from 0530 to 0537 with both C and D being defined at 0530.

2 .3

DAT A GENERATING
BLOCK

The BLOCK pseudo-op given with a numerical argument N wi II
reserve N words of core by placing zeros in them. This pseudo-op
creates binary output, and thus may have a symbolic address label.
Before the N locations are reserved, a check is made to see if
enough space is avai fable for them on the current core page. If
not, this page is assembled and the N locations are reserved on
the next core page. The action here is similar to that of the
CPAGE pseudo-op. Similar restrictions on the argument apply.
/EXAMPLE OF HOW LARGE BLOCK STORAGE
/WITHIN A SUBPROGRAM AREA MAY
/BE ACHIEVED:
/INHIBIT PAGE ESCAPES
/RESERVE 500
/(OCTAL) LOCATIONS

LAP
BLOCK 200
BLOCK 200
BLOCK 100
EAP

/RESUME NORMAL CODING

2-5

As a special use, if the BLOCK pseudo-op is used with a location
tag (but with no argument or a zero argument), no code zeros are
assembled; instead the symbolic address label is made equivalent
to the next relative core location assembled. (This is equivalent
to using a symbolic address label with no instruction on the same
line.)
Examples:

TEXT

LIST,

BLOCK 3

NAMEl,
NAME2,
NAME3,
NAME4,

BLOCK
BLOCK 0

/ASSEMBLES AS
/3 ZEROS WITH
11
/
LIST" DEFINED
/AT THE lST LOCATION
/DEFINES NAME 1 =
/NAME2 = NAME3 =
/NAME4

BLOCK 2

The TEXT pseudo-op is used to obtain packed six-bit ASCII text
strings. Its function and use are a I most exactly the same as for
the BLOCK pseudo-op except that instead of a numerical argument,
the argument is a text string. In particular, a check is made to be
sure that the text string will fit on the current page without being
interrupted by literals, etc.
The text string argument must be contained on the same line as the
TEXT pseudo-op. Any printing character may be used to delineate
the text string. This character must appear at both the beginning
and the end of the string. Carriage return, line feed and form feed
are illegal characters within a text string (or as delineators). All
characters in the string are stored in simple stripped six-bit form.
Thus, a tab character (ASCII 211) will be stored as an 11, which
is equivalent to the six-bit for the letter I. In general, characters
outside the ASCII range of 240-337 should not be used.
Example:
0200
0201
0202
0203
0204
0205
0206
0207
0210
0211

2405
3024
4005
3001
1520
1405
4061
6263
5273
7700

TAG,

2-6

TEXT /TEXT EXAMPLE 123*; ?/

2 .4

EXTERNAL SUBROUTINE
SABR and the Linking Loader possess extensive capabilities for calling external subprograms

and for passing arguments between them. In addition to the foci lities mentioned in this section,
COMMON storage is also available (refer to Section 2 .2).
For example, a user wishes to write a long main program, MAIN, which uses two major
subroutines, S 1 and S2. S 1 requires two arguments and S2 requires one argument. The user would
then write MAIN, S 1 and S2 as three separate programs in the fol lowing fashion:
MAIN,

CLA

/START OF MAIN

END
ENTRY

sl ,

ENTRY

S2!

Sl
BLOCK 2
RETRN Sl
END
S2
BLOCK 2
RETRN S2

END
He would then assemble each of these subprograms with SABR and load all of them with the
Linking Loader.
MAIN would contain statements in the form
CALL
ARG
ARG
CALL
ARG

y
1, S2

Also Sl could contain CALLs to S2 or S2 CALLs to Sl.
In addition, any of the subprograms could make use of DUMMY variables.
During the loading process all of the proper addresses will be saved in tables so that when
the user begins execution of MAIN, the Run-Time Linkage Routines (see Section 3 .3 .4), which were
automatically loaded, will be able to execute the proper reference. Thus, MAIN will be able to
fully use S 1 and S2 and be able to pass data to and receive it from them.
The particular pseudo-operators required to make use of these foci Ii ties are described
next.

2-7

2 .4. l

The CALL and ARG Statements
The CALL and ARG statements are the usual means of calling an external subroutine.

For

example, a subroutine named SUBR with two arguments can be called by another program with the
instruction sequence:
TAG,
Nl I
N2,
ETC,

CALL
ARG
ARG

2 I SUBR
(50
LOCATN

A CALL statement must contain both the number of arguments and the ENTRY point of the
subprogram being cal led in that order and separated by a comma. Arguments may or may not have
address labels. Constant arguments may be specified as literals for the reason explained below. However, true constant arguments may also be specified.
The above instructions are assembled as follows:
CPAGE 6

TAG,

JMS LINK
020X (06)

Nl I

CDF CUR (05)
POINTER

N2,

CDF CUR or CDF 10

LOCATN

/Make sure the fol lowing
/2N + 2 words wi II
/fit on the current core page.
/Ca II the CALL Linkage Routine,
/where 2 =the number of
/arguments and X =
/the loca I number of the
/subprogram being called
/viz., SUBR.
/Field address of argument
/in form of a CDF instruction.
/Address in the literal
/table where the 50 is
/ assemb Ied .
/Field of the argument
/depending on whether it is
/or is not in COMMON.
/Address of argument.

When a subprogram is referenced in a CALL statement, the Run-Time Linkage Routine,
LINK; always executes the transfer to the subprogram as follows.
First, it assumes that the ENTRY point to the subprogram is a two-word block. Into the
first word of this block it places the number of the field where the CALL to the subprogram occurred.
In the second word, it places the address where the CALL occurred, plus 2. In the example above
SUBR would receive a 62Ml where TAG is in field M, and SUBR# would receive the address of Nl.

If there were no arguments, SUBR# would receive the address of ETC. Thus, the two-word block
at the ENTRY point serves as storage for the 15-bit address vector for picking up arguments and also
for returning from the subprogram.

2-8

Execution of the subprogram begins at the first location following the two-word ENTRY
block.
The number of arguments in a CALL sequence must be less than 64

• The ARG statement

may be used only in conjunction with a CALL statement.
When the ARG pseudo-op is used with a literal, as in the above examples, the actual literal
(50 in this case) wi II be generated in the litera I table, and in the location following the CDF CUR,
there will be generated a relocatable pointer to the literal. This is the same as using a literal as a
parameter.
If the .l\.RG statement is used with a true constant argument, the constant itself is assembled
in the location following the CDF instruction. In this case, the CDF is useless and is always just a
meaning less space fi lier.
The advantage of using the ARG - literal method is that it allows a subroutine to pick up
an argument which is sometimes a variable and sometimes a constant.

2 .4.2

The ENTRY and DUMMY Statements
ENTRY

The ENTRY pseudo-op is used at the beginning of a subprogram to
name its entry point, and define this symbol as an external for the
Linking Loader.
The ENTRY statement must occur before the symbolic name of the
entry point appears as a symbolic address label. The actual entry
location must be a two-word reserved space so that both the return
address and field can be saved when the routine is called.
Example:
ENTRY
SUBROU,

SUBROU
BLOCK 2

CLA
For convenience of picking up subprogram arguments following a
CALL statement, an ENTRY acquires all the properties of a
DUMMY variable.
DUMMY

A DUMMY variable is a special type of variable in the FORTRAN/SABR
system. It must be so defined in the subprogram which references it.
When referenced directly a DUMMY variable is treated the same as
any other local symbol. However, when referenced indirect~y it causes
a ca II to the DUMMY Variable Run-Time Linkage Routine. This Linkage Routine assumes that the DUMMY variable is a two-word vector
such that the first word is a 62N 1 (where N =the field of the address
to be referenced) and the second word contains the actua I 12-bit
address to be referenced .

2-9

DUMMY variables are used for passing arguments to and from
subroutines. (See Section 2.4.4.)
Example:

ENTRY
DUMMY
DUMMY

y
BLOCK 2

AI,

M
N

2.4.3

The RETRN Statement
The RETRN statement is used to return from a subprogram to the calling program. The name

of the subprogram being returned from must be specified so that the Return Linkage Routine can determine
the action required / and because a subprogram may have differently named entry points. It is possible
for the careful user to return to the location following the fast colt of any subprogram merely by specifying it in a RETRN statement.
Example:
TAG,

RETRN

SUBROU

Before the RETRN statement is used, the user must be sure to increment the pointer in the
second word of the subprogram entry to the proper point beyond all the arguments following the CALL
statement. An example of how this is done is given below.

2.4.4

Picking up Subprogram Arguments
An advanced technique for picking up subprogram arguments is provided because:

a. Subprogram arguments are two-word addresses and a subprogram CALL is executed by the
Run-Time Linkage Routine.
b.

The calling program and subprogram may reside in different fields.

A subprogram entry point is assumed to have been defined as a two-word reserved block and
defined as an ENTRY. The appearance of the subprogram name in an ENTRY statement gives the twoword block the properties of a DUMMY variable. This means that when the subprogram name is referenced indirectly this generates a call to the DUMMY Variable Run-Time Linkage Routine where the details
of locating and picking up the argument address words are worked out. Thus, the user, need only use
the number sign feature to increment the argument pointer in the second word of the entry point.

2-10

The fol lowing example shows how SUBR would pick up the arguments 50 and LOCATN in the
example and deposit them in LOCl and LOC2.
Example:

/MAIN
MAIN,

PROGRAM
CLA

TAG,
Nl I
N2,
ETC,

CALL
ARG
ARG

2,SUBR
(50
LOCATN

END
/SUBROUTINE

SUBR,

ENTRY
SUBR
DUMMY TEM
2
BLOCK
TAD I

SUBR

DCA
INC
TAD I

TEM
SUBR#
SUBR

DCA
TAD I
DCA
INC

TEM#
TEM
LOCl
SUBR#

/fHIS GIVES YOU THE FIELD ADDRESS
/(CDF CUR) OF THE (50
/I.E., THE CONTENTS OF Nl
/fO FIRST WORD OF DUMMY
/MOVE ARG PTR TO Nl#
/GET ADDRESS OF (50
/I.E., CONTENTS OF Nl#
/fO 2ND WORD OF DUMMY
/PICK UP THE 50
/MOVE ARG PTR TO N2

Similar method to pick up contents of LOCATN.

TEM,

INC
RETRN
BLOCK

SUBR#
SUBR
2

/MO VE PTR FOR RETURN AT ETC

Constant arguments are specified as literals because the subprogram may not know that a
constant argument is being used. Hence, specifying constant arguments as litera Is wi II ensure that
the second word of every assembled argument is actually the address of the argument.
The ARG statement may be used with a constant (e.g., if a constant address is intended).
The following technique may be used if SUBR can assume that the first argument is always
a constant:

2-11

Example:
/MAIN PROGRAM
TAG,
CALL
Nl,
ARG
N2,
ARG
ETC,

2, SUBR
50
LOCATN

END
/SUBROUTINE

SUBR,

ENTRY
DUMMY
BLOCK
INC
TAD I

SUBR
TEM
2
SUBR#
SUBR

DCA
INC

LOCl
SUBR#

2-12

/MOVE ARG PTR TO Nl#
/THIS GETS THE 50
/IMME DIA TEL Y
/GET C(LOCATN) IN
/THE USUAL WAY

CHAPTER 3
THE ASSEMBLED BINARY CODE

Because SABR is not a one-for-one assembler, it is necessary to give a general description of
the type of code which it produces. The ordinary user needs only a general understanding of this topic;
a more detailed discussion is included for the advanced user.

3. l

THE BINARY OUTPUT TAPE
SABR outputs each machine instruction on binary output tape as a 16-bit word contained in

two 8-bit frames of paper tape. The first four bits contain the relocation code used by the Linking
Loader to determine how to load the data word. The last twelve bits contain the data word itself.

Relocation
Code

High Order of
Data Word

first frame
second frame

Low Order of Data Word

The assembled binary tape is preceded and followed by leader/trailer code 200. The checksum is contained in the last two frames of tape before the trailer code. It appears as a normal 16-bit
word as shown below.

High Order of
Checksum

Low Order of Checksum

first frame
second frame

All assembled programs have a relative origin of 0200.

3.2

THE LOADER RELOCATION CODES
The four-bit relocation codes issued by SABR for use by the Linking Loader are al I explained

below. The codes are given in octal.

Absolute

Load the data word at the current loading address. No change
is required.
Example:

0205

5277

0242
0356

7500
0020

3-1

JMP LOC where
LOC is at 0277
(on page)
SMA
20
(a constant)

Simple
Relocation

Add the relocation constant to the word before loading it.
(The relocation constant is 200 less than the actual address
where the first word of the program is Ioaded . ) Items with this
code are a Iways program addresses.
Example:
0376

0520

LOC2

In the above example, LOC2 is at relative address 0520. If the
first word of the program (relative address 0200) is loaded at
1000, then the actual address of A is 1176 and location 1176
will be loaded with the value 1320, which will be the actual
address of LOC2 when loaded.
03

External
Symbol
Definition*

The data word is the relative address of an entry point.
Before entering this definition in the Linkage Tables so that the
symbol may be referenced by other programs at run-time, the
Linking Loader must add the relocation constant to it.
The six frames of paper tape fol lowing the two-frame definition
are the ASCII code for the symbol.
Example:
03

address
address Iow order

c
2
space
space
04

Reorgin*

Change the current loading address to the value specified by the
data word plus the relocation constant.

CDF
Current

The data word is always a 6201 (CDF) instruction which has been
generated automatically by SABR. The code 05 indicates to the
Linking Loader that the number of the field currently being
loaded into must be inserted in bits 6-8 before loading.
Example:
0300
0301

6201
1776

0376

0520

TAD LOC2
where LOC2 is off page so that
the TAD instiUction must be indiiect.

If the program containing this code is being loaded into field 4,
relative location 0300 will be loaded with 6241.
Such an instruction is referred to in this document as CDF Current.
They are generated automatically by SABR when a direct reference
instruction must be assembled as an indirect, and there is the possibility that the current data field setting is different from the field
where the indirect reference occurs.
*Does not appear in assembly listings.

3-2

Subroutine
Linkage
Code

The data word is a special constant enabling the Linking Loader
to perform the necessary linking for an external subroutine call.
(c. f., CALL Pseudo-op, Section 2 .4) The structure of the data
word is shown below.
Bits 0 - 5

Bits 6 - 11

4 - number of arguments _ . 4- local program __,.

fol lowing the CALL

number assigned
to the extern a I
subroutine being
called.

Before the 12-b it, two-part code word is Ioaded into memory, a
global external number wi 11 be substituted for the local external
symbol number in the right half of the data word.
Example:

0200
0201

4033
0307 06

CALL

3, SUB

ARG X
ARG Y
ARG Z
Here, SUB has been assigned the local number 07 during
assembly. At loading time this number wil I be changed to the
global number (for example, 23), which is assigned to SUB. In
this example, 0323 would actually be loaded at relative address

0201.
10

Leader/Trailer*
and
Checksum

This code represents normal leader/trailer. At the first occurrence of this code fol lowing the assembled program, the computer
word contains the checksum.

The data \A/ord is the highest location in Field 1 assigned to
COMMON storage by the program. This item will occur
exactly once in every binary tape and it must be the first word
after the leader. If no COMMON storage has been allocated
in the program, the data word wi 11 be 0177.

Transfer*
Vector

Signifies that reference to an external symbol occurs in the
assembled program. The 12-bit data word is meaningless. The
next six frames contain the ASCII code for the symbol.
The Linking Loader uses this definition to create a transfer table,
whereby local external symbol numbers assigned during assembly
of this particular program can be changed to the global external
symbol number when several programs are being loaded.

3 .3

PAGE ASSEMBLY
SABR assembles page-by-page rather than one instruction at a time. This is accomplished

by bui Iding various tables as instructions are read. When a ful I page of instructions has been collected
*Does not appear in assembly listings.
3-3

{counting literals, off-page pointers and multiple word instructions) the page is assembled and punched.
Several pseudo-ops also cause a page to be assembled.

3.3. 1

Page Format
A normal assembled page of code has a format as shown below.

xooo
Assembled
Instructions

Page Escape
Li tera Is and
Off-Page
Pointers

X377

Page Escape

Literals and off-page pointers are intermingled in the table at the end of the page.

3.3.2

Page Escapes
Under normal circumstances SABR is in Automatic Paging Mode. This mode causes SABR to

connect each assembled core page of code to the next page by an appropriate jump. This is called a
Page Escape . For the Iast page of code, SABR Ieaves the Automatic Paging Mode and issues no Page
Escape. Also., a pseudo-op is available to turn off this Automatic Paging Mode.
There are two types of Page Escapes, depending on whether or not the last instruction is a
skip instruction. If the last instruction was not a skip instruction, the Page Escape is as fol lows:
last instruction (non-skip)
5377 (JMP to x 177)
Iiterals
and
off-page
pointers
x177/NOP

If the last instruction was a skip instruction, the page escape takes four words as follows:
last instruction (a skip)
5376 (JMP to x 176)
5377 (JMP to x 177)
Iiterals
etc.
xl76/SKP
x177/SKP

3-4

3.3.3

Multiple Word Instructions
Certain instructions in the source program require SABR to assemble more than one instruc-

tion (e.g., off-page indirect references and indirect references where a data field re~setting may be
required). In the Iisting, the source instruction wil I appear beside the first of the assembled binary
words.
A difficulty arises when a multiple word instruction follows a skip instruction. In such a
case, extrn instructions must be assembled to enable the skip to be effected exactly as desired by the
programmer.

3.3.4

Run-Time Linkage Routines
The routines described in this section are entirely automatic. The user needs to know nothing

about them except to better understand the program assembly listing.
Many of the multiple word instructions involve use of special linking routines called the
Run-Time Linking Routines. These routines make up a special portion of the 8K FORTRAN/SABR
System. They are used at execution time by al I user programs to carry out the linkage for cal Is to
externai subroutines and for al i the various forms of off-fieid and off-page indirect memory references.
Since the Linkage Routines are needed by al I user programs, their use is entirely automatic.
The user need not consider them either at programming time or at loading time. SABR determines when
cal Is to the Linkage Routines are required in the user's program and automatically generates such cal Is.
The Linking Loader a!·ways automatica! !y !cads the Linkage Routines.
The residence of the Linkage Routines is described in Appendix D.
There are seven Linkage Routines:
a.

change Data Field to current and skip

CDFSKP

change Data Field to one (COMMON) and skip

CD ZS KP

off page indirect reference linkage

OPISUB

off bank (COMMON) indirect reference linkage

OBISUB

DUMMY variable indirect reference linkage

DUMSUB

subroutine CALL linkage

LINK

subroutine RETURN linkage

RTN

The fol lowing is a description of the individual Linkage Routines.
a. CDFSKP is cal led when a direct off-page memory reference, requiring that the data
field be reset to the current field, fol lows a skip-type instruction.

3-5

Example:
Program

Assembled
Code

SZA
DCA LOC

7440
4045
7410
3776

Meaning

call CDFSKP
SKP in case AC = 0 at .-2
execute the DCA via a pointer near the end of
the page.

b. CD ZS KP is cal led when a direct memory reference is made to a location in COMMON
(which is always in Field 1), the action of CDZSKP is the same as that of CDFSKP except that it
always executes a CDF 10 instead of a CDF current.
Example:
Program

SZA
DCA CLOC

Meaning

Assembled
Code
7440
4051
7410
3776

call CDZSKP
SKP in case AC = 0 at .-2
execute the DCA via a pointer near the end of
the page.

OPISUB is cal led when there is an indirect reference to an off page location.

Example:
Program

Assembled
Code

Meaning

DCA I PTR

4062
0300 01
3407

call OPISUB
relative address of PTR
execute the DCA I via 0007

d. OBISUB is called when there is an indirect reference to a location in COMMON. In
such a case it is assumed that the location in COMMON which is being indirectly referenced points to
some location that is also in COMMON.
Example:
Program

DCA I CPTR

Assembled
Code

Meaning

4055
1000
3407

call OBISUB
address of CPTR in Field l
execute the DCA I via 0007

e. DUMSUB is cal led when there is an indirect reference to a DUMMY variable. In such
a case, DUMSUB assumes that the DUMMY variable is a two-word vector in which the first word is a
62N l, where N =the field of the address to be referenced, and the second word is the actual address
to be referenced •

3-6

Example:
Program
DCA ! D UMV AR

Assembled
Code

Meaning

4067
0300 01
3407

ca!! DUMSUB
relative address of DUMVAR
execute DCA I via pointer
in location 0007

f.
LINK is called to execute the linkage required by a CALL statement in the user's
program. When a CALL statement is used, it is assumed that the entry point of the subprogram is named
in the CALL and that this entry point is a two-word, free biock fol lowed by the executable code of the
subprogram. LINK leaves the return address for the CALL in these two words in the same format as a
DUMMY variabie.
Example:
Program
CALL 2, SUBR
ARG X
ARG C
g.

Meaning

Assembled
Code

4033
0205 06
62Ml
0300 01
6211
1007

cal I LINK
code word
X resides in field M
relative address of X
C is in COMMON
absolute address of C

RTN is cal led to execute the linkage required by a RETRN statement in the user's

program.
Example:
Program
RETRN SUBR

3.3.5

Assembled
Code

Meaning

4040
0005 06

call RTN
number of the subprogram being returned from
(SUBR)

Skip Instructions
In page escapes and in multiple word instructions, skip-type instructions must be distin-

guished from non-skipping instructions. For this reason, a special pseudo-op, SK PDF, must be used
to define skip instructions not in the permanent symbol table.
This also explains why both ISZ and INC are included in the permanent symbol table. ISZ
is considered to be a skip instruction and INC is not. INC should be used to conserve space when the
programmer desires only to increment a memory word with no possibility of a skip resulting.
Example l shows the code which is assembled for an indirect reference instruction to an off
page location following an INC instruction and Example 2 shows the same instruction following an ISZ
instruction. In Example 1, it is assumed there is no possibility of the INC instruction actually causing
a skip.

3-7

Example
INC
TAD

POINTR
LOC2

0220
0221
0222
0223

2376
4062
0520
1407

0220
0221
0222
0223
0224
0225

2376
7410
5226
4062
0520
1407

01}

off page indirect execution

Example 2
ISZ
TAD I

3 .4

COUNTR
LOC2

skip to execution
jump over execution

01}

off page indirect execution

PROGRAM ADDRESSES
Since each assembly is relocatable, the addresses specified by SABR always begin at 0200

and all other addresses are relative to this address. At loading time, the Linking Loader will properly
adjust all addresses. For example, if 0200 and 1000 are the relative addresses of A and B, respectively,
in the assembled program, and if A is loaded at 2000, then B will be loaded at 1000 + 1600 or 2600.
Al I programs to be assembled by SABR must be arranged to fit into one field of memory not
counting page 0 of the field or the top page (7600 - 7777) of the field. If a program is too large to
fit into one field, it shou Id be split into several subprograms.
Explicit CDF or CIF instructions are not needed by SABR programs because of the availability of external subroutine calling and COMMON storage. Explicit CDF or CIF instructions cannot
be properly assembled by SABR.

3.5

THE SYMBOL TABLE
Entries in the symbol table are variable in length. A one or two character symbol requires

three symbol table words. A three- or four-character symbol requires four words, and a five- or six-character symbol, five words. Thus, for long programs it may be to the user's advantage to use short symbols wherever possible.
The symbol table, not counting permanent symbols, contains 2644

words of storage. How10
ever, this space must be shared with the table when unresolved forward and external references are
temporarily stored as two-word entries. If we may assume that a program being assembled never has
more than 100

of these unresolved references at any one time, this leaves 2464 words of storage
10
10
for symbols. Using an average of four words per symbol, this al lows room in the program for 616
10
symbols.
Symbol table overflow is a fatal error condition which generates the error message S.

3-8

CHAPTER 4
SABR OPERATING PROCEDURES

This chapter describes how to assemble a program source using SABR. The procedure for
loading a binary tape of an assembled program is described in Chapter 6.

4. l

LOADING SABR IN A BASIC PDP-8 SYSTEM
Procedure
Make sure the Binary Loader is in memory, say in field n.

Set the console switches as fol lows:
Instruction Field = n, Switch Register= 7777.

4.2

Press LOAD AD Dress.

Insert the SABR binary tape into the reader.

If using the high-speed reader, depress Switch Register Bit 0.

Press ST ART.

SABR will now be loaded into memory by the Binary Loader; portions of
SABR wi 11 load into field 0 and field 1.

LOADING SABR IN A DISK MONITOR SYSTEM
Procedure
.J.

Make sure the Disk Monitor is in memory.

(Type CTRL/C or ST ART at 07600. )

When the Monitor responds with a dot, call the system Loader as follows:
.LOAD)

()represents typing the RETURN key)

Insert the SABR binary tape in the reader.

Answer the loading command dialogue as fol lows:
* IN-R:) for high speed reader or *IN-T:) for ASR reader

*
*ST= )
t <CTRL/P> t <CTRL/P>

After typing the second CTRL/P t it is necessary to reposition the tape in the
reader for Pass 2.

t CTRL/C and CTRL/P are typed by holding down the CTRL key while typing the C or P key.
4-1

Procedure

SABR is now loaded into memory, partly in field 0 and partly in field l. It
may be saved on the user's system device by responding to the monitor's dot
as follows:
. SAVE SABR! 0-7177;200)
. SAVE SABl ! 700, 1700-12427; ~

SABR is now saved on the user's system device and may be cal led as follows:
• SABl )
.SABR)
The field l portion must be called first.

4.3

OPERATING SABR

It is assumed that the programmer has written his program in SABR language and punched this
source program on paper tape in ASCII code. The source tape may have been sp Ii t into severe I separate
tapes by placing a PAUSE statement at the end of each section except the last. The last tape must have
an END statement at the end.
After SABR has been ioaded into memory, it is used to assembie the source program. In Pass
l the relocatable binary version of the user's program is created and, at the end of this pass, the symbol
table is either typed or punched, according to whether this listing is to be typed or punched. Pass 2 is
the listing pass. The assembly is carried out as follows.

NOTE

If SABR has been saved on the System I/O device, as
in Section 4.2, it will start automatically at Step 3
below when called into memory. The source tape (first
section) should be inserted in the reader before operation begins.

Procedure
Set the console switches as fol lows:
Data field = 0, Instruction Field = 0, Switch Register = 0200.

Press LOAD AD Dress and START.

SABR now types a sequence of two or three questions;
HIGH SPEED READER?
HIGH SPEED PUNCH?
LISTING ON HIGH SPEED PUNCH?

4-2

Procedure
These questions must be answered with Y if the answer is yes. Any other
answer is assumed to be no. The third question is typed oniy if the second
is answered Y. If the third is answered Y, both the symbol table and the
listing are punched on the high-speed paper tape punch. Otherwise, they
are typed on the teletypewriter. The user need not wait for the ful I question
to be typed before responding.

As soon as SABR has echoed the user's response to the last question, turn on
the punch device and, if it is being used, the ASR reader. If the low-speed
reader is used, the error message E indicates that the user has waited too
Iong before turning the reader on. The user must begin again.

At this point, Pass l begins. SABR reads the source tape and punches the
binary tape. After the binary tape has been completed, SABR types or
punches the program symbol table.

If the source tape is in several sections (separate tapes with PAUSEs at the
end of all except the last), SABR halts at the end of each section. At this
point, insert the next section in the reader and then press CONTinue.

At the end of Pass 1, SABR halts.

If an assembly listing is desired, reposition the beginning of the source tape
in the reader and if using the ASR reader, set it to ST ART, and then press
CONTinue.

At the end of Pass 2, SABR again halts. To restart SABR for assembling
another program, press CONTinue.

To restart SABR at any time, press STOP, set the Switch Register= 0200,
press LOAD AD Dress and START. However, the first pass must always be
repeated.

4.4

After assembling in a Disk Monitor environment, control may be returned
to Monitor by restarting at iocation 7600.

OPERATING PROCEDURE FOR USE AS FORTRAN PASS 2
In addition to being a stand-alone assembler, SABR also serves as Pass 2 of 8K FORTRAN

compilation. For this purpose, the use of SABR is slightly different from that described in Section 4.3.
However, SABR must stil I be loaded into memory as described in Section 4. l or 4.2. This difference in
the operation of SABR is due only to the unusual format of the FORTRAN Compiler Pass l.
The Compiler, in one pass, converts the user's FORTRAN source into a symbolic machine
language program tape. SABR then converts the symbolic tape into relocatable binary. However, the
symbolic tape produced by the Compiler is not a standard format SABR language tape. It is arranged
as shown below.

4-3

L
E
A
D
E
R

s
F

E
p
A
R
A
T

0
R
T
R

Main part of program;
Executable code.

END

Symbol Definitions
Common, Arrays,
Data and
Program Entry
point.

p
A

u
s
E

T
R
A
I
L
E
R

'True Start
3 ft of blank tape

The tape is arranged this way because the data at the end of the tape cannot be inserted in
the midst of the executable code, and some data which should be at the beginning of the tape is not
known until later. Thus, the true start of the symbolic program is near the end of the symbolic tape,
preceded by a segment of leader/trailer code and fol lowed by a PAUSE statement.

To assemble such a tape with SABR, one of three methods must be followed. Actually, the
general procedure is the same as that described in Section 4. 3, but it differs in special details. The
differences are aii covered by the three methods expiained beiow.

4.4. l

Method l
The simplest method is to cut the symbolic tape into two parts. The cut should be made at

the middle of the blank tape which separates the executable code from the symbol definitions. The
latter section of the tape should then be marked "Section 1 11 and the former section (the executable
code} should be marked "Section 2. 11 Assembly then proceeds with the two-part symbolic tape exactly
as described in Section 4. 3.

4.4.2

Method 2
The user may avoid actually cutting the symbolic tape by manipulating the tape as if it were

in two parts as explained above. The tape should initially be inserted in the reader with the separator
blank tape over the read-head. When SABR halts at the PAUSE statement at the physical end of the
tape, the user should reposition the tape, putting the physical beginning of the tape in the reader.
Then press CONTinue. The assembly pass will end at the separator blank tape code. The assembly
listing can be produced in a similar manner, pressing CONTinue to start the listing pass.

4-4

4.4.3

Method 3
The third method requires SABR to pass over the symbolic tape two times for each pass of the

assembiy. However, it ai iows the tape to be inserted at its physi cai beginning. It is based on the fact
that a symbolic tape output by the FORTRAN Compiler has as its physical first line the special pseudoop, FORTR. This pseudo-op has no effect except when a symbolic tape output by the Compiler is assembled using this third method.
Procedure
Insert the symboi i c tape in the reader at its physi cai beginning.

Start SABR as usual.

Sensing the FORTR statement as the first line, SABR ignores all further
data until after it passes over the END statement. SABR then begins the
actual assembly by processing the symbol definitions, etc., which are at
the latter end of the tape.

Then, SABR halts at the PAUSE statement which is at the physical end of
the tape. At this time the user should reposition the symbolic tape in the
reader at the physical beginning of the tape, and then press CONTinue.
SABR now assembles the executable code portion of the tape in the normal
way.

If an assembly listing is desired, proceed as in Method 2 after SABR finishes
the assembly pass.

4-5

CHAPTER 5
THE LINKING LOADER

5. 1

INTRODUCTION
Relocatable binary program tapes produced by SABR assembly are ioaded into memory by using

the 8K System Linking Loader. The Linking Loader is capable of loading and linking a user's program
and subprograms in any fields of memory. It is even capable, in a special way, of loading programs
over itself. The Linking Loader also has options which give storage maps and core availability.
The Linking Loader requires a PDP-8/I, -8/L, -8, -8/S or -5 Computer with at !east 8K
words of core memory.

Either high-speed or ASR paper tape input is acceptable, however, a high-

speed reader is highly recommended.
The software requirements are:
a.

Binary paper tape copy of the Linking Loader

Relocatable binary paper tape copies of both Part 1 and Part 2 of the 8K System Library

c. The relocatable binary paper tapes of the user's own program and subprograms which
have been produced by assembling his programs with SABR.

5.2

LOADING WITH THE LINKING LOADER
Generally speaking, the Linking Loader is capable of loading any number of user and Library

programs into any field of PDP-8 memory. These programs are loaded consecutively via the high-speed
reader (or the ASR reader). The choice of which field to load each program into is a Switch Register
option. Usually, several programs may be loaded into each field.

Because of the space reserved for

the Linkage Routines the available space in field 0 is three pages smaller than in all other fields.
Any COMMON storage reserved by the programs being loaded is allocated in field 1 from
location 0200 upwards. The space reserved for COMMON is obviously subtracted from the available
loading area in field 1. The program reserving the largest amount of COMMON storage must be loaded
first.
The Linking Loader uses the following special method to enable loading data over itself.
When the Linking Loader encounters data which must be loaded over itself, it punches this data onto
paper tape in RIM format. Then, after the user has finished loading all his relocatable binary program
tapes, he simply loads the RIM format tape using the standard RIM loader.
The Run-Time Linkage Routines which are necessary to execute SABR programs (see Section
3. 3.4) are automatically loaded into the required areas of every field by the Linking Loader as a part
of its initialization. For the user, the only required knowledge of these routines is the particular areas
of core they occupy (see Appendix D).

5-1

The 8K System Library subprograms (See Appendix E), which may be used by any SABR program, are loaded in the same way as any other relocatable binary programs. Only those library programs which the user's programs actually call need to be loaded.

5.3

LOADING INFORMATION OPTIONS
During the loading operation with the Linking Loader, two user options are available to

obtain information about what has al ready been loaded. The Switch Register is used to select these
options. Either option may be selected after any program has finished loading.

I WARNING I
If the ASR punch is turned on, it must be turned off
before selecting these options.

The Switch Register bits used are as follows:
BIT 0 = l selects the Core Availability option;
BIT l = 1 selects the Storage Map option.
The Core Avai iabi iity option causes the number of free pages of memory In every fie id of
memory to be typed in a list on the Teletype. For example, if the user has a 16K configuration, a list
like the following might be typed.

0002
0010

0030
0036

(number of free pages in field O)
(number of free pages in field l)
(number of free pages in field 2)
(number of free pages in field 3)

The number of pages initially available in field 0 is 0033 and in all other fields is 0036.
The Storage Map option causes a list of all program entry points to be typed, along with
the actual address at which they have been loaded. The entry points of programs which have been called
but which have not been loaded are also listed along with a U flag for undefined. Such flagged programs must be loaded before execution of the user's programs is possible. The Core Availability list is
automatically appended to the Storage Map. A sample is shown below.
MAIN
READ
WRITE
IOH
SETERR
ERROR
TTY OUT
HS OUT

10200
01055

01066
03031
00000 u
00000 u
00000 u
00000 u

5-2

00000 u

TT YIN
HSIN
FDV

CLEAR
!FAD
FMP
ISTO
STO
FLOT
FAD
DIV
!REM
FSB
FLOAT
FIX
IFIX
CHS
0011
0033

5.4

00000
04722
05247
05131
04632
05074
04447
05210
04010
00000
00000 u
04000
05046
04513
04561
05231

HOW TO LOAD THE LINKING LOADER
The Linking Loader must be ioaded into the highest avaiiabie fieid of memory.
Procedure
Make sure the Binary Loader is in memory, for example, in field m.
2

Let h represent the number of the highest field in the user's configuration.

Set the console switches as follows:
Data Field= h, Instruction Field= m, Switch Register= 7777.

5.5

Press LOAD AD Dress.

Place the binary paper tape of the Linking Loader in the reader.

If using a high-speed reader, depress Switch Register Bit 0.

Press START. The Linking Loader will now be loaded into memory.

OPERATION OF THE LINKING LOADER
The Linking Loader is used to load the user's relocatable programs and 8K Library

subprograms as outlined below.
NOTE
The program or subprogram which uses the largest
amount of COMMON storage should be loaded
first. (The Library subprograms do not use COMMON.)

5-3

Procedure
After the Linking Loader has been loaded into the highest memory field, h,
the user should set the console switches as fol lows: Data Field= h, Instruction
Field= h, Switch Register= 0200.

Press LOAD ADDress.

Place the relocatable binary tape for the first program to be loaded in the
reader. Position the tape with leader code in the reader.

Set Switch Register to 0000. Then, if loading via the ASR reader is required,
raise Switch Register Bit 6. If the user does not have a high-speed punch, he
should raise Switch Register Bit 7. Finally, set Switch Register Bits 9-11 to
the number of the field into which the first program or subprogram is to be
loaded.
Switch Register*

I ~ I I21 1 1 1~ 1 1° I: I I I
3

1°

V'l

..,

.......

0..,
Cl>
"U

..,

Cl>

;::c

Cl>

0
a_

..,Cl>

;::c

'"---....._,..---)
Number of
Loading Field

g..

Example:

If the user wishes to load his first program into field 3, and if he has no highspeed VO device, then he should set the Switch Register to 0063 before the next step.

Press START.

The user's relocatable binary program wi 11 now be loaded. When loading is
completed, the Linking Loader halts.

The user may now either load another program or select one of the options in
steps 9 and 10.

To load another program, insert the program relocatable binary tape in the
reader, set Switch Register Bits 9-11 to the number of the field the program
is to be loaded into, and then press CONTinue.

To select the Core Availability option, set Switch Register Bit 0 = 1, and
press CONTinue.

To select the Storage Map option, set Switch Register Bit 1 = 1, and press
CONTinue.

If the ASR punch is turned on for possible RIM format data punching, as explained in Section 5.2, ensure that it is turned off before selecting either of
the options. Turn it on again after the typing of the option is completed.
*Al I other Switch Register bits are irrelevant.

5-4

Procedure

The user may continue loading more programs as in step 8 after using either
of the options.
Any time the Linking Loader halts, the user may access memory directly via
the DEPosit and EXAMine console switches. After this is done the Linking
Loader may be restarted via the console switches at location 7200 (in the
highest field, where the Linking Loader resides).

5-5

CHAPTER 6

DEMONSTRATION PROGRAM

The following demonstration program is aSABRprogramshowingthe use ofthe libraryroutines.
The program is written to add two integer numbers, convert the result into floating-point, and type the
result in both integer and floating-point format. The source program was written and listed using the
Symbolic Editor; the Disk Monitor System was used during assembly; and the assembled program was then
ioaded and run using the 8K Linking Loader.
The system configuration consisted of a PDP-8/I with 8K words of core, DF32 Disk, ASR33
Teletype, and high-speed reader and punch. The Disk Monitor System, Symbolic Editor, and SABR
Assembler were available on the disk. The ASR33 paper tape reader was used during assembly for
demonstration (printout) purposes.

Demonstration Program

Program

Comment
After writing the source program it was printed and punched using the Symbolic Editor.

ENTRY START
START,

CALL
TAD
TAD
DCA
CALL
ARG
CALL
ARG
CALL
ARG
ARG
CALL
ARG
CALL
ARG
CALL
ARG
HLT

FORMT,

TEXT

2
2

BLOCK
END

0,0PEN
A

/INITIALIZE IO DEVICES
/COMPUTE C = A + B

c
J,FLOAT /CONVERT TO FLOATING POINT

1, STO
D

2,WRITE
N
FORMT
1,IOH

/INITIALIZE THE IO HANDLER
/DEVICE NUMBER 1 = TELETYPE
/FORMAT SPECIFICATION
/TYPE THE INTEGER NUMBER

c
1, I OH

/TYPE THE FLOATING POINT NR

1, I OH
0
"C 'THE

/COMPLETE THE IO

ANSWERS ARE',IS,F7.2)"

6-1

Demonstration Program
Comment
CTRL/C was typed after the asterisk
to return control to the Disk Monitor.

*
.SABl

SABR was transferred from the disk
into core.

• Sl\BK

The source program tape was placed
in the teletype reader.

When started, SABR printed its
identification and initial dialogue
questions which were answered.

PDP-8 SABR DEC-08-A282-12
HIGH SPEED READER? N
HIGH SPEED PUNCH? Y
LI S TI NG 0 1-.J HI GH SP D:: D P !J r-.J CH ? N

The TTY reader must be set to START
within 3 seconds after typing N to the
last question. Otherwise, as was the
case here, the error message wi ti
appear, and SABR must be restarted
at location 0200, as was done here.

E AT
HIGH S?~ED READER?
N
HIGH SPEED PUNCH? Y
LISTING ON HIGH SPEED PUNCH? N

The initial dialogue questions are
repeated and again answered.
After typing the N to the last question, the TTY reader was immediately
set to ST ART and assembly commenced.

0257

026:,~

FLOAT
FORMT
IOH

N
OPEN
START
STD
WRITE

The Symbol Table concluded the
assembly.

0 261
0262
00t2l0EXT
0240
0000EXT
0256
0000EXT
0200EXT
000f3EXT
000(2lEXT

Here the source program tape was
again placed in the TTY reader and
the CONTinue switch was depressed.
The program listing was printed.

6-2

Demonstration Program

ENTRY START
0200
0201
0202
0203
02l;;J4
020 5
02t;;J6
020 7

0 2 l:?J
0211
0212
0213
0214
0215
0216
0217
0220
0221
0222
0223
0224
0225
0226
0227
0230
023i
0232
0233
0234
0235
0236
0237

S T_l\RT,

4033
001~2

1257
1260
3261
40 3 3
010 3
6201
0261
4033
010 4
6201
0262
4033
0205
6201
0256
6201
0240
4033
0106
6201
0261
4033
010 6
6201
0262
4033
010 6
621 1
0000
7402

0240
52J 4 7
0241
2410
0242 - 0540
0243 011 6
0244 2327
0245 0522
0246 2340
0247 0122
0250 0547
0251
5411
0252
6554
0253 0 667
0254
5662
0255
5100
0256 0001
0257 02102
0260
0002
0261
0000
0262 0000
0263 0000
0264 0000

CALL

0, OPEN

/I NI TI ALI ZE

TAD
TAD
DCA
CALL

/COMPJTE c

ARG

Ci\LL

1, STO

ARG

CALL

2, WRITE /I NI TI ALI ZE

ARG

/DEVICE NUMBER 1

ARG

FORMT

/FORMAT SPEC I FI CATI ON

CALL

1, I OH

/TYPE THE: INTEGER NUMBER

ARG

CALL

1, I OH

ARG

CALL

1, I OH

ARG

IO DEVICES

06
05
01
06
05
01
06
05
01
05
01
06
05
01
06
05
01

=A + B

c
1,FLOAT /CONVERT TO FLO_C\TI NG POINT

TH~

IO HANDLER

= TELETYPE

/TYPE THE FLOATING POINT NR

/COMPLETE TH2: IO

HLT
FORMT,

TEXT "C 'THE ANSWERS ARE',I5,F7.2)"

N,
A,
g,

1
2
2
0
BLOCK

c,
o,

END

6-3

Demonstration Program
Comment

Program

The 8K Linking Loader was loaded
into core using the Binary Loader,
and started at location 0200 of field l.
When started, the Linking Loader
printed its identification.
PDP-8 LINKING LOADER DEC-08-A283-05

ST,~RT

OPEl\J
FLO,AT
STO
WRITE
IO:-!

READ
SETERR
ERRO~

TTY OUT
HS OUT
TTYIN
HSIN
FDV

CLEAR
IFAD

FMP
ISTO
FLOT
FAD
DIV
IREM
FSB
FIX
IFIX
CHS
ABS
IABS

Library Tape Part l was loaded into
core by placing the tape in the TTY
reader, setting the reader to START,
and pressing CONTinue.

0100(2}
06125
05034
04444
0130 2
03142
01271
06200

After loading the library subprograms,
Switch Register bit l was set to l, and
CONTinue was pressed to get the
storage page of the programs and subprograms loaded into core.

06303
0 6027
0 6~j 5 5
060 2!0
0 6045
1

0 4 71 1
05227
0 51 1 6
04623
0 50 61
0 5153
0 4010
0 5445
05616
0 42l 0 0
0 4510
04556
0 521 1
0 5636
05670

MPY

05400

IRDS'tJ
CKI 0
EXIT
CLRERR

05713
0 6121
0 6142
0 6231

0004
0036

The last two numbers represent the
number of free {available) pages in
each core field -- 0004 free pages
in field 0, and 0036 free pages in
field l.

6-4

Comment

Program

To execute the compiled program,
the 5'witch Register was set to 01000,
the starting address of the main program (determined from the Storage
Map).
The LOAD ADDress switch was pressed and then START switch was pressed.

THE ANSWERS ARE

The program ran as planned, producing
the desired results.

6-5

APPENDIX A
ASCII* CHARACTER SET

Character

Code

Character

Code

Character

NULL
BELL
TAB
LINE FEED
FORM
RETURN
SPACE

200

0
l

A
B

$
%
&

*
+

207
211
212

2
3

260
261
262
263

214

264

215

265

240

266

241
242
243
244
245
246
247
250
251
252
253
254
255

267
270
271
272
273
274
275
276
277

G
H

8
9

>
?

c
D

I
J
K

L
M
N
0
p
Q

256

v
w
x

257

307

310
311
312
313
314
315
316
317
320
321
322
323
324
325

326
327
330
331
332
333

334
335
336
337
377

RUBOUT

A-1

301

302
303
304
305
306

rL
]

*An abbreviation for U.S.A. Standard Code for Information Interchange.

Code

APPENDIX B
PERMANENT SYMBOL TABLE

Memory Reference Instructions
Mnemonic

Code

AND

0000

TAD

1000

ISZ
INC

2000
2000

DCA
JMS

3000
4000

JMP
I

5000
0400

Operation

Event Time

combine C(AC) and C(MEM) by logical AND
and store result in AC
combine C(AC) and C(MEM) by two's
complement addition and store result in AC
with carry added to the LINK
increment C(MEM) and skip if result is 0
same as ISZ except should be used only when it
is known that an actual skip cannot occur (see
Section 3 .3 .5)
deposit C(AC) into MEM and clear the AC
jump to subroutine (actually deposit the current
value of the PC into MEM and jump to /V\E.M + 1)
jump to MEM location
indirect memory reference
Operate Microinstructions: Group 1

l""I A

' - L...r"\

CLL
CMA
CML
RAR
RTR
RAL
RTL
IAC
CIA
STA
STL
NOP

7200
7100
7040
7020
7010
7012
7004
7006
7001
7041
7240
7120
7000

clear AC
clear LINK
comp Iement AC
complement LINK
rotate AC and LINK 1 bit right
rotate AC and LINK 2 bits right
rotate AC and LINK l bit left
rotate AC and LINK 2 bits left
increment AC
negate AC (CMA IAC combined)
set AC (C LA CMA combined)
set LINK (CLL CML combined)
no operation

1
2
2
4*
4*
4*
4*
3
1 ,3
1,2
1,2

Operate Microinstructions: Group 2
CLA
SMA
SZA
SPA
SNA
SNL
SZL

7600
7500
7440
7510
7450
7420
7430

clear AC
skip if AC negative
skip if AC zero
skip if AC positive or zero
skip if AC non-zero
skip if LINK non-zero
skip if LINK zero

* 3 for PDP-8

B-1

1
1
l

1
1
1

Operate Microinstructions: Group 2 (Cont)

Mnemonic

Code

Operation

Event Time

SKP
SPC
OSR

7410
7710
7404

l
l

HLT

7402

skip unconditiona I ly
(SPA CLA combined}
inclusive OR switch register with C(AC);
result to AC
halt
IOT Microinstructions

Program Interrupt
ION
IOF
Keyboard/Reader
KSF
KRB
Teleprinter/Punch
TSF

TLS
High-Speed Reader (Type PC02)
RSF
RRB
RFC
High-Speed Punch (Type PC03)
PSF
PLS

6001
6002

turn interrupt on
turn interrupt off

6031
6036

skip if keyboard/reader flag= 1
c fear AC & read keyboard buffer, and
cf ear keyboard flag

6041
6046

skip if teleprinter/punch flag = 1
load teleprinter/punch buffer,
select and print, and c Iear teleprinter/
punch flag

6011
6012
6014

skip if reader flag = l
read reader buffer and c Iear flag
clear flag and buffer and fetch character

6021
6026

skip if punch flag =l
clear flag and buffer, load and punch
Pseudo-Operators

ABYSM
ARG
BLOCK
CALL
COMMN
CPA GE

DECiM
DUMMY
EAP
END
ENTRY
FORTR
IF
LAP
OCTAL
OP DEF
PAGE

Direct Absolute Symbol Definition
Argument for Subroutine Cal I
Reserve Storage Block
Ca 11 Externa I Subroutine
Common Storage Definition
Check if Page Will Hold Data
Dec i ma I Conversion
Dummy Argument Definition
Enter Automatic Paging Mode
End of Program
Define Program Entry Point
Assemble FORTRAN Tape
Conditional Assembly
Leave Automatic Paging
Octal Conversion
Define Non-Skip Operator
Terminate the Page

B-2

3
4

Pseudo- Operators (Cont)
Mnemonic

Operation

Code

PAUSE

Pause for Next Tape

REORG
RETRN
SK PDF
TEXT

Terminate Page and Reset Origin
Return from External Subroutine
Define Skip-Type Operator
Text String
Floating-Point Accumulator

ACH
Ar't..A

1-\\,,..IV\

ACL

20*
21*
22*

high-order word
middle woid
low-order word

*The Floating Point Accumulator is in field 1.

B-3

APPENDiX C
ERROR MESSAGES

C. l

SABR
Because SABR is a one-pass automatic paging assembler for binary relocatable programs,

object errors are difficult to correct. If there are errors in the source, the assembled binary code will
be virtually useless. Both errors E and S are fatal; assembly halts when they are encountered. The
other types of errors are not fatal, but they cause the line in which they occur to be treated as a comment and thus essentially ignored. An address label on such a line will remain undefined and no space
is reserved in the binary output for the erroneous data.
During the assembly pass error messages are typed on the teletype as they occur.
Example:

LOC

-+-0004

This means that an error of type C has occurred at the fourth instruction after the location tag
LOC.

This line count includes comment lines and blank lines.
During the listing pass, the error is typed in the address field of the instruction line.
The following error messages may occur.
A

Too many or too few ARGs follow a CALL statement.

An i Ilegal character appears on the line. This could possibly be an 8 or 9 in an
octal digit string or an alphabetic character in a digit string.

A symbol is multiply defined (occurs only during Pass 1). It is impossible to
resolve multiple definitions during Pass 2; therefore, listings of programs which
contain multiple definitions wil I have unmarked errors.
An illegal syntax has been used. Below are listed the types of illegal syntax
that may occur.

A pseudo-op with improper arguments.

A quote mark with no argument.

A non-terminated text-string.

A memory reference instruction with improper address.

An i Ilegal combination of micro-instructions.

There is no END statement.

C-1

one of the following:
a.

The symbol table has overflowed. This can be corrected by using fewer

symbols, using shorter symbols, or by breaking the program into smaller parts.
b.

Common storage has been exhausted.

/vbre than 64 different user-defined symbols have occurred in a core page.

/vbre than 64 externa I symbo Is have been dee la red •

One further type of error may occur. This is an undefined symbol. Because SABR is a onepass assembler, an undefined symbol cannot be determined unti I the end of the assembly pass, so the
error diagnostic UNDF is given in the symbol table listing. (Refer to the discussion of the Symbol
Table at the end of Appendix F .)
C .2

LINKING LOADER
If during the process of loading a program or subprogram the Linking Loader encounters an

error, the user is notified by an error message; the partially loaded program or subprogram is ignored,
removed from the field, and core is freed. The error messages are typed out in the form
ERROR

xx xx

where XXXX is the error code number.
Error Code

Explanation

0001

More than 64 10 subprogram names have been seen by
the Loader (6410 subprogram names is the capacity of
the Loader's symbol table}.

0002

The current field is full, or load was to nonexistent
memory.

0003

The current subprogram has too large a COMMON
storage assignment. (Subprogram with largest common
storage declaration must be loaded first.) This is a
semi-fatal error. Re-initialize the Linking Loader as
explained below and reload the programs in the proper
order.

0004

Checksum error in input tape. If the error persists, reassembly is necessary.

0005

Illegal Relocation Code has been encountered. This
can occur only if the relocatable binary tape is bad or
if the user is using it improperly (e.g., not starting at
the beginning of the tape, or reader error, or punch
error). If the error persists, re-assembly is necessary.

Recovery from errors 2, 4, and 5 is accomplished by repositioning the tape in the reader to
the leader code at the beginning of the subprogram and then pressing CONTinue. When attempting to
recover from one of these errors, no other program should be loaded before reloading the program which

C-2

caused the error. Obviously, on Error 2 a different field should be selected before pressing CONtinue.
The entire loadina Process mav be restarted via the console switches. at anvr time
re- - - - bv
-- /
- "'

initializing the Linking Loader. To do this, set the console switches as fol lows: Data Field = h (the
field where the Linking Loader resides), Instruction Field = h, Switch Register = 6200; then press LOAD
AD Dress and ST ART.

C.3

LIBRARY PROGRAM
During execution, the Library programs check for certain errors and type out the appropriate

error messages in the form
11

XXXX 11 ERROR AT LOC NNNN

where XXXX specifies the type of error, and NNNN is the location of the error. When an error is
encountered, execution stops, and the error must be corrected.
When multiple error messages are typed, the location of the last error message is relevant to
the user program. The other error messages are to subprograms cal led by the statement at the relevant
location.
Explanation

Error Code
11

ALOG 11

Attempt to compute log of negative number

ATAN 11

Result exceeds capacity of computer

D1vz

Attempt to divide by 0

EXP

Result exceeds capacity of computer
11

FIPW

FMTl 11

Error in raising a number to a power
Multiple decimal points
E or • in integer

FMT3 11

II lega I character in I, E, or F field

FMT4 11

Multiple minus signs

FMT5

Invalid FORMAT statement

FLPVv'll

Negative number raised to floating power

FPNT 11

Floating - point error may be caused by:
Division by zero; floating - point overflow;
attempting to fix too large a number.

SQRT 11

Attempt to square root a negative number

C-3

To pinpoint the location of a Library execution error:
Step

Procedure
From the Storage Map, determine the next lowest
numbered location (external symbol) which is the
entry point of the program or subprogram containing
the error.

Subtract in octal the entry point location of the program or subroutine containing the error from the LOC
of the error in the error message.

From the assembly symbol table, determine the
relative address of the external symbol found in step
l and add that relative address to the result of step 2.

The sum of step 3 is the relative address of the error,
which can then be compared with the relative addresses of the numbered statements in the program.

C-4

APPENDIX D
FREE PAGE 0 LOCATIONS

Because the Library Linkage Routines must be in core when SABR assembled programs are run,
certain core locations are not available as fol lows:
Field 0
Fi e Id 0 , l , 2 , . . .

Locations 0400 - 0777
Locations 0007 and 0033 - 0073

Thus in every field of memory the following page 0 locations are available to the user:

0000 - 0006
0010 - 0017
0023 - 0032
0074 - 0177

for interrupts, debugging, etc.
auto-index registers
arbitrary
arbitrary

Locations 20, 21, 22 in field l are used for the Floating-Point Accumulator. The user
should use these locations with great care.
When using the Library routines, locations 20-32 in the field where the routines reside, are
used for temporary storage by the routines.
Locations 176 and 177 in the field where the I/O handler routines (IOH) reside are used for
temporary storage by the 1/0 handler.

D-1

APPENDIX E
THE LIBRARY SUBPROGRAMS

The Library is a set of subprograms which may be CALLed by any FORTRAN/SABR program.
The relocatable binary versions of these subprograms are arranged in two paper tapes for the convenience
of the user. Part 1 contains those subprograms which are used by almost every FORTRAN/SABR program.
All the Library subprograms are described below.
Many of the subprograms reference the Floating-Point Accumulator located at ACH, ACM,
ACL (20, 21, 22 of field 1),

E. l

INPUT/OUTPUT
READ is called to initialize the I/O handler before reading data. WRITE is called to initia-

lize the I/O handler before writing data. IOH is called for each item to be read or written. IOH must
also be called with a zero argument to terminate an input-output sequence. (Refer to Chapter 6.)
All of these programs require that the Floating-Point Accumulator be set to zero before they
are called.
Examples:
rAI I

\.,.r'\L.L.

ARG
ARG
CALL
ARG

2, READ
(n

/n =DEVICE NUMBER
/fa =ADDR OF FORMAT

l ,IOH
data l

/data l =ADDR OF HIGH
/ORDER WORD OF

/FLOATING POINT NUMBER
CALL
ARG

1,IOH
data 2

CALL
ARG

1,IOH
0

CALL
ARG
ARG

2,WRITE
(n
fa

The following device numbers are currently implemented:
1. Teletype keyboard/printer
2. High-speed reader/punch

E-1

E .2

FLOATING-POINT ARITHMETIC
FAD is called to add the argument to the Floating-Point Accumulator
CALL
ARG

1, FAD
addres

FSB is called to subtract the argument from the Floating-Point Accumulator.
CALL
ARG

l ,FSB
addres

FMP is called to multiply the Floating-Point Accumulator by the argument.
CALL
ARG

l ,FMP
addres

FDV is called to divide the Floating-Point Accumulator by the argument.
CALL
ARG

l ,FDV
addres

CHS is called to change the sign of the Floating-Point Accumulator
r'-r'\L
A 1 1
.. L.

rue

Vt'-11..I

All of the above programs leave the result in the Floating-Point Accumulator. The address
of the high-order word of the floating-point number is 11 addres 11 •
STO is cal led to store the contents of the Floating-Point Accumulator in the argument address
CALL
ARG

l ,STO
storag

/storag =ADDRESS WHERE
/RESULT IS TO BE PUT

IFAD is called to execute an indirect floating point add to the Floating-Point Accumulator.
CALL
ARG

l ,IFAD
ptr

/ptr =2-word POINTER
/TO HIGH ORDER
/ADDRESS OF FLOATING

/POINT ARGUMENT
ISTO is called to execute an indirect floating point store.
CALL
ARG

l ,ISTO
ptr

CLEAR is ca Iled to c fear the Floating-Point Accumulator.
CALL

0 ,CLEAR

E-2

FLOT is caiied to convert the integer contained in the AC (processor accumulator) to a
floating point number and store it in the Floating-Point Accumulator.

FIX is called to convert the number in the Floating-Point Accumulator to a 12-bit signed
integer and leave the result in the AC.
CALL

O,FIX

ABS leaves the absolute value of the floating point number at 11 addr 11 in the Floating-Point
Accumulator.
CALL
ARG

E .3

1,ABS
addr

INTEGER ARITHMETIC
MPY is called to multiply the integer contained in the AC by the integer contained in

addr. 11 The result is left in the AC.
CALL
ARG

1,MPY
addr

DIV is called to divide the integer contained in the AC by the integer contained in 11 addr. 11
The result is left in the AC.
CALL
ARG

1,DIV
addr

!REM leaves the remainder from the last executed integer divide in the AC.
CALL
ARG

1 ,IREM

(The argument is ignored.)
!ABS leaves the absolute value of the integer contained in 11 addr 11 in the AC.
CALL
ARG

1 ,IABS
addr

IRDSW reads the value set in the console switch register into the AC.
CALL

0 ,IRDSW

E-3

E .4

SUBSCRIPTING
SUBSC is called to compute the address of a subscripted variable. The address is left in the

AC. When SUBSC is cal led, it assumes that the AC contains the first dimension of the array. This
dimension should be positive if the subscripted variable is an integer, and negative if the subscripted
variable is a floating point number.
Example:
Assume S is a 20 X 20 floating-point array.
8
8

E .5

(20

TAD
CIA
CALL
ARG

3 ,SUBSC
i1

ARG

base

/i 1 =ADDRESS OF 2ND
/SUBSCRIPT
/i2 =ADDRESS OF lST
/SUBSCRIPT
/BASE ADDRESS
/OF ARRAY

FUNCTIONS
SQRT leaves the square root of the floating-point number at 11 addr 11 in the Floating-Point

Accumulator.
CALL
ARG

l ,SQRT
addr

SIN, COS, TAN leave the specified function of the floating-point argument at 11 addr 11 in
the Floating-Point Accumulator.
CALL
ARG

l ,SIN
addr

AT AN leaves the arctangent of the floating-point number at 11 addr 11 in the Floating-Point
Accumulator.
CALL
ARG

1 ,ATAN
addr

ALOG leaves the natural logarithm of the floating-point number of 11 addr 11 in the FloatingPoi nt Ace umu lator.
CALL
ARG

l ,ALOG
addr

E-4

EXP raises 11 e 11 to the power specified by the floating-point number at 11 addr 11 and leaves
the result in the Floating-Point Accumulator.
CALL
ARG

1,EXP

addr

Ai I of these subprograms require that the Fioating-Point Accumulator be set to zero before

they are called.

E.6

POWERS (IIPOW I IFPOW I FIPOW I FFPOW)
These routines are called by FORTRAN to implement exponentiation. The address of the

first operand is in the AC (floating-point or processor depending on mode), and the address of the
second is an argument. The address of the result is in the appropriate AC upon return.

Function
Name

Mode of
Operand l
(Base)

Mode of
Operand 2
(Exponent)

Mode of
Result

II POW

Integer
Integer
Floating point
Floating point

Integer
Floating point
Integei
Floating point

Integer
Floating point
Floating point
Floating point

IFPOW
FIPOW
FFPOW

CALL
ARG
E.7

2,FFPOW
addr 2

/ADDRESS OF OPERAND 2

LIBRARY ORGANIZATION
Part l.

IOH 11
FLOAT 11

contains
contains

INTEGER 11
UTILITY 11
11
ERROR11

contains
contains
contains

IOH I READ I WRITE
FAD I FSB I FMP I FDV I STO, FLOT, FLOAT,
FIX, IFIX, IFAD, ISTO, CHS, CLEAR
IREM, ABS, IABS, DIV, MPV, IRDSW
TTYIN, TTYOUT, HSIN, HSOUT, OPEN, CKIO
SETERR, CLRERR, ERROR

contains
contains
contains
contains
contains

SUBSC
IIPOW, IFPOW I FIPOW, FFPOW I EXP I ALOG
SQRT
SIN, COS, TAN
ATAN

Part 2.

SUBSC 11
POWERS 11
11
SQRT 11
11
TRIG 11
11
ATAN 11
11

E-5

E .8

DECTAPE I/O ROUTINES
RTAPE and WTAPE (read tape and write tape) are the DECtape read and write subprograms

for the 8K FORTRAN and 8K SABR systems. The subprograms are furnished on one relocatable binarycoded paper tape which must be loaded into field 0 by the 8K Linking Loader, where they occupy one
page of core.
RTAPE and WTAPE allow the user to read and write any amount of core-image data onto
DECtape in absolute, non-file-structured data blocks. tv\any such data blocks may be stored on a
single tape, and a block may be from 1 to 4096 words in length.
RTAPE and WTAPE are subprograms which may be called with standard, explicit CALL statements in any 8K FORTRAN or SABR program. Each subprogram requires four arguments separated by
commas. The arguments are the same for both subprograms and are formatted in the same manner. They
specify the following:
a.

DECtape unit number (from 0 to 7).

b. Number of the DECtape block at which transfer is to start. The user may direct the
DECtape service routine to begin searching for the specified block in the forward direction rather
than the usual backward direction by making this argument the two's complement of the block number.
c.

Number of words to be transferred (1 < N < 4096).

Core address at which the transfer is to start.

In 8K FORTRAN, the CALL statements to RTAPE and WTAPE are written in the following
format (arguments are taken as decimal numbers):
CALL RTAPE (6, 128,388,LOCA)
In 8K SABR, they are written in the following format (arguments may be either octal or decimal numbers):
CALL 4, WTAPE
ARG (6
ARG (200
ARG (604
ARG LOCB

/WOULD BE SAME FOR RTAPE
/DATA UNIT NUMBER
/STARTING BLOCK NUMBER IN OCTAL
/WORDS TO BE TRANSFERRED IN OCTAL
/CORE ADDRESS, START OF TRANSFER

In these examples, LOCA and LOCB may or may not be in COMMON.
As a typical example of the use of RTAPE and WTAPE, assume that the user wants to store
the four arrays A, B, C, and Don a tape with word lengths of 2000, 400, 400, and 20 respectively.
Since PDP-8 DECtape is formatted with 1612 blocks (numbered 1-2700 octal) of 129 words each (for a
total of 207,948 words), A, B,C, and D will require 16, 4, 4, and 1 blocks respectively. Each array
must be stored beginning at the start of some DECtape block. The user may write these arrays on tape
as follows:

E-6

CALL WTAPE (O, 1,2000,A)
CALL WTAPE (0,17,400,B)
CALL WTAPE (O ,21 ,400 ,C)
CALL WTAPE (O ,25 ,20 ,D)
The user may also read or write a large array in sections by specifying only one DECtape
block (129 words) at a time. For example, B could be read back into core as follows:
CALL RTAPE (0, 17,258,B(l))
rA• ' RTAoi: (O I i 9 I 1 ?9 Bf?59)\I
CALL RTAPE (O ,20, 13 ,B(388))
""'

... ...

...

I .....

,.....

As shown above, it is possible to read or write less than 129 words by starting at the beginning of a DECtape block. It is impossible, however, to read or write starting in the middle of a block.
For example, the last 10 words of a DECtape block may not be read without reading the first 119 words
as well.
A DECtape read or write is normally initiated with a backward search for the desired block
number. To save searching time, the user may request RT APE or WTAPE to start the block number
search in the forward direction. This is done by specifying the negative of the block number. This
should be used only if the number of the next block to be referenced is at least fourteen block numbers
greater than the last block number used. For example, if the user has just read array A and now wants
array D, he may write:
CALL RTAPE (O, 1 ,2000 ,A)
CALL RTAPE (0,-25,20,D)

E. 9

DISK I/O ROUTINES (preliminary)
ODISK and CDISK (open disk and close disk) and RDISK and WDISK (read disk and write

disk) are the four DECdisk (DF32/DS32) input and output subprograms for the 8K FORTRAN and 8K
SABR systems. They are furnished on one relocatable binary-coded paper tape which is loaded into
core using the Linking Loader, where they occupy eight pages of core.
E.9.1

ODISK and CDISK
ODISK is used to open (activate) a file (named using the Linking Loader D function) so that

the file can be read or written using RDISK or WDISK. CDISK will close (deactivate) a file which was
opened with ODISK so that the contents of the file cannot be altered.
The ODISK and CDISK subprograms may be called with standard, explicit CALL statements,
in any 8K FORTRAN or 8K SABR program. ODISK requires one argument when opening a file. However, it requires two arguments when specifying or changing the size (in blocks) of a file. CDISK
always requires only one argument.

E-7

The first argument of both ODISK and CDISK is the logical number (from 1 through 10
inclusive} of the file as it was named using the Linking Loader. (Refer to Section H .3 .1 for a discussion
of logical file numbers.) The second argument to ODISK is the number of blocks (from 1 through 128)
to be saved for the fi Ie •
In 8K FORTRAN, the CALL statements to ODISK and CDISK are written in the following
format (arguments must be decimal integer numbers}:
CALL ODISK (1)
when opening a file, or
CALL OD ISK (l ,5}
when specifying or changing the size of a fi I e, and
CALL CDISK (1)
when closing an opened file.
In 8K SABR, the CALL statements to ODISK and CDISK are written in the following
format (arguments may be either octal or decimal numbers}:
CALL 1, ODISK
ARG (1
ARG (5

/LOGICAL FILE NUMBER
/NUMBER OF BLOCKS, OCTAL

when specifying or changing the size of a file, and
CALL 1 ,CDISK
ARG (1

/LOGICAL FILE NUMBER

when closing an opened file.
ODISK prepares the file named for data transfer. When running the user program using the
Disk Monitor System, ODISK uses Disk Monitor 1/0 and the three scratch blocks on disk zero for a
window whenever a fi I e is opened .
All open files should be closed before terminating program execution, thus preserving the
contents of the files.

E.9.2

RDISK and WDISK
The RDISK and WDISK subprograms may be called with standard, explicit CALL statements

in any BK FORTRAN or 8K SABR program. The ODISK subprogram must be used to open the file concerned before using the RDISK or WDISK subprograms.

E-8

Each of these subprograms requires four arguments, arranged as listed below:
1.

Logical file number (determined using the Linking Loader D function).

')
~.

Logical block of the file number (block number of the file where data transfer is to

Number of words to be transferred (from 1 through 4096)

Core address where data transfer is to start (field O) •

begin),

Both RDISK and WDISK require the arguments above.
!n 8K FORTRAN, the CALL statements to RDISK and WDISK are written in the fol !owing

format (arguments are taken as decimal numbers):
CALL RD ISK (4 ,2 ,55, LOCA)
when reading file 4, beginning with block 2, transferring 55 words, starting at the location of tag
LOCA, which may be the name of an array defined in a DIMENSION statement. WDISK would be
formatted in the same fashion.
In 8K SABR, the CALL statements to RDISK and WDISK are written in the fol lowing format
(arguments may be either octal or decimal numbers):
CALL 4, RDISK
ARG (4
ARG (2
ARG (55
ARG LOCA

/SAME FOR WD IS K
/LOGICAL FILE NUMBER
/BLOCK 0 F FILE
/WO RDS TO TRANSFER, OCTAL
/CORE ADDRESS OF START, FIELD 0

WDISK would be formatted in the same fashion.
A variable number of words may be transferred. It is not necessary to transfer in 200-word
blocks, as with the Disk /v\onitor System.

E-9

APPENDIX F
SAMPLE OF AN ASSEMBLY LISTING

This program is offered only to i I lustrate many of the features and formats of a SABR program.

The program cannot be run.
PDP-8 SABR DEC-08-A2B2
High Speed Reader ? Y
High Speed Punch? Y
Listing on High Speed Punch?
DTCA
DTSF
LOC
MUL
NAME
POINTR
SUB
St
St2
S2
S3
S4
TAG

67620P
67710P
OOOOUNDF
OOOOEXT
lOOOCOM
1013
0200EXT
0202
0214
0214
0227
0233
0177ABS
0400
0401
0402
/SAMPLE OF SABR CODE
6762
6771

0200
0201

0202
0203
0204
0205
0206
0207
0210

0177

OP DEF
SKPDF
/ABSYM
ABSYM

200

NAME,

0000
0000

0000
4067
0200 01
1407
7106
7006
6211

DTCA
DTSF
LOC
TAG
DECIM
COMMN
ENTRY
DUMMY
LAP
BLOCK

SUB,

st/

F-1

6762
6771
176
177

8
SUB

x
2

EAP
OCTAL
0
TAD I

SUB

CLL RTL;

RTL

DCA

NAME#

0211
0212
0213

3776
6201 05
2775

0214
0215
0216
0217
0220
0221
0222
0223
0224

4033
0302 06
6201 05
0400 01
6201 05
0374 01
6201
7777
1373

st 2,
S2,

INC

POINTR

CALL

3,MUL

ARG

(20

ARG

-1

TAD

(D-49
LOC, 1

0225
0226

1372
5200

0227
0230
0231
0232
0233
0234

S3,

0372
0373
0374
0375
0376
0377

4233
0004
0200
0371 01
6762
5377
0037
7501
7717
0020
1013 01
1001
7000

0400
0401
0402
0403
0404
0405
0406
0407
0410

0000
0214 01
2301
1520
0075
4052
5777
6465
6600

0411

U/ I

0412
0413
0576
0577

5376
5377
7410
7410

1000
1001
1002
1003
1004

7410
7410
5206
4062
0214 01

0371

PAUSE
TAD
JMP
CPAGE
JMS
4
NAME
(37
DTCA

S4,

PAGE
0
St2
TEXT

Y,
Z,

L. 771

(-"?
SUB
4
S4

"SAMP @ = */?456

DTSF

REORG
SKP
TAD I

F-2

1000
St2

1005
1006
1007
1010
1011
1012
1013
1176
1177

1407
1377
6211
3776
4040
0001 06
0000
1000
0333

POINTR,

TAD
DCA

(333
NAME

RETRN

SUB

END

For a multiple word instruction the actual instruction line is typed beside the first
instruction.
0650
0651
0652
0653
0654

6201
5774
7106
7006
7006

LOC2,

JMP

NAME

/OFF PAGE

CLL RTL; RTL; RTL

For an erroneous instruction, the error flag appears in the address field. The instruction
is not assembled.
0700

7200

0701

7402

N2,

CLA
CLL SKP

HLT

The page escape and literal and off-page pointer table are typed with nothing except the
correct address, value and loader code.
0770
0771

F. 1

7006
7500

0772

5376

0773
0774
0775
0776
0777

5377
0200 01
0020
7410
7410

N3,

RTL
SMA

THE SYMBOL TABLE
Symbols are listed in alphabetic order at the end of the assembly pass (Pass 1) with their

relative addresses beside them.
The following flags are added to special types of symbols.
ABS
COM
OP

The address is absolute.
The address is in COMMON.
The symbol is an operator.

F-3

EXT
UNDF

The symbol is an external and thus, may or may not be defined. If not
defined there is no difficulty; it is in another program.
The symbol is not an external symbol and has not been defined in the program.
This is a programmer error. No earlier diagnostic can be given because it is
not known unti I the end of Pass 1 that the symbol is undefined.
A location is reserved for the instruction containing the undefined symbol, but
nothing is placed in it.

F-4

APPENDIX G
OPERATING PROCEDURES

This appendix is a condensation of Chapter 4. The figures referenced (in parentheses) are
found in the PDP-8/I System User's Guide, DEC=08=NGCC-D.

G. l

LOADING THE SABR ASSEMBLER
Procedure
Load the SABR Assembler using BIN (See Figure B-2); IF= 1
SR== 7777. When loaded, parts of the Assembler wi 11 be in
field 0 and field 1.
To load the Assembler on the disk, proceed in step sequence,
otherwise, begin at step 4, below.

With the Disk Monitor in memory, cal I the Disk System
Loader by typing:
.LOAD
and load the SABR assembler onto the disk (see PDP-8/I
Disk/DECtape Monitor System, DEC-D8-SDAB-D).

Save the Assembler by typing:
.SAVE SABR!0-7177; 200
.SAVE SABl !~, 1700 - 12427;)

1t)e
G.2

ASSEMBLING (Pass 1)
See section 4.4 for alternate methods of assembling.
Procedure

Step

Insert source program tape into the tape reader.

Set DF==O, IF==O, SR==0200, press LOAD ADD, START, and
answer SABR's initial dialogue.

Turn the appropriate punch and reader ON; the tape reads
in and the binary tape is punched.

If the program is in sections, when a PAUSE is encountered,
insert the next section of tape into the tape reader and
press CONT; assembly is completed when SABR halts after
producing the relocatable binary tape.
SABR may be restarted to assemble another program by
starting over at step 4 above •
SABR may be restarted at any time by pressing STOP and
starting over at step 4.

G-1

~ ;.J.z.if
171.~31)/~'

Procedure
To generate an assembly listing, proceed in step sequence,
otherwise, begin at step 9.

G .3

Insert the source program tape(s) into the reader and press
CONT.

LOADING THE LINKING LOADER
Procedure
Set DF=highest field in the configuration, IF= 1, SR=7777,
and press LOAD ADD.

G .4

Insert Linking Loader tape into the appropriate reader: if
ASR reader, turn reader ON; if high-speed reader, set
SR=3777.

Press START; the Linking Loader will be read into core
memory.

LOADING PROGRAMS AND SUBPROGRAMS
Step

Procedure

Set DF and IF=to DF in step 9 above, SR=0200, and
press LOAD ADD.

Insert relocatable binary tape (first, program or subprogram with largest amount of COMMON storage) into
the reader with Ieader code over reader head •

Set SR as explained in Section 5 .5.

Press START; the relocatable binary program wil I be

Ioaded into core memory •
Repeat from step 13 for subsequent program or subprogram tapes ~select an option (core
availability or storage map) as explained in Section 5.3.

G .5

EXECUTING THE SABR PROGRAM
Step

Procedure

Set DF and IF=to field of MAIN program, and SR=to
starting address of MAIN program (determined from
storage map) •

Turn punch ON and/or insert data tape in reader, as
required.

Press LOAD ADD and ST ART.

Program execution will begin.

G-2

APPENDIX H
DISK LINKING LOADER

H. l

INTRODUCTION
The Disk Linking Loader (LLDR) is used to load and execute 8K FORTRAN compiled and 8K

SABR assembled user programs when the system configuration includes one or more DECdisks and the

Disk Monitoi System. Such user progrnms exist as a main program with several subprograms (including
necessary 8K library subprograms), all of which must be on punched paper tape in relocatable binary
format. LLDR loads these multiple-part programs in a page-wise relocatable manner, and links all
calls to and returns from external subprograms.
The user communicates with LLDR via the keyboard in a simple, straightforward manner;
LLDR types *OPT - and the user responds with a one-letter code which causes LLDR to perform one of
seven possible functions (operations).
LLDR, unlike the standard BK Linking Loader (Chapter 5), is entirely keyboard oriented and
makes extensive use of the disk. For example:
a. It allows user programs to be loaded over LLDR itself by utilizing temporary disk storage
in the Disk Monitor System environment.
b. it provides two ievels of program overlaying so that much larger programs can be run.
Up to eight files (programs and subprograms) can be loaded into each overlay area. Overlay files are
saved on the disk and called into core as needed at program execution time.
c. It provides several utility and convenience features such as storage map listing, a
listing of necessary subprograms not present in core, a listing of available (unoccupied) core, and
automatic program starting .
d. It includes load-time monitoring via the keyboard rather than the console switches,
and several other minor features.
LLDR accepts paper tape input only, from either the low- or high-speed readers, as do both
the 8K FORTRAN and BK SABR systems. However, the user program (during execution) can use both
DECdisk and DECtape for input/output.
The operating system (Run-Time Linkage Routines) necessary for execution of BK FORTRAN
and BK SABR programs is contained within the LLDR program, and its use is entirely automatic.
Two loading techniques are provided: normal loading and overlay loading. In normal loading, each file is loaded into a separate core area where it remains during execution. In overlay loading, several files are sequentially loaded into the same core area and saved on the disk. At execution
time, each file is brought from the disk into core when it is needed. LLDR provides two levels of
overlay, and each al lows up to eight files per overlay level. A normally loaded program may call a
program in either overlay level, and a program in either overlay level may cal I a program in the other
level.

H-1

The fol lowing main stipulations should be remembered when using LLDR.
a. A program in an overlay level may not call another external program in the same overlay level, except as explained in Section H .4.
b. Common storage (i.e., data storage accessible by al I programs and subprograms) is
always located in field l.
c. The program or subprogram which requests the largest amount of common storage must
be loaded first.
d.

No one program or subprogram may be greater than 4K in length.

e. Programs may not be loaded across field boundaries, although they may be loaded into
any available field.

Overlay files may not be loaded over LLDR, although normal files may be.

LLDR requires a PDP-8/I, -8/L, -8, or -8/S computer with at least 8K words of core, an
ASR-33 Teleprinter, and at least one DECdisk. A high-speed paper tape reader is optional but highly
recommended. LLDR can use al I avai Iable core memory and disk storage.

H .2

LOADING, SAVING, AND STARTING LLDR
LLDR is furnished on punched paper tape in binary-coded format, and is loaded into field 0

by the standard Binary Loader (refer to PDP-8/I System User's Guide, DEC-08-NGCC-D).
Before using LLDR or saving it as a systems program on the disk, it should be properly
initialized for the amount of core available and for the type of paper tape reader to be used. LLDR is
initially set for a basic configuration of SK words of core and a high-speed paper tape reader. With
any other configuration, LLDR should be started and initialized as explained in Section H .2 .2. Complete loading, saving, and cal ling procedures are given below for both basic and expanded configurations.
The following procedures assume that the user is familiar with the Disk N\onitor System, and that the
system is avai Iable for use.

H .2 .1

Basic Configuration
The user with SK of core and a high-speed reader should use the following procedures.

Determine that the Disk l\.A.onitor is in memory. (Type CTRL/C* or START at 07600 .)

When N\onitor responds with a dot, call the system loader by typing
.LOAD)

( ) represents typing the RETURN key)

*CTRL/C is typed by holding down the CTRL key while typing the C key.

H-2

Insert the LLDR binary tape in the high-speed reader.

Answer the loading command dialogue as follows:

*IN-R: )

*
*ST= )
t <CTRL/P> t <CTRL/P>

Keys shown within angle brackets
are not echoed on the teleprinter
when typed by the user.

After each up-arrow which is typed by the Monitor, the user types CTRL/P by holding down the CTRL
key while typing the P key; this is equivalent to pressing the CONT switch when loading manually.
e.

LLDR is now loaded into core; save it on the disk by typing
..:.SAVE LLDR!0-6777;200)

LLDR may now be called to load relocatable binary programs by typing
.LLDR ~

H.2.2

Expanded Configuration
The user, with any configuration other than the basic configuration mentioned above, should

use the fol lowing procedure:
a.

Determine that the Disk Monitor is in memory. (Type CTRL/C or START at 07600 .)

When Monitor responds with a dot, call the system loader by typing
.:.LOAD)

Insert the LLDR binary tape in the appropriate reader.

Answer the loading command dialogue as fol lows:
*IN-R: ~

-*-

(R:
T:

for high-speed reader
for ASR reader)

*ST=7400)
_! <CTRL/P > t <CTRL/P >
e. LLDR is now loaded into core. It automatically starts at location 7400, causing it to
type out its initialization questions. Answer the questions as shown below.
*GIVE SIZE OF MEMORY IN K-12)

(user typed 12)

*HIGH SPEED READER? Y

(user typed Y)

When answering the first question, the user should type the amount of available core memory after K-;
the user should type Y for yes, or N for no in answer to the second.
f.

When the above questions have been properly answered, LLDR may be saved on the disk

by typing
.SAVE LLDR!0-6777;200 ~
g.

LLDR may now be called to load relocatable binary programs by typing
.LLDR)

H-3

whereas the LLDR system program wil I be transferred from disk storage into core memory, and automatically started (executed) so that it types out its version number and *OPT-. It then waits for the user to
specify which of the seven optional functions is to be performed. The version number and option request might appear as shown below.
PDP-8 DISK w LINKING w LOADER w DEC-08-A2C7-03
*OPTLLD R is now in core, started, and ready for use .
H .3

LLDR Functions
When LLDR has been initialized and started as described in the preceding section, it types

its program version number (also found on the paper tape identification label) and option statement and
then waits for the user to specify the desired function to be performed. For example:
PDP-8 DEC-08-A2B4-02
*OPTThe user's response to *OPT- is in the form of a one-letter code fol lowed by the RETURN key. LLDR's
functions and corresponding one-letter function codes are listed below.
Code

Function

Core avai labi Iity listing

Disk file assignment

Exit with halt

Normal loading

Storage map Iisting

Overlay loading

s
u

Start main program
Un loaded program Ii sting

Functions may be cal led whenever needed or desired, except that the M, U, and S functions
must not be cal led first.
Upon completion of a function (except E or S), LLDR will request another by repeating the
option statement (*OPT-).
Any error made by the user when responding to an option statement will cause LLDR to type
a question mark, ignore the response, and repeat the option statement.
LLDR may be stopped (e.g., to make a program patch) and restarted without altering the
state of the computer by using the console STOP switch and restarting at 00600. This method may be
used at any time after completion of any function other than D, except during overlay loading or
wh i Ie a tape is actual Iy being read .

H-4

At any time during the use of LLDR (except while a tape is being read in), control may be
returned to the Disk Monitor. This is done by typing CTRL/C; however, when CTRL/C is typed, all
data temporarily stored on the disk is lost.

H .3 .1

Disk File Assignment Function (D)

If the user's programs or subprograms create or use disk data files with the RDISK and WDISK
library functions, the D function must be the first function used. The D function performs the preliminary job of entering the names of user files into the disk directory. This prepares the way for using the

RDISK and WDISK iibrary functions, which aiiow the user to read and write data on the disk at execution ti me.
Use of the D-function proceeds as shown below:
PDP-8 DEC-08-A2B4-0l
*OPT-D )
*FILES-ABC, WXYZ, Ml I /\Kl., 5H, R, 3, p)
*OPTwhere a directory entry is assigned to each of the eight fi!e names. File names may be from one to four
characters in length, and up to ten files may be specified. All such files must be named in one execution of the D function.
The order in which the data files are named for the D function is especially important. The
reason for this is that when the user's program references disk data files using the RDISK and WDISK
library functions, he must reference these files not by name but by logical number (1,2, ... , 10).
This logical number is determined by the order in which he names the files for the D function. For
example, if files have been named in the D function as shown in the previous example, the user's program wi 11 reference fi Ie M1 by statements of the form
CALL RDISK (3 I

•••)

because M 1 was the third fi Ie named.
Before using the D function the user should study thoroughly the operation and use of the
RDISK and WDISK I ibrary functions in Section E-9.
The disk directory will accommodate ten file names. If the directory is too full to accommodate al I files named, a meaningful error message is printed by LLD R. In the example above, if the
directory had room for only four files, the error message
DISK WILL NOT HOLD 5H & FOLLOWING FILES
would have been printed. If this happens, the entire D function request is ignored and LLDR prints
another *OPT- to allow the user to repeat the D function with fewer files or to specify a different
function.

H-5

After the D function has been performed, LLDR will again print *OPT- for the user to continue with the process of loading his program. After the D function has been used or when a different
function has been called, the D function is no longer available--if called a second time or after a
different function, it is treated as an illegal function code.
Again, if the D function is to be used, it must be the first function used. If it is not chosen
as the first function, it is not available for use until a fresh image of LLDR is brought into core from the
disk.

H.3.2

Loading Functions (L and O)
The two loading functions, L for normal loading and 0 for overlay loading, are avai Iable

for use at any time. These are the principal functions of LLDR--to load relocatable programs for execution. These functions use the standard technique of link-loading as described in Chapter 5, which
applies specifically to the relocatable binary code (Chapter 3) produced by the 8K FORTRAN/SABR
system.
Programs and subprograms may be loaded in any order and into any field. The only restrictions are listed below.
a.

The subprogram which requires the largest amount of common storage must be loaded

first .
b. No subprogram may be loaded across a core field boundary; i.e., no subprogram may be
longer than 4K in length.
c.

A maximum of 64 subprograms may be loaded, including multiple entry points for single

programs.
LLDR loads subprograms in the order presented and into the field specified (see below) from
the lowest available memory upward. Common storage is allocated in the lower portion of field 1
before loading actually starts. A maximum of 3840 words of common storage fills field 1.
LLDR loads in a page-wise relocatable fashion (each program begins at the start of a new
core page), establishing external links so that each subprogram is properly executed.

H .3 .2. 1

Normal Loading (L)
In normal loading, the user's program is loaded directly into core memory where it remains

available for, throughout, and after execution. The core area occupied by each normally loaded
program is the property of that program, and no other program can be loaded into its core area.

To perform normal loading, the user responds to *OPT..___,,.,.. with the letter L. When this is
done, LLDR types a request for the number of the field in which the user wishes to load. This specified
field must exist in the configuration. For example:

H-6

*OPT-L)
*FIELD-2)
Had field 2 been nonexistent, the fol !owing \Vou!d have occurred:
*OPT-L)
*FIELD-2)

?
*OPTwhere LLDR ignored the user's response, typed the question mark, and repeated the option statement.
When LLDR is satisfied with user response, it then types an up-arrow. At this point LLDR

wi 11 pause and wait for the user to pi ace his reiocatabie binary tape in the tape reader, and to type
CTRL/P which causes LLDR to load the program into core. When the program has been loaded, LLDR
will type another up-arrow and pause for user response. If the user wishes to load another program into
the same field, he need only place the tape in the reader and then type another CTRL/P (or press the
CONTinue switch and then type CTRL/P is using the low-speed tape reader). When the user no longer
wishes to load into the same field, he should respond to the up-arrrow by typing the RETURN key, and
LLDR will type another option statement.
The user may respond to an up-arrow with CTRL/N, which causes LLDR to by-pass the next
program on a multi-program tape. This situation may, for example, occur with a library subprogram
tape.
A typical example of normal loading is shown below, where three programs are loaded into
field 0 and two into field 1, with one program being by-passed.
*OPT-L)
*FIELD-0)

*..!.<CTRL/P> _!_ <CTRL/P> l_ <CTRL/P> _!_)
*OPT-L)
*FIELD-1 )
* .!..<CTRL/P> t <CTRL/N > _!_ <CTRL/P> _! )
*OPT-

If the low-speed reader had been used in the example above, the CONTinue switch would have been
pressed just before each CTRL/key combination.

H .3 .2 .2

Overlay Loading (O)
Overlay loading allows the user to load as many as 16 subprograms into the same core area.

The user may load one or two overlay levels (each 0 function cal I constitutes an overlay level) of subprograms (files) with up to eight files per level. Overlay loading is possible only when no two subprograms of the same level need to be in core at the same time, i.e., they do do not ca 11 each other.

H-7

Al I subprograms loaded during the operation of an 0 function are loaded into the same core
area (overlay level) and automatically saved in separate files on the disk. At execution time each file
is called back into core as needed. No protection is given to the file of this overlay level that was
previously in core. It is completely overwritten in core. Overlay files should use common storage for
data which must remain in core.
Files in a given level may be loaded in any order, provided they are all loaded during the
same execution of 0 function. Files in a given level need not be the same length; enough core is
allocated for the largest file in the level.
Loading with the 0 function is quite similar to loading a string of programs in the same field
using the L function. An example is given below, where three files are loaded into the first level and
two files into the second level, with one file being passed over.
*OPT-0)
*FIELD-1 ~
:..! <CTRL/P >~ <CTRL/P > .! <CTRL/P >_!_ J
*OPT-0)
*FIELD-1)
* t <CTRL/P >_!_ <CTRL/N >_! <CTRL/P >_! )
*OPTLoading of a single overlay level is terminated with the RETURN key. Loading of an overlay level wi 11
automatically be terminated after eight files have been loaded.
As with the L function, if the low-speed reader had been used in the example above, the
CONTinue switch would have been pressed just before each CTRL/key combination.
When the main program is removed from core, linkage to its overlay files is broken. Therefore, for subsequent execution, files must be reloaded with the main program.

H .3 .2 .3

Error Messages
When LLDR detects an error during loading of a program, it types an error message of the

fol lowing form:

ERROR OOOn
where n is a number from 1 to 6, representing the type of error detected. If the error is fatal, control
returns to the Disk Monitor. If it is not fatal, the user may be able to continue loading (see below).
Error

Fatal?

Attempt to load more than 64 subprograms

Yes

Field overflow

Subprogram with largest common assignment
not loaded first

Yes

Error No.

H-8

Error

Error No.

Fata!?

Checksum error

Improper or damaged tape or reader error

Disk overflow

A discussion of each non-fatal error is given below.
Error 2 - During normal loading, loading may be continued in a different memory field.
During overlay loading, the entire overlay level must be reloaded into a different memory field.
Errors 4 and 5 - During either type of loading, the user may reposition the faulty tape in the
wl/D1 ;••. . I. I.. espon .."' e
I.. ea do
... r ,..,,....,..j
..... 1.... f.I y1no
I'"'"' r-i
..... ;u..

r·"v t-h
11 e ....
• 1ewI ut-' a •• ow . If the error persists , reassemb i y or hardware
I

..... -

....

maintenance wi 11 be necessary.
Error 6 - Occurs during normal loading only when the user is loading into the upper portion
of field O; the program which caused the error must be loaded into a different field. During overlay
loading, the current overlay level will be closed with only the files that were loaded successfully. The
file which caused the overflow (the last file read) and succeeding files will have to be loaded normally.

H .3 .3

Uti Iity Functions (C, M, and U)

H .3 .3 .1

Core Availability (C)
The user may at any time request a list of the number of pages available for loading in each

core field. The following example assumes that the user has a 16K computer (4 fields):
*OPT-C ~

0033
0036
0036
0036
*OPT-The numbers listed are the octal number of free pages left in fields 0, 1, 2, and 3, respectively.

H . 3 .3 . 2

Storage Map (M)
During the link-loading process, LLDR builds a list of external symbols; i.e., main program

and subprogram entry points and their actual starting addresses. This list forms a complete storage map
of all programs loaded, as shown below:

H-9

*OPT-M)
MAIN
READ
WRITE
IOH
SET ERR
TTYIN

10200
01055
01066
03031
00000
00000

FLOAT
FIX

05046
04513

*OPTStarting addresses are expressed in five octal digits - the first digit represents the memory field and the
other four the address in that field. The U means that the stated subprogram has been called but has
not been loaded, and therefore must be loaded before successful execution is possible.
Listing of the storage map may be prematurely terminated by typing CTRL/P.

H .3 .3 .3

Unloaded Program Listing (U)
This function is used to obtain a list of those subprograms which must still be loaded before

successful execution is possible. All symbols flagged with a U in a storage map listing will be listed
as shown below:
*OPT-U
SET ERR
TTYIN
TTY OUT
HSIN
HSOUT
*OPTThis listing may also be prematurely terminated by typing CTRL/P.

H.3.4

Exit Functions (E and S)
The E function is used to cause a halt after al I loading is complete. The S function is used

to automatically start execution of the loaded program at the beginning of the main program.
Both of these functions signal LLDR that loading is complete. They each cause any data
which has been temporarily saved on the disk (except overlay files) to be read into core.
When the E function is used, LLDR reads in al I data temporarily stored on the disk and then
halts. The user's entire program (except overlay files) will be in core, ready for patching, execution,
or saving on the disk.

H-10

When the S function is used, LLDR checks for a subprogram called MAIN (such as a
FORTRAN main program). If found, execution will automatically start at the starting address of MAIN.

If MAIN is not found / the S function is executed as an E function.
H .4

TECHNIQUES, OVERLAY LOAD ING
In general, any group of subprograms which do not call each other (either directly or in-

directly) may be loaded into the same overlay level. A typical situation fol lows:
MAIN

A
B

c
D
E,F,G,H

contains calls to
contains cal Is to
contains cal Is to
contains calls to
contains calls to
contain no external calls

A, BI c I DI E
""'

Ll 1C1

D,G

DI E, H
E

The above combination may be loaded as fol lows:
Normal

Overlay 0

MAIN

Overlay 1

G
H

If D contains a call to any other than E, it would be better to load D normally and put E in overlay 1.
If F were to call B, the above loading situation would not work; A would be calling B indirectly, and
these two are in the same overlay level.

It is possible, however, to call another program in the same overlay level only if the called
program never attempts to return to the calling program. In this way, simple chaining may be achieved.
For example, a very long FORTRAN main program can be split into sections with each section terminated by a cal I to the next. Such a situation is shown below.
MAIN

calls A, B, C and is terminated by a call to MAIN2

MAIN2

calls A, B, C and is terminated by a call to MAIN3

MAIN3

calls A, B, C and is terminated by a call to MAIN4

MAINS

cal Is A, B, C and stops

A, B,C

contain no external calls

The above combination may be loaded as fol lows:

H-11

Overlay 0

Overlay l

MAIN
MAIN2
MAIN3

A
B

MAINS
When the MAIN program is contained in an overlay area, the E function cannot be used unless MAIN is loaded last into the overlay level. The S function wil I work with the above combination
since it works regardless of the order in which the segments of MAIN are loaded.
With FORTRAN programming alone, a subprogram other than a MAIN program may not be
chained. However, this is possible with careful assembly language programming. An example of such
programming is shown below, where SUB is split into a two-part chain, SUB and SUB2. MAIN is a
standard FORTRAN program containing cal Is to SUB in the form:
CALL SUB (A 1, A2, A3)
SUB is written as a standard FORTRAN program which does part of the work for the entire subroutine
chain, including processing arguments Al and A2. It is written with two arguments and concludes with
CALL SUB2(Z
where Z is any dummy argument. After SUB has been compiled and before the intermediate compiler
symbolic is assembled, it should be edited to include the insertions enclosed in brackets.

[X,
SUB,

COMMN2]
ENTRY SUB
BLOCK 2

TAD
DCA
TAD
TAD
DCA
CALL
ARG
END

SUB

/SAVE RETURN FIELD

x
(-2
SUB#

/-2* NO. OF ARGS TO BE PASSED
/SAVE ARGUMENT ADDRESS

l ,SUB2

SUB2 is also a standard FORTRAN program containing the latter portion of the entire subroutine, including the processing of argument A3. The actual contents of SUB2 is coded in FORTRAN just as if
it were a subroutine taking one argument. After SUB2 has been compiled, the compiler symbolic output is edited as shown below:

H-12

COMMN 2]
ENTRY SUB2
SUB2,
BLOCK 2
rI TAD
x
SUB2
I DCA
TAD
SUB2#
DCA

/REPLACE ARG POINTER

/CONTINUE WITH NORMAL FORTRAN
/CODE, CONCLUDING WITH
RETRN

SUB2

END

H .5

USER PROGRAM EXECUTION

If the user chooses not to execute his program automatically with the S function, he may
determine the exact address for the start (using the storage map or assembly listing), and execute his
program, using the console switches or the Disk Monitor.
At execution ti me, the Run-Ti me Linkage Routines (see Section 3 .3 .4) must be in core.
These routines accomplish the necessary linkage for all calls to and returns from external subprograms,
aii off-page indirect references, and all off-field references (including those to common and passing
subroutine arguments).

If 1 during execution of a user program, a call is made to a nonexistent program or subprogram,
an unconditional halt wil I occur and control will return to the Disk Monitor. This error is fatal. All
other execution-time errors are covered in Section C-3.
Program execution may be terminated at any time by typing CTRL/C. However, when
CTRL/C is typed, all overlay files stored on the disk are lost.

H .6

STORAGE ALLOCATION
The following core availability map allows the user to plan his loading.
Field 0

0000-0777

Used by the Run-Time Linkage Routines and not available
to the user for loading.

1000-4377

Avai Iable for any loading.

4000-7577

Residence of LLDR during loading. Available for normal
loading (by automatic use of temporary disk storage), but
not available for overlay loading.

7600-7777

Disk Monitor permanent residence.

H-13

INDEX

ABSY/v·\ Pseudo-Op

1-6; 2-4; B-2

Addresses, Program

3-8

Constants 1=8

ASCII 1-9

Alpha Constants 1-9

Illegal Alpha 1-9

Alphabetic Characters 1-2

Numeric 1-9

Alphanumerics 1-2, -6

Conversion, Numeric 1-10

ARG Pseudo-Op 2-11; B-2

Core Availability Option 5-2

ASCII Constants 1-9

CPAGE Pseudo-Op 2-2; B-2

Character Set App. A

Data Generating 2-5, -6

Assembled Binary Code Ch. 3

DECIM Pseudo-Op 2-1; B-2

Assembly

Demonstration Programs 6-1

Control 2-4

DUMMY Pseudo-Op 2-9; B-2

FORTRAN Pass 2 4-3, -4, -5
Listing 6-3, -4; F-1 ,

-4

Page 3-3

Variables 2-7
DUMSUB Linkage Routine 3-6
EAP Pseudo-Op 2-2; B-2

Pass l 4-2; G-1

Page Escapes 3-4

Automatic Paging Mode

Elements / Statements 1-3

Enter 2-2

END Pseudo-Op 2-1; B-2

Leave 2-2

ENTRY Pseudo-Op 2-9; B-2

Binary Output, Relocatable 3-1

Error Messages App • C

Blank Lines 1-5

Library Programs C-3

BLOCK Pseudo-Op 2-5; B-2

Linking Loader C-2

CALL Pseudo-Op 2-8; B-2

SABR C-1

CDF Current 3-2

Escapes, Page 3-4

CDFSKP Linkage Routine 3-5

Equivalent Symbols 1-6

CDZSKP Linkage Routine 3-6

Executing a SABR Program, Operating

Character Set 1-2; App. A

Procedures

II legal 1-2

G-2

External, Subroutines 2-7, -8, -9

Codes, Loader Relocation 3-1, -2, -3

Flags, Symbol Table 1-8

Comments 1-4

Floating-Point Accumulator B-3; App. E

COMMN Pseudo-Op 1-6; 2-4

FI oati ng- Point Arithmetic, Library Subprograms

COMMON Storage 2-4, -7; 5-1

E-2

High 3-3

Format Effectors 1-5

INDEX (Cont)

Format, Page 3-4

LAP Pseudo-Op 2- 2; B-2

FORTRAN Pass 2

Page Escapes 3-4

Assembly Methods 4-4, -5

Legal Characters 1-2

Operating Procedures 4-3, -4, -5

Library Subprograms App. E.

FORTR Pseudo-Op 4-5; B-2

Error Messages C-3

Free Core, Page 0, D-1

Floating-Point Arithmetic E-2

FUNCTIONS, Library Subprograms E-4

Functions E-4

Hardware, Required

Input/Output E-1

SABR 1-1

Integer Arithmetic E-3

Linking Loader 5-1

Organization E-5

IF Pseudo-Op 2-3; B-2

Powers E-5

Illegal

Subscripting E-4

Alpha Constants 1-9

LINK Linkage Routine 3-7

Characters 1-2

Linking Loader Ch. 5

Incrementing Operands 1-7

Information Options 5-2

Input/Output, Library Subprograms· E-1

Error Messages C-2

Instructions

Demonstration Of

6-4, -5

IOT 1-3

System Requirements 5-1

Memory Reference 1-3, B-1

Loading 5-1, -3

Skip 3-7

Memory Map Option 5-2

Micro (Group 1 and 2) 1-3; B-1, -2

Operation Of 5-3

Multiple Word 3-5
Sequence of 1-5
Integer Arithmetic, Library Subprograms E-3
Introduction

Listing
Symbol Table 1-8; F-1, -3, -4
Pass 2, Assembly 6-3; F-1, -2, -3
Literals 1-9, -10

SABR 1-1

Loader Relocation Codes 3-1

Linking Loader 5-1

Loading Procedures

IOH Routines D-1; E-1

Information Options 5-2

IOT Instructions 1-3; B-2

Linking Loader 5-1; G-2

Labels

Programs and Subprograms G-2

Example Of 1-3
Language, SABR Ch. 1

SABR: Basic 4-1; G-1
Disk Monitor 4-1

INDEX (Cont)

Pass 2, Listing 4-2

Memory Map Option, Linking Loader 5-2

FORTRAN 4-3 -4, -5

Memory Reference Instructions 1-3; B-1
Methods, FORTRAN Pass 2 Assembly 4-4, -5

PAUSE, Pseudo-Op 2-1; B-3

Microinstructions (Group 1 and 2) 1-3; B-1

Permanent Symbols 1-6; App. B

-2

Mode, Numeric Conversion 1-10

POWERS, Library Subprograms E-5

Multiple Word Instructions 3-5

Program Addresses 3-8

Nul I Lines 1--5

Programs, Loading Procedures G=2

Numeric Constants 1-9

Pseudo-Operators

Numeric Conversion, Mode 1-10

Arguments To 1-4

OBISUB Linkage Routine 3-6

Description Of Ch. 2

OCTAL Pseudo-Op 2-1; B-2

List Of B-2, -3

OPDEF Pseudo-Op 1-6; 2-4; B-2

REORG Pseudo-Op 2-2; B-3

Operands 1-4

Relocatable Binary 3-1

Incrementing 1-7
Operating Procedures App. G

Assemb Ii ng G~ 1

Requirements
Hardware and Software 1- l
Software 1= 1

Executing the SABR Program 4-2; G-2

RETRN Pseudo-Op 2-1 O; B-3

FORTRAN Pass 2, 4-3 -4, -5

RTN Linkage Routine 3-7

Loading Programs and Subprograms G-2

Run-Time Linkage Routines 3-5

Loading SABR G-2

SABR

In Core 4-1

Error Messages C- 1

In Disk Monitor System 4-1

Language 1- 1

Operation, Linking Loader 5-3

Operating Prodecures Ch. 4

Operators 1-3

System Requirements 1-1; App. G

Options, Switch Register 5-4

Sample of Assembly Listing App. F

Core Availability 5-2

Sequence of Instructions 1-5

Storage Map 5-2

SKIP Instructions 3-7

Tape Reader 4-1; 5-3

SKPDF Pseudo-Op 1-6; 2-4; B-3

Page Assembly 3-3

Software Required

Escapes 3-4

Linking Loader 5-1; App. G

Format 3-4

SABR 1-1; App. G

PAGE Pseudo-Op 2-2; B-2
Pass 1, Assembly 4-2

Special Characters 1-2

INDEX (Cont)

Statements 1-2, -5

Text Pseudo- Op 2-6; B-3

Comments 1-4

User-Defined Symbols 1-6

Elements 1-3

Variables, DUMMY 2-7, -9

Format Of 1-3
Format Effectors 1-5
Labels 1-2
Operands 1-4
Operators 1-3
Storage, COMMON 2-7; 3-3; 5-1
Storage Map, Option 5-2
Subprogram Arguments, Picking Up 2-10
Subroutines, External 2-8, -10
Subscripting, Library Subprograms E-4
Switch Register Options 5-4
Core Availability 5-2
Storage Map 5-2
Tape Reader 4- 1; 5-3, -4
Symbolic Tape Editor 1-2; 6-1
Symbol Definition 2-4
Symbols 1-6, -7, -8
Equivalent 1-6
External Definition 3-2
Incrementing Operands 1-7
Permanent 1-6; B- 1 , -2, -3
Symbol Table Listing F-1, -3, -4
Types of 1-6
User-Defined 1-6
Symbol Table 3-8
Listing F-1, -3, -4
Special Flags 1-8
System Configuration
SABR 1-1
Linking Loader 5-1

HOW TO OBTAIN SOFTWARE INFORMATION

Announcements for new and revised software, as well as programming notes, software probiems, and documentation corrections are published by Software Information Service in the following newsletters.

Digital Software News for the PDP-8 Family
Digital Software News for the PDP-9/15 Family
PDP-6/PDP- l 0 Software Bulletin

These newsletters contain information applicable to software available from Digital's Program Library.

Please complete the card below to place your name on the newsletter mailing list.

Questions or problems concerning DEC Software should be reported to the Software Specialist at your nearest DEC
regional or district sales office. In cases where no Software Specialist is available, please send a Software Trouble
Report form with details of the problem to:
Software Information Service
Digital Equipment Corporation
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 01754
These forms, which are available without charge from the Program Library, should be fully filled out and accompanied by teletype output as well as listings or tapes of the user program to facilitate a complete investigation. An
answer will be sent to the individual and appropriate topics of general interest will be printed in the newsieiter.

New and revised software and manuals, Software Trouble Report forms, and cumulative Software Manual Updates
are available from the Program Library. When ordering, inciude the document number and a brief description of
the program or manual requested. Revisions of programs and documents will be announced in the newsletters and
a price list will be included twice yearly. Direct all inquiries and requests to:
Program Library
Digital Equipment Corporation
146 Main Street, Bldg. 3-5
Maynard, Massachusetts 017 54
Digital Equipment Computer Users Society (DECUS) maintains a user Library and publishes a catalog of programs
as well as the DECUSCOPE magazine for its members and non-members who request it. For further information
please write to:
DECUS
Digital Equipment Corporation
146 Main Street
Maynard, Massachusetts 01754

Send Digital's software newsletters to:
Name _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
Company Name _ _ _ _ _ _ _ _ _ _ _ _ _ __
Address _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ___

(zip code)
PDP-8/1 0
PDP-8/L 0
LINC-8 0
PDP-12 0
PDP-9 0
PDP-15 0
PDP-IO 0
OTHER 0
My system serial number is _ _ _ _ _ _ _ _ _ _ _ (if known)
My computer is a

Please specify

-------

·•·•········•············•······························································ Fold Here ....................................................................................... .

..•..............•...•..................................................... Do Not T car - Fold Here and Staple ················································ ............. .

FiRST CLASS
PERMIT NO. 33
MAYNARD, MASS.

BUSINESS REPLY MAIL
NO POSTAGE STAMP NECESSARY IF MAILED IN THE UNITED STATES

Postage will be paid by:

mamanmn
Digital Equipment Corporation
Software Quality Control
Building 12
146 Main Street
Maynard, Mass. 01754

8K SABR

READER'S COMMENTS

DEC-08-ARXB-D

Digital Equipment Corporation maintains a continuous effort to improve the quality and usefulness of its
publications. To do this effectiveiy :ve need user feedback - your critical evaluation of this manual.

Please comment on this manual's completeness, accuracy, organization, usability, and readability.

DEC also strives to keep its customers informed of current DEC software and publications. Thus, the following periodically distributed publications are available upon request. Please check the appropriate boxes for a current issue of the
publication(s) desired.

Software Manual Update, a qu·arterly collection of revisions to current software manuals.

User's Bookshelf, a bibliography of current software manuals.

Program Library Price List, a list of currently available software programs and manuals.

Name - - - - - - - - - - - - - - - Organization
S t r e e t - - - - - - - - - - - - - - Department
City _ _ _ _ _ _ _ _ _ _ _ S t a t e - - - - - - - - - - - - - - Zip or Country _ _ _ __

........................................................................................ Fold Here ........................................................................................ .

................................. ... ....................................... Do Not ·r car · Fold Here and Stapl·e ............................................................. ..

FIRST CLASS
PERMIT NO. 33
MAYNARD, MASS.
BUSINESS REPLY MAIL
NO POSTAGE STAMP NECESSARY IF MAILED IN THE UNITED STATES

Postage will be paid by:

mamanma
Digital Equipment Corporation
Software Quality Control
Building 12
146 Main Street
Maynard, Mass. 01754

Digital Equipment Corporation
Maynard, Massachusetts

printed in U.S.A.

~nmnomo