Digital PDFs

XX-0B44E-6E-1

May 2000

214 pages

Original

10MB

Document:	PDP-15/10 Software System Introduction Part I
Order Number:	XX-0B44E-6E
Revision:	1
Pages:	214
Original Filename:	http://bitsavers.org/pdf/dec/pdp15/PDP-15_10_Software_System.pdf

OCR Text

PART I
PDP-15/10 SOFfWARE SYSTEM
INTRODUCTION

1.1 GENERAL
Software for the PDP- IS/ 10 consists of the COMPACT and the BASIC 1/0 Monitor Software Systems, which are
designed to operate in a paper tape (or card) environment.

I. I. I COMPACT System
COMP ACT is a complete programming system for the PDP-IS/ I 0 and includes a symbolic assembler, text editor,
debugging routine, various utility routines, and a library of mathematical routines. With COMPACT software, the user
can prepare symbolic programs on-line using the Text Editor, assemble them using the CAP-IS Assembler, and
execute them under the control of the debugging routine (ODT). The utility and mathematical routines in the system
can be incorporated into user programs as required. With the addition of a DECtape transport and control unit, users
can use the FAST-IS System to store and retrieve frequently used system and user programs. In addition, users can
convert PDP-8 programs (PAL-D, PAL-III, or MACR0-8) to PDP-IS assembly language with the aid of 8TRAN.

1.1.2 Basic 1/0 Monitor System
The more sophisticated Basic 1/0 Monitor Software System, available to users with expanded PDP-15 / l 0 Computers
(Refer to Table 1-1), is a complete system for the preparation, compilation, assembly, debugging, and operation of
relocatable programs.
Powerful system programs include FORTRAN IV, FOCAL, a sophisticated macro assembler (MACR0-15), an on-line
debugging system (DDT), an on-line editor (EDIT-IS), and a peripheral interchange program (PIP). A versatile and
flexible input-output programming system (IOPS) frees the user from the need to create device handling subroutines
and from the concerns of device timing. The Basic 1/0 Monitor is upward compatible with the Advanced and
Background/Foreground Monitors of the PDP-IS series; thus, all programs prepared for the Basic 1/0 Monitor can be
run using the Advanced or Background/Foreground Monitors. In addition, users having PDP-8 programs (PAL-D,
PAL-III or MACR0-8) may convert them to PDP- IS assembly language with the aid of 8TRAN.

1-l(Part I)

l. l.3 Hardware Requirements

Minimum hardware configurations applicable to the COMPACT and Basic 1/0 Monitors are provided in Table 1-l.
l.2 COMPACT SYSTEM PROGRAMS

The COMPACT System consists of the programs described in the following paragraphs.
1.2.l Assembler
The PDP-15/10 Assembler (CAP-15) is a two-pass assembler that requires less than 3K of core memory. It processes
source programs to produce an executable binary code. The Assembler makes machine language programming easier,
faster, and more efficient. It pennits the programmer to use mnemonic symbols to represent instruction operation
codes, locations, and numeric data. The programmer can direct the Assembler's processing by means of a powerful set
of pseudo operations. An output listing that shows the programmer's source coding, as well as the binary object code
produced by the Assembler, can be obtained.
1.2.2 Text Editor
The Text Editor provides for the creation/modification of source programs and other ASCII text material. Commands
issued from the teletype direct the Editor to bring a group of lines from the input device to an internal buffer. The
user can then, by means of additional commands, examine, delete, and change the contents of the buffer, and insert
new text at any point in the buffer. When a block of lines has been edited, it is punched on the output device.
The Editor is most frequently used to modify PDP-15/10 source programs, but it can also be used to edit any
symbolic text. The Editor operates with either high-speed or low-speed paper tape devices, and occupies
approximately 20001 0 locations of core memory. Any additional memory is used for buffers.

Table 1-1. PDP-15/10 Hardware Configurations

PDP-15/10 Hardware
Configuration
Applicable
Software

COMPACT
COMPACT
FAST 15
BASIC 1/0
MONITOR

Basic 4K System

ASR33

ASR33
DECtape

~
~

Expanded SK System*
ASR 33
High-speed
Reader/punch

ASR 33
High-speed
Reader/punch
DECtape

~
~
~

*Additional options include the CR03B Card Reader, the 6470 or 647F Line Printer, and the TU20, TU20A, TU30, and TU30A
Magnetic Tape Transports.

l-2(Part I)

1.2.3 Debugging Routine
Octal Debugging Technique (ODT) is a debugging aid that allows the user to conduct an interactive, on-line debugging
session using octal numbers and teletype commands. When errors are found, the programmer can correct them on-line
and execute the program immediately to test the correction. Thus, ODT can be used to compose a program on-line
and check it out as composition progresses. Manual operation of console controls is not required to operate ODT; all
functions are initiated by typing commands on the teletype.

1.2.4 Utility Routines
Utility routines in the COMPACT Software System include a FAST-15 system for DECtape handling, a hardware
readin mode (HRM) punch routine, paper-tape handling routines, teletype 1/0 routines, an octal dump routine, and a
memory scan routine. These routines are briefly described in the following paragraphs.

1.2.4.1 FAST-15 System - FAST-15 (Fast Acquisition of System Tape) is a loading system used to retrieve
frequently used programs from DECtape and to create system tapes. The main advantages of the system are speed and
ease of access. Equipment required for use of FAST-15 includes a Type TC02 DECtape Control Unit and a Type
TU55 DECtape Transport.
The FAST-15 system tape, as distributed by Digital Equipment Corporation, contains commonly used system
programs such as the Symbolic Editor, the CAP-15 Assembler, and ODT. Since these programs can be called from
DECtape with only a small bootstrap, paper tape handling is eliminated. FAST-15 is not restricted to system
programs; it also can be employed very conveniently for frequently accessed user programs.

l.2.4.2 HRM Puncher - The Hardware Readin Mode (HRM) Puncher is a self-relocating, paper tape dump program.
It can be loaded by means of the PDP-15/10 Hardware Readin (HRI) facility into any block of core memory. When

loaded, the HRM Puncher relocates itself and punches out an area of memory, specified by the user, in the HRI
format.
1.2.4.3 Paper Tape Handling Routines - The paper tape handling routines include PTLIST, a paper tape list and
PTDUP, a paper tape duplicator. PTLIST can be used to read an ASCII coded paper tape from either the high-speed
or low-speed paper tape reader, and provide a character-by-character listing on the teletype. PTDUP can be used to
duplicate/verify ASCII or binary tapes using the high-speed paper tape reader and punch.

l.2.4.4 Teletype Input/Output Routines - The teletype input/output routines include the Teletype I/O Conversion
(TICTOC) Package and the Decimal and Octal Print Package.
TICTOC is used to read 8-bit ASCII code from the teletype and pack it as a 6-bit trimmed ASCII code, and vice versa.
Routines in the package fall into three main categories: input, output, and formatting.
The decimal and octal print routines are subroutines that can be used to dump the accumulator in either
signed-decimal or octal format.
1.2.4.5 Octal Dump Routine - The octal dump routine allows the user to obtain hard copy or paper tape output
showing the contents of any register or set of registers that he specifies. The user specifies the registers that are to be
dumped via the teletype keyboard.

J-3(Part I)

1.2.4.6 SCAN Routine - SCAN is a small program used to scan areas of memory for a particular bit configuration.
The user specifies the start and stop address for the area to be scanned, the bit configuration to look for, and the bit
positions to be tested. When scanning the area, if a match is found, the address of the match and the matching word
are printed.

1.2.S Mathematical Routines
The COMP ACT Software System mathematical routines are grouped into four major packages: Integer Arithmetic,
Trigonometric Functions, Floating Point, and Floating Point 1/0. The Integer Arithmetic routines allow PDP-15/10
users, without the EAE option, to write programs using simulated multiply and divide instructions. The
Trigonometric Package provides the user with a large repertoire of trigonometric functions in both single- and
double-precision. The Floating Point Package allows the user with the capability of inputting and outputting decimal
data in floating point format. All floating point data transfers are handled through a software accumulator.

1.2.6 STRAN
8TRAN is used to translate programs written in Pal III, Pal-D, or MACR0-8 assembly language to MACR0-15
assembly language for assembly and execution in the PDP-15 Basic Software environment. The translator produces a
straightforward translation that clearly indicates the parts of the translated program to be reviewed in light of the
PDP- l 5's greater word length and more powerful instruction set. It does not simulate the PDP-8 or produce directly
executable code.
A full description is contained in PDP-15 8TRAN Manual (DEC-15-ENZA-D).

l.3 BASIC 1/0 MONITOR
The PDP-15/10 Basic 1/0 Monitor System simplifies the handling of input/output functions and facilitates the
creation, debugging, and use of PDP-15/10 programs. It allows overlapped input/output and computation, as well as
simultaneous operation of a number of asynchronous peripheral devices, while freeing the user from the need to
create device handling subroutines. The Monitor, operating in conjunction with the Input/Output Programming
System (IOPS), provides a complete interface between the user's programs and the peripheral hardware.
The Monitor accepts 1/0 commands from the system or user programs and supervises their execution. By calling upon
the device manipulation routines of IOPS, it provides for simultaneous 1/0 and computation.
The Basic 1/0 Monitor contains:
a. Routines for its own initialization and control
b. Tables to allow communication between the Monitor, system programs, user programs, and the Input/Output
Program System
c. The CAL Handler, which is used to dispatch to the appropriate Monitor and 1/0 subroutines
d. Device handlers for the teletype and clock.
The Monitor resides in lower core and occupies about 880 1 0 locations.

l-4(Part I)

1.3.l Input/Output Programming System (IOPS)
The Input/Output Programming System (IOPS) consists of an 1/0 control routine and individual hardware device
handling subroutines that process file and data level commands to the devices. These handlers exist for all standard
PDP-1 S/ l 0 peripherals (see Part II, Chapter 4 ).
The 1/0 control routine accepts user program commands and transfers control to the appropriate device handlers.
These device handlers are responsible for transferring data between the program and 1/0 devices, for initiating the
reading or writing of files, for the opening and closing of files, and for the performance of all other functions peculiar
to a given hardware device. They are also responsible for ignoring functions which they are incapable of handling (for
example, trying to rewind a card reader, or skipping files on a non-file-oriented device). All device handlers operate
either with or without the Automatic Priority Interrupt (API) option.
1.3.2 System Programs
The Basic 1/0 Monitor, in addition to the IOPS, is complemented by the following system programs:
a. FORTRAN IV Compiler, Object Time System, and Science Library
b. MACRO-IS Assembler
c. FOCAL
d. Dynamic Debugging Technique (DDT) Program
e. Text Editor Program (EDITOR)
f. Peripheral Interchange Program (PIP)
g. Linking Loader (LOADER)
h. 8TRAN, PDP-8 to PDP-1 S Translator
i. Chain Builder Program (CHAIN)
j. Chain Execute Program (EXECUTE)

k. PUNCH-IS

1.3.2.1 FORTRAN IV Compiler - The PDP-1 S FORTRAN IV compiler is a two-pass system that accepts statements
written in the FORTRAN IV language and produces a relocatable object program capable of being loaded by the
Linking Loader. It is completely compatible with USA FORTRAN IV, as defined in USA Standard X3.9-1966, with
the exception of the following features, which were modified to allow the compiler to operate in 8192 words of core
storage:
a. Complex arithmetic is not legal.
b. Adjustable array dimensions are not allowed at source level, but can be implemented by calling
dimension-adjustment subroutines.

1-S(Part I)

c. Blank Common is treated as Named Common except when object program is used in chaining.
d. The implied DO feature is not included for the DATA statement.
e. Maximum of 5 characters in Hollerith constants.
f. Specification statements must be strictly positioned and ordered.
The FORTRAN IV compiler operates with the PDP-15/10 program interrupt or API facilities enabled. It generates
programs that operate with the program interrupt or API enabled and can work in conjunction with assembly language programs that recognize and service real-time devices. Subroutines written in either FORTRAN IV or MACR015 assembly language can be loaded with and called by FORTRAN IV main programs. Comprehensive source language diagnostics are produced during compilation, and a symbol table is generated for use in on-line debugging with
DDT.
The PDP-15 FORTRAN IV Compiler, Object Time System, and Science Library are described fully in the FORTRAN
IV Manual (DEC-15-KFZA-D).

1.3.2.2 FOCAL - The FOCAL (Formulating On-Line Calculations in Algebraic Language) compiler is an on-line
interactive (conversational) system for non-programmers which makes use of standard mathematical notation and
short imperative English command statements to solve user problems.
Arithmetic capabilities of FOCAL permit single commands to generate random numbers and to evaluate standard
functions including: square root, absolute value, sign, integer, and natural exponent of any number; sine, cosine,
arctangent, and Naperian log value.
With FOCAL, user defined mathematical operations are executed by a single command. In addition, specific
hardware, text format, and error codes can be defined to facilitate user needs.
The FOCAL compiler is described fully in the PDP-15 FOCAL Manual (DEC-15-KJZA-D).

1.3.2.3 MACR0-15 - The MACR0-15 Assembler provides PDP-15/10 users with highly sophisticated macro
generating and calling facilities within the context of a symbolic assembler. MACR0-15 is described in detail in the
MACR0-15 Assembler Manual (DEC-15-AMZA-D). Some of the prominent features of MACR0-15 include:
a. The ability to (1) define macros

(2) define macros within macros (nesting)
(3) re-define macros (in or out of macro definitions)
(4) call macros within macro definitions
(5) have macros call themselves (recursion)
b. Conditional assembly based on the computational results of symbols or expressions.

l-6(Part l)

c. Repeat functions.
d. Boolean manipulation.
e. Optional octal and symbolic listings.
f. Two forms of radix control (octal, decimal) and two text modes (ASCII and 6-bit trimmed ASCII).
g. Global symbols for easy linking of separately assembled programs.
h. Choice of output format: relocatable, absolute binary (check summed); or full binary capable of being loaded
via the hardware READIN switch.
i. Ability to call input/output system macros that expand into IOPS calling sequences.
1.3.2.4 Dynamic Debugging Technique (DDT) Program - DDT provides on-line debugging facilities within the
PDP-15/lO Basic Software System, enabling the user to load and operate his program in a real-time environment while
maintaining strict control over the running of each section. DDT allows the operator to insert and delete breakpoints,
examine and change registers, patch programs, and search for specific constants or word formats.
The DDT-15 breakpoint feature allows for the insertion and simultaneous use of up to four breakpoints, any one (or
all) of which may be removed with a single keyboard command. The search facility allows the operator to specify a
search through any part or all of an object program with a printout of the locations of all registers that are equal (or
unequal) to a specified constant. This search feature also works for portions of words as modified by a mask. With
DDT-15, registers may be examined and modified in either instruction format or octal code, and addresses may be
specified in symbolic relative, octal relative, or octal absolute. Patches may be inserted in either source language or
octal.
DDT-15 is described more fully in the PDP-15 Utility Program Manual (DEC-15-YWZA-D).

1.3.2.5 Text Editor Program - The Text Editor of the PDP-15/10 Basic 1/0 Monitor provides the ability to read
alphanumeric text from an input device (paper tape reader), to examine and correct it, and to write it on an output
device. It can also be used to create new symbolic programs.
The Editor operates on lines of symbolic text delimited by carriage return (CR) or ALT MODE characters. These lines
can be read into a buffer, selectively examined, deleted or modified, and written out. New text may be substituted,
inserted, or appended.
For further details on the Text Editor, refer to the PDP-15 Utility Programs Manual (DEC-15-YWZA-D).

1.3.2.6 Peripheral Interchange Program (PIP) - The primary function of PIP is to facilitate the manipulation and
transfer of data files from any input device to any output device. It can be use<l to segment or combine files, perform
code conversions, and copy tapes.
Directions for the use of PIP- I 5 can be found in the PDP-I 5 Utility Programs Manual (DEC-15-YWZA-D).

1.3.2. 7 Linking Loader - The Linking Loader loads any PDP-15, FORTRAN IV, or MACR0-15 object program
which exists in relocatable format (or absolute format if pseudo-ops .ABS and .FULL are not used). Its tasks include

l-7(Part l)

loading and relocation of programs, loading of called subroutines, retrieval and loading of implied subroutines, and
building and relocation of the necessary symbol tables. Its operation is discussed in the PDP-1 S Utility Program
Manual (DEC-1 S-YWZA-D).

1.3.2.8 8TRAN - 8TRAN is used to translate programs written in Pal III, PAL-Dor MACR0-8 assembly language to
MACRO-IS assembly language for assembly and execution within the PDP-IS Basic Software environment. The
Translator produces a straightforward translation that clearly indicates the parts of the translated program requiring
review, in light of the PDP-I S's greater word length and more powerful instruction set. It does not simulate the PDP-8
or produce directly executable code. A full description is contained in PDP-1 S - 8TRAN Manual (DEC-15-ENZA-D).

1.3.2.9 Chain Builder and Execute Programs - The Chain Builder and Execute programs provide the user with a
capability for program segmentation which allows for multiple core overlap of executable code and certain types of
data areas. A more complete description of the Chain Builder and Execute programs is given in the Utility Programs
Manual (DEC-15-YWZA-D).
1.3.2.10 PUNCH-15- PUNCH-IS is a utility program used to output selected areas of core memory onto paper tape
for use with the PDP-15 Basic 1/0 Monitor. The primary applications of PUNCH-15 in the Basic 1/0 Monitor
environment are:
a.

System program modification

+.DAT SLOT reassignment

c. Production of an executable user program core load on a single paper tape. This is particularly useful when
that core load consists of a main program and several subroutines/library routines, the constant reloading of which
may be time consuming.
The PDP-lS Utility Programs Manual (DEC-lS-YWZA-D) describes PUNCH-IS more fully.

l-8(Part I)

PARTil
COMPACT SOFIWARE

PART II
COMPACT SOFIWARE

CHAPTER 1
COMPACT ASSEMBLER (CAP-15)

1.1 INTRODUCTION

The CAP-15 Assembler processes symbolic source programs to produce a binary code that can be executed by the
PDP-15/10 computer. It normally processes source programs in two passes and requires 3K of core memory. The
Assembler makes machine language programming for the PDP-15/10 easier, faster, and more efficient. It permits the
programmer to use mnemonic symbols to represent instruction operation codes, locations, and numeric quantities;
thus, the programmer can easily refer to any point in his program without knowing actual machine locations.
The programmer can direct the Assembler's processing by means of a powerful set of pseudo-operation (pseudo-op)
instructions. These pseudo-ops can be used for program control, to reserve blocks of core storage, to set the radix for
numerical interpretation by the Assembler, to handle strings of text characters in 6-bit ASCII code, to define
symbolic addresses, and many other functions that are described in detail in Section 1.4.
An output listing, showing both the programmer's source coding and the object program produced by the Assembler,
can be printed. This listing includes all the symbols used by the programmer and their assigned values. If assembly
errors are detected, erroneous lines are marked with meaningful letter codes, which can be interpreted by referring to
Section 1.5.4. Operating procedures are found in Section 1.5.

The Assembler normally processes a source program in two passes; that is, it reads the same source program twice,
producing a printed listing and/or outputting the object code during the second pass. (However, if the listing and
binary output utilize the same device, then a third pass is required to produce the binary output.) Assembler coding
for the two passes is resident in memory at the same time. PASSI and PASS2 are almost identical in operation. The
main function of PASSI is to identify locations that are to be assigned symbols, and to construct a symbol table.
PASS2 uses the information derived by PASSI (and left in memory) to produce the final listing object code output.

1.2 HARDWARE REQUIREMENTS AND OPTIONS
The Assembler operates with PDP-15/10 systems having at least 4K of core memory and a console Teletype (ASR33).
With the addition of a high-speed paper tape reader/punch, the user can significantly decrease assembly time and also
have the capability of simultaneously producing a listing and binary output during PASS2.

I- ](Part II)

l.3 ASSEMBLY LANGUAGE ELEMENTS

1.3. l Program Statements

A single statement may be written on a 72-character teletype line, in which case a carriage return delimits the
statement. Since a carriage return is not a printable character, it is represented in this manual as)
STATEMENT)
Several statements may be written on a single line, separated by semicolons:
STATEMENT;STATEMENT;STATEMENT)
In this case, the statement line ends with a carriage return character, but semicolons are used as internal statement
delimiters. Thus, if a statement is followed by another statement on the same line, it must end with a semicolon.
A statement may contain up to four fields separated by a space, spaces, or tab characters. These four fields are the
label (or tag) field, the operation field, the address field, and the comments field. Because the space and tab characters
are not printed, the space is represented by L-1 , and the tab by
in this manual. Tabs are normally set ten spaces
apart on most teletype machines, and are used to line the fields up in columns in the source program listing.

-I

The basic statement format is as follows:
LABEL

-I OPERATION -I ADDRESS -I /COMMENTS)

where each field is delimited by a tab or space, and each statement is terminated by a semicolon or carriage return.
The comments field is preceded by a tab (or space) and a slash(/).
An assembly statement may have an entry in any or all of the four fields. The following forms are acceptable:

TAG)
TAG -I OP)
TAG -I OP -I ADDR)
TAG -I OP -I ADDR L-1 comments)
TAG -I OP L-1 comments)
TAG -j -I ADDR)
TAG -I -I ADDR '--'comments;
TAG -I comments)
-I OP)
-I OP -I ADDR)
-I OP -I ADDR -I comments)
-I OP -I comments
-I -I ADDR)
-I -I ADDR -I comments;
/comments;
-I comments)
Note that when a label field is not used, its delimiting tab is written, except for Jines containing only comments. When
the operation field is not used, its delimiting tab is written if an address field follows, except in label only and
comments only statements.

l-2(Part II)

A label (or tag) is a symbolic address created by the programmer to identify the statement. When a label is processed
by the Assembler, it is said to be defined. A label can be defined only once. The operation code field may contain a
machine mnemonic instruction code, a pseudo-op code, a number, or a symbol. The address field may contain a
symbol, number, or expression which is evaluated by the Assembler to form the address portion of a machine
instruction. In some pseudo-operations, this field is used for other purposes, as explained later in this manual.
Comments are usually short explanatory notes that the programmer adds to statement as an aid in analysis and
debugging. Comments do not affect the object program or assembly processing; they are merely printed in the
program listing. Comments must be preceded by a slash (/);the slash must be preceded by one of the following:

a. Space
b. Tab
c. Semicolon
d. Carriage return

1.3.2 Symbols
A symbol consists of a string of alphabetic or alphanumeric characters (including periods and percent signs). The frrst
character of a symbol must be a letter, a period or a percent sign. The first character of a symbol in the label field
must not be a digit. A period can not be used alone as a symbol in the label field. The letter X, if used alone, can only
appear in an address field (see 1.3.4.4).

The following symbols are legal:
MARKI

.. 1234

%50.99

.A
.%

P9.3

INPUT

.SA

L@Bl

The colon(:) and@ are illegal characters.
The first character may not be a digit.
See section 1.3.4.4.

The following symbols are illegal:
TAG:!
SABC

Only the first six characters of a symbol are meaningful to the Assembler, but the programmer may use more for his
own information. If he writes,
SYMBOLl
SYMBOL2
SYMBOL3
as the symbolic labels on three different statements in his program, the Assembler will recognize only SYMBOL and
type error flags on the lines containing SYMBOLl, SYMBOL2 and SYMBOL3 because to the Assembler they are
duplicates of SYMBOL.

l.3.2.1 Evaluation of Symbols - When the Assembler encounters a symbol during processing of a source language
statement, it evaluates the symbol by reference to two tables: The user's symbol table and the permanent symbol

l-3(Part II)

table. The user's symbol table contains all symbols defined by the user. The user defines symbols by using them as
labels or by direct assignment statements (see Section l.3.2.2). A label is defined when first used, and cannot be
redefined. When a label is defined by the user, it is given the current value of the Location Counter (see Section
1.3.4).
The Assembler has, in its permanent symbol table, definitions of the symbols for all of the PDP-15/10 memory
reference instructions, operate instructions, index and limit register instructions, and some input/output transfer
instructions. (See Appendix B for a complete list of these instructions.) Both the permanent symbol table and the
user's symbol table reside in storage in alphanumeric sequence. The permanent symbol table symbols can be used in
the operation field of a statement without prior definition by the user.
Example

When the LAC symbol appears in the operation field of a statement,
the Assembler treats it as an op code rather than a symbolic address. It
has a value of 2000008 , which is taken from the operation code
definition in the permanent symbol table.

The user can use instruction mnemonics (see Appendix B) or the pseudo-instruction mnemonics code (see Appendix
C) as symbol labels. For example,
DZM -j DZM ...... Y)
where the label DZM is entered in the user's symbol table and given the current value of the Location Counter, and
the op code DZM is given the value 140000 from the permanent symbol table. The user must be careful, however, in
his usage of these dual purpose (field dependent) symbols. Symbols in the operation field will be interpreted as either
instruction codes or pseudo-ops, rather than as a symbolic label, if they are in the permanent symbol table. In the
following example, several symbols with values have been entered in the user's symbol table and the permanent
symbol table. The sample coding shows how the Assembler uses these tables to form object program storage words.

Permanent Symbol Table

User Symbol Table

Symbol

Value

Symbol

Value

TAGl

100

LAC

200000

TAG2

200

DAC

040000

DAC

300

JMP

600000

If the following statements are written,

TAG! -j

TAG2

l-4(Part II)

the following code is generated by the Assembler

DAC -j TAG2)

040200

-j LAC -j DAC)

200300

DAC

-j JMP

-I TAGl)

600100

-j TAGl)

000100

-I

000300

1.3.2.2 Direct Assignment Statements - The programmer may define a symbol directly in the user's symbol table by
means of a direct assignment statement, written in the form:
SYMBOL=n)
or
SYMl =SYM2)
where n is any number or expression. There should be no spaces before the symbol or between the symbol and the
equal sign. (The operation field is assumed to be to the right of the equal sign, unless it is followed by a space or a tab,
in which case the address field is assumed.) The Assembler enters the symbol in the user's symbol table, along with
the assigned value. Symbols entered in this way can be redefined. The following are legal direct assignments:
Z=28;A= l;B=~
A symbol can also be assigned a symbolic value:

A=4)
B=A)
The symbol B is given the value 4. Direct assignment statements do not generate storage words in the object program. In general, it is good programming practice to define symbols before using them in statements which generate storage words. The Assembler will interpret the following sequence without trouble.

Z=S)
Y= Z)
V=Y)
-j LAC ....., V L..1 /SAME AS LAC 5

A symbol may be defined after use. For example,
-jLAC ...._. Y)
Y= I.J
This is called a forward reference, and is resolved properly in PASS2. When first encountered in PASS 1, the LACY
statement is incomplete because Y is not yet defined. Later in PASS 1, Y is given the value 1. In PASS 2, the Assembler
finds that Y = l in the symbol table, and forms the complete storage word.

1-S(Part II)

Since the Assembler operates in two passes, only one-step forward references are allowed. The following forward
reference is illegal:
-\ LAC ...._. Y,)
Y= Z)

Z= I )
In the listing, during PASSI, the line which contains Y = Z will be printed with an A error code indicating a direct
assignment error.

1.3.2.3 Undefined Symbols - If any symbols remain undefined at the end of PASSI of assembly, they are
automatically defined as the addresses of successive registers following the end of the program (i.e., following the
highest program counter value encountered by the Assembler). All statements that referenced the undefined symbol
will be flagged as undefined. One memory location is reserved for each undefined symbol with the initial contents of
the reserved location being unspecified.
Example

Location
Counter

Source Statement

Generated
Code

100

-\LAC UNDEF 1)

200104

101

-! LAC UNDEF3

200106

102

-j LAC UNDEF4

200107

103

-! LAC UNDEF2

200105

Comments

'
> Flagged as an error

-I.END

1.3.3 Numbers

1.3.3.1 Integer Values - An integer is a string of digits, with or without a leading sign. Negative numbers are
represented internally in two's complement form. The range of integers is as follows.

Unsigned
Signed

0-+ 262143, 0
0 -+ 1310711 0

(777777 8 )
(377777 8 )

e.g., 218 -1
e.g.,± 2 17 -1

An octal integer* is a string of digits (0-7), signed or unsigned. If a non-octal digit is encountered (8 or 9) it will be
flagged as a numerical error.

*Initiated by usage of .OCT pseudo-instruction and is also the initial assumption if no radix control pseudo-instruction was encountered (see
Section 1.4.4).

l-6(Part II)

Example

Coded Value

Generated Value (Octal)

Comment

-5

777773

two's complement

3347

003347

3779

000000

error

A decimal integer* is a string of digits (0-9), signed or unsigned.

Example

Coded Value

Generated Value (Octal)

Comment

-8

777770

two's complement

256

000400

1.3.3.2 Expressions - Expressions are strings of symbols and numbers separated by arithmetic or Boolean operators.
Expressions represent unsigned numeric values ranging from 0 to 2 18 -1. All arithmetic is performed in unsigned
integer arithmetic (two's complement), modulo 2 18 . Division by zero is regarded as division by one and results in the
original dividend; this condition will not be regarded as an error. Fractional remainders are ignored. The value of an
expression is calculated by substituting the numeric value for each element (symbol) of the expression and performing
the specified operations.
The following are allowable operators to be used with expressions:

Character
Function

Name
Plus

Symbol

Minus

Addition (two's complement)
Subtraction (convert to two's complement
and add)

Asterisk

Multiplication (unsigned)

Slash

Division (unsigned)

*Initiated by usage of DEC pseudo-op (see Section 1.4.4).

1-7(Part II)

Character
Function

Name

Symbol

Ampersand

Logical AND

Exclamation point

Inclusive OR

Back slash

Boolean

Exclusive OR
Comma

•

Operations are perfonned from left to right (i.e., in the order in which they are encountered). For example, the
assembly language statement A + B*C + D/E - F "' G is equivalent to the following algebraic expression
(((((A+ B)* C) + D)/E) - F)* G.

Example
Assume the following symbol values:

Symbol

Value (Octal)

000002

000010

000003

000005

The following expressions would be evaluated according to the rules above.

Expression

1-8(Part II)

Evaluation (Octal)

A+B-C

000007

A/B+A *C

000006

B/A-2"' A-I

000003

A&B

000000

C+A&D

000005

B*D/A

000024

B*C/A*D

000074

(The remainder of A/B is lost)

The expression (A + B) * (C + D) cannot be represented in one source statement. To circumvent this problem, at least
one of the members of the expression should be defined by a direct assignment statement (see Section 1.3.2.2).
Examples

E=A+B)
C+D*E

/Represents (A + B) * (C + D))

or
E=A+B)
F=C+D)

E*F

/Represents (A + B) * (C + D~

1.3.4 Address Assignments
As source program statements are processed, the Assembler assigns consecutive memory locations to the storage
words of the object program. This is done by reference to the Location Counter, which is initially set to zero, and
incremented by one each time a storage word is formed in the object program. Some statements, such as machine
instructions, cause only one storage word to be generated, incrementing the Location Counter by one. Other
statements, such as those used to enter data or text, or to reserve blocks of storage words, cause the Location Counter
to be incremented by the number of storage words generated.
1.3.4. l Referencing the Location Counter - The programmer may directly reference the Location Counter by using
the period symbol(.) in the address field. He can write
~ JMP ~ .-1)
which will cause the program to jump to the storage word the address of which was previously assigned by the
Location Counter. The Location Counter can be set to another value by using the .LOC pseudo-op, as described in
1.4.8.

1.3.4.2 Direct Addressing - Direct Addressing occurs when bits 4 and 5 of the Memory Reference Instruction word
(see Figure 1-1) are set to 0. The machine action defined by the operation code field is applied to the operand
specified by the 12-bit address field. The 12-bit address field allows addressing of any location of the 4,096 word
memory page. Access to locations outside the current page is gained via execution of indexed or indirect address
instructions.

1.3.4.3 Indirect Addressing - To specify an indirect address, which can be used only in memory reference
instructions, the programmer writes an asterisk (*) immediately following the operation field symbol. This sets the
Defer Bit (bit 4) for the storage word (see Figure 1-1) and the contents of the address field point to a location (within
the current page) which contains the 15-bit effective address.
The error flag S results if an asterisk (*) suffixes a non-memory reference instruction.

1-9(Part II)

Two examples of legal indirect addressing are as follows:

-j TAD* -I A)
-j LAC* -j B)
The following example is illegal .

....j CLA*)

Indirect addressing cannot be specified in non-memory reference
instructions.

NOTE

A symbol followed by an asterisk and by another symbol is interpreted as a
multiplication operation (e.g., LAC*SYMBOL).

Operation
Code

•

17
Address Field

X= Indexed Address Indicator
*=

Indirect Address Indicator

Figure 1-l. Memory Reference Instruction Format

l.3.4.4 Indexed Addressing - Indexed Addressing is specified by bit 5 (indirect bit) of the Memory Reference
Instruction word (see Figure 1-1) being set to l. Indexed Addressing can be either Duect or Indirect. For Direct
Indexed Addressing, bit 4 is 0 and bit 5 is l. In this mode, the effective address is taken as the sum of the 12-bit
address field (bits 6-17) and the contents of the Index Register. The Index Register is an 18-bit hardware register
which may contain a signed 2's complement number.

Indirect Indexed Addressing is specified by bits 4 and S being set to l. In this mode of addressing, the effective
address (EFA) is taken as the contents of the location specified by the sum of the 12-bit address field and the Index
Register.

1-l O(Part II)

Example

Contents
C(30) = 37
C(XREG) = 20
C(57) = 101

LAC* -j 30,X
/C(30) + C(XREG) = EFA
/37 + 20= EFA
/57 = EFA
/C(57) =AC
/101 =AC

Using a comma X(,X) in the address field causes bit 5 of the current address value to be exclusively ORed with l 0000.
For example, if bit 5 of the previous address value was 1, this operation sets it to 0. The Assembler takes this into
consideration when the operation field value and the address field value are combined (see 1.3.6.2).
In all cases when X is used, the value 10000 is used to perform the operation specified.* Standard usage of X is shown
below.
Example

A=50
LAC --! A,X
Storage word generated 210050
Using X to denote Index Register usage causes the following restrictions:
a. X cannot be used as a TAG.
b. X may not be used more than once in an expression.
c. X can only appear in an address field.
d. X cannot be used with a .DSA statement.

1.3.4.5 Literals - Symbolic data references in the operation and address fields may be replaced with direct representation of the data enclosed in parentheses**. This inserted data is called a literal. The Assembler sets up the address
link, therefore one less statement is necessary in the source program.
The following examples show how literals may be used, and their equivalent statements. The information contained
within the parentheses, whether a number, symbol, expression, or machine instruction, is assembled and assigned
consecutive memory locations after the highest location used by the program (including registers reserved for
undefined symbols). The address of the generated word will appear in the statement that referenced the literal.
Duplicate literals are stored only once so that many uses of the same literal in a given program result in only one
memory location being allocated for that literal. Nested literals are not allowed and result in an L error diagnostic at
assembly time.
*The symbol X should only be used for indexing purposes. After the expression has been evaluated, bit 5 will be set to l regardless of
the expression result.
**The opening parenthesis [ ( ] is mandatory while the closing parenthesis [ ) ] is optional. The operation field is assumed to be to the
right of the opening parenthesis [ ( ) unless it is followed by a space or tab, in which case the address field is assumed.

1-1 l(Part II)

Equivale~ Statements

Usage of Literal
-I ADD....., (1)

-I ADD....., ONE
ONE

-I LAC L-1 (TAG

-t 1
-I LAC ....... TAGAD

TAGAD -ITAG
-I LAC ....., (DAC -I TAG)

-I LAC L-1 (JMP

-I .+2)

-I LAC....., INST
INST

-I DAC -I TAG

HERE

-I LAC ....., INST

INST

-j JMP ....., HERE+2

The following sample program illustrates how the Assembler handles literals.

Location Counter

Source Statement

Generated Code

-.j .LOC ._, 100

-I LAC ....., ( 100)

200110

101

-I DAC ....., 100

040100

102

-I LAC ....., (JMP ....., .+5)

200111

103

-I LAC....., (TAGl)

200110

104

-I LAC....., (JMP....., TAGl)

200112

105

-I LAC ....., (JMP ....., TAG2)

200112

100

TAGl

TAG2=TAG1
106
107

DAC

-I LAC ....... (JMP)

200113

-I LAC ....., (DAC -I DAC)

200114

-j .END

1-l 2(Part II)

Location
Counter

Source St.atement

Generated
Code

Generated Literals
110

000100

111

600107

112

600100

ll3

600000

114

040107

1.3.5 Statement Fields
1.3.5.1 Label Field - If the user wishes to assign a symbolic label to a statement, to facilitate references to the
storage word generated by the Assembler, he may do so by beginning the source statement with any desired symbol.
The symbol must be terminated by a space (spaces) or tab, or a statement terminating semicolon, or carriage return.

Examples
TAG ....., any value;
TAG .._, any value)
TAG

--1 any value)

TAG;

These examples are equivalent to coding

TAG)

TAG

--j 0)

in that a word of all Os will be output with
the symbol TAG associated with it.
Symbols used as labels are defined in the user's symbol table with a numerical value equal to the present value of the
Location Counter. A label is defined only once; if it was previously defined by the user, the current definition of the
symbol will be flagged as a multiple definition error. All references to a multiply-defined symbol will be made to the
first value encountered by the Assembler.

Example

location
Counter

Statement

Storage Word

Generated

100

-I LAC --1 B

200103

101

--j LAC -IC

200104

--jA

200100

102

-!LAC

Notes

error, multiple definition
first value of A referenced

l- l 3(Part II)

Location
Counter

103
104

Statement

Storage Word
Generated

-I 0
c -I 0

Notes

000000
000000

Anything more than a single symbol to the left of the label-field delimiter is an error; it will be flagged and ignored.
The following statements are illegal.

-I

TAG+l
LAS)
LOC* 2 -j RAR)
Redefinition of certain symbols can be accomplished by using direct assignments; that is, the value of a symbol can be
modified. If an Assembler permanent symbol or user symbol (which was defined by a direct assignment) is redefined,
the value of the symbol can be changed without causing an error message. If a user symbol, which was first defined as
a label, is redefined by either a direct assignment or by using it again in the label field, it will cause an error.

Example

Coding

Generated Value (Octal)

sets current value of A to 3

A=3

-I LAC -I A

200003

-I DAC -j A

040003
redefines value of A to 4

A=4

-j LAC -j A

200004

-j DAC -jA

040004

B=A

--\DAC -j B
PSF=700201

Comments

illegal usage; a label cannot be
redefined
040105
to redefine possibly incorrect
permanent symbol definition.

*Assume that this instruction will occupy location 105.
1.3.5.2 Operation Field - Whether or not a symbol label is associated with the statement, the operation field must be
delimited on its left by a space(s) or tab. If it is not delimited on its left, it will be interpreted as the label field. The
operation field may contain any symbol, number, or expression which will be evaluated as an 18-bit quantity using
unsigned arithmetic modulo 2 1 8 • In the operation field, machine instruction op-codes and pseudo-op mnemonic

l -l 4(Part II)

symbols take precedence over identically named user defined symbols. The operation field must be terminated by one
of the following characters:

-I or L...1 (s)

field delimiters

;or;

statement delimiters

Examples

-I ISZ)
-I .+3 L...I)

TAG

L...1 CMA!CMLJ

-I TAG/5+'fAG2;

1.3.5.3 Address Field -The address field, if used in a statement, must be separated from the operation field by a tab,
or space(s). The address field may contain any symbol, number, or expression which will be evaluated as an 18-bit
quantity using unsigned arithmetic, modulo 2 18 • If op-code or pseudo-op code symbols are used in the address field,
they must be user defined, otherwise they will be undefined to the Assembler and cause an error message. The symbol
X cannot be user def"med (see section 1.3.4.4). The address field must be terminated by one of the following
characters:

-I or L...1 (s)

field delimiters

)Or;

statement delimiters

Examples
TAG2

-I DAC -j .+3 -j /COMMENT)
-I

-I TAG2/5 + 3

&...1 (s))

-1 ISZ -I TAG2, x + 4)
-I JMP -I BEGIN)

-I TAD -I A; -j DAC -I ~
In the last example, a tab or space(s) is required after the semicolon so the Assembler can interpret DAC as being the
operation field rather than the label field.
An error condition will exist if the bank and page bits (3, 4 & 5) of the address do not match the bank and page bits
of the bank currently being assembled into and the extended memory bits of the address are not zero;-

Examples
Location
(octal)

Instruction

1000

-!LAC ....... 100

1001

-I DAC

Comments

wiu not came enm me_.

L...I

101

1-15(Part II)

Location

Instruction

(octal)

Comments

1002

-I JMS L...I 250

will not cause error messages

1003

-.j ISZ L-1 40146

will cause a bank error message code B
(see 1.3.5.3)

1004

-I LAC L...I 10100

will cause a bank error message code
(see 1.3.5.3)

1.3.5.4 Comments Field - Comments may appear anywhere in a statement. They must begin with a slash (/)
immediately preceded by a
L...I

·space(s)

(s)

tab
carriage return (end of previous line)

)

semicolon
Comments are terminated only by a carriage return.

Examples
L...1

(s) / THIS IS A COMMENT)

TAG 1

-.j LAC .._. /after the ; is still a comment)

/THIS IS A COMMENT)

-I RTR L-..1 /COMMENT)
-I RTR; -! RTR; /THIS IS A COMMENT)

--I

Observe that
A/COMMENT) is not a comment, but rather an operation field expression. A line that is
completely blank; that is, there is no data between two sets of) (s), will be treated as a comment by the Assembler.
1.3.6 Statement Evaluation
When the Assembler evaluates a statement, it examines the first cbaracter position in the label, operation, and address
fields for a numeric character. (Comments fields are ignored.) If the first character of a field is numeric, the contents
of the whole field is treated as a number.
1.3.6.l Numbers - Numbers are not field dependent. When the Assembler encounters a number (or expression) in
the operation or address fields (a number in the frrst character position of the label field is illegal), it uses those values
to form the storage word. The following statements are equivalent:

l-l 6(Part II)

-I 200000 ...... 10)
-j 10 +LAC)

-I LAC .__. I 0)
All three statements cause the Assembler to generate a storage word containing 2000 I 0.
A statement can consist of a number or expression which generates a single 18-bit storage word; for. example,

-I 23;

L.J

45; &....I 357; &....I 62)

This group of four statements generates four words interpreted under the current radix.
Zero words are generated by statements containing only labels. For example,
A;B;C;D;E)
generates five words set to zero, which can be referenced by the labels defined.

1.3.6.2 Word Evaluation - When the Assembler encounters a symbol in a statement field, it determines the value of
the symbol by referring to the user's symbol table and the permanent symbol table, according to the priority list
shown below. The value of a storage word is computed by combining the 18-bit operation field quantity with the
18-bit address field quantity, in the following manner.
[(OPERATION FIELD+ (ADDRESS FIELD & 017777))] =Value of Word
0-17
0-17
If the operation code is one of the 9-bit operators (see Appendix B), the word value is computed as follows.

[(OPERATION FIELD+ (ADDRESS FIELD & 777))}= Value of Word
0-17
0-17
Extensive error checking is performed on the address field value to ensure correct results. If the page bit of the
address field is different from that of the program counter, the line is flagged with a 'B' error code at assembly time,
indicating a page error. Page bits are always set to 0 if the address value is legal and Index Register usage is not
specified.
Example:
Error

Flag

LOC

Object
Code

00100

210000

10000

040100

Source Code

.LOC 100
LAC B /ADDRESS OF BON A DIFFERENT PAGE

.LOC 10000
DAC A /ADDRESS OF A ON A DIFFERENT PAGE

The LAW instruction and the 9-bit operators check the low order 5 bits (or 9 bits) of the address value for validity by
AN Ding off the low order bits and checking the result for 0 or for equality with the 'AND' value.
l-l 7(Part II)

{address field & 760000] = 760000 or 000000
OR
IF A 9 BIT OPERATOR
{address field & 777000] = 777000 or 000000
Example

/The value for -I is 777777.
/When this value is ANDed
/with 760000, the result
/is 760000 and valid.

LAW-I

If the ADDRESS FIELD and 7600008 does not equal 760000 8 or 000000, any erroneous results produced are
flagged by the Assembler. This validity check is performed only if an operation field and an address field are present.
If the ADDRESS FIELD ANDed with 7600008 does not equal 7600008 or 000000, any erroneous results produced
are flagged by the Assembler. This validity check is performed only if an operation field and an address field are

present.
If the instruction is a Memory Reference instruction, the low order 5 bits of the address value are examined to make
sure they are not set*. If any of the 5 bits are set, the line is flagged with an E error code during assembly.

-I 2 -I 5 -I /GENERATES 000007
The value of a symbol depends on whether it is in the label field, or the address field. The Assembler attempts to
evaluate each symbol by running down a priority list, depending on its field as shown.

Current value
Location Counter

Address Field

Operation Field

Label Field

I . Pseudo-op
2. Direct assignment in user
symbol table.

User symbol table,
(in eluding direct
assignments)

2. Undefined
3. Permanent symbol table
4. User symbol table
5. Undefined
This means that if a symbol is used in the address field, it must be defined in the user's symbol table; otherwise, it is
undefined.
In the operation field, pseudo-ops take precedence and cannot be redefined. Direct assignments allow the user to
redefine machine op codes, as shown in the example below.

*Does not include LAW instruction.

l -l 8(Part II)

Example

DPOSIT = DAC)
The user can use machine instruction codes and Assembler pseudo-op codes in the label field and refer to them later
in the address field.

1.4 PSEUDO OPERATIONS
In the discussion of symbols in the previous chapter, it was stated that the Assembler has, in its permanent symbol
table, definitions of the symbols for all the PDP-15 memory reference instructions, operate instructions, and some
IOT instructions (listed in Appendix B) which may be used in the operation field without prior definition by the user.
Also contained in the permanent symbol table are a class of symbols called pseudo-operations (pseudo-ops) which,
instead of generating instructions or data, direct the assembler on how to proceed with the assembly.
By convention, the first character of every pseudo-op symbol is a period (.). This convention is used to prevent the
programmer from inadvertently using, in the operation field, a pseudo-instruction symbol as one of his own.
Pseudo-ops may be used only in the operation field.

1.4.1 Program Control (.END)

One pseudo-op that must be included in every source program is .END, which must be the last statement in the
program. This statement marks the physical end of the source program and may also contain the location of the first
instruction in the object program to be executed at run-time.
The .END statement is written in the general form

-l .END._. START)
where START may be a symbol, number, or expression whose value is the address of the first program instruction to
be executed.
If a starting address is not present in an .END statement, the program will halt after being loaded and the user must
manually start his program.

The following are legal .END statements:
-\ .END ..... BEGIN + 5)
-\.END &....1 200)

-j .END)
1.4.2 Program Segments (.EOT)
If the input source program is physically segmented, each segment except the last must terminate with an .EOT
(end-of-tape) statement. The last segment must terminate with an .END statement. For example, if the input source

l-l 9(Part II)

program is prepared on three different tapes, the first two are terminated by .EOT statements, and the last by an
.END statement. The .EOT statement is written without label and address field, as follows.

-j .EOT)
1.4.3 Reserving Storage Words In The Object Program
The programmer may reserve blocks of storage words, or single words, for use during program execution.
1.4.3.1 Reserving Blocks of Storage (.BLOCK) - .BLOCK reserves a block of memory equal to the value of the
expression contained in the address field. If the address field contains a numerical value, it will be evaluated according
to the radix in effect. The symbolic elements of the expression must have been defined previously; otherwise, phase
errors (see Section 1.5.5) might occur in PASS2.
The user may reference the first location in the block of reserved memory by defining a symbol in the label field. The
initial contents of the reserved locations are unspecified.

Label Field

Operation Field

Address Field

User Symbol

.BLOCK

Predefined Expression

Examples

BUFF -j .BLOCK '--' 12

.-j .BLOCK '--' A-B+65)
1.4.3.2 Reserving Single Storage Words - Storage words set to zero are set up as
A

-j O; -j O; -j 0)

In this way, three words are set to zero, starting at A. Storage words set to zero are also set up by statements
containing only labels:
A;B;C;D;E)
1.4.4 Radix Control (.OCT and .DEC)
The initial radix (base) used in all number interpretation by the Assembler is octal (base 8). In order to allow the user
to express decimal values, and then restore to octal values, two radix setting pseudo-ops are provided.

Pseudo-op Code

.OCT
.DEC

l-20(Part II)

Meaning

Interpret all succeeding numerical values in base 8 (octal)
Interpret all succeeding numerical values in base IO (decimal)

These pseudo-instructions must be coded in the operation field of a statement. All numbers are decoded in the
current radix until a new radix control pseudo-instruction is encountered. The programmer may change the radix at
any point in a program.

Source Program

Generated Value (Octal)

-I 100

-.j LAC

Radix in Effect

200100

000025

200144

000423

-.j 76

000076

-.j 85

000000

error

-.j 25

Uri tUtl volue ;,
assumed to be octal

-.j .DEC

-I 100

-.j LAC
-.j 275
-.j .OCT

1.4.5 Text Handling (.SIXBT)

Label Field

Operation Field

SYMBOL

.SIXBT

Address Field

delimiter

character string

delimiter

.SIXBT denotes 6-bit trimmed ASCII characters, which are formed by truncating the leftmost bit of the
corresponding 7-bit ASCII character. The Assembler converts the characters to their appropriate numerical equivalent
(see Appendix A). The characters are packed, three per word left justified, with unused bits set to zero.

I st character

2nd character

3rd character

Spaces or tabs prior to the text delimiter are ignored. Any printing character may be used as the text delimiter,
except). The text delimiter must be present on both ends of the text string, otherwise, the user may get more
characters than desired; however,) may be used to terminate the test string. After a carriage return, a new .SIXBT
pseudo-op is necessary to continue with 6-bit text.

l-21 (Part II)

Examples

Source

Generated Code
010203
040500
576162
655700
010203
040000

..j.SIXBT ..j /ABCDE/)
TAG

..j.SIXBT -j -/125/-)

-I /ABCD)

..j .SIXBT
1.4.6 Defining A Symbolic Address (.DSA)

.DSA (define symbol address) is used in the operation field when it is desired to create a word composed only of an
address field. It is especially useful when a user symbol is also an instruction or pseudo-op symbol.
Label Field

Operation Field

Address Field

User Symbol

.DSA

Any expression

Examples

JMP -\ LAC -\ TAG

-I .DSA ..j JMP

-I

Equivalent methods of defining the user symbol JMP to be in the
address field.

-I JMP

1.4. 7 Conditional Assembly (.I FOE F, .I FUND and .ENDC)
It is often useful to assemble some parts of the source program on an optional basis. This is done by means of
conditional assembly statements of the form:

-I .IF... -I expression
The pseudo-op may be one of the two conditional pseudo-ops shown below, and the address field may contain any
symbol or expression. If the condition is satisfied, that part of the source program starting with the statement
immediately following the conditional statement and up to the .ENDC (end conditional) pseudo-op is assembled. If
the condition is not satisfied, this coding is not assembled (it is treated as comments).
The two conditional pseudo-ops (sometimes called IF statements) and their meanings are shown below.
Conditional
Pseudo-op

I-22(Part II)

Assemble
IF xis:

.IFDEF

defined (present) in user symbol table

.!FUND

undefined (not present) in user's symbol table

In the following sequence, the pseudo-op .IFDEF is satisfied, and the source program coding between .IFDEF and
ENDC is assembled.

Y=5)

-I .IFDEF -I Y)
-!LAC.._. A)

-j DAC .._. B)
-j .ENDC)
Conditional statements may be nested. For each IF statement there must be a terminating .ENDC statement. If the
outermost IF statement is not satisfied, the entire group is not assembled. If the first IF is satisfied, the succeeding
coding is assembled. If another IF is encountered, however, its condition is tested, and the succeeding coding is
assembled only if the second IF statement is satisfied. Logically, nested IF statements are like AND circuits. If the
first, second and third conditions are satisifed, then the coding that follows the third nested IF statement is
assembled.

Example

-I .IFDEF .._. W)
-j LAC -I TAG)
-I .IFUND ....... Y)
-I DAC -I TAGl)

conditional l initiator

-1.ENDC)

conditional 2 terminator

-I .IFDEF Z)
-j DAC -I TAG2)

conditional 3 initiator

-1.ENDC)

conditional 3 terminator

-1.ENDC)

conditional l terminator

&...1

conditional 2 initiator

IF statements may be activated by means of a strip of tape which precedes the regular source tapes. The strip tape
may take on the following form:
Strip Tape

A=O

/Causes A and B to be

B=O

/entered into the user's symbol table.

-1.EOT

/Terminates the strip tape

1-23(Part II)

Source Program
~ JFDEF

-j A

/This coding would be assembled.

-j .ENDC

-I .IFUND -I B
/This coding would not be assembled.

-j .ENDC
-j .END
Since the purpose of the strip tape is to enter definitions into the user's symbol table, it is only necessary to have the
strip tape read in PASS I of assembly.

1.4.8 Setting The Location Counter (.LOC)

Label Field

Operation Field

Address Field

Not Used

.LOC

Predefined symbolic
expression, or number

The .LOC pseudo-op sets or resets the Location Counter to the value of the expression contained in the address field.
The symbolic elements of the expression must have been defined previously; otherwise, phase errors might occur in
PASS2. The .LOC pseudo-op may be used anywhere and as many times as required. If the .LOC pseudo-op is not
used, the assembler assumes location 0 as the starting point of the program.

Examples
Location Counter

I-24(Part II)

Instruction

100

-..j .LOC ._. l 00

100

-.f LAC a....1 TAG l

Location Counter

Instruction

101

-I DAC '-' TAG2

102

-I LOC L-1

102

-j LAC L-1 B

103

-I DAC L-1 C

107

-f .LOC '-' A+5

lIO

-I LAC...., C
-I DAC D

111

-j LAC L-1 E

112

-I DAC ,_, F

107

L-1

1.4.9 Listing Control (.XLIST and .LIST)
The following assembler listing controls are effective only when a listing was requested in the command string
(Section 1.5.2).
The .XLIST statement causes the assembler to stop listing the assembled program. The listing printout actually starts
at the beginning of PASS2; therefore, to suppress all of the program listing, .XLIST must be the first statement in the
program. If only a part of the program listing is to be suppressed, the .XLlST statement can be inserted at any point
to stop listing from that point.
The .LIST statement, which is normally used following an .XLIST statement, causes the assembler to resume the
listing at the point at which it is encountered.

1.4. IO Object Output Control (.FULL)

Label Field

Operation Field

Address Field

Not Used

.FULL

Not Used

The .FULL pseudo-op causes hardware readin-mode binary output to be produced (see Section 1.5.4.3 for a·
description of the normal binary output). It must appear before any coding (except comments), otherwise, it will be
flagged and ignored. The program is assembled as unchecksummed binary code and each physical record of output
contains nothing other than 18-bit binary storage words generated by the Assembler. The Assembler will cause the
address of the .END statement to contain a punch in channel 7, thereby allowing the output to be loaded via
hardware readin mode. If no address is specified in the .END statement, a halt (rather than a jump) will be output as
the last word.

1-25(Part II)

The following specific restrictions apply to programs assembled in .FULL mode output
.LOC

Should be used only at the beginning of the program .

.BLOCK

May be used only if no literals appear in the program, and must
immediately precede .END.
Undefined symbols may be used if no literals appear in the program.
Literals may be used only if the program has no undefined symbols.

1.4.11 Size of Program (.SIZE)
When the Assembler encounters .SIZE, it outputs, at that point, the address of the last location plus one occupied by
the object programs. This is normally the length of the object program (in octal).
Label Field

Operation Field

Address Field

User Symbol

.SIZE

Not Used

Example
Generated Code

Source Code
-..j .LOC -..j 100)

00100
00100

000105

-..j .SIZE)

00101

200103

-..j LAC -..j A)

00102

040104

-..jDAC -j B)

00103

000000

A -jO)

00104

000000

B -..j 0)

000000

-..j .END)

1.5 OPERATING PROCEDURES

1.5.1 Loading Procedures
The loading procedure depends on whether the Teletype paper tape reader or the high-speed paper tape reader is used
as the loading device.

1-26(Part II)

1.5.1.1 Teletype Reader as Loading Device - Place paper tape of the hardware readin low-speed binary loader (see
Appendix D) into the teletype reader. Engage the start switch on the reader. Enter 7700 into the ADDRESS switches
(17700 if computer has 8K of memory). Press 1/0 RESET and then press READIN. When the computer halts
(AC=777777), disengage the start switch, place the Assembler binary tape into keyboard reader, engage the start
switch, and press START.
1.5.1.2 High-Speed Reader as Loading Device - Place paper tape of the Hardware Readin High-speed Binary Loader
(see Appendix D) into the high-speed reader. Enter 7720 into the ADDRESS switches (17720 if computer has 8K of
memory). Press 1/0 RESET and then press READIN. When computer halts (AC = 777777) place the Assembler
binary tape into high-speed reader and press START.

1.5.2 COMMAND String
After the Assembler has been loaded into memory, it will type out a sequence of messages. The user's responses to
these messages indicate the options and devices that are to be used for the current assembly.

1.5.2.l Binary Option - The first message typed is *BIN -. If no binary output is desired, the user responds with
carriage return()).
If binary output is desired, the user types L) if the binary device is the low-speed punch, or H) if the binary device
is the high-speed punch. If any other character is typed it is ignored and the message (*BIN-) is repeated.*

1.5.2.2 Listing Option - The second message typed is *LST-. If no listing is desired, the user responds with). If a
listing is desired, the user types L) to produce the listing on the teleprinter, or H) to produce the listing on the
high-speed punch. If any other character is typed, it is ignored and the message (*LST-) is repeated.*

1.5.2.3 Symbol Table Option - The third message typed is *SMB-. If no symbol table output is desired, the user
responds with). If symbol table output is desired, the user types L) to produce the symbol table on the teleprinter,
or types H) to produce the symbol table on the high-speed punch. If any other character is typed it is ignored and
the message (*SMB-) is repeated.*
1.5.2.4 Source Input Device - The last message typed is *SRC-. If the source program is to be read from the
keyboard reader, the user types L). If it is to be read from the high-speed reader, the user responds with H) . If H
or L is not specified, or if any other character is typed, it is ignored and the message (*SRC-) is repeated.* When the
carriage return is typed, the Assembler starts reading from the input device.

1.5.2.5 Error Printout - Normally, all errors that are encountered by the Assembler are output to the listing device
and to the Teletype, if it is not the listing device. If no error printout is desired on the Teletype, it can be suppressed
by typing N in addition to the normally typed character on any of the four command string messages prior to typing
the carriage return.

*Note, that the character just prior to the carriage return()) is the one accepted by the Assembler, and therefore, if it is desirea to
change it, it may be retyped before the) is typed.

l-27(Part II)

Examples

(All characters underlined are typed by the Assembler)
*BIN -H)

Binary on high-speed punch.

*LST-N)

No listing, and suppress error printout on Teletype.

*SMB - L)

Symbol printed on Teletype.

*SRC-H)

Source input from high-speed reader.

*BIN -F)

Illegal request, message repeated.

*BIN-L)

Binary output on Teletype punch.

*LST-L)

Listing on Teletype.

*SMB-L)

Symbol table printout on Teletype.

*SRC- L)

Source input from keyboard reader.

NOTE

Normally, the Assembler requires two passes to assemble any program:
however, if the binary output device and the listing device are the same
device, then three passes are required. PASS2 produces the listing and PASS3
produces the binary object code. PASS2 always produces the symbol table,
if requested. When assembly is completed, the Assembler returns to the
command string processor for additional assemblies.
1.5.3 Continuation and Termination Control
Two control characters are available to control Assembler processing: CTRL C and CTRL P. They are typed by
holding the CTRL key and striking either C or P. The Assembler echoes these characters as tC or tP.
CTRL C may be typed when it is desired to prematurely terminate assembly and return to the beginning of assembly.
The Assembler echoes tC and returns to the command string processor. CTRL C may also be used to return to the
beginning of the command string if it is desired to change the options before starting the assembly. (Note that the
keyboard reader start switch must be disengaged before typing CTRL C.)
Examples

(All characters underlined are typed by the Assembler)
TAG!

l-28(Part 11)

ooo4s:oo200 tc

A multiple definition error occurred and the user prematurely terminated the assembly by typing CTRL C.
*BIN-)
*LST-H)
*SMB-tC)
*BIN- L)
CTRL C was typed to change the binary request from no binary output to binary output on the Teletype.
Two pseudo-ops control the termination of a segment or total source input to assembly. They are .EOT and .END
(see Sections 3.1 and 3.2).
When .EOT is encountered by the Assembler it outputs EQT tP on the Teletype. It then waits for the user to load the
next tape into the tape reader and type CTRL P to continue.
When .END is encountered and another pass is required, the Assembler types END OF PASS tP. When the tape is
reloaded in the reader for the next pass, the user types CTRL P to proceed.

l.5.4 Assembly Output

l.5.4.1 Symbolic Listing - If the user requests a symbolic listing, via the command string, the Assembler will
produce an output listing on the requested output device. (Teletype or high-speed punch)
The body of the listing will be formatted as follows.
Error Flags

Location

Object Code

xxx

xxxxx

xxxxxx

Source Statement

where
Error Flags

Errors encountered by the Assembler (see Section 1.5.5)

Location

= Location assigned to the binary code.

Object Code

= The contents of the location (in octal)

NOTE

Locations and object codes assigned for literals are listed following the
program.

When 56 lines have been encountered or when a form feed is encountered, the Assembler precedes the following
output with a new page number. In the case of a form feed the Assembler also outputs three up arrows (ttt)

l-29(Part II)

preceding the new page number, to indicate that a form feed caused the new page. At the end of the assembly listing
will be an error line count indicating any errors encountered during assembly.

*al N*LST-L
*S"lb-L

>FSt<C-H

uUP1..

it.1~126;00127

uNiJ

idki l 31d

ENV Oi PASS

TP
tP
P.AuE

/THIS IS A COMPLETE SA~PLE PROGRAM LISTING.
/THE LAST COLUMN (COMMENTS> CONTAINS THE
/PARAGRAPH NU1'ibEl1 IN THIS i"IANUAL WHERE
/FULL EXPLANATIONS MAY BE FOUND.
I
.1..0C
iihiid 144

id0id310

100

.DEC

/l .4.4

ABC

100;

DEF

-3;

+75

ADDR

.DSA

TAGl
5

201/J

.OCT
001.02
;0i() 1lcJ3

l(Jl(Jl 04
11"6105

777775
000075
000112

11:100!01 iil
011HH:l21::1

000010

A:l0

.~LOCK

11.4.4

/l.4.6
/1.4.3.1
/1.3.2.2

8:20

C:A
TAGl
TAG2

.SIZE
.SIXBT

/ABCD/

• IFDEi

id!r1 l l 2
Jt.j 113

000133

JJ114

.0400012l

,JJJ l 15

k1":l0 illid 1

.Jitil16

:J0ill0f02

010203

11.4.8

.ENDC
• lfUND

11.4.1.1
/1.4.5

/l.4.7

1
lU~l17

id id l2.k2
00121

ic);(i122
0012~

,Y)

IJJ124
l{jicJl25
&i1Jl26

D.11

ldiU1~7

.ENDC

6ich:H21
00icHhHil
201ill2ii'.l
22i0010
200131
34ic1l 32

00131

200126
000100
600100

3 ERROR LINES

l-30(Part II)

DAC

~0fJl30

21U0 l ;o5
~klfil 1icM

iU.0132

J,v)p

DUPL
uUPL

.+2

11.4.7
11.3.4.l

DAC

11.3.5.3

LAC
LAC*
LAC
TAU
UNO
LAC
LAC

.END

/l.3.4.3

C100
11.3.4.5
CJ.'1P ABC>
/l.3.2.3
ABC+5

UUP1..

11.3.5.1

ABC

. /l .4.1

l.5.4.2 Symbol Table Output - After the assembly listing has been typed, the A~mbler will output a symbol table
(if requested) which lists all user defined symbols. There will be two symbol lists; the first will be an alphabetically

ordered list of the symbols and the second will be a list in numerical value order. The symbol table listing is useful in
tracing or debugging a program for which the programmer does not have a complete assembly listing (symbols defined
by labels will have a value of 5 octal digits and symbols defined by direct assignments will have a value of 6 octal
digits).
Sample Symbol Table Listing

PAGE

IOIOohH iO
ir'.l0 l tJiJ

AoC
tu)uH

..:;

.Oc:ihl4

Ji60tJ20
.U00Jlir'J

uAC

0i01G0

DEF

'CJ./J l lll2

IAGl
IAG2
UNi>

00112

JUPL

l(JIQ1~6

..l0113
tllid 130

PAGE

oHi00 l 0

d
AdC

0~.0i010
id~0fU2~

001.00

DEF

'1010~

ADUR
TAul
TAG2
DAC

0i0104
.hH 12
0.0113

.iJuPL
~Nu

00120
"0126
!00130

1.5.4.3 Object Program Output - If the user requests binary output, the normal object code produced by the
Assembler is a binary paper tape which can be loaded at run time by either of the hardware readin binary loaders (see
Appendix D). The format of the binary output is as follows:
Block Heading - (three binary words)
WORD I

Starting address to load the block body which follows.

WORD2

Number of words in the block body (two's complement).

WORD3

Checksum of block body (two's complement). It also includes Word 1
and Word 2 of the block heading.

Block Body - (n binary words)
The block body contains the binary data to be loaded under block heading control.

l-3l(Part II)

Starting Block - (two binary words)

WORD 1

Locations to start execution of program. It is distinguished from the
block heading by having bit 0 set to 1 (negative).

WORD2

Dummy word

If the value of the expression of the .END statement is equal to zero, the provided loader halts before transferring

control to the object program, thereby allowing manual intervention by the user. (See Section 1.4. l 0 for an alternate
form of binary output.)

1.5.5 Error Detection and Flagging
The Assembler examines each source statement for possible errors. The statement which contains the error will be
flagged by one or several letters in the left-hand margin of the line. The following table shows the error flags and their
meanings.
Flags

Meaning

Error in direct symbol table assignments ignored (see Section 2.5.1 ).

Memory bank error (see Section 1.3.5.3).

The statement contains a reference to a multiply-defined symbol. It is
assembled with the first value defined.

Erroneous results may have been produced. Will also occur on
undefined .END value (see Section 1.3.5.3).

Literal phase error; literal encountered in PASS2 does not equal any
literal encountered in PASSI.

An attempt is made to define a symbol which has already been defined.
The symbol retains its original value.

Error in number usage.

Operand error. Non-Memory Reference Instruction has an address
value.

Phase error; PASSI value does not equal PASS2 value of a symbol.
PASSI value will be used.

1-32(Part II)

Flag

Meaning

Questionable line

Symbol error; an illegal character was encountered and ignored.

TAG or LABEL error.
1. Unrecognizable character in TAG field.
2. A period used alone in a TAG field.
3. A TAG begins with a number.
4. Xis used as a TAG.

An undefined symbol was encounterd.

Illegal use of Index Register.
1. X occurs in a .DSA statement.

2. X occurs more than once in an expression.

3. X occurs in a TAG or OP-Code field.
In addition to flagging error lines, the Assembler, during PASS 1, will print the following conditions.

Example

Condition

Multiple definitions

ABC

Direct assignment forward references

A=B

Undefined symbols

UNDF

00100; 00125

06255

The following condition will cause assembly to be terminated prematurely in PASSI.

Message

Cause

TABLE OVERFLOW

Too many symbols and/or literals

l.5.6 Internal Operations

l.5.6. l Symbol Table Capacity - The Assembler occupies approximately 3000 10 memory locations, leaving about
1090 10 registers free for symbols and literals. (The Assembler determines the physical size of memory of the

l-33(Part II)

computer; therefore, if it is an 8K machine, 5190 1 0 registers are available for table space.) Each symbol defined by
the user requires three memory locations and each literal requires one memory location. For a 4K PDP-15/10 this
means that about 360 1 0 symbols (or 3001 0 symbols and 190 1 0 literals) may be used before overflow occurs.

1.5.6.2 Halts - Normally, the Assembler does not halt for any reason except if an unknown program interrupt
occurs. If this happens the machine halts with the status word in the AC. In order to clear the condition the user must
deposit, in the location the program counter is pointing to, the instruction to clear the flag that caused the unknown
interrupt, and then press CONTINUE.

l-34(Part II)

CHAPTER 2
COMPACT TEXT EDITOR

2.1 INTRODUCTION

The PDP-15/10 Compact Text Editor provides for the creation and/or modification of source programs and other
ASCII text material. Commands issued from the Teletype direct the Editor to bring a group of lines from the input
device to an internal buffer. The user may then, by means of additional commands, examine, delete, and change the
contents of the buffer, and insert new text at any point in the buffer. When a block of lines has been edited, it is
punched on the paper tape reader.
Editor operation codes are divided into two basic categories: control instructions and Editor commands. Control
instructions determine whether the Editor is to be used to create new ASCII material (input level) or to modify
existing text (edit level). Within the edit level there are four Editor command classes: 1/0 requests, pointer
manipulation, editing requests, and examination requests.
The Editor is most frequently used to modify PDP-15/10 source programs, but it also can be used to edit any
symbolic text. The Editor operates with either high-speed or low-speed (ASR33) paper tape devices, and occupies
approximately 2000 10 locations of core memory. Any additional memory is used for buffers.
Appendix E provides a concise summary of Editor commands. Appendix F contains a simple procedure for creating
ASCII text using the Editor. Appendix G contains examples of Editor operation, written for the user who is not
fa~iliar with the Editor, but wishes to edit ,an ASCII tape immediately. The user should at least read 2.3, Operating
Procedures, before attempting to use the Editor as described in Appendix G. A more comprehensive, annotated
example of an actual editing session is contained in Appendix H.

2.2 FUNCTIONAL DESCRIPTION

2.2.1 Input Format

The following paragraphs describe the control levels, operation code formats, and data modes for the Editor.

2-1 (Part II)

2.2.l.l Control Levels - The PDP-15/10 Compact Editor operates on one of two control levels: input or edit. The
input level is used to create new text material; on this level the Editor interprets lines from the Teletype as text to be
added to an open block. Instructions are available to conveniently change the control level. The edit level is used to
modify existing text; on this level the Editor accepts and acts upon control .words and data strings to bring in lines of
text, to change, delete, or replace the line currently in the work area, and to insert single or multiple lines after the
current line.

2.2. l. 2 Operation Code Format - The format for all Editor operations consists of the operation code, followed by a
L.-1
(space), followed by arguments where applicable. The space is a blank delimiter which is considered by the
Editor to be a part of the command itself, not part of the argument string which follows the command. Legal
abbreviations are indicated with square brackets in this manual. Certain commands (e.g., FIND and RETYPE) require
the presence of arguments. Others (DELETE, NEXT) may take explicit arguments at the option of the user. Optional
arguments are given in parentheses. For a description of the command language see 2.4.

2.2. l.3 Data Mode - The Editor can accept input from a maximum of two devices* in addition to the keyboard. The
first device normally holds previously prepared text upon which changes are to be carried out. The second, the
subsidiary input device, is usually the medium through which additional, previously prepared text is inserted in the
object text.
Data from the input device is made available for editing in block form. A user-specified portion of the input is held in
a core buffer for editing until the user requests that the contents of the buffer be added to the output text.
Commands to the Editor are performed on that portion of the text currently in the buffer. Lines may be accessed
repeatedly until the buffer is emptied by the user. The lines of text in the buffer are made available for modification
by manipulating a software pointer (see Figure 2- l ).
-

POINTER

----I

PSEUOO LINE

LINES OF ASCII TEXT

Figure 2-1. Line Buffer and Software Pointer
2.2.2 Output Format
The teleprinter is used by the Editor to echo user requests, to make responses to those requests, and to print error
messages. Edited text is punched out on either the low-speed (ASR33) or the high-speed paper tape punch (if
available and requested by the user during the initialization sequence). Edited source programs are punched in a form
that is ready to be read by the Assembler; parity is not punched; channel 8 8 is always punched (see Figure 2-2).
*The low-speed (ASR33) paper tape reader, and a high-speed paper tape reader (if available).

2-2(Part II)

, . . . . - - - - - - - - - - - - CHANNEL
CHANNEL
~------- CHANNEL

r----------

8 [ALWAYS PUNCHED)
7
6

~------

CHANNEL 5
. - - - - - - - - CHANNEL 4
TAPE-SPEED SPROCKET HOLES
, . . . . - - - - - - CHANNEL 3

r-------

,....------ CHANNEL

-........._....c::--- CHANNEL

DI REC Tl ON OF
TAPE MOVEMENT

2
I

eoooooooo

--FRAME

• 0 0 0 0 0 0 0 0
• 0 0 0 0 0 0 0 0
•00000000
• 0 0 0 0 0 0 0 0

- - FRAME
- - FRAME

n+I
n+2

eoooeoeoe

- - OCTAL
- - OCTAL

•
•

0 0 0 • 0 0 • 0
• 0 0 0 0 0 0 0

••000000•

••oooooeo
•

- - ETC.
- - OCTAL

- - OCTAL
- - OCTAL
- - OCTAL

000• NULL FRAME
0 I 5• CARRIAGE RETURN
0 I 2• LINE FEED
100•@
101• A
t 02• B

0 0 0 0 0 0 0 0

0 •

e•

HOLE POSITION
HOLES PUNCHED

Figure 2-2. PDP-15/10 ASCII Tape Format
2.3 OPERATING PROCEDURES
2.3. l Loading Procedure

Prior to loading the Editor, proceed as follows to generate leader:
Turn Teletype switch to OFF LINE.
Press punch switch ON.
Press HERE IS key several times to generate leader.
Press punch switch OFF.
Turn Teletype switch to ON LINE.
The loading procedure used depends on whether the low-speed (ASR33) or high-speed paper tape reader is used as the
loading device.

2.3. l. l Low-Speed Reader - The low-speed reader loading procedure is as follows:
Place paper tape containing low-speed, hardware readin binary loader in the Teletype reader.
Turn reader switch to ON
Set ADDRESS Switches to 7700 (17700 for 8K systems).

Press I/O RESET and READIN. The computer will halt (AC=777777).

2-3(Part II)

Tum reader switch to OFF and place binary tape of Editor in Teletype reader.
Tum reader switch to ON and press START. The Editor will be loaded into memory and will type EDITOR on the
Teletype.

2.3.1.2 High-Speed Reader - The high-speed reader loading procedure is as follows:
Place paper tape containing high-speed, hardware readin binary loader in the high-speed paper tape reader.
Set ADDRESS Switches to 7720 ( 17720 for 8K systems).
Press 1/0 RESET and READIN. The computer will halt (AC=777777).
Place binary tape of Editor in high-speed reader and press START. The Editor will be loaded into memory and will
type EDITOR on the Teletype.

2.3.1.3 Loader Halts AC = 777777

Program loaded.

AC = non-zero

Checksum error on last block loaded. Reposition tape to blank frame prior to
beginning of last block and press START to read again. To ignore error, press
CONTINUE.

2.3.2 Initialization
The Editor always begins with control at the edit level and assumes that the user wishes to modify some existing text.
When first loaded, or when restarted, the Editor types EDITOR, followed by the 1/0 initialization request sequence
(the underlined portion is typed by the Editor).
Select input device:
Select subsidiary device:
Select output device:

INTXT*)
H,L)
GETXT*)
H,L)
OUTXT*)
H,L

User specifies high-speed (H)
or low-speed (L) paper tape
device,) = carriage return.

The user's response to INTXT* initializes the Editor to handle the READ command. If it is L) ,the low-speed reader
on the Teletype is used to read ASCII text into the block buffer for editing. If the response is H), the Editor expects
to find the ASCII text on the high-speed paper tape reader. Similarly, the user's response to GETXT* initializes the
Editor to handle the GET command by selecting the high- or low-speed reader as the subsidiary device. (The user may
specify that the same device be used for both READ and GET.) The response to OUTXT* selects the output device,
initializing the Editor to handle the WRITE command on the high- or low-speed punch. To obtain a listing of the
edited text, the user must specify the ASR33 punch as the output device by responding to OUTXT* with L).

The carriage return is the signal to the Editor to process each step in the initialization. The Editor interprets only the
letter immediately preceding the carriage return to be the desired device selection (i.e., HXL) is equivalent to L) ).

2-4(Part II)

If that letter is not an H or L, the Editor assumes that a mistake has been made and repeats the initialization message.
If the user's response specifies the device that is not ON, or is not even part of the system, the Editor goes into an
indefmite loop, the first time it attempts to use that device, waiting for a response from a device that is not available.
In that event, the Editor may be restarted (press STOP, set the Address switches to 22 8 , and press START) and
reinitialized.

2.3.3 Operation
All text in the buffer is available for editing until a WRITE command is issued or until each individual line is deleted.
Using a block buffer has the advantage of rapid correction of command errors. If the user finds that he has typed the
wrong command, he can immediately correct it since the buffer has not yet been added to the output file.
If the ASR33 is the output device, the punch switch must be OFF except when actually outputting edited text (see
Paragraph 2.3.3.2). This avoids contamination of the output tape by command echoing and typing of Editor
commands. The Teletype punch must be specified as the output device to obtain a listing.

2.3.3.1 Editor Break (CTRL P) - Frequently, having made a mistake in his command string, the user may wish to
stop processing and reissue his command. When the user types the break character CTRL P (formed by depressing the
CTRL key while striking P) during command processing, the normal instruction sequence is interrupted as soon as
processing of the current line has been completed.
Control is transferred from the command processor to the edit command decoder. The line after the line that was
being processed when CTRL P was typed is left in the work area as the current line for examination or modification.
The Editor then awaits a new command from the keyboard.
The break character (echoed as b) results in program restart when the Editor is waiting for a command. On input
level, the break character results in a transfer of control to the edit level (see paragraph 2.4.1.2).

2.3.3.2 Editor Continue - If the INTXT* device is the Teletype, the Editor halts after a READ. The user must turn
the Teletype reader switch to OFF and then press CONTINUE. Similarly, if the GETXT* device is the Teletype
reader, the Editor halts after a GET to allow the user to turn the reader switch to OFF and press CONTINUE. If the
OUTXT* device is the Teletype punch, the Editor halts twice in the execution of a WRITE. After the
command-terminating carriage return, the Editor halts to allow the user to turn on the punch, thus avoiding
contamination of his output tape. When punching stops, the Editor halts and waits for the user to press CONTINUE
as a sign that the Teletype punch has been turned off, and messages may be typed safely.

2.3.3.3 Editor Recovery - If a paper tape reader out-of-tape condition occurs during the execution of a command,
the user can recover by typing CTRL R (formed by depressing the CTRL key while striking R).
2.3.3.4 Using the Erase and Kill Characters - The Editor allows the use of two keyboard characters for correction of
the line currently being typed by the user. The RUBOUT key (erase character) results in the deletion of the
immediately preceding character. The Editor echoes a backslash(\) for each RUBOUT typed. CTRL U ("kill line"
character) results in deletion of the entire line typed so far. CTRL U is formed by depressing the CTRL key while
striking the letter U. The Editor echoes an at sign (@) each time CTRL U is typed.

2.3.3.5 Editor Restart - To restart the Editor at the beginning of the initialization sequence, the user should press
STOP, set the Address switches to 228 , and press 1/0 RESET and START.

2-S(Part II)

If, during command processing (especially FIND and LOCATE), the Editor attempts to move the current-line pointer

past the end of the block buffer, it is assuming the user has made a mistake and types
END OF BUFFER REACHED BY:
followed by the command string. The user must issue a TOP request if further modifications to the current block are
required.
If the user requests SIZE ..._. n, where n is greater than the number of full-length lines that the block buffer can hold,

the Editor types
CAPACITY WARNING
The user may respond with CTRL U to proceed as though the SIZE ..._. n request were never issued, or he can type a
carriage return to ignore the warning and continue. If the user types a carriage return, the command is processed and
the Editor's buffer capacity may be exceeded.
CAPACITY WARNING is also typed when the Editor calculates that execution of a command will result in more than
SIZE or the internal parameter SYSMAX (whichever is greater) lines in the buffer. Again, the user has the option of
killing the command (CTRL U) or of processing it anyway (carriage return). The Editor calculates the number of lines
in the buffer without regard to their length. For example, with a 4K system, the Editor could easily hold 30 lines of
CAP-15 assembly language instructions without comments. Thus
>SIZE L-1 30
CAPACITY WARNING
>READ

>
would be extremely dangerous if the user intends to add extensive commands throughout his code with the APPEND
command. Since the APPEND command does not change the number of lines in the buffer, the user would receive no
additional capacity warnings unless he attempted to use INSERT, INPUT, or GET.

2.3.4 Error Recovery
Operator command errors are detected by the Editor. The message
NOT A REQUEST:
is typed, followed by the command string in error. The user should then retype his command in the correct form.

2.4 COMMAND LANGUAGE

2.4.1 Control Commands
Editor control commands consist of three commands that cause the Editor to enter the input level, and one command
to enter the edit level. (Control is initially at the edit level). These commands are described in the following paragraphs.

2-6(Part II)

2.4.1.1 Transfer from Edit to Input Level - Any one of the following three commands causes the Editor to transfer
control from the edit level to the input level.
a. Carriage return ())typed as the first character on a line.
b. The INSERT command (see Section 2.4.4.4) with no arguments. In this case the current line is added to the
output before the control level is changed.
c. The OVERLAY ( '--' n) command (see Section 2.4.4. 7). In this case, n lines (or the current line only if n is
omitted) are deleted from the buffer before the control level is changed.

2.4.1.2 Transfer from Input to Edit Level - A carriage return typed as the first character of a line when operating in
the input level causes the Editor to transfer control from input level to edit level. The user can also change the control
level by typing CTRL P (formed by depressing the CTRL key while striking P). When control has been transferred to
the edit level, the Editor awaits the next command from the Teletype. The line after the line that was being processed
when CTRL P was typed is left in the work area as the current line, ready for examination or modification.

2.4.2 Editor Commands
2.4.2.1 SIZE [S] ( L.....1 n)
- Set the total number of lines that will occupy a buffer to n. The SIZE command may
be issued at any time, and takes effect when the next group of lines is loaded into the buffer via a READ command.
The integer variable n is initially set to 20 10 for a 4K system, or 55 10 for an 8K system, and must always be set
greater than 1.

2.4.2.2 READ) - Reads sequential lines from the input device, loading them into the buffer as they are
encountered, until the number of lines in the buffer is equal to the argument specified in the SIZE request. The
pointer is set to the first line of the buffer when the operation is complete (see Figure 2-3). The READ request will
not be accepted if any lines remain in the current buffer. The buffer must have been cleared by DELETE requests or a
WRITE command.*
POINTER

-r-------,

PSEUDO LINE (BLANK!

LINES READ FROM INPUT
DEVICE AS SPECIFIED
BY SIZE REQUEST

Figure 2-3. Line Buffer After READ Command

*If the input device runs out of tape, the user may terminate the processing of this command by typing a CTRL Ron the Teletype.

2-7(Part II)

If more than one block of text is to be read, the READ command inputs the first line of the next block into an
intermediate buffer. This means that if the user changes tapes on the input device between READs, he will find the

"next line" from the previous tape at the top of the block buffer, followed by SIZE- I lines from the new tape. This
does not occur if GET is used to input the new tape into the block buffer, since GET does not use an intermediate
buffer.

NOTE
If the input device is the low-speed paper tape reader, the Editor halts at the

beginning of a READ to allow the user to turn on the reader. Also, the Editor halts at the completion of a READ to allow the user to turn off the
reader; the user should press CONTINUE to proceed.

2.4.2.3 GET [G] ( L-11 n) ) - The next n lines from the subsidiary input device are added to the buffer below the
current line. When command processing is complete, the nth line read is left in the work area as the current line. If n
is omitted, it is assumed to be I. The pointer remains at the last line read (see Figure 2-4).

2.4.2.4 RENEW) - The contents of the block buffer are written on the output device and a new block is read into
core.*

2.4.2.5 WRITE) - Punches the current contents of the block buffer on the OUTXT device, and clears the buffer.*

2.4.2.6 CLOSE) - This command must be preceded by a WRITE request. The remainder of the input file is then
written on the output device.*

2.4.3 Pointer Manipulation
The pointer is a software device that places the current line in a work area to facilitate editing. After a page of text
has been read, the pointer is positioned at the top of the block buffer to allow the user to insert text before the first
line in that block. This is done by keeping a pseudo line in the buffer which is never punched out. Thus, the first line
of the text in a block is the "second" line in the buffer. Consequently, the first line of the block of text in the buffer
is shifted one position to the right when examined with the PRINT command. This "space" is not printed. The
following paragraphs describe commands that enable the user to manipulate the pointer.

2.4.3. l TOP [T]) - This command moves the pointer to the beginning of the edit block buffer. The first line of the
buffer becomes the current line (see Figure 2-5).

*If the output device is the low-speed punch, the Editor Continue feature provides protection against output contamination (see paragraph 2.3.3.2).

2-8(Part II)

POINT ER
(BEFORE ·~

POINT ER
(AFTER ~

PSUEOO LINE

CURRENT LINE
BEFORE GET

}

11- 1 LINES

-71 th LINE (CURRENT
LINE AFTER GET)

GET

LAST LINE
IN BUFFER

1/ ISSUEO WITH POINTER WITHIN BUFFER

POINTER
•-----(BEFORE GETl""1
-PSEUDO LINE

PSEUDO LINE

I}~-". .

POINTER
_
(AFTER GE~~------4

-7/lhLINE

--------i! }~~:·:.:: "

POINTER

(BEFORE GETf"':t-

I
I

------- -

LAST LINE IN BUFFER

I
:I
th
GEfl-=-L-------- -71 LINE

POINTER _ I(AFTER

GET 7/ ISSUED WITH POINTER AT BOTTOM OF BUFFER

GET 7/ ISSUEO WITH POINTER AT TOP OF BUFFER

NOTE: If GET n attempts to exceed the maximum line buffer size, CAPACITY WARNING will be typed. The user
may type a carriage return () ) to ignore the warning and continue, or he may type CTRL U to proceed as though the
command were never issued.

Figure 2-4. Line Buffer After GET Command

NOTE

If the input device is the low-speed paper tape reader, the Editor halts at the
beginning of a GET to allow the user to turn on the reader. Also, the Editor
halts at the completion of a GET to allow the user to tum off the reader; the
user should press CONTINUE to proceed.

2-9(Part II)

POINTE~-.--------,

(AFTERT0°Pl;--71.--------1

PSEUDO LINE
FIRST LINE OF TEXT

POINTER
(BEFORE TO~-,...-----~

POSSIBLE POSITION OF
CURRENT LINE BEFORE
TOP COMMAND

Figure 2-5. Line Buffer After TOP Command
2.4.3.2 NEXT [NJ ( ..._. n)) - The pointer is moved past the next n lines, beginning with the line currently in the
work area. Line N+ I is brought into the work area for modification (see Figure 2-6). If omitted, n is assumed to be 1.
If the command results in the pointer moving past the last line of buffer, the error message
END OF BUFFER REACHED BY:
NEXTn
is printed and the pointer is moved to the top of the buffer.
-

PSEUDO LINE
FI RS T LINE OF TEXT

P 0 I NT E_FL_) __ ~
( BE F 0 RE NE X T-,--------l--------4

CURRENT LINE

POINTE_~
(AFTER NEXT!
--J....-------4

}~-· "'"
LINE 7)+1 (NEW "CURRENT" LINE!

Figure 2-6. Line Buffer After NEXT Command

2.4.3.3 FIND [F] ..._. string) - This command searches the buffer for the next line that begins with the character
group "string". The search begins with the line following the current line. If the search is successful, the line beginning
with "string" is brought into the work area (see Figure 2-7). If the search is unsuccessful, the END OF BUFFER
message is printed and the pointer is moved to the top of the buffer. "String" may contain any number of characters.

2.4.3.4 LOCATE (L] ..._. string) - This command searches the buffer for the next occurrence of a line that
contains the character group "string." The search begins with the line following the current line. If the search is
successful, the line containing "string" is brought into the work area (see Figure 2-8). If the search is unsuccessful, the
END-OF-BUFFER message is printed and the pointer is moved to the top of the buffer. "String" may contain any
number of characters.

2-IO(Part II)

POINTER
(BEFORE FINO~-.,__ _ _ _ __

POINTE_~j----------t

(AFTER FIND)

PSEUDO

LI NE

CURRENT LINE (BEFORE
FIND IS EXECUTED l

CURRENT LINE (AFTER
FIND IS EXECUTED>
OR LINE BEGINNING WITH
"STRING"

FIND STRING WHEN "STRING" IS FOUND

!AFTERPOINTE
FIND'1•.!!.........:,,
-1_ _ _ _ _ __.

POINTE,.':!R_~-------t

(BEFORE

FIN~o--------

PSEUDO LINE (CURRENT LINE
AFTER FIND IS EXECUTED>

CURRENT LINE (BEFORE
FIND IS EXECUTED)

FIND STRING WHEN "STRING" IS NOT FOUND

Figure 2-7. Line Buffer After FIND Command

2.4.3.S BOTTOM [B]) - This command moves the pointer to the beginning of the last line in the buffer. The last
line is typed on the Teletype (see Figure 2-9).

2.4.3.6 SEARCH L...J string - The entire input tape is searched for the next occurrence of a line beginning with the
character group "string." This command must be preceded by a WRITE to clear the block buffer. If the search is
successful, the line beginning with "string" is brought into the work area with the remainder of the buffer left empty
for inputting new text or large inserts for the subsidiary input device using the GET command.

NOTE

If the output device (OUTXT*) is the Teletype punch (L), the Editor will
halt at the beginning of a SEARCH to allow the user to turn on the punch.
The user should press CONTINUE to proceed with the SEARCH. When the
line beginning with "string" has been found, the Editor will again halt to
allow the user to tum off the punch. Again, the user should press
CONTINUE to proceed.
1,,,.;;l

2-11 (Part II)

......-

PSEUDO LINE

POINTE
(BEFORE LOCAT

CURRENT LINE
!BEFORE LOCATE IS EXECUTED)

POINTE
(AFTER LOCAT

rr~

LOCATE

STRING WHEN"STRING"IS FOUND

POINTER
(AFTER LOCAT~1---------i

POINTE~t---------t

(BEFORE LOCATE)

CURRENT LINE
(AFTER LOCATE IS EXECUTED)
OR LINE CONTAINING"STRING"

PSEUDO LINE
(CURRENT LINE AFTER LOCATE
is EXECUTED)

CURRENT LINE
(BEFORE LOCATE IS EXECUTED!

LOCATE STRING WHEN "STRING" IS NOT FOUND

Figure 2-8. Line Buffer After LOCATE Command

2.4.4 Editing Requests
The following paragraphs describe commands that enable the user to accomplish his actual editing task.

2.4.4.1 RETYPE [R] "'-' line) - The character string "line" replaces tne current line. The new line is left in the
work area and may be modified.
2.4.4.2 APPEND [A] "'-' string) - "String" is added to the current line following the last character preceding the
terminating carriage return. Thus, to add a comment to the current line
JMSGETNUM)
the command might be
A ......,

2- l 2(Part II)

-I /Get decimal argument. )

NOTE: -I indicates a tab

POINT1.~E~R~...t-------1

!BEFORE BOT~~-------1

PSEUDO LINE
FIRST LINE OF TEXT

-CURRENT LINE
(BEFORE BOTTOM IS EXECUTED)

POINTER
(AFTER

LINE IN BUFFER
eon·~oM;;,l:-i~=====j -LAST
(CURRENT LINE AFTER BOTTOM
IS EXECUTEOI

Figure 2-9. Line Buffer After BOTTOM Command
The new current line would be:
JMS GETNUM -1 /Get decimal argument.)
If "string" is absent, the current line is unchanged. When the current line is changed, the new line is left in the work

area and may be modified.

2.4.4.3 CHANGE [CJ L-1 qstring1qstring2q) - In the current line, the first character group (stringl) which
matches that occurring between the first pair of delimiting characters (q's in this case) is replaced by the character
group (string2) appearing between the second pair of delimiting characters. The delimiting characters are chosen by
the user and can be any character (including blank) that does not appear in either of the character strings. Both
"stringl" and "string2" may contain any number of characters, including zero. If VERIFY is ON (see paragraph
2.4.5.2), the program will print the new current line on the Teletype when the request change has been accomplished.
The new line is left in the work area and may be modified.

Examples

Current Line:

NXTLIN

a. In the comment, spell PRINT properly,
REQUEST:
NEWLINE:

CHANGF.a....JRN/RIN/)
NXTLIN

JMSTYPOUT

/PRNT THE LINE.

JMSTYPOUT

/PRINT THE LINE.

JMSTYPOUT

/PRINT TifE LINE.

JMP*TYPOUT

/PRINT THE LINE.

b. Make the "JMS" a "JMP".

REQUEST:
NEWLINE:

CHANGE '-' XSXP*X)
NXTLIN

2-13(Part II)

c. Delete the "T" in the tag,
REQUEST:
NEWLINE:

C '-'/Tl/)

JMP* TYPOUT

NXLIN

/PRINT THE LINE.

2.4.4.4 INSERT [I) '-' line - The character string "line" is inserted below the current line. The pointer is stepped
to place "line" in the work area and "line" becomes the new current line (see Figure 2-10). The program remains at
the edit level when the request processing is completed.

(BEFORE INS ERTi
POINT

POINT·~
(AFTER INSE RTl

PSEUDO LINE

CURRENT LINE (BEFORE INSERT COMMAND)
INSERTED LINE OR CURRENT LINE AFTER
INSERT COMMAND.

Figure 2-10. Line Buffer After INSERT Command
2.4.4.5 GET [G] ( ..._. n)) -This command adds n lines from subsidiary input device to buffer below the current
line (see Section 2.4.2.3). If the subsidiary device runs out of tape, the user may terminate processing of this
command by typing CTRL R on the Teletype.

2.4.4.6 DELETE [DJ ( '-' n)) - The next n lines, including the current line, are deleted from the buffer. The line
following the last line deleted becomes the current line (see Figure 2-11 ). If n is omitted, only the current line is
deleted. If n is large enough to cause the pointer to pass the end of the buffer, the END-OF-BUFFER message is
printed and the pointer is moved to the top of the buffer.

2.4.4.7 OVERLAY [O] ( ..._. n)) - Starting with the current line, n lines (or the current line only, if n is omitted)
are deleted from the buffer and the control level is changed to input level. When control returns to edit level, the
pointer will be positioned at the last line typed by the user.

2.4.5 Examination Requests
The following paragraphs describe commands that allow the user to examine text within the block buffer.

2.4.5. I PRINT [P) ( 1-1 n)) - The next n lines from the buffer, including the current line, are printed on the
Teletype. The pointer is left at the last line printed (see Figure 2-12); n is assumed to be one if omitted.
If, as a results of the request, the pointer moves past the last line of the buffer, the error message

END OF BUFFER REACHED BY:
PRINT n
2- I 4(Part II)

is printed and the pointer is moved to the top of the buffer (see Figure 2-12).

-PSEUDO LINE

PQINTER
J------ ·--(BEFORE DEl.~1------~

I
I
I

POINTE~

CURRENT LINE (BEFORE DELETE)

}~-, ""' """'
LINE

71+ 1 (NEW"CORRECT"UNE)

""" '""" !. .________,
DELETE

71 THAT OOES NOT REACH END Of BUFFER

PSEUDO LI NE
(CURRENT LINE AFTER DELETE)

CURRENT LINE (BEFORE DELETE)

I
I
I

LINES DELETED

DELETE

I
I

L------~
1/ THAT REACHES END OF BUFFER

Figure 2-11. Line Buffer After DELETE Command

O~~

2.4.5.2 VERIFY [V] L-1 (
))- Set Editor response according to the parameter. When VERIFY is ON, text
lines are printed in response to certain .editing requests, as follows:
a. The line brought into the work area as a result of a FIND or LOCATE request is printed.
b. The last line of the buffer, brought in by the BOTTOM request, is printed.
c. The new line resulting from a CHANGE request is printed.

When VERIFY is OFF, only error messages are printed. After the Editor is loaded initially, VERIFY is ON. The
verify request without arguments is equivalent to requesting VERIFY ON.

2- I 5(Part II)

-PSEUOO LINE

POINTE~Rm:iot:======l

CURRENT LINE (BEFORE PRINT)

(BEFORE PRlNTl

}

POINTER
(AFTER PRl~l-------t

11 THAT

DOES

11- 2 LINES PRINTED

_ 71th LINE PRINTED OR CURRENT
LINE AFTER

NOT REACH END OF

BUFFER

POINTER
(AFTER PRl~-1----------4

_PSEUDO LINE
(CURRENT LINE AFTER PRINT)

POINTER
(BEFORE P~-....-------1

CURRENT LINE (BEFORE P"INTl

LINES PRINTED*

*
PRINT

END OF BUFFER MESSAGE TYPED

11 THAT REACHES END OF BUFFER
Figure 2-12. Line Buffer After PRINT Command

O~~

'°"""

2.4.5 .3 BRIEF._. (
))- Set Edit°' verification acoonling to the parameter. BRIEF ON
in the ahr..viated printing of the current line when responding to edit requests. An attempt is made to print only the tag, operation code, and address fields of lines brought into the word area as a result of the FIND, LOCATE, and BOTTOM
requests. The printing of the new line resulting from a CHANGE request is terminated at the last newly inserted
character. BRIEF is initially set to OFF. The setting of this indicator is of no consequence when VERIFY is OFF.
The brief request without arguments is equivalent to BRIEF ON.

2- l 6(Part II)

CHAPTER 3
OCTAL DEBUGGING TECHNIQUE

3.1 INTRODUCTION
Octal Debugging Technique (ODT) is a debugging aid that allows the user to conduct an interactive, on-line debugging
session using octal numbers and Teletype commands. The program, which is self-contained and completely
independent, is designed to run on a basic 4K or 8K PDP-15/10 with an ASR33 and will operate with PI or API
enabled. ODT is written in CAP-15 Assembly Language and may be assembled along with the user programs with
which it is to run. All symbols that are internal to ODT begin with a percent sign (%). Thus, if the user avoids
beginning his own symbols with this character, multiple definition of symbols will not occur when ODT is assembled
with the user program. However, ODT would normally be assembled separately from the user's programs and then
loaded into memory only when desired. Standard versions which are delivered in object program form may be used
for the debugging of any programs as long as enough memory is available to load ODT at the proper locations.

Using ODT, the programmer can conduct a debugging session, and when errors are found, correct them on-line and
execute the program immediately to test the correction. Thus, ODT can be used to compose a program on-line and
check it out as composition progresses. Manual operation of console controls is not required to operate ODT; all
functions are initiated by typing commands on the Teletype .. The source coding for ODT is so designed that many
features can be removed by defining parameters at assembly time. This will be of prime interest to users who are
checking out large programs and wish to conserve core storage.
Appendix I provides a convenient listing of the Teletype code in octal form. Appendix J is a concise listing of ODT
commands.

3.2 GENERAL DESCRIPTION
All input to ODT is initiated through the console Teletype keyboard. In general, the user types commands to ODT
which control the following operations:
Starting the user program at any point.

3-l(Part II)

Stopping and subsequently continuing the user program at selected points called breakpoints.
Examining and/or modifying the AC and Link at a breakpoint.
Examining and/or modifying any memory locations (registers).
Searching user defined areas for registers with specified bit configurations.
Punching out areas of memory to be loaded later for more debugging and/or execution.
Initializing buffer areas to any desired value.
Selection of hardware options such as a high-speed or low-speed paper tape punch.
There are two outputs from ODT: teleprinter and paper tape punch (either high-speed punch or ASR33 punch). The
teleprinter is used by ODT to type all information concerning register contents and error indications. This output is
very useful since it contains the user's commands as well as ODT responses, which can be analyzed off-line if
necessary. The paper tape punch is used to save or dump the user program, or portions of it, as the user desires. The
tapes generated when dumping a program are fully checksummed and can be loaded by the PDP-15 /I 0 Loader. The
format of this tape is shown in Figure 3-1. Each memory word consists of three data frames of six bits each. The
program start address is distinguished from the other load addresses by having bit 0 set to I.

- -Leader ($F command)
2 feet

Figure 3-1. PDP-15/10 Loader Format
3.2. l Operational Organization

ODT can be loaded and used in four basic ways:
It can be loaded immediately following the user program, with control being given to ODT after it is loaded.
It can be loaded after the user program has been running.
It can be loaded before the user program(s) and initiated by manual control after the user program(s) has been
running.
It can be loaded by itself and used as a stand-alone program.

3-2(Part II)

Block Start Address

# of words in block (2's complement)
Checksum of block including the previous
two words (2's complement)

Data Block
Block Body (any length)

(k 1; k 2 $0 command)

Block Spacer (10 Frames)
More Data Blocks

Last Data Block

Program Start Address
Dummy word

Terminal Block (k$T command)
Trailer

.,,...__--__,..,,._
Figure 3-1. PDP-15/10 Loader Format (cont)

3-3(Part II)

In any case, ODT is initiated by starting the processor at location %0DT. The distributed version is loaded and
initiated at location %0DT (6000 for 4K and 16000 for 8K). However, the user may assemble versions of ODT which
are loaded at any location in memory. Also, by choice of .END statements, ODT may be initiated automatically by
the loader or may require manual initiation at location %0DT. Once initiated, the user has control over running and
stopping his program with keyboard commands. If ODT breakpoints are requested, the following modifications are
made to the user program:

Each instruction that the user designates as a breakpoint is replaced with JMS* AUTOX. This instruction gives
control to ODT when the breakpoint is encountered, and is removed when a breakpoint is executed so that the user
will see his own instructions if he examines these locations or dumps memory.
An auto-index register, the address of which may be examined and/or modified by using the $V command, is used
by ODT and may not be modified or referenced by the user program.

If the PI (program interrupt) is used, ODT must, during breakpoints, intercept all TIY interrupts and pass all other
to the user. Therefore the instruction at location 1 is replaced with the instruction JMP %INT. Location I is
restored when a dump is performed but it is not restored during other ODT operations.

3.2.2 Functional Organization
The internal organization of ODT is almost totally modularized into independent subroutines. In general, the internal
structure consists of major functions, as follows:
Initialization
TTY Character Input/Output
TIY Message Output
PI Intercept
Command Decode
Command Execution

3.2.2.1 Initialization - This function is performed when ODT is started at location %0DT. The tasks performed
include disabling the entire system (similar to an 1/0 reset), resetting miscellaneous registers internal to ODT, clearing
all breakpoint requests, clearing hardware and software flags, and initiating the command decode function.

3.2.2.2 Teletype Character Input/Output - All teletype communication is performed a character at a time; that is,
the input and output is not buffered. This means that the user must not type commands unless ODT is waiting for
one. If characters are typed while a requested operation is in progress, they will not be seen by ODT and will not be
echoed on the teleprinter.

3-4(Part II)

3.2.2.3 Teletype Message Output - This function is performed by repeated use of the teletype character output
routine. The functions performed include the typing of addresses, register contents, operational messages and error
messages.

3.2.2.4 Pl Interrupt - If the PI is enabled, the Teletype will interrupt when it is ready for service. Therefore, during a
breakpoint, ODT must be able to handle the Teletype 1/0 interrupts. This function is performed with the help of the
PI Intercept (%INT) routine. This routine is entered, instead of the users interrupt program, on every program
interrupt which occurs while ODT has control. If the interrupt is not from the Teletype, control continues to the user
trap program.

3.2.2.5 Command Decoder- The functions performed by the command decoder include:
Reading the keyboard.
Assembling and saving all octal numbers typed by the user.
Detecting the command character.
Allocating control to the proper command execution subroutines.
Detection of illegal numbers and commands.

3.2.2.6 Command Execution - In general, each command has a separate entry point into the command execution
routines. These routines are entered from the command decoder. They exit either to the user's program ($G, SC) or
back to the command decoder to interpret the next command.

3.3 COMMAND LANGUAGE

The following conventions are used throughout this section and the remainder of this chapter:
$represents either the ALT MODE key or the dollar sign (S) key($ is echoed).
k represents an octal number of six or less digits. When used to specify an address. k is five or less
octal digits.
n represents a single octal digit.
) represents carriage return

+ represents line feed.

t (up arrow) is formed by depressing the shift key and typing N.
Any underlined text in the examples refers to that which is typed by ODT.

3-S(Part II)

Aindicates a blank or space.
Priority levels are specified as follows:
Highest
0
Hardware API levels.

Lowest

2
3
4
5
6
7
8

Optional
Software API levels.

PI level
Program level

When a command is typed by the user, ODT responds in one of the following ways:
If the command does not require a typed response, ODT will respond with a carriage return and line feed () .j..) to
indicate that the command has been accepted. On this type of command, the user should not type anything on the
keyboard until after ODT has typed the) .j... Anything typed after the command and before the) .j.. will be ignored;
in fact, it won't be echoed on the teleprinter.

On commands that request the contents of a register to be displayed, ODT performs the typed response but does
not follow with the,) .j..; this allows modification and/or other commands to be typed on the same line.
On commands that cause multiple words, messages, or multiple lines to be typed, ODT terminates the output
with)H.
Illegal or incorrect commands are ignored, and ODT responds with ? ) .j...
3.3. l Register Examination and Modification
The following paragraphs describe all ODT register examination and modification commands.

3.3. l. I Open a Register (k/ j) - The command k/ causes ODT to open register k. This means that ODT fetches the
contents of register k, types the contents on the teleprinter, and then leaves the register open for modification. Once a
register is opened, the user may close it, or he may type a new octal value for the register and then close it. ODT
allows only one register to be open at any particular time. Thus, another register-opening command will close the
open register.
The / command causes the last referenced register to be opened. This allows the user to open and close register N,
perform some non-register-opening commands (possibly including the execution of his program), and then type / to
have register N opened again.
Example

400/002066) .!_
39?).j..
3765/01264317/001326)±
/001326)1
0/001472

3-6(Part II)

Open and close register 400
9 is illegal
Open and close register 3765 and then
open register 17
Re-examine register 17
Open register 0

3.3.1.2 Close Register () ) - Typing a carriage return ()) closes any open register. If a register is open and the user
types k) the value k will be stored in the open register and the register will be closed.
Example

300/000206) .!.
300/000206 3333) 1.
/003333 ).:!:..
0/001472 279?)-l-

If a register is open, typing another register-opening command will close it.
Example

300/003333 426/ 010000

Closes 300 and opens 426

3.3.1.3 Close Register and Open Next (.j..) - The line feed (.j..) works the same as the carriage return()) except that
after modifying (optional) and closing the open register, the next sequential register is opened and its location and
contents are printed. Note that on the 4K machine, location 0 follows 7777.
Example

300/003333 126 .j..)
00301/700301

Change contents of location 300 to 1268
and open 301.

Note that ODT types addresses as five digits and register contents as six digits with no leading zero suppression.
3.3.1.4 Close Register and Open Previous (t) - The up arrow (t) acts like the line feed (.j..) except that the register
previous to the currently open register is opened.
Example

30/700314t ).J_
0027 /700002

Open register 30. The t causes
ODT to open register 27.

3.3.l.5 Close Register and Interpret Contents as Address <~l - The left arrow (~)closes (optionally modifies) the
currently open register. It then interprets the contents of the register as a memory referencing instruction. ODT then
opens the register referenced by the address portion of that instruction. The indirect bit is not tested so that if the
register contained the instruction LAC* 200 the following would occur:

326/22020~)

00200/nnnnnn-

Ignore the indirect bit and
open register 200.

3.3. l .6 Open AC and Link Register ($A) - Typing SA at any time opens and displays the register holding the current
AC value. The value may be modified if desired.

3-7(Part II)

Typing ~ opens the next register which holds the Link status in bit 17. This may also be modified.

3.3.1.7 Display Pl and API Status ($J) - Typing $J causes ODT to sample and print the current Pl and API status
registers. Since both of these may be running, the values may be constantly changing.
Example

$1.J..±.
Pl/nnnnnn) .j,
APl/nnnnnn )H
The PI is sampled with an IORS instruction and the API is sampled with the RPL instruction. Definitions of the bits
in these registers may be found in the PDP-15 Reference Manual.

3.3.2 Execute Instruction k (k$X)
The instruction k will be executed. The saved AC and Link are restored before execution and saved again after. As
long ask is not a CAL, JMP or JMS, control will remain in ODT. On a JMS or CAL, control returns to ODT on the
instruction following the JMS or CAL, even if control is transferred to a routine that normally contains a breakpoint.
The JMP A instruction will be executed as though the command k$G were given, where k is address A. All other
instructions, except for the DBR instruction and the multiply /divide instructions of the EAE class, may be executed
in this manner. Note that a skip instruction that actually skips causes the current breakpoint return address to be
incremented by one.
ODT responds with ) .j, as soon as the $X command is accepted. When ODT is ready to accept another command, it
types another carriage return - line feed sequence.
Example

140012$X).j,

)-1-

User executes a DZM 12.

3.3.3 Setting User Start Address

3.3.3.1 Open User Start Address ($Z) - Typing $Z opens a register internal to ODT and types its contents. The user
can then type the start address of his program followed by . Then the command $G (see Section 3.3.3.2) will
transfer control to the address specified. (The initial content of the $Z register is unspecified.)
Example

$ZOOOOOO 200) .j,

Set start address equal to 000200.

3.3.3.2 Transfer to User Start Address or Address k ($G, k$G) - This command is used to start the user program at
any location. The command k$G starts the program at location k. The command $G starts the program as specified
by the $Z register.

3-8(Part II)

The normal use of $G is to initially start the user program. If breakpoints have been specified, they are set before
control goes to the user program. The AC and link are unspecified unless the user has used the $A command to set
particular values. The AC and link may also be specified by executing AC modifying instructions with the $X
command. The status of the PI and API are not altered by a $G command.
Another use of $G is to restart the user program at any point after a breakpoint has been encountered. That is,
instead of continuing after the breakpoint with a $C, the user can start at some other location by typing the $G
command. The user's AC and link are restored for a $C command. Again, the Pl and API are not disturbed.

3.3.4 Using Breakpoints
ODT allows the user to specify up to four breakpoints at selected locations in his program (except location l). The
breakpoint facility of ODT provides a means of suspending the program operation at any desired point and then
examining the status of the program through the use of the other ODT commands.

3.3.4. l Setting Breakpoints (k$B, k$nB) - To set a breakpoint at location k, the user types k$nB, where n is I. 2, 3,
or 4. If breakpoint n is already set at some location, it is moved to location k. The command k$B will assume
breakpoint I.

3.3.4.2 Removing Breakpoints ($nB, $8) - To remove breakpoint n, the user can type $nB where n is I, 2, 3 or 4.
To remove all breakpoints, the user can type $B.

NOTE

The above described commands do not immediately alter the breakpoints.
The mechanism is set so that the desired changes occur at the next $G or $C
command.

Example
$B)l
300$4B) .!.
1760$B) J:..

Remove all breakpoints
Set breakpoint 4 at location 300.
Set breakpoint l at location 1760.

3.3.4.3 Continue from Breakpoint ($C, k$C) - The $C command causes the user's instruction at the breakpoint to
be executed and control to be returned to the user program. The k$C command does the same thing except that a
loop count is set so that the break does not occur until the kth time it is encountered. This command may be given
only while a breakpoint is in progress. Note that even though the loop count may not be satisfied, ODT is still entered
and has control for about 200 µs on every pass through the breakpoint. During this time, re-entrancy may occur (see
paragraph 3.3.4.9). Note that $C is the same as l $C.

3.3.4.4 Kill Pl and API During Breaks ($K) - Typing $K causes the Pl and API to be disabled during all breakpoints.
Their status is saved and restored at the $C command. The $K command stays in effect until the $U command is
given.

3-9(Part II)

3.3.4.S Allow Pl and API During Breaks ($U) - Typing $U re-enables the PI and API after having been disabled by
the $K command. This is the nonnal operating mode for ODT.

3.3.4.6 Vary Autoindex ($V) - The operation of breakpoints requires the use of one autoindex register. The user
may not use this register if ODT is used. Normally, ODT uses location 17 but the user may change this with the $V
command. Typing $V opens a register internal to ODT, which holds the address of the autoindex register. The user
may alter this value to any number in the range 10 through 17. This alteration, if perfonned, should be accomplished
before the first $G command is given to ODT. The user should also note that the autoindex register is used by ODT
even if breakpoints are not specified. Starting ODT at location %0DT will reset the autoindex location to 17.
Restarting ODT at %REST will leave the autoindex specification unaltered.

3.3.4.7 Open Re-entrant PC List ($Y) - This command, which is really only meaningful after a re-entrancy error,
opens the first breakpoint return address. Typing .). opens the succeeding PC values. The list is terminated by an IOT
class instruction (op code 70). Note that the values in this list are the addresses+ 1 of the active breakpoints.
3.3.4.8 Breakpoint Operation - When, during execution, the user program encounters the location of the breakpoint,
control is transferred immediately to ODT. ODT saves the user program AC and link. The status of the teleprinter flag
is also saved. From this point on, until a $G or $C command is given, the user does not have control of the Teletype.
ODT then replaces all break instructions and types the breakpoint number and location as follows:

).J. $Bn .6k).).
where n is the breakpoint number and k is the address of the breakpoint.
ODT is now in control and is waiting for the user to type commands. The user can examine and change any registers.
The current AC and link may be examined and modified by using the $A command. The user should note that the
instruction at the breakpoint has not been executed. It will not be executed until a $C command is given. The user
can remove and/or move any breakpoints at this time.
At a breakpoint, while ODT has control, the status of the PI and API is not altered (except for short periods of time
as described below). Thus, a breakpoint at priority n causes ODT to be entered and executed at that priority. This
means all processing at that priority level and below will stop.
However, all processing at priority levels above the level of the breakpoint will continue. It may be that the user will
want all processing to stop at a breakpoint. This may be accomplished in two ways. The user may set the breakpoint
in a section of code that is executed at the highest priority. If this cannot be done, the user can type the command $K
to "kill" PI and API processing at the next breakpoint. The PI and API status is saved and then restored on the $C
command. The $K command can be typed at any time that ODT has control; it stays in effect until the $U command
is typed and vice versa.

3.3.4.9 Breakpoint Restrictions - The following restrictions apply when using breakpoints:
a.

Breakpoints may not be placed at the following types of instructions:
( I) Instructions that are modified by the user program.

3-lO(Part II)

(2) Instructions that are executed out of line by an XCT.
(3) Instructions or data that are not actually executed, such as instructions used as arguments following a

JMS.
(4) An instruction that uses indirect addressing and is the first indirect following a DBR instruction.

Breakpoints on other indirect instructions are not restricted.
(5) An instruction in an area operating in the extend mode.
(6) On any of the exit sequence instructions of a PI or API routine. For example, a breakpoint must not be

set on any of the instructions .
ION

DBR
JMP*

in the exit of a Pl routine.
b. If a breakpoint is placed on a CAL instruction, the breakpoint will occur as normal. However, when the $C
command is given, the breakpoint will be removed before control returns to the user. Thus, a breakpoint on a CAL
instruction may only be executed once. A breakpoint may be set on any of the other memory referencing instructions
(op codes 04 through 60), any operate instruction (op code 74), and the EAE instructions (op code 64). The IOT
instructions are discussed in the next paragraph.
c. There is one IOT class instruction that cannot be at a loc~tion where a breakpoint is set; this is the DBR
instruction. Generally, this instruction requires that it be followed by a JMP* referencing the location holding the PC,
link and extend mode status. This will not occur if the DBR instruction is at the breakpoint, since many ODT
instructions will be executed after the DBR is executed and before the user regains control.
If the $K command is in effect, breakpoints should not be set at the IOT instructions listed below. If a breakpoint is
set at these instructions, the results will be as indicated.

IORS

The PI bit will always indicate that the Pl is disabled.

IOF

No operation

ION

Could cause ODT to fail if any breakpoints are defined within an
interrupt routine.

SPI

SPI will operate correctly if the API was off when the break was
executed; however, if the API was enabled, then ODT raises the priority
to level 0 (highest priority). This may cause SPI not to skip when the
user expects it to do so.

ISA

If the API was disabled when the break was executed and if the ISA is
enabling the API, ODT may fail if any breakpoints are defined within
the API level routines. If the API was enabled when the break was
executed, then the API will be enabled and active at level 0 when the
ISA is executed. ODT will then DBK. Thus, if the user's ISA has

3-1 l(Part II)

requested an interrupt at levels 4-7, the ISA will work. If the ISA has
turned off the API, the ISA will operate correctly. But if the ISA has
tried to raise the priority to levels 1-7, the ISA will operate as a no-op.
RPL

If the API was off when the break was executed, RPL will work.

Otherwise, RPL will read a word into the AC which indicates that the
API is enabled and level 0 is active.
d.

There are several timing considerations for setting breakpoint instructions.
( 1) If the user is operating with no PI or API, then setting a breakpoint merely stops his program. Hence,
all 1/0 in progress will be terminated until the $C command is given.
If a XX$C command is given, the breakpoint is not executed again until the XXth occurence. The user

must realize that on every pass over the breakpoint, ODT is entered, the user instruction is executed by
ODT, and then control is returned to the user program. Therefore, the instruction at the breakpoint,
which normally takes from 1.5 to 12 µs to execute, will actually take about 200 µs to execute.
(2) If the use is running with PI and API enabled, then several additional timing restrictions must be
considered.
(a.) Assume the user inserts breaks only at the program level (that is, at non-PI, API levels). When a
breakpoint is encountered, independent of the status of the iteration loop count, the PI is
disabled for short periods of less than 80 µs and the API is temporarily raised to level 0 (highest
priority) for periods of less than 50 µs.
(b.) Assume that the user inserts multiple breakpoints on the same priority level. When the breakpoint
is encountered at level n, ODT will operate at that level (except for the short intervals, described
in paragraph (2) (a) above where ODT operates at level 0). Therefore as long as the break is in
progress, which could be several minutes, all interrupts on that priority and below will be
inhibited.
(c.) Several possibilities exist if the user places breakpoints on more than one level. In general, ODT is
designed such that only one breakpoint may be in progress at any given time. However, it is
possible that a breakpoint is executed at level N and then, before ODT can remove the other
breakpoints, another breakpoint is encountered at a level M, where M has a higher priority than
level N. ODT always detects this situation, but there is no way to recover to the extent that the
user can continue his program. However, ODT will do the following:
Disable Pl, API
Print the message ODT REENTERED
Halt
The user may press CONTINUE to restart ODT or he may manually restart ODT at location
%REST ( 16011 on the distributed version). Thus, the user may examine, with ODT commands,
the status of the program. In particular, he may use the $Y (see below) command to open up the
first location of list of PC values of all breakpoints which are now active. By stepping down the
list, he may determine which breaks occurred and in which order. The list may have from two to
four entries, but is terminated with an IOT class instruction (op code 70). At first glance, it may
appear that if a break is in progress, then the next break which interrupts the currently active
break will cause ODT to terminate. However, the following may occur:

3- l 2(Part II)

A breakpoint instruction is executed at some level. This saves the PC (return address which is
the breakpoint address + 1).
A higher priority level becomes active and takes control away from ODT, possibly before
ODT can even save the AC. Of course, this may occur at any time before ODT removes the
other breakpoints.
Another break is executed on the higher level. This instruction will save the PC in the
location following the PC of the interrupted breakpoint. If this breakpoint routine (in ODT)
executes one more instruction, then ODT will detect the re-entrancy and the re-entrant error
process will be initiated. However, this second breakpoint may be interrupted by a higher
level request before it executes its first instruction (following the breakpoint instruction). Of
course, the higher level routine may contain another breakpoint; etc.
Jn any case, the following can be guaranteed:
In no case will the PC value of an executed breakpoint be lost.
Since there are a maximum of four breakpoints, either some breakpoint will get far enough
to detect the re-entrancy or the breakpoint at the highest level will be executed. This means
all other breaks are locked out so that the re-entrancy will be detected.
On multiple entries, all AC values after the first will be lost. The first AC may or may not be
lost.
Thus, in case of re-entrancy, ODT will not collapse and the user will be able to determine what
happened. The user may not issue a $C command. The $G must be used to restart the program.
3.3.5 Searching Operations
ODT has the capability to search a specified area of memory for any bit configuration in any specified bit position.
When a match is found, ODT types the location and the contents of the memory register. The search is performed
between a low and high limit with the following algorithm:
a. Get the memory word
b. Mask with the search mask (see below)
c. Compare with k from the k$W command
d. If equal, print location address and unmasked contents. If not equal, go to next memory register.
The following command sequence is used to set the mask, low limit, high limit, and initiate the search:
$M

Open mask register for modification
Open low limit register
Open high limit register

k$W

Search fork
3-13(Part II)

The mask and limits must be set before the k$W command is given. Otherwise, whatever is in these locations will be
used.
Note that a core dump between the low and high limits is performed if the user specifies a mask of 0 and then types
0$W ($W is the same as a 0$W).
Example

Suppose the user wants to locate all JMS instructions between location 200 and 1500. The command sequence would
be as follows:
$M 077336 740000-i-)

User opens and changes the mask.

07745/00100 200.J.)

The line feed on previous line opens low limit

07746/007000 1500).±.

Opens and changes high limit.

100000$W )-1-

User types JMS op code and then

00300/ 102030) .j.

ODT starts the search

01476/126750) H

There are two JMS instructions
JMS

2030 at location 300 and

JMS* 6750 at location 1476.
NOTE

The address of low-limit and high-limit registers may vary depending upon
the assembled version ofODT.

3.3.6 Initialize Buffers ($1)
This command is used to load a specified value into any contiguous block of memory. The user defines the block to
be loaded by setting start and stop addresses into the low and high limits of the word search routine. This means the
user must first open the mask with $M and then type a line feed ( .j.) to open the low limit. Once the start and stop
addresses have been set up, typing k$1 causes the value k to be loaded into the buffer. Typing $1 is the same as typing
0$1.
Example

$M 740000.j.)
07745/000200 433.j.)
07746/007000 624) +
400000$1).J..1.

3-14(Part II)

User opens MASK and then types .j. to open low limit. Change to 433.
Open high limit and change to 624. Fill buffer with 400000.

NOTE

The address of low-limit and high-limit registers may vary depending upon
the assembled version ofODT.

3.3.7 Paper Tape Output
The following paragraphs describe the ODT paper tape output commands.

3.3. 7. I High-Speed/Low-Speed Punch Available ($H, $L) - Typing $H indicates to ODT that all punching is to be
performed on the high-speed punch. Typing $L indicates to ODT that all punching is to be performed on the
low-speed punch (ASR Teletype). Since the low-speed punch and the teleprinter are physically connected, the user
always has the punch turned off until after a command to punch has been given to ODT. ODT halts to allow the user
to turn the punch on and off at the correct times. The user operates the CONTINUE switch on the operator's console
to resume ODT operations.

3.3.7.2 Feed Leader ($F) - The $F command causes about 2 feet of leader to be punched either on the high- or
low-speed punch (depending on whether $H or $L has been seen by ODT). For the low-speed punch, ODT halts to
allow the user to turn on the punch. The user then operates the CONTINUE lever. When the punching is complete
ODT halts again so that the punch may be turned off before the next command is typed. The punch must be off
when the keyboard is used or else the commands will be punched as they are typed. The halts are not performed
when the high-speed punch is used.

NOTE

The initiation of any punching permanently disables the PI and APL The
user's program may not be continued with $C. The program must be
restarted with a $G.

3.3.7.3 Dump Block of Memory (k 1 ; k 2 $0) - This command causes the dumping of all locations between k 1 and
k2 , inclusive. If the low-speed punch is used, halts occur before and after punching. If k 2 is not specified, an error is
assumed.
All dumping is performed in the PDP-15/10 Loader format (see Figure 3-1). The main feature is that noncontiguous
blocks may be punched, each with a separate checksum. After the last block there is a special start block that tells the
loader to stop loading and transfer control to a specified address. This terminal block is punched by using the k$T
command. As many blocks as desired may be dumped before the k$T is issued.

3.3.7.4 Terminate Punching (k$T) - The k$T command causes the terminal block (see paragr~ph 3.3.7.3) to be
punched. If $T is typed, the loader halts when loading is complete. If k$T is typed, the loader gives control to
location k when loading is completed.

3-1 S(Part II)

3.4 OPERATING PROCEDURE
This section contains paragraphs describing the ODT loading procedure, start-up procedure, conditional assembly, and
error recovery.

3.4. l Loading Procedure
ODT can be loaded in several different ways. Since the user will have the symbolic source tape of ODT, he may
assemble it into any convenient, contiguous block of memory. The pseudo-op .LOC must be supplied by the user.
ODT would not, in general, be assembled below location IOI; ODT may not, under any circumstances, be assembled
below location 21. ODT may either be assembled and loaded with the user program, or assembled separately and
loaded with the user program. It could even be loaded after the user program has been running. ODT may be
assembled in the .FULL mode if desired.
ODT will also be supplied as a binary tape that can be loaded by the PDP-15/10 Loader. Assuming that the loader is
already in memory, proceed as follows:
Press IO RESET
Set the Address switches to 7700 or 772'J ( 17700 or 17720, if 8K). Note that these two addresses are for the two
versions of the Loaders; the first one for the low-speed and the second for the high-speed paper tape reader.
Press START. ODT will be initiated automatically by the loader unless a checksum error occurs. The distributed
binary version loads at location 6000 (16000 if 8K) and runs almost up to the PDP-15/10 Loader. The addresses for
starting and restarting are:
%0DT

= 6000 j

%REST = 6011

%0DT = 160001
4K systems or

8K systems

%REST = 16011

3.4.2 Start-Up Procedure
Several start-up procedures may be used with ODT. The distributed binary version starts automatically when loaded.
User-assembled versions are started in a manner dependent upon the .END statement used. The source tape will not
have an .END statement. Instead, it will have the .EQT pseudo operation. When the .EQT is encountered the user
may continue with another segment of his program or he may insert a segment consisting of the .END statement. In
any case, the .END determines the start-up procedure as follows:

.END %0DT - ODT will start automatically when the binary tape is loaded .
.END X - The user program will start at location X when the binary tape is loaded. The user will have to manually
give control to ODT at location o/oODT .
.END -The loader will halt when loading is complete. The user must manually start at o/oODT when he is ready.
If ODT is in control initially, the user program is initiated with the $G command. Normally, control will then stay
in the user program until a breakpoint is executed.
In general, when ODT is started manually, an IO reset should be performed first.

3-l 6(Part II)

When ODT is initiated, it types) H to indicate that it is ready to accept commands.

3.4.3 Conditional Assembly
The ODT source program is designed such that certain features may be deleted under the control of assembly
parameters. At assembly time, a parameter tape may be read before the ODT source tape in pass 1. This tape contains
definitions of parameters that allow the user to generate a particular version of ODT. The parameter tape need not be
read on pass 2; however, reading it on pass 2 will provide a list of all defined parameters for the assembly.
A parameter is either "defined" or "undefined." Defined means that the parameter name appears in the Assembler
symbol table. "Undefined" means that it does not appear in the symbol table. None of the parameters are defined in
the ODT source code. If the source tape is assembled without a parameter tape, all of the parameters listed below will
be undefined; thus a version with all features will be produced. To define a parameter, %%J for example, include the
assignment statement
%%1=0)

on the parameter tape. The parameter tape should be terminated by the .EOT pseudo-op. A complete description of
how to define a parameter is given in Paragraph 1.4. 7.
The following paragraphs list the parameters available for ODT and their effect.
3.4.3. l %%J - If undefined, the command $J (display Pl and API status) is available. If defined, $J is not available,
thus saving approximately 21 ( 17 1 0 ) locations.
3.4.3.2 %%X - If undefined, the $X (execute instruction k) command is available. If defined, the $X command is
not available, thus saving about 46 (38 10 ) locations.
3.4.3.3 %%MULD - If undefined, there is no restriction on setting a breakpoint at a multiply or divide EAE class
instruction. However, if the user's PDP-15/10 does not have the EAE option or if the user does not want to set
breakpoints at multiply or divide EAE instructions (the other EAE instructions are not restricted), then 25 (21 1 0 )
locations may be saved by defining %%MULD
3.4.3.4 %%WM - If undefined, the word searching commands $M, $W, and the $1 command are available. If defined,
these commands are not available, thus saving about 54 (441 0 ) locations. If the user defines %%WM, the utility
program SCAN may be used to perform the word searching or scanning function.

3.4.3.5 %%KU - If undefined, the commands $K and $U are available to control the PI and API at breakpoints. If
defined, these commands are not available, saving about 23 ( 19 1 0 ) locations. In this case, $U is always in effect.

3.4.3.6 %%OFT - If undefined, the commands $H, $L, $D, $F, and $T are available to dump segments of memory in
the PDP-15/10 Loader format on either the high-speed ($H) or low-speed ($L) paper tape punches. If defined, these
commands are not available, saving about 213 (139 1 0 ) locations. When these commands are not available, dumping
may be performed, in the HRI format, by the HRM Puncher utility program.

3.4.3.7 %%V - This parameter controls the $V command for varying the autoindex number. If %%Vis undefined,
the $V command is available. But if %%OB (see below) is defined, ODT will not need an autoindex register and the

3-17 (Part II)

$V command will not be available even if %%V is undefined. If %%V is defined, then the $V command is not
available. In this case, ODT assumes register 17*. If %%Vis defined and %%OB is undefined, approximately 30 (24 1 0 )
locations are saved. But if %%OB is defined, these 30 locations are included in the locations saved.

3.4.3.8 %%OB - This parameter controls the number of breakpoints allowed in ODT. As described in the paragraph
about the $B command (see paragraph 3.3.4.1 ), there are four breakpoints available. This is the case when %%OB is
undefined.
If %o/oOB is defined, there is only one breakpoint available. This has several consequences which are listed below. It is
assumed that the reader is familiar with the description of the breakpoint commands and restrictions described in
paragraph 3.3.4.

a. The command $B removes the breakpoint. (If $nB is typed, ODT ignores the n).
b. The command k$B sets the breakpoint at location k. If it was already set somewhere else, it is moved to
location k. (If k$nB is typed, ODT ignores the n).
c. When the breakpoint is encountered, the message printed is ") .j. $B ~ k) .j." where k is the address of the
break.
d. The re-entrancy problem does not exist when only one breakpoint is available.
e. The command $Y (open re-entrant PC list) is not available.
f. No autoindex register is used; thus, the command $V (vary autoindex) is not available.
g. Approximately 214 (140 10 ) locations are saved.

3.4.3.9 %AUTOX - The default assumption is 17 for which autoindex register to use. The parameter %AUTOX may
be defined to change the default assumption; that is, if %AUTOX is undefined, then register 17 is used unless changed
by the $V command. If %AUTOX is defined as
%AUTOX=XX
where XX is an octal number in the range of 10-17, then the default assumption will be autoindex register XX.

3.4.4 Error Recovery

3.4.4.1 Runaway Program - If the user program does not execute a breakpoint, ODT will not regain control. The
following sequence of operations may be performed to give ODT control:
Depress PROGRAM STOP.
Depress IO RESET
*The assumed autoindex register may be changed by a parameter definition. (See %AUTOX)

3-l 8(Part II)

Set the address of%REST in the Address switches (6011or16011 on the distributed version).
Depress START
This will cause ODT to restore the breakpoint instructions and transfer control to the command decoder. The $G
command must be used to start the user's program.
3.4.4.2 Re-Entrancy - The problem of re-entrancy on breakpoints is discussed in Section 3.3.4.9. ODT always
detects the re-entrancy and prints an error message. At this time the user cannot determine if the PI and/or API were
enabled when the re-entrancy occurred. However, the other PI and API status bits may be examined with the $J
command. The $Y command is used to open the first entry of a list of PC values with the following meanings:
PC 1
PC 2

Adr + l of first breakpoint to occur
Adr + l of next breakpoint to occur

PCn = Adr + l of nth breakpoint to occur (maximum of four)
70XXXX = End of PC list
After the first value has been printed (by the $Y), each successive value may be printed by using the ,j. command. The
end of the list is indicated by an IOT class instruction (70XXXX).
After a re·entrancy error, the user must again specify all breakpoints before restarting his program with the $G
command.

3.4.4.3 Breakpoint Entry Error - In the event of a software error in the user program or in ODT, it is possible that
control could come to ODT as though a breakpoint had been executed. ODT would not be able to determine which
breakpoint was executed; ODT would, therefore, proceed as follows:
Disable PI, API
Print BAD BREAK ENTRY
Halt
ODT can be continued by pressing CONTINUE or by manually starting at locatio!l %REST. Since, after a software
failure, the user program or ODT may be destroyed, the user should reload both his program and ODT.
Another situation, which is not a software error but an operational error, will cause the above error procedure to
occur. This is the execution of the instruction at the breakpoint address, out of line, with an XCT. If this is the case,
reloading is not necessary but the user's program may be restarted only with the $G command.

3.4.4.4 Command Error - If a command is illegal or contains illegal characters, ODT ignores the command and
responds with ? ) ,!.. If, while typing a command, the user changes his mind, he can cause ODT to ignore the
command in the following ways:
Type an 8 or 9.
Type two ALT MODEs (or two $s)

3-l 9(Part II)

Type CTRL
Type an illegal command character (e.g., "E").
3.5 ODT ASSEMBLY INSTRUCTIONS
ODT will be delivered both as an object program tape and as a source tape. The object program tape is loaded and
initiated at location %0DT ( 6000 for 4K and 16000 for 8K). This version includes all ODT commands defined in this
manual.
The source tape is provided to allow the user freedom to assemble ODT at different locations, in the .FULL or
PDP-15/10 Loader format, and with certain features and options deleted or changed.
Paragraph 3.4.3 described all of the conditional features of ODT. The largest version of ODT (no parameters defined
except possibly for %AUTOX) is approximately 950 1 0 locations. The smallest version (all parameters defined except
possibly %AUTOX) is approximately 530 1 0 locations.
The source tape has no .LOC statement and does not have an .END statement. If the user assembles ODT by itself a
.LOC and .END must be provided on separate tapes (segmented source tapes are described in paragraph 1.4.2 (Part
II)). The main reason for leaving off the .LOC and .END is to allow the user freedom to change the .LOC or .END or
follow ODT with user programs without requiring an editing of the ODT source tape.
The following shows the coding for the tapes necessary to assemble ODT at location 200 in the .FULL mode with
automatic initiation and with the dumping and word searching commands deleted.
Tape 1

%%DFT = 0).1%%WM = 0).1-j .EOT ).!Tape 2

-j .FULL).1-j .LOC -j 200) +
.EOT ) .!Tapes 3, 4. 5

Supplied ODT source tape (segmented into three pieces).
Tape 6

-j .END-j%ODT) .!On pass 1 of the assembly, tapes I through 6 must be read by the Assembler. On successive passes, it is not necessary
to include tape 1; however, doing so will cause the parameter assignments to appear on the assembly listing. All tapes
other than the parameter tape must be read on all passes.

3-20(Part II)

CHAPTER 4
COMPACT UTILITY ROUTINES

4. I INTRODUCTION
This section provides descriptions of all utility routines supplied with the PDP-IS/JO COMPACT Software System.
These routines include a FAST-IS system for DECtape handling, a hardware readin mode (HRM) punch routine,
paper-tape handling routines, Teletype 1/0 routines, an octal dump routine, and a scan routine used to search core
memory.

4.2 FAST-15

4. 2.1 General Description
FAST-IS (Fast Acquisition of System Tape) is a loading system for use in the PDP-lS/10 COMPACT Software
System to retrieve frequently used programs from DECtape and to create system tapes. The main advantages of the
system are speed and ease of access.
The equipment required for use of FAST-1 S includes a basic PDP-IS/IO with 4,096 words of core memory, one Type
TC02 DECtape Control unit, and one Type TUSS DECtape Transport.
The FAST-IS system tape, as distributed by Digital Equipment Corporation, contains commonly used system
programs such as the Symbolic Editor, the CAP-IS Assembler, and ODT. Since these can be called from DECtape
with only a small bootstrap, paper-tape handling is eliminated. This results not only in a significant time savings, but
also in increased reliability. FAST-IS is by no means restricted to systems programs; it can be employed very
conveniently for frequently accessed user-created programs. This chapter contains complete directions for use of the
FAST-IS system tape, as well as directions for adding user programs to the system.

4.2.2 The FAST System
The FAST System includes four programs: the FAST Loader, the High Writer, the Low Writer, and the Reader.

4-1 (Part II)

In normal use, once the FAST Loader and FAST Writer have prepared a DECtape for system use, only the FAST
Reader need be used; this program is commonly designated FAST.

4.2.2.1 FAST Loader - The FAST Loader writes a table of contents or directory onto block 1 of a certified
DECtape. The directory consists of 18 three-word entries; one entry for each of the 18 accumulator (AC) switches.
Each directory entry specifies three parameters for the program to be stored and retrieved, under the control of an
AC switch, as follows:
The first location in memory occupied by the program (load point),
The number of locations allocated in memory,
The starting location of the program.

4.2.2.2 FAST Writer -The FAST Writer transfers a program from core memory to DECtape as specified by the table
in the first DECtape block. Program selection is determined by the leftmost AC switch in the 1 position. The FAST
Writer exists in both high and low versions. The high version, which occupies location 7 600 - 7777 8 , is normally used.
However, it is permissable for programs read by the FAST Reader to overlay the first 1008 locations of the Reader
(e.g., locations 7600 - 7677 8 ). So that such programs may be written on a FAST DECtape, a low version of the FAST
writer, which occupies locations 100 - 3008 , is used.

4.2.2.3 FAST Reader - The FAST Reader transfers a program from the DECtape to the computer memory in the
locations specified by the table in the first DECtape block. Program selection is determined by the leftmost AC switch
in the l position.

4.2.3 The FAST Loader
The FAST Loader writes a table of contents (directory) of predetermined programs onto the first block of certified
DECtape. This table determines the order of programs on DECtape and can be modified by the user. This table, which
is located at the end of the FAST Loader, contains the following information for all programs:

Its first location in memory (load point).
The total number of words to be loaded.
Its starting address.
One program or core image is assigned to each of the 18 10 accumulator switches. To each switch, 32 10 DECtape
blocks are assigned which allows 176008 words to be written and/or read. The number of blocks assigned to each
switch is not variable, but the number of words read or written is variable. Switch 17 has 30 1 0 blocks assigned. This
allows 170008 words to be written instead of 176008 .
To load the directory onto the first block of the DECtape, the user should first modify the table of contents in the
FAST Loader as necessary. This is most conveniently done by using the Symbolic Editor and punching a revised
source tape. Using the CAP-15 Assembler, prepare a binary object tape. Select the DECtape for unit I-WRITE. Load

4-2(Part II)

the FAST Loader using the PDP-i5/IO Loader; it will stop with all ls in the AC. Press CONTINUE to execute the
writing of the directory on block l. If an error occurs, the Loader will return to the beginning and halt with all ls in
the AC; otherwise, it will halt with the AC clear (all Os).

4.2.4 The FAST Writer
This description assumes that the DECtape which is about to be prepared is ready; that is, the FAST Loader has been
appropriately modified and the directory has been written onto block l of the DECtape.
To transfer a program from memory to the DECtape:
Load the program into memory.
Place the desired DECtape on a transport and select unit I-WRITE.
Press 1/0 RESET.
Place the Writer in the paper tape reader, set the ADDRESS switches to 7600 (l 7600 for 8K systems), and press
START*. The Writer will stop with all ls in the AC.
Set the DATA (AC) switches to select the desired program. Selection is controlled by the leftmost switch that is
up; all switches to the right of that one are ignored. (No switches up is equivalent to switch 0 up).
Press CONTINUE to execute the transfer. If no error occurs, the computer will halt with the AC clear. If an error
occurs, the Writer will return to the beginning and stop with all ls in the AC.

4.2.5 The FAST Reader
The FAST Reader (commonly designated FASn occupies memory locations 7600 - 7777 8 (17600-17777 8 for 8K
systems). FAST destroys the 668 locations immediately preceding the first location of FAST before the program is
read from the DECtape. However, the program read-in may overlay all of these 668 locations. If several programs are
to be loaded through repeated use of FAST, only the last program to be loaded may overlay the 668 locations
mentioned above since each call to FAST destroys these locations.

4.2.5.l Standard FAST System Tape -The FAST System DECtape (prepared for system program retrieval)
distributed by Di~tal Equipment Corporation, allocates the first five DATA (AC) switches as follows:
Leftmost

Switch Up

Program
Symbolic Tape Editor
CAP-15 Assembler

ODT (with all commands
defined)

*If the area of core to be written will overlay the Writer, then the user must read in the low version of the Writer and set the Address
switches to 1008 .

4-3(Part II)

Leftmost
Switch Up

Program

Paper Tape Lister

Paper Tape Duplicator

The user can, however, prepare different system DECtapes and working program DECtapes by the method described
in paragraphs 4.2.3, 4.2.4, and 4.2.6.
All of the standard system programs as distributed by DEC are initiated automatically by FAST after they are loaded.
All system programs, except for ODT, are not normally loaded into memory with user programs; thus, their memory
requirements are of no concern. ODT is loaded with user programs. This program is an overlay of FAST*, and must
be loaded by FAST after other programs have been loaded. The memory requirement for ODT is exactly the same as
the object program distributed on paper tape.

Program

Load Address

Start Address

Size

ODT

6000

6000 (o/cODT)

1665

6011 (%REST)

4.2.5.2 The FAST S~rt - These instructions are for use with DECtapes (similar to the one distributed by DEC) that
have been properly prepared.
Place the FAST System DECtape on a transport and select unit I-WRITE LOCK.
Assuming FAST is already in memory, set the ADDRESS switches to 7600 (17600 for 8K systems) and press
START (this is known as FAST Start). The computer will halt with all Is in the AC.
If FAST is not in memory, perform the following:

Press I/O RESET.
Place the FAST binary tape in the paper tape reader.
Set the ADDRESS switches to 7600 (17600 for 8K systems) and press START.
The computer will halt with all Is in the AC.
Set the AC switches to select the program desired, according to the table of contents (written onto block 1 by the
FAST loader) associated with the DECtape.

*The first 1008 locations of FAST can be overlayed by a program read from DECtape.

4-4(Part II)

Press CONTINUE. The DECtape will rewind to block l, search forward to the appropriate block, and transfer the
program. If no error occurs, control will transfer to the start of the program. If an error occurs, the computer halts
with all 1s in the AC. Check for an incorrect setting of the DECtape switches. The DECtape status flags can be
examined for possible malfunctions.
If the program just loaded did not overlay any part of FAST, then FAST is ready for the procedure beginning with
"Set the AC switches". However, if the program did overlay FAST, FAST must be reloaded.

NOTE
The program selected is determined by the leftmost AC switch that is up; the
positions of other switches to the right are ignored by FAST. No switches up
is equivalent to switch 0 up.

4.2.6 Example Writing A FAST System Tape
This example shows how to add a program to the standard system DECtape. The program to be added uses 426
locations beginning with location 2000. The start address is 2010. The binary tape is in the CAP-15 Loader format.

4.2.6.1 Preparing the Directory - The standard systems tape has a directory with five programs assigned to switches
0 through 4. This example will use switch 5 for the new program. The directory may be changed in two ways. The
best method is to edit the source tape of the FAST Loader and assemble the new FAST Loader as described in
paragraph 4.2.3. This always generates a hard copy of the new directory. The other method, which may be faster but
is prone to errors, is to load the FAST Loader into memory, manually alter the proper directory entries and then
execute the FAST Loader. In any case, the first 6 entries in the directory are initially as follows:

TABLE
ZERO

ONE

TWO

TABLE
I

7477
22
1
7477
22
16000
1700
16000

THREE

FOUR

1000
22
I
2000
22

/DUMMY WCATION
/SWITCH 0: SYMBOLIC TAPE EDITOR
/STARTING ADDRESS
/SWITCH I PDP-15/10 ASSEMBLER

/SWITCH 2 ODT-15
/OVERLAYS FAST

/SWITCH 3 PAPER TAPE LISTER

/SWITCH 4 PAPER TAPE DUPLICATOR

4-S(Part II)

FIVE

/SWITCH S UNASSIGNED
7477
22

The standard system programs will remain as they are. Therefore, the entries for switches 0 through 4 need not be
changed. The entry for switch 5 will be changed to

FIVE

2000
426
2010

/PROGRAM LOADED AT 2000
/USES 426 WORDS
/INITIATE AT 20 l 0

When the directory has been properly prepared, it may be written on DECtape as described in paragraph 4.2.3.

4.2.6.2 Writing the Program on DECtape - To add the program to the system DECtape, proceed as follows:
Place the prepared DECtape on unit 1 - WRITE.
Place the binary tape containing the program in the paper tape reader.

Set the ADDRESS switches to 7720 (or 7700) and press IO RESET and then START.
Set the ADDRESS switches to 7600 ( 17600 for 8K systems), place the HRI tape for the FAST Writer in the paper
tape reader, and press READIN.

Set AC switch 5 up (all others down).
Press CONTINUE. The DECtape will rewind to block 1 and then will search forward to the proper block and write
out the program.
If no errors occur, the Writer will halt with Os in the AC. If it halts with all Is, an error has occurred. The program
will be written again if the user presses CONTINUE.

When the DECtape stops and the program halts with the 0 in the AC, the writing is complete. To check that the
program may be read again, proceed as follows:

Change the DECtape unit to unit 1 - WRITE LOCK.
Load the FAST Reader by placing the tape in the paper tape reader. Set the Address switches to 7600 (17600 for
8K systems) and press READIN.

Set AC switch 5 up. Press CONTINUE to read the program. Control should be transferred to location 2010 when
loading is complete. If an error occurs, the reader will halt with all ls in the AC. In this case, press CONTINUE to
read program A again.

4-6(Part II)

4.2.7 Assembling The High And Low Writers
As mentioned previously, the FAST Writer exists in two versions. The high version occupies locations 7600- 7777 8
(17600 - I 7777 8 for 8K systems) and the low version occupies locations JOO - 277 8 • There is only one source tape for
the two versions of the Writer. The source code is conditionalized so that if the symbol LOW is defined, the low
version is produced at assembly time; if undefined, the high version is produced. (See Part II, paragraph 1.4.7, for a
complete description of conditional assemblies.) Thus, if the source tape is assembled by itself, the high version is
produced since LOW is undefined. The low version is produced by preceding the source tape with a parameter tape
containing the following two statements.

LOW=O)
--j.EOT)

4.3 HRM PUNCHER

4.3.l General Description
The Hardware Readin Mode (HRM) Puncher is a self-relocating dump program, written in CAP-15 Assembly
Language. It can be loaded by means of the PDP-15/10 hardware readin (HRI) facility (see PDP-15 Reference Manual)
into any block of memory. Once loaded, the HRM Puncher relocates itself and punches out a block of contiguous
memory locations, specified by the user, in the HRI format. The HRM Puncher operates anywhere in up to 8K of
memory.

4.3.2 Output Format
Binary output is punched in the HRI format which consists of the data words followed by the hardware readin word.
The data words correspond to consecutive memory words from the start address through the stop address. Each word
is punched as three frames. Each frame has channel 8 punched and channel 7 not punched. Channels I through 6
contain six bits of the data word. The three frames correspond to the memory word as follows:

Frames

xxxxxx
xxxxxx
xxxxxx

t!t:
0

Bits 0-5
Bits 6-11
Bits 12-17

Channels 1-6
Channe I 7 not punched
Channe I 8 a !ways punched

Following all the data words is the HRI word which halts the processor. This word is different than the data words in
that channel 7 of the third frame is punched. The HRI word is punched as follows:

4-7(Part II)

•

74}
00
740040 = HLT

•

The HRI reads this word and executes the instruction, i.e., the processor halts.

4.3.3 Functional Description
The HRM Puncher is self-relocating and self-initializing. It uses several locations external to itself. These are:

Location 0
Locations 7766-7777, see below; (17766-17777 for 8K).

The source tape may be assembled under the control of two parameters to get four versions of the HRM puncher. The
parameters are defined (or left undefined) at assembly time by means of a parameter tape (See Paragraph 1.4.7). The
parameters are:
SLOW

Defined for low-speed punch (ASR).
Undefined for high-speed punch.

Defined to punch leader/trailer.
Undefined to punch no leader or trailer.

The four versions are described below. The distributed object program, in the .FULL mode, is for the low-speed
punch and produces leader/trailer.

4.3.3. 1 High-Speed Punch With No Leader/Trailer - For this version, both parameter~ are undefined. The leader and
trailer should be provided manually. This version requires 73 8 locations for the puncher, plus locations 0 and
7770-7777 for pointers and temporary storage.

4.3.3.2 High-Speed Punch With Leader/Trailer - This version is assembled with V2 defined and SLOW undefined.
The memory requirements are 115 8 locations for the puncher and locations 0 and 7766-7777.

4.3.3.3 Low-Speed Punch (ASRI With No Leader/Trailer - This version is assembled with only SLOW defined. The
memory requirements are 1028 locations for the puncher and locations 0 and 7770-7777. The leader and trailer
should be provided manually.

4.3.3.4 Low-Speed Punch With Leader!Trailer - Both V2 and SLOW are defined to assemble this version. The
memory requirements are 1248 locations for the puncher and locations 0 and 7766-7777.

4-8(Part II)

4.3.3.5 HRM Puncher - The HRM puncher is divided into two almost equal parts. The first part performs the
relocation and initialization (which includes the reading of the switches to set the start and stop addresses for the
dump). The second part performs the actual dumping process.

4.3.4 Operating Procedure
The following paragraphs contain procedures for loading and starting the HRM puncher.

4.3.4.1 Loading Procedure - Proceed as follows to load the HRM puncher:
Press IO RESET.
Place the HRM puncher in the reader.
Set the address of where the HRM puncher is to be loaded in the ADDRESS switches.
Press the READIN key.
The load address must not be 0. The address must be selected so that the HRM puncher does not overlay locations
7766-7777. The size of the puncher is defined for the various versions in Section 4.3.3.

4.3.4.2 Start-up Procedure - The HRM puncher can be started and/or restarted at any time after it has been loaded.
The procedure is:
Manually generate leader if necessary.
Press IO RESET.
Initiate at the selected load address by pressing START. The program will halt immediately.
Load the start address for the dump into the AC switches. Press CONTINUE. The program will halt immediately.
Load the stop address for the dump into the AC switches (this address must be greater than the start address). Press
CONTINUE.
When punching is complete, the program will halt. Manually generate the trailer if necessary.
To repeat, or to punch out another area, repeat the procedure from the beginning.

4.4 PAPER TAPE HANDLING ROUTINES

4.4. l Paper Tape Lister (PTLIST)
The Paper Tape Lister (PTLIST) is used to read an ASCII-coded paper tape from either the high-speed or low-speed
paper tape reader, and to provide a character-by-character listing on the Teletype. Carriage return and line feed

4-9(Part II)

characters must be punched on the tape if these operations are to take place (they are not handled automatically by
PTLIST). If a tab is encountered by PTLIST, it is converted to the appropriate number of spaces. Each tab stop is
assumed to be every tenth print position.
When the program has been loaded by the PDP-IS/IO Loader, it will type the following message on the Teletype:
PLEASE READY THE INPUT DEVICE AND SET THE AC SWITCH.
If input is to be from the low-speed (ASR) paper-tape reader, the user should set the reader switch to the ON
position, set the AC switches to 400000, and depress the CONTINUE switch to start the listing. The program enters a
wait loop when the reader runs out of tape. Inserting more input will cause the reader to continue. Typing CTRL U
will cause the program to terminate*.

If input is to be from the high-speed paper tape reader, the user should set the AC switches to 0 and depress the
CONTINUE switch to start the listing. The listing will terminate when the reader runs out of tape. At termination, the
program types whatever is left in the input buffer and halts at location 2 I. Press CONTINUE to restart.

4.4.2 Paper Tape Duplicator (PTDUP)
The Paper Tape Duplicator (PTDUP) is used to duplicate and/or verify ASCII or binary paper tapes using the
high-speed paper tape reader and punch. The program can also be used to punch a title on a tape that is being

duplicated.
After the program has been loaded using the PDP-15/10 Loader, it prints the following message:
SWITCH? (M, V, or D)
The user should iype the letter of the function that he wishes to perform, as desCribed in the following paragraphs.

NOTE
At the end of the job the program halts at location 21. Press CONTINUE to
restart.

M (Master Tape Duplicator) - This switch allows the user to type in a title and have that title punched in readable

format preceding his duplicated tape. The following characters have a special meaning while typing in a title line.

ASCII Character

Action

2 I 2 LINE FEED

Punch a title line and return for more input.

215 CARRIAGE RETURN

Punch a title line and return for more input.

377 RUB OUT(S)

Ignore the previous character(s)

*CTRL U is formed by depressing the CTRL key while striking the U key.

4-lO(Patt II)

ASCII Chsacter

Action

375ALTMODE

Punch title line and begin duplicating the tape (acts like D switch after
it punches the title).

225 CTRL U

Ignore whatever is typed on this line and begin to type a new line.

D (Duplicate a Tape) - The following message is printed:

PARITY
The user should type Y if he wants even parity to be generated. Any other character will cause the paper tape image
to be duplicated as is. If the high-speed reader runs out of tape, the program wiJI terminate when it has completed
punching the input buffer.

NOTE

The tape being duplicated must have at least one inch of trailer. The last ten
frames are not duplicated.
V (Verify a Tape) - A parity check is performed on each frame. Frames not having even parity cause PARITY
ERROR to be printed. The program will stop with the reader positioned at one frame past the frame in error.
Depressing CONTINUE will cause the program to continue verifying.

At the end of duplication the input frame count and output frame count are printed in octal, as follows:
INPUT FRAME COUNT xxxxxx .
OUTPUT FRAME COUNT xxxxxx
If the input and output frame counts are unequal, an error has occurred. The user should then restart the job.

NOTE

If the punch runs out of tape, the programs halts with all 1s in the AC. Refill
the punch and continue.

The following message is always printed when verification has been completed.
PARITY ERROR= nnnnnn
where nnnnnn indicates the number of error frames in octal.

4.5 TELETYPE INPUT/OUTPUT ROUTINES
The Teletype Input/Output routines include the Teletype 1/0 Conversion (TICTOC) and the Decimal and Octal Print
packages. Each of the routines in these packages is described in the following paragraphs. Each package should be
assembled along with the program with which it is to run. The Teletype must be initialized by the user program to

4-1 l(Part II)

enable use of the Teletype I/O routines. This is accomplished by the statement TLS+ l 0 at the beginning of the user
program (see PDP-15 Reference Manual).

4.5.l Teletype 1/0 Conversion (TICTOC) Package
The Teletype Input/Output Conversion (TICTOC) Package is used to convert 8-bit ASCII code to a 6-bit trimmed
ASCII code, and vice versa. Formatting facilities are also available. Routines in the package fall into three main
categories: input, output, and formatting. The routines in each of these categories are described in the following
paragraphs, along with input and output formats and character sets.

NOTE

When using output and formatting routines, ensure that all characters have
been printed before halting. This can be accomplished by means of the "skip
if teleprinter flag set" instruction TSF (see PDP-15 Reference Manual).

4.5.1.1 Input Routines - The input routines include %TIC and %TICI. %TIC is used to input a string of characters
from the Teletype and pack them three to a word into memory. The first, second, and third characters (in order of
arrival) are packed into the left, middle, and right sections of a word. Subroutine %TIC requires two arguments for
execution; the calling sequence is as follows:

%TIC

(PC-I)
(PC)
(PC+ l)
(PC+ 2)

LAW
JMS
.DSA
(Return)

(Stop Character)
%TIC
(Buffer Area)

Stop Character may be the trimmed ASCII (see Section 5.1.4) or 8-bit ASCII of any character. %TIC will only look at
the rightmost six bits of the Stop Character. When this character is typed, %TIC stores it with the rest of the text and
returns to the calling program at location (PC + 2). Buffer Area is the address of the first location of a block of
storage into which the incoming text is to be packed. When the Stop Character (terminating character) is
encountered, it is packed and the rest of the word filled out with zeros, if necessary. The user is not allowed to use
the character @ as part of his text.
Four teletype keys have special meaning for %TIC:
LINE FEED (ASCII 212) - causes %TIC to ignore what has been typed and start over again. The input buffer
address is reinitialized to receive the new text, and %TIC outputs a carriage-return and line feed to the Teletype. The
tab count is set to 0.
AT SIGN (@ASCII 300) - delimits the text externally. The @performs the same function as the Stop Character.
The rightmost six bits are stored with the rest of the text and, if necessary, the word is filled out with zeros. Return is
made to location PC+ 2 of the calling program (see calling sequence).
TAB KEY (ASCII 211) - TICTOC keeps a tab count. When the TAB key is struck, the Teletype spaces to the next
TAB stop. The spaces are stored in the input buffer with only the rightmost six bits being packed. The tab count is
then set to 0.

4- I 2(Part II)

CARRIAGE RETURN (ASCII 215)-The tab count is cleared and a Teletype carriage return and line feed are
executed. No data is packed and %TIC keeps listening. %TIC will not stop listening until either the Stop Character or
an@ (at sign) has been typed. %TIC uses subroutine %TICI to get an 8-bit ASCII character. %TIC restores the AC and
link before returning. Refer to Section 5.1.4 for list of valid %TIC characters.

%TICJ Calling sequence:
(PC)

JMS

(PC+ I)

(Return)

%TICI

/Subroutine Call
/Return with 8-bit ASCII
character in AC.

Subroutine %TICI inputs a single 8-bit ASCII character from the Teletype. %TIC1 uses subroutine %TOCI to echo
the character that was typed on the keyboard.

4.5.l.2 Output Routines -The output routines include %TOC, %TOCI, and %TDIG.

%TOC - Subroutine %TOC is used to type out a string of text in the same format as described for %TIC (three
6-bit characters per word). Each 6-bit set is tested for being less than 408 . If the 6-bit set is less than 408 , then 3008 is
added to it to form an 8-bit ASCII character. 2008 will be added to the 6-bit set if it is 40 8 or greater. When the 8-bit
character has been built, %TOC will use subroutine %TOC1 to print it. The calling sequence is as follows:
(PC- 1)

LAW

(Stop Character)

(PC)

JMS

%TOC

(PC+ I)

.DSA

(Buffer Address)

(PC+ 2)

(Return)

The AC and link are restored by subroutine %TOC. The buffer address is the address of the first word of the block of
storage containing the text. The last character in the typing string is the Stop Character, which is not typed. As in
%TIC, the rightmost six bits of the Stop Character are the only bits considered. If an at sign (@)is encountered, it will
act as if it were the Stop Character.

%TOCJ - Subroutine %TOCI outputs an 8-bit ASCII character to the teleprinter. A tab count is kept by %TOCI
which is used by subroutines %TABIT and %TIC. No assumption is made concerning the position of the teleprinter.
%TOCI types one character at the current position of the Teletype. The 8-bit character is not examined in any way
before printing on the Teletype. The calling sequence is as follows:
(PC)

JMS

(PC+ I)

(Return)

%TOCI

/Call with 8-bit character
in AC.
/Return with AC and link
unchanged.

%TD/G - Subroutine %TDIG is used to type a single digit. The calling sequence is as follows:

4- I 3(Part II)

(PC)

JMS

(PC+ I)

(Return)

%TDIG

/AC must already be loaded.
/AC and link unchanged.

The AC must be set before calling %TDIG. The leftmost fourteen bits are stripped from the AC. The four bits left are
added to 2608 to form an 8-bit ASCII character. %TDIG uses subroutine %TOC l to print the character.

4.5.l.3 Formatting Routines - The formatting routines include %CARR, %TABIT, and %SPACE. The calling
sequence and description of each routine follows.

%CARR - Calling sequence:

(PC)

JMS

(PC+ 1)

(Return)

o/cCARR

/Subroutine call

A carriage return (ASCII 215) and line feed (ASCII 212) are printed. The tab count is then set to 0.
%TABIT- Calling sequence:

(PC)

JMS

(PC+ 1)

(Return)

%TABIT

/Subroutine call

The Teletype spaces to the next tab stop. The tab count is then set to 0. This count is kept automatically. %TIOCN,
the number of spaces in a tab, is normally assembled as 8 (decimal), but may be altered by the user.
%SPACE - Calling sequence:

(PC)

JMS

(PC+ l)

(Return)

%SPACE

/Subroutine call

A space (ASCII 240) is printed on the Teletype. The accumulator and link are unchanged by calling the formatting
routines.

4.5.I.4 Input Format and Character Set - The input characters read from the Teletype buffer by %TICl and %TIC
are 8-bit ASCII. They are converted to 6-bit ASCII by stripping off the high-order 2-bits. The character set is as shown
in Appendix I.

4.5. l .5 Output Format and Character Set - The 6-bit code is expanded to 8-bit ASCII by adding 2008 or 3008 to it.
If the 6-bit number is less than 408 , 3008 is added; otherwise, 200 8 is added (see Appendix A).

4.5.2 Decimal And Octal Print Package
The Decimal and Octal Print Routines (%DIP, %OPT, %OPS, and %0PZ) are subroutines which dump the
accumulator in either signed-decimal or octal mode.

4-14(Part II)

4.5.2.l Decimal Integer Print (%DIP) -This routine prints the signed decimal equivalent of an 18-bit binary number.
The binary number to be printed must be loaded into the AC before calling %DIP. Insignificant Os (leading Os) are not
printed; the routine types a space in place of each leading zero. In the case of a negative number, before typing the
first significant figure, the routine types a minus sign. If the number is positive, the sign is omitted (space) and
understood to be plus.
4.5.2.2 Octal Print Subroutines (%OPS, %OPT,' %0PZ) - The Octal Print subroutines type the contents of the
accumulator as an octal number, suppressing nonsignificant leading zeros if desired. Two methods of suppression are
available through two entry points.
a. %OPS - The (JMS %OPS) entry causes leading zeros to be suppressed; the printed number occupying only the
number of spaces needed to print all significant digits (left justified).
b. o/oOPT - The (JMS %OPT) entry causes leading zeros to be suppressed by blanks; the printed number is right
justified in six spaces.
c. o/oOPZ - The (JMS o/oOPZ) entry causes no zero suppression to take place.

4.5.2.3 Operation - Input to the subroutines is provided in the accumulator by the user.
The subroutines are called directly:
(PC - 1)

LAC

NUM

(PC)

JMS

%DIP/o/oOPZ/o/oOPS/o/oOPT

(PC+ 1)

(Return)

Control is returned to the user at location PC + 1.
The link and AC are restored upon return to the calling program.
The character set for the decimal and octal print routines is as follows.
%DIP: -,0,1, .. 9,Space
o/oOPS/o/oOPT 0,1, ... 7,Space
o/oOPZ: 0,1, .. 7
Examples

Input (AC)

%DIP

%OPT

%0PZ

%OPS

777777.

-1

777777

377777

131071

377777

400000

-131072

400000

4- l 5(Part II)

Input (AC)

%DIP

%OPT

%0PZ

%OPS

000000

400055

-131027

400055

000055

4.6 OCTAL DUMP ROUTINE

4.6.1 General Description
The Octal Dump Program allows the user to obtain either Teletype hard copy or paper tape output showing the
contents of any register or set for registers that he specifies. The user specifies which registers are to be dumped via
the Teletype keyboard as described in the following paragraph.

4.6.2 Input Format And Character Set
To obtain the dump of all memory locations between registers A (represented by xxx .. ) and B (represented by yyy .. )
where A~ B, the user must type xxx .. -yyy .. -. A minus sign must be placed after each address; there is no need to type
leading zeros in the address.

Examples

To dump from I 08 to 25 8 , type: I 0-25To dump from 1008 to 13778 , type: 100-1377To obtain the contents of a single register, type the address twice, followed by a minus sign each time, as usual.
Examples

To obtain 10, type: I 0-10To obtain 7725, type: 7725-7725To obtain the contents of all registers from address A up to the end of memory (7777 in a 4K machine, 17777 in an
BK machine) type address A followed by a slash.
Example

1500/
To print out memory locations between two registers A and B, type the smallest register first and the largest second,
each followed by a minus sign.
Example
If A < B, type: A - B If A > B, type: B - A -

4-l6(Part II)

If the user types 6 characters and the sixth character is not a slash(/), or minus(-) the program will interpret this as
an illegal character. The highest possible address is 7777 8 (or 177778 for an 8K machine).
Always place a minus sign after each address. If an illegal (not octal) character is typed, the program does not
recognize it and types back a question mark(?) followed by a carriage return and line feed.
The input character set includes the numbers 0,1, ... 7, and the special characters I and-.

4.6.3 Output Format And Character Set
The output format appears as follows
LLLLL

xxxxxx

LLLLL

xxxxxx

where the LLLLLs represent a five-character octal address in increments of I Os , and the xxxxxx 's represent the
contents (in octal) of locations LLLLL to LLLLL + 7s .
Example

00100

123014 530126 430765

00110

453123

110476

693530 511430 610110

104572

271246

312467

764213

222104 631221

745311

113457

If all locations in any line or group of lines of at least I Os locations are equal, the output would appear as follows.

00120

123456 711111

740040 000000 000000

123123

740040

123123

600000

000000

740040 omitted
00160

740040 740040

The output character set includes only the digits 0 through 78 •

4.6.4 Operation
DUMP is normally loaded starting at location 7300 (17300 for an 8K machine). The user can change this starting
address by reassembling the DUMP program, preceded by a strip tape. The strip tape satisfies conditional assembly
statements in the DUMP program by defining a start address. This is accomplished by setting the symbol LOW to the
desired starting address.
Example

LOW= 22,J

-j .EOT,J
(DUMP program follows)
4- l 7(Part II)

The above example would cause the DUMP program to be assembled starting at location 22. If the symbol LOW is not
defined, the default start address is location 7 300 ( 17300 for an 8K machine).
After loading the program starts automatically. It types a carriage return/line feed and prints the following message:
OUTPUT DEVICE =
The user must respond with H if the output is to go to the high-speed paper tape punch. Any other character will
cause the program to assume the teleprinter as the output device. The program then outputs a carriage return/line
feed and goes into a wait loop. The user should then type his parameters.
After the user's parameters have been satisfied, the program restarts automatically with a carriage return/line feed and
waits for more parameters. If the user wishes to change the output device, he must restart the DUMP program at the
original start address.

4.7 SCAN ROUTINE

4. 7 .1 General Description
SCAN is a small ( 1008 locations) program used to scan areas of memory for a particular bit configuration. The user
specifies the start and stop address for the area to be scanned, the bit configuration to look for, and the bit positions
to be tested (e.g., a mask). SCAN then scans the area. When a match is found, the address of the match is printed
along with the unmasked matching word. Proper selection of the operating parameters allows SCAN to be used as a
dump. SCAN operates in either 4K or 8K without reassembly.

4.7.2 Output Format
The output of SCAN is a sequence of addresses and their contents printed on the Teletype. When SCAN is initiated, it
prints a carriage return()) and a line feed(.!-). As matches are found, the addresses and contents are printed as:
AAAAA/XXXXXX) ,i.
where AAAAA is the address (5 octal digits) and XXXXXX is the unmasked contents (6 octal digits). After the last
line has been printed, another carriage return/line feed sequence is printed and SCAN halts.

Example
The following is an example of a SCAN output to search an area for all JMS instructions:

Remarks

)+
00200/ I 00600 ~ .!-

4- l 8(Part II)

JMS 600 at location 200.

Remarks (cont)

00377/100670) i

JMS 670 at location 377.
END of SCAN.

4. 7.3 Functional Description

SCAN is a self-contained and self-initializing program. Once loaded, it may be started and restarted at any time. SCAN
never alters any locations outside of itself. SCAN consists of a main program and three subroutines. The actual scan is
accomplished by the main program. All output is performed by the three closed subroutines. The SCAN process is
performed for all locations between the beginning address and the end address as follows:
Fetch the contents of the core location.
Mask* it with the contents of MASK (specified by user).
Compare the result with the contents of WORD (specified by user).
If not equal, repeat for next core location. If the compare is equal, print the address and the unmasked
contents.

In general, SCAN operates at 1/0 speed. If no matches are detected, a 4K area is scanned in the time it takes to print
two carriage return/line feed sequences.
SCAN may be used as a dump or as a search for any particular bit configurations. For example, the user may:
Locate all instructions with a particular op code (such as all JMSs or all XCTs).
Locate all references to a particular address.
Locate all indirect references.

4. 7.4 Operating Procedure

The following paragraphs describe the loading procedure and start-up procedure for the SCAN program.

4. 7.4. l Loading Procedure - SCAN is loaded by the hardware readin (HRI) facility of the PDP-15 / l 0. The procedure
is as follows.

Load the paper tape into the reader.
Set the address of BEGIN in the Address switches. This address is 7700 ( 17700 for 8K) on the supplied binary
tape.
Depress the IO RESET key.

*"Mask" indicates an l 8·bit Boolean AND function.

4- I 9(Part II)

Depress the READIN key.
When loading is complete, the processor will halt.

4.7.4.2 Start-Up Procedure - The following procedure can be used at any time to start or restart SCAN. The purpose
of this process is to set four parameters for SCAN. These are, in the order that they are set:
Location
4K

7701
7702

17701
17702

WORD
MASK

7703
7704

17703
17704

BEGLOC
ENDLOC

/Bit configuration to look for.
/Specifies which bits to test.
/(! to test, a 0 to not test.)
/First location to test.
/Last location to test.

Initially, the first three values are 0 and the last is 7777. Once a value has been set, it does not have to be set again for
another run if it is desired to use the same value again.
The Start-up procedure is as follows:
Press IO RESET key.
Set the beginning address of SCAN into the ADDRESS switches. This is 7700 for 4K or 17700 for 8K.
Press EXAMINE to store the address in the AR register.
Set the desired value for WORD into the AC switches. Operate the DEPOSIT NEXT key. This sets the value from
the AC switches into memory.

NOTE

If it is not desired to change the present value of the parameter, step d may
be replaced with the operation EXAMINE NEXT.

Repeat the preceding step for MASK, BEGLOC, and ENDLOC.
Operate the START key. This starts SCAN at location BEGIN.
To search for all references to 7200 between locations 500 and 1500:

(WORD)= 7200
(MASK) = 7777
(BEGLOC) = 500
(ENDLOC) = 1500

4-20(Part II)

To dump all of the area between the BEGLOC and ENDLOC, specify WORD = 0 and MASK = 0. This causes a match,
and associated printout, to occur at every location.
If a restart is desired and no parameters are to be set, only the first three steps and the sixth step need be performed.
After the first three steps have been performed, the sixth step may be performed at any time to start the program.

4-21 /22(Part II)

CHAPTER 5
COMPACT MATHEMATICAL ROUTINES

5.1 INTRODUCTION
All mathematical routines in the PDP-15/10 COMPACT Software System library are described in this section. These
routines are grouped in four major packages: Integer Arithmetic, Trigonometric Functions, Floating Point, and
Floating Point 1/0. Table 5-1 on page 5-19 provides a concise listing of their important characteristics. Each package,
if used, must be assembled and loaded with the user program. The size and number of tapes furnished for each
package are as follows.

Size (decimal)

Number of Tapes

Integer Arithmetic

184

Trigonometric Functions

760

Floating Point

574

Floating Point 1/0

480

Package

Separate tapes are supplied for each of the four Integer Arithmetic routines. Trigonometric Functions are on two
tapes - one single precision and one double precision. Each of the above tapes is terminated with a .EOT pseudo-op.
All symbols that are internal to the math routines begin with a percent sign(%) so that the user can avoid using them
as his own symbols.
If any of the trigonometric functions are not required by a user program, the user can specify that they are not to be
assembled by means of a strip tape. The strip tape contains symbols that are used in conditional assembly statements,
and is ended with a .EOT pseudo-op. This procedure can save a great deal of core space, and symbol table space, for

5-l(Part II)

users who do not require all of the trigonometric functions. The following symbols (set equal to 0) should be included
on the strip tape, as desired, to specify that a function should not be assembled.
Single-Precision Functions
Function

SQRT

Size (decimal)

Symbol

%NSQRT

SIN

cos

%SIN (internal routine)

%NSIN
%NCOS
(both of the above)

EXP

%NEXP

ALOG
ALOG 10
%LOGS (internal routine)

II
11
51

%NA LOG
%NLGIO
(both of the above)

ATAN

%NATAN

TANH

%NTANH

POLY

%NPOLY

Double Precision Functions

DSQRT

%NDSQ

DSIN

ocos

7
12

%DSIN (internal routine)

%NDSIN
%NDCOS
(both of the above)

DEXP

%ND EXP

DLOG
DLOGIO
%DLOGS (internal routine)

12
12
59

%NDLG
%NDLIO
(both of the above)

DATAN

%NDTAN

DPOLY

%ND POL

For example, if a user program did not require the functions SQRT, TANH, and ALOG 10, and wished to conserve
core space, he would assemble a strip tape containing the following symbols before assembling his own program and
the trigonometric functions.
%NSQRT=O~

%NTANH=O.J

5-2(Part II)

%NLG10=0.J

-j .EOT .J
Assuming that the double-precision trigonometric functions were not required, the following functions would then be
assembled.
SIN

cos
%SIN

EXP
ALOG
%LOGS
ATAN

POLY
The same procedure applies to the Floating Point Package and to Floating Point I/O. Both the single precision and
double precision sections of the Floating Point Package require that the general floating point section be assembled
with them. However, if either the single precision or double precision section of the Floating Point Package is not
required by a user program, the user can delete them by including the symbols %NSING or %NDOUB on the strip
tape. In like manner, if either the floating point input (FLIP) program or the floating point output (FLOP) program is
not required in the Floating Point 1/0 Package, the symbols %NFLIP or %NFLOP, respectively, should be included on
the strip tape.

5.2 INTEGER ARITHMETIC SIMULATION
The Integer Arithmetic Simulation routines include multiply (MULT), logical multiply (LMUL), divide (DIV), and
logical divide (LDIV). The purpose of these routines is to allow PDP-15I10 users to program using simulated multiply
and divide instructions. A description of each routine is given in the following paragraphs. The reader is referred to
Table 5-1 (Page 5-19) for a tabular summary of all COMPACT mathematical routines. Included in the summary for
each routine is the name, mnemonic, calling sequence, function, errors, accuracy, timing, storage requirements, and
pertinent comments.

5.2. l Multiply Routines (MUL T/LMUL)
The purpose of MULT is to multiply a signed 18-bit multiplicand by a signed 18-bit multiplier and produce a signed
36-bit product. Return is made with the low-order product in the AC and the high-order product in location
%MHIGH.

5-3(Part II)

The purpose of LMUL is to multiply a logical (unsigned) 18-bit multiplicand by a logical 18-bit multiplier and
produce a logical 36-bit product. Return is made to the calling program with the low-order product in the AC and the
high-order product in location %LMHY.
The calling sequence for the multiply routines is as follows.
PC-I

LAC

Multiplier

JMS

MULT (or LMUL)

PC+I

LAC

Multiplicand

PC+2

(Return)

The algorithm used by the multiply routines is as follows: the least significant bit of the multiplier is tested; if it is
equal to I, the multiplicand is added to the developing product, which is then shifted right one bit position; if it
equals 0, no addition is made before the shift. The process is repeated until all the bits of the multiplier, in order from
least significant to most significant, have been processed.

Example

For this example, assume that register MP! is the multiplier, MP2 is the multiplicand, and that the product will be
developed in registers MP3 and MPI, combined. Each of these registers is 5 bits long for purposes of illustration. The
multiplier (MP!) is equal to 9, and the multiplicand (MP2) is equal to 4.

Comments

MP3

MP1

MP2

00000

01001

00100

Initial state of registers. The least significant bit of MP! is tested.

00100

01001

00100

Since it is a I, the contents of MP2 are added to MP3.

00010

00100

The combined contents of MP3 and MP! are shifted right one position. The least
significant bit of MP! is tested.

00001

00010

00100

Since it is a 0, no addition takes place, the combined contents of MP3 and MP! are
shifted right one position, and the least significant bit of MP! is tested again.

00000

10001

00100

Since it is a 0, no addition takes place, the combined contents of MP3 and MP! are
shifted right one position, and the least significant bit of MP I is tested again.

00100

10001

00100

Since it is a I, the contents of MP2 are added to the contents of MP3.

00010

01000

00100

00001

00100

The combined contents of MP3 and MPI are shifted right one position and the
least significant bit of MP 1 is tested again.
Since it is a 0, no addition takes place, the combined contents of MP3 and MP1 are
shifted right one position, and the multiplication is complete (with a product of 36
in locations MP3 and MPI combined).

5-4(Part II)

5.2.2 Divide Routines (OIV/LDIV)
The purpose of DIV is to divide a signed 36-bit dividend by a signed 18-bit divisor. The signed 18-bit quotient that is
developed is returned to the AC. The remainder, signed the same as the dividend, is returned in location %REM. When
the magnitude of the divisor is equal to or less than that of the high-order dividend, no divis,on takes place since the
quotient cannot be expressed by an I 8-bit signed integer. In this case, the program exits with the link bit set to I. If
division takes place, the link bit is set to 0 prior to exit.
The purpose of LDIV is to divide a logical (unsigned) 36-bit dividend by an unsigned 18-bit divisor. The I 8-bit
quotient that is developed is returned to the AC. The remainder is returned in location %LREM. When the high-order
dividend is greater than or equal to the divisor, division does not take place, and the link bit is set to I. If division
takes place, the link bit is set to 0 prior to exit.
The calling sequence for the divide routines is as follows.
PC-I

LAC

Dividend (high order)

JMS

DIV (or LDIV)

PC+I

LAC

Dividend (low order)

PC+2

LAC

Divisor

PC+3

(Return)

The algorithm used by both division routines is binary long division, where the quotient is determined by a
subtraction process. Unlike decimal long division, where a single quotient digit can be one of lO numbers, in binary
long division the quotient digit is either I or 0. To determine this digit, the divisor is subtracted from the dividend and
if the remainder is negative, the quotient digit is 0, the remainder is ignored, the divisor is moved one place to the
right with respect to the dividend, and the process is repeated. If the remainder is positive, the quotient is I and the
remainder is used as the next dividend, as in ordinary long division. In either case, the divisor is moved one place to
the right and the next bit from the original dividend is included with the new dividend. The following example
demonstrates the algorithm. The divisor is equal to 5, and the dividend is equal to 45 8 •

Example

Step I

IOI I 100101
IOI

Result of division is negative; therefore, quotient is 0.
Disregard remainder. Move divisor one place to right.

-111

Step 2

101 /100101
I 01
+100

Result of division is positive; therefore, quotient is I. Retain
remainder as new dividend and bring down next digit from
dividend.

5-S(Part II)

Step 3

011
101 / 100101
101
1000
101
+l l

Step 4

Oil I

Result of subtraction is positive; therefore, quotient is l.
Retain remainder as new dividend and bring down next digit.

Final quotient is 111 with a remainder of 0 l 0.

101 /100101
101
1000
101
111
101
+010

In implementing the algorithm, the divide routines rotate the dividend left instead of moving the divisor right. No
division occurs if the high-order dividend is greater than or equal to the divisor. This eliminates situations where the
high-order dividend is divisible by at least one (which, if allowed to continue, would produce erroneous results).

NOTE

DIV does not check for the overflow condition that will occur when the
high-order dividend is zero, the low-order dividend is greater than 377777
(bit zero is set), and the divisor is equal to 1.

5.3 TRIGONOMETRIC FUNCTIONS

Detailed algorithms for all trigonometric functions in the PDP-15/10 COMPACT Software System Mathematical
Library are described in this chapter. Most of the functions are computed by methods of approximation as described
by Cecil Hastings in his book Approximation for Digital Computers.
The trigonometric routines must be assembled along with the user program. The Floating Point Package (paragraph
5.4) must also be assembled to enable use of the trigonometric routines. Execution time for the trigonometric
routines is greatly reduced if the EAE option is available since multiplication and division can then be accomplished
by the hardware.
The reader is referred to Table 5-1 for a tabular summary of all PDP-15/ 10 mathematical routines. Included in the
summary for each routine is the name, mnemonic, calling sequence, function, errors, accuracy, timing, and pertinent
comments.

5.3. l Square Root (SORT, DSORT}

The calling sequence for the square root routines is as follows.
PC

5-6(Part II)

JMS

SQRT (or DSQRT)

PC+l

.DSA

PC+2

(Error Return)

PC+3

(Normal Return)

ARG (+400000 if indirect)

If the argument (ARG) is negative, return is made to the error return (PC+2) with the argument in the floatir.
accumulator and the AC set equal to 1; otherwise, return is made to location PC+3 with the square root of th_
argument in the floating accumulator.

A first-guess approximation of the square root of the argument is obtained as follows.
If the exponent (EXP) of the argument is odd:

P0 = .5

(~)
( EXP-I)
2
+ ARG
2

If the exponent (EXP) of the argument is even:

( EXP)

(EXP · t)

Po = .5 -2- +ARG - 2 -

Newton's iterative approximation is then applied three times for SQRT or four times for DSQRT.

ARG)
pi + I = I /2 ( pi + T

5.3.2 Sine And Cosine (SIN, COS, DSIN, DCOS)
The calling sequence for the sine and cosine routines is as follows.
PC

JMS

SIN (or COS, DSIN, DCOS)

PC+I

.DSA

ARG (+400000 if indirect)

PC+2

(Error Return)

PC+3

(Normal Return)

If the integer portion of the product (ARG*7T/2) is too large (i.e., the exponent of the product is greater than 21 8 ),
return is made to the error return (PC+2) with the AC set equal to 3. Otherwise, return is made to location PC+3 with
the sine or cosine of the argument in the floating accumulator.

The argument is converted to quarter circles by multiplying by 2/1T. The low two bits of the integral portion
determine the quadrant of the argument and produce a modified value of the fractional portion (F) as follows.

5-7(Part II)

Low2 Bits

Quadrant

00
01
10
11

I
II

Modified Value (Z)

F
1-F
-F
-(1-F)

Ill
IV

Z is then applied to the following polynominal expression:

sin x

where n

='t

~1=0

C2i +

I I)
z2i +

=4 for SIN and COS, and n = 6 for DSIN and DCOS. The values of Care as follows.
SIN, COS

DSIN,DCOS

C 1 = .157080 x 101

= .157079633 X l 0 1

C3 = -.645964 x l 0°

=-.645964097 x 10°

Cs = 796897 x l 0-1

= .796926260 x 10- 1

C7 = -.467377 x I 0-2

c, = -.468175300 x 10-2

C9 = .151484 x 10-3

C9 = . I 60438400 x 10- 3
C 1 1 ::: -.359518435 x 10-5
C 13 = .544652850 x 10-7

The argument for COS and DCOS routines is adjusted by adding 'lf/2. The sine function is then used to compute the
cosine as follows.

cos x = sin ( !!__ + x)
2

5.3.3 Exponential (EXP, DEXP)
The calling sequence for the exponential routines is as follows.

JMS

EXP (or DEXP)

PC+l

.DSA

ARG (+400000 if indirect)

PC+2

(Error Return)

PC+3

(Normal Return)

5-8(Part II)

If the integer portion of the product ARG*log 2 e is too large (i.e., the exponent of the product is greater than 21 8 ),
return is made to the error return (PC+2) with the AC set equal to 3. Otherwise, return is made to location PC+ 3 with
the exponential of the argument in the floating accumulator.

The function ex is calculated as

which will have an integral portion (I) and a fractional portion (F). Then

where

and n = 6 for EXP, and n = 8 for DEXP. The values of C are as follows.

EXP

DEXP

C0 = .100000 x 101

C0 = . I 00000000 x I 0 1

c, = .346574 x 10°

c. = .346573590 x 10°

C2 = .600566 x I 0- 1

C2 = .600566267 x 10-1

C3 = .69380 I x I 0-2

C3 =.693801368 x 10-2

C4 = .601130 x 10-3

C4 = .601133075 x 10-3

Cs= .416700" lo-4

Cs = .416670330 x 10-4

c6 = .240977 x 10-s

C6 = .240678700 x 1o-s
C7 = .119610000 x 10-6
C8 = .518000000 x 10-8

5.3.4 Natural And Common Logarithms (ALOG, ALOG10, DLOG, DLOG10)
The calling sequence for the logarithm routines is as follows.
PC

JMS

ALOG (or ALOGIO, DLOG, DLOGlO)

5-9(Part II)

ARG (+400000 if indirect)

PC+l

.DSA

PC+2

(Error Return)

PC+3

(Normal Return)

If the argument is Jess than or equal to zero, return is made to the error return (PC+2) with the AC set equal to 2.
Otherwise, return is made to location PC+3 with the result in the floating accumulator.

The exponent of the argument is saved as one greater than the integral portion of the result. The fractional portion of
the argument is considered to be a number between l and 2. Z is computed as follows:

x- ('L
x+

then

log2x = 1/2 +

(~=
1

C2i +I z2i + 1)

where n = 2 for ALOG and ALOG IO, and n = 3 for DLOG and DLOG IO.
The values of C are as follows.

ALOG & ALOG10

DLOG & DLOG10

C 1 = .288539 x 101

C1 = .288539007 x 10 1

C3 = .961471 x 10°

C3 = .961800762 x 10°

Cs = .598979 x I 0°

Cs = .576584342 x JOO
C7 =.434259751 X JOO
for ALOG and DLOG

and

log 10 x = (log 2 x) (log 10 2)

for ALOG I 0 and DLOG l 0.

5.3.5 Arc Tangent (ATAN, DAT AN)
The calling sequence for the arc tangent routines is as follows.

5- IO(Part II)

JMS

ATAN (or DAT AN)

PC+!

.DSA

ARG ( +400000 if indirect)

PC+2

(Return)

Return is made to location PC+2 with the arc tangent of the argument in the floating accumulator. There are no error
conditions.
For x less than or equal to I, Z = x, and

arc tangent x

=(i:

C2i + I z2i + l )

i= 0

where n = 8 for ATAN, and n = 3 for DATAN. For x greater than I, Z = l/x, and

arc tangent x = 7f/2

_(~

~=O

C2i + l z 2i +

~
)

where n =7 for ATAN, and n = 3 for DATAN. The values of C are as follows.

ATAN

DAT AN

C1 = .999999 x 10°

c. = .999215000 x 10°

= -.333299 x 10°

C3 = -.321181900 x 10°

= .199465 x 10°

C5 = .146276600 x 10°

= -.139085 x 10°

C7 = -.389929000 x I 0- 1

C9 = .964200 x I 0-1
C1 I

= -.559099 x 10-1

C13 = .218612 x 10-1

c. 5 = -.405406 x 10-2

To get full 34-bit accuracy in DAT AN, the tangent of the first approximation is taken, and the small angle theory is
then used to minimize the error angle as follows:
tan (P - x)

arctan x = P -

l+xtanP

where P is the result of the first approximation.

5.3.6 Hyperbolic Tangent (TANH)

5-11 (Part II)

The calling sequence for the hyperbolic tangent routine is as follows.
PC

JMS

TANH

PC+l

.DSA

ARG (+400000 if indirect)

PC+2

(Error Return)

PC+3

(!'formal Return)

If the integer portion of the product x * log 2 e is too large (i.e., the exponent of the product is greater than 21 8 ),
return is made to the error return (PC+2) with the AC set equal to 3. Otherwise, return is made to location PC+3 with
the hyperbolic tangent of the argument in the floating accumulator.
The function

tanh lxl=

fi_
\

l+e21xl}

ex is computed as 2x log2e which will have an integral portion (I) and a fractional portion (F). Then,

and n = 6.

where

The values of C are as follows.

C0 = .100000 x 101

C4 = .601130 x 10-3

C1 = .346574 x 10°

C5 = .416700 x 10-4

C2 = .600566 x I 0- 1

= .240977 x 10-5

C3 = .693801 x 10-2

5.3.7 Polynomial Evaluation (POLY, DPOLY)
The calling sequence for the polynomial evaluation routine is as follows.
PC

JMS

POLY (or DPOLY)

PC+l

.DSA

PUST

PC+2

(Return)

5- l 2(Part II)

The value of Z must be loaded in the floating accumulator prior to calling POLY or DPOLY. PLIST refers to a list of
constants stored in contiguous locations within the calling program as follows:
PLIST

/2's complement of number or terms
/Last term

/First term
Return is made to location PC+2 with the result in the floating accumulator. There are no error conditions.
The polynomial

x=f~

\i=O

C2i+ I z2i+

)

is evaluated as follows.

x = Z(C0 + Z 2 (C 1 •••

5.4 FLOATING POINT PACKAGES
The purpose of the Floating Point Package is to allow PDP-IS/IO users to program for floating-point data using
simulated floating-point instructions. The Floating Point Package must be assembled and loaded with the user
program at run time. All simulated floating-point instructions exist as subroutines within this package and must be
called as such by the user program.
The reader is referred to Table 5-l for a tabular summary of all PDP-15/IO mathematical routines. Included in the
summary for each routine is the name, mnemonic, calling sequence, function, errors, accuracy, timing, storage
requirements, and pertinent comments. The following paragraphs provide a brief description of the floating-point
accumulator, and single-precision, double-precision, and general floating-point operations.

5.4.l Floating Point Accumulators
The floating accumulator is a software accumulator which is included in the Floating Point Package. It is a three-word
accumulator, %FACl being the label of the first word, %FAC2 the second, and %FAC3 the third. Floating-point data
is stored in the floating accumulator in the following format; negative mantissae are indicated by the setting of bit 0
of word %F AC2.

5-l 3(Part II)

% FAC I

EXPONENT (2 'S COMPLEMENT)

% FAC 2

SIGN OF MANTISSA
HIGH-ORDER MANTISSA

% FAC 3

LOW-ORDER MANTISSA

Used by both single- and double-precision routines, this format is also the general format of double-precision
numbers. Single-precision numbers have a different format and must be converted before and after use in the floating
accumulator. This conversion is taken care of automatically by the floating arithmetic load and store routines FLAC
and FDAC (see Table 5-l ). The format of single-precision numbers is as follows.

LOW ORDER MANTISSA

DATA WORD I

DATA WORD

EXPONENT (2 'S COMPLEMENn

8 9

SIGN OF MANTISSA

21...___. .__H_IG_H_o_R_D_E_R_M_A_N_T_is_s_A_ _ _ _ _ _ _ ____.
0

5.4.2 Single Precision Operations
Single-precision routines perform floating-point operations on double-word quantities as described in paragraph 5.4.1.
The arithmetic operations are performed with one operand in the floating accumulator and the other operand taken
from storage. The computed result is developed in the floating accumulator, and is accurate to 26 bits.
All routines in the single-precision floating-point package are summarized in Appendix A. In general, the calling
sequence for arithmetic routines is as follows:

JMS

SUBR

.DSA

ARG2

(Return)

S- I 4(Part II)

where ARG2 is the address of the first location of the argument from storage. The single-precision load and store
routines (FLAC and FDAC) must be used to load ARG I into the floating accumulator prior to calling an arithmetic
routine, and to store the result upon return.

5.4.3 Double Precision Operations
Double-precision routines perform floating-point operations on triple-word quantities as described in paragraph 5.4. 1.
The arithmetic operations are performed with one operand in the floating accumulator and the other operand taken
from storage. The computed result is developed in the floating accumulator, and is accurate to 34 bits.
All routines in the double precision floating point package are summarized in Appendix A. The calling sequence is the
same as for single-precision routines except that the double-precision load and store routines (DLAC and DDAC) must
be used for loading and storing the floating accumulator.

5.4.4 General Floating Point Operations
General floating-point routines perform the actual computation for arithmetic operations set up by the individual
single- and double-precision routines. All arithmetic operations are performed with one operand in the floating
accumulator and the other in the held accumulator (%HAC 1 - %HAC3). All floating-point operations are performed
on normalized operands.
All general floating point routines are summarized in Table 5- 1. The calling sequence for these routines is simply
PC

JMS

PC+I

(Return)

SUBR

except for the unfloat routine %FUNF. This routine is called as follows:

JMS

PC+I

(Error Return)

PC+2

(Normal Return)

%FUNF.

where the error return (PC+ I) is taken if there is an attempt to unfloat a number that has an integer portion too large
for an 18-bit word. If the error return is taken, the AC is set equal to 3. Otherwise, return is made to PC+2 with the
result in the AC.

5.5 FLOATING POINT 1/0

5.5.1 General Description
The floating-point input/output (FLIO) program is designed to allow the user to input and output decimal data in
floating-point format. The program is designed to operate with either the high-speed paper tape reader/punch or the
ASR33 Teletype. All data transfers are handled through the floating accumulator. FLIO should be assembled and

5-I 5(Part II)

loaded along with the Floating Point Package (FPOINn and the user program. The user program must initialize
(select) the device used by FLIO; for example, to select teleprinter output, "TSL+ 10" must appear at the beginning
of his program (See PDP-15 Reference Manual).

For input, the user program must specify the device as the only argument when calling FLIP (floating-point input).
The input field is delimited by:
Tab

211

Line Feed

212

Carriage Return

215

Slash

257

The second number of the exponent will also delimit the field. Spaces and illegal characters are ignored until the first
digit is encountered. If the decimal number is too large, the rightmost extra digits are ignored until the exponent or a
delimiting character is typed.
For output, the user program must specify the device and the number of digits to be output when calling FLOP
(floating-point output). Output takes place at the current device position; that is, spacing and other formatting
functions are not automatically performed.
The following represents device assignments for both input and output.
ASR33 Teletype

000000

High-Speed Reader/Punch

400000

The calling sequence for FLIP is as follows.
JMS
xxxxxx
Next Instr.

FLIP

/Subroutine call
/Device specification
/Return

FLOP

/Subroutine call
/Device specification plus no.
/of digits to be output (octal)
/Return

The calling sequence for FLOP is as follows.
JMS
xxxxxx+n
Next Instr.

NOTE

Again, for both input and output, data transfers are processed through the
floating accumulator.

5-l 6(Part II)

5.5.2 Input Format And Character Set
The input format is flexible. The normal form is± .DDDD ... ± EE where the Ds are decimal digits representing the
mantissa (a maximum of nine digits) and the Es are decimal digits representing the exponent (two digits). Any
meaningful variation of the normal form is permitted on input. If the decimal point is omitted, the number is assumed
to be integral If more than one decimal point is in the input stream, the last one to appear is the one that will be
used. A plus or minus sign after at least one number of the mantissa field will signal the program that the next two
digits are exponent digits.
The input character set is as follows.
+
E

Space

Tab

0, l .... 9

Carriage Return
Line Feed

Examples
123.123E4

1231230.

.123123-4

.0000123123

-1.23.123

-123.123

+123123

123123.

5.5.3 Output Format And Character Set
The output format is rigid. The user can specify as many digits to be output as he wishes between one and nine
(inclusive).* After the last mantissa is typed/punched, a plus is output if the exponent is positive and a minus sign is
output if the exponent is negative. The output character set includes the following.

0, I. ..9

Space

*If the argument to FLOP is 0, an "x" is printed, if the argument specifies more than nine decimal digits, an "x" is printed along with the
first nine digits.

5-l 7(Part II)

Examples

. l 2312300+()4
.123-4

.1+10
All output is rounded. For example,
JMS

FLOP

6
indicates that the user wishes to output six decimal digits. If his internal number is . I 23456789E I 2, his output would
be rounded as follows .

. 123457+E12

5-18(Part II)

Table S-1. Summary of PDP-IS/IO COMPACT Mathematical Routines
Accuracy
RoutirwName

Mnemonic

Calling Sequence

Function

Errors•

Bits

Timing***
Non-EAE

EAE

184

Integer Arithmetic
Multiply

MULT

Logical Multiply

LMUL

Divide

DIV

Logical Divide

WIV

LAC Multiplier
JMS MULT/LMUL
LAC Multiplicand

LAC Dividend (high order)
JMS DlV/LDIV
LAC Dividend (low order)
LAC Divisor

l*J

None

447 µs

N.A.

I * J

None

404µs

N.A.

l/J

l>J,
Link set
l>J,
link set

483µs

N.A.

41411s

N.A.

l/J

760

Square Root:
SP Square Root
DP Square Root

SQRT
DSQRT

JMS SQRT/DSQRT
.DSA Address of arg.
(Error Return)
(Normal Return)

Sine:
SP Sine
DP Sine

SIN
DSIN

JMS SIN/DSIN
.DSA Address of arg.
(Error Return)
(Normal Return)

cos
DCOS

Exponential:
SP Exponential
DP Exponential

DEXP

Natural Logarithm:
SP Natural Logarithm
DP Natural Logarithm

ALOG
DLOG

:le footnotes at the end of table.

EXP

xi 12
x112

Comments
Signed high-order product in
location %MHIGH.
Low-order product in AC.
Unsigned high-order product
in location %LMHY.
Low-order product in AC.
Signed remainder in location
%REM.
Unsigned remainder in Jocation%LREM.

Trigonometric Functions

Cosine:
SP Cosine
DP Cosine

Storage
(Decimal)

Any trigonometric function
can be deleted by conditional
assembly, see Chapter l

#1
#1

26
34

8.488 ms
13.907 ms

3.522 ms
5.S8S ms

34
35

sin (x)
sin (x)

#3
#3

26
34

11.460 ms
17.481 ms

4.814 ms
9.441 ms

7
7

Calls POLY and %SIN**
Calls DPOLY and %DSIN**

JMS COS/DCOS
.DSA Address of arg.
(Error Return)
(Normal Return)

cos (x)
cos (x)

#3
#3

26
34

11.895 ms
18.135 ms

5.269 ms
9.894ms

II
12

Calls POLY and %SIN**
Calls DPOLY and %DSIN"*

JMS EXP/DEXP
.DSA Address of arg.
(Error Return)
(Normal Return)

#3
#3

26
34

14.471 ms
27.107 ms

7.944 ms
9.377 ms

69
86

#2
#2

26
34

ll.099ms
17.928 ms

5.817 ms
6.461 ms

II
12

JMS ALOG/DLOG
.DSA Address of arg.
(Error Return)
(Normal Return)

loBeX
IOBeX

Calls POLY and %LOGS**
Calls DPOL Y and %DLOGS**

Table 5-1. Summary of PDP-15/10 COMPACT Mathematical Routines(cont)

Routine Name

Calling Sequence

Mnemonic

Function

Errors*

Timing***

Accuracy

Bits

Storage

Non·EAE

EAE

(Decimal}

Comments

Trigonometric Functions
(cont)
Common Logarithm:
SP Common Logarithm
DP Common Logarithm

ALOGIO
DLOGIO

JMS ALOGJO/DLOGlO
.DSA Address of arg.
(Error Return)
(Normal Return)

Jog,oX
Jog, ox

#2
#2

26
34

11.099 ms
17.928 ms

5.817 ms
6.461 ms

JI
12

Calls POLY and %LOOS"*
CallsDPOLY and %DLOGS**

Arc Tangent:
SP Arc Tangent
DP Arc Tangent

ATAN
DATAN

JMS ATAN/DATAN
.DSA Address of arg.
(Normal Return)

tan- 1 (x)
tan- 1 (x)

None
None

26
34

14.115 ms
54.540ms

7.736 ms
24.075 ms

50
90

Calls POLY
Calls POLY and %DSIN* *

Hyperbolic Tangent

TANH

JMS TANH
.DSA Address of arg.
(Error Return)
(Normal Return)

tanh(x)

17.910ms

9.737 ms

Polynomial Evaluation:
SP Poly. Eva!.

POLY

JMS POLY/DPOLY

None

N.A.

••••

DP Poly. Eva!.

DPOLY

.DSA PLIST
(Normal Return)

None

N.A.

*****

•••••

PUST -N
Cn
Cnt

Floating Point
SP Arithmetic:
Add
Subtract
Multiply

:ee footnotes at the end of table.

=°I: C2i+l z2i+I
i=O

/2's comp. of no. of terms
/Last term
Cn-t

/First term

574
FAD
FSUB
FMPX

Called by SIN, COS, ALOG
ALOGlO, and ATAN
Called by DSIN, DCOS,
DLOG DLOG 10, and
DATAN

Arg. 1fFL.AC)
Augend
Minuend
Multiplicand

Arg.2
Addend
Subtrahend
Multiplier

JMSSUBR
.DSAArg. 2

A+B
A-B
A"B

_,.FL.AC
-+FL.AC
-+FL.AC

None
None
None

26
26
26

415.5 µs
513µs
1.482 ms

393µs
457.5 µs
421.5 µs

Either SP or DP Arithmetic
can be deleted by conditional
assembly. If either of these is

Table 5-1. Summary of PDP-I 5/10 COMPACT Mathematical Routines (cont)

Routine Name
Divide
Reverse Subtract
Reverse Divide
LQad
Store

Timing***
Non-EAE

Storage

Function

Value

Divisor
Minuend
Dividend
Address
Address

R/B -+FL.AC
B-A -+FL.AC
B/A -+ FL.AC
Arg. -+FL.AC
FL.AC-+ Arg.

None
None
None
None
None

26
26
26
N.A.
N.A.

L986ms
510µs
2.211 ms
105µs
79.5 µs

481.5 µs
445.5 µs
556.5 µs
105µs
79.5 µs

Arg. }(FL.AC)

Arg.2

DFAD
DSUB
DMPY
DDVD
DSBR
DDVR
DLAC
DDAC

Augend
Minuend
Multiplicand
Dividend
Subtrahend
Divisor

Addend
Subtrahend
Multiplier
Divisor
Minuend
Dividend
Address
Address

A+B -+FL.AC
A-B -+FL.AC
A*B -+ FL.AC
A/B -+FL.AC
B-A -+FL.AC
B/A -+FL.AC
Arg. -+FL.AC
FL.AC-+ Arg.

None
None
None
None
None
None
None
None

34
34
34
34
34
34
N.A.
N.A.

463.5 µs
516µs
2.031 ms
2.336 ms
519 µs
2.840ms
94.S µs
69 µs

385.5 µs
450its
411 µs
471 µs
435 µs
696µs
94.5 µs
69µs

%ANEG.
%FNEG
%SWITCH

JMS
JMS
JMS

%ANEG.
%FNEG.
%SWITCH

-(AC)-AC
-(FL.AC) .... FL.AC
(FL.AC);=(Held AC)

None
None
None

N.A.
N.A.
N.A.

7.5µs
12µs
57 JJ.S

7.5 µs
12µs
57 µs

'!WM

JMS

'?nFM.

(FL.AC)*(Held AC)

None

2.1 ms

216µs

'if·FG.
%FA.
%FNOR.

JMS
JMS
JMS

%FG.
%FA.
%FNOR.

(FL.AC)+(Held AC)

None
None
None

N.A.
34
N.A.

38µ.s
207 µs
363µs

38µs
201 µs
60µs

%FH.
'if·FS.
'Jl,FlR.
%FD.
%FLOT.
%FUNF.

JMS
%FH.
JMS
%FS.
JMS
%FIR.
JMS
%FD.
'!WLOT.
JMS
JMS
%FUNF.
(Error Return)
(Normal Return)

None
None
None
None
None
#3

N.A.
N.A.
N.A.
34
N.A.
N.A.

21 µs
42µ.s
39 JJ.S
2.1 ms
405µs
46µs

21 µs
42µs
39µs
276µs
95 µs
96jlS

FDVD
FSBR
FDVR
FLAC
FDAC

DP Arithmetic
Add
Subtract
Multiply
Divide
Reverse Subtract
Reverse Divide
Load
Store

Ac:curac:y

Calling Saquance

Mnemonic
Dividend
Subtrahend
Divisor

Value

JMSSUBR
.DSA Arg. 2

JMS SUBR
.DSA Arg. 2

Errors•

Bits

EAE

(Decimal)

Comments
assembled, the General Floating Point tape must also be assembled. See Chapter I.

General Operations:
Negate Hardware AC
Negate Floating AC
Exchange Floating AC
with Held AC
General Floating
Multiply
Get Address
Gen. Floating AC
Normalize Floating
AC
Hold Floating Add
Sign Control
Round and Insert Sign
Gen. Floating Divide
Float
Unfloat

(FL.AC)-(Held AC)

(Held AC)/(FL.AC)
(AC)-(FL.AC)
(FL.AC) .... (AC)

See footnotes at the end of table.

:c....

.......

383

FLIP

JMS FLIP
xxxxxx

/Subr call

/Devke

Input FL.PT.

••••••

N.A.

Sum in Floating Accumulator

Quotient in Floating AccumuIa tor

480

FllJQting Point 1/0
Floating Point Input

Product in Floating Accumulator

181

Either FLIP or FLOP can be
deleted by conditional assembly, see Chapter I.

......

Table 5-1. Summary of PDP-15/10 COMPACT Mathematical Routines (cont)

Routine Name

Mnemonic

Floating Point Output

FLOP

JMS FLOP
xxxxxx

Calling Sequence

Function

/Suhr.call
/Device + no. of
digits to be
oµtput

Output FL.PT.

•For errors designated as #n, the error number is stored in the AC, and the Error Return is taken. Errors result Jn attempts to
I) take the square root of a negative number, 2) take the Logx of an argument <ll, and 3) unfloat a number, the integer
portion of which is too large for an 18-bit word.
**Internal routines. %SIN and %DSIN called by SIN/COS, DSIN/DCOS, and DATAN. %LOGS and %DLOGS called by
ALOG/ALOG JO, DLOG/DLOGJO. Storage requirements for%SIN, %DSIN, %LOGS, and %DLOGS are 55,69,51,and 59
locations, respectively .
..-timings indicated are estimated and represent average-to-worst-case times.
••••Non-EAE = 2.862 ms+ 1.S81C;EAE =1.128 ms+ .896C.C =number of coefficients.
·•••*Non-EAE = 4.322 ms+ 2.482C; EAE • 1.076 ms+ .876C. C =number of coefficients.
•••••Flip will ignore any characters after the first nine significant digits, and will continue to do so until a delimiting character is
read.

Errors•
None

Accuracy
Bits

N.A.

Timing***
Non-EAE
N.A.

EAE

Storage
(Decimal)

N.A.

183

Comments
Device is specified as:
000000 for 1TY
400000 for high-speed papertape reader punch

APPENDIX A
CHARACTER SET
Printing
Character

7-bit
ASCII

IOO
IOI

c
D
E

F
G

H
I

J
K
L
M
N
0
p

Q
R

s
T

w
x
y

z
lt
\

Jt
tt
-t

Null (Blank)
Horizontal Tab
Line Feed
Vertical Tab
Form Feed
Carriage Return
Rubout
Notes:

I02
I03
104
I05
I06
107

6-bit
Trimmed
ASCII**

oott
01
02
03
04
05
06
07

I IO

111
112
113
114
115
116
117
120
121
122
123
124
125
126
127
130
131
132
133
134
135
136
137
000
011
012
013
014
015
177

11
12
13
14
15
16
17
20
21
22
23
24
25
26
27
30
31
32
33
34
35
36
37

Printing
Character

7-bit
ASCII

(Space)
!

040
041
042
043
044
045
046
047
050
051
052
053
054
055
056
057
060
061
062
063
064
065
066
067
070
071
072
073
074
075
076
077

"t
#t
$t
%
&

(
)

*
+
'

I
0
l

2
3
4
5

6
7
8
9

:t
'

<t
=

>t
?t

6-bit
Trimmed
ASCII**

40
41
42
43
44
45
46
47
50

51
52
53
54
55
56

57
60
61
62
63
64
65
66
67
70
71
72
73
74
75
76
77

All other characters are illegal and are flagged and ignored

t = Illegal as source except in a comment or text; Ignored by %TOC and %TOC I.
tt =Ignored by %TOC and %TOCI.

** = .SIXBT pseudo-op text value.

A-l/A-2(Part II)

APPENDIX B
PERMANENT SYMBOL TABLE
Memory Reference

Operate (Cont)

CAL
DAC
JMS
DZM
LAC
XOR
ADD
TAD
XCT
ISZ
AND
SAD
IMP

CLC
LAS

000000
040000
100000
140000
200000
240000
300000
340000
400000
440000
500000
540000
600000

Operate

OPR
NOP
CMA
CML
OAS
RAL
RAR
IAC
HLT

SMA
SZA
SNL
SML
SKP
SPA
SNA
SZL
SPL
RTL
RTR
SWHA
CLL
STL
CCL
RCL
RCR
CLA

740000
740000
740001
740002
740004
740010
740020
740030
740040
740040
740100
740200
740400
740400
741000
741100
741200
741400
741400
742010
742020
742030
744000
744002
744002
744010
744020
750000

LAT

GLK
LAW

750001
750004
750004
750010
760000

1/0 States
IOT

IORS
CAF

700000
700314
703302

Interrupt

IOF
ION

700002
700042

Teletype Keyboard

KSF
KRB
KRS

700301
700312
700322

Teletype Teleprinter

TSF
TCF
TLS

700401
700402
700406

Paper Tape Reader

RSF
RCF
RSA
RRB
RSB

700101
700102
700104
700112
700144

Paper Tape Punch

PSF
PCF
PSA
PSB

700201
700202
700204
700244

B-l(Part II)

Index and Limit Register Instructions

AAS
AAC
AXS
CLLR
AXR

720000
723000
725000
736000
737000

CLX
PAL
PAX
PLA
PLX

735000
722000
721000
730000
731000

Index and Limit Register Instructions (cont)

PXA
PXL

Mode Switching and Listing Instructions

EBA
DBA
SBA

707724
707722
707721

Index Register

B-2(Part II)

724000
726000

10000

APPENDIX C
SUMMARY OF PSEUDO-OPS
Format

Paragraph

Function

.BLOCK

1.4.3.l

label -j .BLOCK

-j expAi

.DEC

l.4.4

-j .DEC)

.DSA

1.4.6

label -I .DSA ..._. exp..)

Defines a user symbol which 1s to be
used only in the address field.

.END

1.4.1

-j .END ..._.START)

Must terminate every source program.
ST ART is the address of the first
instruction to be executed.

.ENDC

1.4.7

-j.ENDC)

Terminates conditional coding in .IF
statements.

.EOT

1.4.2

-j.EOT)

Must terminate physical program
segments, except the last, which is
terminated by .END.

.FULL

1.4.10

-j.FULL)

Produces hardware readin binary tapes.

1.4.7

-.j.IFxxx ..._.exp)

.LIST

1.4.9

-j .LIST)

Resume printing of assembly listing .

.LOC

1.4.8

-j .LOC ..._.exp)

Sets the Location Counter to the value
of the expression.

.OCT

1.4.4

-j.OCT)

Sets the prevailing radix to octal.
Assumed at start of every program.

.SIXBT

1.4.S

label -j .SIXBT ~ /text/)

.SIZE

1.4.11

label

.XLIST

1.4.9

.IFDEF
.IFUND

-I .SIZE)

-j .XLIST)

Reserves a block of storage words equal
to the expression. If a label is used, it
references the first word in the block.
Sets prevailing radix to decimal.

If a condition is satisfied, the source
coding following the .IF statement, and
terminating with an .ENDC statement, is
assembled.

Input text strings in 6-bit trimmed
ASCII, with first character as delimiter.
The Assembler outputs the address of
last location plus one occupied by the
object program.
Suppress printing of assembly listing .

C-l/C-2(Part II)

APPENDIX D
PDP-15/10 HARDWARE READIN BINARY LOADERS
I

/POP-15/10 HARDWARE REAOIN LOADERS
I

/DEFINING %LOW PRODUCES THE LOW SPEED VERSION
/OTHERWISE. THE HIGH SPEED VERSION IS PRODUCED.
I

/LOW SPEED READER VERSION:
/HARDWARE READIN TO 7700 (17700 IF aK>, WHEN IT HALTS,
/PLACE BINARY PROGRAM TAPE IN LOW SPEED READER
/ANO PRESS START, WITH BANK/PAGE MOOE SWITCH IN PAGE POSITION,
I

/IIGH SPEED READER VERSION:
/HARDWARE REAOIN TO 7720 <17720 IF 8K>, WHEN IT HALTS,
/PLACE BINARY PROGRAM TAPE IN HIGH SPEED READER
/ANO PRESS START, WITH BANK/PAGE MODE SWITCH IN PAGE POSITION,
I

/LOADER HALTS:
I

/AC=777777 - PROGRAM LOADED.
IAC:NONlERO - CHECKSUM ERROR ON LAST BLOCK LOADED,
I
REPOSITION TAPE AT BLANK FRAME PRIOR TO
I
BEGINNING OF LAST BLOCK AND PRESS START
I
TO REREAD.
I
TO IGNORE ERROR, PRESS CONTINUE.
I

CAF:703302
RSF=700101
RSB=700144
RRB=71110112
KSF=700301
KRH=700312
KRS=71!10322
,FULL
SKPFLG:RSF
RDSLCT:RSB
ROBFR=RRB
.L 0 c 11120
• I FOEF %LOW
SKPFLG:KSF
RDSLCT=KRS
RORFR:KRB
.LOC 17700
.ENDC
LONXBK

CAF

OlM LOCKSM
-dMS LDREAU
DAC LOSTAO
SPA
JMP LDXFR
JMS LDREAO
DAC LDWOCT
JMS LDREAO
LDNXWD JMS LDREAO
OAC• LDSTAD

/CLEAR FLAGS
/CHECKSUMMING LOCATION
/GET A WORD
/BLOCK HEADING-LOADING ADDRESS
/START BLOCK
1WORO COUNT (2'S COMPLEMENT)
/LOAD DAT A INTO

D-1 (Part II)

LDXFR

LDREAD

LORD A

LDC TR
LDT MP
LDMSK
LDCKSM
LDSTAD
LOWD CT

IS2 LOSTAO
!Si! LDWDCT

/MEMORY
/FINISHED LOADING

JMP LONXWO
TAD LDCKSM

/NO
/ADD INTO CHECKSUM

SH
HLT

/CHECKSUM ERROR

JMP LONXBK
f)AC LOWDCT
ISl LDWDCT
JMP* LDSTAD
CLC ! HLT

/EXECUTE START ADDRESS
/MANUALLY START USER PROGRAM

fll

• IFDEF %LOW
LAW -3
DAC LOCTR
f);!M LOTMP
,ENOC
TAD LDCKSM
DAC LDCKSM
RD SL CT
SKPFLG
JMP • -1
ROB FR
, I FDEF %LOW
TAD LDMSK
SPA!CLL
JHP LDRDA
TAD LDTMP
ISl LDCTR
SKP!RTL
.ENDC
JMP• LDREAD
. I FDEF %LOW
RTL
RTL
DAG LDTMP
JMP LDROA
0
fll

777600

.ENDC

0
0
fll

.END

D-2(Part II)

/WAIT FOR READER
/READ BUFFER
/BINARY FRAME
/YES

/NO

/ACCUMULATE 3 FRAMES
/INTO 1 BINARY WORD
/PACK COUNTER
/BINARY WORD
IBJNARY FRAME MASK
/CHECKSUM
1LOADING1STARTING ADDRESS
/WORD COUNT

APPENDIX E
SUMMARY OF EDITOR COMMANDS
Transfer from Edit Level to Input Level

Carriage return as first character.

>I)

INSERT command with no arguments.
OVERLAY command.*

Transfer from Input Level to Edit Level

)

Carriage return as first character.

CTRLP

(echoes tP)

Text Input/Output
S( &...1 n))

(SIZE) Sets number of lines to occupy buffer.

READ)

Reads sequential lines from input device; number
of lines specified by SIZE command.

G( &...1 n))

(GET) Adds next n lines from subsidiary device to
the buffer.

RENEW)

Writes contents of buffer and reads new block.

WRITE)

Punches contents of buffer on output device and
clears buffer.

CLOSE)

Writes remainder of input file on output device;
must be preceded by a WRITE command.

Pointer Manipulation

(TOP) Moves pointer to top of buffer.

N( &...1 n))

(NEXT) Moves pointer past next n lines.

F L..I string )

(FIND) Searches buffer to find next line that
begins with string.

L ..._. string)

(LOCATE) Searches buffer to find next line that
contains string.

(BOTTOM) Moves pointer to final line in buffer.

*n lines are deleted from the buffer before the control level is changed.

E-1 (Part II)

SEARCH 11-1 string)

Searches input file for next line that begins with
string.

Editing Requests
R 11-1 line)

(RETYPE) Replaces the current line with "line."

A 11-1 string)

(APPEND) Adds "string" to current line.

C 11-1 qstring I qstring2q)

(CHANGE) Replaces string! with string2.

I 11-1 line)

(INSERT) Inserts "line" after current line.

G( 11-1 n))

(GET) Adds n lines from subsidiary input device
after current line.

D( 11-1 n))

(DELETE) Deletes next n lines from buffer.

0( 11-1 n))

(OVERLAY) Deletes next n lines from buffer and
transfers control to input level.

Examination Requests
(PRINT) Prints next n lines on Teletype.
(VERIFY) Causes text lines to be printed in
response to certain commands.
(BRIEF) Causes abbreviated printing of VERIFY
lines.

E-2(Part II)

APPENDIX F
CREATING ASCII TEXT WITH THE EDITOR
(Using ASR33 Reader/Punch)
This appendix contains procedures intended for the user who is not familiar with the Editor, but wishes to create
ASCII text on paper tape immediately. The basic procedure includes loading the paper tape containing the Editor into
core memory, completing an initialization sequence, typing the desired text on the Teletype, and punching out the
text (which has been held in the Editors buffer). It is assumed that the low-speed (ASR33) paper tape reader and
punch are to be used. Appendix D contains detailed examples of basic Editor operation, and Appendix E contains a
comprehensive Editor demonstration.

To punch leader:
l. Turn Teletype switch to OFF LINE.
2. Press Teletype punch switch ON.
3. Press HERE IS key several times to generate leader.
4. Press Teletype punch switch OFF.
5. Turn Teletype switch ON LINE.
To load the Editor:
6. Place the tape containing the low-speed, hardware readin binary loader in the Teletype reader.
7. Press Teletype reader switch ON.
8. Set ADDRESS switches to 7720 (l 7720 for 8K systems).
9. Press 1/0 RESET and READIN. The computer will halt (AC= 777777).
IO. Press reader switch OFF and place binary tape of Editor in Teletype reader.
11. Press reader switch ON and press START. The Editor will be loaded into memory (this takes about 3-1/2
minutes using the ASR33 reader).

To initialize the Editor:
When loading is complete, the Editor types
EDIT-15
INTXT*
12. You type:
The Editor types:

L)
GETXT*

F-1 (Part II)

13. You type:
The Editor types:
14. You type:
The Editor types:
followed by:

L)
OUTXT*

L)
EDIT

15. You type a carriage return following the right angle bracket, and the Editor responds
INPUT

To create your ASCII text:
16. Just type it in. Each line should consist of less than 72 characters followed by a carriage return. If you make a
mistake, don't worry about it now. Make a note of it and, when you have finished creating your text, you can edit it
using basic Editor commands illustrated in Appendix D. The Editor will store your ASCII text in a 20-line buffer
(55-line buffer for 8K systems). When you have typed enough lines to fill the buffer or when you have finished
creating your text (whichever occurs first), proceed with the following steps.

To punch out the contents of the buffer:
I 7. Type a carriage return as the first character on a line. The Editor will respond
EDIT

>
I 8. You type WRITE) following the right angle bracket.
I 9. Press Teletype punch switch ON.
20. Press CONTINUE. The Editor will punch out the buffer. When punching stops
21. Press the Teletype punch switch OFF.
22. Press CONTINUE. The Editor will respond with

>
23. If you wish to continue creating text, return to step 15. If you are finished creating text, return to steps I
through 5 to generate trailer in the same manner that you generated leader.

F-2(Part II)

APPENDIX G
EXAMPLES OF EDITOR OPERATION
This appendix is intended for the user who is not acquainted with the Editor, but has an ASCII tape that he wishes to
modify. Basic editing can be accomplished using only two I/O commands (READ and WRITE) and six editing
commands (INSERT, RETYPE, DELETE, PRINT, NEXT, and TOP). Most editing jobs can be accomplished using
these simple commands; the following editing session illustrates their use.
(I) To get things started, load the Editor into memory and initialize it by following the procedure in paragraph

2.3.
(2) Make sure the Teletype reader switch is OFF.
(3) Place the tape containing the ASCII text to be edited in the Teletype reader.
(4) Type READ)
(5) The Editor will halt. Turn the Teletype reader switch ON.
The Editor will read 20 1 0 lines (55 1 0 for 8K systems) into a block buffer ltnd halt. Turn the Teletype reader switch
OFF and press CONTINUE. The Editor will type a right angle bracket (>) in the left margin. This is the signal that
the Editor has finished processing your last command (READ) and is ready to accept your next command. You can
now perform the following editing operations.
To obtain a listing of the text in the Editor's buffer:
a. Type TOP)
This moves the current-line pointer to the beginning of the buffer.
The Editor types >
b. Type PRINT ....., 22)
This commands the Editor to print 22 lines, starting with the current line. The number 22 includes the 20 lines
in the buffer, one pseudo line at the beginning of the buffer, and one extra line to make sure you get
everything.
The Editor types:
(Line I of original text)

(Line 20 of original text)
END OF BUFFER REACHED BY:
PRINT 22

>
To insert a line at the top of the buffer (before line 1 or original text):
a. TypeTOP)
The Editor types >
*If an out-of-tape condition occurs during the READ (e.g., the ASCII text is less than 20 10 lines), you can continue by typing CTRL R.

G-1 (Part II)

b. Type INSERT L..I (Line to be added before line 1)
To verify this change:
a. TypeTOP)
The Editor types>
b. Type PRINT ......, 3)

The Editor types:
(Line 1 to be added)
(Line I of original text)

>
To change line 5 instead of verifying as above:
a. Type NEXT ......, 4)
This moves the current-line pointer from line I of the original text to line 5 of the original text.
The Editor types >
b. Type PRINT)
This will allow you to verify that the current-line pointer is in fact pointing at the line you wish to modify.
The Editor types:
(Line 5 of original text)

>
c. Type RETYPE ~ (New version of line 5))
The Editor types>
To add four lines after line 7:
a. Type NEXT ~ 2
This moves the current-line pointer to line 7.
The Editor types>
b. Type INSERT)
The Editor types INPUT
You type:
(Line I to be inserted))
(Line 2 to be inserted))
(Line 3 to be inserted))
(Line 4 to be inserted))
The Editor types:

)
EDIT)

>
To verify your insertion:

G-2(Part II)

a. TypeTOP)
The Editor types>
b. Type NEXT L-1 8)
(You typed 8 as the parameter for NEXT since you wish to move the pointer down through seven lines of
original text plus the line you added at the top.)
The Editor types >
c. Type PRINT L-1 6)
The Editor types:
(Line 7 of original text)
(Line 1 to be inserted)

(Line 4 to be inserted)
(Line 8 of original text)

>
To delete lines 11, 13,and 14:
a. Type NEXT ..._. 3)
The Editor types >
b. Type PRINT)
(To verify the position of the current-line pointer.)
The Editor types:
(Line 11 of original text)

>
c. Type DELETE)
The Editor types >
(The current-line pointer is now positioned at line 12, the line after the line deleted.)
d. Type NEXT)
The Editor types>
e. Type DELETE L-1 2)
The Editor types>
Now you have added 5 lines to the original 20 lines read, deleted 3 lines, and modified 1 line. You now have 22 lines
in your buffer. To verify that this block is as you want it, issue a TOP command followed by a PRINT L...1 24
command.
To punch out the edited block:
a. Type WRITE)
The Editor halts. Turn on the Teletype punch and press CONTINUE.
The Editor punches out the block buffer (tabs will not appear in the listing) and halts.

G-3(Part II)

Turn off the Teletype punch and press CONTINUE.
The Editor types>
You now have the option of returning to step 4 to fill the buffer and edit more previously prepared text, or using the
INSERT command to fill the buffer. In either case, if the buffer is filled again, a WRITE must precede further
READs. If starting a new editing job, type CTRL P to clear all buffers before beginning.

G-4(Part II)

APPENDIX H
EDITOR DEMONSTRATION
This appendix provides a comprehensive example of an actual editing session. Before attempting to follow the
example in this appendix, the user should perform the operations outlined in Appendices F and G, and read sections
2.3 and 2.4 to familiarize himself with the Editor. The first page shows the editorial changes that are to be made to an
ASCII tape. The next several pages show hardcopy output of the editing session, and the last page shows an assembly
listing of the newly edited file. The following comments are keyed to the two pages containing the hard-copy output
of the editing session.
1) After initializing the Editor to READ text from the high-speed paper tape reader, to GET subsidiary input
from the Teletype reader, and to WRITE edited text on the high-speed punch, the user must READ a block of text
before editing can begin.

2) User types a carriage return to ignore CAPACITY WARNING.

3) The user has misspelled a command. It must be retyped in order to be executed. After the command has been
executed, PRINT is used to verify the results.
4) The use of command abbreviations saves time.
5) The user should have simply typed-

I L-1 ISZ PACK /BUMP TO "TO" ADDRESS

instead of changing from edit level to input level, typing the line, followed by another carriage return to get back to
edit level.
6) Verification is truncated with BRIEF ON, and prints out only up to the character changed. This saves time on
long editing jobs.
7) Verification can be eliminated by the confident user.
8) The end of the buffer has been reached. The user must now issue a WRITE followed by a READ to continue.

9) The editing task is now complete. The user issues a WRITE to finish punching out his edited tape.

H-l(Part II)

Input File Marked For Editing

-a

.,,,_

f) Jl)~TE~~~~o4R•

.•

~o•DJP-r;;i.,A

rre. ,. .

un=.1t
~;..:::,.. ,
tJ
a'( •1suattouT 1NE PACK, 7-BITt;;HARS TO IOPS ASCII.f:.a.,_,~ft'b~"..-~~~,!~ ~L~ ··~T
/CALL: JMS
PACK
Al.~/l(f, 1' 'f>0
,_.,.& TllC S
I
FROM
/5l"Aitl'•FjU1'VrA1''tAY,
"'~T" TE'\M~1',.Af
I
TO .Jt / •r.1. -.,- ,., ourf'CJT' A•IUy.
t'ul"P"'" Co, ,_.
PACK
~
LAC~
PACK
/GET fRQ,•l A . ESS.
0
.st.
ptel\..w;
PrRQ,'I}
/GIVE
TO
mo;
PO
NTEH.
1
LAC•
PACK
/uET AuURESS Oi TO ARRAY.
D~~,1
PTO
/GIVE TO OUTPUT POINTER.
~~
PLBH
/SAV'i. AS START ADIJRESS.
~
PACK
/dU~P TO RETURN.
PLOOPl LAW
17773
/SET UP
DAC
PK5CHR
/rCHAb,Tlt ,oftll)TIR,
PLOOP<'... ~
Pl0 ROi1
/GET NEXT WOHD IN INPUT ARRAY.
Sttl>~~
~JNOUtl(
/TER1"1INATOR7
~
J1"1P

i-'LOOP7

lSZ

PCLOS
PFkOM

IBZ:

Pflii811t , ,

DAC

PWHD3
PRALJ... JI
PK5i'R - '

Jfl!S

ISZ
JMP
ID(>OP2
LAC
PWRD2
RAL/,lL
DAC
PWRD2
LAC
PWRDl
I sz

JMP
PCLOS ~AW
OZflt MOJSAD
JMP

J°'l p,.C~~~*
• END

H-2(Part II)

1'1'tHIH T8 NEHI wen.w.

/SET UP TO ROTATE.
"'I
/5 CHARS IN t;J- •
/NO, GET ANOTHER.
/WORD PAIR COMPLETE.
/CLEAR PAIR BIT 35.

'('

JI(

1777 3

PK5CHR
Pl.OOP7

/INCO~PLETE

~!gK

/LESS END AilDRESS.
/RETURN TO CALLER •

PTO
PTCl-.-oo
PL~l

t.»~"""'~'
.
n.r>f'~#, lAIJ
Pl.Ii~
CMA

/YES, GO CLOSE OUTPUT ARRAY.
/POINT TO NEXT WORD.

/GET FIRST WD OF PA~.
/oIT 0 OF WD 2.
~\.('
/INSERT FIRST WD IN OUT ARRAY.
/a Ui1P OUT ADDRESS.
/GO SET UP NEXT PAIR.
/i1AKE SURE PAIR IS CO~PLETE.

RAL~

DAC*

+tiQ; il<I12.'4'

PAIH.
/FORM WORD PAIR COUNTA""'>
/START ADDRESS.
'C/

Hard-Copy Output of Editing Session

EDIT-15/10
I NlXT*

GETXT*
L

OU TX I"'

Eull

>KiAD
>f!ND /SUdROUT
/SUBROUTINE PACK, 7-dil CHARS 10 !OPS ASCII.
>OVERLAY 1
ItWuT

/SUBROUTINE PACK, 7-dlT LEFT-ALJJUSTED CHARS TO NON-HEADERED !OPS
/~5CII.
ON riETURN, AC HOLJS TOTAL WORDS OCCUPIEu &Y PACKEJ ARRI
CAPACITY ~ARNING
/A WOriD OF ALL l'S MUSI TERMINATE THE INPUT CUNPACKEJ) ARRAY.
CAPACITY wARNING
EUil

>LOCATE FHO;'l
I

FHOi'1

/START OF INPUT ARRAY.
>NEXT
>APENU
/START OF OUTPUT ARRAY.
NOT A REQUEST:
APEND
/START OF OUTPUT ARRAY.
>APPENu
/START OF OUTPUT ARRAY.
>PRINT 1
I
TO
/START OF OUTPUT ARRAY.
>L LAC
LAC
PACK
/uET FROi"i ADRESS.
>CtiANGE LAC/LAC*/
LAC
PACK
/GET FROl1 AurlC:SS.
>CHANG ii: /LAC/LAC*/
PACK
LAC*
/GET iRO"'I AiJttESS.
>C I i-;DR/ ttDDR/
PACK
LAC*
/GET rnoM AuDrlESS.
>NEXT 1
>INSERT
CAPACITY wARNING
>APP~Nu

I NPLIT

ISZ PACK

/bU~P

TO "IO"

AuJ~ESS.

Cf;PACIIY WARNING

H-3(Part II)

Hard-Copy Output of Editing Session (cont)

EDIT

>PitlNI

ISZ PACK
/BUMP TO "TO" AUDRESS.
>BklEF ON
>C .I .I.
ISZ PACK
>Pt(! NI
ISZ PACK
1au~P TO "To" ADDRESS.
>L PLi3H
DAC
PLBH
>BHI~F OFF
>PHI NI
DAC
PLBH
/SAVE AS START ADDRESS.
>DELETE 2
>fHINI
PLOOPl LAW
17773
/SET UP
>N l
>A
/5-CHARACTER COUNTEH.
>L C
/IEfMI NA TOR 7
SAC
C-1
>VERIFY OFF
>C /SAC/SAD/
>C /8\C-1/ENDCHR/
>V ON
>PRINT
/TEiMINATOR?
SAD
ENDCHR
>N
>D

>N
>P
ISZ

PF'ROi"I

/POINT TO NEXT WORD.

ISZ

PFR01'1

/POINT TO NEXT WORD.

>!)

>P
>f PL

ENLI OF BUFFER REACHEi.> BY:
PL
>W
>nEAD

>i t'L

PLOOf'7 Jf"IS
PHAL7
>N
>HETYPE ISZ PK5CHR
>N
>C ,LP,PL,
JMP
PLOOP2
>N 2
>CHANGE /LILI CLL
RALi CLL

H-4(Part II)

5 \/5 CHARS IN?

/NO, GET ANOTHER.
/CLEAR PAIR BIT 35.

Hard-Copy Output of Editing Session (cont)

>N 2
>P
LAC
PWRDl
>C IH.IIrl/
LAC
Plllrl D1
>L HAL
HALlCLL
>C I I CLL//
RAL
>L J,1P
J1'1P
PLi00Pl
>CHANGE /ld0/00/
JMP
PLOOP1
>N
>

/GET FIRST lllD OF PAR.•
/1.:iET FIRST Wu Or PAIR.
/olT 0 OF Wl.> 2.
/olT 0

OF WiJ 2.
/GO SET UP NEXT PAIR.
/GO SET UP NEXT PAIR.

1 NPUT

DZt'1 PWRD3

/FILL PAIH WITH ZEROES.

EDIT
>L
CLAICLL

>R ENDCHH

>N
>R
>L PTO

TAD• PACK

TAD
PTO
>INSERT ISZ PACL\K
CAPACITY WARNING
>BOTI01•l
>OllERLAY
INPUT
Piri01VJ

WORD PAIR COUNT
/FORM WORD PAIR COUNT.

/FOR~

LAW -1

/START ADDRESS.
/LESS END ADDRESS.

.ENiJ
IO

CltPACIIY WARNING
PTO
CBPACITY WARNING
PK5CHR iO
CAPACITY WARNING
.END
CAPACITY WARNING
EDIT
>Ii/RITE

H-S(Part II)

Edited Output File

PLEASE READY THE INPUT DEVICE AND SET TH~ AC SWITCH
/SUBROUTINE PACK, 7-BIT LErT-AUJUSIED CHARS TO NON-HEADERED IOPS
/ASCII. ON RETURN, AC HOLDS TOTAL WORDS OCCUPIEu BY PACKEU ARRAY.
/A ~OrlD OF ALL 1"5 MUST TERMINATE THE INPUT <UNPACKED> ARRAY.
/CALL: J~S
PACK
I
flWi'<l
/START OF INPUT ARRAY.
I
IO
/START Or OUTPUT ARRAY.
PACK
JO
LAC*
PACK
/GET FROM ADDRESS.
DAC
PFH0:1
/GIVE TO FRO~ POINTER.
ISZ PACK
/SUMP TO "IO" AUURESS.
LAC*
PACK
/GET ADDRESS OF TO ARRAY.
DAC
PTO
/GIVE TO OUTPUT POINTER.
PLOOPl LAw
17773
/SET UP
DAC
PK5CHR /$-CHARACTER COUNTER.
/GET NEXT WORD IN INPUT ARRAY.
PLOOP2 LAC*
PFROM
SAD
ENDCHR
/TER1'1INATOR?
J1'<1P
PCLOS
/YES, GO CLOSE OUTPUT ARRAY.
ISZ
Pl'ROM
/POINT TO NEXT WORD.
DAC
PWRD3
/SET UP TO ROTATE.
PLOOP7 JMS
PRAL7
/5 CHARS IN?
ISZ PKSCHR
/NO, GET ANOTHER.
Jl'lP
PLOOP2
/WORD PAIR COMPLETE.
LAC
PWRD2
/CLEAR PAIR BIT 35.
RALi CLL
PWRD2
DAC
/GET FIRST WD OF PAIR.
PWrlD)
LAC
/BIT 0 OF WD 2.
RAL
/INSERT FIRST WD IN OUT ARRAY.
DAC*
PTO
/BU~P OUT ADDRESS.
ISZ
PIO
/GO SET UP NEXT PAIR.
J1'1P
PLOOP 1
/1'1AKE SURE PAIR IS COi'1PLETE.
PCLOS
LA~
17773
/FILL PAIH WITH ZEROES.
DZi'I PWRD3
SAD
PK5CHR
/INCO~PLETE PAIR.
JMP
PLOOP7
ENiJCHR LAVI -1
/FORM WORD PAIR COUNT.
/START AUDRESS.
TAD* PACK
/STA.rt! ADDRESS.
TAD
PLBH
C~A

TAD
PTO
ISZ PACK
J,'1P*
PACK
PFl10i'l

PTO
P!<5CHR

.END

H-6(Part II)

/LESS END ADDRESS.
/RETURN TO CALLER.

APPENDIX I
ASCil CHARACTER SET
MODEL 33ASR TELETYPE CODE (ASCII) IN OCTAL FORM
Character

A
B

c
D

E
F
G
H
I
J
K

L
M
N
0

p
Q

u
v

z
0
1
2
3
4
5

6
7
8
9

8-BitCode
(in Octal)

301
302
303
304
305
306
307
310
31 l
312
313
314
315
316
317
320
321
322
323
324
325
326
327
330
331
332

Character

#
$
%
&

'
(
)

*
+
'

I
:

<
=

>
?
@

[

J
t

260

261
262
263
264
265
266
267
270*
271*

Leader/Trailer
Line-Feed O.)
Carriage-Retumi)>
Space (A)
Rub-out
Blank (Null)

ALT MODE($)
Tab

8-BitCode
(in Octal)

241*
242*
243*
244
245*
246*
247*
250*
251*
252*
253*
254*
255*
256*
257
272*
273
274*
275*
276*
277
300*t
333*
334*
335*
336
337
200**
2I 2t
215t
240
377**
000**
373
21 It

*Not recognized by ODT.
**Ignored by ODT.
tHave special meaning to TIC & TOC.

1-1/1-2 (Part II)

APPENDIX J
ODT COMMAND SUMMARY
Register Examination and Modification

k/
I
)
k)

.t.
t
$A
$1

Open register k.
Open last referenced register.
Close register.
Store k and close register.
Close register and open next register .
Close register and open previous register.
Close register and interpret contents as address.
Open AC and Link registers.
Display PI and API status.

Execute Instruction
k$X

Execute instruction k.

Setting User Start Address
$Z
$G
k$G

Open user start address.
Start program at address specified in $Z register.
Start program at location k.

Breakpoint Commands
k$B
k$nB
$B
$nB
$C
k$C

$K
$U

$V
$Y

Set breakpoint I at location k.
Set breakpoint n at location k.
Remove all breakpoints.
Remove breakpoint n.
Execute instruction at breakpoint and continue
with user program.
Execute instruction at breakpoint and continue
with user program when the breakpoint has been
encountered for the kth time.
Disable PI and API during all breakpoints.
Enable PI and API during all breakpoints.
Vary autoindex (must be between l 0 and 17).
Open re-entrant PC list (after re-entrancy error).

Searching Operations
$M
k$W

Open mask register for modification.
Search for bit configuration k.

J-1 (Part II)

Initialize Buffers
$1
k$I

Initialize (set) block of memory locations to zero.*
Initialize (set) block of memory location to k. *

Paper Tape Output

$H
$L
k1; k2 $D
$T
k$T

Select high-speed punch.
Select low-speed punch.
Dump (punch) all locations between k 1 and k2 ,
inclusive.
Punch terminal block; loader will halt when
loading complete.
Punch terminal block; loader will transfer to
location k when loading complete.

*User must first specify high and low limits of block by means of$M command.
**If low-speed punch ($L), ODT halts to allow user to turn-on punch.

J-2(Part II)

PARTW
Basic 1/0 MONITOR

.-1~-·

PART Ill
BASIC I/0 MONITOR

CHAPTER I
THE PDP-15 MONITOR ENVIRONMENT

1.1 MONITOR FUNCTIONS
The Basic 1/0 Monitor greatly simplifies the task of programming 1/0 functions by providing an interface between
system or user programs and the external world of 1/0 devices. Upward compatibility exists between the Basic 1/0
Monitor and the other monitors of the PDP-15 Software System; programs written to operate under control of the
Basic 1/0 Monitor will also operate, without modification, under control of the Advanced and
Background/Foreground Monitors. The Monitors, by means of the Input/Output Programming System (IOPS) and
Program Interrupt (Pl) or Automatic Priority Interrupt (API), allow simultaneous operation of multiple 1/0 devices
along with overlapping computations.

Certain features such as the general Monitor environment, data handling, and logical/physical 1/0 device associations,
are common to all three Monitors. Detailed information on the Advanced and Background/Foreground Monitors will
be found in manuals DEC-15-MR2A-D and DEC-15-MR3A-D respectively.

1.1.2 General 1/0 Communication
A general communication required to accomplish an 1/0 task is the same for all three Monitor systems (see Figure
1-1). A system or user program initiates an 1/0 function by means of a Monitor command (system macro), which is
interpreted by a CAL handler within the Monitor as a legitimate 1/0 call. (See the PDP-15 Reference Manual
DEC-15-BRZA-D for a description of the CAL instruction.) The 1/0 call includes a logical 1/0 device number as one
of its arguments. The Monitor establishes the logical/physical 1/0 device association by means of its Device
Assignment Table (.DAT). When this has been accomplished, the Monitor passes control to the appropriate device
handler subroutine to initiate the 1/0 function and returns control to the system or user program. The system or user
program retains control until an interrupt (Pl or API) occurs, at which time it relinquishes control to the device
handler to perform and/or complete the specified 1/0 function. Computations or other processing can be performed
by the system or user program while waiting for an interrupt. This feature allows the programmer to make optimum
use of available time.

1-1 (Part III)

DATA

j_
SYSTEM OR
USER PROGRAM

VIA CAL
HANDLER

Vi A CAL
MONITOR

l
I/0 DEVICE HANDLER

INITIATION

CONTROL RETURN

VIA PI
OR API
I/0 DEVICE

INTERIWPT

Figure 1-1. General I/O Communication in Monitor Environment

1.1.3 Command, Control, and Data Flow
Figure 1-2 provides a more detailed representation of the Monitor environment, with emphasis on command, control,
and data flow. As shown, the user can initiate a command via the Teletype which is interpreted by a Command
Processor within the system program (or user program if so designed).

Each system or user program must internally set up line buffers (except when using Dump mode, discussed later) to
be used in transmitting data to or from the external environment. Each line buffer of n words consists of a two-word
header (referred to as a header word pair) and n -2 words of data. The system or user program can exercise control on
output by modifying the header word pair, or it can verify on input by examining the header word pair. The use of
line buffers is discussed in more detail later in this chapter.

Monitor 1/0 commands (system macros) are written as part of the system or user program. In FORTRAN IV source
programs, these commands are in the form of READ and WRITE statements (refer to the PDP-IS FORTRAN IV
Manual, DEC-1 S-KFZA-D). These statements are translated by the compiler into the proper calling sequences for the
FORTRAN Object Time System which provides the required Monitor calls at execution time. In MACRO-IS source
programs, Monitor 1/0 commands are written as system macros within the system or user program. These system
macros are expanded at assembly time and include a CAL initiated Monitor call that contains the logical device
number as one of the arguments.

At execution time, Monitor calls are processed by the CAL Handler within the Monitor. Non-I/0 functions are then
further processed by the Monitor Control routine and I/O functions are processed by the I/O Control routine (see
Figures 1-2 and 1-3). A complete description of each of these commands is given in Chapter 2. If the original
command involved is an I/O function, the I/O control routine checks the Device Assignment Table to associate the
logical I/O device (specified by the system macro) to a physical I/0 device. The logical/physical device associations
can be modfied either by reassembly or with the aid of utility program PUNCH l S.

When the logical/physical 1/0 device association has been established, the Monitor passes control to the appropriate
1/0 device handler which initializes itself, initiates I/O, and returns control to the system or user program. As
mentioned previously, the system or user program retains control until the specified device causes an interrupt (Pl or
API). At this point, it relinquishes control to the device handler to continue or complete the specified 1/0 operation.

l-2(Part III)

MONITOR ENVIRONMENT

EXTERNAL ENVIRONMENT

ERROR MESSAGES AND COMMAND ACKNOWLEDGEMENTS

USER

ERROR
MESSAGES

COMMANOS
>-----+--TELETYPE
COMMA NOS
11
ACKNOWLEDGEMENTS

•

...

SYSTEM
OR USER
PROGRAM

COMMAND
PROCESSOR

MON I TOR ERROR
DIAGNOSTIC (,MEDl
PROGRAM

OUTPUT CONTROL

I
I
I

HEADER

MON I TOR

INPUT
VERIFICATION

TERRORS

MONITOR
COMMANDS

WORD PAIR

DEVICE ASSIGNMENT
TABLE I.DAT l

(SYSTEM MACROS)

Loo.

I OATA

DATA

LINE
BUFFER

I
I
I
I
I
I
I

~ONTROL

CAL
HANDLER

. .1J

NON !/0
FUNCTIONS
~~~~~

MONI TORJ
CONTROL
ROUTINE

1-I/O FUNCTIONS~
ERRORS

-1
l /0 CONTROL
ROUTINE

MONITOR

I
I

11;,;s- - - - - I

IL _____ _

I
I
I

INITIALIZATION

1/0
OE VICE

DATA AND CONTROL

INITIATION

INTERRUPT

ERRORS

I
.1

I/0 DEVICE HANDLER

L __________ _

_ _ _ _ _ _ _J

I
I

INPUT/OUTPUT PROGRAMMING SYSTEM (!OPS)

Figure 1-2. Command, Control, and Data Flow in the Basic 1/0 Monitor Environment

Function
Code

Command

.INIT
2

.DLETE, .RENAM, and .FSTAT

.SEEK

.ENTER

.CLEAR

.CLOSE

.MTAPE

.READ and .REALR

:WRITE and .REALW

.WAIT and .WAITR

.TRAN

.TIMER

Functions processed by

.EXIT

monitor control routine

.SETUP

.IDLE and .IDLEC

.RLXIT

Functions processed
by 1/0 control routine

Figure 1-3. Monitor Commands and Function Codes

NOTE

.INIT, .READ, .WRITE, .WAIT, .WAITR, .CLOSE, .TIMER, and .EXIT are
recognized by all three Monitors.
.SEEK, .ENTER, .FSTAT, .RENAM, .DLETE, .TRAN, .CLEAR, and
.MTAPE are recognized by the Advanced and Background/Foreground
Monitors (and are ignored by the Basic 1/0 Monitor) .
.SETUP is used by the Monitors in setting up the 1/0 skip chain and API
channel registers (see Chapter 4) .
.IDLE, .IDLEC and .RLXIT are recognized by the Background/Foreground
Monitor only.

l-4(Part III)

In either case, control is returned to the system or user program at the point where it was interrupted. The system or
user program, by means of a .WAIT system macro (described in Chapter 2), can determine whether an input or output
operation has been completed. If the transfer of data from or to the system or user program line buffer has been
completed, program execution continues; if the transfer has not been completed, control is returned to the .WAIT
macro.
Additional buffering is provided by the individual device handlers as required. All device handlers are non-resident in
the sense that only those handlers required by the system or user program are loaded into core.

l.2 LINE BUFFERS
As mentioned in the preceding general description of the Monitor environment, each system or user program must
internally set up line buffers to be used in transmitting data to or from the external environment. An exception to
this rule is when data is transmitted in the Dump mode (described in paragraph 1.3.1.4). Each line buffer of n words
(always even) should be set up to consist of a two-word header (termed a header word pair) followed by n-2 words of
data as shown in Figure 1-4.

Word 0

First Word of Line Buffer Header

Word l

Second Word of Line Buffer Header

Word 2

First Word of Data Area

Word n-1

Last Word of Data Area

Figure 1-4. Line Buffer Structure

A system or user program should contain at least one line buffer for each device that is to be used. This buffer is used
to set up output lines before transmittal to an output device, or to receive input lines from the associated input
device. The Monitor accepts commands (system macros) from system or user programs to initiate input to the line
buffers and to write out the contents of line buffers. Complete descriptions of these commands are given in Chapter
2. Line buffers are internal to, and must be defined by, each system or user program. The header word pair within a
line buffer is detailed in Figure 1-5. The .BLOCK pseudo operation may be used to reserve space for a line buffer. A
tag is required to allow referencing by individual .READ and .WRITE macros. For example:
.DEC
LINEIN

.BLOCK 52

/creates 52-word line
/buffer named LINEIN.

LINOUT

.BLOCK 52

/creates 52-word line
/buffer named LINOUT.

Before output, the user must set the appropriate word pair count in bits 1 through 8 of word zero in the line buffer if
they have not already been set by a device handler on input. This count overrides the word count passed to IOPS by

1-S(Part III)

0
HEADER,
WORD 0

1-1GNORE CHECKSUM
ON BINARY INPUT

---···--8

I l

9---11

12. 13

T----------1----.

fOUNT
I

!
,

14----17

v
I

110 MODE
I

.r--'~-----..------

:~:gE:A~~;g~~I~· INCLUDING ------~
FOR !OPS BINARY ONLY
}
I CORRESPONDS TO 7-9 PUNCH
ON BINARY CARDSI

---------------_,

VALIDITY BITS:
00 • DATA CORRECT
01 • PAR! TY ERROR

10 • CHECK SUM ERROR }
11 = SHORT LINE
(BUFFER OVERFLOW)

110 MOOE.
0000 = !OPS BINARY
000 1 = IMAGE BINARY
0010 •!OPS ASCII
00 1 I • IMAGE ALPHANUMERIC

-----------'

0100 = DUMP
}
0101 = EOF (LOGICAL)*
0110= EOM (PHYSICAL!*
0111 =TAPE LABEL*

-------------'

* !OPS AND IMAGE MODES ONLY
MM ALSO DUMP MODE FOR 9 CHANNEL
MAGNETIC 4 APES
0---------------------------~17

Hf ADER,
WORD1

CHECKSUM•
TWO'S COMPLl;MENT OF HEADER WORD 0 PLUS DATA
WORDS CO= CHECKSUM NOT COMPUTED!

---~

Figure 1-5. Format of Header Word Pair

the .WRITE macro. (The word count must still be specified in the .WRITE macro for each data mode; however, it
only has meaning in Dump mode since there is no header word pair.) In IOPS binary mode (discussed in Paragraph
1.3.l.2), bits 9 through 11 should be set to IOI if the output will ultimately be on cards. The checksum word, the
second word in the header, need not be set by the user since checksums are computed by IOPS.

Before input, the user should not be concerned with the header word pair since they will be set by IOPS to enable the
user to determine what has happened after input has terminated.
On input, the word count specified in the .READ macro is used by IOPS to determine the maximum number of
locations to be occupied by the data being read. If the word count is exceeded before input is terminated, or if there
is a parity or checksum error, IOPS sets the appropriate validity bits in header word 0 to indicate the error.
After input, the user should check the validity bits in word 0 of the line buffer header to determine if the data was
read without error. If multiple errors are detected, priority is given to a parity error over a checksum error. IOPS
ignores checksum errors on binary input if bit 0 of word 0 of the line buffer header is set to 1. IOPS sets the 1/0
mode bits (bits 14 through 17 of word 0 of the line buffer header) to: 6 (01102 ) if it senses a physical end-of-medium
(such as end-of-tape in the paper-tape reader), or 5 (0101 2 ) if it senses a logical end-of-file during an IOPS binary
read.

l -6(Part III)

When choosing a word count (that is, the maximum line buffer size) to specify in system macros, both the set of
possible devices and the mode of data transmission must be considered. The maximum line buffer sizes (including
2-word header) for standard peripheral devices, along with applicable data modes, are listed in Table 1-1.
Table 1-1. Maximum Line Buffer Sizes
Maximum Line
Buffer Size

Device

Data Modes*

Notes

PR (paper tape reader)

All

34 1 0 sufficient only if mode 2. Headers
accepted for mode O; generated for modes 1,
2& 3.

PP (paper tape punch)

All

341 0 sufficient only if mode 2. Headers
output only for mode 0.

TT (Teletype)

2 & 3 only

Allows for 80 10 characters.
Headers generated on input.
Headers not output on output.

CD (card reader)

2 only

Headers generated on input.

LP (line printer)

2 only

Allows for 125 1 0 characters. No headers
output.

MT0-7 (magnetic tape)

0, 2& 5

Modes 0 and 2 allow for several line buffers
(logical records) per physical block.

*See paragraph 1.3 below.

1.3 DATA MODES
The Input/Output Programming System allows data transmission to or from a system or user program in five different
modes.

Mode
IOPS Binary

Code*
0

Image Binary
IOPS ASCII

Image Alphanumeric

Dump

9-Channel Dump

*Bits 14 through 17 of Header Word 0, specified by system macro and set by IOPS.

I-7(Part III)

1.3.1 IOPS Modes

The two IOPS data modes include IOPS ASCII and IOPS binary as shown in Figure 1-6 on paper tape, and de"scribed
in the following paragraphs.
TAPE CHANNEL
87654
321
FEED

DIRECT I ONl
OF APE
MOVEMENT

0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0

0 0

------

0 0 0
0 0 0
0 0 0

0 0

0
0

t.____ 7-BIT ASCII CODE

l.__________ PARITY BIT lEVEN PARITY)
IOPS ASCII
TAPE CHANNEL

8 7 6 5 4

3 2

FEED

DIRECTION]
OF TAPE
MOVEMENT

e 0 0 0 0 o 0 0 0
e O o o O o O O O
e o o o O o o o O
•

0 0 0 0

t•I 6-BIT BYTE
2nd 6-BIT BYTE
3rd 6-BIT BYTE

0 0 0 0

1 t

,_.,,, " """ "°"
PARITY BIT lODD PARITY)

' - - - - - - - - - MUST ALWAYS BE PUNCHED
lOPS BINARY

Figure 1-6. IOPS Mode Data on Paper Tape

1.3.1.1 IOPS ASCII - 7-bit ASCII is used by IOPS to accommodate the entire 128-character revised ASCII set
(Appendix A). All alphanumeric data, whatever its original form on input (ASCII, Hollerith, etc.) or final form on
output, is converted internally and stored as 5/7 ASCII. "5/7 ASCII" refers to the internal packing and storage
scheme. Five 7-bit ASCII characters are packed in two contiguous locations as shown in Figure 1-7 and can be stored
as binary data on any bulk storage device. Input requests involving IOPS ASCII should be made with an even word
count to accommodate the paired input.
ASCII data is input to or output from IOPS ordinarily, via the Teletype or paper tape, although it may exist in 5/7
ASCII form on any mass storage device. IOPS ASCII is defined as a 7-bit ASCII character with even parity in the
eighth (high order) bit, in keeping with USA standards. IOPS performs a parity check on input of IOPS ASCII data
prior to the 5/7 packing. On output, IOPS generates the correct parity.
Non-parity IOPS ASCII occurs in data originating at a Model 33, 35, or 37 teletype, without the parity option. This
data always appears with the eighth (high order) bit set to 1. Apart from parity checking, the IOPS routines handle
IOPS ASCII and non-parity IOPS ASCII data identically.

1-8(Part III)

1ST CHARACTER

WORD 0

21'40 CHARACTER

3RD CHARACTER
1-4

o--.

WORD 1 3RD CHARACTER
5-7

3-----------9

10 - - - - - - - - · 16

4TH CHARACTER

5TH CHARACTER

4 l - UNUSED

Figure 1-7. 5/7 ASCII Packing Scheme

An alphanumeric line consists of an initial form control character (line feed, vertical tab, or form feed), the body of
the line, and a carriage return (CR) or ALT MODE. CR (or ALT MODE) is a required line terminator in IOPS ASCII
mode. Control character scanning is performed by some device handlers for editing or control purposes (see paragraph
4.4 for effects of control characters on specific devices).

1.3. l.2 IOPS Binary - IOPS Binary data is blocked in an even number of words, with each block preceded by a
two-word header. On paper tape (see Figure 1-6), IOPS binary uses six bits per frame, with the eighth channel always
set to 1, and the seventh channel containing the parity bit (odd parity) for channels 1 through 6 and channel 8. The
parity feature supplements the checksumming as a data validity provision it?- paper tape IOPS binary.

1.3. l.3 Image Modes - Image Mode data is read, written, and stored in the binary or alphanumeric form of the
source or terminal device, one character per word, as shown in Figures 1-8 and 1-9. No conversion, checking, or
packing is permitted, and character scanning is generally omitted.

1.3.1.4 Dump Mode - Dump mode data is always binary. Dump mode is used to output from or load directly into
any core memory area, bypassing the use of line buffers. Each dump mode statement has arguments defining the core
memory area to be dumped. Dump mode is normally used with bulk storage devices, although it is also possible to use
it with paper tape output and input.

1.3.l.5 9-Channel Dump Mode - This mode is always binary and is exclusively used with 9 channel magnetic tape
transports. This mode is designed to take advantage of all 8 data bits in each frame of the tape and thus insure
maximum thruput for each word transferred. A full discussion of this data mode is provided in paragraph 4.4.6.

1.3.2 Input/Output Data Mode Terminators
Input/output terminators for each of the data modes are summarized in Table 1-2.

1.4 SYSTEM TABLES

System tables used by each of the monitor systems include the Device Assignment Table (.DAT), and the System
Communication Table (.SCOM). These tables are discussed in the following paragraphs.

l-9(Part III)

TA PE CHANNEL

a 1 6 5 4

3 2

FEED

oooocoooe
o e o c o • o e o
e o o o e o e o c

DIRECTION!
OF TAPE
MOVEMENT

0 Q 0

0 0

y
L A L L EIGHT CHANNELS USED
I MAGE ALPHANUMERIC

TAPE CHANNEL
87654
321
FEED

DIRECTION
OF TAPE
MOVEMENT

• o e o e

c o •

•

• o e o e o e o e
•

•

• 0

•

0 & 0

t
·o;

6-BIT 81 NARY CODE
13 FRAMES/WORD)
IGNORED

L - - - - - - - - MUST ALWAYS BE PUNCHED
I MAGE Bl NARY

Figure 1-8. Image Mode Data on Paper Tape

WORD COUNT

~------~ ~ ~~~~E:..IR
A
C

C
O

/a
3

ORD C 0 UNT

HEADER
WORD PAIR

ABC;

IN5/7ASCII

a
c

;

!OPS ASCII

ABC
FOUR 8-BIT
CHARACTERS
!RIGHT
JUSTIFIED)

IMAGE ALPHANUMERIC

Figure 1-9. IOPS ASCII and Image Alphanumeric Data in Line Buffers

1.4.l Device Assignment Table (.DAT)
Both FORTRAN IV and MACR0-15 coded user programs, as well as the system programs, specify 1/0 operations
with commands to logical I/O devices. One of the Monitor's functions is to relate these logical units to physical

1-1 O(Part III)

Table 1-2. Input/Output Data Mode Terminators

I
N
p

IOPS
ASCII

IOPS
Binary

Carriage Return

Word Pair Count

ALT MODE

Image
Alphanumeric

Image
Binary

Dump

Word Count

End of Medium

Word Pair Count**

Word Count*

End of File**

Word Count

End of Medium

End of File**

Word Pair Count

Word Count

9-Channel
Dump

Word Count*
End of File**

Carriage Return

u
T

ALT MODE

Word Pair Count Word Pair Count

u Word Pair Count***

(except on teletype)

*A short Line Indicator will be placed by IOPS into the validity bits ( 12 and 13) of Header Word 0 if the maximum size of the line
buffer is reached before an End-0f-File.
**Bulk storage only.
***If the word pair count is not greater than I, the output line is ignored. If the word pair count is greater than 1, it has no effect and a
carriage return or ALT MODE are the only legal line terminators.

devices. To do this, the Monitor contains a Device Assignment Table (.DAT) which has "slot" numbers that
correspond directly to logical 1/0 device numbers. Each .DAT slot contains the physical device unit number (if
applicable) along with a pointer to the appropriate device handler.
All I/O communication in the Monitor environment is accomplished by the logical/physical device associations
provided by the Device Assignment Table (see paragraph 3.3.2).
1.4.2 System Communication Table (.SCOM)
The System Communication Table (.SCOM) provides a list of registers that can be referenced by the Monitor, IOPS,
and system programs. A complete list of .SCOM entries, and the purpose of each, is given in Table 1-3. The System
Communication Table begins at location I 008 •
1.5 SPECIFYING DEVICES USED TO LINKING LOADER

When writing a MACR0-15 program that uses Monitor commands (system macros), it is necessary to use the .IODEV
pseudo operation somewhere in the program to specify to the Linking Loader which .DAT slots are to be used. The
.IODEV pseudo-op causes a code to be generated that is recognized by the Linking Loader and used to load device
handlers associated with specified .DAT slots. FORTRAN IV programs cause the compiler to generate this code based

1-11 (Part III)

Table 1-3. System Communication Table (.SCOM) Entries

Purpose

Word
.SCOM

First free register below resident portion of System Bootstrap (constant)

.SCOM+ 1

First free register above resident Monitor (constant)

.SCOM+2

First free register

.SCOM+3

Last free register

.SCOM+4

Hardware options available:
Bit 0
Bit 1
Bit 2
Bit 3
Bit 6
Bit 7

l = API
1 = EAE
1 =TTY is 35/37

1 = Non-resident Monitor in core
I = 9-channel, 0 = 7-channel Magnetic tape
l = Page Mode, 0 "" Bank Mode

.SCOM+5

System program starting location

.SCOM+ 6

User starting location (bits 3 through 17), and:
Bit 0
Bit 1
Bit 2

.SCOM + 7-11 8

1 =DDT Load
1 = G Load
1 = No-symbol-table Load

Device numbers of Linking Loader's devices. These are used to avoid loading user handlers already
in core for the Loader itself.

. SCOM + 12-158 Transfer vectors associated with API software level channel registers 40 through 43 8 •
.SCOM + 16

Contains PC on keyboard interrupts.

.SCOM + 17

Contains AC on keyboard interrupts.

.SCOM +20

Bit 0
Bit 3-17

1 "" 12K, 20K or 28K System

=First Free Register in extra 4K Page

on the units specified in READ and WRITE statements. If a variable is used in a FORTRAN program to specify an
1/0 unit, handlers will be loaded for all positive .DAT slots that have handlers assigned. The .IODEV pseudo-op has
the following form:
.IODEV 3, 5, 6
where the MACR0-15 program containing this statement can use .DAT slots 3, 5, and 6. An error message is
generated if a slot called for by a program is unassigned. The MACR0-15 Assembler Manual (DEC-15-AMZA-D)
provides a full discussion of pseudo-ops.

l- l 2(Part III)

CHAPTER 2
USER PROGRAM COMMANDS (SYSTEM MACROS)

User program commands or system macros that apply to the Basic 1/0 Monitor are described in this chapter for
convenient reference. Because of the upward compatability of Monitor systems, all 1/0 Monitor commands (system
macros) are also used in the Advanced and Background/Foreground Monitor environments.

NOTE

When executing a system macro, the monitor makes no attempt to save the
user's accumulator and link bit.

2.1 BASIC 1/0 MONITOR COMMANDS (SYSTEM MACROS)
The following commands are available for use in programs that are to operate in the Basic 1/0 Monitor environment.
Each command is described in detail in the paragraphs that follow.

Purpose

Name
.INIT

Initializes the device and device handler.

.READ

Transfers data from the device to the line buffer.

.WRITE

Transfers data from the line buffer to the device .

.WAIT

Checks availability of the user's line buffer and waits if busy .

.WAITR

Checks availability of the user's line buffer, and provides transfer
address for busy return.

.CLOSE

Terminates use of a file .

2-1 (Part III)

Name

Purpose

.TIMER

Calls and uses real-time clock .

.EXIT

Returns control to the Monitor.

2.2 INIT (INITIALIZE)
FORM:

.INIT a, F, R

VARIABLES:

a= Device Assignment Table (.DAT) slot number (in octal radix)
F = File Type:

{ 0 = Input File
l =Output File

R = User Restart Address* (should be in every .INIT statement)
EXPANSION:

LOC

CAL+ F 7_s + a9-17

LOC+ 1

/The CAL handler will place the unit number (if
/applicable) associated with .DAT slot a into bits
/0 through 2 of this word.**

LOC+2

LOC+3

/Maximum size of line buffer associated with .DAT
/slot a for example, 255 10 for DECtape.***

DESCRIPTION: The macro .INIT causes the device and device handler associated with .DAT slot a to be initialized .
.INIT must be given prior to any 1/0 commands referencing .DAT slot a; a separate .INIT command must be gi\;en for
each .DAT slot referenced by the program. Each initialized .DAT slot constitutes an open file to the device handler
and must be .CLOSEd. Since a .DAT slot may refer to only one type of file (input or output), only one file type
specification (0 or l) may be made in an .INIT statement. If a .DAT slot first references an input file, then an output
file (or vice versa), a second .INIT command must be executed to change the transfer direction prior to the actual
data transfer command.
2.3 READ
FORM:

.READ a, M, L, W

VARIABLES:

a= .DAT slot number (octal radix)

M =Data mode

0 = lOPS Binary
1 = Image Binary
2 = IOPS ASCII
3 = Image Alphanumeric
4= Dump Mode

L = Line Buffer address
W =Line buffer word count (decimal radix), including the two-word header
*Has meaning only for .INIT commands referencing slots used by Teletype (the last .INIT command encountered for any slot
referencing the keyboard or teleprinter takes precedence). When the user types CTRL P, control is transferred to R. For example, the
Linking Loader takes advantage of this feature to restart the system when a new medium has been placed in the input device.
**Has no direct effect upon the user's program, but should be noted so that no attempt will be made to use LOC +I as a constant.
***Size is returned by the handler so that the program, in a device-independent environment, can use it to properly set up line buffers.

Page 2-2(Part III)

EXPANSION:

LOC
LOC+ I

LOC+2

.DEC
LOC+3

/CAL Handler will place unit number (if applicable)
/into bits 0 through 2.

/Decimal radix

-W

DESCRIPTION: The .READ command is used to transfer the next line of data from the device assigned to .DAT slot
a to the line buffer in the user's program. In the operation, M defines the mode of the data to be transferred; L is the
address of the line buffer; and W is the number of words in the line buffer (including the two-word header).
Since 1/0 operations and internal data transfers may proceed asynchronously with computation, a .WAIT command
must be used after a .READ command before the user attempts to use the data in the line buffer or to read another
line into it.
When a .READ (non-dump mode) has been completed, the program should interrogate bits 12 through 13 of the first
word of the line buffer header to ascertain that the line was read without error. Bits 14 through 17 should be checked
for end-of-medium and end-of-file conditions.

2.4 .WRITE
FORM:

WRITE

a, M, L, W

VARIABLES:

a= .DAT slot number (octal radix)

M= Data mode

0 = IOPS Binary
l = Image Binary
2 = IOPS ASCII
3 = Image Alphanumeric
4= Dump Mode

L = Line buffer address
W =Line buffer word count (decimal radix), including the two-word header
EXPANSION:

LOC

CAL+ M6-s + a9-11

LOC+ 1

LOC+2

DEC
LOC+3

/CAL Handler will place the unit number (if appli/cable) associated with .DAT slot a into bits
/0 through 2.

/Decimal radix

-W

2-3(Part III)

DESCRIPTION: .WRITE is used to transfer a line of data from the user's line buffer to the device associated with
DAT slot a.
WAIT must be used after a .WRITE command, before the line buffer is used again, to insure that the transfer to the
device has been completed.
On non-bulk storage devices, headers are output along with the data in IOPS binary mode only (bit 9 and 11 of
header word 0 should be set to 1). On bulk storage devices, headers are output along with the data in all modes except
dump mode. In image modes, the header space cannot be used for data, even though the headers are not written out.
The word pair count in the header takes precedence over maximum size (or word count) in all modes and must be
inserted by the user.
For both .READ and .WRITE macros, dump mode causes the transfer of the specified core area to or from one record
on magnetic or paper tape. One or more blocks on DECtape or disk may be occupied by a single dump command. A
subsequent .WRITE in dump mode will utilize the unfilled portion of the last block.

2.5 .WAIT
FORM:

WAIT

VARIABLES:

a= .DAT slot nuwher (octal radix)

EXPANSION:

LOC

CAL+ a9 _ 17

LOC+ 1

/The CAL Handler will place the unit number (if
/applicable) associated with .DAT slot a into bits
/0 through 2.

DESCRIPTION: .WAIT is used to detect the availability of the user's line buffer (being filled by .READ or emptied
by .WRITE). If the line buffer is available, control is returned to the user immediately after the .WAIT macro
expansion (LOC + 2). If the transfer of data has not been completed, control is returned to the .WAIT macro ..WAIT
must also be used after the .TRAN command.

2.6 .WAITR

FORM:

WAITR

a, ADDR

VARIABLES:

a = .DAT slot number (octal radix)
ADDR =Address to which control is passed if line buffer is not available for use.

EXPANSION:

2-4(Part III)

LOC

CAL+ 10008 + ag-11

LOC+ 1

LOC+2

ADDR

/The CAL Handler will place the unit number (if
/applicable) associated with .DAT slot a into bits
/0 through 2.

DESCRIPTION: .WAITR is also used to detect the availability of the user's line buffer. If the buffer is available,
control is returned to the user immediately after the .WAITR macro expansion (LOC + 3). If the transfer of the data
has not been completed, however, control is given to the instruction at ADDR. It is the user's responsibility to return
to the .WAITR to again check the availability of the buffer.

2.7 .CLOSE

FORM:

.CLOSE

VARIABLES:

a= .DAT slot number (octal radix)

EXPANSION:

LOC

CAL+ a9_ 17

LOC+ I

/The CAL Handler will place the unit (if applicable)
/associated with .DAT slot a into bits 0 through 2.

DESCRIPTION: When action has been initiated (.INIT or .SEEK or .ENTER) on a file (whether the device is
file-oriented or not) this action must be terminated by a .CLOSE command.
On input, it is assumed that the user is finished with the file when the .CLOSE macro is used, so the file is closed. On
output, all associated output is allowed to finish and then an EOF (end-of-file) line is output before the file is finally
closed. If a refers to a file-oriented device, any earlier file of the same name and extension, as currently referenced, is
deleted from its directory after the new file is written.

2.8 .TIMER
n,C

FORM:

TIMER

VARIABLES:

n =Number of clock increments (decimal radix)
C = Address of subroutine to handle interrupt at end of interval

EXPANSION:

LOC

CAL

LOC+ I

LOC+2

c
.DEC

LOC+3

/Decimal radix

-n

DESCRIPTION: .TIMER is used to set the real-time clock to n increments and to start it. Each clock increment
represents l/60s for 60 Hz systems and l/50s for 50 Hz systems.
C + 1 is the location to which control is given when the Monitor services the clock interrupt. The coding at C should
be in subroutine form; for example,

2-5(Part III)

/C + 1 is reached via JMS

DACSAVEAC
. } Must not contain any Monitor CALs
.

in Basic or Advanced Software Systems

LACC

/Restore Link

RAL
LAC SAVEAC
XIT

/Restore AC

JMP* C

so that control will return to the originally-interrupted sequence when the interval-handling routine has been
completed. The Monitor automatically reenables the interrupt system before transferring control to C + 1. If the user
wishes to initiate another interval at the completion of the previous interval in the subroutine specified to .TIMER, he
may do so as follows:
LAC (desired interval in 2's complement)
DAC* (7
LACC

/Restore Link

RAL
LAC SA VEAC

/Restore AC

CLON

/Turn on clock

JMP* C
2.9 .EXIT

FORM:

.EXIT

EXPANSION:

LOC

CAL

LOC +I

DESCRIPTION: .EXIT provides the standard method for returnit. 6 to the Monitor after completion of a system or
user program. In the Basic 1/0 Monitor environment, it causes a program halt; in the Advanced Monitor environment,
it causes the non-resident monitor to be reloaded. When the reloading process has been completed, the Monitor types
MONITOR
$

on the teleprinter, indicating that it is ready to accept the next command. In the Background/Foreground Monitor
environment, the effect of the .EXIT depends upon whether it occurs in a BACKGROUND or a FOREGROUND job.

2-6(Part III)

CHAPTER 3
BASIC INPUT/OUTPUT MONITOR FUNCTIONS

3.1 GENERAL

The Basic 1/0 Monitor simplifies the programming of input and output functions in the basic paper-tape environment.
It serves as an interface between the system and user programs and the external world of device hardware, relying
upon the routines and capabilities of the Input/Output Programming System (IOPS) to relieve the programmer of
writing his own device and data handling subroutines. The 1/0 Monitor allows simultaneous operation of many 1/0
peripherals and overlapped computation. Since upward compatibility exists between the Monitor systems, user
programs that are written to operate under control of the 1/0 Monitor will also operate, without modification, under
control of the Advanced and Background/Foreground Monitors.

The Input/Output Monitor is designed to take advantage of the Automatic Priority Interrupt (API) if it is present on
the system. Both the 1/0 skip chain for the Program Interrupt Control (PIC) and the API channels are set up to
handle all devices which have been requested by the user. All unused channels are tied to an error routine to detect
spurious interrupts.

The reader is referred to Chapter 1 for a general discussion of the Monitor environment, and to Chapter 2 for detailed
descriptions of user program commands (system macros) available in the 1/0 Monitor environment.

3.2 PROGRAMMING EXAMPLE
The following example illustrates the use of system macros with MACR0-15 programs in the Basic 1/0 Monitor
environment. The example inputs a line of data from the Teletype keyboard, and outputs the same line of data to the
Teletype. The arguments used by the system macros are given symbolic names (via MACR0-15 direct assignment
statements) to facilitate recall for the programmer, and to change the arguments easily, if desired. Note the use of the
pseudo-ops..TITLE, .IODEV, .BLOCK, and .END, in addition to the system macros. The assembly listing that
follows the example shows how the system macros are expanded at assembly time. (The reader may wish to compare
these expansions with the system macro descriptions in Chapter 2.)

3-1 (Part Ill)

• TITLE ECHO
TT1:2

TT0=4U
OUT:l
1N:l6

l0P;):2

2,4
TTO,OUT,RESTRT
TTI, I N,riESTRT
.ki::IU)
TTI,IOPS,oUFFEk,34
• vlA1 T TTI
.•RUE TTO,IOPS,dUFFEH,34

.lOlJEV
• INI T
• INI T

START

eiEGIN

.wAIT

TIO

BEGIN
.CLOSE TTI
.CLOSE TTO
JillP
START
.BLOCK 42
START
.END

Ji'lP
flESTHT

dUJ'l''EH

/lNlTlALIZE TELETYPE OUTPUT
/ANU INPUT
/INPUT IOPS ASCII FRO~ TELETYPE
/vlAIT UNTIL INPUT CO~PLETE
/uUTPUT SA~~ DATA ON TELETYP~
/~Alt UNTIL OUTPUT CO~PLETE
/LOOP TO INPUT AGAIN
/TER~INATE INPUT
/TERMINATE OUTPUT
/RETURN TO REINITIALIZE
/SET UP TELETYPE BUFFER C34 DEC>
/END OF ECHO PROGRA~

ASSEMBLY LISTING:
PAGE

ECHO
.TITLE ECHO
VH1JOJ002
QIP,(11411'/J
00J0vHil1
00J(ll0011J
11l'1JCl!002

A.
A.
A
A.
A

TTI =2
TT0=4111l
OUT=1
I N=l'/J
IOPS=2

• I OOEV
llJ001/JVJ
0011!00
0011!01
00002
00011l3

R
START
R '11Clll 410 A. •G

1211Hlf114
flJr210 V'l5
OJOJ006
"10007
011l010J
00'1!10
00011
lil0012

R 91OJ111riHll2
R i10J0Vlfil1
R '110V'l025
R 01H'litl011J
R
R OJ0?~rn2
R Cl!0Cl!i1111l
R V!Cl!(l!Ql32

R '110!i!001 A. •G
R PIOJVl025 R *G
R 011JQl0VJl1J A •G
A. ttG
A. ttG
R ttG
A. ttG
BEGIN
A •G
A. •G
R •G

2.4

TTO,OUT,RESTRT
• IN IT
CAL+OUT•1Hi0 TT0&777
1
RESTRT+0
0
.INIT
TTI, IN,RESTRT
CAL+ I Ntt1f1100J TTI&777
1
RESTRT+11J
Ill
,RfA.D
TT I, I OPS, BUF'F'ER, 3.4
CAL+ I OPS•10Jli!IPI TTI &777

Hl
BUFFER

.DEC
00013 R 777736 A ttG
00014 R C/!01-'1002 A ttG
00015 R 0100J'1112 A ttG
00016 R 0Pl24111l A •G
00017 R 17l0riHH1 A •G
0ril02fll R 17l0'1l032 R •G
ril0021 R 777736 A ttG

3-2(Part III)

-34
TTI
.WA.IT
CAL TT I &777
12
.WRITE
TTO,IOPS,eUFFER,34
CAL+IOPS•10011J TT0&777

11
BUFFER
.DEC
-34
.WAIT

TTO

00022 R ~11!0410 A oG
00023 R Q!(IJCIJ012 A ttG
00024 R 6'H'lv.l!e R
REST RT
000215 R
00025 R 01H~002 A ttG
li'l0Pl26 R 0(210flll7l6 A oG
00027
00030!
00031
00032
PAGE

R lilriJV'!410 A ttG
R lll!illl!C106 A ttG
R 601111,100 R
R
BUFFER
A
!i'l0V!Vl00 R

CAL TT0&777
12
JMP
REG IN
.CLOSE
TT I
CAL TTI&777
6

.CLOSE
TTO
CAL TT0&777
6

JMP

START
.BLOCK
42
.ENO
START
0 ERROR LINES

ECHO

AEGIN
AUF'FER
IN
IOPS
OUT
RESTIH
START
TTI
TTO
PAGE 3

00'1!10 R
lll0032 R
0'110000 A
000002 A
fi'HIJ0001 A
00025 R
00000 R
PIV.'10002 A
000410 A
ECHO

IN
START
OUT
IOPS
TT I
REGIN
REST RT
RUFFER
TTO

00000Cll
00000
0V.HHHJ1
000002
000002
000HI
00025
00032
0P10410

R
A
A

A
R

R
R
A

3.3 OPERATING THE BASIC 1/0 MONITOR SYSTEM
The reader is referred to the PDP-15/10 Users Guide (DEC-15-GG lA-D) for detailed operating procedures for system
programs in the 1/0 Monitor environment. The following PDP-15 Software System manuals contain additional
detailed information on system programs.
Manual

Document Number

Utility Programs

DEC-15-YWZA-D

MACRO-IS Assembler

DEC-15-AMZA-D

FORTRAN IV

DEC-15-KFZA-D

FOCAL

DEC-15-KJZA-D

8TRAN

DEC-15-ENZA-D

This section contains descriptions of loading programs, device assignments, and error detection and handling.

3-3(Part III)

3.3.1 Loading Programs in the Basic 1/0 Monitor Environment
In the paper tape system, each system program accompanied by the necessary I/O device handlers and an appropriate
version of the Monitor, is punched on a separate paper tape in absolute format. See Figure 3-1 for memory maps of
the Basic I/O Monitor System.
The eleven system tapes supplied are:

FORTRAN IV

DDT (without Patch File capabilities)

FOCAL-15

DDT (with Patch File capabilities)

MACR0-15

8TRAN (PDP-8 -to PDP-15 Translator)

PIP-15

CHAIN

Text Editor

EXECUTE

Linking Loader

In addition, the utility program PUNCH-15 which provides the ability to dump an executable core load and .ABS
loader onto paper tape, is provided with all paper tape systems.
At the beginning of each tape is a Bootstrap Loader in hardware READIN mode. By setting the starting address of the
Loader on the console ADDRESS switches to 17720 (8K), depressing 1/0 RESET, and then depressing the READIN
switch, these system tapes may be loaded.

Since the tapes also contain appropriate versions of the Basic I/O Monitor and the necessary 1/0 device handlers, the
system programs listed above can be loaded, ready for operation, in a single step.

Once the system program has been loaded and takes control, 'the individual system program operating procedures
come into use. (See PDP-15/10 Users Guide, DEC-15-GG IA-D.)

User programs, however, normally exist in relocatable form, as output from FORTRAN IV or MACR0-15, these
tapes do not contain copies of the Monitor. To load these programs, the Linking Loader or DDT should be loaded
first. The user should then initiate loading of his main program followed by all required subprograms. By loading
subprograms in order of size (largest first, smallest last), the µser has a better chance of satisfying core requirements
for his program in systems with extended core memory. The Monitor (including the device handlers) contained on the
Linking Loader or DDT tape may be used with user programs, and the Linking Loader or DDT can be used to load
the necessary device handlers as well as the user's object programs.

3-4(Part III)

MEMORY MAP A
4K TO 32K
(4K INCREMENTS)

SYSTEM PROGRAMS

FORTRAN JI:
MACR0-15
EDITOR
PIP-15

MEMORY MAPB·LINKING LOADER, CHAIN
4K TO 32K
(4K INCREMENTS)

BOOTSTRAP
LOADER IN
HRM FORMAT

• SCOM

4810
• SCOM ANO • SCOM + 3

USER
PROGRAMS
FOCAL-15
AND 8 TRAN

SYSTEM
PROGRAM

•

BOOTSTRAP
LOADER 1111
HRM FORMAT

• SCOM+3

SYSTEM
PROGRAM
TABLE
SPACE

GLOBAL
SYMBOL
TjLE
• SCOM+2

•

SYSTEM PROGRAM
DEVICE HANDLER

LINKING LOADER
OR CHAIN
• SCOM +2

PAPER TAPE
READER HANDLER

SYSTEM PROGRAM
DEVICE HANDLER

• SCOM+1

• SCOM+I

110 M()lollTOR
WITH TELETYPE ·1111
AND
TELETYPE ·OUT
DEVICE HANDLERS

110 MONITOR
WITH TELETYPE•IN
AND
TELETYPE-OUT
DEVICE HANDLERS

Refer to Section 4.4 for sizes for device
handlers.

NOTE:
IN THE CASE OF CHAIN THE
PAPER TAPE PUNCH l!ANOLER
IS ALSO RESIDENT IN CORE.

Refer to Memory Map D for results of Linking
Loader

Figure 3-1. BASIC 1/0 Monitor System Memory Maps

3-S(Part III)

MEMORY MAP C-DOT TAPE

4K TO UIC

MEMORY MAP 0-USER PROGRAM READY
TO BE EXECUTED

4K TO 52 k

(4 K INCREMENTS)

(4K INCREMENTS)
BOOTSTRAP
LOADER IN
HRM FORMAT

4810
BOOTSTRAP
• SCOM
•

SCOM

DOT
USER
PROGRAM!S)

• SCOM+3
USER DEVICE
HANDLER

UJER
PROGRAMS

USER DEVICE
HANDLER

• SCOM +3

GLOBAL AND
DDT
SYMBOLJTABLES
• SCOM+2

LINKING
LOADER

-------

lb.l

LINKING LOADER
DEVICE HANDLER

PAPER TAPE
PUNCH HANDLER

------LINKING LOADER
DEVICE HANDLER

PAPER TAPE
READER HANDLER
• SCOM+t

la.I

1/0 MONITOR
WITH TELETYPE-IN
ANO
TELETYPE-OUT
DEVICE HANDLERS

1/0
MONITOR
(INCLUDING
TELETYPE
HANDLER)

Refer to Memory Map E for results of Link
Loading in DDT mode.

(FORTRAN IV or MACR0-15) output may
make use of core as low as they point.

Paper Tape Punch Handler is only present in
DDT versions with patch file capabilities.

a. If the user program did not have any device
handlers in common with the Linking
Loader.

Refer to Section 4.4 for sizes of device handlers.
.SCOM+1 and .SCOM+2 both point to one of
two places and non-BLOCK DATA COMMON

b. If the user program did have at least one
device handler in common with the Linking
Loader.

Figure 3-1. BASIC 1/0 Monitor System Memory Maps (Cont)

3-6(Part III)

MEMORY MAP F (EXECUTE)

MEMORY MAP E

4K TO ll2K
(4K INCREMENTS)

4K TO 32K
(4K INCREMENT)
BOOTSTRAP

BOOTSTRAP

• SCOM

• SCOM
DDT

EXECUTE

USER
PROGRAMISl

USER
PROGRAM($)

USER/DOT
DEVICE HANDLER

LI BR ARY
PROGRAM(Sl

USER/DOT
DEVICE HANDLER

USER DEVICE
HANDLE RISI
• SCOM + 3
NAMED COMMON

• SCOM+ 3

DOT CREATED
SYMBOLS AND
PATCH SPACE
• SCOM +2
DDT
SYMBOL
TABLE

BLANK COMMON

7JIIIUDI£DJTJ • SCOM +I
LINKING LOADER
DEVICE HANDLER
LINKING LOADER
DEVICE HANDLER

•

LINKING LOADER
BLOCK TRANSFER
ROUTINE

SCOM+2

EXECUTE'S
DEVICE
HANDLER

I /0 MONITOR

1/0 MONITOR

(INCLUDING
TELETYPE
HANDLER l

(INCLUDING
TELETYPE
HANOLERl

BB Oto

Refer to Section 4.4 for sizes for device
handlers.
Non BLOCK DATA COMMON (FORTRAN IV
of MACR0-15 output) may make use of core as
low as the DDT symbol table. However, trouble
will occur if the user requests DDT to create
symbols or make patches that cause overlaying
of the COMMON area.
The Linking Loader device handlers would have
been used to satisfy user device requests.

When the user types in the name of the XCT
File to be run, .EXECUTE brings in the first
chain from paper tape.
FORTRAN programs pass on data in blank
common starting at .SCOM + 2. Macro programs
pass on data between .SCOM + 2 and .SCOM +

3.
A call from the running chain to bring in
another chain is effected by transferring control
back to Execute.

Figure 3-1. BASIC 1/0 Monitor System Memory Maps (Cont)

3-7(Part III)

3.3.2· Device Assignments
The device assignment table used by the Basic I/O Monitor is fixed in length and in the assignments it contains. It is
composed of two sections; the upper section is for use by all system programs except PIP, the lower section is
referenced by all user programs and PIP.
The upper portion of the .DAT contains 13 slots, referenced as -1 through -15 8 • The lower section has 8 slots
numbered I through l 0 8 • The standard assignments for the device assignment table for user programs and system
programs other than PIP are shown in Figure 3-2. Figure 3-3 illustrates PIP assignments, which include the Card
Reader (CR03B) and Line Printer as standard devices.
Device

Handler*

Use

Paper Tape Punch
Paper Tape Reader
Paper Tape Reader
Paper Tape Punch
Paper Tape Punch
TTY Printer
Paper Tape Reader
Paper Tape Reader

(PPA.)
(PRA)
(PRA.)
(PPA)
(PPB.)
(TTA.)
(PRB.) •
(PRA.)
(TIA.)

-5
-4

TTY Keyboard
0
Paper Tape Punch
0
Paper Tape Reader

-3
-2
-1

TTY Printer
TTY Keyboard
Paper Tape Reader

(TTA.)
(TIA.)
(PRA.)

Editor Output
8TRAN Input
Editor Input
8TRAN Output
MACRO, FORTRAN IV Output
MACRO, FORTRAN IV Listing
MACRO, FORTRAN IV Input
DDT Patch-file Input and
Editor Secondary Input
MACRO Secondary Input
Not Used
Output (.DDT, Chain)
Not Used
System Input (Linking Loader,
DDT Chain and Execute)
Teleprinter Output
Keyboard Input
System Device (Linking Loader,
DDT and Chain)

TTY Printer
TTY Keyboard
Paper Tape Reader
TTY Printer
Paper Tape Punch
Paper Tape Reader
Paper Tape Punch
Paper Tape Reader

(TIA.)
(TTA.)
(PRA.)
(TIA.)
(PPA.)
(PRA.)
(PPA.)
(PRA.)

.DAT Slot
.DATBG

-15
-14
-13
-12
-11
-10

-7
-6

DAT

.DAT
I
2
3
4

s
6
7

(PPA.)
(PRA.)

Teleprinter Output
Keyboard Input
Input
Listing
Output
Scratch
Scratch
Scratch

DATND=.
Figure 3-2. Device Assignment Table (.DAT) for the Basic 1/0 Monitor
The negative .DAT slot assignments for use by system programs may be changed by DEC. For example, .DAT slot -IO
might be associated with a card reader, while .DAT slot -12 could be assigned to a line printer. Provision for changing
positive .DAT slot assignments for use by relocatable user programs is included in the PUNCH 15 utility program (see
section 3.3.S). For example, a line printer handler (LPA) or a card reader handler (CDB) could be added.
*See Section 4.4 for a description of the handlers.

3-8(Part Ill)

Device

Handler

Use

Teletype

(TIA.)

Input/Output

Teletype

(TIA.)

Input/Output

Paper Tape Reader

(PRA.)

Input

Line Printer

(LPA.)

Output

Paper Tape Punch

(PPA.)

Output

Card Reader

(CDB.)

Input

P!iper Tape Punch

(PPA.)

Output

Paper Tape Reader

(PRA.)

Input

.DAT Slot

Figure 3-3. Device Assignment Table (.DAT) for PIP

3.3.3 Error Detection and Handling
Comprehensive error checking is provided by the Linking Loader and the Input/Output Programming System.
Detailed lists of errors that may occur are given in Appendix C and D, respectively. The other system programs also
provide comprehensive error checking. Refer to the appropriate PDP-15 manual (see paragraph 1.3 in Part I of this
manual).
3.3.4 Control Character Commands in the Basic 1/0 Monitor Environment
All control character commands recognized by the Basic 1/0 Monitor are summarized in Table 3-1. These commands
(except RUBOUT) are formed by holding down the CTRL key while striking a letter key. The character or characters
echoed on the Teletype and the resulting action is given in the table for each command.
3.3.5 Modifying System Programs and Building Executable User Core Loads in the Basic 1/0 Monitor Environment
The capability of modifying or patching system programs in the Basic 1/0 Monitor environment is provided by the
utility program PUNCH 15. PUNCH 15 allows for producing an executable core load on paper tape in .ABS format
with the standard .ABS loader (HRM 17720 of the highest available core bank) on the front of the tape. This is
particularly useful when the core load consists of a relocatable main program, subroutines and library routines, the
repetitive loading of which tends to be time consuming. It is further possible to specify the number of .ABS tapes to
be output ( 1-9) for convenience of tape handling.
The areas of memory output by PUNCH 15 are 0 up to .SCOM + 2 (the first free cell) and .SCOM + 3 (last free cell)
up to the .ABS loader ( 17720 modulo 8K). PUNCH 15 is loaded (1/0 RESET and READ IN) at 17720 of the highest
core bank available. It loads and relocates part II of itself in free core, i.e., from the cell in .SCOM + 2 and up. When
loaded it types the following message on Teletype:
P, TorS?)

>
3-9(Part III)

Table 3-1. Control Character Commands
Command

Echo

Action

CTRLS

t S

Starts user program after Linking Loader has brought it into core via a
LOAD command.

CTRLT

CTRL T is applicable only when using DDT and transfers control to
DDT which types
DDT

>
to indicate its readiness for another DDT command. All previous DDT
conditions remain intact (breakpoints, register modifications, etc.).
CTRLR

t R

Allows the user to continue when an IOPS4 (device not ready) error
occurs. The user must first ready the device, and then type CTRL R.

CTRLP

t P

Forces control to last address specified in the .INIT command
referencing Teletype. Used by system programs to reinitialize or restart.
Generates EQT code - used to terminate Teletype input.

CTRLD
CTRLU

Cancels current line on Teletype (input or output).

RUBOUT

Cancels last character input from Teletype (not applicable with DDT).

The user is expected to type CTRL P or CTRL T or CTRL S to define the starting location of the program to be
punched followed by carriage return (1 tape) or a number (1-9) specifying the total number of tapes into which
binary output is to be divided.
All tapes output by PUNCH 15 are loaded by pressing 1/0 RESET and the READIN key with ADDRESS switches set
to 17720 of the highest core bank.

CTRL T should be used if DDT is part of the core load.
CTRL P should be used for all other system programs and user programs which have already initialized (.INIT) the
teletype with a restart address at the time PUNCH 15 was executed.
CTRL S should be used only if the core load was output by PUNCH 15 after Linking Loader operation at the
moment when the loader itself was expecting the CTRL S command.
All tapes output by PUNCH 15 are loaded by pressing 1/0 RESET and the READIN key with ADDRESS switches to
17720 of the highest core bank.

3.3.6 Modifying User .DAT Slots in the 1/0 Monitor Environment
Although it is not possible to reassign negative .DAT slots at load time in the Basic 1/0 Monitor environment for
system programs (reassembly is required), PUNCH 15 provides this capability for the user or positive .DAT slots.

3-lO(Part III)

For example, Figure 3-2 lists the standard .DAT slot assignments. A relocatable user program wanting to use a card
reader (CDB.) or line printer (LPA.) whose handlers are included in the paper tape library has no way of doing so
unless the positive .DAT table can be modified. The modification procedure is as follows:
1. Load the Linking Loader (or DDT) tape into core (HRM 17720 modulo 8K).
2. Stop the computer and modify the appropriate .DAT slot cell* according to the Loader-1/0 correspondence
table below. For example, if .DAT slot 7 is to be assigned to the line printer using handler LPA, cell 144 should be
changed to 000007.
3. Load PUNCH into core (HRM 17720 modulo 8K).
4. Type CTRL Pin response to the query from PUNCH: P, Tor S?
5. Load the resultant punched tape into core (HRM 17720 modulo 8K).
6. The Loader will restart ready for acceptance of typed program names to be loaded. If .DAT slot 7 is referenced
by the user program (e.g., IODEV 7), LPA for the line printer will be loaded from the 1/0 Library.

Loader - 1/0 Correspondence

Handler
TIA.
PRA.
PRB.
PPA.
PPB.
PPC.
LPA.
CDB.
MTF

.DAT Slot Value

2
3
4

5
6
7
11

13*

*.DAT table (cell 0) begins at location 135
**With unit number in bits 0-2.

3-l l/3-12(Part III)

CHAPTER 4
1/0 DEVICE HANDLERS

This chapter contains information that is essential for a good understanding and proper use of 1/0 device handlers for
the Basic 1/0 Monitor system. Included is a general description of 1/0 hardware and API software level handlers, a
complete section on writing special 1/0 device handlers, a summary of 1/0 handlers acceptable to system programs,
and_ a summary of standard 1/0 handler features.
It is assumed that the reader is familiar with all related material in the PDP-15 Reference Manual (DEC-15-BRZA-D),
especially Chapter 5, Organization of the Input/Output Processor. It is also assumed that the reader is familiar with
the PDP-15 Monitor environment and other pertinent information contained in this manual.

4.1 DESCRIPTION OF 1/0 HARDWARE AND API SOFTWARE LEVEL HANDLERS

4. I .1 1/0 Device Handlers

All communications between user programs and 1/0 device handlers are made via CAL instructions (see Chapter 2)
followed by argument lists. The CAL Handler in the Monitor performs preliminary setups, checks on the CAL calling
sequence, and transfers control via a JMP instruction to the entry point of the device handler. When the control
transfer occurs, the AC contains the address of the CAL in bits 3 through 17 and bits 0, 1, and 2 indicate the status of
the Link, extend mode and memory protect, respectively, at the time of the CAL. Note that the content of the AC at
the time of the CAL is not preserved.
On machines that have an API, the execution of a CAL instruction automatically raises the priority to the highest

software level (level 4). Control passes to the handler while it is still at level 4, allowing the handler to complete its
re-entrant procedures before debreaking (DBK) from level 4. This permits the handler to receive re-entrant calls from
software levels higher than the priority of the program that contained this call. If a device handler does not contain
~ntrant procedures, system failure caused by inadvertent re-entries can be prevented by remaining at level 4 until
control is returned to the user.
If the non-reentrant method is used, the debreak and restore (DBR) instruction should be executed just prior to the
JMP* which returns control to the user, allowing debreak from level 4 and restoring the conditions of the Link,

4-1 (Part III)

extend mode, and memory protect. Any IOTs issued at the CAL level (level 4 if API present, mainstream if no API)
should be executed immediately before the
DBR
XCT .+1
JMP*
exit sequence to ensure that the exit takes place before the interrupt from the issued IOT occurs. (The XCT is
necessary to ensure that the 3 cycles requested by the API on a debreak operation occur in the instruction after the
DBR.)
The CAL instruction must not be used at any hardware priority level (API or PIC), since interrupts to these levels are
not closed out by the execution of a CAL and recovery is not possible from such sequences of events as
a. An 1/0 flag coming up during a CAL at level 7,
b. Control going to the 1/0 device handler at level 3,
c. The handler at level 3 CALing and thus destroying the content of location 00020 for the previous CAL.
The highest API software level (level 4) is also used for processing CALs and care must be taken when executing CALs
at this level. For example, a routine that is CAL'd from level 4 must know that if a debreak (DBR or DBK) is issued,
control will return to the calling program at a level lower than 4. The calling routine will also debreak; however, this
second debreak will not be from level 4 but from the next highest active level.

4.1. l .1 Setting Up the Skip Chain and API (Hardware) Channel Registers - When the Monitor is loaded, the Program
Interrupt Control (PIC) skip chain and the Automatic Priority Interrupt (API) channels are set up to handle the
Teletype keyboard, teleprinter and clock interrupts, only. The skip chain contains the other skip IOT instructions,
but indirect jumps to an error routine result if a skip occurs, as follows:

SKP
SKP
JMP*
SKP
JMP*
SKP
SKP
JMP

PRA

/Skip if Reader flag.

INITl
LPT
SKP
INT2
TTl

/INT 1 contains error address.
/Skip if line printer flag.
/INT2 contains error address.
/Skip if Teletype flag.

TELINT

/To Teletype interrupt handler.

All unused API channels also contain JMPs to the error address.
When a device handler is called for the f"rrst time via an .INIT user program command, it must call a Monitor routine
(.SETUP) to set up its skip chain entry or entries and API channel, prior to performing any 1/0 functions. The calling
sequence is as follows.

4-2(Part III)

CAL

SKP

JOT

DE VINT
(normal return)

/N = API channel register 40 through 77 (see section
/4.1.3 for standard channel assignments), 0 if device
/not connected to APL
/.SETUP function code.
/Skip JOT for this device.
/Address of interrupt handler.

DEVINT exists in the device handler in the following format.
DEVPIC

DE VINT

DVSTON
DEVION

DAC
LAC*
DAC
LAC
JMP
JMP
DAC
LAC

DEVAC
(0
DEVOUT
DEVION
DVSTON
DEVPIC
DEVAC
DEVINT

/SAVE AC.
/SAVE PC, LINK, EX.MODE, MEM.PROT.
/FORCE ION AT DISMISSAL.
/PICENTRY.
/API ENTRY, SAVE AC.

DAC
DEVOUT
IORS
SMA!CLA
LAW
17740
TAD
DEVION
DAC
DVSWCH
DEVCF
ION

/SAVE PC, LINK, EX.MODE, MEM.PROT.
/CHECK STATUS OF PIC
/FOR RESTORATION AT DISMISSAL.
/PIC OFF, BUILD IOF IOT.
/PICON, BUILD ION IOT.

IOF
DEVIOT

/DISABLE PIC TO INSURE
/DISMISSAL BEFORE INTERRUPT
/FROM THIS JOT OCCURS

/DISMISS ROUTINE
LAC
DAC
LAC
DVSWCH
ION
DBR
JMP*

(JMP DEVPIC
DEVINT
DEVAC

DEVOUT

/CLEAR DEVICE DONE FLAG
/ENABLE PIC SO THAT OTHER DEVICES
/AREN'T SHUT OUT.

/RESTORE DEVINT IN
/CASE API DISABLED.
/RESTORE AC
/IONORIOF
/DEBREAK AND RESTORE CONDITIONS
/OF LINK, EX.MODE AND MEM.PROT.

Since the auto-index registers and EAE registers are not used by the standard 1/0 device handlers, it is not necessary
to save and restore them.
The Monitor routine (.SETUP) checks the skip chain for the instruction which matches SKP IOT; if there is a match it
places the address, DEVINT, in the appropriate transfer vector (INTn) and places JMS* INTn in the corresponding
API channel register. If a match cannot be found, IOPS outputs the following error message,
.IOPS 05 XXXXXX

4-3(Part III)

indicating that the skip IOT in the CAL calling sequence at location XXXXXX was not in the skip chain.
Refer to paragraph 4.2.4 for the method of incorporating new handlers and associated skip chain entries into the
Monitor.

4.1.2 API Software Level Handlers

4.1.2.1 Setting Up API Software Level Channel Registers - When the Monitor is loaded, the API software-level
channel registers ( 40 through 43) are initialized to
JMS*
JMS*
JMS*
JMS*

/LEVEL4
/LEVEL 5
/LEVEL 6
/LEVEL 7

.SCOM+l2
.SCOM+l3
.SCOM+l4
.SCOM+l5

where the .SCOM registers are at absolute locations 00112 through 00115 and contain the address of an error routine.
Therefore, prior to requesting any interrupts at these software priority levels, the user must modify the contents of
the .SCOM registers so that they point to the entry point of the user's software level handlers.

Example:
.SCOM = 100
LAC
DAC*

(LVSINT
(.SCOM+l3

LVSINT exists in the user's area in the following format:

LVSINT 0
DAC
SAVSAC
/SAVE AUTO INDEX REGISTERS
/IF LEVEL 5 ROUTINES
/USE THEM AND LOWER LEVEL
/ROUTINES ALSO USE THEM
/SAVE MQ AND STEP COUNTER
/IF SYSTEM HAS EAE AND IT
/IS USED AT DIFFERENT LEVELS.

/RESTORE SAVED REGISTERS.
DBR
XCT
.+l
LVSINT
JMP*

4-4(Part III)

/PC, LINK AND MEM. PROT.
/SAVE AC

/DEBREAK FROM LEVEL 5
/AND RESTORE LINK AND MEM.PROT.

4.1.2.2 Queueing - High priority /high data rate/short access routines cannot perform complex calculations based on
unusual conditions without holding off further data inputs. To perform the calculations, the high priority program
segment must initiate a lower priority (interruptable) segment to perform the calculations. Since, in general, many
data handling routines will be requesting calculations, there will exist a queue of calculation jobs waiting to be
performed at the software level. Each data handling routine must add its job request to the appropriate queue (taking
care to raise the API priority level as high as the highest level that manipulates the queue before adding the request)
and issue an interrupt request (ISA) at the corresponding software priority level. The general flow chart, Figure 4-1
depicts the structure of a software level handler involved with queued requests.
LV51NT
SAVE PC.LINK.AC.
AUTO-INDEX REGS,
MO. STEP COUNTER
AND CONDITIONS
OF EXTEND MODE
AND MEMORY PROTECT

RESTORE SAVED
REG IS TE RS

RAISE TO HIGHEST
LEVEL THAT
MANIPULATES
LEVEL 5 QUEUE

REMOVE ENTRY
FROM QUEUE

DllK llACK
TO LEVEL 5

GO HONOR THIS

'-----< JOll REQUEST VIA
AJMS

Figure 4-1. Structure of API Software Level Handler

Care must be taken about which routines are called when a software level request is honored; that is, if a called
routine is "open" (started but not completed) at a lower level, it must be reentrant or errors will result.

NOTE

The standard hardware 1/0 device handlers do not contain reentrant
procedures and must not be reentered from higher software levels.
New resident handlers for Power Fail, Memory Parity, nonexistent memory
violation, and Memory Protect violation have been incorporated into the
system and effect an IOPS error message if the condition is detected. (see
Appendix E for IOPS errors). The user can, via a .SETUP, tie his own
handler to these skip IOT or API channel registers.
4-S(Part III)

4.1.3 Standard API Channel/Priority Assignments

Channel

Device

4-6(Part Ill)

Option
Number

Priority

Channel
Register

Software Priority

DECtape

TC02

MAG tape

TC59

RESERVED

Paper Tape Reader

Clock overflow

Power Fail

KF15

Parity

MPlSB

Display (LP flag)

VP15

Card Reader

CR03B

Line Printer

LP15A/LP15C

A/D

AFOl/ADCl/9

DB99/DB98

DB09

RESERVED

Data Phone

DP09A

DECdisk

RF15

DISK Pack

RP15

Plotter

565

61
2

Channel

Device

Multi-Station
TTY Control

Priority

Channel
Register

LT19A (Teleprinter)

LT19A (Keyboard)

Option
Number

NOTE: Channels 26-33, 36 and 37 and corresponding Channel Registers are reserved.

4.2 WRITING SPECIAL 1/0 DEVICE HANDLERS
This section contains information prepared specifically to aid those users who plan to write their own special I/O
device handlers for the Basic I/O Monitor system.
Although special handlers cannot be incorporated directly into the Basic Monitor System, they can be designed to run
with user programs in the Basic Monitor environment (see paragraph 4.2.4). If a user wishes to incorporate a special
handler into a systems program (for example, a card punch handler for MACRO 15), he must purchase the source
tape and assemly listings, modify them symbolically, and reassemble using at least a I 6K machine.
It is assumed that the user is familiar with paragraph 4.1. To summarize, the handler is entered via a JMP from the
Monitor as a result of a CAL instruction. The contents of the AC contain the address of the CAL in bits 3 through 17.
Bit 0 contains the Link, bit 1 contains the extend mode status, and bit 2 contains the memory protect status. The
previous contents of the AC and Link are lost.
To show the steps required in writing an I/O device handler, a complete handler (Example B) was developed with the
aid of a skeleton handler (Example A). This handler is a non-reentrant type (discussed briefly at the beginning of this
chapter) and uses the Debreak and Restore instruction (DBR) to leave the handler at software priority level 4 (if API),
and restore the status of the Link, extend mode, and memory protect. Example A is referenced by part numbers to
illustrate the development of Example B, a finished Analog to Digital Converter (ADC) I/O Handler. The ADC
handler shown in Example B, was written for a hypothetical basic analog-to-digital converter with IOT instructions as
shown in Table 4-1. This handler is used to read data from the ADC and store it in the user's line buffer. The handler
shown in Example Bis for instructional purposes only; it has not been thoroughly tested.
TABLE 4-1. Hypothetical A/D Converter IOT Instructions
Mnemonic
Symbol

Octal

Operation

Code

Executed

ADSF

701301

Skip if converter flag is set. This flag is connected to the program
interrupt.

ADSC

701304

Select and convert. The converter flag is cleared and a conversion of an
incoming voltage is initiated. When the conversion is complete, the
converter flag is set.

ADRB

701312

Read converter buffer. Places the content of the buffer in the AC, left
adjusted. The remaining AC bits are cleared. The converter flag is
cleared.
4-?(Part III)

The reader, while looking at the skeleton of a specialized handler as shown in Example A, should make the following
decisions about his own handler (The decisions made in this case are in reference to developing the ADC handler):
a. Services that are required of the handler (flags, receiving or sending of data, etc.). By looking at the ADC IOT's
shown in Table 4-2, it can be seen that there are three IOT instructions to be implemented. These instructions are:
Skip if Converter Flag Set; Select and Convert; and Read Converter Buffer.
The only service the ADC handler performs is that of receiving data and storing it in user specified areas. This handler
will have a standard 256-word buffer.
b. Data Modes used (for example, !OPS ASCII, etc.J. Assuming that there is only one data mode for this device,
mode specification is unnecessary in Example B.
c. Which I/O macros are needed for the handler's specific use, that is, .INIT, .CLOSE, .READ, etc. These are fully
described in Chapter 2 of this manual. For an ADC, the user would be concerned with three of the macros.
.INIT would be used to set up the associated API channel register and the interrupt skip IOT sequence in the
Program Interrupt (PIC) skip chain. This is done by a CAL (N) as shown in Part III of Example A, where (N) is.
the channel address. The standard device/ API channel associations can be found in paragraph 4.1.3 .
.READ is used to transfer data from the ADC. When the .REAi{ macro is issued, the ADC handler will initiate
reading of the specified number of data words and then return control to the user. The analog input data
received is in its raw form; it is up to the programmer to convert the data to a usable format .
.WAIT detects the availability of the user's buffer area and ensures that the 1/0 transfer is completed. It would
be used to ensure a complete transfer before processing the requested data.
d. Implementation of the AP/ or PIG interrupt service routine. Example A shows an API or PIC interrupt service
routine that handles interrupts, processes the data and initiates new data requests to fully satisfy the .READ macro
request. Note that the routines in Example A will operate with or without APL Example B used the routines exactly
as they are shown in Example A.
During the actual writing of Example B, consideration was given to the implementation of the 1/0 Macros in the new
handler in one of the following ways:
(1) Execute the function in a manner appropriate to the given device as discussed in (c) . .INIT, .READ,

.WAIT were implemented into the ADC handler (Example B) under the subroutine names ADINIT,
ADREAD, ADWAIT.
Wait for completion of previous 1/0. (Example B shows the setting of the ADUND switch in the
ADREAD subroutine to indicate 1/0 underway.)
(2) Ignore the function if meaningless to the device. See Example B (.FSTAT results in JMP ADIGN2) in
the dispatch table DSPCH. For ignored macros, the return address must be incremented depending
upon the argument string after the CAL. The number of arguments for each macro is shown in Chapter

2.
(3) Issue an error message in the case where it is not possible to perform the 1/0 function. (An example
would be trying to execute an .ENTER on the paper tape reader.) In Example B the handler jumps to
DVERR6 which returns to the Monitor with a standard error code in the AC.

4-8(Part III)

After the handler has been written and assembled, the user must assemble the handler and splice it to the 1/0 Library
Tape. This procedure is described in 4.2.4.

4.2.1 Discussion of Example A by Parts
Part 1

Stores CAL pointer and argument pointer; also picks up function code
from argument string.

Part 2

By getting proper function code in Part 1 and adding a JMP DSPCH,
the CAL function is dispatched to proper routine.

Part 3

This is the .SETUP CAL used to set up the API channel register and PIC
skip chains. Section 4.1.3 of this manual shows the standard device/API
associations.

Part 4

Shows the API and PIC handlers. It is suggested these be used as shown.

Part 5

This area reserved for processing interrupt and performing any
additional 1/0.

Part 6

Interrupt dismiss routine.

Part 7

Increments argument pointer in bypassing arguments of ignored macro
CAL's.

4.2.2 Example A, Skeleton 1/0 Device Handler

'5i

i:i..

/SPECIALIZED I/O HANDLER
/CAL ENTRY ROUTINE
• GL00L DEV.
• MED:3
DAC
DVCALP
DEV.
DAC
DVARGP
DVARGP
ISZ
DVARGP
LAC*
(77777
AND
DVARGP
ISZ
TAD
(JMP DSPCH>
DAC
DSPCH
DSPCH
xx
JMP
DVI NIT
JMP
DVFSAT
JMP
DVSEEK
DVENTR
JPIP
JMP
DVCLER
DVCLOS
JPIP
DVMTAP
JMP
DVREAD
JMP
DVWRTE
JMP
DVWAIT
JMP
DVTRAN
JMP

/MUST BE OF FORM AAA •
I.MED CMONITOR ERROR DIAGNOSTIC>
/SAVE CAL POINTER
/AND ARGUMENT POINTER
/POINTS TO FUNCTION CODE
/GET CODE
/REMOVE UNIT I !F 4PPLICABLE
/POINTS TO CAL + 2
/DISPATCH WITH
/MODIFIED JUMP
/1 = • !NIT
/2 :: .FSTAT, .DELET, .RENAM
/3 = .SEEK
/4 : .ENTER
15 : .CLEAR
/6 :: .CLOSE
17 = .MTAPE
/10 = .READ
/ 11 = • WRITE
/12 = • WAIT
/13 = .TRAN

4-9(Part III)

/ILLEGAL FU~CTIONS IN ABOVE TABLE CODED AS:
/
JMP
DVERR6
/FUNCTION CODE ERROR
DVERR6 LAW 6
JMP*
C. MED+l >

/ERROR CODE 6
/TO MONITOR

/DATA MODE ERROR
DVERR7 LA•4 7
JMP*
C.MED+l>

/ERROR CODE 7

/DEVICE NOT READY
DVERR4 LAC
<RETURN>
DAC*
C. MED>
LAC C4 >
JMP*
C .MED+l)

/l/O UNDERWAY LOOP
DVBUSY D9R
JMP*
D\fCALP
/~ORMAL

DVCK

/TO MONITOR

/RETURN (ADDRESS IN HANDLER
/TO RETURN TO WHEN NOT READY
/CONDITION HAS 9EEN REMOVED>
/ERROR CODE 4
/TO MONITOR

/8RiAK FROM LEVEL 4
/LOOP ON CAL

RETURN FROM CAL
D9R

JMP*

DVARt;P

/BREAK FROM LEVEL 4
/RETURN AFTER CAL AND
/ARGUMENT STRING.

/THE DVINIT ROUTINE MUST INCLUDE
/A. SETUP FOR
/EACH FLAG CO~NECTED TO PIC CAT BUILD TIME>
/ONE OF THESE MAY ALSO BE THE API SETUP CALL.
/THE SETUP CALLING SEQUENCE IS:
DVINIT

CAL

= API CHANNEL REGISTER

/C40-77>; 0 IF NOT

/CONNECTED TO ~PI
16

SKP IOT
DBVINT

/IOPS FUNCTION CODE
/SKIP !OT TO TEST THE FLAG
/ADDRESS OF INTERRUPT
/HANDLER CAPI OR PIC>

/THIS SPACE CAN BE USED FOR I/O SUBROUTINES
/INTERRUPT HANDLER FOR API OR PIC
DEVPIC DAC
DEVAC
/SAVE AC
LAC*
(0)
/SACE PC, LINK,
DAC
DEVOUT
/MEM. PROT.
LAC
DEVI ON
/FORCE ION AT DISMISSAL
JMP
DVSTON
DEVINT JMP
DEVPIC
/PIC ENTRY
DAC
DEVAC
/API ENTRY, SAVE AC
/SAVE PC, LINK,
LAC
DEVI 'H
DAC
DEVOUT
/MEIWI. PROT.
I ORS
/CHECK STATUS OF PIC

4-IO(Part III)

SMA !CLA
LAW
17740
TAD
DEVI ON
DAC
DVS~~CH
DEVCF
ION

/FOR ~ESTORATION AT
/DISMISSAL
/CLEAR FLAG
/EN.t\9LE PIC

/THIS IS THE AREA DEVOTED TO PROCESSING INTERRUPT AND
{ /PERFORMIN9 ANY ADDITIONAL I/O DESIRED.
~
!OF
/DISABLE PIC TO !~SURE
DEVIOT
/DISMISSAL ~EFORE
/INTERRUPT FROM THIS
/IOT OCCURS
~

/INTERRUPT HANDLER DISMISS RTE
DVDISM LAC
CJMP DEVPIC>
DAC
DEVI NT
LAC
DEVAC
DVSWCH ION
DBR
JMP*
DEVOUT

/RESTORE PIC ENTRY
/RESTORE AC
/ION OR IOF
/DEBREAK AND RESTORE
/LINK, MEM. PROT.

/IF THE HANDLER USES THE AUTO-INDEX
/OR
/EAE REGISTERS, THEIR CONTENTS
/SHOULD BE
/SAVED AND RESTORED.
/FU~CTIONS POSSIBLY IGNORED SHOULD
/CONTAIN PROPER INDEXING TO BYPASS
/ARGUMENT STRING.

~ {DVI GN2

I SZ
JMP

DVARGP
DVCK

/BYPASS FILE POINTER

4.2.3 Example B, Special 1/0 Handler for Hypothetical A/D Converter

/ADC IOT'S
AOSF:701301
ADSC:701304
ADRB:701312
/ADSF:SKIP IF CONVERTER FLAG IS SET
/ADSC:SELECT ANO CONVERTCADC FLA3 IS LEARED
/AND A CONVERSION IS INITIALITED>
/ADRB:READ CONVERTER BUFFERCPUTS CONTENTS IN AC>
/CAL ENTRY ROUTINE
/ADC. IS GLOBAL NAME FOR HANDLER
.GLOBL ADC.
.MED::S
/MEDCMON!TOR ERROR DIAGNOSTIC)
DAC ADCALP
/SAVE CAL POINTER
ADC.
DAC ADARGP
/AND ARGUMENT POINTER
ISZ ADARGP
/POINTS TO FUNCTION CODE
LAC* ADARGP
/GET CODE
ISZ ADARGP
/POINTS TO CAL + 2
TAD CJMP DSPCH>
/DISPATCH WITH
DAC DSPCH
4-11 (Part III)

/MODIFIED JU:-!P
XX
/1 = • !NIT
JMP ADINIT
/2 = .FSTAT, • DEL ET, .RENAM
JMP AOir;N2
/3 = .SEEK
JMP ADIGN2
/4 = .ENTER
JMP ADl'RR6
/5 = .CLEAR
JMP ADFRR6
16 = .CLOSE
JMP ADOK
11 = .MTAPE
JMP ADOK
/10 = .READ
JMP ADREAD
I 11 = • WRITE
JMP ADERR6
JMP ADWAIT
/12 = • WAIT
JMP ADERR6
/13 = • TRAN
/ILLEGAL FUNCTIONS IN ABOVE TABLE CODED AS:
I
JMP ADERR6
/FUNCTION CODE ERROR
/ER~OR CODE 6
ADERR6 LAW 6
/TO MONITOR
JMP* C.MED+l)
/DATA MODE ERROR
/ERROR CODE 7
ADERR7 LAW 7
/TO MONITOR
JMP* C.MED+l >

DSPCH

/THE ADINT ROUTINE MUST INCLUDE A

I .SETUP

FOR
/EACH FLAG ASSOCIATED WITH THE
/DEVICE
/THE .SETUP CALLIN3 SEQUENCE IS:
ADINIT ISZ ADARGP
/IDY TO RET ST~DRD BUFF SIZE
.DEC
LAC C256
/STANDARD BUFFER SIZE CDECIMAL)

.ocT

DAC* ADARr;P
ISZ ADARllP
CAL 57

/PUT BACK STANDARD 3UFFER SIZE

/57:,Pl CHANNEL REGISTER
/(40-77>: 0 IF NOT
/CON~ECTED TO API
ADCKSM 16
I.SETUP !OPS FUNCTION CODE
ADSF
ADC BP
/ADC SKIP IOI TEST THE FLAG
ADLBHP ADCI tH
/ADDRESS OF INTERRUPT
/HANDLER CAPI OR PIC>
/ERASES CALL FUNCTION IN CASE
ADUND
LAC .+2
ADl4RC
DAC .-5
/OF FURTHER .!NIT'S THEY WILL BE
ADWPCT JMP ADSTOP
/IGNORED BY THIS JMP TO ADSTOP
/WHERE THE I/O UNDERWAY SWITCH
/IS CLEARED AND 1LL 1/0 IS
/TERMINATED
/THE PREVIOUS SIX TA~S IN THE CAL AREA ARE USED FOR TEMP
/STORAGE DURING THE ACTUAL .READ FUNCTION
/ADCKSM IS FOR STORING THE CHECKSUM
/ADDRP IS THE CURRENT BUFFER POINTER
/ADLRHP IS T~E LINE BUFFER HEADER POINTER
/ADUND IS FOR DEVICE UNDERWAY SWITCH
/ADR~C IS USED AS A -WORD COUNT REGISTER
/ADRWCT IS USED TO STORE CURRENT WORD PAIR COUNT
/STOP ADC ROUTINE CLEARS I/O U~DER WAY SWITCH
ADSTOP DZM ADUND
/ADC 14AIT LAC ADUND
SNA
JMP ADOK

4-12(Part III)

/I/.0 UNDER~AY LOOP
ADBUSY DBR
J~P* ADCALP
ADR EAD

LAC ADIJ~D
SZA!CMA
JMP ADBUSY
DAC ADUND
LAC* ADARGP
DAC ADDBP
D4C ADLRHP
ISZ ADARflP
LAC* ~DR~8P
D4C ADR'~C
ISZ ADl\R:1P
DZM AD'~PCT
DZM ADCKSM
ISZ ADDBP
ISZ ADDBP

ADSC
/NORMAL RETURN FROM CAL
ADOK
DBR
JMP* ADARGP

/CHECK TO SEE IF l/O IS UNDERWA~
/IF NOT SET IT WITH -1
/IT WAS SET, 80 9ACK TO CAL
/SET IT
/GET LINE BUFFER HEADER POINTER
/STORE !T
/ALSO STORE IT FOR LATER HEADER
/INCREMENT ARG. POINTER
/GET -L.B.W.C.<2'S COMP>
/STORE IT IN WORD COUNT REGISTER
/I~CREMENT FOR EXIT FROM .READ
/ZERO WORD PAIR COUNT REG.
/ZERO CHECKSUM ~EG.
/GET PAST HEADER PAlq
/NOW POINTING AT 2EGINNING OF
/BUFFER
/START UP DEVICE
/BREAK FROM LEVEL 4
/RETURN AFTER CAL

/INTERRUPT HANDLER FOR API OR PIC
/SAVE AC
ADCPIC DAC ADCAC
/SAVE PC, LINK,
LAC* (0)
DAC ADCOUT
/MEM. PROT.
/FORCE ION AT DISMISSAL
LAC ADCION
JMP ADSTON
ADCINT JMP ADCPIC
/PIC ENTRY.
/API ENTRv, SAVE ~C
DAC ADCAC
/SAVE PC, LINK,
LAC ADCINT
DAC ADCOUT
/MEM. PROT.
/RESTORE PIC ENTRY qECAUSE PIC
LAC CJMP l\DCPIC>
DAC ADCINT
/A JMS CALL NOT A JUMP
/CHECK STATUS OF PIC
!ORS
/FOR RESTORATION AT
SMA!CLA
/DISMISSAL
LAW 17740
TAD Al)CION
ADSTON DAC ADSWCH
/READ CONVERTER BUFFER
ADRB
/ENABLE PIC FOR OTHER DEVICES
ADC! ON I ON
/STORE DATA IN USER BUFFER
DAC* ADDBP
/INC. BUFFER POINTER
ISZ ADDBP
/INC. WORD PAIR COUNTER
ISZ ADWPCT
TAD ADCKSM
/ADD CHECKSUM
DAC ADCKSM
/STORE IT
I IS l/O COMPLETE
ISZ ADRINC
/NO KEEP GOING
JMP ADCONT
/YES, COMPUTE WORD COUNT PAIR
LAC ADWPCT
/ADD ONE MAY 9E ODD
TAD Cl
/DIVIDE BY TWO, AT SAME TIME
RTL
RTL
RTL
RTL

/ADJUST IT, FOR HEADER

4- l 3(Part III)

ADC ONT

AND <377000
DAC* ADLBHP
ISZ ADLBHP
TAD .l\DCKSM
DAC* ADLBHP
DZM ADUND
JMP ADDISM
I OF
ADSC

/AU. SET
/STORE IN HEADER #l
/INC. TO STORE CKSUM
/ADO WORD PAIR COUNT
/STORE IN HEADER 12
/CLEAR DEVICE UNDERWAY
/EXIT
/DISABLE PIC TO INSURE
/DISMISSAL ?EFORE
/I~TERRUPT

/INTERRUPT HANDLER DISMISS RTE
ADDISM LAC ADCAC
ADSWCH ION
DBR
JMP* ADCOUT
ADCAIP 0
ADARGP 0
ADCOUT 0
ADCAC
0

/IOT OCCURS

FROM THIS

/RESTORE AC
/ION OR IOF
/DEBREAK 4ND RESTORE
/LINK, MEM. PROT.
/ADC CAL POINTER
/l\DC 4RGUEMENT POINTER
/PC,L,MP
/AC SAVED HERE

/IF THE HANDLER USES THE AUTO-INDEX
/OR
/EAE REGISTERS, THEIR CONTENTS
/SHOULD BE
/SAVED AND RESTORED.
/FUNCTIONS POSSIBLY IGNORED SHOULD
/CONTAI~ PROPER INDEXING TO BYPASS
/ARriUMENT STRING.
ADIGN2 ISZ ADARGP
/BYPASS FILE POINTER
JMP ADOK
.END

4- I 4(Part III)

4.2.4 Incorporating User-Generated 1/0 Handlers
4.2.4.1 The following example shows how to incorporate a user generated 1/0 handler for a special or non-standard
device (e.g., Type 350 incremental plotter) into the standard l/O Library. This means that the Linking Loader, DDT,
or DDTNP must be modified to recognize the new handler. The changes to be made include .DAT Slot modification
and skip chain modification. If the user wishes to use his own handler name, instead of one of the standard handler
names (e.g., PLA., instead of LPA.), a radix SO value in the Loader 1/0 Configuration (IOC) Table must be changed.
4.2.4.2 Procedure:
a.

Load the Linking Loader as follows:
(I) Set the ADDRESS Switches to 17720 of the highest core bank.
(2) Place the Linking Loader tape in the reader.
(3) Press RESET then READIN.

b. Stop the computer at the point where the Loader types ts and modify the appropriate .DAT Slot (Slot 0 begins
at location 135) according to the Loader IOC Table (Table 4-2). Suppose .DAT Slot 4 is chosen and the user wishes to
substitute his plotter handler for handler LPA .. Cell 141 (.DAT Slot 4) should be changed from 000001 (Loader code
for TIA.) to 000007 (Loader code for LPA.).
c.

Modify the Monitor Skip Chain (Table 4-3) as follows:
(l) Determine the proper Skip IOT instruction code (octal) assigned to the device, either from the User's

Guide DEC-I S-H2DA-D (in the case of standard DEC 1/0 devices) or from engineering documentation
supplied to the customer (for customer specified devices).
(2) Load the new IOT in the desired Skip Chain location. In case of the plotter mentioned above, the Skip
IOT (from the User's Guide) is 702401 (PLSF). The skip associated with LPA. is at location 1533 as
shown in Table 4-3. Location 1533 should be changed from 706501 (LSDF) to 702401 (PLSF).
If the user has named his plotter handler LPA., he may proceed to step E below.

d. If the user prefers a different name* (e.g., PLA.), the radix 50 value associated with the LPA handler in the
IOC Table (Table 4-2) must be changed. Compute the new radix 50 value as follows:
( l) Find P in column l of Table 4-4 (062000).
(2) Find L in column 2 (000740).
(3) Find A in column 3 (000001).
(4) The resultant radix 50 value is 062741.
(5) The fourth character of the mnemonic PLA. is the period (.).
This is signified to the Loader by making the sign bit of the computer word (bit 0) a 1. The resulting
radix 50 value is now 462741.
*The new name must consist of from one to three alphabetic characters followed by a period.

4-l 5(Part Ill)

(6) Hence the contents of location 4SOS (Loader) or S264 (DDT) should be changed to 46274 l.
e.

Load PUNCH-1 S as follows:
(l) Set the ADDRESS Switches to 17720 of the highest core bank.

(2) Place the PUNCH-1 S tape in the reader.
(3) Press RESET then READIN.
f.

When the PUNCH-l S query "tP, TORS?" appears on the teletype, type CTRL P.

g. When PUNCH-1 S replies with a >, type the number of output tapes desired (in this case 1 - up to 9 if
desired.)
h. To load the resultant tape(s), proceed as in Step a. The computer stops after loading each tape except the
last. Press CONTINUE when the next tape is in the reader. Program execution begiAs automatically when the end of
the last tape is reached.
i. Assemble the new handler in relocatable binary form. Remove the end-of-file (EOF) block using PIP-IS with
the "W" switch option. Attach this tape to the end of the first tape in the I/O Library.

Table 4-2. Linking Loader IOC Table

Handler

TIA.
PRA.
PRB.
PPA.
PPB.
PPC.
LPA.
CDA.
CDB.
CDC.
MTF.
DRA.
DRB.
DRC.
DRD.

Code
Slot

Radix

Location

Loader/DDTNP

DDT

Value

4477
4500
4501
4502
4503
4504
4S05
4506
4507
4SIO
4511
4Sl2
4513
4514
4Sl5

5256
5257
5260
5261
5262
5263
S264
526S
5266
S267
S270
5271
5272
5273
5274

500041
463321
463322
463201
463202
463203
446601
411S4l
411S42
411543
452146
415721
415722
415723
415724

I
2
3
4

s
6
7
10
ll
12
13
14
IS
16
17

tThe IOC Table Locations shown above are subject to change in later versions of DDT and the Linking Loader. For information regarding
these changes, see the Reader Service Card at the end of this manual.

4-16(Part III)

Table 4-3. Basic 1/0 Monitor Skip Chain*
Locationt

Contents

Mnemonic

1521
1522
1523
1524
1525
1526
1527
1530
1531
1532
1533
1534
1535
1536
1537
1540
1541
1542
1543
1544
1545
1546
1547
1550
1551
1552
1553
1554
1555
1556
1557
1560
1561
1562
1563
1564
1565
1566
1567

703201
741000
621517
707341
741000
621604
700001
741000
600476
706701
741000
621575
706721
741000
621576
706501
741000
621572
700101
741000
621573
700201
741000
621574
700301
741000
601 I JO
700401
741000
601344
701741
741000
621600
701701
741000
621601
702701
741000
621602

SPF AL
SKP
JMP*INT6
MTSF
SKP
JMP* INT13
CLSF
SKP
JMPCLKPIC
RCSF
SKP
JMP*INT4
RCSD
SKP
JMP*INT5
LSDF
SKP
JMP* INTI
RSF
SKP
JMP* INT2
PSF
SKP
JMP* INT3
KSF
SKP
JMPTIINT

TSF
SKP
JMPTOINT
MPSNE
SKP
JMP* INT7
MPSK
SKP
JMP* INTlO
SPE
SKP
JMP* INTI l

Device Meaning

POWER FAIL

MAGTAPE DONE

CLOCK OVERFLOW

CARD COLUMN READY

CARD DONE

LINE PRINTER DONE

PAPER TAPE READER DONE'

PAPER TAPE PUNCH DONE

KEYBOARD READY

TELEPRINTER DONE

NON-EXISTENT MEMORY

MEMORY PROTECT VIOLATION

MEMORY PARITY ERROR

*This skip chain memory map applies only when the Linking Loader, DDT, DDTNP, CHAIN, EXECUTE or relocatable user programs are in core.
tThe skip chain locations shown above are subject to change in later versions of the associated software. For information regarding these
changes, see the Reader Service Care at the end of this manual.

4-l 7(Part III)

Table 4-4. Radix 508 Values

x -.
A
B

c
D

E
F
G

H
I
J
K
L
M
N
0
p

Q
R

s
T

v
w
x

z
%

0
I

2
3
4
5
6
7
8
9

003100
006200
011300
014400
017500
022600
025700
031000
034100
037200
042300
045400
050500
053600
056700
062000
065100
070200
073300
076400
101500
104600
107700
113000
116100
121200
124300
127400
132500
135600
140700
144000
147100
152200
155300
160400
163500
166600
171700

-XA

c
D

E
F
G

H
I
J
K
L
M
N
0
p

Q
R

s
T

v
w
x

z
%

0
1
2
3
4
5
6
7
8
9

000050
000120
000170
000240
000310
000360
000430
000500
000550
000620
000670
000740
001010
001060
001130
001200
001250
001320
001370
001440
001510
001560
001630
001700
001750
002020
002070
002140
002210
002260
002330
002400
002450
002520
002570
002640
002710
002760
003030

--X
A

c
D

E
F
G
H
I

J
K
L
M
N
0
p

Q
R

s
T

v
w
x
y

z
%

0
1
2
3
4
5
6
7
8
9

000001
000002
000003
000004
000005
000006
000007
000010
000011
000012
000013
000014
000015
000016
000017
000020
000021
000022
000023
000024
000025
000026
000027
000030
000031
000032
000033
000034
000035
000036
000037
000040
000041
000042
000043
000044
000045
000046
000047

4.3 1/0 HANDLERS ACCEPTABLE TO SYSTEM PROGRAMS
This section lists the .DAT slot requirements of system programs, the uses made of the .DAT slots, and the 1/0
handlers that may be assigned to each. It is imperative that one and only one 1/0 handler for a device be in core at the

4-l 8(Part III)

same time; that is, PRA and PRB should not be brought in together since there is no communication between the two
interrupt handlers.

NOTE

All system programs use .DAT slots -2 and -3 for teletype input and output,
respectively. These assignments cannot be changed.

4.3.1 FORTRAN IV
.DAT Slot

Use

-I I

Input

Handler

TTA
PRA
PRB*
MTF (non-file oriented)
CDB

-12

Listing

TTA*
LPA
PPA
MTF (non-file oriented)

-13

Output

PPA
PPB*
PPC
MTF (non-file oriented)

4.3.2 MACR0-15
MACR0-15 is identical to FORTRAN IV, with two exceptions: a If .ABS binary output is requested on .DAT slot
-13, PPC. cannot be used. b When the P option (in the command string) is used, .DAT slot -IO is the secondary input
device:

TTA*
PRA
PRB
CDB
4.3.3 FOCAL
.DAT Slot

Use

Handler

Input

PRA*
PRB
CDB

*DEC Standard assignment.

4-19(Part III)

.DAT Slot

Use

Handler

Output

PPA*
LPA

4.3.4 Editor

4.3.5

.DAT Slot

Use

Handler

-IO

Secondary Input

TIA
PRA*
PRB
CDB

-14

Input

TIA
PRA*
PRB
COB

-15

Output

LPA
TIA
PPA*

.DAT Slot

Use

Handler

-1

System
Library

PRA*

-4

Input

PRA*

-5

External
User Library

Linking Loader and DDT

NONE*
PRA

4.3.6 PIP
.DAT Slot

3
4
5
6
7
IO

*DEC Standard assignment

4-20(Part III)

Use

Handler

1/0
1/0
Input
Output
Output
Input
Output
Input

TIA*
TIA*
PRA*
LPA*
PPA*
COB*
PPA*
PRA*

4.3.7 Chain Builder
.DAT Slot

Use

Handler

-1

System Library
User Programs
External
Library
Output

PRA*
PRA*
NONE*

-4

-5
-6

PPA*
PPB
PPC

4.3.8 Chain Execute
.DAT Slot

Use

Handler

-4

Chain Input

PRA*

.DAT Slot

Use

Handler

-14
-15

Output
Input

PPA*
PRA*
CDB

4.3.9 STRAN

4.4 SUMMARY OF STANDARD 1/0 HANDLER FEATURES

4.4.I TTA·(Teletype)
a. Functions - All function descriptions (except READ and WRITE) refer to action taken when either the
teleprinter or the keyboard is addressed.
Mnemonic

.INIT

Code

Action
(l)

Return standard buffer size (34 1 0 ).

(2)

Assign return addresses for certain control characters from contents of
CAL ADDRESS +2. Bits 0 through I in CAL + 2 address are set to
designate caller:
B0,1=01
BO, I = 10
BO, I= 00

caller = Monitor
caller= DDT
caller = any other user

(3) Set I/O UNDERWAY indicator.
(4) Print carriage return and line feed (CR/LF).

*DEC Standard Assignment

4-21 (Part III)

Mnemonic

Code

.DLETE
.REN AM
.FSTAT
.SEEK
.ENTER
.CLEAR
.CLOSE

2
2
2
3
4
5
6

.MTAPE
.READ

7
10

.WRITE

.WAIT
.TRAN

12
13

Action

Ignored
Ignored
Ignored
Ignored
Ignored
Ignored
(1) Set 1/0 underway indicator.
(2) Print CR/LF.
(3) Wait on .CLOSE for completion of 1/0.
Ignored
(1) Set 1/0 underway indicator .
(2) Set up to accept characters from keyboard.
(1) Set 1/0 Underway indicator.
(2) Print line.
Allow input at output to finish .
IOPS06-lllegal function .

b. Legal Data Modes
IOPS ASCII (Mode 2)
IMAGE ALPHANUMERIC (Mode 3)
c. Vertical Carriage Control Characters
Output -

(I ) Line feed (12 8 )

Input -

IOPS ASCII: Ignore all leading line feeds; otherwise output
IMAGE: Output
(2) Others (vertical tab, form feed) output
Inserted in buffer

d. Horizontal Ca"iage Control Characters

Tab (11 8 ) in or out
IOPS ASCII

(1) Model 33 - output sufficient number of spaces to place the next typed

Image Alphanumeric

character in column 11, ... , 71. Insert only 11 8 in buffer on input.
(2) Model 35 - input, insert 11 8 in buffer. Output, print tab.
Input, insert I 18 in buffer. Output, print tab.

e. Program Control Characters-IN

Stop current 1/0
to Teletype.
Decode character
and echo on Teleprinter.*

(I) CTRL C transfers control to the address specified as return in the
.INIT performed by the Monitor.

*Character will be ignored (not echoed) in cases (I), (2), and (3), if respective . .INIT has not been performed.

4-22(Part Ill)

(2) CTRL P transfers control to the address specified as return in the .INIT
performed by the user (other than the Monitor or DDT).
(3) CTRL T transfers control to the address specified as return in the .INIT
performed by DDT.
(4) CTRL S transfers control to the address specified in .SCOM + 6.

f. Data Control Characters-JN

All characters inserted in buffer as 7-bit characters.

Image Alphanumeric
IOPS ASCII

(l) Rubout. Delete previous character typed. Type out reverse slash (\).

(2) CTRL U. Delete entire line typed so far. Type out commercial at (@). If
output is UNDERWAY, printing is terminated and a CR/LF is output.

g. Data Control Characters-OUT (both modes)

Ignore RUBOUT ( 177 8 ) and NULL (00).
In Image Alpha mode, a RUBOUT should be used to fill the last word pair when an odd number of characters is
to be output.
h. Errors (no program-initiated recovery)

IOPS6 - Illegal Function
IOPS7 - Illegal Data Mode
i. Program Sfze
469 1 0 registers (this is included in resident MONITOR).
j. Teletype J/O - Can be requested only from mainstream in API systems, since the Teletype is not connected to
the APL

4.4.2 PP (Paper Tape Punch)
a. Functions
Mnemonic

.INIT

Action

Code
(l)

Return standard buffer size (52 1 0 ).

(2)

.SETUP - no APL

(3)

Punch two fanfolds of leader.

4-23(Part III)

Mnemonic

Code

Action

.DLETE

.RENAM

.FSTAT

.SEEK

IOPS6 - Illegal function

.ENTER

Ignored

.CLEAR

Ignored

.CLOSE

Ignored
Ignored
Ignored

(1)

Allow previous output to terminate.

(2)

Punch EOF if IOPS Binary

(3)

Punch two fanfolds of trailer

(4)

Allow trailer punching to terminate

.MTAPE

Ignored

.READ

IOPS6 - Illegal function

.WRITE

(1)

Allow previous output to terminate.

(2)

Output buffer

.WAIT

Allow previous output to terminate.

.TRAN

IOPS6 - Illegal function

b. Legal Data Modes
IOPS Binary

(mode 0)

IMAGE Binary

(mode 1)

IOPSASCII

(mode 2)

IMAGE ALPHANUMERIC

(mode 3)

Dump

(mode 4)

c. Vertical Control Characters (/OPS ASCII only)
May appear only as first character of line, if elsewhere in line will be ignored; if no vertical control character at
beginning of line, a line feed (012) will be used.

4-24(Part III)

012

Line feed

013

Vertical tab, followed by four deletes ( 177)

014

Form feed, followed by 408 nulls (000)

d. Horizontal Control Characters (/OPS ASCII only)

011

Horizontal tab, followed by one delete (I 77)

e. Recoverable E"ors
No tape in punch - Monitor error IOPS4
(1)

Put tape in punch

(2)

Type CTRL R

(1)

.SEEK

(2)

.READ

(3)

.TRAN

f. Unrecoverable E"ors

IOPS6 - Illegal Function:

IOPS7 - Illegal Data Mode
g. Program Size

PPA. (all data modes)

397 decimal registers

PPB. (all except IOPS ASCII)

270 decimal registers

PPC. (IOPS binary only)

210 decimal registers
NOTE

In API systems, the paper tape punch can be called only from mainstream,
since the punch is not connected to the API.

4.4.3 PR (Paper Tape Reader)
a. Functions
Mnemonic

.INIT

Action

Code
(1)

Return standard line buffer size (52 1 0 )

(2)

.SETUP API channel register 508

4-25(Part III)

Mnemonic

Action

Code

(3)

Clear 1/0 UNDERWAY indicator

Ignored

.DLETE

.RENAM

.FSTAT

.SEEK

Ignored

.ENTER

IOPS6 - Illegal function

.CLEAR

IOPS6 - Illegal function

.CLOSE

Allow previous input to finish and then clear 1/0 UNDERWAY
indicator.

.MT APE

Ignored

.READ

Ignored
Ignored

(I)

Allow previous input to be completed.

(2)

Input line or block of data (see modes below).

.WRITE

IOPS6 - Illegal function

.WAIT

Allow previous input to be completed before allowing user program to
continue.

.TRAN

IOPS6 - Illegal function

b. Legal Data Modes
IOPS ASCII
(mode 2)

(I)

Constructs line buffer header, computing:
Word pair count
Data mode
Data validity bits

4-26(Part III)

(2)

Packs characters into the line buffer in 5/7 ASCII, checking parity
(eighth bit, even) on each character.

(3)

Allows vertical form control characters. (FF, LF, VT) only in character
position I of the line buffer. Otherwise, ignored.

(4)

Terminates reading on CR or line buffer overflow. In the latter case,
tape is moved past the next CR to be encountered.

IOPS Binary
(Mode 0)

Image
Alphanumeric
(Mode 3)

(I)

Reads binary data in alphanumeric mode, checking parity (seventh
hold, odd) on each frame.

(2)

Accepts line buffer header at head of input data, modifying data
validity bits if parity or checksum errors (or short line) have occurred.

(3)

Terminates reading on overflow of word pair count in line buffer
header or word count in .READ macro, whichever is smaller, moving
tape to end of line or block if necessary.

(l)

Constructs line buffer header, computing:
Word pair count
Data mode

(2)

Stores characters, without editing, or parity checking in the line buffer,
one per register.

(3)

Terminates reading as a function of .READ macro word count.

Image Binary
(Model)

Same as Image Alphanumeric, except a binary read is issued to the
PTR.

Dump
(Mode 4)

Same as Image Alphanumeric except, no header is constructed; loading
begins at the core address specified in the .READ macro. A binary read
is issued to the PTR.

NOTE

An end of tape condition causes the PTR interrupt service routine to
terminate the input line, turning off the 1/0 UNDERWAY program indicator

and marking the header (data mode bits) as an EOM (end of medium) for all
modes except Dump.
c. Unrecoverable Errors

(IOPS6)

Illegal Function

Illegal Data Mode

(I)

.ENTER

(2)

.CLEAR

(3)

.WRITE

(4)

.TRAN
(IOPS7)

4-27(Part III)

d. Program Size

PRA. (all data modes)

436 decimal registers

PRB. (IOPS ASCII only)

287 decimal registers

4.4.4 COB. (CARD READER CR03B)

NOTE

When the CR03B is used as an EDITOR input, a blank card must be placed
immediately after the End of File (EOF) card (card column I punched
12-11-0-1-2-3-4-5-6-7-8-9). If an EOF card is not used to end a deck, the
blank card is not required.

a. Functions
Mnemonic

Code

.INIT

Action
(I)

Return standard buffer size (36 1 0 )

(2)

Call .SETUP to update skipchain with PIC servicer addresses for column
ready and card done flags and to place API servicer address in location
55 8 (API channel 13).

Ignored

.DLETE

.RENAM

.FSTAT

.SEEK

Ignored

.ENTER

IOPS6 - Illegal function

.CLEAR

IOPS6 - Illegal function

.CLOSE

Allow previously requested input to terminate

.MTAPE

Ignored

.READ

.WRITE

4-28(Part III)

Ignored
Ignored

(I)

Allow previously requested input to terminate .

(2)

Ensure that device is ready.

(3)

Initiate input of next card.
IOPS6 - Illegal function

Action

Mnemonic

Code

.WAIT

Allow previously requested input to terminate.

.TRAN

IOPS6 - Illegal function

b. Legal Data Modes

IOPS ASCII (mode 2) Eighty card columns are read and interpreted as Hollerith (029 or 026) data, mapped into the
corresponding 64-graphic subset of ASCII, and stored in the user's line buffers in 5/7 format (36 1 0 locations required
to store an 80 column card). Compression of internal blanks to tabs and truncation of trailing blanks is not
performed; all 80 characters appearing on the card are delivered to the caller's line buffer. In addition, a carriage
return (015 8 ) character is appended to the input line. A total of 81 ASCII characters are thus returned by the handler
in IOPS ASCII mode.
All illegal punch configurations (i.e., those not appearing in the 029 or 026) cause an IOPS4 error. In response to the
error, the user can:
(a) correct the card, reinsert it into the reader input hoppers, and type CTRL R to restart the read operation;
(b) attempt to restart read (CTRL R) without correcting the card;
(c) remove the card from the read hopper and restart read (CTRL R).
The single addition to the Hollerith set, one made necessary by the constraints of system programs, is the provision
for the internal generation of the ALT MODE terminator. The appearance of a 12-1-8 punch (multiple-punched A/8)
on the card is mapped into the standard PDP-15 ALT MODE character (175 8 ) in the user's line.
When card processing is complete, word 1 of the header is constructed and stored in the caller's line buffer area. Word
2 of the header, the checksum location, is never disturbed by the card reader handler in IOPS ASCII mode.
Refer to Appendix B for a listing of legal Hollerith codes and their corresponding ASCII graphics.
c. Recoverable Errors

IOPS4 - Reader Not Ready
(I)

Hopper Empty

(2)

Stacker Full

(3)

Feed check, LIGHT CHECK (may be hardware failure)

(4)

Read Check, DARK FAIL (may be hardware failure)

(5)

Reader not ready:
(a) Stop button depressed.
(b) Start button not depressed.

(6)

Validity check.

(7)

Pick Fail card selected but not passed from hopper to read stations.
4-29(Part Ill)

d. Unrecoverable Errors

IOPS6 - Illegal Function
(I)

.ENTER

(2)

.CLEAR

(3)

.WRITE

(4)

.TRAN

e. Program Size
CD B. occupies 41 0 1 0 locations.
f. Assembly Procedures

CDB. exists in the I/O Library in a version which recognizes DEC 029 code (Appendix B). It also is supplied to the
user as a source tape to allow an alternate assembly of the handler which will recognize DEC 026 code.
The following is the procedure for assembling CDB. in 026 mode:
(I) Prepare to assemble the CDB. source tape using MACR0-15 with the "P" option.*

(2) Enter the following parameter on the Teletype:
DEC026 = 1 (CTRL D)
(3) After assembly, remove the end-of-file (EOF) block from the new CDB. tape by using PIP with the "W"
switch option.
(4) This tape becomes the first tape in the I/O Library.

4.4.5 LPA•(647 LINE PRINTER)
a. Function
Mnemonic

.INIT

Action

Code
(l)

Return standard line buffer size (52 1 0 )

(2)

.SETUP - API channel register 56 8

(3)

Clear line printer buffer

(4)

Form feed

*Assembly without the "P" option produces the 029 version of the handler.

4-30(Part Ill)

Mnemonic

Code

.DLETE

.RENAM

.FSTAT

.SEEK

IOPS6 - Illegal function

.ENTER

Ignored

.CLEAR

Ignored

.CLOSE

Action

Ignored
Ignored
Ignored

(1)

Allow previous output to terminate

(2)

Form feed

(3)

Allow form feed to terminate

.MTAPE

Ignored

.READ

IOPS6 - Illegal function

.WRITE

(1)

Allow previous output to terminate

(2)

Output line

.WAIT

Allow previous output to terminate

.TRAN

IOPS6 - Illegal function

b. Legal Data Modes
IOPS ASCII
(mode 2)
c. Vertical Control Characters (when first character of line)
12

Print every line

Print every second line

Print every third line

Print every sixth line

Print every tenth line

Print every twentieth line

4-3l(Part III)

Overprint

Form feed

d. Horizontal Control Characters (anywhere in line)

Horizontal tab - converted to N spaces, where N is the number necessary to have the next
character in column I I , 21, 31 , 41 , ...

e. Recoverable Errors
Monitor error message, .IOPS4. Make device ready, then type CTRL R
to continue.

Device not
ready
f. Unrecoverable Errors

Ml ;tor error message, .IOPS6 XXXXXX, where XXXXXX is address
of ror CAL.

Illegal function

Illegal data
mode

(I)

.SEEK

(2)

.READ

(3)

.TRAN
Monitor error message, .IOPS7 XXXXXX, where XXXXXX is address
of error CAL.
Any mode other than IOPS ASCII.

g. Program Size

297 (decimal) registers.

4.4.6 MTF. (TU20, TU20A, TU30, TU30A Tape Transports)
a. Introduction
MTF. is a general magnetic tape 1/0 handler intended principally to meet the immediate needs of PDP-15 FORTRAN
users. It will accept data-transfer requests in data modes used by the FORTRAN IV Object Time System and in
addition,
honor the following FORTRAN statements:

will

REWINDu
ENDFILE u
BACKSPACEu

4-32(Part III)

The magnetic tape facility is also available, via MTF., to MACRO-IS users. It should be noted. however, that legal
data modes and functions are limited to those required in the FORTRAN environment and the special 9-channel
dump mode.

b. Functions
Mnemonic

Action

Code

.INIT

(I)

Return standard buffer size (56 1 0 = 708 ).

(2)

Call .SETUP for API channel 45, if first time through.

(3)

Set transfer direction for drive referenced. Direction is specified by bit
8 of CAL instruction:
Bit 8 = I, this drive is output;
Bit 8 = 0, this drive is input.
Set default conditions for this drive, viz:

(4)

(a) This drive will transfer in odd parity.
(b) This drive will transfer at 800 BPI.
(c) This drive is 7-channel.*
.OPER

Ignored

.FSTAT

Ignored

.DLETE

Ignored

.REN AM

Ignored

.SEEK

(I)

If the specified drive has been INITed and transfer direction is input,
return to caller.

(2)

If the specified drive has been INITed and transfer direction is not

input, IOPS6.

.ENTER

(3)

If the specified drive has not been INITed and transfer direction is
· input, perform .INIT for this drive (input) .

(I)

If the specified drive has been INITed and transfer direction is output

return to caller.

*Unless the assembly parameter had been used to change the default assumption to 9-<:hannel. Variable MT9CHN need only be defined
for 9-<:hannel paper-tape systems.

4-33(Part Ill)

b. Functions (cont)
Mnemonic

Code

(2)

Action
If the specified drive has been INITed and transfer direction is not

output, IOPS6.
(3)

If the specified drive has not been INITed and transfer direction is

output, perform .INIT of this drive (output) .
.CLEAR

.CLOSE

Ignored
(I)

Wait for current 1/0 to complete .

(2)

Is this drive an input drive?
Yes:
No:

(3)

Is this drive open for transfers?
No:
Yes:

.MTAPE

return to caller.
continue.

(4)

Write two EOF markers (7-channel = 17 8 ; 9-channel = 23 8 ), hang
on CAL during transfers.

(5)

Backspace one record, hang on CAL during backspace. This positions
the read head between the two EOF markers written.

(6)

Set switch indicating this drive is not open for transfers.

(7)

Return to caller at CAL + 2

(I)

Wait for current 1/0 to complete.

(2)

Interpret subfunction (bits 5-8) of CAL instruction.
Subfunctions:
0- Rewind

(a) Set indicator that 1/0 is not underway.
(b) Issue rewind to this drive.
I - Undefined

IOPS6.
2 - Backspace Record

4-34(Part III)

Mnemonic

Action

Code

(a) Set indicator that 1/0 is underway.
(b) Issue backspace-one-record to this drive. Note: the backspace is
attempted at the density currently assigned to this drive (via .INIT or
.MTAPE 10-17).
3 - Backspace File
IJlegal: IOPS6
4-Write EOF
5 - Space Forward Record

(a) Check 1/0 transfer direction for this drive.
i.

Output: IOPS 6

ii. Input: continue.

(b) Set indicator that I/O is underway
(c) Issue space-forward-one-record to this drive.
6 - Space Forward File

Illegal: IOPS6
7 - Space to Logical EOT

Illegal:

IOPS error, code 6.

I 0-17 - Describe Tape Configuration

Update format descriptor bits for this drive. Note that these bits will
be reset by any subsequent .INIT to this drive.
1/0 transfers (including space operations) will be performed in the
parity, density, and channel-count specified in .MTAPE l 0-17, thus:
10:7-channel, even,
11 :7-channel, even,
12:7-channel, even,
14:7-channel, odd,
15:7-channel, odd,
16: 7-channel, odd,
13:9-channel, even,
17:9-channel, odd,

.READ

(I)

200 BPI
556 BPI
800 BPI
200 BPI
556 BPI
800 BPI
800 BPI
800 BPI

Wait for current l/O to complete.
4-35(Part Ill)

Mnemonic

Code

Action

(2)

Ensure that requested drive is available (IOPS4 if not).

(3)

Parameters:
(a) Buffer address:

from CAL sequence

(b) Word count (maximum):
(c) Tape format:

from CAL sequence

from previous .INIT or .MTAPE 10-17.

(4)

Data Format: Data are transferred directly from magtape to the user's
core buffer area. No editing of any kind is performed by MTF.

(5)

Parity Errors: Hardware parity errors during an input transfer cause the
proper bits to be set in word 0 of the line buffer header (bits 12-13 =
0 I if an error occurred). No reread is attempted.

(6)

End of File, End of Tape
(a) End of File: When an end-of-file marker (EOF) is encountered on
the tape being read, an EOF pseudo-line is constructed and stored
in the user's line buffer area.* This EOF line consists of a two-word
header and one data word, thus:
header word 0:
header word I :
data word 0:
data word I:

002005
775773
000000
Unchanged

(b) Physical End of Tape: When an end-of-tape is encountered on the
tape being read, an end-of-medium line is constructed and stored in
the user's line buffer area.* The EOM line consists of a two-word
header and one data word, thus:
header word 0:
header word l :
data word 0:
data word l:

002006
775772
000000
unchanged

An attempt is then made to backspace over the record being read.
The assumption is made that the record was written through the
end-of-tape.

.WRITE

(1)

Wait for current 1/0 to complete.

(2)

Ensure that requested drive is available (IOPS 4 if not).

*The action described takes place only if the 1/0 transfer had been performed in Mode 0 or Mode 2 (see Section C). For error activity in
Mode 5 transfers, see Section C.3.

4-36(Part III)

Mnemonic

Code

Action

(3)

Parameters:
(a) Line buffer address:

.WAIT

.WAITR

.TRAN

from CAL sequence.

(b) Word count:

from bits 1-8 of line buffer header.*

from previous .INIT or .MTAPE 10-17.

(4)

Data Format: Data are transferred directly from the user's line buffer
area to the device. No editing of any kind is performed by MTF. Each
. WRITE results in exactly one physical record on tape.

(5)

Write errors: 5 attempts are made to rewrite the record. Before each
attempt, a 3-inch length of tape is erased. If the fifth try is unsuccessful, an IOPS 4 return is made to the monitor. The user then has
the option of trying to write 5 more times (CTRL R) or restarting
with a fresh tape.

(6)

End of Tape: When EOT is detected during writing, an error return
(IOPSI 5) is made to the monitor.

(I)

Wait for current 1/0 to complete.

(2)

Return to CAL + 2
Determine whether current 1/0 is complete .

(1)

If 1/0 is incomplete, transfer to location specified in CAL+ 2.

(2)

If 1/0 is complete, return to CAL + 3.

The .TRAN request (for forward-direction Transfers only) is honored
and results in a binary transfer from or to the next physical record on
tape. The device-address agreement in the CAL sequence (CAL + 2) is
ignored by the .TRAN processor.

c. legal Data Modes

( 1)

IOPS Binary (Mode 0) - 7 or 9 channel transports

(2)

IOPS ASCII (Mode 2) - 7 or 9 channel transports

(3)

9~Channel Dump (mode 5)- 9 channel transports only.

(a) This mode is designed to take advantage of all 8 data bits in each
frame of 9-channel tape and thus ensure maximum throughput for
*In Modes 0 and 2 only (see Section C). For Mode 5 transfers, the word count is taken from the CAL sequence.

4-37(Part Ill)

Mnemonic

Code

Action

each word transferred. Whereas normal (Modes 0 and 2) transfers
require three tape frames for each PDP-IS register, transfer in Mode
5 require only two frames per word. Only 16 bits of each PDP-15
word, however, are read from or written on tape. Word format for
Mode S is described below:
Parity Bits (P0 , P1 ): Bits 0 and I of the PDP-15 word are used
as parity bits for the 2 8-bit data bytes (D 0 and D 1 ),
respectively, in the low-order portion of the register. During an
output (write) transfer, these bits are ignored. The hardware will
generate the proper parity for each data frame. During an input
(read) transfer, these bits are set to the actual parity values for
the two data frames as read from tape.
Data Bytes (D0 , D1 ): Bytes 2-9 and 10-17 hold the values of
two adjacent 8-bit frames. During reads and writes, D0 is the
first frame transferred. If a record containing an odd number of
frames is read, the final frame is stored in D0 ; D 1 is set to binary
zeroes. An even number of frames are always written during an
output operation.
(b) Error Treatment in Mode 5
Because no line-buffer header is present in data transferred in Mode
5, there is presently no facility for indicating 1/0 errors to the user
program, particularly during input transfers. MTF. will do its best
to transfer correct data but, if it fails, cannot so inform the calling
program.
Read Errors
Parity Errors: 5 attempts are made to reread the error record. If
the record cannot be read successfully, the data resulting from
the fifth reread attempt are left in the user's buffer area.
Buffer Overflow: If the number of words in the record being
read exceeds the count given in the CAL sequence, transfer is
halted.
1 :
End of File: No data are transferred. The read head is
positioned preceding the EOT refledive spot.
Write Errors
As described in paragraph e. under .WRITE.
(c) Restrictions
Mode 5 is a legal 1/0 mode only if the referenced transport is
9-channel. An attempt to use Mode 5 in reading or writing a
4-38(Part IJI)

Mnemonic

Code

Action

7-channel drive results in an error return (IOPS7) to the Monitor.
Requests for transfers in modes other than 0, 2, and 5 result in
error returns (IOPS7) to the Monitor.
d. Program Size

MTF. currently occupies 11448 PDP-15 registers.

4-39/4-40(Part III)