mikpe.pdp10-tools/erlang/apps/as/doc/macro10.txt

MACRO-10 Assembler Syntax
*************************

This is a brief summary of the basic lexical and syntactic elements
of the MACRO-10 Assembler.  Directives and macros are not covered.

Numbers
=======
Numbers are in octal by default.

The default number radix can be changed using the RADIX directive.

A single-digit number is always interpreted as radix 10, that is,
"9" denotes decimal 9 not an error even if RADIX is 8 or 2.

The radix of a number can also be specified by prefixing the number
with ^B for binary, ^O for octal, or ^D for decimal. (These are two-
character sequences "up-arrow" "B" etc, not control characters.)

Hexadecimal radix is not supported.

The K, M, and G suffixes append 3, 6, or 9 zeros, respectively, to
a number.  (This has surprising effects outside of RADIX 10.)

Floating-point numbers are expressed in decimal with a mandatory decimal
point and optional exponent, e.g. 3.14, 3.14E9, 3.14E-7, -3.14E+7.

A "Bn" suffix shifts the number left by 35-n bits.  Excess bits are lost.
E.g. "1B17" is "000001 000000".

Using the "_" infix binary operator in an expression "X_Y" shifts X Y bits
to the left if Y is non-negative, and -Y bits to the right if Y is negative.
The operands X and Y may be any expressions, including symbols.

Prefixing a number by ^L (a two-character sequence, not a control character)
yields the number of leading zeros in the binary representation of the number.
E.g., ^L0 generates 36 decimal, ^L153 generates 29 decimal, and ^L-153 generates 0.

Literals
========
A literal is a piece of code within square brackets.  The enclosed code is
emitted to the literals pool (normally at the end of the program), and the
literal is replaced by the code's address.

Literals can be nested.

A dot (".") in a literal refers to the statement containing the outermost
literal.

Symbols
=======
Symbols may be composed of letters, digits, dots ("."), dollar signs ("$"),
and percent signs ("%").  Other characters are considered delimiters.

A symbol may not start with a digit.

If the first character is a dot, the second must not be a digit or another dot.
(MACRO-10 may generate symbols starting with two dots.)

Only the first six characters are significant.

Labels
------
A symbol in the first (left-most) field of a statement suffixed by a colon (":")
defines a label for that location.  A dobule colon ("::") marks the label internal.
There may be more than one label in a statement.  Labels may not be redefined.

Direct-Assignment Symbols
-------------------------
A symbol is created and associated with a value using "symbol=expression" or
"symbol=:expression" (in this case the symbol is internal).  These symbols may
be redefined at any time.

Variable Symbols
----------------
Suffixing a symbol with a number sign ("#") makes it a variable symbol.  Such
symbols are allocated by the assembler, either following the literals pool, or
into the code via VAR directives.

Symbol Visibility
-----------------
Symbols are marked local (INTERNAL) via the INTERN directive, or by following
the label or direct-assignment definition with an extra colon, i.e., "::" or "=:".

Symbols are marked global (EXTERNAL) via the EXTERN directive, or by mentioning
the symbol at least once with a "##" suffix, .e.g. FLAG7## .

Expressions
===========
Numbers and symbols can be combined with arithmetic and logical operators, or grouped
between angle brackets "<" and ">" to form expressions.

Operators
---------
The infix binary arithmetic operators are "+" (addition), "-" (subtraction),
"*" (multiplication), and "/" (division).

The infix binary logical operators are "&" (bit-wise and), "!" (bit-wise or), and
^!" (bit-wise xor), while  "^-" is the bit-wise one's complement prefix operator.

The infix binary operator ",," combines the low 18 bits of its operands to form a
36-bit word.

All operators operate on and generate 36-bit values.

Priority
--------
Unary operators and shifts have highest priority: "+", "-", "^-", "^D", "^O",
"^B", "B" (binary shift), "_" (underscore shift, "^F", "^L", "E", "K", "M", "G".
The K, M, and G postfix operators are performed first.

Logical binary operators have second-highest priority, and are evaluated left to
right: "!", "^!", "&".

Multiplicative operators have second-lowest priority, and are evaluated left to
right: "*", "/".

Additive operators have lowest priority: "*", "-".

[TODO: what is the priority of ",,"?  I suspect lower than additive operators]

Strings
=======
String operands to data emitting directives, such as ASCII, ASCIZ, COMMENT,
or SIXBIT, use user-selectable delimiters: DtextD, .e.g, /foo/ or "bar".  The
delimiter cannot be a blank, and it cannot be quoted inside the text.

A string of at most five 7-bit ASCII characters can be assembled into a single
left-justified word, by enclosing it with double quotes, "text".

A string of at most six SIXBIT characters can be assembled into a single
left-justified word, by exclosing it with single quotes, 'text'.

Comments
========
A comment starts with a semi-colon ";" and extends to the end of the line.

A comment extends to the next line if the first line ends with CTRL-_ (control
underscore, ASCII 037), followed by CR-LF.

Instruction Statements
======================
The general form of an instruction statement is

        label:  mnemonic accumulator,address    ; comment
or      label:  mnemonic accumulator,           ; comment
or      label:  mnemonic address                ; comment

An address may be prefixed with "@" to indicate indirect addressing, and/or
suffixed by an index register with parentheses "(indexreg)" to indicate
indexed addressing.  E.g.,

        ADD 17,@100(3)

Either operand may be a symbol or an expression.

Absent operands are treated as all-bits-zeros.

I/O instructions are similar, except the "accumulator" operand is replaced by
a "device" operand, and the binary layout is different.

There cannot be more than one instruction statement in a line.

References
==========
DECSYSTEM20 MACRO ASSEMBLER Reference Manual, AA-4159C-TM, Apr. 1978.
DECSYSTEM10 MACRO ASSEMBLER Reference Manual, AA-C780C-TB, Apr. 1978.
MACRO-6 Assembly Language Programming Manual, DEC-6-0-TP-MAC-LM-FP-ACT01, Feb. 1965.