z80asm

The list is passed to "z80lexer", that in turn calls "z80preprocessor" to handle file includes, and then splits the input into tokens.

The stream of tokens is passed on to <CPU:Z80::Assembler::Parser> that parses the input and generates the object image in CPU::Z80::Assembler::Program. Assembly macro expansion is handled at this stage by CPU::Z80::Assembler::Macro.

The assembly program is composed by a list of CPU::Z80::Assembler::Segment, each representing one named section of code. Each segment is composed by a list of CPU::Z80::Assembler::Opcode, each representing one assembly instruction.

The output object code is returned as a string.

If the $CPU::Z80::Assembler::verbose variable is set, an output listing is generated by CPU::Z80::Assembler::List on standard output.

Assembly is done in five steps:

1.

input is preprocessed, scanned and split into tokens

2.

tokens are parsed and converted to lists of opcodes

3.

addresses for each opcode are allocated

4.

relative jumps are checked for out-of-range jumps and replaced by absolute jumps if needed

5.

object code is generated for each opcode, computing all expressions used; the expressions are represented by CPU::Z80::Assembler::Expr.

z80asm_file

z80preprocessor

The list is passed to the Asm::Preproc that takes care of file includes and handles the %line and #line lines generated by external preprocessors like cpp or nasm.

The result is a Iterator::Simple::Lookahead of Asm::Preproc::Line objects that contain each of the input lines of the input.

z80lexer

It calls "z80preprocessor" to split the input into a Iterator::Simple::Lookahead of Asm::Preproc::Line objects representing the source lines of the Z80 assembly language program.

It returns a stream of Asm::Preproc::Token objects for each assembly token in the input.

Each token contains a type string, a value and a Asm::Preproc::Line object pointing at the input line where the token was found.

z80masm

  z80masm sourcefile [destfile]

The z80masm program, installed as part of this module, calls the z80asm_file() function to assemble an input source file, generates an output binary file, and produce an assembly listing on standard output.

Input line format

    ; comment beginning with ;
    # comment beginning with # as first char on a line
    [LABEL [:]] INSTRUCTION [: INSTRUCTION ...] [; optional comments]
    LABEL [:]
    LABEL = EXPRESSION [; ...]

Preprocessing

Tokens

reserved words

  Asm::Preproc::Token('word', 'word', $line)

All the reserved words and symbols are returned in lower case letters.

strings

  Asm::Preproc::Token(STRING => $string, $line)

Single- or double-quoted strings are accepted. The quote character cannot be used inside the string. The returned string has the quotes stripped.

identifiers

  Asm::Preproc::Token(NAME => $name, $line)

The program identifiers must start with a letter or underscore, and consist solely of letters, underscores and numbers. There is a special case "$" identifier that represents the current location counter.

Identifiers are returned with case preserved, i.e. the assembler is case-sensitive for labels and case-insensitive for assembly reserved words.

numbers

  Asm::Preproc::Token(NUMBER => $decimal_number, $line)

Numbers are converted to decimal base from one of the following formats:

• Decimal numbers are sequences of the digits 0..9, e.g. 159.
• Hexadecimal number are sequences of the digits 0..9 and letters A..F, prefixed by one of "0x", "$" or "#"; or prefixed by a zero digit, if the number starts by a letter, and suffixed by a "H" character, e.g. 0xFA21, 0FA21H, 10H, $FA21, #FA21.
• Binary numbers are sequences of the digits 0..1, prefixed by one of "0b" or "%"; or suffixed by "B", e.g. 0b10, 10B, %10.

Z80 assembly

relative jumps

The DJNZ and JR instructions take an address as their destination, not an offset. If you need to use an offset, do sums on $. Note that $ is the address of the *current* instruction. The offset needs to be calculated from the address of the *next* instruction, which for these instructions is always $ + 2.

A relative jump instruction can always be used. The assembler automatically replaces it with an absolute jump if the distance is too far, or if the given flag is not available, e.g. "jr po,NN". A "djnz NN" instruction is converted to "dec b:jp nz,NN" if the distance is too far.

stop

This extra instruction (which assembles to 0xDD 0xDD 0x00) is provided for the convenience of those using the CPU::Emulator::Z80 module.

Pseudo-instructions

Accepts a list of expressions, and evaluates each as a byte to load to the object file.

defw

Accepts a list of expressions, and evaluates each as a 16-bit word to load to the object file, in little-endian order.

defm, deft

Accepts a list of literal strings, either single- or double-quoted. The quoted text can not include the quotes surrounding it or newlines. The characters are loaded to the object file.

defmz

Same as "defm", but appends a zero byte as string terminator after each string.

defm7

Same as "defm", but "inverts" (i.e. bit 7 set) the last character of the string, as string terminator.

equ, =

Labels are created having the value of the address they are created at.

Alternatively labels may be assigned expressions by using "equ" or "=". The expressions use the Perl operators and can refer to other labels by name, even if they are defined further on the file. The "$" can be used in the expression to represent the current location counter.

  label      = $ + 8
  otherlabel = label / 2 + 3

org

Tell the assembler to start building the code at this address. If it is not the first instruction of the assembly, the gap to the previous location counter is filled with $CPU::Z80::Assembler::fill_byte. If absent, defaults to 0x0000.

include

Recursively include another file at the current source file.

Macros