An assembler converts symbolic opcodes and labels into executable machine code.
Symbolic opcodes such as LOAD or ADD cannot be executed directly, the need to be converted to machine executable bytecode. These get converted by using a lookup table.
Labels need to be converted to memory addresses. Because symbolic assembly code may have a non-specific location, using labels makes it easier to produce output from compiler and by hand.
The PL/0 assembler is very simple. It supports all the pl1 symbolic opcodes, and direct labels.
The following program is a program that generates the fibonacci sequence.
# Firstly we JMP to main, because the execution starts at offset 0
JMP main
# This contains f[n]
m:
1
# This contains f[n+1]
n:
1
# The number of iterations to calculate
count:
20
# The main function which is a loop to generate the next Fibonacci number.
main:
LOAD m
PRINT
LOAD n
ADD
LOAD n
SAVE m
SAVE n
# Decrement the counter
PUSH 1
LOAD count
SUB
DUP
SAVE count
JLT main
This code can be compiled to machine executable code:
-- Assemble the above fibonacci program
$ ./pl0_assembler.py < examples/fibonacci.pl1a
[16, 5, 1, 1, 20, 6, 2, 50, 6, 3, 45, 6, 3, 7, 2, 7, 3, 10, 1, 6, 4, 46, 12, 7, 4, 20, 5]
-- Feed the machine code to the virtual machine
$ ./pl0_assembler.py < examples/fibonacci.pl1a | ./pl0_machine.py
1
1
2
3
5
8
13
21
34
55
89
144
233
377
610
987
1597
2584
4181
6765
-- Machine State --
Sequence: [16, 5, 10946, 17711, 0, 6, 2, 50, 6, 3, 45, 6, 3, 7, 2, 7, 3, 10, 1, 6, 4, 46, 12, 7, 4, 20, 5]
Stack: []
Offset: 27