A simple educational Register-based VM, Assembler, and Disassembler.

reg-vm.js

// "Fetch, decode, eval!"
// A simple Register-based VM, Assembler, and Disassembler.
// by Dmitry Soshnikov <[email protected]>

// This virtual machine (VM) consists of registers (data storage),
// and operations (instructions) which operate on the registers.

// --------------------------------------------------------------
// Registers.
// --------------------------------------------------------------

// Storage: actual regs contents, initially empty (0).
var regStore = {ax: 0, bx: 0, cx: 0, dx: 0};
// Array to map register name to a number (index of the array), to be encoded
// in the instructions: ax: 0, bx: 1, etc.
var regs = Object.keys(regStore);

// --------------------------------------------------------------
// Instructions. Known as "opcodes" (operation codes).
// --------------------------------------------------------------

var instructionSet = {
  halt: 0, // stops the program
  mov : 1, // moves contents to a register
  add : 2 // sums reg1 to reg2 and stores back in reg1
};

// --------------------------------------------------------------
// Assembler. Compiles mnemonics into a machine code.
// --------------------------------------------------------------

function assemble(program) {
  return program.map(function(command) {
    var
      parsed = command.replace(',', '').split(/\s+/);
      instruction = parsed[0],
      op1 = parsed[1],
      op2 = parsed[2];
    
    // Each instruction is encoded as: [opcode, op1, op2], which in memory
    // represented as HEX code 0x<opcode><op1><op2>, e.g. 0x1014 -- mov ax, 20.
    var machineInstruction = [
      instructionSet[instruction],
      // In our instructions, firt op is always a register.
      regs.indexOf(op1) > -1 ? regs.indexOf(op1) : 0,
      // Second can be a register or an immediate value (stored as HEX).
      regs.indexOf(op2) > -1 ? regs.indexOf(op2) :
        op2 ? Number(op2).toString(16) : 0,
    ];
    
    // Actual memory data, representing code in HEX.
    memoryDump.push('0x' + machineInstruction.join('').toUpperCase());
    
    return machineInstruction;
  });
}

// --------------------------------------------------------------
// Decoder. Decodes instructions from memory.
// --------------------------------------------------------------

function decode(instruction) {
  // In real machine, here goes decoding of encoded instruction number in
  // memory. Here we have a simple version: [instr, reg, (reg|hex-value)].
  return [
    instruction[0],
    instruction[1],
    parseInt(instruction[2], 16)
  ];
}

// --------------------------------------------------------------
// Evaluating module of CPU.
// --------------------------------------------------------------
function eval(decoded) {
  var
    instruction = decoded[0],
    op1 = decoded[1],
    op2 = decoded[2];
  switch (instruction) {
    case 0: // halt, stop execution.
      console.log('halt');
      isRunning = false;
      break;
    case 1: // mov
      regStore[regs[op1]] = op2;
      // Debug disassembler at running.
      console.log('mov ', regs[op1], ', ', op2);
      break;
    case 2: // add
      regStore[regs[op1]] = regStore[regs[op1]] + regStore[regs[op2]];
      console.log('add', regs[op1], ', ', regs[op2]);
      break;
  }
}

// --------------------------------------------------------------
// Fetcher.
// --------------------------------------------------------------

// Instructions pointer, points to the current command to be executed.
var ip = 0;

// Fetches the next command from the program (code segment in memory).
function fetch() {
  return program[ip++];
}

// --------------------------------------------------------------
// Testing.
// --------------------------------------------------------------

// Test assembly program.
var assemblyCode = [
  'mov ax, 10',
  'mov bx, 20',
  'add ax, bx',
  'halt'
];

// Generate machine code.
var memoryDump = [];
var program = assemble(assemblyCode);

// Check generated machine code.
// ['0x10A', '0x1114', '0x201', '0x000']
console.log('Code segment:', memoryDump);

// CPU at work.
var isRunning = true;
while (isRunning) {
  // Main execution loop: fetch, decode, eval.
  var instruction = fetch();
  var decoded = decode(instruction);
  eval(decoded);
}

// Check registers contents after program execution.
// {ax: 30, bx: 20, cx: 0, dx: 0}
console.log('Registers:', regStore);

In this case it's not about JS, it's about machine ;) JS here used as just a quick language for testing.