/** * Zilog Z80 emulator chip — TDD spec. * * The Z80 is binary-compatible with the 8080 plus extensions, so the * 8080 tests' structure carries over. This file focuses on: * 1. The Z80-specific bus protocol (M1̅ / MREQ̅ / IORQ̅ / RFSH̅) * 2. Z80-only instructions (EX, EXX, DJNZ, IX/IY, block ops, IM 0-2) * 3. NMI behaviour (pushes PC, vectors to 0x0066) * * The 8080-subset instructions are NOT re-tested here — once both chips * are implemented, a shared "8080-subset suite" should run against both. */ import { describe, it, expect, beforeEach, afterEach } from 'vitest'; import { BoardHarness } from '../src/BoardHarness.js'; import { chipWasmExists, hex8, hex16 } from '../src/helpers.js'; const CHIP = 'z80'; const skip = !chipWasmExists(CHIP); const CLOCK_HZ = 4_000_000; const CLOCK_NS = Math.round(1e9 / CLOCK_HZ); function fullPinMap() { const m = { M1: 'M1', MREQ: 'MREQ', IORQ: 'IORQ', RD: 'RD', WR: 'WR', RFSH: 'RFSH', HALT: 'HALT', WAIT: 'WAIT', INT: 'INT', NMI: 'NMI', RESET: 'RESET', BUSREQ: 'BUSREQ', BUSACK: 'BUSACK', CLK: 'CLK', VCC: 'VCC', GND: 'GND', }; for (let i = 0; i < 16; i++) m[`A${i}`] = `A${i}`; for (let i = 0; i < 8; i++) m[`D${i}`] = `D${i}`; return m; } async function bootZ80(program) { const board = new BoardHarness(); await board.addChip(CHIP, fullPinMap()); board.installFakeRom(program, { addrPrefix: 'A', addrWidth: 16, dataPrefix: 'D', dataWidth: 8, rd: 'RD', rdActiveLow: true, cs: 'MREQ', // only respond when MREQ̅ is asserted csActiveLow: true, baseAddr: 0, }); const ram = board.installFakeRam(0x8000, { addrPrefix: 'A', addrWidth: 16, dataPrefix: 'D', dataWidth: 8, rd: 'RD', wr: 'WR', cs: 'MREQ', baseAddr: 0x8000, }); board.setNet('WAIT', true); // not waiting board.setNet('INT', true); // INT̅ deasserted (active-low on Z80) board.setNet('NMI', true); // NMI̅ deasserted board.setNet('BUSREQ', true); board.setNet('RESET', false); board.advanceNanos(CLOCK_NS * 4); board.setNet('RESET', true); // Do NOT advance after RESET deassert — the caller has its own // advanceNanos loop, and may want to poke RAM contents first // (same lesson as bootCpu in the 8080 tests). return { board, ram }; } describe('Zilog Z80 chip', () => { describe('pin contract', () => { it.skipIf(skip)('registers all 40 named pins', async () => { const board = new BoardHarness(); await expect(board.addChip(CHIP, fullPinMap())).resolves.toBeDefined(); board.dispose(); }); }); describe('reset', () => { it.skipIf(skip)('first M1 fetch is from 0x0000', async () => { const board = new BoardHarness(); await board.addChip(CHIP, fullPinMap()); const m1Fetches = []; board.watchNet('M1', (low) => { if (low === false) m1Fetches.push(board.readBus('A', 16)); }); board.installFakeRom([0x00, 0x00, 0x76], { // NOP NOP HALT rd: 'RD', cs: 'MREQ', csActiveLow: true, }); board.setNet('WAIT', true); board.setNet('INT', true); board.setNet('NMI', true); board.setNet('BUSREQ', true); board.setNet('RESET', false); board.advanceNanos(CLOCK_NS * 4); board.setNet('RESET', true); board.advanceNanos(CLOCK_NS * 30); expect(m1Fetches[0], 'first M1 fetch').toBe(0x0000); board.dispose(); }); }); describe('M1 cycle', () => { it.skipIf(skip)('asserts M1̅ + MREQ̅ + RD̅ during opcode fetch', async () => { const board = new BoardHarness(); await board.addChip(CHIP, fullPinMap()); let sawAllAsserted = false; board.watchNet('M1', (state) => { if (state === false) { // Snap the other signals at the same instant if (board.getNet('MREQ') === false && board.getNet('RD') === false) { sawAllAsserted = true; } } }); board.installFakeRom([0x00, 0x76], { rd: 'RD', cs: 'MREQ', csActiveLow: true }); board.setNet('WAIT', true); board.setNet('INT', true); board.setNet('NMI', true); board.setNet('BUSREQ', true); board.setNet('RESET', false); board.advanceNanos(CLOCK_NS * 4); board.setNet('RESET', true); board.advanceNanos(CLOCK_NS * 30); expect(sawAllAsserted, 'M1̅, MREQ̅, RD̅ asserted simultaneously during fetch').toBe(true); board.dispose(); }); it.skipIf(skip)('asserts RFSH̅ during the refresh phase of M1', async () => { const board = new BoardHarness(); await board.addChip(CHIP, fullPinMap()); let rfshSeen = false; board.watchNet('RFSH', (state) => { if (state === false) rfshSeen = true; }); board.installFakeRom([0x00, 0x00, 0x76], { rd: 'RD', cs: 'MREQ', csActiveLow: true }); board.setNet('WAIT', true); board.setNet('INT', true); board.setNet('NMI', true); board.setNet('BUSREQ', true); board.setNet('RESET', false); board.advanceNanos(CLOCK_NS * 4); board.setNet('RESET', true); board.advanceNanos(CLOCK_NS * 30); expect(rfshSeen, 'RFSH̅ must pulse low after M1 fetch').toBe(true); board.dispose(); }); }); describe('Z80-only instructions', () => { // Z80 mnemonic constants — only those used in tests below. const LD_A_n = 0x3E; const LD_BC_nn = 0x01; const LD_DE_nn = 0x11; const LD_HL_nn = 0x21; const LD_IX_nn = 0xDD; const _IX_LD_nn = 0x21; // DD 21 nn nn const EX_DE_HL = 0xEB; const EXX = 0xD9; const DJNZ = 0x10; const LDIR = 0xED; const _LDIR = 0xB0; // ED B0 const LD_aHL_n = 0x36; const LD_addr_A = 0x32; const HALT = 0x76; it.skipIf(skip)('EX DE, HL swaps register pairs', async () => { // LD HL, 0x1234 ; LD DE, 0x5678 ; EX DE, HL ; LD (0x8000), A is awkward // because we can't read HL/DE directly. Use this instead: // LD HL, 0xAA00 ; LD DE, 0xBB00 ; EX DE, HL ; LD (HL), 0x77 ; HALT // After EX, HL = 0xBB00 (in our RAM range) so we write to 0xBB00. // Wait, 0xBB00 is in our RAM (0x8000+) — yes. const program = new Uint8Array([ LD_HL_nn, 0x00, 0xAA, LD_DE_nn, 0x00, 0xBB, EX_DE_HL, LD_aHL_n, 0x77, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0xBB00)).toBe(0x77); board.dispose(); }); it.skipIf(skip)('DJNZ decrements B and jumps while non-zero', async () => { // LD A, 0 ; LD B, 5 ; LOOP: INC A ; DJNZ LOOP ; LD (0x8000), A ; HALT // Expected: A = 5 stored at 0x8000. const INC_A = 0x3C; const program = new Uint8Array([ LD_A_n, 0x00, 0x06, 0x05, // LD B, 5 INC_A, // LOOP: DJNZ, 0xFD, // jump back -3 to LOOP LD_addr_A, 0x00, 0x80, // LD (0x8000), A HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 500; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x8000)).toBe(5); board.dispose(); }); it.skipIf(skip)('LDIR copies a memory block from HL to DE', async () => { // Pre-load source: 4 bytes at 0xC000..0xC003. Then LDIR HL=0xC000, // DE=0x9000, BC=4. After: 4 bytes copied to 0x9000..0x9003. const program = new Uint8Array([ LD_HL_nn, 0x00, 0xC0, // LD HL, 0xC000 LD_DE_nn, 0x00, 0x90, // LD DE, 0x9000 LD_BC_nn, 0x04, 0x00, // LD BC, 0x0004 LDIR, _LDIR, // ED B0 HALT, ]); const { board, ram } = await bootZ80(program); ram.poke(0xC000, 0x11); ram.poke(0xC001, 0x22); ram.poke(0xC002, 0x33); ram.poke(0xC003, 0x44); for (let i = 0; i < 500; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x11); expect(ram.peek(0x9001)).toBe(0x22); expect(ram.peek(0x9002)).toBe(0x33); expect(ram.peek(0x9003)).toBe(0x44); board.dispose(); }); it.skipIf(skip)('LD A, (IX+d) reads via IX with signed displacement', async () => { // Pre-load 0xCD at 0xA005. Set IX = 0xA000. LD A, (IX+5) → A=0xCD. // Then LD (0x9000), A so we can verify. const program = new Uint8Array([ LD_IX_nn, _IX_LD_nn, 0x00, 0xA0, // DD 21 00 A0 — LD IX, 0xA000 0xDD, 0x7E, 0x05, // DD 7E 05 — LD A, (IX+5) LD_addr_A, 0x00, 0x90, // LD (0x9000), A HALT, ]); const { board, ram } = await bootZ80(program); ram.poke(0xA005, 0xCD); for (let i = 0; i < 400; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0xCD); board.dispose(); }); it.skipIf(skip)('EXX swaps the main register set with the shadow set', async () => { // LD HL, 0x1111 // EXX ; swap → HL = shadow (0x0000 after reset shadow init) // LD HL, 0x9000 ; main HL now 0x9000 (was the shadow) // EXX ; swap back → original HL = 0x1111 in main set // LD (HL), 0x77 ; writes to 0x1111... wait, main HL is 0x1111 // ; that's not in our RAM range (0x8000+). // Restructure: use two HL values both in RAM range. // LD HL, 0x9100 ; EXX ; LD HL, 0x9200 ; EXX ; LD (HL), 0x77 ; HALT // After: write to 0x9100 (the original main HL). const program = new Uint8Array([ LD_HL_nn, 0x00, 0x91, // LD HL, 0x9100 (main) EXX, // → main set goes to shadow LD_HL_nn, 0x00, 0x92, // LD HL, 0x9200 (this is now the new "main") EXX, // → swap back; main HL = 0x9100 LD_aHL_n, 0x77, // LD (HL), 0x77 → write 0x77 to 0x9100 HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 300; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9100)).toBe(0x77); // Verify the OTHER write didn't happen (shadow set's HL=0x9200 // was never written via LD (HL), 0x77 in the shadow context). expect(ram.peek(0x9200)).toBe(0x00); board.dispose(); }); }); describe('interrupts', () => { it.skipIf(skip)('NMI̅ falling edge pushes PC and vectors to 0x0066', async () => { // EI ; loop: NOP ; JR -1 // ISR at 0x0066: LD A, 0xAB ; LD (0x9000), A ; HALT const program = new Uint8Array(0x80); program.fill(0x00); program[0x00] = 0xFB; // EI program[0x01] = 0x00; // NOP program[0x02] = 0x18; program[0x03] = 0xFD; // JR -3 → loop program[0x66] = 0x3E; program[0x67] = 0xAB; // LD A, 0xAB program[0x68] = 0x32; program[0x69] = 0x00; program[0x6A] = 0x90; // LD (0x9000), A program[0x6B] = 0x76; // HALT const { board, ram } = await bootZ80(program); // Run a few cycles to enter the loop. for (let i = 0; i < 50; i++) board.advanceNanos(CLOCK_NS); // Pulse NMI̅ low (active low) → falling edge triggers interrupt. board.setNet('NMI', false); board.advanceNanos(CLOCK_NS * 4); board.setNet('NMI', true); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0xAB); board.dispose(); }); it.skipIf(skip)('IM 1 + INT̅ vectors to 0x0038', async () => { // EI ; IM 1 ; loop: NOP ; JR -1 // ISR at 0x0038: LD A, 0x39 ; LD (0x9000), A ; HALT const program = new Uint8Array(0x80); program.fill(0x00); program[0x00] = 0xFB; // EI program[0x01] = 0xED; program[0x02] = 0x56; // IM 1 program[0x03] = 0x00; // NOP loop program[0x04] = 0x18; program[0x05] = 0xFD; // JR -3 program[0x38] = 0x3E; program[0x39] = 0x39; // LD A, 0x39 program[0x3A] = 0x32; program[0x3B] = 0x00; program[0x3C] = 0x90; program[0x3D] = 0x76; // HALT const { board, ram } = await bootZ80(program); for (let i = 0; i < 50; i++) board.advanceNanos(CLOCK_NS); // INT̅ active-low: drive low to request interrupt. board.setNet('INT', false); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); board.setNet('INT', true); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x39); board.dispose(); }); it.skipIf(skip)('IM 2 + INT̅ uses I:byte to vector through a table', async () => { // Set up: // I = 0x40, vector byte = 0x00 (our chip approximates the bus // data byte as 0x00 since we don't model an INTA cycle), so // vector table address = 0x4000. Place ISR pointer (0x6000) // there. ISR writes 0xC2 to 0x9000 and HALTs. const program = new Uint8Array(0x8000); program.fill(0x00); program[0x00] = 0x3E; program[0x01] = 0x40; // LD A, 0x40 program[0x02] = 0xED; program[0x03] = 0x47; // LD I, A program[0x04] = 0xED; program[0x05] = 0x5E; // IM 2 program[0x06] = 0xFB; // EI program[0x07] = 0x00; // NOP (loop) program[0x08] = 0x18; program[0x09] = 0xFD; // JR -3 → 0x07 // Vector table at I:00 = 0x4000 → ISR @ 0x6000 program[0x4000] = 0x00; program[0x4001] = 0x60; // ISR at 0x6000: LD A, 0xC2 ; LD (0x9000), A ; HALT program[0x6000] = 0x3E; program[0x6001] = 0xC2; program[0x6002] = 0x32; program[0x6003] = 0x00; program[0x6004] = 0x90; program[0x6005] = 0x76; const { board, ram } = await bootZ80(program); // Let LD A,I + LD I,A + IM 2 + EI execute, then enter the loop. for (let i = 0; i < 80; i++) board.advanceNanos(CLOCK_NS); // Pulse INT̅ low. board.setNet('INT', false); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); board.setNet('INT', true); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000), 'ISR sentinel must reach RAM via IM 2 vectoring').toBe(0xC2); board.dispose(); }); }); describe('CB-prefix bit ops', () => { const HALT = 0x76; const LD_addr_A = 0x32; const CB = 0xCB; const LD_HL_nn = 0x21; const LD_BC_nn = 0x01; it.skipIf(skip)('SET n, A turns on the right bit', async () => { // LD A, 0x00 ; SET 7, A ; LD (0x9000), A ; HALT // Expected: A = 0x80, stored at 0x9000. const program = new Uint8Array([ 0x3E, 0x00, // LD A, 0x00 CB, 0xFF, // SET 7, A (op = 11_111_111 = 0xFF) LD_addr_A, 0x00, 0x90, // LD (0x9000), A HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x80); board.dispose(); }); it.skipIf(skip)('RES n, A turns off the right bit', async () => { // LD A, 0xFF ; RES 0, A ; LD (0x9000), A ; HALT // Expected: A = 0xFE. const program = new Uint8Array([ 0x3E, 0xFF, CB, 0x87, // RES 0, A (op = 10_000_111 = 0x87) LD_addr_A, 0x00, 0x90, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0xFE); board.dispose(); }); it.skipIf(skip)('RLC A rotates left circular', async () => { // LD A, 0x81 ; RLC A ; LD (0x9000), A ; HALT // 0x81 = 1000_0001 → rotate left circular → 0000_0011 = 0x03 (bit 7 // wrapped to bit 0). const program = new Uint8Array([ 0x3E, 0x81, CB, 0x07, // RLC A LD_addr_A, 0x00, 0x90, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x03); board.dispose(); }); it.skipIf(skip)('SRL A shifts right logical with zero into MSB', async () => { // LD A, 0x81 ; SRL A ; LD (0x9000), A ; HALT // 0x81 → 0x40 (low bit 1 falls into CF; MSB filled with 0) const program = new Uint8Array([ 0x3E, 0x81, CB, 0x3F, // SRL A LD_addr_A, 0x00, 0x90, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x40); board.dispose(); }); it.skipIf(skip)('SRA A shifts right arithmetic, sign-extending', async () => { // LD A, 0x80 ; SRA A ; LD (0x9000), A ; HALT // 0x80 → 0xC0 (sign bit propagates) const program = new Uint8Array([ 0x3E, 0x80, CB, 0x2F, // SRA A LD_addr_A, 0x00, 0x90, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0xC0); board.dispose(); }); it.skipIf(skip)('DAA after BCD ADD adjusts the result', async () => { // LD A, 0x09 ; LD B, 0x07 ; ADD A, B ; DAA ; LD (0x9000), A ; HALT // 9 + 7 = 16 (BCD): raw 0x10 + DAA correction 0x06 = 0x16. const program = new Uint8Array([ 0x3E, 0x09, // LD A, 0x09 0x06, 0x07, // LD B, 0x07 0x80, // ADD A, B 0x27, // DAA LD_addr_A, 0x00, 0x90, HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 200; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x16); board.dispose(); }); it.skipIf(skip)('ADC HL, BC adds register pair with carry', async () => { // LD HL, 0x1000 ; LD BC, 0x2000 ; OR A,A (clear CF) ; ADC HL,BC ; // LD A, H ; LD (0x9000), A ; LD A, L ; LD (0x9001), A ; HALT // After ADC HL=0x3000. Store H and L separately. const program = new Uint8Array([ LD_HL_nn, 0x00, 0x10, // LD HL, 0x1000 LD_BC_nn, 0x00, 0x20, // LD BC, 0x2000 0xB7, // OR A — clears CF (and other flags except SZP) 0xED, 0x4A, // ADC HL, BC 0x7C, // LD A, H LD_addr_A, 0x00, 0x90, // LD (0x9000), A 0x7D, // LD A, L LD_addr_A, 0x01, 0x90, // LD (0x9001), A HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 400; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x30); // H expect(ram.peek(0x9001)).toBe(0x00); // L board.dispose(); }); it.skipIf(skip)('RLD rotates a low nibble between A and (HL)', async () => { // LD A, 0x12 ; LD HL, 0xC000 ; (ram[0xC000] poked to 0x34) ; RLD ; // LD (0x9000), A ; HALT // Before: A = 0x12, mem = 0x34 // RLD: A_low (0x2) → mem_low; mem_high (0x3) → A_low; mem_low (0x4) → mem_high. // After: A = 0x13, mem = 0x42. const program = new Uint8Array([ 0x3E, 0x12, // LD A, 0x12 LD_HL_nn, 0x00, 0xC0, // LD HL, 0xC000 0xED, 0x6F, // RLD LD_addr_A, 0x00, 0x90, // LD (0x9000), A HALT, ]); const { board, ram } = await bootZ80(program); ram.poke(0xC000, 0x34); for (let i = 0; i < 300; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0x13); // A expect(ram.peek(0xC000)).toBe(0x42); // mem board.dispose(); }); it.skipIf(skip)('CPIR scans memory for accumulator match', async () => { // Pre-poke 0x9100=0x11, 0x9101=0x22, 0x9102=0x33, 0x9103=0x44. // LD A, 0x33 ; LD HL, 0x9100 ; LD BC, 0x0004 ; CPIR ; // After CPIR: HL stops one past 0x9102 (the match position). HL=0x9103. // Store H and L to verify HL. const program = new Uint8Array([ 0x3E, 0x33, // LD A, 0x33 LD_HL_nn, 0x00, 0x91, // LD HL, 0x9100 LD_BC_nn, 0x04, 0x00, // LD BC, 0x0004 0xED, 0xB1, // CPIR 0x7C, // LD A, H LD_addr_A, 0x00, 0x80, // LD (0x8000), A 0x7D, // LD A, L LD_addr_A, 0x01, 0x80, // LD (0x8001), A HALT, ]); const { board, ram } = await bootZ80(program); ram.poke(0x9100, 0x11); ram.poke(0x9101, 0x22); ram.poke(0x9102, 0x33); ram.poke(0x9103, 0x44); for (let i = 0; i < 600; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x8000)).toBe(0x91); // H expect(ram.peek(0x8001)).toBe(0x03); // L = 0x03 (one past match) board.dispose(); }); it.skipIf(skip)('BIT 7, A sets ZF when bit clear, clears when bit set', async () => { // LD A, 0x00 ; BIT 7, A ; JR Z, +taken ; LD A, 0xFF (should NOT run) // taken: LD A, 0xAA ; LD (0x9000), A ; HALT const program = new Uint8Array([ 0x3E, 0x00, // LD A, 0x00 CB, 0x7F, // BIT 7, A — ZF=1 0x28, 0x02, // JR Z, +2 (skip the next 2 bytes) 0x3E, 0xFF, // (skipped) LD A, 0xFF 0x3E, 0xAA, // taken: LD A, 0xAA LD_addr_A, 0x00, 0x90, // LD (0x9000), A HALT, ]); const { board, ram } = await bootZ80(program); for (let i = 0; i < 300; i++) board.advanceNanos(CLOCK_NS); expect(ram.peek(0x9000)).toBe(0xAA); board.dispose(); }); }); /* ZEXDOC end-to-end integration run lives in its own file (`zexdoc.test.js`) — it needs a much longer time budget than the unit suite. */ });