Added 10 more opcodes into the CPU interpreter

This commit is contained in:
2026-08-08 14:42:38 -05:00
parent faef20555b
commit a1ff5f280f
3 changed files with 140 additions and 9 deletions
+47 -4
View File
@@ -10,18 +10,61 @@
- read_prg/read_chr/write_chr with debug_assert guards
- Split from_file (thin I/O wrapper) / parse_data (pure parser) for testability
- 8 unit tests, all passing (header, PRG mapping, error cases, CHR-RAM roundtrip)
- Bus module done (src/bus.rs):
- Bus trait (read/write) so the CPU is decoupled from the concrete memory layout
- NesBus: 2KB mirrored CPU RAM, cartridge PRG routing, PPU/APU/SRAM/expansion regions stubbed (return 0)
- CPU skeleton done (src/cpu.rs):
- Flag bit constants, registers (A/X/Y/SP/PC/P), set/clear/flag helpers, set_zp_flags
- Memory helpers: fetch, fetch16, read, write, read16, push, pop
- All 10 addressing mode helpers, including the JMP ($xxxx) page-wrap bug and page-cross cycle penalties
- First 3 opcodes implemented (JMP abs, LDX imm, STX zp)
- step dispatch shell + trace in nestest format
- set_test_start for the nestest harness
- nestest harness working (main.rs): loads nestest.nes, runs from $C000 with CYC:7, dumps trace to ours.log
- First 5 trace lines match the nestest reference log exactly (PC, regs, P, SP, CYC), then it panics on the next unimplemented opcode (JSR 0x20), which names the next target
## Next
- CPU (6502 core) - biggest remaining piece for this milestone
- Then bus to connect CPU to RAM/cartridge
- Verify CPU with nestest log diff + blargg suite before any graphics
- GUI (egui + winit) and PPU come after the CPU milestone
Finish the official CPU. Nestest analysis findings: the full 8991-line log tests illegal opcodes too (225 distinct opcodes total; first illegal is 0x04 at line 5004). So the official-only CPU is verified against lines 1-5003, then panics on 0x04 by design. Official opcodes never seen in the log: BRK 0x00, CLI 0x58.
CPU implementation batches (by family, ordered by nestest log frequency):
- Batch 1 - Subroutines + stack: JSR 0x20 (push ret-1 high then low), RTS 0x60 (pop low then high), PHA 0x48, PLA 0x68, PHP 0x08, PLP 0x28 (B-flag on push rule), RTI 0x40, BRK 0x00
- Batch 2 - Branches (all 8): BEQ F0, BNE D0, BCC 90, BCS B0, BPL 10, BMI 30, BVC 50, BVS 70. One shared method per flag. +1 cycle if taken, +1 more if target crosses page
- Batch 3 - Loads: LDA 8 modes, LDX 5 modes, LDY 5 modes (share load + set_zp_flags)
- Batch 4 - Stores: STA 7 modes, STX 3, STY 3 (share store core, no flag changes)
- Batch 5 - Compares: CMP 8 modes, CPX 3, CPY 3 (set C/Z/N, no store)
- Batch 6 - Arithmetic: ADC 8 modes, SBC 8 modes (N/V/C/Z, decimal flag inert), BIT 2 modes
- Batch 7 - Logic: AND, ORA, EOR (8 modes each)
- Batch 8 - Inc/Dec: INC/DEC 4 modes each (RMW), INX, INY, DEX, DEY
- Batch 9 - Shifts: ASL, LSR, ROL, ROR (accumulator + 4 memory modes each, RMW)
- Batch 10 - Flags + misc: CLC 18, SEC 38, CLI 58, SEI 78, CLD D8, SED F8, CLV B8, NOP EA, JMP abs 4C, JMP ind 6C, transfers TAX/TAY/TXA/TYA/TSX/TXS
Rules for each batch:
- Add the opcode's length to opcode_len in the same edit (never let them drift)
- Re-diff after each batch (cargo run > ours.log + awk column compare)
- Check base cycles per mode, feed page-cross extra into the cycle total
Verification for this phase:
- All 151 official opcodes implemented
- Trace matches lines 1-5003 field-for-field (PC/A/X/Y/P/SP/CYC)
- cargo build clean
- Commit + update STATE.md
After official CPU is done (follow-ups):
- Add MMC1 (mapper 1) to cartridge so blargg official_only.nes / all_instrs.nes can run (authoritative official check)
- Run blargg mapper-0 suite (coredump, cpu_timing_test, ram_after_reset, registers)
- Then the illegal-opcode phase for the full 8991-line nestest pass
- Then bus/PPU/timing, GUI (egui + winit) come after the CPU milestone
## Decisions
- No external dependencies yet, keeping it pure std until the GUI milestone
- GUI choice: egui + winit (decided, not yet used)
- Scope: cartridge + CPU first, verified headless before graphics
- Bus is a trait (impl Bus) so the CPU works against any memory layout (real bus, test bus, debug bus)
- Flags stored as a raw u8 with named bit-mask constants (FLAG_CARRY etc.)
- opcode length is a growing match in opcode_len() (returns u8); migrate to a full 256-entry table when it gets big
- trace diff compares PC + A/X/Y/P/SP/CYC, ignoring disassembly/PPU columns
- Mapper 0 first, other mappers (MMC1, CNROM) implemented after CPU works
- Test ROMs are homebrew/open, no ethics issue; game backups stay in gitignored roms/backup