#!/usr/bin/env python3 """ GBA Function Matcher - Matches C code against original ROM bytes. Uses agbcc (GCC 2.95) to compile C and compares byte-by-byte. Key insight: agbcc's code generation depends heavily on C types. - Struct member access (p->field) generates different code than array access (p[i]) - Pointer type matters (int* vs struct S* vs volatile int*) - Function must have observable side effects to prevent dead code elimination """ import os import re import struct import subprocess import sys import tempfile ROM_PATH = "bit Generations - Orbital (Japan) (En).gba" ROM_BASE = 0x08000000 AGBCC = "agbcc/agbcc" GCC = "arm-gnu-toolchain-13.3.rel1-darwin-arm64-arm-none-eabi/bin/arm-none-eabi-gcc" OBJCOPY = "arm-gnu-toolchain-13.3.rel1-darwin-arm64-arm-none-eabi/bin/arm-none-eabi-objcopy" def read_rom(): with open(ROM_PATH, "rb") as f: return f.read() def try_compile_and_match(c_code, addr, size): """Try to compile C code and match against ROM bytes""" rom = read_rom() offset = addr - ROM_BASE orig = rom[offset:offset+size] with tempfile.NamedTemporaryFile(suffix=".c", mode="w", delete=False) as f: f.write(c_code) c_path = f.name s_path = c_path.replace(".c", ".s") o_path = c_path.replace(".c", ".o") bin_path = c_path.replace(".c", ".bin") try: # Compile with agbcc result = subprocess.run([AGBCC, "-O2", c_path], capture_output=True, text=True, timeout=30) if not os.path.exists(s_path): return None, None, "COMPILE_FAIL" # Assemble with gcc (supports Thumb16 properly) result = subprocess.run([GCC, "-mthumb", "-mcpu=arm7tdmi", "-c", s_path, "-o", o_path], capture_output=True, text=True, timeout=30) if result.returncode != 0 or not os.path.exists(o_path): return None, None, "ASM_FAIL" # Extract binary result = subprocess.run([OBJCOPY, "-O", "binary", "-j", ".text", o_path, bin_path], capture_output=True, text=True, timeout=30) if not os.path.exists(bin_path): return None, None, "OBJCOPY_FAIL" with open(bin_path, "rb") as f: binary = f.read() # Trim to exact size if len(binary) > size: binary = binary[:size] matching = sum(1 for a, b in zip(binary, orig) if a == b) rate = matching / size if size > 0 else 0 return binary, orig, rate except Exception as e: return None, None, f"ERROR: {e}" finally: for p in [c_path, s_path, o_path, bin_path]: try: if os.path.exists(p): os.unlink(p) except: pass def get_function_asm(addr, size): """Get the agbcc assembly for a function""" rom = read_rom() offset = addr - ROM_BASE orig = rom[offset:offset+size] # Simple Thumb16 disassembly lines = [] i = 0 while i < size - 1: hw = struct.unpack_from('> 8) & 7 imm = hw & 0xFF lines.append(f" movs r{rd}, #{imm}") elif (hw & 0xF800) == 0x3000: rd = (hw >> 8) & 7 imm = hw & 0xFF lines.append(f" adds r{rd}, #{imm}") elif (hw & 0xF800) == 0x3800: rd = (hw >> 8) & 7 imm = hw & 0xFF lines.append(f" subs r{rd}, #{imm}") elif (hw & 0xF800) == 0x2800: rd = (hw >> 8) & 7 imm = hw & 0xFF lines.append(f" cmp r{rd}, #{imm}") elif (hw & 0xF800) == 0x6000: rd = hw & 7 rn = (hw >> 3) & 7 imm = ((hw >> 6) & 0x1F) * 4 lines.append(f" str r{rd}, [r{rn}, #{imm}]") elif (hw & 0xF800) == 0x6800: rd = hw & 7 rn = (hw >> 3) & 7 imm = ((hw >> 6) & 0x1F) * 4 lines.append(f" ldr r{rd}, [r{rn}, #{imm}]") elif (hw & 0xFFC0) == 0x4400: rd = ((hw >> 4) & 8) | (hw & 7) rs = (hw >> 3) & 0xF lines.append(f" mov r{rd}, r{rs}") elif (hw & 0xFFC0) == 0x1800: rd = hw & 7 rn = (hw >> 3) & 7 rs = (hw >> 6) & 7 lines.append(f" adds r{rd}, r{rn}, r{rs}") elif (hw & 0xFFC0) == 0x1A00: rd = hw & 7 rn = (hw >> 3) & 7 rs = (hw >> 6) & 7 lines.append(f" subs r{rd}, r{rn}, r{rs}") elif (hw & 0xFFC0) == 0x4280: rd = hw & 7 rs = (hw >> 3) & 7 lines.append(f" cmp r{rd}, r{rs}") elif (hw & 0xFFC0) == 0x4300: rd = hw & 7 rs = (hw >> 3) & 7 lines.append(f" orrs r{rd}, r{rs}") elif (hw & 0xFFC0) == 0x4040: rd = hw & 7 rs = (hw >> 3) & 7 lines.append(f" eors r{rd}, r{rs}") elif (hw & 0xFFC0) == 0x4080: rd = hw & 7 rs = (hw >> 3) & 7 lines.append(f" lsls r{rd}, r{rs}") elif (hw & 0xFFC0) == 0x40C0: rd = hw & 7 rs = (hw >> 3) & 7 lines.append(f" lsrs r{rd}, r{rs}") elif (hw & 0xFF00) == 0xD000: cond = (hw >> 8) & 0xF offset_val = hw & 0xFF if offset_val & 0x80: offset_val -= 256 target = addr_here + 4 + offset_val * 2 conds = ["eq","ne","cs","cc","mi","pl","vs","vc","hi","ls","ge","lt","gt","le","al"] c = conds[cond] if cond < len(conds) else f"x{cond}" lines.append(f" b{c} 0x{target:08X}") elif (hw & 0xF800) == 0xE000: offset_val = hw & 0x7FF if offset_val & 0x400: offset_val -= 2048 target = addr_here + 4 + offset_val * 2 lines.append(f" b 0x{target:08X}") else: lines.append(f" .hword 0x{hw:04X}") i += 2 if size % 2 == 1: lines.append(f" .byte 0x{orig[size-1]:02X}") return '\n'.join(lines)