RE Tools

RE Tools

Navigation: Home · Tools · RE Tools

Why IRA and Ghidra are painful here

IRA

• IRA is an Amiga program. It is meant to run on real Amiga hardware or in WinUAE / FS-UAE, not as a native Windows .exe.
• The listing EXTRACTED/mm2.asm in this repo was already produced by IRA V2.11. You can search and read it in any editor without running IRA yourself.
• To run IRA yourself: copy mm2 onto an Amiga hardfile, run IRA from Amiga, point it at the executable, export .asm. Expect setup friction (Kickstart, tools, paths).

Ghidra

• Ghidra has a 68000 CPU module, but no built-in Amiga hunk loader.
• Importing mm2 directly usually fails or shows garbage, because the file is a hunk container (headers, relocations, multiple segments), not a flat ROM image.
• Fixes that work on Windows:
  1. Import flat segments extracted by this repo (recommended below).
  2. Install a third-party Amiga hunk loader extension (search GitHub for ghidra amiga hunk loader — names change; pick one that mentions LoadSeg/hunk).

---

What works out of the box (Python)

From the MM2 folder:

pip install -r tools\requirements.txt

# Full linear disassembly (~50k insns) — Capstone + hunk parser
python tools\disasm_m68k.py mm2

# Segment binaries + addresses for Ghidra
python tools\extract_segments.py mm2

Outputs:

Tool	Output	Use
disasm_m68k.py	EXTRACTED/mm2.capstone.asm	Quick grep, opcode search
extract_segments.py	EXTRACTED/ghidra/mm2_code_*.bin, mm2_segments.json	Ghidra / IDA raw import
(existing)	EXTRACTED/mm2.asm	Best static listing (IRA labels, DC for data)

Data codec helpers (items.dat, str.dat)

Python:

python tools\mm2_codec.py items-decode items.dat items.csv
python tools\mm2_codec.py items-encode items.csv items.dat
python tools\mm2_codec.py str-decode str.dat str.txt
python tools\mm2_codec.py str-encode str.txt str.dat

C (compile + run):

gcc tools\mm2_codec.c -O2 -o tools\mm2_codec.exe
tools\mm2_codec.exe items-decode items.dat items.csv
tools\mm2_codec.exe items-encode items.csv items.dat
tools\mm2_codec.exe str-decode str.dat str.txt
tools\mm2_codec.exe str-encode str.txt str.dat

play is only an AmigaDOS script (assigns fonts/libs, runs mm2). Disassemble mm2, not play.

---

Ghidra: import extracted code (no plugin)

1. Run python tools\extract_segments.py mm2.

2. File → Import File → EXTRACTED/ghidra/mm2_code_00.bin.

3. Format: Raw binary.

4. Language: Motorola 68000:BE:32:default (big-endian 68000).

5. Base address: 0 (see mm2_segments.json → first code segment load_addr).

6. After import: Window → Memory Map — confirm base 00000000.

7. Navigate → Go To → 00000018 (entry after stub; see manifest entry_points).

8. Symbol → Create Label at that address (e.g. start).

9. Import mm2_code_01.bin at base 0x27adc (see manifest — data/BSS sit between the two code hunks in the real load map).

   Alternative: if you only care about code and want addresses to match mm2.capstone.asm, place the overlay at 0x2593c (immediately after the first code hunk, skipping data).

Limitations: relocations are not applied. Jump targets and A4-based Amiga OS calls may look wrong until you fix them manually or use a hunk loader plugin.

---

MM2 layout (for navigation)

Item	Offset (seg 0)	Notes
Hunk header	file 0x0	Not in extracted .bin
Code start	file 0x20 / load 0x0	First byte of mm2_code_00.bin
jmp main	load 0x18	4ef9 0002 48ae → game init ~0x248ae
Data	load 0x2593c	mm2_data_00.bin
Overlay code	load 0x27adc	mm2_code_01.bin (or 0x2593c in code-only views)

Main game init is around 0x248ae in the first code segment (search mm2.asm for LAB_248AE or capstone listing).

---

If you still want IRA-quality labels

Use EXTRACTED/mm2.asm as your primary reference. Re-run IRA only if you need a fresh export after patching the binary.

For interactive decompilation on Windows without Amiga: Ghidra + extracted segments, or consider Binary Ninja / IDA with the same flat binaries and manifest bases.