From dd08001c42b886fcd669cc32a26b98091d3677ab Mon Sep 17 00:00:00 2001
From: Wei-ju Wu <weiju.wu@gmail.com>
Date: Fri, 7 Nov 2014 16:06:19 -0800
Subject: [PATCH] started disassembler

---
 hunktools/disassemble.py | 109 +++++++++++++++++++++++++++++++++++++++
 hunktools/hunk.py        |  16 ++++--
 2 files changed, 122 insertions(+), 3 deletions(-)
 create mode 100644 hunktools/disassemble.py

diff --git a/hunktools/disassemble.py b/hunktools/disassemble.py
new file mode 100644
index 0000000..74237e2
--- /dev/null
+++ b/hunktools/disassemble.py
@@ -0,0 +1,109 @@
+import struct
+
+
+ADDR_MODES = {
+    '000': 'Dn', '001': 'An', '010': '(An)', '011': '(An)+', '100': '-(An)',
+    '101': '(d16,An)', '110': '(d8,An,Xn)', '111': 'EXT'  # -> ADDR_MODES_EXT
+}
+
+ADDR_MODES_EXT = {
+    '000': '(xxx).W', '001': '(xxx).L', '100': '#<data>',
+    '010': '(d16,PC)', '011': '(d8,PC,Xn)'
+}
+
+OPCODE_CATEGORIES = {
+    '0000': 'bitops_movep_imm', '0001': 'move.b', '0010': 'move.l', '0011': 'move.w',
+    '0100': 'misc', '1101': 'add_addx'
+}
+
+ADD_OPMODES = {
+    '000': ('b', 'ea+dn->dn'), '001': ('w', 'ea+dn->dn'),'010': ('l', 'ea+dn->dn'),
+    '100': ('b', 'dn+ea->ea'), '101': ('w', 'dn+ea->ea'),'110': ('l', 'dn+ea->ea')
+}
+
+
+def is_move(category):
+    return category in {'move.b', 'move.w', 'move.l'}
+
+
+def operand(size, mode_bits, reg_bits, data, offset):
+    result = ""
+    added = 0
+    mode = ADDR_MODES[mode_bits]
+    size = size.upper()
+
+    if mode == 'EXT':
+        mode = ADDR_MODES_EXT[reg_bits]
+        regnum = int(reg_bits, 2)
+        if mode == '#<data>':
+            data_offset = offset + 2
+            if size == 'L':
+                imm_value = struct.unpack('>i', data[data_offset:data_offset+4])[0]
+                added += 4
+            elif size == 'W':
+                data[data_offset:data_offset+2]
+                imm_value = struct.unpack('>h', data[data_offset:data_offset+2])[0]
+                added += 2
+            else:
+                raise Exception('unsupported size: ', size)
+            result = "#%d" % imm_value
+    else:
+        regnum = int(reg_bits, 2)
+        result = mode.replace('n', str(regnum))
+    return result, added
+    
+
+def disassemble_move(bits, data, offset):
+    category = OPCODE_CATEGORIES[bits[0:4]]
+    total_added = 0
+    # dst: reg|mode
+    dst_op, added = operand(category[-1], bits[7:10], bits[4:7], data, offset)
+    total_added += added
+
+    # src: mode|reg
+    src_op,added = operand(category[-1], bits[10:13], bits[13:16], data, offset)
+    total_added += added
+
+    return ((category, src_op, dst_op), total_added)
+
+
+def disassemble_add(bits, data, offset):
+    total_added = 0
+    reg = "D%d" % int(bits[4:7], 2)
+    size, operation = ADD_OPMODES[bits[7:10]]
+    ea, added = operand(size, bits[10:13], bits[13:16], data, offset)
+    total_added += added
+
+    if operation == 'ea+dn->dn':
+        src = ea
+        dst = reg
+    elif operation == 'dn+ea->ea':
+        src = reg
+        dst = ea
+    else:
+        raise Exception('Unknown operation for add')
+
+    return (('add.%s' % size, src, dst), total_added)
+
+
+def disassemble(data, offset):
+    bits = "{0:016b}".format(struct.unpack(">H", data[offset:offset+2])[0])
+    # first step categorize by looking at bits 15-12
+    opcode = bits[0:4]
+    category = OPCODE_CATEGORIES[opcode]
+    if is_move(category):
+        op, added = disassemble_move(bits, data, offset)
+    elif category == 'add_addx':
+        if bits[7] == 1 and bits[10:12] == '11':  # addx
+            raise Exception('addx not supported yet')
+        else:
+            op, added = disassemble_add(bits, data, offset)
+    else:
+        print("category: ", category)
+        raise Exception('TODO')
+    return (op, 2 + added)
+
+
+def print_instruction(address, op):
+    opcode, src, dst = op
+    print("$%08x:\t%8s %8s, %8s" % (address, opcode, src, dst))
diff --git a/hunktools/hunk.py b/hunktools/hunk.py
index ad73b5e..8e8036e 100755
--- a/hunktools/hunk.py
+++ b/hunktools/hunk.py
@@ -5,6 +5,8 @@ an Amiga Hunk file.
 import argparse
 import struct
 
+from disassemble import disassemble, print_instruction
+
 HUNK_BLOCK_UNIT         = b'\x00\x00\x03\xe7'
 HUNK_BLOCK_NAME         = b'\x00\x00\x03\xe8'
 HUNK_BLOCK_CODE         = b'\x00\x00\x03\xe9'
@@ -153,11 +155,19 @@ def parse_hunkfile(hunkfile):
         print("%d blocks read: " % len(blocks))
         for i, block in enumerate(blocks):
             if block[0] == 'NAME':
-                print("%d: '%s' -> '%s'" % (i, block[0], block[1]))
+                print("Block %d: '%s' -> '%s'" % (i, block[0], block[1]))
             elif block[0] == 'BSS':
-                print("%d: '%s' -> %d" % (i, block[0], block[1]))
-            else:
+                print("Block %d: '%s' -> %d" % (i, block[0], block[1]))
+            elif block[0] == 'CODE':
                 print("%d: '%s'" % (i, block[0]))
+                code = block[1]
+                offset = 0
+                while offset < len(code):
+                    op, size = disassemble(code, offset)
+                    print_instruction(offset, op)
+                    offset += size
+            else:
+                print("Block %d: '%s'" % (i, block[0]))
 
 
 if __name__ == '__main__':