Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
MorsGames
GitHub Repository: MorsGames/sm64plus
Path: blob/master/first-diff.py
7853 views
1
#!/usr/bin/env python3
2
import os.path
3
import argparse
4
from subprocess import check_call
5
6
# TODO: -S argument for shifted ROMs
7
8
parser = argparse.ArgumentParser(
9
description="find the first difference(s) between the compiled ROM and the baserom"
10
)
11
versionGroup = parser.add_mutually_exclusive_group()
12
versionGroup.add_argument(
13
"-j",
14
"--jp",
15
help="use original Japanese version",
16
action="store_const",
17
const="jp",
18
dest="version",
19
)
20
versionGroup.add_argument(
21
"-u",
22
"--us",
23
help="use United States version",
24
action="store_const",
25
const="us",
26
dest="version",
27
)
28
versionGroup.add_argument(
29
"-e",
30
"--eu",
31
help="use European (PAL) version",
32
action="store_const",
33
const="eu",
34
dest="version",
35
)
36
versionGroup.add_argument(
37
"-s",
38
"--sh",
39
help="use Shindou (Rumble) version",
40
action="store_const",
41
const="sh",
42
dest="version",
43
)
44
parser.add_argument(
45
"-m", "--make", help="run make before finding difference(s)", action="store_true"
46
)
47
parser.add_argument(
48
"-c",
49
"--count",
50
type=int,
51
default=1,
52
help="find up to this many instruction difference(s)",
53
)
54
parser.add_argument(
55
"-n", "--by-name", type=str, default="", help="perform a symbol or address lookup"
56
)
57
parser.add_argument(
58
"-d", "--diff", action="store_true", help="run ./diff.py on the result"
59
)
60
args = parser.parse_args()
61
diff_count = args.count
62
63
version = args.version
64
65
if version is None:
66
version = "us"
67
best = 0
68
for ver in ["us", "jp", "eu", "sh"]:
69
try:
70
mtime = os.path.getmtime(f"build/{ver}/sm64.{ver}.z64")
71
if mtime > best:
72
best = mtime
73
version = ver
74
except Exception:
75
pass
76
print("Assuming version " + version)
77
78
if args.make:
79
check_call(["make", "-j4", "VERSION=" + version, "COMPARE=0"])
80
81
baseimg = f"baserom.{version}.z64"
82
basemap = f"sm64.{version}.map"
83
84
myimg = f"build/{version}/sm64.{version}.z64"
85
mymap = f"build/{version}/{basemap}"
86
87
if os.path.isfile("expected/" + mymap):
88
basemap = "expected/" + mymap
89
90
required_files = [baseimg, myimg, mymap]
91
if not os.path.isfile(baseimg):
92
print(baseimg + " must exist.")
93
exit(1)
94
if not os.path.isfile(myimg) or not os.path.isfile(mymap):
95
print(
96
myimg
97
+ " and "
98
+ mymap
99
+ " must exist. Try rerunning with --make to build them."
100
)
101
exit(1)
102
103
mybin = open(myimg, "rb").read()
104
basebin = open(baseimg, "rb").read()
105
106
if len(mybin) != len(basebin):
107
print("Modified ROM has different size...")
108
exit(1)
109
110
if mybin == basebin:
111
print("No differences!")
112
if not args.by_name:
113
exit(0)
114
115
116
def search_map(rom_addr):
117
ram_offset = None
118
last_ram = 0
119
last_rom = 0
120
last_fn = "<start of rom>"
121
last_file = "<no file>"
122
prev_line = ""
123
with open(mymap) as f:
124
for line in f:
125
if "load address" in line:
126
# Example: ".boot 0x0000000004000000 0x1000 load address 0x0000000000000000"
127
if "noload" in line or "noload" in prev_line:
128
ram_offset = None
129
continue
130
ram = int(line[16 : 16 + 18], 0)
131
rom = int(line[59 : 59 + 18], 0)
132
ram_offset = ram - rom
133
continue
134
prev_line = line
135
136
if (
137
ram_offset is None
138
or "=" in line
139
or "*fill*" in line
140
or " 0x" not in line
141
):
142
continue
143
ram = int(line[16 : 16 + 18], 0)
144
rom = ram - ram_offset
145
fn = line.split()[-1]
146
if "0x" in fn:
147
ram_offset = None
148
continue
149
if rom > rom_addr or (rom_addr & 0x80000000 and ram > rom_addr):
150
return f"in {last_fn} (ram 0x{last_ram:08x}, rom 0x{last_rom:06x}, {last_file})"
151
last_ram = ram
152
last_rom = rom
153
last_fn = fn
154
if "/" in fn:
155
last_file = fn
156
return "at end of rom?"
157
158
159
def parse_map(fname):
160
ram_offset = None
161
cur_file = "<no file>"
162
syms = {}
163
prev_sym = None
164
prev_line = ""
165
with open(fname) as f:
166
for line in f:
167
if "load address" in line:
168
if "noload" in line or "noload" in prev_line:
169
ram_offset = None
170
continue
171
ram = int(line[16 : 16 + 18], 0)
172
rom = int(line[59 : 59 + 18], 0)
173
ram_offset = ram - rom
174
continue
175
prev_line = line
176
177
if (
178
ram_offset is None
179
or "=" in line
180
or "*fill*" in line
181
or " 0x" not in line
182
):
183
continue
184
ram = int(line[16 : 16 + 18], 0)
185
rom = ram - ram_offset
186
fn = line.split()[-1]
187
if "0x" in fn:
188
ram_offset = None
189
elif "/" in fn:
190
cur_file = fn
191
else:
192
syms[fn] = (rom, cur_file, prev_sym, ram)
193
prev_sym = fn
194
return syms
195
196
197
def map_diff():
198
map1 = parse_map(mymap)
199
map2 = parse_map(basemap)
200
min_ram = None
201
found = None
202
for sym, addr in map1.items():
203
if sym not in map2:
204
continue
205
if addr[0] != map2[sym][0]:
206
if min_ram is None or addr[0] < min_ram:
207
min_ram = addr[0]
208
found = (sym, addr[1], addr[2])
209
if min_ram is None:
210
return False
211
else:
212
print()
213
print(
214
f"Map appears to have shifted just before {found[0]} ({found[1]}) -- in {found[2]}?"
215
)
216
if found[2] is not None and found[2] not in map2:
217
print()
218
print(
219
f"(Base map file {basemap} out of date due to renamed symbols, so result may be imprecise.)"
220
)
221
return True
222
223
224
def hexbytes(bs):
225
return ":".join("{:02x}".format(c) for c in bs)
226
227
228
# For convenience, allow `./first-diff.py <ROM addr | RAM addr | function name>`
229
# to do a symbol <-> address lookup. This should really be split out into a
230
# separate script...
231
if args.by_name:
232
try:
233
addr = int(args.by_name, 0)
234
print(args.by_name, "is", search_map(addr))
235
except ValueError:
236
m = parse_map(mymap)
237
try:
238
print(
239
args.by_name,
240
"is at position",
241
hex(m[args.by_name][0]),
242
"in ROM,",
243
hex(m[args.by_name][3]),
244
"in RAM",
245
)
246
except KeyError:
247
print("function", args.by_name, "not found")
248
exit()
249
250
found_instr_diff = []
251
map_search_diff = []
252
diffs = 0
253
shift_cap = 1000
254
for i in range(24, len(mybin), 4):
255
# (mybin[i:i+4] != basebin[i:i+4], but that's slightly slower in CPython...)
256
if diffs <= shift_cap and (
257
mybin[i] != basebin[i]
258
or mybin[i + 1] != basebin[i + 1]
259
or mybin[i + 2] != basebin[i + 2]
260
or mybin[i + 3] != basebin[i + 3]
261
):
262
if diffs == 0:
263
print(f"First difference at ROM addr {hex(i)}, {search_map(i)}")
264
print(
265
f"Bytes: {hexbytes(mybin[i : i + 4])} vs {hexbytes(basebin[i : i + 4])}"
266
)
267
diffs += 1
268
if (
269
len(found_instr_diff) < diff_count
270
and mybin[i] >> 2 != basebin[i] >> 2
271
and not search_map(i) in map_search_diff
272
):
273
found_instr_diff.append(i)
274
map_search_diff.append(search_map(i))
275
if diffs == 0:
276
print("No differences!")
277
if not args.by_name:
278
exit()
279
definite_shift = diffs > shift_cap
280
if not definite_shift:
281
print(str(diffs) + " differing word(s).")
282
283
if diffs > 100:
284
if len(found_instr_diff) > 0:
285
for i in found_instr_diff:
286
print(f"Instruction difference at ROM addr {hex(i)}, {search_map(i)}")
287
print(
288
f"Bytes: {hexbytes(mybin[i : i + 4])} vs {hexbytes(basebin[i : i + 4])}"
289
)
290
if version == "sh":
291
print("Shifted ROM, as expected.")
292
else:
293
if not os.path.isfile(basemap):
294
if definite_shift:
295
print("Tons of differences, must be a shifted ROM.")
296
print(
297
"To find ROM shifts, copy a clean .map file to "
298
+ basemap
299
+ " and rerun this script."
300
)
301
exit()
302
303
if not map_diff():
304
print(f"No ROM shift{' (!?)' if definite_shift else ''}")
305
if args.diff:
306
diff_args = input("Call ./diff.py with which arguments? ") or "--"
307
if diff_args[0] != "-":
308
diff_args = "-" + diff_args
309
if "w" in diff_args and args.make:
310
diff_args += "m" # To avoid warnings when passing -w, also pass -m as long as -m was passed to first-diff itself
311
312
check_call(
313
[
314
"python3",
315
"diff.py",
316
f"-{version[0]}",
317
diff_args,
318
search_map(found_instr_diff[0]).split()[1],
319
]
320
)
321
322