2
# Test tool to compare Capstone output with llvm-mc. By Nguyen Anh Quynh, 2014
6
from pathlib import Path
11
print("Test %s" % fname)
12
fpath = Path(fname) if isinstance(fname, str) else fname
14
if fpath.exists() is False:
16
for f in fpath.iterdir():
20
with fpath.open() as f:
23
if not lines[0].startswith('# '):
24
print("ERROR: decoding information is missing")
27
# skip '# ' at the front, then split line to get out hexcode
28
# Note: option can be '', or 'None'
30
# print lines[0][2:].split(', ')
31
(arch, mode, option) = lines[0][2:].split(', ')
32
mode = mode.replace(' ', '')
33
option = option.strip()
36
"CS_ARCH_ARM": CS_ARCH_ARM,
37
"CS_ARCH_AARCH64": CS_ARCH_AARCH64,
38
"CS_ARCH_MIPS": CS_ARCH_MIPS,
39
"CS_ARCH_PPC": CS_ARCH_PPC,
40
"CS_ARCH_SPARC": CS_ARCH_SPARC,
41
"CS_ARCH_SYSZ": CS_ARCH_SYSZ,
42
"CS_ARCH_X86": CS_ARCH_X86,
43
"CS_ARCH_XCORE": CS_ARCH_XCORE,
44
"CS_ARCH_RISCV": CS_ARCH_RISCV,
45
"CS_ARCH_TRICORE": CS_ARCH_TRICORE,
46
"CS_ARCH_ALPHA": CS_ARCH_ALPHA,
47
"CS_ARCH_HPPA": CS_ARCH_HPPA,
51
"CS_MODE_16": CS_MODE_16,
52
"CS_MODE_32": CS_MODE_32,
53
"CS_MODE_64": CS_MODE_64,
54
"CS_MODE_MIPS32": CS_MODE_MIPS32,
55
"CS_MODE_MIPS64": CS_MODE_MIPS64,
57
"CS_MODE_ARM": CS_MODE_ARM,
58
"CS_MODE_THUMB": CS_MODE_THUMB,
59
"CS_MODE_ARM+CS_MODE_V8": CS_MODE_ARM + CS_MODE_V8,
60
"CS_MODE_THUMB+CS_MODE_V8": CS_MODE_THUMB + CS_MODE_V8,
61
"CS_MODE_THUMB+CS_MODE_MCLASS": CS_MODE_THUMB + CS_MODE_MCLASS,
62
"CS_MODE_THUMB+CS_MODE_V8+CS_MODE_MCLASS": CS_MODE_THUMB+CS_MODE_V8+CS_MODE_MCLASS,
63
"CS_MODE_LITTLE_ENDIAN": CS_MODE_LITTLE_ENDIAN,
64
"CS_MODE_BIG_ENDIAN": CS_MODE_BIG_ENDIAN,
65
"CS_MODE_64+CS_MODE_LITTLE_ENDIAN": CS_MODE_64 + CS_MODE_LITTLE_ENDIAN,
66
"CS_MODE_64+CS_MODE_BIG_ENDIAN": CS_MODE_64 + CS_MODE_BIG_ENDIAN,
67
"CS_MODE_MIPS32+CS_MODE_MICRO": CS_MODE_MIPS32 + CS_MODE_MICRO,
68
"CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS32 + CS_MODE_MICRO + CS_MODE_BIG_ENDIAN,
69
"CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN+CS_MODE_MICRO": CS_MODE_MIPS32 + CS_MODE_MICRO + CS_MODE_BIG_ENDIAN,
70
"CS_MODE_BIG_ENDIAN+CS_MODE_V9": CS_MODE_BIG_ENDIAN + CS_MODE_V9,
71
"CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS32 + CS_MODE_BIG_ENDIAN,
72
"CS_MODE_MIPS32+CS_MODE_LITTLE_ENDIAN": CS_MODE_MIPS32 + CS_MODE_LITTLE_ENDIAN,
73
"CS_MODE_MIPS64+CS_MODE_LITTLE_ENDIAN": CS_MODE_MIPS64 + CS_MODE_LITTLE_ENDIAN,
74
"CS_MODE_MIPS64+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS64 + CS_MODE_BIG_ENDIAN,
75
"CS_MODE_RISCV32": CS_MODE_RISCV32,
76
"CS_MODE_RISCV64": CS_MODE_RISCV64,
77
"CS_MODE_TRICORE_110": CS_MODE_TRICORE_110,
78
"CS_MODE_TRICORE_120": CS_MODE_TRICORE_120,
79
"CS_MODE_TRICORE_130": CS_MODE_TRICORE_130,
80
"CS_MODE_TRICORE_131": CS_MODE_TRICORE_131,
81
"CS_MODE_TRICORE_160": CS_MODE_TRICORE_160,
82
"CS_MODE_TRICORE_161": CS_MODE_TRICORE_161,
83
"CS_MODE_TRICORE_162": CS_MODE_TRICORE_162,
84
"CS_MODE_BIG_ENDIAN+CS_MODE_QPX": CS_MODE_BIG_ENDIAN+CS_MODE_QPX,
85
"CS_MODE_HPPA_11": CS_MODE_HPPA_11,
86
"CS_MODE_HPPA_20": CS_MODE_HPPA_20,
87
"CS_MODE_HPPA_20W": CS_MODE_HPPA_20W,
91
("CS_ARCH_X86", "CS_MODE_32"): 0,
92
("CS_ARCH_X86", "CS_MODE_64"): 1,
93
("CS_ARCH_ARM", "CS_MODE_ARM"): 2,
94
("CS_ARCH_ARM", "CS_MODE_THUMB"): 3,
95
("CS_ARCH_ARM", "CS_MODE_ARM+CS_MODE_V8"): 4,
96
("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_V8"): 5,
97
("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_MCLASS"): 6,
98
("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_V8+CS_MODE_MCLASS"): 7,
99
("CS_ARCH_AARCH64", "0"): 8,
100
("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN"): 9,
101
("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_MICRO"): 10,
102
("CS_ARCH_MIPS", "CS_MODE_MIPS64"): 11,
103
("CS_ARCH_MIPS", "CS_MODE_MIPS32"): 12,
104
("CS_ARCH_MIPS", "CS_MODE_MIPS64+CS_MODE_BIG_ENDIAN"): 13,
105
("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN"): 14,
106
("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN+CS_MODE_MICRO"): 14,
107
("CS_ARCH_PPC", "CS_MODE_BIG_ENDIAN"): 15,
108
("CS_ARCH_SPARC", "CS_MODE_BIG_ENDIAN"): 16,
109
("CS_ARCH_SPARC", "CS_MODE_BIG_ENDIAN+CS_MODE_V9"): 17,
110
("CS_ARCH_SYSZ", "0"): 18,
111
("CS_ARCH_XCORE", "0"): 19,
112
("CS_ARCH_MIPS", "CS_MODE_MIPS32R6+CS_MODE_BIG_ENDIAN"): 20,
113
("CS_ARCH_MIPS", "CS_MODE_MIPS32R6+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN"): 21,
114
("CS_ARCH_MIPS", "CS_MODE_MIPS32R6"): 22,
115
("CS_ARCH_MIPS", "CS_MODE_MIPS32R6+CS_MODE_MICRO"): 23,
116
("CS_ARCH_M68K", "0"): 24,
117
("CS_ARCH_M680X", "CS_MODE_M680X_6809"): 25,
118
("CS_ARCH_EVM", "0"): 26,
119
("CS_ARCH_BPF", "CS_MODE_LITTLE_ENDIAN+CS_MODE_BPF_CLASSIC"): 30,
120
("CS_ARCH_BPF", "CS_MODE_LITTLE_ENDIAN+CS_MODE_BPF_EXTENDED"): 31,
121
("CS_ARCH_BPF", "CS_MODE_BIG_ENDIAN+CS_MODE_BPF_CLASSIC"): 32,
122
("CS_ARCH_BPF", "CS_MODE_BIG_ENDIAN+CS_MODE_BPF_EXTENDED"): 33,
123
("CS_ARCH_RISCV", "CS_MODE_RISCV32"): 45,
124
("CS_ARCH_RISCV", "CS_MODE_RISCV64"): 46,
125
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_110"): 47,
126
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_120"): 48,
127
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_130"): 49,
128
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_131"): 50,
129
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_160"): 51,
130
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_161"): 52,
131
("CS_ARCH_TRICORE", "CS_MODE_TRICORE_162"): 53,
132
("CS_ARCH_PPC", "CS_MODE_BIG_ENDIAN+CS_MODE_QPX"): 54,
133
("CS_ARCH_ALPHA", "CS_MODE_LITTLE_ENDIAN"): 55,
134
("CS_ARCH_ALPHA", "CS_MODE_BIG_ENDIAN"): 56,
135
("CS_ARCH_HPPA", "CS_MODE_HPPA_11+CS_MODE_BIG_ENDIAN"): 57,
136
("CS_ARCH_HPPA", "CS_MODE_HPPA_20+CS_MODE_BIG_ENDIAN"): 58,
139
# if not option in ('', 'None'):
140
# print archs[arch], modes[mode], options[option]
142
for line in lines[1:]:
143
# ignore all the input lines having # in front.
144
if line.startswith('#'):
146
if line.startswith('// '):
148
# print("Check %s" %line)
149
code = line.split(' = ')[0]
152
if code.find('//') >= 0:
154
hex_code = code.replace('0x', '').replace(',', '').replace(' ', '').strip()
156
hex_data = bytes.fromhex(hex_code)
157
fpath = Path("fuzz/corpus/%s_%s" % (os.path.basename(fname), hex_code))
158
if fpath.parent.exists() is False:
159
fpath.parent.mkdir(parents=True)
160
with fpath.open('wb') as fout:
161
if (arch, mode) not in mc_modes:
162
print("fail", arch, mode)
165
fout.write(mc_modes[(arch, mode)].to_bytes(1, 'little'))
167
except Exception as e:
168
print(f"skipping: {hex_code} with: {e}")
172
if __name__ == '__main__':
173
if len(sys.argv) == 1:
174
fnames = sys.stdin.readlines()
176
test_file(fname.strip())
178
# print("Usage: ./test_mc.py <input-file.s.cs>")
179
test_file(sys.argv[1])