1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
|
#!/usr/bin/env python3
# (barebones) utility to generate tests and search encodings and mnemonics
# TODO: proper command line argument parsing, and help
import re, sys, codecs
N_SAMPLES = 4 # number of samples for each encoding
from arm64test import lift, ATTR_PTR_AUTH, path_il_h
if not sys.argv[1:]:
sys.exit(-1)
arch = None
def disassemble(addr, data):
global arch
if not arch:
arch = binaryninja.Architecture["aarch64"]
(tokens, length) = arch.get_instruction_text(data, addr)
if not tokens or length == 0:
return None
return disasm_test.normalize("".join([x.text for x in tokens]))
def print_case(data, comment=""):
ilstr, attributes = lift(data)
il_lines = ilstr.split(";")
if len(il_lines) == 2 and len(ilstr) < 60:
il_lines = [ilstr]
# print("\t(b'%s', " % ("".join(["\\x%02X" % b for b in data])), end="")
print(" (b'%s', " % ("".join(["\\x%02X" % b for b in data])), end="")
for i, line in enumerate(il_lines):
if i != 0:
# print("\t\t\t\t\t\t ", end="")
print(" " * (4 * 6 + 1), end="")
print("'%s" % line, end="")
if i != len(il_lines) - 1:
print(";' + \\")
# comment = comment or ""
# comment += " %s" % len(il_lines)
comment = " # " + comment if comment else ""
attr = ''
if attributes:
# attr = ", \"%s\"" % repr(list(attributes)[0])
if ATTR_PTR_AUTH in attributes:
attr = ", ATTR_PTR_AUTH"
print("'%s),%s" % (attr, comment))
def gather_samples(mnems, encodings):
encodings = [x.upper() for x in encodings]
global N_SAMPLES
fpath = "./disassembler/test_cases.txt"
with open(fpath, "rt") as fp:
lines_read = fp.read()
mnems = [re.compile(x, re.IGNORECASE) for x in mnems]
samples = 0
current_encoding = None
# not_sample_line_pat = re.compile(r"^// (\w*) .*", re.IGNORECASE)
encoding_line_pat = re.compile(r"^// (\w*_\w*?) .*", re.IGNORECASE)
# sample_line_pat = re.compile(r"^(..)(..)(..)(..) (.*)$")
sample_line_pat = re.compile(r"^([\dA-F]{2})([\dA-F]{2})([\dA-F]{2})([\dA-F]{2}) (.*)$", re.IGNORECASE)
for i, line in enumerate(lines_read.splitlines()):
_line = line.strip().upper()
if _line.startswith("// NOTE:"):
continue
if _line.startswith("// SYNTAX:"):
continue
if _line.startswith("// https:"):
continue
if _line.startswith("// HTTPS:"):
continue
if _line.startswith("// 1101010100|L=0|OP0=00|OP1=011|CRN=0011|CRM=0100|1|OPC=00|RT=11111"):
continue
if _line.endswith("// TCOMMIT"):
continue
if _line.endswith("// DRPS"):
continue
if _line.endswith("// ERET"):
continue
if _line.endswith("// ERETAA"):
continue
if _line.endswith("// ERETAB"):
continue
if _line.endswith("// PSSBB"):
continue
if _line.endswith("// SSBB"):
continue
if _line.endswith("// PSSBB_DSB_BO_BARRIERS"):
continue
# if re.match(r"^// .*? .*", line):
m = encoding_line_pat.match(line)
if m:
# m = re.match(r"^// (.*?) .*", line)
# example:
# // BFCVT_Z_P_Z_S2BF 01100101|opc=10|0010|opc2=10|101|Pg=xxx|Zn=xxxxx|Zd=xxxxx
current_encoding = m.group(1)
samples = 0
continue
# if not_sample_line_pat.match(line):
# continue
if line.startswith("//"):
continue
# m = re.match(r"^(..)(..)(..)(..) (.*)$", line)
m = sample_line_pat.match(line)
if m:
# example:
# 658AB9BB bfcvt z27.h, p6/m, z13.s
if samples >= N_SAMPLES:
continue
(b0, b1, b2, b3, instxt) = m.group(1, 2, 3, 4, 5)
data = codecs.decode(b3 + b2 + b1 + b0, "hex_codec")
# if not (instxt==mnem or instxt.startswith(mnem+' ')):
# mnemonic_match = [x for x in mnems if instxt.lower().startswith(x.lower()) or current_encoding.lower().startswith(x.lower())]
mnemonic_match = [
x for x in mnems if x.search(instxt) or x.search(current_encoding)
]
encoding_match = current_encoding.upper() in encodings
if not (mnemonic_match or encoding_match):
continue
# if samples == 0:
# print('\t# %s' % encoding)
# print("\t# %s %s" % (instxt.ljust(64), current_encoding))
print(" # %s %s" % (instxt.ljust(64), current_encoding))
print_case(data)
samples += 1
continue
print("unable to parse line (%d): %r" % (i + 1, line))
sys.exit(-1)
# generate lifting tests for a given mnemonic
# example:
# ./test_gen mnemonic ld1
# regex matching (ignoring case) for mnemonics
if sys.argv[1] == "mnemonic":
mnems = sys.argv[2:]
for mnem in mnems:
print("searching for mnemonic -%s-" % mnem, file=sys.stderr)
gather_samples([mnem], [])
# exact match (ignoring case) for encodings
elif sys.argv[1] == "encoding":
encnames = sys.argv[2:]
for encname in encnames:
print("searching for encoding -%s-" % encname, file=sys.stderr)
gather_samples([], [encname])
elif sys.argv[1] == "mte":
mnems = [
"addg",
"cmpp",
"gmi",
"irg",
"ldg",
"dgv",
"ldgm",
"st2g",
"stg",
"stgm",
"stgp",
"stgv",
"stz2g",
"stzg",
"stzgm",
"subg",
"subp",
"subps",
]
gather_samples(mnems, [])
elif sys.argv[1] == "recompute_arm64test":
with open("arm64test.py") as fp:
lines = [x.rstrip() for x in fp.readlines()]
i = 0
preserve = False
with open(path_il_h, "rt") as f:
LIFT_PAC_AS_INTRINSIC = "'#define LIFT_PAC_AS_INTRINSIC 1\n'" in f.readlines()
# print(f"{LIFT_PAC_AS_INTRINSIC=!r}", file=sys.stdout)
while i < len(lines):
if "testing that select PAC instructions lift to " in lines[i]:
if "testing that select PAC instructions lift to intrinsics" in lines[i]:
preserve = not LIFT_PAC_AS_INTRINSIC
elif "testing that select PAC instructions lift to NOP" in lines[i]:
preserve = LIFT_PAC_AS_INTRINSIC
# print(f"{LIFT_PAC_AS_INTRINSIC=!r} {preserve=!r}", file=sys.stdout)
if preserve:
print(lines[i])
i += 1
continue
m = re.match(r"^(?:\t| {4})\(b\'\\x(..)\\x(..)\\x(..)\\x(..)\'.*$", lines[i])
if m:
while i + 1 < len(lines) and re.match(r"^\s+\'.*$", lines[i + 1]):
lines[i] += lines[i + 1]
del lines[i + 1]
# if i + 1 < len(lines):
# if re.match(r"^\s+\'.*$", lines[i + 1]):
# while i + 1 < len(lines) and re.match(r"^\s+\'.*$", lines[i + 1]):
# lines[i] += lines[i + 1]
# del lines[i + 1]
# m = re.match(r"^(?:\t| {4})\(b\'\\x(..)\\x(..)\\x(..)\\x(..)\'.*($\n^.*)*?$\n(?=^ {4}\(b)", lines[i], re.M)
if not m:
print(lines[i])
i += 1
continue
(b0, b1, b2, b3) = m.group(1, 2, 3, 4)
comment = None
m = re.search(r"# (.*)$", lines[i])
if m:
comment = m.group(1)
data = codecs.decode(b0 + b1 + b2 + b3, "hex_codec")
print_case(data, comment)
i += 1
while lines[i].startswith("\t\t\t\t\t\t") or lines[i].startswith(" " * (4 * 6)):
i += 1
|
