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
|
# SPDX-License-Identifier: GPL-2.0-or-later
# FRR DWARF structure definition extractor
#
# Copyright (C) 2020 David Lamparter for NetDEF, Inc.
import sys
import os
import subprocess
import re
import argparse
import json
structs = [
"xref",
"xref_logmsg",
"xref_threadsched",
"xref_install_element",
"xrefdata",
"xrefdata_logmsg",
"cmd_element",
]
def extract(filename="lib/.libs/libfrr.so"):
"""
Convert output from "pahole" to JSON.
Example pahole output:
$ pahole -C xref lib/.libs/libfrr.so
struct xref {
struct xrefdata * xrefdata; /* 0 8 */
enum xref_type type; /* 8 4 */
int line; /* 12 4 */
const char * file; /* 16 8 */
const char * func; /* 24 8 */
/* size: 32, cachelines: 1, members: 5 */
/* last cacheline: 32 bytes */
};
"""
pahole = subprocess.check_output(
["pahole", "-C", ",".join(structs), filename]
).decode("UTF-8")
struct_re = re.compile(r"^struct ([^ ]+) \{([^\}]+)};", flags=re.M | re.S)
field_re = re.compile(
r"^\s*(?P<type>[^;\(]+)\s+(?P<name>[^;\[\]]+)(?:\[(?P<array>\d+)\])?;\s*\/\*(?P<comment>.*)\*\/\s*$"
)
comment_re = re.compile(r"^\s*\/\*.*\*\/\s*$")
pastructs = struct_re.findall(pahole)
out = {}
for sname, data in pastructs:
this = out.setdefault(sname, {})
fields = this.setdefault("fields", [])
lines = data.strip().splitlines()
next_offs = 0
for line in lines:
if line.strip() == "":
continue
m = comment_re.match(line)
if m is not None:
continue
m = field_re.match(line)
if m is not None:
offs, size = m.group("comment").strip().split()
offs = int(offs)
size = int(size)
typ_ = m.group("type").strip()
name = m.group("name")
if name.startswith("(*"):
# function pointer
typ_ = typ_ + " *"
name = name[2:].split(")")[0]
data = {
"name": name,
"type": typ_,
# 'offset': offs,
# 'size': size,
}
if m.group("array"):
data["array"] = int(m.group("array"))
fields.append(data)
if offs != next_offs:
raise ValueError(
"%d padding bytes before struct %s.%s"
% (offs - next_offs, sname, name)
)
next_offs = offs + size
continue
raise ValueError("cannot process line: %s" % line)
return out
class FieldApplicator(object):
"""
Fill ELFDissectStruct fields list from pahole/JSON
Uses the JSON file created by the above code to fill in the struct fields
in subclasses of ELFDissectStruct.
"""
# only what we really need. add more as needed.
packtypes = {
"int": "i",
"uint8_t": "B",
"uint16_t": "H",
"uint32_t": "I",
"char": "s",
}
def __init__(self, data):
self.data = data
self.classes = []
self.clsmap = {}
def add(self, cls):
self.classes.append(cls)
self.clsmap[cls.struct] = cls
def resolve(self, cls):
out = []
# offset = 0
fieldrename = getattr(cls, "fieldrename", {})
def mkname(n):
return (fieldrename.get(n, n),)
for field in self.data[cls.struct]["fields"]:
typs = field["type"].split()
typs = [i for i in typs if i not in ["const"]]
# this will break reuse of xrefstructs.json across 32bit & 64bit
# platforms
# if field['offset'] != offset:
# assert offset < field['offset']
# out.append(('_pad', '%ds' % (field['offset'] - offset,)))
# pretty hacky C types handling, but covers what we need
ptrlevel = 0
while typs[-1] == "*":
typs.pop(-1)
ptrlevel += 1
if ptrlevel > 0:
packtype = ("P", None)
if ptrlevel == 1:
if typs[0] == "char":
packtype = ("P", str)
elif typs[0] == "struct" and typs[1] in self.clsmap:
packtype = ("P", self.clsmap[typs[1]])
elif typs[0] == "enum":
packtype = ("I",)
elif typs[0] in self.packtypes:
packtype = (self.packtypes[typs[0]],)
elif typs[0] == "struct":
if typs[1] in self.clsmap:
packtype = (self.clsmap[typs[1]],)
else:
raise ValueError(
"embedded struct %s not in extracted data" % (typs[1],)
)
else:
raise ValueError(
"cannot decode field %s in struct %s (%s)"
% (cls.struct, field["name"], field["type"])
)
if "array" in field and typs[0] == "char":
packtype = ("%ds" % field["array"],)
out.append(mkname(field["name"]) + packtype)
elif "array" in field:
for i in range(0, field["array"]):
out.append(mkname("%s_%d" % (field["name"], i)) + packtype)
else:
out.append(mkname(field["name"]) + packtype)
# offset = field['offset'] + field['size']
cls.fields = out
def __call__(self):
for cls in self.classes:
self.resolve(cls)
def main():
argp = argparse.ArgumentParser(description="FRR DWARF structure extractor")
argp.add_argument(
"-o",
dest="output",
type=str,
help="write JSON output",
default="python/xrefstructs.json",
)
argp.add_argument(
"-i",
dest="input",
type=str,
help="ELF file to read",
default="lib/.libs/libfrr.so",
)
args = argp.parse_args()
out = extract(args.input)
with open(args.output + ".tmp", "w") as fd:
json.dump(out, fd, indent=2, sort_keys=True)
os.rename(args.output + ".tmp", args.output)
if __name__ == "__main__":
main()
|
