metasm 1.0.3 → 1.0.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- checksums.yaml.gz.sig +3 -0
- data.tar.gz.sig +0 -0
- data/Gemfile +3 -2
- data/metasm.gemspec +3 -2
- data/metasm.rb +4 -1
- data/metasm/compile_c.rb +2 -2
- data/metasm/cpu/arc/decode.rb +0 -21
- data/metasm/cpu/arc/main.rb +4 -4
- data/metasm/cpu/arm/decode.rb +1 -5
- data/metasm/cpu/arm/main.rb +3 -3
- data/metasm/cpu/arm64/decode.rb +2 -6
- data/metasm/cpu/arm64/main.rb +5 -5
- data/metasm/cpu/bpf/decode.rb +3 -35
- data/metasm/cpu/bpf/main.rb +5 -5
- data/metasm/cpu/bpf/render.rb +1 -12
- data/metasm/cpu/cy16/decode.rb +0 -6
- data/metasm/cpu/cy16/main.rb +3 -3
- data/metasm/cpu/cy16/render.rb +0 -11
- data/metasm/cpu/dalvik/decode.rb +4 -26
- data/metasm/cpu/dalvik/main.rb +20 -2
- data/metasm/cpu/dalvik/opcodes.rb +3 -2
- data/metasm/cpu/{mips/compile_c.rb → ebpf.rb} +5 -2
- data/metasm/cpu/ebpf/debug.rb +61 -0
- data/metasm/cpu/ebpf/decode.rb +142 -0
- data/metasm/cpu/ebpf/main.rb +58 -0
- data/metasm/cpu/ebpf/opcodes.rb +97 -0
- data/metasm/cpu/ebpf/render.rb +36 -0
- data/metasm/cpu/ia32/debug.rb +39 -1
- data/metasm/cpu/ia32/decode.rb +111 -90
- data/metasm/cpu/ia32/decompile.rb +45 -37
- data/metasm/cpu/ia32/main.rb +10 -0
- data/metasm/cpu/ia32/parse.rb +6 -0
- data/metasm/cpu/mcs51/decode.rb +1 -1
- data/metasm/cpu/mcs51/main.rb +11 -0
- data/metasm/cpu/mips/decode.rb +8 -18
- data/metasm/cpu/mips/main.rb +3 -3
- data/metasm/cpu/mips/opcodes.rb +1 -1
- data/metasm/cpu/msp430/decode.rb +2 -6
- data/metasm/cpu/msp430/main.rb +3 -3
- data/metasm/cpu/openrisc.rb +11 -0
- data/metasm/cpu/openrisc/debug.rb +106 -0
- data/metasm/cpu/openrisc/decode.rb +182 -0
- data/metasm/cpu/openrisc/decompile.rb +350 -0
- data/metasm/cpu/openrisc/main.rb +70 -0
- data/metasm/cpu/openrisc/opcodes.rb +109 -0
- data/metasm/cpu/openrisc/render.rb +37 -0
- data/metasm/cpu/ppc/decode.rb +0 -25
- data/metasm/cpu/ppc/main.rb +6 -6
- data/metasm/cpu/ppc/opcodes.rb +3 -4
- data/metasm/cpu/python/decode.rb +0 -20
- data/metasm/cpu/python/main.rb +1 -1
- data/metasm/cpu/sh4/decode.rb +2 -6
- data/metasm/cpu/sh4/main.rb +25 -23
- data/metasm/cpu/st20/decode.rb +0 -7
- data/metasm/cpu/webasm.rb +11 -0
- data/metasm/cpu/webasm/debug.rb +31 -0
- data/metasm/cpu/webasm/decode.rb +321 -0
- data/metasm/cpu/webasm/decompile.rb +386 -0
- data/metasm/cpu/webasm/encode.rb +104 -0
- data/metasm/cpu/webasm/main.rb +81 -0
- data/metasm/cpu/webasm/opcodes.rb +214 -0
- data/metasm/cpu/x86_64/compile_c.rb +13 -9
- data/metasm/cpu/x86_64/parse.rb +1 -1
- data/metasm/cpu/z80/decode.rb +0 -27
- data/metasm/cpu/z80/main.rb +3 -3
- data/metasm/cpu/z80/render.rb +0 -11
- data/metasm/debug.rb +43 -8
- data/metasm/decode.rb +62 -14
- data/metasm/decompile.rb +793 -466
- data/metasm/disassemble.rb +188 -131
- data/metasm/disassemble_api.rb +30 -17
- data/metasm/dynldr.rb +2 -2
- data/metasm/encode.rb +8 -2
- data/metasm/exe_format/autoexe.rb +2 -0
- data/metasm/exe_format/coff.rb +21 -3
- data/metasm/exe_format/coff_decode.rb +12 -0
- data/metasm/exe_format/coff_encode.rb +6 -3
- data/metasm/exe_format/dex.rb +13 -3
- data/metasm/exe_format/elf.rb +12 -2
- data/metasm/exe_format/elf_decode.rb +59 -1
- data/metasm/exe_format/main.rb +2 -0
- data/metasm/exe_format/mz.rb +1 -0
- data/metasm/exe_format/pe.rb +25 -3
- data/metasm/exe_format/wasm.rb +402 -0
- data/metasm/gui/dasm_decomp.rb +171 -95
- data/metasm/gui/dasm_graph.rb +61 -2
- data/metasm/gui/dasm_hex.rb +2 -2
- data/metasm/gui/dasm_main.rb +45 -19
- data/metasm/gui/debug.rb +13 -4
- data/metasm/gui/gtk.rb +12 -4
- data/metasm/main.rb +108 -103
- data/metasm/os/emulator.rb +175 -0
- data/metasm/os/main.rb +11 -6
- data/metasm/parse.rb +23 -12
- data/metasm/parse_c.rb +189 -135
- data/metasm/preprocessor.rb +16 -1
- data/misc/openrisc-parser.rb +79 -0
- data/samples/dasm-plugins/scanxrefs.rb +6 -4
- data/samples/dasm-plugins/selfmodify.rb +8 -8
- data/samples/dbg-plugins/trace_func.rb +1 -1
- data/samples/disassemble-gui.rb +14 -3
- data/samples/emubios.rb +251 -0
- data/samples/emudbg.rb +127 -0
- data/samples/lindebug.rb +79 -78
- data/samples/metasm-shell.rb +8 -8
- data/tests/all.rb +1 -1
- data/tests/expression.rb +2 -0
- data/tests/graph_layout.rb +1 -1
- data/tests/ia32.rb +1 -0
- data/tests/mips.rb +1 -1
- data/tests/preprocessor.rb +18 -0
- metadata +124 -6
- metadata.gz.sig +0 -0
data/metasm/exe_format/main.rb
CHANGED
@@ -217,6 +217,8 @@ class ExeFormat
|
|
217
217
|
|
218
218
|
def shortname; self.class.name.split('::').last.downcase; end
|
219
219
|
|
220
|
+
def inspect; "<Metasm::ExeFormat %s @%x>" % [shortname, object_id]; end
|
221
|
+
|
220
222
|
module IntToHash
|
221
223
|
# converts a constant name to its numeric value using the hash
|
222
224
|
# {1 => 'toto', 2 => 'tata'}: 'toto' => 1, 42 => 42, 'tutu' => raise
|
data/metasm/exe_format/mz.rb
CHANGED
@@ -11,6 +11,7 @@ require 'metasm/decode'
|
|
11
11
|
module Metasm
|
12
12
|
class MZ < ExeFormat
|
13
13
|
MAGIC = 'MZ' # 0x4d5a
|
14
|
+
MAGIC.force_encoding('BINARY') if MAGIC.respond_to?(:force_encoding)
|
14
15
|
class Header < SerialStruct
|
15
16
|
mem :magic, 2, MAGIC
|
16
17
|
words :cblp, :cp, :crlc, :cparhdr, :minalloc, :maxalloc, :ss, :sp, :csum, :ip, :cs, :lfarlc, :ovno
|
data/metasm/exe_format/pe.rb
CHANGED
@@ -11,8 +11,9 @@ require 'metasm/exe_format/coff'
|
|
11
11
|
module Metasm
|
12
12
|
class PE < COFF
|
13
13
|
MAGIC = "PE\0\0" # 0x50450000
|
14
|
+
MAGIC.force_encoding('BINARY') if MAGIC.respond_to?(:force_encoding)
|
14
15
|
|
15
|
-
attr_accessor :coff_offset, :signature, :mz
|
16
|
+
attr_accessor :coff_offset, :signature, :mz, :productid
|
16
17
|
|
17
18
|
def initialize(*a)
|
18
19
|
super(*a)
|
@@ -26,7 +27,14 @@ class PE < COFF
|
|
26
27
|
def decode_header
|
27
28
|
@cursection ||= self
|
28
29
|
@encoded.ptr = 0x3c
|
29
|
-
|
30
|
+
lfanew = decode_word(@encoded)
|
31
|
+
|
32
|
+
try_rich_sz = [0x400, lfanew].min
|
33
|
+
rich_ary = []
|
34
|
+
rich_ary << decode_word(@encoded) while @encoded.ptr < try_rich_sz
|
35
|
+
@productid = decode_productid(rich_ary)
|
36
|
+
|
37
|
+
@encoded.ptr = lfanew
|
30
38
|
@signature = @encoded.read(4)
|
31
39
|
raise InvalidExeFormat, "Invalid PE signature #{@signature.inspect}" if @signature != MAGIC
|
32
40
|
@coff_offset = @encoded.ptr
|
@@ -38,6 +46,20 @@ class PE < COFF
|
|
38
46
|
super()
|
39
47
|
end
|
40
48
|
|
49
|
+
RICH_MAGIC = 0x68636952 # "Rich"
|
50
|
+
def decode_productid(ary)
|
51
|
+
return unless idx = ary.index(RICH_MAGIC) and xorkey = ary[idx+1]
|
52
|
+
ary = ary[0, idx-1].map { |dw| dw ^ xorkey }
|
53
|
+
return unless idx = ary.rindex(0x536E6144) # "DanS"
|
54
|
+
ary = ary[idx+2..-1]
|
55
|
+
out = []
|
56
|
+
until ary.empty?
|
57
|
+
ar1 = ary.shift
|
58
|
+
out << { :id => ar1 >> 16, :ver => ar1 & 0xffff, :count => ary.shift.to_i }
|
59
|
+
end
|
60
|
+
out
|
61
|
+
end
|
62
|
+
|
41
63
|
# creates a default MZ file to be used in the PE header
|
42
64
|
# this one is specially crafted to fit in the 0x3c bytes before the signature
|
43
65
|
def encode_default_mz_header
|
@@ -226,7 +248,7 @@ EOS
|
|
226
248
|
a.each { |aa|
|
227
249
|
next if aa == Expression::Unknown
|
228
250
|
dasm.auto_label_at(aa, 'seh', 'loc', 'sub')
|
229
|
-
dasm.addrs_todo <<
|
251
|
+
dasm.addrs_todo << { :addr => aa }
|
230
252
|
}
|
231
253
|
super(dasm, di)
|
232
254
|
else
|
@@ -0,0 +1,402 @@
|
|
1
|
+
# This file is part of Metasm, the Ruby assembly manipulation suite
|
2
|
+
# Copyright (C) 2006-2009 Yoann GUILLOT
|
3
|
+
#
|
4
|
+
# Licence is LGPL, see LICENCE in the top-level directory
|
5
|
+
|
6
|
+
|
7
|
+
require 'metasm/exe_format/main'
|
8
|
+
require 'metasm/encode'
|
9
|
+
require 'metasm/decode'
|
10
|
+
|
11
|
+
|
12
|
+
module Metasm
|
13
|
+
# WebAssembly
|
14
|
+
# leb integer encoding taken from dex.rb
|
15
|
+
class WasmFile < ExeFormat
|
16
|
+
MAGIC = "\0asm"
|
17
|
+
MAGIC.force_encoding('binary') if MAGIC.respond_to?(:force_encoding)
|
18
|
+
|
19
|
+
SECTION_NAME = { 1 => 'Type', 2 => 'Import', 3 => 'Function', 4 => 'Table',
|
20
|
+
5 => 'Memory', 6 => 'Global', 7 => 'Export', 8 => 'Start',
|
21
|
+
9 => 'Element', 10 => 'Code', 11 => 'Data' }
|
22
|
+
|
23
|
+
TYPE = { -1 => 'i32', -2 => 'i64', -3 => 'f32', -4 => 'f64',
|
24
|
+
-0x10 => 'anyfunc', -0x20 => 'func', -0x40 => 'none' }
|
25
|
+
|
26
|
+
EXTERNAL_KIND = { 0 => 'function', 1 => 'table', 2 => 'memory', 3 => 'global' }
|
27
|
+
|
28
|
+
# begin WTF
|
29
|
+
OPCODE_IMM_COUNT = Hash.new(0)
|
30
|
+
[2, 3, 4, 0xc, 0xd, 0x10].each { |op| OPCODE_IMM_COUNT[op] = 1 }
|
31
|
+
OPCODE_IMM_COUNT[0x11] = 2
|
32
|
+
(0x20..0x24).each { |op| OPCODE_IMM_COUNT[op] = 1 }
|
33
|
+
(0x28..0x3e).each { |op| OPCODE_IMM_COUNT[op] = 2 }
|
34
|
+
(0x3f..0x42).each { |op| OPCODE_IMM_COUNT[op] = 1 }
|
35
|
+
# 0x43 followed by uint32, 0x44 followed by uint64 (float constants)
|
36
|
+
# end WTF
|
37
|
+
|
38
|
+
|
39
|
+
class SerialStruct < Metasm::SerialStruct
|
40
|
+
# TODO move uleb/sleb to new_field for sizeof
|
41
|
+
new_int_field :u4, :uleb, :sleb
|
42
|
+
end
|
43
|
+
|
44
|
+
class Header < SerialStruct
|
45
|
+
mem :sig, 4, MAGIC
|
46
|
+
decode_hook { |exe, hdr| raise InvalidExeFormat, "E: invalid WasmFile signature #{hdr.sig.inspect}" if hdr.sig != MAGIC }
|
47
|
+
u4 :ver, 1
|
48
|
+
end
|
49
|
+
|
50
|
+
class Module < SerialStruct
|
51
|
+
uleb :id
|
52
|
+
fld_enum :id, SECTION_NAME
|
53
|
+
uleb :payload_len
|
54
|
+
attr_accessor :edata, :raw_offset, :name
|
55
|
+
|
56
|
+
def decode(exe)
|
57
|
+
super(exe)
|
58
|
+
@raw_offset = exe.encoded.ptr
|
59
|
+
@edata = exe.encoded[exe.encoded.ptr, @payload_len]
|
60
|
+
exe.encoded.ptr += @payload_len
|
61
|
+
end
|
62
|
+
end
|
63
|
+
|
64
|
+
attr_accessor :endianness
|
65
|
+
|
66
|
+
def encode_u4(val) Expression[val].encode(:u32, @endianness) end
|
67
|
+
def decode_u4(edata = @encoded) edata.decode_imm(:u32, @endianness) end
|
68
|
+
def sizeof_u4 ; 4 ; end
|
69
|
+
def encode_uleb(val, signed=false)
|
70
|
+
v = val
|
71
|
+
out = EncodedData.new
|
72
|
+
while v > 0x7f or v < -0x40 or (signed and v > 0x3f)
|
73
|
+
out << Expression[0x80 | (v&0x7f)].encode(:u8, @endianness)
|
74
|
+
v >>= 7
|
75
|
+
end
|
76
|
+
out << Expression[v & 0x7f].encode(:u8, @endianness)
|
77
|
+
end
|
78
|
+
def decode_uleb(ed = @encoded, signed=false)
|
79
|
+
v = s = 0
|
80
|
+
while s < 10*7
|
81
|
+
b = ed.read(1).unpack('C').first.to_i
|
82
|
+
v |= (b & 0x7f) << s
|
83
|
+
s += 7
|
84
|
+
break if (b&0x80) == 0
|
85
|
+
end
|
86
|
+
v = Expression.make_signed(v, s) if signed
|
87
|
+
v
|
88
|
+
end
|
89
|
+
def encode_sleb(val) encode_uleb(val, true) end
|
90
|
+
def decode_sleb(ed = @encoded) decode_uleb(ed, true) end
|
91
|
+
attr_accessor :header, :modules, :type, :import, :function_signature,
|
92
|
+
:table, :memory, :global, :export, :start_function_index,
|
93
|
+
:element, :function_body, :data, :code_info
|
94
|
+
|
95
|
+
def initialize(endianness=:little)
|
96
|
+
@endianness = endianness
|
97
|
+
@encoded = EncodedData.new
|
98
|
+
super()
|
99
|
+
end
|
100
|
+
|
101
|
+
def decode_type(edata=@encoded)
|
102
|
+
form = decode_sleb(edata)
|
103
|
+
type = TYPE[form] || "unk_type_#{form}"
|
104
|
+
if type == 'func'
|
105
|
+
type = { :params => [], :ret => [] }
|
106
|
+
decode_uleb(edata).times {
|
107
|
+
type[:params] << decode_type(edata)
|
108
|
+
}
|
109
|
+
decode_uleb(edata).times {
|
110
|
+
type[:ret] << decode_type(edata)
|
111
|
+
}
|
112
|
+
end
|
113
|
+
type
|
114
|
+
end
|
115
|
+
|
116
|
+
# return the nth global
|
117
|
+
# use the @global array and the @import array
|
118
|
+
def get_global_nr(nr)
|
119
|
+
glob_imports = @import.to_a.find_all { |i| i[:kind] == 'global' }
|
120
|
+
return glob_imports[nr] if nr < glob_imports.length
|
121
|
+
nr -= glob_imports.length
|
122
|
+
@global[nr]
|
123
|
+
end
|
124
|
+
|
125
|
+
# return the nth function body
|
126
|
+
# use the @function_body array and the @import array
|
127
|
+
def get_function_nr(nr)
|
128
|
+
func_imports = @import.to_a.find_all { |i| i[:kind] == 'function' }
|
129
|
+
return func_imports[nr] if nr < func_imports.length
|
130
|
+
nr -= func_imports.length
|
131
|
+
@function_body[nr]
|
132
|
+
end
|
133
|
+
|
134
|
+
def type_to_s(t)
|
135
|
+
return t unless t.kind_of?(::Hash)
|
136
|
+
(t[:ret].map { |tt| type_to_s(tt) }.join(', ') << ' f(' << t[:params].map { |tt| type_to_s(tt) }.join(', ') << ')').strip
|
137
|
+
end
|
138
|
+
|
139
|
+
def decode_limits(edata=@encoded)
|
140
|
+
flags = decode_uleb(edata)
|
141
|
+
out = { :initial_size => decode_uleb(edata) }
|
142
|
+
out[:maximum] = decode_uleb(edata) if flags & 1
|
143
|
+
out
|
144
|
+
end
|
145
|
+
|
146
|
+
# wtf
|
147
|
+
# read wasm bytecode until reaching the end opcode
|
148
|
+
# return the byte offset
|
149
|
+
def read_code_until_end(m=nil)
|
150
|
+
if m
|
151
|
+
raw_offset = m.raw_offset + m.edata.ptr
|
152
|
+
edata = m.edata
|
153
|
+
else
|
154
|
+
edata = @encoded
|
155
|
+
end
|
156
|
+
|
157
|
+
while op = edata.decode_imm(:u8, @endianness)
|
158
|
+
case op
|
159
|
+
when 0xb
|
160
|
+
# end opcode
|
161
|
+
return raw_offset
|
162
|
+
when 0xe
|
163
|
+
# indirect branch wtf
|
164
|
+
decode_uleb(edata).times { decode_uleb(edata) }
|
165
|
+
decode_uleb(edata)
|
166
|
+
when 0x43
|
167
|
+
edata.read(4)
|
168
|
+
when 0x44
|
169
|
+
edata.read(8)
|
170
|
+
else
|
171
|
+
OPCODE_IMM_COUNT[op].times { decode_uleb(edata) }
|
172
|
+
end
|
173
|
+
end
|
174
|
+
raw_offset
|
175
|
+
end
|
176
|
+
|
177
|
+
def decode_header
|
178
|
+
@header = Header.decode(self)
|
179
|
+
@modules = []
|
180
|
+
end
|
181
|
+
|
182
|
+
def decode
|
183
|
+
decode_header
|
184
|
+
while @encoded.ptr < @encoded.length
|
185
|
+
@modules << Module.decode(self)
|
186
|
+
end
|
187
|
+
@modules.each { |m|
|
188
|
+
@encoded.add_export(new_label("module_#{m.id}"), m.raw_offset)
|
189
|
+
f = "decode_module_#{m.id.to_s.downcase}"
|
190
|
+
send(f, m) if respond_to?(f)
|
191
|
+
}
|
192
|
+
func_imports = @import.to_a.find_all { |i| i[:kind] == 'function' }
|
193
|
+
export.to_a.each { |e|
|
194
|
+
next if e[:kind] != 'function' # TODO resolve init_offset for globals etc?
|
195
|
+
idx = e[:index] - func_imports.length
|
196
|
+
next if not fb = function_body.to_a[idx]
|
197
|
+
@encoded.add_export(new_label(e[:field]), fb[:init_offset], true)
|
198
|
+
}
|
199
|
+
# bytecode start addr => { :local_var => [], :params => [], :ret => [] }
|
200
|
+
# :local_var absent for external code (imported funcs)
|
201
|
+
@code_info = {}
|
202
|
+
import.to_a.each { |i|
|
203
|
+
next unless i[:kind] == 'function'
|
204
|
+
@code_info["#{i[:module]}_#{i[:field]}"] = { :params => i[:type][:params], :ret => i[:type][:ret] }
|
205
|
+
}
|
206
|
+
function_body.to_a.each { |fb|
|
207
|
+
@code_info[fb[:init_offset]] = { :local_var => fb[:local_var], :params => fb[:type][:params], :ret => fb[:type][:ret] }
|
208
|
+
}
|
209
|
+
global_idx = import.to_a.find_all { |i| i[:kind] == 'global' }.length - 1
|
210
|
+
global.to_a.each { |g|
|
211
|
+
@code_info[g[:init_offset]] = { :local_var => [], :params => [], :ret => [g[:type]] }
|
212
|
+
@encoded.add_export new_label("global_#{global_idx += 1}_init"), g[:init_offset]
|
213
|
+
}
|
214
|
+
element.to_a.each { |e|
|
215
|
+
@code_info[e[:init_offset]] = { :local_var => [], :params => [], :ret => ['i32'] }
|
216
|
+
}
|
217
|
+
data.to_a.each { |d|
|
218
|
+
@code_info[d[:init_offset]] = { :local_var => [], :params => [], :ret => ['i32'] }
|
219
|
+
}
|
220
|
+
end
|
221
|
+
|
222
|
+
def decode_module_type(m)
|
223
|
+
@type = []
|
224
|
+
decode_uleb(m.edata).times {
|
225
|
+
@type << decode_type(m.edata)
|
226
|
+
}
|
227
|
+
end
|
228
|
+
|
229
|
+
def decode_module_import(m)
|
230
|
+
@import = []
|
231
|
+
decode_uleb(m.edata).times {
|
232
|
+
i = {}
|
233
|
+
i[:module] = m.edata.read(decode_uleb(m.edata))
|
234
|
+
i[:field] = m.edata.read(decode_uleb(m.edata))
|
235
|
+
kind = decode_uleb(m.edata)
|
236
|
+
i[:kind] = EXTERNAL_KIND[kind] || kind
|
237
|
+
case i[:kind]
|
238
|
+
when 'function'
|
239
|
+
i[:type] = @type[decode_uleb(m.edata)] # XXX keep index only, in case @type is not yet known ?
|
240
|
+
when 'table'
|
241
|
+
i[:type] = decode_type(m.edata)
|
242
|
+
i[:limits] = decode_limits(m.edata)
|
243
|
+
when 'memory'
|
244
|
+
i[:limits] = decode_limits(m.edata)
|
245
|
+
when 'global'
|
246
|
+
i[:type] = decode_type(m.edata)
|
247
|
+
i[:mutable] = decode_uleb(m.edata)
|
248
|
+
end
|
249
|
+
@import << i
|
250
|
+
}
|
251
|
+
end
|
252
|
+
|
253
|
+
def decode_module_function(m)
|
254
|
+
@function_signature = []
|
255
|
+
idx = 0
|
256
|
+
decode_uleb(m.edata).times {
|
257
|
+
@function_signature << @type[decode_uleb(m.edata)]
|
258
|
+
@function_body[idx][:type] = @function_signature[idx] if function_body
|
259
|
+
idx += 1
|
260
|
+
}
|
261
|
+
end
|
262
|
+
|
263
|
+
def decode_module_table(m)
|
264
|
+
@table = []
|
265
|
+
decode_uleb(m.edata).times {
|
266
|
+
@table << { :type => decode_type(m.edata), :limits => decode_limits(m.edata) }
|
267
|
+
}
|
268
|
+
end
|
269
|
+
|
270
|
+
def decode_module_memory(m)
|
271
|
+
@memory = []
|
272
|
+
decode_uleb(m.edata).times {
|
273
|
+
@memory << { :limits => decode_limits(m.edata) }
|
274
|
+
}
|
275
|
+
end
|
276
|
+
|
277
|
+
def decode_module_global(m)
|
278
|
+
@global = []
|
279
|
+
decode_uleb(m.edata).times {
|
280
|
+
@global << { :type => decode_type(m.edata), :mutable => decode_uleb(m.edata), :init_offset => read_code_until_end(m) }
|
281
|
+
}
|
282
|
+
end
|
283
|
+
|
284
|
+
def decode_module_export(m)
|
285
|
+
@export = []
|
286
|
+
decode_uleb(m.edata).times {
|
287
|
+
flen = decode_uleb(m.edata)
|
288
|
+
fld = m.edata.read(flen)
|
289
|
+
kind = decode_uleb(m.edata)
|
290
|
+
kind = EXTERNAL_KIND[kind] || kind
|
291
|
+
index = decode_uleb(m.edata)
|
292
|
+
@export << { :field => fld, :kind => kind, :index => index }
|
293
|
+
}
|
294
|
+
end
|
295
|
+
|
296
|
+
def decode_module_start(m)
|
297
|
+
@start_function_index = decode_uleb(m.edata)
|
298
|
+
end
|
299
|
+
|
300
|
+
def decode_module_element(m)
|
301
|
+
@element = []
|
302
|
+
decode_uleb(m.edata).times {
|
303
|
+
seg = { :table_index => decode_uleb(m.edata),
|
304
|
+
:init_offset => read_code_until_end(m),
|
305
|
+
:elems => [] }
|
306
|
+
decode_uleb(m.edata).times {
|
307
|
+
seg[:elems] << decode_uleb(m.edata)
|
308
|
+
}
|
309
|
+
@element << seg
|
310
|
+
@encoded.add_export new_label("element_#{@element.length-1}_init_addr"), @element.last[:init_offset]
|
311
|
+
}
|
312
|
+
end
|
313
|
+
|
314
|
+
def decode_module_code(m)
|
315
|
+
@function_body = []
|
316
|
+
idx = 0
|
317
|
+
decode_uleb(m.edata).times {
|
318
|
+
local_vars = []
|
319
|
+
body_size = decode_uleb(m.edata) # size of local defs + bytecode (in bytes)
|
320
|
+
next_ptr = m.edata.ptr + body_size
|
321
|
+
decode_uleb(m.edata).times { # nr of local vars types
|
322
|
+
n_vars_of_this_type = decode_uleb(m.edata) # nr of local vars of this type
|
323
|
+
type = decode_type(m.edata) # actual type
|
324
|
+
n_vars_of_this_type.times {
|
325
|
+
local_vars << type
|
326
|
+
}
|
327
|
+
}
|
328
|
+
code_offset = m.raw_offset + m.edata.ptr # bytecode comes next
|
329
|
+
m.edata.ptr = next_ptr
|
330
|
+
@function_body << { :local_var => local_vars, :init_offset => code_offset }
|
331
|
+
@function_body.last[:type] = @function_signature[idx] if function_signature
|
332
|
+
@encoded.add_export new_label("function_#{@function_body.length-1}"), @function_body.last[:init_offset]
|
333
|
+
idx += 1
|
334
|
+
}
|
335
|
+
end
|
336
|
+
|
337
|
+
def decode_module_data(m)
|
338
|
+
@data = []
|
339
|
+
decode_uleb(m.edata).times {
|
340
|
+
idx = decode_uleb(m.edata)
|
341
|
+
initoff = read_code_until_end(m)
|
342
|
+
data_len = decode_uleb(m.edata)
|
343
|
+
data_start_ptr = m.raw_offset + m.edata.ptr
|
344
|
+
data = m.edata.read(data_len)
|
345
|
+
data_end_ptr = m.raw_offset + m.edata.ptr
|
346
|
+
|
347
|
+
@data << { :index => idx, :init_offset => initoff, :data => data }
|
348
|
+
@encoded.add_export new_label("data_#{@data.length-1}_init_addr"), initoff
|
349
|
+
@encoded.add_export new_label("data_#{@data.length-1}_start"), data_start_ptr
|
350
|
+
@encoded.add_export new_label("data_#{@data.length-1}_end"), data_end_ptr
|
351
|
+
}
|
352
|
+
end
|
353
|
+
|
354
|
+
def decode_module_0(m)
|
355
|
+
# id == 0 for not well-known modules
|
356
|
+
# the module name is encoded at start of payload (uleb name length + actual name)
|
357
|
+
m.name = m.edata.read(decode_uleb(m.edata))
|
358
|
+
f = "decode_module_0_#{m.name.downcase}"
|
359
|
+
send(f, m) if respond_to?(f)
|
360
|
+
end
|
361
|
+
|
362
|
+
def decode_module_0_name(m)
|
363
|
+
# TODO parse stored names of local variables etc
|
364
|
+
end
|
365
|
+
|
366
|
+
def cpu_from_headers
|
367
|
+
WebAsm.new(self)
|
368
|
+
end
|
369
|
+
|
370
|
+
def init_disassembler
|
371
|
+
dasm = super()
|
372
|
+
function_body.to_a.each { |fb|
|
373
|
+
v = []
|
374
|
+
fb[:local_var].map { |lv| type_to_s(lv) }.each { |lv|
|
375
|
+
v.last && lv == v.last.last ? v.last << lv : v << [lv]
|
376
|
+
}
|
377
|
+
v.map! { |sublist|
|
378
|
+
# i32 ; i32 ; i32 ; i32 ; i32 ; i32 ; i64 -> 5 * i32 ; i64
|
379
|
+
sublist.length > 3 ? "#{sublist.length} * #{sublist.first}" : sublist.join(' ; ')
|
380
|
+
}
|
381
|
+
dasm.add_comment fb[:init_offset], "proto: #{fb[:type] ? type_to_s(fb[:type]) : 'unknown'}"
|
382
|
+
dasm.add_comment fb[:init_offset], "vars: #{v.join(' ; ')}"
|
383
|
+
}
|
384
|
+
global.to_a.each { |g|
|
385
|
+
dasm.add_comment g[:init_offset], "type: #{type_to_s(g[:type])}"
|
386
|
+
}
|
387
|
+
dasm.function[:default] = @cpu.disassembler_default_func
|
388
|
+
dasm
|
389
|
+
end
|
390
|
+
|
391
|
+
def each_section
|
392
|
+
yield @encoded, 0
|
393
|
+
end
|
394
|
+
|
395
|
+
def get_default_entrypoints
|
396
|
+
global.to_a.map { |g| g[:init_offset] } +
|
397
|
+
element.to_a.map { |e| e[:init_offset] } +
|
398
|
+
data.to_a.map { |d| d[:init_offset] } +
|
399
|
+
function_body.to_a.map { |f| f[:init_offset] }
|
400
|
+
end
|
401
|
+
end
|
402
|
+
end
|