metasm 1.0.1 → 1.0.2

Sign up to get free protection for your applications and to get access to all the features.
Files changed (235) hide show
  1. checksums.yaml +7 -0
  2. data/.gitignore +1 -0
  3. data/.hgtags +3 -0
  4. data/Gemfile +1 -0
  5. data/INSTALL +61 -0
  6. data/LICENCE +458 -0
  7. data/README +29 -21
  8. data/Rakefile +10 -0
  9. data/TODO +10 -12
  10. data/doc/code_organisation.txt +2 -0
  11. data/doc/core/DynLdr.txt +247 -0
  12. data/doc/core/ExeFormat.txt +43 -0
  13. data/doc/core/Expression.txt +220 -0
  14. data/doc/core/GNUExports.txt +27 -0
  15. data/doc/core/Ia32.txt +236 -0
  16. data/doc/core/SerialStruct.txt +108 -0
  17. data/doc/core/VirtualString.txt +145 -0
  18. data/doc/core/WindowsExports.txt +61 -0
  19. data/doc/core/index.txt +1 -0
  20. data/doc/style.css +6 -3
  21. data/doc/usage/debugger.txt +327 -0
  22. data/doc/usage/index.txt +1 -0
  23. data/doc/use_cases.txt +2 -2
  24. data/metasm.gemspec +22 -0
  25. data/{lib/metasm.rb → metasm.rb} +11 -3
  26. data/{lib/metasm → metasm}/compile_c.rb +13 -7
  27. data/metasm/cpu/arc.rb +8 -0
  28. data/metasm/cpu/arc/decode.rb +425 -0
  29. data/metasm/cpu/arc/main.rb +191 -0
  30. data/metasm/cpu/arc/opcodes.rb +588 -0
  31. data/{lib/metasm → metasm/cpu}/arm.rb +7 -5
  32. data/{lib/metasm → metasm/cpu}/arm/debug.rb +2 -2
  33. data/{lib/metasm → metasm/cpu}/arm/decode.rb +13 -12
  34. data/{lib/metasm → metasm/cpu}/arm/encode.rb +23 -8
  35. data/{lib/metasm → metasm/cpu}/arm/main.rb +0 -3
  36. data/metasm/cpu/arm/opcodes.rb +324 -0
  37. data/{lib/metasm → metasm/cpu}/arm/parse.rb +25 -13
  38. data/{lib/metasm → metasm/cpu}/arm/render.rb +2 -2
  39. data/metasm/cpu/arm64.rb +15 -0
  40. data/metasm/cpu/arm64/debug.rb +38 -0
  41. data/metasm/cpu/arm64/decode.rb +289 -0
  42. data/metasm/cpu/arm64/encode.rb +41 -0
  43. data/metasm/cpu/arm64/main.rb +105 -0
  44. data/metasm/cpu/arm64/opcodes.rb +232 -0
  45. data/metasm/cpu/arm64/parse.rb +20 -0
  46. data/metasm/cpu/arm64/render.rb +95 -0
  47. data/{lib/metasm/ppc.rb → metasm/cpu/bpf.rb} +2 -4
  48. data/metasm/cpu/bpf/decode.rb +142 -0
  49. data/metasm/cpu/bpf/main.rb +60 -0
  50. data/metasm/cpu/bpf/opcodes.rb +81 -0
  51. data/metasm/cpu/bpf/render.rb +41 -0
  52. data/metasm/cpu/cy16.rb +9 -0
  53. data/metasm/cpu/cy16/decode.rb +253 -0
  54. data/metasm/cpu/cy16/main.rb +63 -0
  55. data/metasm/cpu/cy16/opcodes.rb +78 -0
  56. data/metasm/cpu/cy16/render.rb +41 -0
  57. data/metasm/cpu/dalvik.rb +11 -0
  58. data/{lib/metasm → metasm/cpu}/dalvik/decode.rb +35 -13
  59. data/{lib/metasm → metasm/cpu}/dalvik/main.rb +51 -2
  60. data/{lib/metasm → metasm/cpu}/dalvik/opcodes.rb +19 -11
  61. data/metasm/cpu/ia32.rb +17 -0
  62. data/{lib/metasm → metasm/cpu}/ia32/compile_c.rb +5 -7
  63. data/{lib/metasm → metasm/cpu}/ia32/debug.rb +5 -5
  64. data/{lib/metasm → metasm/cpu}/ia32/decode.rb +246 -59
  65. data/{lib/metasm → metasm/cpu}/ia32/decompile.rb +7 -7
  66. data/{lib/metasm → metasm/cpu}/ia32/encode.rb +19 -13
  67. data/{lib/metasm → metasm/cpu}/ia32/main.rb +51 -8
  68. data/metasm/cpu/ia32/opcodes.rb +1424 -0
  69. data/{lib/metasm → metasm/cpu}/ia32/parse.rb +47 -16
  70. data/{lib/metasm → metasm/cpu}/ia32/render.rb +31 -4
  71. data/metasm/cpu/mips.rb +14 -0
  72. data/{lib/metasm → metasm/cpu}/mips/compile_c.rb +1 -1
  73. data/metasm/cpu/mips/debug.rb +42 -0
  74. data/{lib/metasm → metasm/cpu}/mips/decode.rb +46 -16
  75. data/{lib/metasm → metasm/cpu}/mips/encode.rb +4 -3
  76. data/{lib/metasm → metasm/cpu}/mips/main.rb +11 -4
  77. data/{lib/metasm → metasm/cpu}/mips/opcodes.rb +86 -17
  78. data/{lib/metasm → metasm/cpu}/mips/parse.rb +1 -1
  79. data/{lib/metasm → metasm/cpu}/mips/render.rb +1 -1
  80. data/{lib/metasm/dalvik.rb → metasm/cpu/msp430.rb} +1 -1
  81. data/metasm/cpu/msp430/decode.rb +247 -0
  82. data/metasm/cpu/msp430/main.rb +62 -0
  83. data/metasm/cpu/msp430/opcodes.rb +101 -0
  84. data/{lib/metasm → metasm/cpu}/pic16c/decode.rb +6 -7
  85. data/{lib/metasm → metasm/cpu}/pic16c/main.rb +0 -0
  86. data/{lib/metasm → metasm/cpu}/pic16c/opcodes.rb +1 -1
  87. data/{lib/metasm/mips.rb → metasm/cpu/ppc.rb} +4 -4
  88. data/{lib/metasm → metasm/cpu}/ppc/decode.rb +18 -12
  89. data/{lib/metasm → metasm/cpu}/ppc/decompile.rb +3 -3
  90. data/{lib/metasm → metasm/cpu}/ppc/encode.rb +2 -2
  91. data/{lib/metasm → metasm/cpu}/ppc/main.rb +17 -12
  92. data/{lib/metasm → metasm/cpu}/ppc/opcodes.rb +11 -5
  93. data/metasm/cpu/ppc/parse.rb +55 -0
  94. data/metasm/cpu/python.rb +8 -0
  95. data/metasm/cpu/python/decode.rb +136 -0
  96. data/metasm/cpu/python/main.rb +36 -0
  97. data/metasm/cpu/python/opcodes.rb +180 -0
  98. data/{lib/metasm → metasm/cpu}/sh4.rb +1 -1
  99. data/{lib/metasm → metasm/cpu}/sh4/decode.rb +48 -17
  100. data/{lib/metasm → metasm/cpu}/sh4/main.rb +13 -4
  101. data/{lib/metasm → metasm/cpu}/sh4/opcodes.rb +7 -8
  102. data/metasm/cpu/x86_64.rb +15 -0
  103. data/{lib/metasm → metasm/cpu}/x86_64/compile_c.rb +28 -17
  104. data/{lib/metasm → metasm/cpu}/x86_64/debug.rb +4 -4
  105. data/{lib/metasm → metasm/cpu}/x86_64/decode.rb +57 -15
  106. data/{lib/metasm → metasm/cpu}/x86_64/encode.rb +55 -26
  107. data/{lib/metasm → metasm/cpu}/x86_64/main.rb +14 -6
  108. data/metasm/cpu/x86_64/opcodes.rb +136 -0
  109. data/{lib/metasm → metasm/cpu}/x86_64/parse.rb +10 -2
  110. data/metasm/cpu/x86_64/render.rb +35 -0
  111. data/metasm/cpu/z80.rb +9 -0
  112. data/metasm/cpu/z80/decode.rb +313 -0
  113. data/metasm/cpu/z80/main.rb +67 -0
  114. data/metasm/cpu/z80/opcodes.rb +224 -0
  115. data/metasm/cpu/z80/render.rb +59 -0
  116. data/{lib/metasm/os/main.rb → metasm/debug.rb} +160 -401
  117. data/{lib/metasm → metasm}/decode.rb +35 -4
  118. data/{lib/metasm → metasm}/decompile.rb +15 -16
  119. data/{lib/metasm → metasm}/disassemble.rb +201 -45
  120. data/{lib/metasm → metasm}/disassemble_api.rb +651 -87
  121. data/{lib/metasm → metasm}/dynldr.rb +220 -133
  122. data/{lib/metasm → metasm}/encode.rb +10 -1
  123. data/{lib/metasm → metasm}/exe_format/a_out.rb +9 -6
  124. data/{lib/metasm → metasm}/exe_format/autoexe.rb +1 -0
  125. data/{lib/metasm → metasm}/exe_format/bflt.rb +57 -27
  126. data/{lib/metasm → metasm}/exe_format/coff.rb +11 -3
  127. data/{lib/metasm → metasm}/exe_format/coff_decode.rb +53 -20
  128. data/{lib/metasm → metasm}/exe_format/coff_encode.rb +11 -13
  129. data/{lib/metasm → metasm}/exe_format/dex.rb +13 -5
  130. data/{lib/metasm → metasm}/exe_format/dol.rb +1 -0
  131. data/{lib/metasm → metasm}/exe_format/elf.rb +93 -57
  132. data/{lib/metasm → metasm}/exe_format/elf_decode.rb +143 -34
  133. data/{lib/metasm → metasm}/exe_format/elf_encode.rb +122 -31
  134. data/metasm/exe_format/gb.rb +65 -0
  135. data/metasm/exe_format/javaclass.rb +424 -0
  136. data/{lib/metasm → metasm}/exe_format/macho.rb +204 -16
  137. data/{lib/metasm → metasm}/exe_format/main.rb +26 -3
  138. data/{lib/metasm → metasm}/exe_format/mz.rb +1 -0
  139. data/{lib/metasm → metasm}/exe_format/nds.rb +7 -4
  140. data/{lib/metasm → metasm}/exe_format/pe.rb +71 -8
  141. data/metasm/exe_format/pyc.rb +167 -0
  142. data/{lib/metasm → metasm}/exe_format/serialstruct.rb +67 -14
  143. data/{lib/metasm → metasm}/exe_format/shellcode.rb +7 -3
  144. data/metasm/exe_format/shellcode_rwx.rb +114 -0
  145. data/metasm/exe_format/swf.rb +205 -0
  146. data/{lib/metasm → metasm}/exe_format/xcoff.rb +7 -7
  147. data/metasm/exe_format/zip.rb +335 -0
  148. data/metasm/gui.rb +13 -0
  149. data/{lib/metasm → metasm}/gui/cstruct.rb +35 -41
  150. data/{lib/metasm → metasm}/gui/dasm_coverage.rb +11 -11
  151. data/{lib/metasm → metasm}/gui/dasm_decomp.rb +7 -20
  152. data/{lib/metasm → metasm}/gui/dasm_funcgraph.rb +0 -0
  153. data/metasm/gui/dasm_graph.rb +1695 -0
  154. data/{lib/metasm → metasm}/gui/dasm_hex.rb +12 -8
  155. data/{lib/metasm → metasm}/gui/dasm_listing.rb +43 -28
  156. data/{lib/metasm → metasm}/gui/dasm_main.rb +310 -53
  157. data/{lib/metasm → metasm}/gui/dasm_opcodes.rb +5 -19
  158. data/{lib/metasm → metasm}/gui/debug.rb +93 -27
  159. data/{lib/metasm → metasm}/gui/gtk.rb +162 -40
  160. data/{lib/metasm → metasm}/gui/qt.rb +12 -2
  161. data/{lib/metasm → metasm}/gui/win32.rb +179 -42
  162. data/{lib/metasm → metasm}/gui/x11.rb +59 -59
  163. data/{lib/metasm → metasm}/main.rb +389 -264
  164. data/{lib/metasm/os/remote.rb → metasm/os/gdbremote.rb} +146 -54
  165. data/{lib/metasm → metasm}/os/gnu_exports.rb +1 -1
  166. data/{lib/metasm → metasm}/os/linux.rb +628 -151
  167. data/metasm/os/main.rb +330 -0
  168. data/{lib/metasm → metasm}/os/windows.rb +132 -42
  169. data/{lib/metasm → metasm}/os/windows_exports.rb +141 -0
  170. data/{lib/metasm → metasm}/parse.rb +26 -24
  171. data/{lib/metasm → metasm}/parse_c.rb +221 -116
  172. data/{lib/metasm → metasm}/preprocessor.rb +55 -40
  173. data/{lib/metasm → metasm}/render.rb +14 -38
  174. data/misc/hexdump.rb +2 -1
  175. data/misc/lint.rb +58 -0
  176. data/misc/txt2html.rb +9 -7
  177. data/samples/bindiff.rb +3 -4
  178. data/samples/dasm-plugins/bindiff.rb +15 -0
  179. data/samples/dasm-plugins/bookmark.rb +133 -0
  180. data/samples/dasm-plugins/c_constants.rb +57 -0
  181. data/samples/dasm-plugins/colortheme_solarized.rb +125 -0
  182. data/samples/dasm-plugins/cppobj_funcall.rb +60 -0
  183. data/samples/dasm-plugins/dasm_all.rb +70 -0
  184. data/samples/dasm-plugins/demangle_cpp.rb +31 -0
  185. data/samples/dasm-plugins/deobfuscate.rb +251 -0
  186. data/samples/dasm-plugins/dump_text.rb +35 -0
  187. data/samples/dasm-plugins/export_graph_svg.rb +86 -0
  188. data/samples/dasm-plugins/findgadget.rb +75 -0
  189. data/samples/dasm-plugins/hl_opcode.rb +32 -0
  190. data/samples/dasm-plugins/hotfix_gtk_dbg.rb +19 -0
  191. data/samples/dasm-plugins/imm2off.rb +34 -0
  192. data/samples/dasm-plugins/match_libsigs.rb +93 -0
  193. data/samples/dasm-plugins/patch_file.rb +95 -0
  194. data/samples/dasm-plugins/scanfuncstart.rb +36 -0
  195. data/samples/dasm-plugins/scanxrefs.rb +26 -0
  196. data/samples/dasm-plugins/selfmodify.rb +197 -0
  197. data/samples/dasm-plugins/stringsxrefs.rb +28 -0
  198. data/samples/dasmnavig.rb +1 -1
  199. data/samples/dbg-apihook.rb +24 -9
  200. data/samples/dbg-plugins/heapscan.rb +283 -0
  201. data/samples/dbg-plugins/heapscan/compiled_heapscan_lin.c +155 -0
  202. data/samples/dbg-plugins/heapscan/compiled_heapscan_win.c +128 -0
  203. data/samples/dbg-plugins/heapscan/graphheap.rb +616 -0
  204. data/samples/dbg-plugins/heapscan/heapscan.rb +709 -0
  205. data/samples/dbg-plugins/heapscan/winheap.h +174 -0
  206. data/samples/dbg-plugins/heapscan/winheap7.h +307 -0
  207. data/samples/dbg-plugins/trace_func.rb +214 -0
  208. data/samples/disassemble-gui.rb +35 -5
  209. data/samples/disassemble.rb +31 -6
  210. data/samples/dump_upx.rb +24 -12
  211. data/samples/dynamic_ruby.rb +12 -3
  212. data/samples/exeencode.rb +6 -5
  213. data/samples/factorize-headers-peimports.rb +1 -1
  214. data/samples/lindebug.rb +175 -381
  215. data/samples/metasm-shell.rb +1 -2
  216. data/samples/peldr.rb +2 -2
  217. data/tests/all.rb +1 -1
  218. data/tests/arc.rb +26 -0
  219. data/tests/dynldr.rb +22 -4
  220. data/tests/expression.rb +55 -0
  221. data/tests/graph_layout.rb +285 -0
  222. data/tests/ia32.rb +79 -26
  223. data/tests/mips.rb +9 -2
  224. data/tests/x86_64.rb +66 -18
  225. metadata +330 -218
  226. data/lib/metasm/arm/opcodes.rb +0 -177
  227. data/lib/metasm/gui.rb +0 -23
  228. data/lib/metasm/gui/dasm_graph.rb +0 -1354
  229. data/lib/metasm/ia32.rb +0 -14
  230. data/lib/metasm/ia32/opcodes.rb +0 -873
  231. data/lib/metasm/ppc/parse.rb +0 -52
  232. data/lib/metasm/x86_64.rb +0 -12
  233. data/lib/metasm/x86_64/opcodes.rb +0 -118
  234. data/samples/gdbclient.rb +0 -583
  235. data/samples/rubstop.rb +0 -399
@@ -0,0 +1,709 @@
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
+ module Metasm
8
+ class Heap
9
+ attr_accessor :vm, :range, :ptsz
10
+ attr_accessor :cp
11
+ # hash chunk userdata pointer -> chunk userdata size
12
+ attr_accessor :chunks
13
+ # hash chunk user pointer -> C::Struct
14
+ attr_accessor :chunk_struct
15
+ # the chunk graph: chunk pointer -> [array of chunks addrs pointed]
16
+ attr_accessor :xrchunksto, :xrchunksfrom
17
+
18
+ def initialize(dbg)
19
+ @dbg = dbg
20
+ @dbg.pid_stuff_list << :heap
21
+ @dbg.heap = self
22
+ @range = {}
23
+ @dwcache = {}
24
+ # userdata_ptr => len
25
+ @chunks = {}
26
+ @xrchunksto = {}
27
+ @xrchunksfrom = {}
28
+ @ptsz = dbg.cpu.size/8
29
+ # ptr => C::Struct
30
+ @chunk_struct = {}
31
+ end
32
+
33
+ def pagecache(base, len)
34
+ @dbg.read_mapped_range(base, len)
35
+ end
36
+
37
+ def dwcache(base, len)
38
+ @dwcache[[base, len]] ||= pagecache(base, len).unpack(@ptsz == 4 ? 'L*' : 'Q*')
39
+ end
40
+
41
+ # return the array of dwords in the chunk
42
+ def chunkdw(ptr, len=@chunks[ptr])
43
+ if base = find_range(ptr)
44
+ dwcache(base, @range[base])[(ptr-base)/@ptsz, len/@ptsz]
45
+ end
46
+ end
47
+
48
+ # returns the list of chunks, sorted
49
+ def chunklist
50
+ @chunklist ||= @chunks.keys.sort
51
+ end
52
+
53
+ # dichotomic search of the chunk containing ptr
54
+ # len = hash ptr => length
55
+ # list = list of hash keys sorted
56
+ def find_elem(ptr, len, list=nil)
57
+ return ptr if len[ptr]
58
+
59
+ list ||= len.keys.sort
60
+
61
+ if list.length < 16
62
+ return list.find { |p| p <= ptr and p + len[p] > ptr }
63
+ end
64
+
65
+ window = list
66
+ while window and not window.empty?
67
+ i = window.length/2
68
+ wi = window[i]
69
+ if ptr < wi
70
+ window = window[0, i]
71
+ elsif ptr < wi + len[wi]
72
+ return wi
73
+ else
74
+ window = window[i+1, i]
75
+ end
76
+ end
77
+ end
78
+
79
+ # find the chunk encompassing ptr
80
+ def find_chunk(ptr)
81
+ find_elem(ptr, @chunks, chunklist)
82
+ end
83
+
84
+ def find_range(ptr)
85
+ @range_keys ||= @range.keys.sort
86
+ find_elem(ptr, @range, @range_keys)
87
+ end
88
+
89
+ # { chunk size => [list of chunk addrs] }
90
+ attr_accessor :buckets
91
+ def bucketize
92
+ @buckets = {}
93
+ chunklist.each { |a|
94
+ (@buckets[@chunks[a]] ||= []) << a
95
+ }
96
+ end
97
+
98
+ # find the kernels of the graph (strongly connected components)
99
+ # must be called after scan_xr
100
+ # also find the graph diameter
101
+ attr_accessor :kernels, :maxpath
102
+ def find_kernels(adj = @xrchunksto)
103
+ @maxpath = []
104
+ kernels = {}
105
+ adj.keys.sort.each { |ptr|
106
+ next if kernels[ptr]
107
+ paths = [[ptr]]
108
+ while path = paths.pop
109
+ next if not l = @xrchunksfrom[path.first]
110
+ l.each { |pl|
111
+ next if kernels[pl]
112
+ next if not adj[pl]
113
+ if path.include?(pl)
114
+ kernels[pl] = true
115
+ else
116
+ paths << [pl, *path]
117
+ end
118
+ }
119
+ @maxpath = paths.last if paths.last and paths.last.length > @maxpath.length
120
+ end
121
+ }
122
+ if @maxpath.first and np = (adj[@maxpath.last] - @maxpath).first
123
+ @maxpath << np
124
+ end
125
+ @kernels = []
126
+ while k = kernels.index(true)
127
+ curk = reachfrom(k, adj).find_all { |ok|
128
+ true == reachfrom(ok, adj) { |tk|
129
+ break true if tk == k
130
+ }
131
+ }
132
+ @kernels << curk
133
+ curk.each { |ka| kernels.delete ka }
134
+ end
135
+ end
136
+
137
+ attr_accessor :roots
138
+ # find the root nodes that allow acces to most other nodes
139
+ # { root => [reachable nodes] }
140
+ # does not include single nodes (@chunks.keys - @xrchunksfrom.keys)
141
+ def find_roots(adj=@xrchunksto)
142
+ @roots = {}
143
+ adj.keys.sort.each { |r|
144
+ if not @roots[r]
145
+ l = reachfrom(r, adj, @roots)
146
+ l.each { |t| @roots[t] = true if adj[t] } # may include r !, also dont mark leaves
147
+ @roots[r] = l
148
+ end
149
+ }
150
+ @roots.delete_if { |k, v| v == true }
151
+ end
152
+
153
+ def reachfrom(p, adj = @xrchunksto, roots={})
154
+ return roots[p] if roots[p].kind_of? Array
155
+ hash = {}
156
+ todo = [p]
157
+ while p = todo.pop
158
+ if to = roots[p] || adj[p] and to.kind_of? Array
159
+ to.each { |tk|
160
+ if not hash[tk]
161
+ hash[tk] = true
162
+ todo << tk
163
+ yield tk if block_given?
164
+ end
165
+ }
166
+ end
167
+ end
168
+ hash.keys
169
+ end
170
+
171
+ # create a subset of xrchunksto from one point
172
+ def graph_from(p, adj = @xrchunksto)
173
+ hash = {}
174
+ todo = [p]
175
+ while p = todo.pop
176
+ next if hash[p]
177
+ if adj[p]
178
+ hash[p] = adj[p]
179
+ todo.concat hash[p]
180
+ end
181
+ end
182
+ hash
183
+ end
184
+
185
+ # dump the whole graph in a dot file
186
+ def dump_graph(fname='graph.gv', graph=@xrchunksto)
187
+ File.open(fname, 'w') { |fd|
188
+ fd.puts "digraph foo {"
189
+ graph.each { |b, l|
190
+ fd.puts l.map { |e| '"%x" -> "%x";' % [b, e] }
191
+ }
192
+ fd.puts "}"
193
+ }
194
+ end
195
+
196
+ # chunk ptr => { dwindex => [list of ptrs] }
197
+ attr_accessor :linkedlists, :alllists
198
+ def find_linkedlists
199
+ @linkedlists = {}
200
+ @alllists = []
201
+ @buckets.sort.each { |sz, lst|
202
+ #puts "sz #{sz} #{lst.length}"
203
+ lst.each { |ptr|
204
+ next if not l = @xrchunksto[ptr]
205
+ next if not l.find { |tg| @chunks[tg] == sz }
206
+ dw = chunkdw(ptr)
207
+ dw.length.times { |dwoff|
208
+ next if @linkedlists[ptr] and @linkedlists[ptr][dwoff]
209
+ tg = dw[dwoff]
210
+ next if @chunks[tg] != sz
211
+ check_linkedlist(ptr, dwoff)
212
+ }
213
+ }
214
+ }
215
+ end
216
+
217
+ def check_linkedlist(ptr, dwoff)
218
+ psz = @chunks[ptr]
219
+ fwd = ptr
220
+ lst = [fwd]
221
+ base = find_range(fwd)
222
+ loop do
223
+ if not base or base > fwd or base + @range[base] <= fwd
224
+ base = find_range(fwd)
225
+ end
226
+ break if not base
227
+ fwd = dwcache(base, @range[base])[(fwd-base)/@ptsz + dwoff]
228
+ break if fwd == 0
229
+ return if not cl = @chunks[fwd] # XXX root/tail may be in .data
230
+ return if cl != psz
231
+ break if lst.include? fwd
232
+ lst << fwd
233
+ end
234
+ fwd = ptr
235
+ while pv = @xrchunksfrom[fwd]
236
+ fwd = pv.find { |p|
237
+ next if @chunks[p] != psz
238
+ if not base or base > p or base + @range[base] <= p
239
+ base = find_range(fwd)
240
+ end
241
+ dwcache(base, @range[base])[(p-base)/@ptsz + dwoff] == fwd
242
+ }
243
+ break if not fwd
244
+ break if lst.include? fwd
245
+ lst.unshift fwd
246
+ end
247
+ if lst.length > 3
248
+ lst.each { |p| (@linkedlists[p] ||= {})[dwoff] = lst }
249
+ @alllists << lst
250
+ end
251
+ end
252
+
253
+ # { chunkinarray => { rootptr => [chunks] } }
254
+ attr_accessor :arrays, :allarrays
255
+ def find_arrays
256
+ @arrays = {}
257
+ @allarrays = []
258
+ @buckets.sort.each { |sz, lst|
259
+ next if sz < @ptsz*6
260
+ lst.each { |ptr|
261
+ next if not to = @xrchunksto[ptr]
262
+ # a table must have at least half its storage space filled with ptrs
263
+ next if to.length <= sz/@ptsz/2
264
+ # also, ptrs must point to same-size stuff
265
+ lsz = Hash.new(0)
266
+ to.each { |t| lsz[@chunks[t]] += 1 }
267
+ cnt = lsz.values.max
268
+ next if cnt <= sz/@ptsz/2
269
+ tgsz = lsz.index(cnt)
270
+ ar = to.find_all { |t| @chunks[t] == tgsz }.uniq
271
+ next if ar.length <= sz/@ptsz/2
272
+ ar.each { |p| (@arrays[p] ||= {})[ptr] = ar }
273
+ @allarrays << ar
274
+ }
275
+ }
276
+ end
277
+ end
278
+
279
+ class LinuxHeap < Heap
280
+ # find all chunks in the memory address space
281
+ attr_accessor :mmaps
282
+
283
+ def scan_chunks
284
+ @chunks = {}
285
+ each_heap { |a, l, ar|
286
+ scan_heap(a, l, ar)
287
+ }
288
+ @mmapchunks = []
289
+ @mmaps.each { |a, l|
290
+ ll = scan_mmap(a, l) || 4096
291
+ a += ll
292
+ l -= ll
293
+ }
294
+ end
295
+
296
+ # scan all chunks for cross-references (one chunk contaning a pointer to some other chunk)
297
+ def scan_chunks_xr
298
+ @xrchunksto = {}
299
+ @xrchunksfrom = {}
300
+ each_heap { |a, l, ar|
301
+ scan_heap_xr(a, l)
302
+ }
303
+ @mmapchunks.each { |a|
304
+ scan_mmap_xr(a, @chunks[a])
305
+ }
306
+ end
307
+
308
+ # scan chunks from a heap base addr
309
+ def scan_heap(base, len, ar)
310
+ dw = dwcache(base, len)
311
+ ptr = 0
312
+
313
+ psz = dw[ptr]
314
+ sz = dw[ptr+1]
315
+ base += 2*@ptsz # user pointer
316
+ raise "bad heap base %x %x %x %x" % [psz, sz, base, len] if psz != 0 or sz & 1 == 0
317
+
318
+ loop do
319
+ clen = sz & -8 # chunk size
320
+ ptr += clen/@ptsz # start of next chk
321
+ break if ptr >= dw.length or clen == 0
322
+ sz = dw[ptr+1]
323
+ if sz & 1 > 0 # pv_inuse
324
+ # user data length up to chucksize-4 (over next psz)
325
+ #puts "used #{'%x' % base} #{clen-@ptsz}" if $VERBOSE
326
+ @chunks[base] = clen-@ptsz
327
+ else
328
+ #puts "free #{'%x' % base} #{clen-@ptsz}" if $VERBOSE
329
+ end
330
+ base += clen
331
+ end
332
+
333
+ del_fastbin(ar)
334
+ end
335
+
336
+ def scan_heap_xr(base, len)
337
+ dw = dwcache(base, len)
338
+ @chunks.each_key { |p|
339
+ i = (p-base) / @ptsz
340
+ if i >= 0 and i < dw.length
341
+ lst = dw[i, @chunks[p]/@ptsz].find_all { |pp| @chunks[pp] }
342
+ @xrchunksto[p] = lst if not lst.empty?
343
+ lst.each { |pp| (@xrchunksfrom[pp] ||= []) << p }
344
+ end
345
+ }
346
+ end
347
+
348
+ # scan chunks from a mmap base addr
349
+ # big chunks are allocated on anonymous-mmap areas
350
+ # for mmap chunks, pv_sz=0 pv_inuse=0, mmap=1, data starts at 8, mmapsz = userlen+12 [roundup 4096]
351
+ # one entry in /proc/pid/maps may point to multiple consecutive mmap chunks
352
+ # scans for a mmap chunk header, returns the chunk size if pattern match or nil
353
+ def scan_mmap(base, len)
354
+ foo = chunkdata(base)
355
+ clen = foo[1] & ~0xfff
356
+ if foo[0] == 0 and foo[1] & 0xfff == 2 and clen > 0 and clen <= len
357
+ @chunks[base + foo.length] = clen-4*@ptsz
358
+ @mmapchunks << (base + foo.length)
359
+ clen
360
+ end
361
+ end
362
+
363
+ def scan_mmap_xr(base, len)
364
+ dw = dwcache(base, len)
365
+ lst = dw[2..-1].find_all { |pp| @chunks[pp] }
366
+ @xrchunksto[base] = lst if not lst.empty?
367
+ lst.each { |pp| (@xrchunksfrom[pp] ||= []) << base }
368
+ end
369
+
370
+ attr_accessor :main_arena_ptr
371
+
372
+ # we need to find the main_arena from the libc
373
+ # we do this by analysing 'malloc_trim'
374
+ def scan_libc(addr)
375
+ raise 'no libc' if not addr
376
+
377
+ return if @main_arena_ptr = @dbg.symbols.index('main_arena')
378
+
379
+ unless trim = @dbg.symbols.index('malloc_trim') || @dbg.symbols.index('weak_malloc_trim')
380
+ @dbg.loadsyms 'libc[.-]'
381
+ trim = @dbg.symbols.index('malloc_trim') || @dbg.symbols.index('weak_malloc_trim')
382
+ end
383
+ raise 'cant find malloc_trim' if not trim
384
+
385
+ d = @dbg.disassembler
386
+
387
+ d.disassemble_fast(trim) if not d.di_at(trim)
388
+ if d.block_at(trim).list.last.opcode.name == 'call'
389
+ # x86 getip, need to dasm to have func_binding (cross fingers)
390
+ d.disassemble d.block_at(trim).to_normal.first
391
+ end
392
+ d.each_function_block(trim) { |b|
393
+ # mutex_lock(&main_arena.mutex) gives us the addr
394
+ next if not cmpxchg = d.block_at(b).list.find { |di| di.kind_of? DecodedInstruction and di.opcode.name == 'cmpxchg' }
395
+ @main_arena_ptr = d.backtrace(cmpxchg.instruction.args.first.symbolic.pointer, cmpxchg.address)
396
+ if @main_arena_ptr.length == 1
397
+ @main_arena_ptr = @main_arena_ptr[0].reduce
398
+ break
399
+ end
400
+ }
401
+ raise "cant find mainarena" if not @main_arena_ptr.kind_of? Integer
402
+ @dbg.symbols[@main_arena_ptr] = 'main_arena'
403
+ end
404
+
405
+ def chunkdata(ptr)
406
+ @cp.decode_c_ary('uintptr_t', 2, @dbg.memory, ptr).to_array
407
+ end
408
+
409
+ def each_heap
410
+ if not @cp.toplevel.struct['malloc_state']
411
+ @cp.parse <<EOS
412
+ // TODO autotune these 2 defines..
413
+ #define THREAD_STATS 0
414
+ //#define PER_THREAD
415
+
416
+ #define NFASTBINS 10
417
+ #define NBINS 128
418
+ #define BINMAPSIZE (NBINS/32)
419
+
420
+ struct malloc_state {
421
+ int mutex;
422
+ int flags;
423
+ #if THREAD_STATS
424
+ long stat_lock_direct, stat_lock_loop, stat_lock_wait;
425
+ #endif
426
+ void *fastbinsY[NFASTBINS];
427
+ void *top;
428
+ void *last_remainder;
429
+ void *bins[NBINS * 2 - 2]; // *2: double-linked list
430
+ unsigned int binmap[BINMAPSIZE];
431
+ struct malloc_state *next;
432
+ #ifdef PER_THREAD
433
+ struct malloc_state *next_free;
434
+ #endif
435
+ uintptr_t system_mem; // XXX int32?
436
+ uintptr_t max_system_mem;
437
+ };
438
+
439
+ struct heap_info {
440
+ struct malloc_state *ar_ptr; // Arena for this heap.
441
+ struct _heap_info *prev; // Previous heap.
442
+ uintptr_t size; // Current size in bytes. XXX int32?
443
+ uintptr_t mprotect_size; // Size in bytes that has been mprotected
444
+ };
445
+ EOS
446
+ end
447
+
448
+ ptr = @main_arena_ptr
449
+ loop do
450
+ ar = @cp.decode_c_struct('malloc_state', @dbg.memory, ptr)
451
+ if ptr == @main_arena_ptr
452
+ # main arena: find start from top.end - system_mem
453
+ toplen = chunkdata(ar.top)[1] & -8
454
+ yield ar.top + toplen - ar.system_mem, ar.system_mem, ar
455
+ else
456
+ # non-main arena: find heap_info for top, follow list
457
+ iptr = ar.top & -0x10_0000 # XXX
458
+ while iptr
459
+ hi = @cp.decode_c_struct('heap_info', @dbg.memory, iptr)
460
+ off = hi.sizeof
461
+ off += ar.sizeof if iptr+off == hi.ar_ptr
462
+ yield iptr+off, hi.size-off, ar
463
+
464
+ iptr = hi.prev
465
+ end
466
+ end
467
+
468
+ ptr = ar.next
469
+ break if ptr == @main_arena_ptr
470
+ end
471
+ end
472
+
473
+ def del_fastbin(ar)
474
+ nfastbins = 10
475
+ nfastbins.times { |i|
476
+ ptr = ar.fastbinsy[i]
477
+ while ptr
478
+ @chunks.delete ptr+2*@ptsz
479
+ ptr = @cp.decode_c_ary('void *', 3, @dbg.memory, ptr)[2]
480
+ end
481
+ }
482
+ end
483
+ end
484
+
485
+
486
+ class WindowsHeap < Heap
487
+ attr_accessor :heaps
488
+
489
+ def scan_chunks
490
+ @hsz = @cp.sizeof(@cp.find_c_struct('_HEAP_ENTRY'))
491
+ @chunks = {}
492
+ each_heap { |ar| scan_heap(ar) }
493
+ end
494
+
495
+ # scan all chunks for cross-references (one chunk containing a pointer to some other chunk)
496
+ def scan_chunks_xr
497
+ @xrchunksto = {}
498
+ @xrchunksfrom = {}
499
+ each_heap { |ar| scan_heap_xr(ar) }
500
+ end
501
+
502
+ # scan chunks from a heap
503
+ def scan_heap(ar)
504
+ each_heap_segment(ar) { |p, l|
505
+ scan_heap_segment(p, l)
506
+ }
507
+ scan_frontend(ar)
508
+ scan_valloc(ar)
509
+ end
510
+
511
+ def scan_frontend(ar)
512
+ if ar.frontendheaptype == 1
513
+ laptr = ar.frontendheap
514
+ @chunks.delete laptr # not a real (user) chunk
515
+ 128.times {
516
+ la = @cp.decode_c_struct('_HEAP_LOOKASIDE', @dbg.memory, laptr)
517
+ free = la.listhead.flink
518
+ while free
519
+ @chunks.delete free
520
+ free = @cp.decode_c_struct('_LIST_ENTRY', @dbg.memory, free).flink
521
+ end
522
+ laptr += la.sizeof
523
+ }
524
+ end
525
+ end
526
+
527
+ def scan_valloc(ar)
528
+ each_listentry(ar.virtualallocdblocks, '_HEAP_VIRTUAL_ALLOC_ENTRY') { |va|
529
+ # Unusedbyte count stored in the BusyBlock.Size field
530
+ @chunks[va.stroff + va.sizeof] = va.CommitSize - va.Size
531
+ }
532
+ end
533
+
534
+ def scan_heap_segment(first, len)
535
+ off = 0
536
+ heapcpy = pagecache(first, len)
537
+ while off < len
538
+ he = @cp.decode_c_struct('_HEAP_ENTRY', heapcpy, off)
539
+ sz = he.Size*8
540
+ if he.Flags & 1 == 1
541
+ @chunks[first+off+@hsz] = sz - he.UnusedBytes
542
+ end
543
+ off += sz
544
+ end
545
+ end
546
+
547
+ def scan_heap_xr(ar)
548
+ each_heap_segment(ar) { |p, l|
549
+ scan_heap_segment_xr(p, l)
550
+ }
551
+ end
552
+
553
+ def scan_heap_segment_xr(first, len)
554
+ off = 0
555
+ heapcpy = pagecache(first, len)
556
+ while off < len
557
+ he = @cp.decode_c_struct('_HEAP_ENTRY', heapcpy, off)
558
+ sz = he.Size*8
559
+ ptr = first + off + @hsz
560
+ if he.Flags & 1 == 1 and csz = @chunks[ptr] and csz > 0
561
+ heapcpy[off + @hsz, csz].unpack('L*').each { |p|
562
+ if @chunks[p]
563
+ (@xrchunksto[ptr] ||= []) << p
564
+ (@xrchunksfrom[p] ||= []) << ptr
565
+ end
566
+ }
567
+ end
568
+ off += sz
569
+ end
570
+ end
571
+
572
+ # yields the _HEAP structure for all heaps
573
+ def each_heap
574
+ heaps.each { |a, l|
575
+ ar = @cp.decode_c_struct('_HEAP', @dbg.memory, a)
576
+ yield ar
577
+ }
578
+ end
579
+
580
+ # yields all [ptr, len] for allocated segments of a _HEAP
581
+ # this maps to the _HEAP_SEGMENT further subdivised to skip
582
+ # the _HEAP_UNCOMMMTTED_RANGE areas
583
+ # for the last chunk of the _HEAP_SEGMENT, only yield up to chunk_header
584
+ def each_heap_segment(ar)
585
+ ar.segments.to_array.compact.each { |a|
586
+ sg = @cp.decode_c_struct('_HEAP_SEGMENT', @dbg.memory, a)
587
+ skiplist = []
588
+ ptr = sg.uncommittedranges
589
+ while ptr
590
+ ucr = @cp.decode_c_struct('_HEAP_UNCOMMMTTED_RANGE', @dbg.memory, ptr)
591
+ skiplist << [ucr.Address, ucr.Size]
592
+ ptr = ucr.Next
593
+ end
594
+ ptr = sg.firstentry
595
+ # XXX lastentryinsegment == firstentry ???
596
+ # lastvalidentry = address of the end of the segment (may point to unmapped space)
597
+ ptrend = sg.lastvalidentry
598
+ skiplist.delete_if { |sa, sl| sa < ptr or sa + sl > ptrend }
599
+ skiplist << [ptrend, 1]
600
+ skiplist.sort.each { |sa, sl|
601
+ yield(ptr, sa-ptr)
602
+ ptr = sa + sl
603
+ }
604
+ }
605
+ end
606
+
607
+ # call with a LIST_ENTRY allocstruct, the target structure and LE offset in this structure
608
+ def each_listentry(le, st, off=0)
609
+ ptr0 = le.stroff
610
+ ptr = le.flink
611
+ while ptr != ptr0
612
+ yield @cp.decode_c_struct(st, @dbg.memory, ptr-off)
613
+ ptr = @cp.decode_c_struct('_LIST_ENTRY', @dbg.memory, ptr).flink
614
+ end
615
+ end
616
+ end
617
+
618
+ class Windows7Heap < WindowsHeap
619
+ # 4-byte xor key to decrypt the chunk headers
620
+ attr_accessor :chunkkey_size, :chunkkey_flags, :chunkkey_unusedbytes
621
+ def each_heap_segment(ar)
622
+ if ar.encodeflagmask != 0
623
+ @chunkkey_size = ar.encoding.size
624
+ @chunkkey_flags = ar.encoding.flags
625
+ @chunkkey_unusedbytes = ar.encoding.unusedbytes
626
+ else
627
+ @chunkkey_size = 0
628
+ @chunkkey_flags = 0
629
+ @chunkkey_unusedbytes = 0
630
+ end
631
+
632
+ each_listentry(ar.segmentlist, '_HEAP_SEGMENT', 0x10) { |sg|
633
+ skiplist = []
634
+ each_listentry(sg.ucrsegmentlist, '_HEAP_UCR_DESCRIPTOR', 8) { |ucr|
635
+ skiplist << [ucr.address, ucr.size]
636
+ }
637
+
638
+ ptr = sg.firstentry
639
+ ptrend = sg.lastvalidentry + @hsz
640
+ skiplist.delete_if { |sa, sl| sa < ptr or sa + sl > ptrend }
641
+ skiplist << [ptrend, 1]
642
+ skiplist.sort.each { |sa, sl|
643
+ yield(ptr, sa-ptr)
644
+ ptr = sa + sl
645
+ }
646
+ }
647
+ end
648
+
649
+ def scan_heap_segment(first, len)
650
+ off = 0
651
+ heapcpy = pagecache(first, len)
652
+ while off < len
653
+ he = @cp.decode_c_struct('_HEAP_ENTRY', heapcpy, off)
654
+ sz = (he.Size ^ @chunkkey_size)*8
655
+ if (he.Flags ^ @chunkkey_flags) & 1 == 1
656
+ @chunks[first+off+@hsz] = sz - (he.UnusedBytes ^ @chunkkey_unusedbytes)
657
+ end
658
+ off += sz
659
+ end
660
+ end
661
+
662
+ def scan_frontend(ar)
663
+ return if ar.frontendheaptype != 2
664
+ lfh = @cp.decode_c_struct('_LFH_HEAP', @dbg.memory, ar.frontendheap)
665
+ lfh.localdata[0].segmentinfo.to_array.each { |sinfo|
666
+ sinfo.cacheditems.to_array.each { |ssp|
667
+ next if not ssp
668
+ subseg = @cp.decode_c_struct('_HEAP_SUBSEGMENT', @dbg.memory, ssp)
669
+ scan_lfh_ss(subseg)
670
+ }
671
+ }
672
+ end
673
+
674
+ def scan_lfh_ss(subseg)
675
+ up = subseg.userblocks
676
+ return if not up
677
+ bs = subseg.blocksize
678
+ bc = subseg.blockcount
679
+ list = Array.new(bc) { |i| up + 0x10 + bs*8*i }
680
+
681
+ free = []
682
+ ptr = subseg.freeentryoffset
683
+ subseg.depth.times {
684
+ free << (up + 8*ptr)
685
+ ptr = @dbg.memory[up + 8*ptr + 8, 2].unpack('v')[0]
686
+ }
687
+ @foo ||= 0
688
+ @foo += 1
689
+ p @foo if @foo % 10 == 0
690
+
691
+ up += 0x10
692
+ list -= free
693
+ list.each { |p| @chunks[p+8] = bs*8 - (@cp.decode_c_struct('_HEAP_ENTRY', @dbg.memory, p).unusedbytes & 0x7f) }
694
+ end
695
+
696
+ def scan_chunks_xr
697
+ @xrchunksto = {}
698
+ @xrchunksfrom = {}
699
+ @chunks.each { |a, l|
700
+ pagecache(a, l).unpack('L*').each { |p|
701
+ if @chunks[p]
702
+ (@xrchunksto[a] ||= []) << p
703
+ (@xrchunksfrom[p] ||= []) << a
704
+ end
705
+ }
706
+ }
707
+ end
708
+ end
709
+ end