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
@@ -5,19 +5,27 @@
5
5
 
6
6
 
7
7
  require 'metasm/main'
8
- require 'metasm/ia32'
8
+ require 'metasm/cpu/ia32'
9
9
 
10
10
  module Metasm
11
11
 
12
12
  # The x86_64, 64-bit extension of the x86 CPU (x64, em64t, amd64...)
13
13
  class X86_64 < Ia32
14
14
  # FpReg, SegReg, Farptr unchanged
15
- # XXX ST(15) ?
16
15
 
17
- # Simd extended to 16 regs, xmm only (mmx gone with 80387)
16
+ # XMM extended to 16 regs, YMM
18
17
  class SimdReg < Ia32::SimdReg
19
- double_map 64 => (0..15).map { |n| "mm#{n}" },
20
- 128 => (0..15).map { |n| "xmm#{n}" }
18
+ double_map 64 => (0..7).map { |n| "mm#{n}" },
19
+ 128 => (0..15).map { |n| "xmm#{n}" },
20
+ 256 => (0..15).map { |n| "ymm#{n}" }
21
+
22
+ def val_enc
23
+ @val & 7
24
+ end
25
+
26
+ def val_rex
27
+ @val >> 3
28
+ end
21
29
  end
22
30
 
23
31
  # general purpose registers, all sizes
@@ -125,7 +133,7 @@ class X86_64 < Ia32
125
133
 
126
134
  def str_to_reg(str)
127
135
  # X86_64::Reg != Ia32::Reg
128
- Reg.from_str(str) if Reg.s_to_i.has_key? str
136
+ Reg.s_to_i.has_key?(str) ? Reg.from_str(str) : SimdReg.s_to_i.has_key?(str) ? SimdReg.from_str(str) : nil
129
137
  end
130
138
 
131
139
  def shortname
@@ -0,0 +1,136 @@
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/cpu/x86_64/main'
8
+ require 'metasm/cpu/ia32/opcodes'
9
+
10
+ module Metasm
11
+ class X86_64
12
+ def init_cpu_constants
13
+ super()
14
+ [:i32, :u32, :i64, :u64].each { |a| @valid_args[a] = true }
15
+ end
16
+
17
+ def init_386_common_only
18
+ super()
19
+ # :imm64 => accept a real int64 as :i argument
20
+ # :auto64 => ignore rex_w, always 64-bit op
21
+ # :op32no64 => if write to a 32-bit reg, dont zero the top 32-bits of dest
22
+ [:imm64, :auto64, :op32no64].each { |a| @valid_props[a] = true }
23
+ @opcode_list.delete_if { |o| o.bin[0].to_i & 0xf0 == 0x40 } # now REX prefix
24
+ @opcode_list.each { |o|
25
+ o.props[:imm64] = true if o.bin == [0xB8] # mov reg, <true imm64>
26
+ o.props[:auto64] = true if o.name =~ /^(j.*|loop.*|call|enter|leave|push|pop|ret)$/
27
+ }
28
+ addop 'movsxd', [0x63], :mrm
29
+ addop('cdqe', [0x98]) { |o| o.props[:opsz] = 64 }
30
+ addop('cqo', [0x99]) { |o| o.props[:opsz] = 64 }
31
+ end
32
+
33
+ # all x86_64 cpu understand <= sse2 instrs
34
+ def init_x8664_only
35
+ init_386_common_only
36
+ init_386_only
37
+ init_387_only
38
+ init_486_only
39
+ init_pentium_only
40
+ init_p6_only
41
+ init_sse_only
42
+ init_sse2_only
43
+
44
+ @opcode_list.delete_if { |o|
45
+ o.args.include?(:seg2) or
46
+ o.args.include?(:seg2A) or
47
+ o.args.include?(:farptr) or
48
+ %w[aaa aad aam aas bound daa das into jcxz jecxz
49
+ lds les loadall arpl pusha pushad popa
50
+ popad].include?(o.name.split('.')[0])
51
+ # split needed for lds.a32
52
+ }
53
+
54
+ @opcode_list.each { |o|
55
+ o.props[:auto64] = true if o.name =~ /^(enter|leave|[sl]gdt|[sl]idt|[sl]ldt|[sl]tr|push|pop|syscall)$/
56
+ }
57
+
58
+ addop('cmpxchg16b', [0x0F, 0xC7], 1) { |o| o.props[:opsz] = 64 ; o.props[:argsz] = 128 }
59
+ addop('iretq', [0xCF], nil, :stopexec, :setip) { |o| o.props[:opsz] = 64 } ; opcode_list.unshift opcode_list.pop
60
+ addop 'swapgs', [0x0F, 0x01, 0xF8]
61
+
62
+ addop('movq', [0x0F, 0x6E], :mrmmmx, {:d => [1, 4]}) { |o| o.args = [:regmmx, :modrm] ; o.props[:opsz] = o.props[:argsz] = 64 }
63
+ addop('movq', [0x0F, 0x6E], :mrmxmm, {:d => [1, 4]}) { |o| o.args = [:regxmm, :modrm] ; o.props[:opsz] = o.props[:argsz] = 64 ; o.props[:needpfx] = 0x66 }
64
+ addop('jecxz', [0xE3], nil, :setip, :i8) { |o| o.props[:adsz] = 32 }
65
+ addop('jrcxz', [0xE3], nil, :setip, :i8) { |o| o.props[:adsz] = 64 }
66
+ end
67
+
68
+ def init_sse3
69
+ init_x8664_only
70
+ init_sse3_only
71
+ end
72
+
73
+ def init_sse41_only
74
+ super()
75
+ addop('pextrq', [0x0F, 0x3A, 0x16], :mrmxmm, :u8) { |o| o.props[:needpfx] = 0x66; o.args[o.args.index(:modrmxmm)] = :modrm; o.props[:opsz] = o.props[:argsz] = 64 }
76
+ addop('pinsrq', [0x0F, 0x3A, 0x22], :mrmxmm, :u8) { |o| o.props[:needpfx] = 0x66; o.args[o.args.index(:modrmxmm)] = :modrm; o.props[:opsz] = o.props[:argsz] = 64 }
77
+ end
78
+
79
+ def init_avx_only
80
+ super()
81
+ addop('rdfsbase', [0x0F, 0xAE], 0, :modrmR) { |o| o.props[:needpfx] = 0xF3 }
82
+ addop('rdgsbase', [0x0F, 0xAE], 1, :modrmR) { |o| o.props[:needpfx] = 0xF3 }
83
+ addop('wrfsbase', [0x0F, 0xAE], 2, :modrmR) { |o| o.props[:needpfx] = 0xF3 }
84
+ addop('wrgsbase', [0x0F, 0xAE], 3, :modrmR) { |o| o.props[:needpfx] = 0xF3 }
85
+ end
86
+
87
+ def addop_macrostr(name, bin, type)
88
+ super(name, bin, type)
89
+ bin = bin.dup
90
+ bin[0] |= 1
91
+ addop(name+'q', bin) { |o| o.props[:opsz] = 64 ; o.props[type] = true }
92
+ end
93
+
94
+ def addop_macroret(name, bin, *args)
95
+ addop(name + '.i64', bin, nil, :stopexec, :setip, *args) { |o| o.props[:opsz] = 64 }
96
+ super(name, bin, *args)
97
+ end
98
+
99
+ def addop_post(op)
100
+ if op.fields[:d] or op.fields[:w] or op.fields[:s] or op.args.first == :regfp0
101
+ return super(op)
102
+ end
103
+
104
+ if op.props[:needpfx]
105
+ @opcode_list.unshift op
106
+ else
107
+ @opcode_list << op
108
+ end
109
+
110
+ if op.args == [:i] or op.name == 'ret'
111
+ # define opsz-override version for ambiguous opcodes
112
+ op16 = op.dup
113
+ op16.name << '.i16'
114
+ op16.props[:opsz] = 16
115
+ @opcode_list << op16
116
+ # push call ret jz can't 32bit
117
+ op64 = op.dup
118
+ op64.name << '.i64'
119
+ op64.props[:opsz] = 64
120
+ @opcode_list << op64
121
+ elsif op.props[:strop] or op.props[:stropz] or op.args.include? :mrm_imm or
122
+ op.args.include? :modrm or op.name =~ /loop|xlat/
123
+ # define adsz-override version for ambiguous opcodes (movsq)
124
+ # XXX loop pfx 67 = rip+ecx, 66/rex ignored
125
+ op32 = op.dup
126
+ op32.name << '.a32'
127
+ op32.props[:adsz] = 32
128
+ @opcode_list << op32
129
+ op64 = op.dup
130
+ op64.name << '.a64'
131
+ op64.props[:adsz] = 64
132
+ @opcode_list << op64
133
+ end
134
+ end
135
+ end
136
+ end
@@ -4,8 +4,8 @@
4
4
  # Licence is LGPL, see LICENCE in the top-level directory
5
5
 
6
6
 
7
- require 'metasm/x86_64/opcodes'
8
- require 'metasm/x86_64/encode'
7
+ require 'metasm/cpu/x86_64/opcodes'
8
+ require 'metasm/cpu/x86_64/encode'
9
9
  require 'metasm/parse'
10
10
 
11
11
  module Metasm
@@ -52,6 +52,8 @@ class X86_64
52
52
  return if arg.kind_of? ModRM and ((arg.b and arg.b.val == 16 and arg.i) or (arg.i and arg.i.val == 16 and (arg.b or arg.s != 1)))
53
53
  return if arg.kind_of? Reg and arg.sz >= 32 and arg.val == 16 # eip/rip only in modrm
54
54
  return if o.props[:auto64] and arg.respond_to? :sz and arg.sz == 32
55
+ # vex c4/c5
56
+ return if o.fields[:vex_r] and not o.fields[:vex_b] and (spec == :modrm or spec == :modrmxmm or spec == :modrmymm) and (((arg.kind_of?(SimdReg) or arg.kind_of?(Reg)) and arg.val >= 8) or (arg.kind_of?(ModRM) and ((arg.b and arg.b.val >= 8) or (arg.i and arg.i.val >= 8))))
55
57
  if o.name == 'movsxd'
56
58
  return if not arg.kind_of? Reg and not arg.kind_of? ModRM
57
59
  arg.sz ||= 32
@@ -62,7 +64,13 @@ class X86_64
62
64
  return arg.sz == 32
63
65
  end
64
66
  end
67
+ return if o.name == 'xchg' and spec == :reg and o.args.include?(:reg_eax) and arg.kind_of?(Reg) and arg.sz == 32 and arg.val == 0
68
+
65
69
  super(o, spec, arg)
66
70
  end
71
+
72
+ def check_reserved_name(name)
73
+ Reg.s_to_i[name]
74
+ end
67
75
  end
68
76
  end
@@ -0,0 +1,35 @@
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/cpu/x86_64/opcodes'
8
+ require 'metasm/render'
9
+
10
+ module Metasm
11
+ class X86_64
12
+ def gui_hilight_word_regexp_init
13
+ ret = {}
14
+
15
+ %w[a b c d].each { |r|
16
+ ret["#{r}l"] = "[re]?#{r}x|#{r}l"
17
+ ret["#{r}h"] = "[re]?#{r}x|#{r}h"
18
+ ret["#{r}x"] = ret["e#{r}x"] = ret["r#{r}x"] = "[re]?#{r}x|#{r}[hl]"
19
+ }
20
+
21
+ %w[sp bp si di].each { |r|
22
+ ret["#{r}l"] = ret[r] = ret["e#{r}"] = ret["r#{r}"] = "[re]?#{r}|#{r}l"
23
+ }
24
+
25
+ (8..15).each { |i|
26
+ r = "r#{i}"
27
+ ret[r+'b'] = ret[r+'w'] = ret[r+'d'] = ret[r] = "#{r}[bwd]?"
28
+ }
29
+
30
+ ret['eip'] = ret['rip'] = '[re]ip'
31
+
32
+ ret
33
+ end
34
+ end
35
+ end
@@ -0,0 +1,9 @@
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/main'
8
+ require 'metasm/cpu/z80/decode'
9
+ require 'metasm/cpu/z80/render'
@@ -0,0 +1,313 @@
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/cpu/z80/opcodes'
8
+ require 'metasm/decode'
9
+
10
+ module Metasm
11
+ class Z80
12
+ def build_opcode_bin_mask(op)
13
+ # bit = 0 if can be mutated by an field value, 1 if fixed by opcode
14
+ op.bin_mask = Array.new(op.bin.length, 0)
15
+ op.fields.each { |f, (oct, off)|
16
+ op.bin_mask[oct] |= (@fields_mask[f] << off)
17
+ }
18
+ op.bin_mask.map! { |v| 255 ^ v }
19
+ end
20
+
21
+ def build_bin_lookaside
22
+ # sets up a hash byte value => list of opcodes that may match
23
+ # opcode.bin_mask is built here
24
+ lookaside = Array.new(256) { [] }
25
+ opcode_list.each { |op|
26
+ build_opcode_bin_mask op
27
+ b = op.bin[0]
28
+ msk = op.bin_mask[0]
29
+ next @unknown_opcode = op if not b
30
+ for i in b..(b | (255^msk))
31
+ lookaside[i] << op if i & msk == b & msk
32
+ end
33
+ }
34
+ lookaside
35
+ end
36
+
37
+ def decode_prefix(instr, byte)
38
+ case byte
39
+ when 0xDD; instr.prefix = 0xDD
40
+ when 0xFD; instr.prefix = 0xFD
41
+ # implicit 'else return false'
42
+ end
43
+ end
44
+
45
+ # tries to find the opcode encoded at edata.ptr
46
+ # if no match, tries to match a prefix (update di.instruction.prefix)
47
+ # on match, edata.ptr points to the first byte of the opcode (after prefixes)
48
+ def decode_findopcode(edata)
49
+ di = DecodedInstruction.new self
50
+ while edata.ptr < edata.data.length
51
+ byte = edata.data[edata.ptr]
52
+ byte = byte.unpack('C').first if byte.kind_of?(::String)
53
+ return di if di.opcode = @bin_lookaside[byte].find { |op|
54
+ # fetch the relevant bytes from edata
55
+ bseq = edata.data[edata.ptr, op.bin.length].unpack('C*')
56
+ # check against full opcode mask
57
+ op.bin.zip(bseq, op.bin_mask).all? { |b1, b2, m| b2 and ((b1 & m) == (b2 & m)) }
58
+ }
59
+
60
+ if decode_prefix(di.instruction, edata.get_byte)
61
+ nb = edata.data[edata.ptr]
62
+ nb = nb.unpack('C').first if nb.kind_of?(::String)
63
+ case nb
64
+ when 0xCB
65
+ # DD CB <disp8> <opcode_pfxCB> [<args>]
66
+ di.instruction.prefix |= edata.get_byte << 8
67
+ di.bin_length += 2
68
+ opc = edata.data[edata.ptr+1]
69
+ opc = opc.unpack('C').first if opc.kind_of?(::String)
70
+ bseq = [0xCB, opc]
71
+ # XXX in decode_instr_op, byte[0] is the immediate displacement instead of cb
72
+ return di if di.opcode = @bin_lookaside[nb].find { |op|
73
+ op.bin.zip(bseq, op.bin_mask).all? { |b1, b2, m| b2 and ((b1 & m) == (b2 & m)) }
74
+ }
75
+ when 0xED
76
+ di.instruction.prefix = nil
77
+ end
78
+ else
79
+ di.opcode = @unknown_opcode
80
+ return di
81
+ end
82
+ di.bin_length += 1
83
+ end
84
+ end
85
+
86
+
87
+ def decode_instr_op(edata, di)
88
+ before_ptr = edata.ptr
89
+ op = di.opcode
90
+ di.instruction.opname = op.name
91
+ bseq = edata.read(op.bin.length).unpack('C*') # decode_findopcode ensures that data >= op.length
92
+ pfx = di.instruction.prefix
93
+
94
+ field_val = lambda { |f|
95
+ if fld = op.fields[f]
96
+ (bseq[fld[0]] >> fld[1]) & @fields_mask[f]
97
+ end
98
+ }
99
+
100
+ op.args.each { |a|
101
+ di.instruction.args << case a
102
+ when :i8, :u8, :i16, :u16; Expression[edata.decode_imm(a, @endianness)]
103
+ when :iy; Expression[field_val[a]]
104
+ when :iy8; Expression[field_val[a]*8]
105
+
106
+ when :rp
107
+ v = field_val[a]
108
+ Reg.new(16, v)
109
+ when :rp2
110
+ v = field_val[a]
111
+ v = 4 if v == 3
112
+ Reg.new(16, v)
113
+ when :ry, :rz
114
+ v = field_val[a]
115
+ if v == 6
116
+ Memref.new(Reg.from_str('HL'), nil, 1)
117
+ else
118
+ Reg.new(8, v)
119
+ end
120
+
121
+ when :r_a; Reg.from_str('A')
122
+ when :r_af; Reg.from_str('AF')
123
+ when :r_hl; Reg.from_str('HL')
124
+ when :r_de; Reg.from_str('DE')
125
+ when :r_sp; Reg.from_str('SP')
126
+ when :r_i; Reg.from_str('I')
127
+
128
+ when :m16; Memref.new(nil, edata.decode_imm(:u16, @endianness), nil)
129
+ when :m_bc; Memref.new(Reg.from_str('BC'), nil, 1)
130
+ when :m_de; Memref.new(Reg.from_str('DE'), nil, 1)
131
+ when :m_sp; Memref.new(Reg.from_str('SP'), nil, 2)
132
+ when :m_hl; Memref.new(Reg.from_str('HL'), nil, 1)
133
+ when :mf8; Memref.new(nil, 0xff00 + edata.decode_imm(:u8, @endianness), 1)
134
+ when :mfc; Memref.new(Reg.from_str('C'), 0xff00, 1)
135
+
136
+ else raise SyntaxError, "Internal error: invalid argument #{a} in #{op.name}"
137
+ end
138
+ }
139
+
140
+ case pfx
141
+ when 0xDD
142
+ when 0xFD
143
+ when 0xCBDD
144
+ when 0xCBFD
145
+ end
146
+
147
+ di.bin_length += edata.ptr - before_ptr
148
+
149
+ return if edata.ptr > edata.length
150
+
151
+ di
152
+ end
153
+
154
+ # hash opcode_name => lambda { |dasm, di, *symbolic_args| instr_binding }
155
+ def backtrace_binding
156
+ @backtrace_binding ||= init_backtrace_binding
157
+ end
158
+ def backtrace_binding=(b) @backtrace_binding = b end
159
+
160
+ # populate the @backtrace_binding hash with default values
161
+ def init_backtrace_binding
162
+ @backtrace_binding ||= {}
163
+
164
+ mask = 0xffff
165
+
166
+ opcode_list.map { |ol| ol.basename }.uniq.sort.each { |op|
167
+ binding = case op
168
+ when 'ld'; lambda { |di, a0, a1, *aa| a2 = aa[0] ; a2 ? { a0 => Expression[a1, :+, a2] } : { a0 => Expression[a1] } }
169
+ when 'ldi'; lambda { |di, a0, a1| hl = (a0 == :a ? a1 : a0) ; { a0 => Expression[a1], hl => Expression[hl, :+, 1] } }
170
+ when 'ldd'; lambda { |di, a0, a1| hl = (a0 == :a ? a1 : a0) ; { a0 => Expression[a1], hl => Expression[hl, :-, 1] } }
171
+ when 'add', 'adc', 'sub', 'sbc', 'and', 'xor', 'or'
172
+ lambda { |di, a0, a1|
173
+ e_op = { 'add' => :+, 'adc' => :+, 'sub' => :-, 'sbc' => :-, 'and' => :&, 'xor' => :^, 'or' => :| }[op]
174
+ ret = Expression[a0, e_op, a1]
175
+ ret = Expression[ret, e_op, :flag_c] if op == 'adc' or op == 'sbc'
176
+ ret = Expression[ret.reduce] if not a0.kind_of? Indirection
177
+ { a0 => ret }
178
+ }
179
+ when 'cp', 'cmp'; lambda { |di, *a| {} }
180
+ when 'inc'; lambda { |di, a0| { a0 => Expression[a0, :+, 1] } }
181
+ when 'dec'; lambda { |di, a0| { a0 => Expression[a0, :-, 1] } }
182
+ when 'not'; lambda { |di, a0| { a0 => Expression[a0, :^, mask] } }
183
+ when 'push'
184
+ lambda { |di, a0| { :sp => Expression[:sp, :-, 2],
185
+ Indirection[:sp, 2, di.address] => Expression[a0] } }
186
+ when 'pop'
187
+ lambda { |di, a0| { :sp => Expression[:sp, :+, 2],
188
+ a0 => Indirection[:sp, 2, di.address] } }
189
+ when 'call'
190
+ lambda { |di, a0| { :sp => Expression[:sp, :-, 2],
191
+ Indirection[:sp, 2, di.address] => Expression[di.next_addr] }
192
+ }
193
+ when 'ret', 'reti'; lambda { |di, *a| { :sp => Expression[:sp, :+, 2] } }
194
+ # TODO callCC, retCC ...
195
+ when 'bswap'
196
+ lambda { |di, a0| { a0 => Expression[
197
+ [[a0, :&, 0xff00], :>>, 8], :|,
198
+ [[a0, :&, 0x00ff], :<<, 8]] } }
199
+ when 'nop', /^j/; lambda { |di, *a| {} }
200
+ end
201
+
202
+ # TODO flags ?
203
+
204
+ @backtrace_binding[op] ||= binding if binding
205
+ }
206
+ @backtrace_binding
207
+ end
208
+
209
+ def get_backtrace_binding(di)
210
+ a = di.instruction.args.map { |arg|
211
+ case arg
212
+ when Memref, Reg; arg.symbolic(di)
213
+ else arg
214
+ end
215
+ }
216
+
217
+ if binding = backtrace_binding[di.opcode.basename]
218
+ binding[di, *a]
219
+ else
220
+ puts "unhandled instruction to backtrace: #{di}" if $VERBOSE
221
+ # assume nothing except the 1st arg is modified
222
+ case a[0]
223
+ when Indirection, Symbol; { a[0] => Expression::Unknown }
224
+ when Expression; (x = a[0].externals.first) ? { x => Expression::Unknown } : {}
225
+ else {}
226
+ end.update(:incomplete_binding => Expression[1])
227
+ end
228
+ end
229
+
230
+ # patch a forward binding from the backtrace binding
231
+ def fix_fwdemu_binding(di, fbd)
232
+ case di.opcode.name
233
+ when 'push', 'call'; fbd[Indirection[[:sp, :-, 2], 2]] = fbd.delete(Indirection[:sp, 2])
234
+ end
235
+ fbd
236
+ end
237
+
238
+ def get_xrefs_x(dasm, di)
239
+ return [] if not di.opcode.props[:setip]
240
+
241
+ case di.opcode.basename
242
+ when 'ret', 'reti'
243
+ return [Indirection[:sp, 2, di.address]]
244
+ when /^jr|^djnz/
245
+ # jmp/call are absolute addrs, only jr/djnz are relative
246
+ # also, the asm source should display the relative offset
247
+ return [Expression[[di.address, :+, di.bin_length], :+, di.instruction.args.first]]
248
+ end
249
+
250
+ case tg = di.instruction.args.first
251
+ when Memref; [Expression[tg.symbolic(di)]]
252
+ when Reg; [Expression[tg.symbolic(di)]]
253
+ when Expression, ::Integer; [Expression[tg]]
254
+ else
255
+ puts "unhandled setip at #{di.address} #{di.instruction}" if $DEBUG
256
+ []
257
+ end
258
+ end
259
+
260
+ # checks if expr is a valid return expression matching the :saveip instruction
261
+ def backtrace_is_function_return(expr, di=nil)
262
+ expr = Expression[expr].reduce_rec
263
+ expr.kind_of?(Indirection) and expr.len == 2 and expr.target == Expression[:sp]
264
+ end
265
+
266
+ # updates the function backtrace_binding
267
+ # if the function is big and no specific register is given, do nothing (the binding will be lazily updated later, on demand)
268
+ def backtrace_update_function_binding(dasm, faddr, f, retaddrlist, *wantregs)
269
+ b = f.backtrace_binding
270
+
271
+ bt_val = lambda { |r|
272
+ next if not retaddrlist
273
+ b[r] = Expression::Unknown
274
+ bt = []
275
+ retaddrlist.each { |retaddr|
276
+ bt |= dasm.backtrace(Expression[r], retaddr, :include_start => true,
277
+ :snapshot_addr => faddr, :origin => retaddr)
278
+ }
279
+ if bt.length != 1
280
+ b[r] = Expression::Unknown
281
+ else
282
+ b[r] = bt.first
283
+ end
284
+ }
285
+
286
+ if not wantregs.empty?
287
+ wantregs.each(&bt_val)
288
+ else
289
+ bt_val[:sp]
290
+ end
291
+
292
+ b
293
+ end
294
+
295
+ # returns true if the expression is an address on the stack
296
+ def backtrace_is_stack_address(expr)
297
+ Expression[expr].expr_externals.include?(:sp)
298
+ end
299
+
300
+ # updates an instruction's argument replacing an expression with another (eg label renamed)
301
+ def replace_instr_arg_immediate(i, old, new)
302
+ i.args.map! { |a|
303
+ case a
304
+ when Expression; a == old ? new : Expression[a.bind(old => new).reduce]
305
+ when Memref
306
+ a.offset = (a.offset == old ? new : Expression[a.offset.bind(old => new).reduce]) if a.offset
307
+ a
308
+ else a
309
+ end
310
+ }
311
+ end
312
+ end
313
+ end