metasm 1.0.0 → 1.0.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- checksums.yaml.gz.sig +0 -0
- data.tar.gz.sig +3 -0
- data/.gitignore +3 -0
- data/.hgtags +3 -0
- data/Gemfile +3 -0
- data/INSTALL +61 -0
- data/LICENCE +458 -0
- data/README +29 -21
- data/Rakefile +10 -0
- data/TODO +10 -12
- data/doc/code_organisation.txt +3 -1
- data/doc/core/DynLdr.txt +247 -0
- data/doc/core/ExeFormat.txt +43 -0
- data/doc/core/Expression.txt +220 -0
- data/doc/core/GNUExports.txt +27 -0
- data/doc/core/Ia32.txt +236 -0
- data/doc/core/SerialStruct.txt +108 -0
- data/doc/core/VirtualString.txt +145 -0
- data/doc/core/WindowsExports.txt +61 -0
- data/doc/core/index.txt +1 -0
- data/doc/style.css +6 -3
- data/doc/usage/debugger.txt +327 -0
- data/doc/usage/index.txt +1 -0
- data/doc/use_cases.txt +2 -2
- data/metasm.gemspec +23 -0
- data/{lib/metasm.rb → metasm.rb} +15 -3
- data/{lib/metasm → metasm}/compile_c.rb +15 -9
- data/metasm/cpu/arc.rb +8 -0
- data/metasm/cpu/arc/decode.rb +404 -0
- data/metasm/cpu/arc/main.rb +191 -0
- data/metasm/cpu/arc/opcodes.rb +588 -0
- data/metasm/cpu/arm.rb +14 -0
- data/{lib/metasm → metasm/cpu}/arm/debug.rb +2 -2
- data/{lib/metasm → metasm/cpu}/arm/decode.rb +15 -18
- data/{lib/metasm → metasm/cpu}/arm/encode.rb +23 -8
- data/{lib/metasm → metasm/cpu}/arm/main.rb +3 -6
- data/metasm/cpu/arm/opcodes.rb +324 -0
- data/{lib/metasm → metasm/cpu}/arm/parse.rb +25 -13
- data/{lib/metasm → metasm/cpu}/arm/render.rb +2 -2
- data/metasm/cpu/arm64.rb +15 -0
- data/metasm/cpu/arm64/debug.rb +38 -0
- data/metasm/cpu/arm64/decode.rb +285 -0
- data/metasm/cpu/arm64/encode.rb +41 -0
- data/metasm/cpu/arm64/main.rb +105 -0
- data/metasm/cpu/arm64/opcodes.rb +232 -0
- data/metasm/cpu/arm64/parse.rb +20 -0
- data/metasm/cpu/arm64/render.rb +95 -0
- data/{lib/metasm/mips/compile_c.rb → metasm/cpu/bpf.rb} +4 -2
- data/metasm/cpu/bpf/decode.rb +110 -0
- data/metasm/cpu/bpf/main.rb +60 -0
- data/metasm/cpu/bpf/opcodes.rb +81 -0
- data/metasm/cpu/bpf/render.rb +30 -0
- data/{lib/metasm/ppc.rb → metasm/cpu/cy16.rb} +2 -4
- data/metasm/cpu/cy16/decode.rb +247 -0
- data/metasm/cpu/cy16/main.rb +63 -0
- data/metasm/cpu/cy16/opcodes.rb +78 -0
- data/metasm/cpu/cy16/render.rb +30 -0
- data/metasm/cpu/dalvik.rb +11 -0
- data/{lib/metasm → metasm/cpu}/dalvik/decode.rb +34 -34
- data/{lib/metasm → metasm/cpu}/dalvik/main.rb +71 -4
- data/{lib/metasm → metasm/cpu}/dalvik/opcodes.rb +21 -12
- data/{lib/metasm/mips.rb → metasm/cpu/ebpf.rb} +3 -4
- 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.rb +17 -0
- data/{lib/metasm → metasm/cpu}/ia32/compile_c.rb +23 -9
- data/{lib/metasm → metasm/cpu}/ia32/debug.rb +44 -6
- data/{lib/metasm → metasm/cpu}/ia32/decode.rb +342 -128
- data/{lib/metasm → metasm/cpu}/ia32/decompile.rb +75 -53
- data/{lib/metasm → metasm/cpu}/ia32/encode.rb +19 -13
- data/{lib/metasm → metasm/cpu}/ia32/main.rb +66 -8
- data/metasm/cpu/ia32/opcodes.rb +1424 -0
- data/{lib/metasm → metasm/cpu}/ia32/parse.rb +55 -17
- data/{lib/metasm → metasm/cpu}/ia32/render.rb +32 -5
- data/metasm/cpu/mcs51.rb +8 -0
- data/metasm/cpu/mcs51/decode.rb +99 -0
- data/metasm/cpu/mcs51/main.rb +87 -0
- data/metasm/cpu/mcs51/opcodes.rb +120 -0
- data/metasm/cpu/mips.rb +14 -0
- data/metasm/cpu/mips/debug.rb +42 -0
- data/{lib/metasm → metasm/cpu}/mips/decode.rb +59 -38
- data/{lib/metasm → metasm/cpu}/mips/encode.rb +4 -3
- data/{lib/metasm → metasm/cpu}/mips/main.rb +13 -6
- data/{lib/metasm → metasm/cpu}/mips/opcodes.rb +87 -18
- data/{lib/metasm → metasm/cpu}/mips/parse.rb +1 -1
- data/{lib/metasm → metasm/cpu}/mips/render.rb +1 -1
- data/{lib/metasm/dalvik.rb → metasm/cpu/msp430.rb} +1 -1
- data/metasm/cpu/msp430/decode.rb +243 -0
- data/metasm/cpu/msp430/main.rb +62 -0
- data/metasm/cpu/msp430/opcodes.rb +101 -0
- 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/{lib/metasm → metasm/cpu}/pic16c/decode.rb +6 -7
- data/{lib/metasm → metasm/cpu}/pic16c/main.rb +0 -0
- data/{lib/metasm → metasm/cpu}/pic16c/opcodes.rb +1 -1
- data/metasm/cpu/ppc.rb +11 -0
- data/{lib/metasm → metasm/cpu}/ppc/decode.rb +18 -37
- data/{lib/metasm → metasm/cpu}/ppc/decompile.rb +3 -3
- data/{lib/metasm → metasm/cpu}/ppc/encode.rb +2 -2
- data/{lib/metasm → metasm/cpu}/ppc/main.rb +23 -18
- data/{lib/metasm → metasm/cpu}/ppc/opcodes.rb +11 -6
- data/metasm/cpu/ppc/parse.rb +55 -0
- data/metasm/cpu/python.rb +8 -0
- data/metasm/cpu/python/decode.rb +116 -0
- data/metasm/cpu/python/main.rb +36 -0
- data/metasm/cpu/python/opcodes.rb +180 -0
- data/{lib/metasm → metasm/cpu}/sh4.rb +1 -1
- data/{lib/metasm → metasm/cpu}/sh4/decode.rb +50 -23
- data/{lib/metasm → metasm/cpu}/sh4/main.rb +38 -27
- data/{lib/metasm → metasm/cpu}/sh4/opcodes.rb +7 -8
- data/metasm/cpu/st20.rb +9 -0
- data/metasm/cpu/st20/decode.rb +173 -0
- data/metasm/cpu/st20/decompile.rb +283 -0
- data/metasm/cpu/st20/main.rb +37 -0
- data/metasm/cpu/st20/opcodes.rb +140 -0
- data/{lib/metasm/arm.rb → metasm/cpu/webasm.rb} +4 -5
- 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.rb +15 -0
- data/{lib/metasm → metasm/cpu}/x86_64/compile_c.rb +40 -25
- data/{lib/metasm → metasm/cpu}/x86_64/debug.rb +4 -4
- data/{lib/metasm → metasm/cpu}/x86_64/decode.rb +58 -15
- data/{lib/metasm → metasm/cpu}/x86_64/encode.rb +59 -28
- data/{lib/metasm → metasm/cpu}/x86_64/main.rb +18 -6
- data/metasm/cpu/x86_64/opcodes.rb +138 -0
- data/{lib/metasm → metasm/cpu}/x86_64/parse.rb +12 -4
- data/metasm/cpu/x86_64/render.rb +35 -0
- data/metasm/cpu/z80.rb +9 -0
- data/metasm/cpu/z80/decode.rb +286 -0
- data/metasm/cpu/z80/main.rb +67 -0
- data/metasm/cpu/z80/opcodes.rb +224 -0
- data/metasm/cpu/z80/render.rb +48 -0
- data/{lib/metasm/os/main.rb → metasm/debug.rb} +201 -407
- data/{lib/metasm → metasm}/decode.rb +104 -24
- data/{lib/metasm → metasm}/decompile.rb +804 -478
- data/{lib/metasm → metasm}/disassemble.rb +385 -170
- data/{lib/metasm → metasm}/disassemble_api.rb +684 -105
- data/{lib/metasm → metasm}/dynldr.rb +231 -138
- data/{lib/metasm → metasm}/encode.rb +20 -5
- data/{lib/metasm → metasm}/exe_format/a_out.rb +9 -6
- data/{lib/metasm → metasm}/exe_format/autoexe.rb +3 -0
- data/{lib/metasm → metasm}/exe_format/bflt.rb +57 -27
- data/{lib/metasm → metasm}/exe_format/coff.rb +35 -7
- data/{lib/metasm → metasm}/exe_format/coff_decode.rb +70 -23
- data/{lib/metasm → metasm}/exe_format/coff_encode.rb +24 -22
- data/{lib/metasm → metasm}/exe_format/dex.rb +26 -8
- data/{lib/metasm → metasm}/exe_format/dol.rb +1 -0
- data/{lib/metasm → metasm}/exe_format/elf.rb +108 -58
- data/{lib/metasm → metasm}/exe_format/elf_decode.rb +202 -36
- data/{lib/metasm → metasm}/exe_format/elf_encode.rb +126 -32
- data/metasm/exe_format/gb.rb +65 -0
- data/metasm/exe_format/javaclass.rb +424 -0
- data/{lib/metasm → metasm}/exe_format/macho.rb +218 -16
- data/{lib/metasm → metasm}/exe_format/main.rb +28 -3
- data/{lib/metasm → metasm}/exe_format/mz.rb +2 -0
- data/{lib/metasm → metasm}/exe_format/nds.rb +7 -4
- data/{lib/metasm → metasm}/exe_format/pe.rb +96 -11
- data/metasm/exe_format/pyc.rb +167 -0
- data/{lib/metasm → metasm}/exe_format/serialstruct.rb +67 -14
- data/{lib/metasm → metasm}/exe_format/shellcode.rb +7 -3
- data/metasm/exe_format/shellcode_rwx.rb +114 -0
- data/metasm/exe_format/swf.rb +205 -0
- data/metasm/exe_format/wasm.rb +402 -0
- data/{lib/metasm → metasm}/exe_format/xcoff.rb +7 -7
- data/metasm/exe_format/zip.rb +335 -0
- data/metasm/gui.rb +13 -0
- data/{lib/metasm → metasm}/gui/cstruct.rb +35 -41
- data/{lib/metasm → metasm}/gui/dasm_coverage.rb +11 -11
- data/{lib/metasm → metasm}/gui/dasm_decomp.rb +177 -114
- data/{lib/metasm → metasm}/gui/dasm_funcgraph.rb +0 -0
- data/metasm/gui/dasm_graph.rb +1754 -0
- data/{lib/metasm → metasm}/gui/dasm_hex.rb +16 -12
- data/{lib/metasm → metasm}/gui/dasm_listing.rb +43 -28
- data/{lib/metasm → metasm}/gui/dasm_main.rb +360 -77
- data/{lib/metasm → metasm}/gui/dasm_opcodes.rb +5 -19
- data/{lib/metasm → metasm}/gui/debug.rb +109 -34
- data/{lib/metasm → metasm}/gui/gtk.rb +174 -44
- data/{lib/metasm → metasm}/gui/qt.rb +14 -4
- data/{lib/metasm → metasm}/gui/win32.rb +180 -43
- data/{lib/metasm → metasm}/gui/x11.rb +59 -59
- data/{lib/metasm → metasm}/main.rb +421 -286
- data/metasm/os/emulator.rb +175 -0
- data/{lib/metasm/os/remote.rb → metasm/os/gdbremote.rb} +146 -54
- data/{lib/metasm → metasm}/os/gnu_exports.rb +1 -1
- data/{lib/metasm → metasm}/os/linux.rb +628 -151
- data/metasm/os/main.rb +335 -0
- data/{lib/metasm → metasm}/os/windows.rb +151 -58
- data/{lib/metasm → metasm}/os/windows_exports.rb +141 -0
- data/{lib/metasm → metasm}/parse.rb +49 -36
- data/{lib/metasm → metasm}/parse_c.rb +405 -246
- data/{lib/metasm → metasm}/preprocessor.rb +71 -41
- data/{lib/metasm → metasm}/render.rb +14 -38
- data/misc/hexdump.rb +4 -3
- data/misc/lint.rb +58 -0
- data/misc/objdiff.rb +4 -1
- data/misc/objscan.rb +1 -1
- data/misc/openrisc-parser.rb +79 -0
- data/misc/txt2html.rb +9 -7
- data/samples/bindiff.rb +3 -4
- data/samples/dasm-plugins/bindiff.rb +15 -0
- data/samples/dasm-plugins/bookmark.rb +133 -0
- data/samples/dasm-plugins/c_constants.rb +57 -0
- data/samples/dasm-plugins/colortheme_solarized.rb +125 -0
- data/samples/dasm-plugins/cppobj_funcall.rb +60 -0
- data/samples/dasm-plugins/dasm_all.rb +70 -0
- data/samples/dasm-plugins/demangle_cpp.rb +31 -0
- data/samples/dasm-plugins/deobfuscate.rb +251 -0
- data/samples/dasm-plugins/dump_text.rb +35 -0
- data/samples/dasm-plugins/export_graph_svg.rb +86 -0
- data/samples/dasm-plugins/findgadget.rb +75 -0
- data/samples/dasm-plugins/hl_opcode.rb +32 -0
- data/samples/dasm-plugins/hotfix_gtk_dbg.rb +19 -0
- data/samples/dasm-plugins/imm2off.rb +34 -0
- data/samples/dasm-plugins/match_libsigs.rb +93 -0
- data/samples/dasm-plugins/patch_file.rb +95 -0
- data/samples/dasm-plugins/scanfuncstart.rb +36 -0
- data/samples/dasm-plugins/scanxrefs.rb +29 -0
- data/samples/dasm-plugins/selfmodify.rb +197 -0
- data/samples/dasm-plugins/stringsxrefs.rb +28 -0
- data/samples/dasmnavig.rb +1 -1
- data/samples/dbg-apihook.rb +24 -9
- data/samples/dbg-plugins/heapscan.rb +283 -0
- data/samples/dbg-plugins/heapscan/compiled_heapscan_lin.c +155 -0
- data/samples/dbg-plugins/heapscan/compiled_heapscan_win.c +128 -0
- data/samples/dbg-plugins/heapscan/graphheap.rb +616 -0
- data/samples/dbg-plugins/heapscan/heapscan.rb +709 -0
- data/samples/dbg-plugins/heapscan/winheap.h +174 -0
- data/samples/dbg-plugins/heapscan/winheap7.h +307 -0
- data/samples/dbg-plugins/trace_func.rb +214 -0
- data/samples/disassemble-gui.rb +48 -7
- data/samples/disassemble.rb +31 -6
- data/samples/dump_upx.rb +24 -12
- data/samples/dynamic_ruby.rb +35 -27
- data/samples/elfencode.rb +15 -0
- data/samples/emubios.rb +251 -0
- data/samples/emudbg.rb +127 -0
- data/samples/exeencode.rb +6 -5
- data/samples/factorize-headers-peimports.rb +1 -1
- data/samples/lindebug.rb +186 -391
- data/samples/metasm-shell.rb +68 -57
- data/samples/peldr.rb +2 -2
- data/tests/all.rb +1 -1
- data/tests/arc.rb +26 -0
- data/tests/dynldr.rb +22 -4
- data/tests/expression.rb +57 -0
- data/tests/graph_layout.rb +285 -0
- data/tests/ia32.rb +80 -26
- data/tests/mcs51.rb +27 -0
- data/tests/mips.rb +10 -3
- data/tests/preprocessor.rb +18 -0
- data/tests/x86_64.rb +66 -18
- metadata +465 -219
- metadata.gz.sig +2 -0
- data/lib/metasm/arm/opcodes.rb +0 -177
- data/lib/metasm/gui.rb +0 -23
- data/lib/metasm/gui/dasm_graph.rb +0 -1354
- data/lib/metasm/ia32.rb +0 -14
- data/lib/metasm/ia32/opcodes.rb +0 -872
- data/lib/metasm/ppc/parse.rb +0 -52
- data/lib/metasm/x86_64.rb +0 -12
- data/lib/metasm/x86_64/opcodes.rb +0 -118
- data/samples/gdbclient.rb +0 -583
- data/samples/rubstop.rb +0 -399
@@ -4,17 +4,26 @@
|
|
4
4
|
# Licence is LGPL, see LICENCE in the top-level directory
|
5
5
|
|
6
6
|
|
7
|
-
require 'metasm/ia32/main'
|
7
|
+
require 'metasm/cpu/ia32/main'
|
8
8
|
|
9
9
|
module Metasm
|
10
10
|
class Ia32
|
11
11
|
# temporarily setup dasm.address_binding so that backtracking
|
12
12
|
# stack-related offsets resolve in :frameptr (relative to func start)
|
13
13
|
def decompile_makestackvars(dasm, funcstart, blocks)
|
14
|
+
esp = register_symbols[4]
|
14
15
|
oldfuncbd = dasm.address_binding[funcstart]
|
15
|
-
dasm.address_binding[funcstart] = {
|
16
|
-
|
16
|
+
dasm.address_binding[funcstart] = { esp => :frameptr }
|
17
17
|
patched_binding = [funcstart] # list of addresses to cleanup later
|
18
|
+
|
19
|
+
if blocks.length <= 12
|
20
|
+
blocks.each { |block| yield block }
|
21
|
+
return
|
22
|
+
end
|
23
|
+
|
24
|
+
# for large function, pre-trace and cache esp/ebp for every block start to improve decompilation time
|
25
|
+
|
26
|
+
ebp = register_symbols[5]
|
18
27
|
ebp_frame = true
|
19
28
|
|
20
29
|
# pretrace esp and ebp for each function block (cleared later)
|
@@ -24,16 +33,16 @@ class Ia32
|
|
24
33
|
if not dasm.address_binding[blockstart]
|
25
34
|
patched_binding << blockstart
|
26
35
|
dasm.address_binding[blockstart] = {}
|
27
|
-
foo = dasm.backtrace(
|
36
|
+
foo = dasm.backtrace(esp, blockstart, :snapshot_addr => funcstart)
|
28
37
|
if foo.length == 1 and ee = foo.first and ee.kind_of? Expression and (ee == Expression[:frameptr] or
|
29
38
|
(ee.lexpr == :frameptr and ee.op == :+ and ee.rexpr.kind_of? ::Integer))
|
30
|
-
dasm.address_binding[blockstart][
|
39
|
+
dasm.address_binding[blockstart][esp] = ee
|
31
40
|
end
|
32
41
|
if ebp_frame
|
33
|
-
foo = dasm.backtrace(
|
42
|
+
foo = dasm.backtrace(ebp, blockstart, :snapshot_addr => funcstart)
|
34
43
|
if foo.length == 1 and ee = foo.first and ee.kind_of? Expression and (ee == Expression[:frameptr] or
|
35
44
|
(ee.lexpr == :frameptr and ee.op == :+ and ee.rexpr.kind_of? ::Integer))
|
36
|
-
dasm.address_binding[blockstart][
|
45
|
+
dasm.address_binding[blockstart][ebp] = ee
|
37
46
|
else
|
38
47
|
ebp_frame = false # func does not use ebp as frame ptr, no need to bt for later blocks
|
39
48
|
end
|
@@ -48,11 +57,16 @@ class Ia32
|
|
48
57
|
dasm.address_binding[funcstart] = oldfuncbd if oldfuncbd
|
49
58
|
end
|
50
59
|
|
60
|
+
# add di-specific registry written/accessed
|
61
|
+
def decompile_func_finddeps_di(dcmp, func, di, a, w)
|
62
|
+
a << register_symbols[0] if di.opcode.name == 'ret' and (not func.type.kind_of? C::BaseType or func.type.type.name != :void) # standard ABI
|
63
|
+
end
|
64
|
+
|
51
65
|
# list variable dependency for each block, remove useless writes
|
52
66
|
# returns { blockaddr => [list of vars that are needed by a following block] }
|
53
67
|
def decompile_func_finddeps(dcmp, blocks, func)
|
54
68
|
deps_r = {} ; deps_w = {} ; deps_to = {}
|
55
|
-
deps_subfunc = {}
|
69
|
+
deps_subfunc = {} # things read/written by subfuncs
|
56
70
|
|
57
71
|
# find read/writes by each block
|
58
72
|
blocks.each { |b, to|
|
@@ -69,8 +83,8 @@ class Ia32
|
|
69
83
|
else a |= Expression[k].externals # if dword [eax] <- 42, eax is read
|
70
84
|
end
|
71
85
|
}
|
72
|
-
|
73
|
-
|
86
|
+
decompile_func_finddeps_di(dcmp, func, di, a, w)
|
87
|
+
|
74
88
|
deps_r[b] |= a.map { |ee| Expression[ee].externals.grep(::Symbol) }.flatten - [:unknown] - deps_w[b]
|
75
89
|
deps_w[b] |= w.map { |ee| Expression[ee].externals.grep(::Symbol) }.flatten - [:unknown]
|
76
90
|
}
|
@@ -97,7 +111,7 @@ class Ia32
|
|
97
111
|
}
|
98
112
|
if stackoff # last block instr == subfunction call
|
99
113
|
deps_r[b] |= deps_subfunc[b] - deps_w[b]
|
100
|
-
deps_w[b] |= [
|
114
|
+
deps_w[b] |= register_symbols[0, 3] # standard ABI
|
101
115
|
end
|
102
116
|
}
|
103
117
|
|
@@ -120,13 +134,13 @@ class Ia32
|
|
120
134
|
else a |= Expression[k].externals # if dword [eax] <- 42, eax is read
|
121
135
|
end
|
122
136
|
}
|
123
|
-
|
124
|
-
|
137
|
+
decompile_func_finddeps_di(dcmp, func, di, a, w)
|
138
|
+
|
125
139
|
next true if (a.map { |ee| Expression[ee].externals.grep(::Symbol) }.flatten - [:unknown] - bw).include? r
|
126
140
|
bw |= w.map { |ee| Expression[ee].externals.grep(::Symbol) }.flatten - [:unknown]
|
127
141
|
false
|
128
142
|
}
|
129
|
-
if r ==
|
143
|
+
if r == register_symbols[0] and (rdi || blk.list.last).opcode.name == 'ret'
|
130
144
|
func.type.type = C::BaseType.new(:void)
|
131
145
|
false
|
132
146
|
elsif rdi and rdi.backtrace_binding[r]
|
@@ -180,15 +194,18 @@ class Ia32
|
|
180
194
|
end
|
181
195
|
|
182
196
|
def decompile_blocks(dcmp, myblocks, deps, func, nextaddr = nil)
|
197
|
+
eax, ecx, edx, ebx, esp, ebp, esi, edi = register_symbols
|
198
|
+
ebx, esp, ebp = ebx, esp, ebp # fix ruby unused var warning
|
183
199
|
scope = func.initializer
|
184
200
|
func.type.args.each { |a| scope.symbol[a.name] = a }
|
185
201
|
stmts = scope.statements
|
186
202
|
blocks_toclean = myblocks.dup
|
187
203
|
func_entry = myblocks.first[0]
|
204
|
+
di_addr = nil
|
188
205
|
until myblocks.empty?
|
189
206
|
b, to = myblocks.shift
|
190
207
|
if l = dcmp.dasm.get_label_at(b)
|
191
|
-
stmts << C::Label.new(l)
|
208
|
+
stmts << C::Label.new(l).with_misc(:di_addr => b)
|
192
209
|
end
|
193
210
|
|
194
211
|
# list of assignments [[dest reg, expr assigned]]
|
@@ -196,7 +213,11 @@ class Ia32
|
|
196
213
|
# reg binding (reg => value, values.externals = regs at block start)
|
197
214
|
binding = {}
|
198
215
|
# Expr => CExpr
|
199
|
-
ce = lambda { |*e|
|
216
|
+
ce = lambda { |*e|
|
217
|
+
ret = dcmp.decompile_cexpr(Expression[Expression[*e].reduce], scope)
|
218
|
+
dcmp.walk_ce(ret) { |ee| ee.with_misc(:di_addr => di_addr) } if di_addr
|
219
|
+
ret
|
220
|
+
}
|
200
221
|
# Expr => Expr.bind(binding) => CExpr
|
201
222
|
ceb = lambda { |*e| ce[Expression[*e].bind(binding)] }
|
202
223
|
|
@@ -221,6 +242,7 @@ class Ia32
|
|
221
242
|
# returns an array to use as funcall arguments
|
222
243
|
get_func_args = lambda { |di, f|
|
223
244
|
# XXX see remarks in #finddeps
|
245
|
+
# TODO x64
|
224
246
|
bt = dcmp.dasm.backtrace(:esp, di.address, :snapshot_addr => func_entry, :include_start => true)
|
225
247
|
stackoff = Expression[[bt, :+, @size/8], :-, :esp].bind(:esp => :frameptr).reduce rescue nil
|
226
248
|
args_todo = f.type.args.to_a.dup
|
@@ -269,19 +291,20 @@ class Ia32
|
|
269
291
|
|
270
292
|
# go !
|
271
293
|
dcmp.dasm.decoded[b].block.list.each_with_index { |di, didx|
|
294
|
+
di_addr = di.address
|
272
295
|
a = di.instruction.args
|
273
296
|
if di.opcode.props[:setip] and not di.opcode.props[:stopexec]
|
274
297
|
# conditional jump
|
275
298
|
commit[]
|
276
299
|
n = dcmp.backtrace_target(get_xrefs_x(dcmp.dasm, di).first, di.address)
|
277
300
|
if di.opcode.name =~ /^loop(.+)?/
|
278
|
-
cx = C::CExpression[:'--', ceb[
|
301
|
+
cx = C::CExpression[:'--', ceb[ecx]]
|
279
302
|
cc = $1 ? C::CExpression[cx, :'&&', ceb[decode_cc_to_expr($1)]] : cx
|
280
303
|
else
|
281
304
|
cc = ceb[decode_cc_to_expr(di.opcode.name[1..-1])]
|
282
305
|
end
|
283
306
|
# XXX switch/indirect/multiple jmp
|
284
|
-
stmts << C::If.new(C::CExpression[cc], C::Goto.new(n))
|
307
|
+
stmts << C::If.new(C::CExpression[cc], C::Goto.new(n).with_misc(:di_addr => di_addr)).with_misc(:di_addr => di_addr)
|
285
308
|
to.delete dcmp.dasm.normalize(n)
|
286
309
|
next
|
287
310
|
end
|
@@ -290,7 +313,7 @@ class Ia32
|
|
290
313
|
# mov cr0 etc
|
291
314
|
a1, a2 = di.instruction.args
|
292
315
|
case a1
|
293
|
-
when Ia32::CtrlReg, Ia32::DbgReg, Ia32::SegReg
|
316
|
+
when Ia32::CtrlReg, Ia32::DbgReg, Ia32::TstReg, Ia32::SegReg
|
294
317
|
sz = a1.kind_of?(Ia32::SegReg) ? 16 : 32
|
295
318
|
if not dcmp.c_parser.toplevel.symbol["intrinsic_set_#{a1}"]
|
296
319
|
dcmp.c_parser.parse("void intrinsic_set_#{a1}(__int#{sz});")
|
@@ -298,11 +321,11 @@ class Ia32
|
|
298
321
|
f = dcmp.c_parser.toplevel.symbol["intrinsic_set_#{a1}"]
|
299
322
|
a2 = a2.symbolic(di)
|
300
323
|
a2 = [a2, :&, 0xffff] if sz == 16
|
301
|
-
stmts << C::CExpression.new(f, :funcall, [ceb[a2]], f.type.type)
|
324
|
+
stmts << C::CExpression.new(f, :funcall, [ceb[a2]], f.type.type).with_misc(:di_addr => di_addr)
|
302
325
|
next
|
303
326
|
end
|
304
327
|
case a2
|
305
|
-
when Ia32::CtrlReg, Ia32::DbgReg, Ia32::SegReg
|
328
|
+
when Ia32::CtrlReg, Ia32::DbgReg, Ia32::TstReg, Ia32::SegReg
|
306
329
|
if not dcmp.c_parser.toplevel.symbol["intrinsic_get_#{a2}"]
|
307
330
|
sz = a2.kind_of?(Ia32::SegReg) ? 16 : 32
|
308
331
|
dcmp.c_parser.parse("__int#{sz} intrinsic_get_#{a2}(void);")
|
@@ -310,7 +333,7 @@ class Ia32
|
|
310
333
|
f = dcmp.c_parser.toplevel.symbol["intrinsic_get_#{a2}"]
|
311
334
|
t = f.type.type
|
312
335
|
binding.delete a1.symbolic(di)
|
313
|
-
stmts << C::CExpression.new(ce[a1.symbolic(di)], :'=', C::CExpression.new(f, :funcall, [], t), t)
|
336
|
+
stmts << C::CExpression.new(ce[a1.symbolic(di)], :'=', C::CExpression.new(f, :funcall, [], t).with_misc(:di_addr => di_addr), t).with_misc(:di_addr => di_addr)
|
314
337
|
next
|
315
338
|
end
|
316
339
|
end
|
@@ -319,8 +342,8 @@ class Ia32
|
|
319
342
|
when 'ret'
|
320
343
|
commit[]
|
321
344
|
ret = nil
|
322
|
-
ret = C::CExpression[ceb[
|
323
|
-
stmts << C::Return.new(ret)
|
345
|
+
ret = C::CExpression[ceb[eax]] unless func.type.type.kind_of? C::BaseType and func.type.type.name == :void
|
346
|
+
stmts << C::Return.new(ret).with_misc(:di_addr => di_addr)
|
324
347
|
when 'call' # :saveip
|
325
348
|
n = dcmp.backtrace_target(get_xrefs_x(dcmp.dasm, di).first, di.address)
|
326
349
|
args = []
|
@@ -353,9 +376,9 @@ class Ia32
|
|
353
376
|
end
|
354
377
|
end
|
355
378
|
commit[]
|
356
|
-
binding.delete
|
357
|
-
e = C::CExpression[f, :funcall, args]
|
358
|
-
e = C::CExpression[ce[
|
379
|
+
binding.delete eax
|
380
|
+
e = C::CExpression[f, :funcall, args].with_misc(:di_addr => di_addr)
|
381
|
+
e = C::CExpression[ce[eax], :'=', e, f.type.type].with_misc(:di_addr => di_addr) if deps[b].include? eax and f.type.type != C::BaseType.new(:void)
|
359
382
|
stmts << e
|
360
383
|
when 'jmp'
|
361
384
|
#if di.comment.to_a.include? 'switch'
|
@@ -369,12 +392,12 @@ class Ia32
|
|
369
392
|
# to.delete addr
|
370
393
|
# next if not l = dcmp.dasm.get_label_at(addr)
|
371
394
|
# sw.body.statements << C::Goto.new(l)
|
372
|
-
|
395
|
+
# }
|
373
396
|
# stmts << sw
|
374
397
|
a = di.instruction.args.first
|
375
398
|
if a.kind_of? Expression
|
376
399
|
elsif not a.respond_to? :symbolic
|
377
|
-
stmts << C::Asm.new(di.instruction.to_s, nil, [], [], nil, nil)
|
400
|
+
stmts << C::Asm.new(di.instruction.to_s, nil, [], [], nil, nil).with_misc(:di_addr => di_addr)
|
378
401
|
else
|
379
402
|
n = di.instruction.args.first.symbolic(di)
|
380
403
|
fptr = ceb[n]
|
@@ -385,12 +408,10 @@ class Ia32
|
|
385
408
|
args = get_func_args[di, fptr.type]
|
386
409
|
else
|
387
410
|
proto = C::Function.new(C::BaseType.new(:void))
|
388
|
-
fptr = C::CExpression[[fptr], C::Pointer.new(proto)]
|
411
|
+
fptr = C::CExpression[[fptr], C::Pointer.new(proto)].with_misc(:di_addr => di_addr)
|
389
412
|
args = []
|
390
413
|
end
|
391
|
-
ret = C::Return.new(C::CExpression[fptr, :funcall, args])
|
392
|
-
class << ret ; attr_accessor :from_instr end
|
393
|
-
ret.from_instr = di
|
414
|
+
ret = C::Return.new(C::CExpression[fptr, :funcall, args].with_misc(:di_addr => di_addr)).with_misc(:di_addr => di_addr)
|
394
415
|
stmts << ret
|
395
416
|
to = []
|
396
417
|
end
|
@@ -404,7 +425,7 @@ class Ia32
|
|
404
425
|
end
|
405
426
|
# need a way to transform arg => :frameptr+12
|
406
427
|
arg = di.backtrace_binding.keys.grep(Indirection).first.pointer
|
407
|
-
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol['intrinsic_lgdt'], :funcall, [ceb[arg]], C::BaseType.new(:void))
|
428
|
+
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol['intrinsic_lgdt'], :funcall, [ceb[arg]], C::BaseType.new(:void)).with_misc(:di_addr => di_addr)
|
408
429
|
when 'lidt'
|
409
430
|
if not dcmp.c_parser.toplevel.struct['interrupt_descriptor']
|
410
431
|
dcmp.c_parser.parse('struct interrupt_descriptor { __int16 offset0_16; __int16 segment; __int16 flags; __int16 offset16_32; };')
|
@@ -414,29 +435,29 @@ class Ia32
|
|
414
435
|
dcmp.c_parser.parse('void intrinsic_lidt(struct interrupt_table *);')
|
415
436
|
end
|
416
437
|
arg = di.backtrace_binding.keys.grep(Indirection).first.pointer
|
417
|
-
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol['intrinsic_lidt'], :funcall, [ceb[arg]], C::BaseType.new(:void))
|
438
|
+
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol['intrinsic_lidt'], :funcall, [ceb[arg]], C::BaseType.new(:void)).with_misc(:di_addr => di_addr)
|
418
439
|
when 'ltr', 'lldt'
|
419
440
|
if not dcmp.c_parser.toplevel.symbol["intrinsic_#{di.opcode.name}"]
|
420
441
|
dcmp.c_parser.parse("void intrinsic_#{di.opcode.name}(int);")
|
421
442
|
end
|
422
443
|
arg = di.backtrace_binding.keys.first
|
423
|
-
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol["intrinsic_#{di.opcode.name}"], :funcall, [ceb[arg]], C::BaseType.new(:void))
|
444
|
+
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol["intrinsic_#{di.opcode.name}"], :funcall, [ceb[arg]], C::BaseType.new(:void)).with_misc(:di_addr => di_addr)
|
424
445
|
when 'out'
|
425
446
|
sz = di.instruction.args.find { |a_| a_.kind_of? Ia32::Reg and a_.val == 0 }.sz
|
426
447
|
if not dcmp.c_parser.toplevel.symbol["intrinsic_out#{sz}"]
|
427
448
|
dcmp.c_parser.parse("void intrinsic_out#{sz}(unsigned short port, __int#{sz} value);")
|
428
449
|
end
|
429
|
-
port = di.instruction.args.grep(Expression).first ||
|
430
|
-
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol["intrinsic_out#{sz}"], :funcall, [ceb[port], ceb[
|
450
|
+
port = di.instruction.args.grep(Expression).first || edx
|
451
|
+
stmts << C::CExpression.new(dcmp.c_parser.toplevel.symbol["intrinsic_out#{sz}"], :funcall, [ceb[port], ceb[eax]], C::BaseType.new(:void)).with_misc(:di_addr => di_addr)
|
431
452
|
when 'in'
|
432
453
|
sz = di.instruction.args.find { |a_| a_.kind_of? Ia32::Reg and a_.val == 0 }.sz
|
433
454
|
if not dcmp.c_parser.toplevel.symbol["intrinsic_in#{sz}"]
|
434
455
|
dcmp.c_parser.parse("__int#{sz} intrinsic_in#{sz}(unsigned short port);")
|
435
456
|
end
|
436
|
-
port = di.instruction.args.grep(Expression).first ||
|
457
|
+
port = di.instruction.args.grep(Expression).first || edx
|
437
458
|
f = dcmp.c_parser.toplevel.symbol["intrinsic_in#{sz}"]
|
438
|
-
binding.delete
|
439
|
-
stmts << C::CExpression.new(ce[
|
459
|
+
binding.delete eax
|
460
|
+
stmts << C::CExpression.new(ce[eax], :'=', C::CExpression.new(f, :funcall, [ceb[port]], f.type.type), f.type.type).with_misc(:di_addr => di_addr)
|
440
461
|
when 'sti', 'cli'
|
441
462
|
stmts << C::Asm.new(di.instruction.to_s, nil, [], [], nil, nil)
|
442
463
|
when /^(mov|sto|lod)s([bwdq])/
|
@@ -448,15 +469,15 @@ class Ia32
|
|
448
469
|
blk = C::Block.new(scope)
|
449
470
|
case op
|
450
471
|
when 'mov'
|
451
|
-
blk.statements << C::CExpression[[:*, [[ceb[
|
452
|
-
blk.statements << C::CExpression[ceb[
|
453
|
-
blk.statements << C::CExpression[ceb[
|
472
|
+
blk.statements << C::CExpression[[:*, [[ceb[edi]], pt]], :'=', [:*, [[ceb[esi]], pt]]].with_misc(:di_addr => di_addr)
|
473
|
+
blk.statements << C::CExpression[ceb[edi], :'=', [ceb[edi], :+, [sz]]].with_misc(:di_addr => di_addr)
|
474
|
+
blk.statements << C::CExpression[ceb[esi], :'=', [ceb[esi], :+, [sz]]].with_misc(:di_addr => di_addr)
|
454
475
|
when 'sto'
|
455
|
-
blk.statements << C::CExpression[[:*, [[ceb[
|
456
|
-
blk.statements << C::CExpression[ceb[
|
476
|
+
blk.statements << C::CExpression[[:*, [[ceb[edi]], pt]], :'=', ceb[eax]].with_misc(:di_addr => di_addr)
|
477
|
+
blk.statements << C::CExpression[ceb[edi], :'=', [ceb[edi], :+, [sz]]].with_misc(:di_addr => di_addr)
|
457
478
|
when 'lod'
|
458
|
-
blk.statements << C::CExpression[ceb[
|
459
|
-
blk.statements << C::CExpression[ceb[
|
479
|
+
blk.statements << C::CExpression[ceb[eax], :'=', [:*, [[ceb[esi]], pt]]].with_misc(:di_addr => di_addr)
|
480
|
+
blk.statements << C::CExpression[ceb[esi], :'=', [ceb[esi], :+, [sz]]].with_misc(:di_addr => di_addr)
|
460
481
|
#when 'sca'
|
461
482
|
#when 'cmp'
|
462
483
|
end
|
@@ -465,8 +486,8 @@ class Ia32
|
|
465
486
|
when nil
|
466
487
|
stmts.concat blk.statements
|
467
488
|
when 'rep'
|
468
|
-
blk.statements << C::CExpression[ceb[
|
469
|
-
stmts << C::While.new(C::CExpression[ceb[
|
489
|
+
blk.statements << C::CExpression[ceb[ecx], :'=', [ceb[ecx], :-, [1]]].with_misc(:di_addr => di_addr)
|
490
|
+
stmts << C::While.new(C::CExpression[ceb[ecx]], blk).with_misc(:di_addr => di_addr)
|
470
491
|
#when 'repz' # sca/cmp only
|
471
492
|
#when 'repnz'
|
472
493
|
end
|
@@ -475,7 +496,7 @@ class Ia32
|
|
475
496
|
bd = get_fwdemu_binding(di)
|
476
497
|
if di.backtrace_binding[:incomplete_binding]
|
477
498
|
commit[]
|
478
|
-
stmts << C::Asm.new(di.instruction.to_s, nil, nil, nil, nil, nil)
|
499
|
+
stmts << C::Asm.new(di.instruction.to_s, nil, nil, nil, nil, nil).with_misc(:di_addr => di_addr)
|
479
500
|
else
|
480
501
|
update = {}
|
481
502
|
bd.each { |k, v|
|
@@ -490,6 +511,7 @@ class Ia32
|
|
490
511
|
binding.update update
|
491
512
|
end
|
492
513
|
end
|
514
|
+
di_addr = nil
|
493
515
|
}
|
494
516
|
commit[]
|
495
517
|
|
@@ -512,9 +534,9 @@ class Ia32
|
|
512
534
|
dcmp.dasm.decoded[b_].block.list.each { |di|
|
513
535
|
di.backtrace_binding = nil
|
514
536
|
}
|
515
|
-
|
537
|
+
}
|
516
538
|
end
|
517
|
-
|
539
|
+
|
518
540
|
def decompile_check_abi(dcmp, entry, func)
|
519
541
|
a = func.type.args || []
|
520
542
|
a.delete_if { |arg| arg.has_attribute_var('register') and arg.has_attribute('unused') }
|
@@ -4,7 +4,7 @@
|
|
4
4
|
# Licence is LGPL, see LICENCE in the top-level directory
|
5
5
|
|
6
6
|
|
7
|
-
require 'metasm/ia32/opcodes'
|
7
|
+
require 'metasm/cpu/ia32/opcodes'
|
8
8
|
require 'metasm/encode'
|
9
9
|
|
10
10
|
module Metasm
|
@@ -195,27 +195,28 @@ class Ia32
|
|
195
195
|
case k
|
196
196
|
when :jmp; {:jmp => 0x3e, :nojmp => 0x2e}[v]
|
197
197
|
when :lock; 0xf0
|
198
|
-
when :rep; {'repnz' => 0xf2, 'repz' => 0xf3, 'rep' => 0xf2}[v]
|
198
|
+
when :rep; {'repnz' => 0xf2, 'repz' => 0xf3, 'rep' => 0xf2}[v]
|
199
|
+
when :jmphint; {'hintjmp' => 0x3e, 'hintnojmp' => 0x2e}[v]
|
200
|
+
when :seg; [0x26, 0x2E, 0x36, 0x3E, 0x64, 0x65][v.val]
|
199
201
|
end
|
200
202
|
}.compact.pack 'C*'
|
201
|
-
pfx << op.props[:needpfx] if op.props[:needpfx]
|
202
203
|
|
203
204
|
if op.name == 'movsx' or op.name == 'movzx'
|
204
205
|
pfx << 0x66 if size == 48-i.args[0].sz
|
206
|
+
elsif op.name == 'crc32'
|
207
|
+
pfx << 0x66 if size == 48-i.args[1].sz
|
205
208
|
else
|
206
209
|
opsz = op.props[:argsz]
|
207
210
|
oi.each { |oa, ia|
|
208
211
|
case oa
|
209
|
-
when :reg, :reg_eax, :modrm, :
|
212
|
+
when :reg, :reg_eax, :modrm, :mrm_imm
|
210
213
|
raise EncodeError, "Incompatible arg size in #{i}" if ia.sz and opsz and opsz != ia.sz
|
211
214
|
opsz = ia.sz
|
212
215
|
end
|
213
216
|
}
|
214
|
-
pfx << 0x66 if (
|
215
|
-
|
216
|
-
|
217
|
-
opsz = op.props[:opsz]
|
218
|
-
end
|
217
|
+
pfx << 0x66 if (op.props[:opsz] and size == 48 - op.props[:opsz]) or
|
218
|
+
(not op.props[:argsz] and opsz and size == 48 - opsz)
|
219
|
+
opsz ||= op.props[:opsz]
|
219
220
|
end
|
220
221
|
opsz ||= size
|
221
222
|
|
@@ -226,7 +227,7 @@ class Ia32
|
|
226
227
|
adsz ||= size
|
227
228
|
# addrsize override / segment override
|
228
229
|
if mrm = i.args.grep(ModRM).first
|
229
|
-
if not op.props[:adsz] and ((mrm.b and mrm.b.sz
|
230
|
+
if not op.props[:adsz] and ((mrm.b and mrm.b.sz == 48 - adsz) or (mrm.i and mrm.i.sz == 48 - adsz))
|
230
231
|
pfx << 0x67
|
231
232
|
adsz = 48 - adsz
|
232
233
|
end
|
@@ -240,10 +241,12 @@ class Ia32
|
|
240
241
|
postponed = []
|
241
242
|
oi.each { |oa, ia|
|
242
243
|
case oa
|
243
|
-
when :reg, :seg3, :seg3A, :seg2, :seg2A, :eeec, :eeed, :regfp, :regmmx, :regxmm
|
244
|
+
when :reg, :seg3, :seg3A, :seg2, :seg2A, :eeec, :eeed, :eeet, :regfp, :regmmx, :regxmm, :regymm
|
244
245
|
# field arg
|
245
246
|
set_field[oa, ia.val]
|
246
247
|
pfx << 0x66 if oa == :regmmx and op.props[:xmmx] and ia.sz == 128
|
248
|
+
when :vexvreg, :vexvxmm, :vexvymm
|
249
|
+
set_field[:vex_vvvv, ia.val ^ 0xf]
|
247
250
|
when :imm_val1, :imm_val3, :reg_cl, :reg_eax, :reg_dx, :regfp0
|
248
251
|
# implicit
|
249
252
|
else
|
@@ -251,7 +254,7 @@ class Ia32
|
|
251
254
|
end
|
252
255
|
}
|
253
256
|
|
254
|
-
if !(op.args & [:modrm, :
|
257
|
+
if !(op.args & [:modrm, :modrmmmx, :modrmxmm, :modrmymm]).empty?
|
255
258
|
# reg field of modrm
|
256
259
|
regval = (base[-1] >> 3) & 7
|
257
260
|
base.pop
|
@@ -265,6 +268,8 @@ class Ia32
|
|
265
268
|
postponed.first[1] = Expression[target, :-, postlabel]
|
266
269
|
end
|
267
270
|
|
271
|
+
pfx << op.props[:needpfx] if op.props[:needpfx]
|
272
|
+
|
268
273
|
#
|
269
274
|
# append other arguments
|
270
275
|
#
|
@@ -273,7 +278,7 @@ class Ia32
|
|
273
278
|
postponed.each { |oa, ia|
|
274
279
|
case oa
|
275
280
|
when :farptr; ed = ia.encode(@endianness, "a#{opsz}".to_sym)
|
276
|
-
when :modrm, :
|
281
|
+
when :modrm, :modrmmmx, :modrmxmm, :modrmymm
|
277
282
|
if ia.kind_of? ModRM
|
278
283
|
ed = ia.encode(regval, @endianness)
|
279
284
|
if ed.kind_of?(::Array)
|
@@ -295,6 +300,7 @@ class Ia32
|
|
295
300
|
when :mrm_imm; ed = ia.imm.encode("a#{adsz}".to_sym, @endianness)
|
296
301
|
when :i8, :u8, :u16; ed = ia.encode(oa, @endianness)
|
297
302
|
when :i; ed = ia.encode("a#{opsz}".to_sym, @endianness)
|
303
|
+
when :i4xmm, :i4ymm; ed = ia.val << 4 # u8
|
298
304
|
else raise SyntaxError, "Internal error: want to encode field #{oa.inspect} as arg in #{i}"
|
299
305
|
end
|
300
306
|
|