diff options
Diffstat (limited to '')
-rw-r--r-- | libraries/LuaJIT-1.1.7/jit/dis_x86.lua | 622 |
1 files changed, 622 insertions, 0 deletions
diff --git a/libraries/LuaJIT-1.1.7/jit/dis_x86.lua b/libraries/LuaJIT-1.1.7/jit/dis_x86.lua new file mode 100644 index 0000000..b9ca150 --- /dev/null +++ b/libraries/LuaJIT-1.1.7/jit/dis_x86.lua | |||
@@ -0,0 +1,622 @@ | |||
1 | ---------------------------------------------------------------------------- | ||
2 | -- LuaJIT x86 disassembler module. | ||
3 | -- | ||
4 | -- Copyright (C) 2005-2011 Mike Pall. All rights reserved. | ||
5 | -- Released under the MIT/X license. See luajit.h for full copyright notice. | ||
6 | ---------------------------------------------------------------------------- | ||
7 | -- This is a helper module used by the LuaJIT machine code dumper module. | ||
8 | -- | ||
9 | -- Sending small code snippets to an external disassembler and mixing the | ||
10 | -- output with our own stuff was too fragile. So I had to bite the bullet | ||
11 | -- and write yet another x86 disassembler. Oh well ... | ||
12 | -- | ||
13 | -- The output format is very similar to what ndisasm generates. But it has | ||
14 | -- been developed independently by looking at the opcode tables from the | ||
15 | -- Intel and AMD manuals. The supported instruction set is quite extensive | ||
16 | -- and reflects what a current generation P4 or K8 implements in 32 bit | ||
17 | -- mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3 and even privileged | ||
18 | -- instructions. | ||
19 | -- | ||
20 | -- Notes: | ||
21 | -- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported. | ||
22 | -- * No attempt at optimization has been made -- it's fast enough for my needs. | ||
23 | -- * The public API may change when more architectures are added. | ||
24 | -- | ||
25 | -- TODO: | ||
26 | -- * More testing with arbitrary x86 code (not just LuaJIT generated code). | ||
27 | -- * The output for a few MMX/SSE opcodes could be improved. | ||
28 | -- * Adding x64 support would be straightforward. | ||
29 | -- * Better input API (iterator) and output API (structured access to instr). | ||
30 | ------------------------------------------------------------------------------ | ||
31 | |||
32 | local type = type | ||
33 | local sub, byte, format = string.sub, string.byte, string.format | ||
34 | local match, gmatch, gsub = string.match, string.gmatch, string.gsub | ||
35 | |||
36 | -- Map for 1st opcode byte. Ugly? Well ... read on. | ||
37 | local map_opc1 = { | ||
38 | --0x | ||
39 | [0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es", | ||
40 | "orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*", | ||
41 | --1x | ||
42 | "adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss", | ||
43 | "sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds", | ||
44 | --2x | ||
45 | "andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa", | ||
46 | "subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das", | ||
47 | --3x | ||
48 | "xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa", | ||
49 | "cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas", | ||
50 | --4x | ||
51 | "incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR", | ||
52 | "decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR", | ||
53 | --5x | ||
54 | "pushVR","pushVR","pushVR","pushVR","pushVR","pushVR","pushVR","pushVR", | ||
55 | "popVR","popVR","popVR","popVR","popVR","popVR","popVR","popVR", | ||
56 | --6x | ||
57 | "pusha/pushaw","popa/popaw","boundVrm","arplWmr", | ||
58 | "fs:seg","gs:seg","o16:","a16", | ||
59 | "pushVi","imulVrmi","pushBs","imulVrms", | ||
60 | "insb","insd/insw","outsb","outsd/outsw", | ||
61 | --7x | ||
62 | "joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj", | ||
63 | "jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj", | ||
64 | --8x | ||
65 | "arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms", | ||
66 | "testBmr","testVmr","xchgBrm","xchgVrm", | ||
67 | "movBmr","movVmr","movBrm","movVrm", | ||
68 | "movVmg","leaVrm","movWgm","popVm", | ||
69 | --9x | ||
70 | "nop|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR", | ||
71 | "xchgVaR","xchgVaR","xchgVaR","xchgVaR", | ||
72 | "cwde/cbw","cdq/cwd","call farViw","wait", | ||
73 | "pushf/pushfw","popf/popfw","sahf","lahf", | ||
74 | --Ax | ||
75 | "movBao","movVao","movBoa","movVoa", | ||
76 | "movsb","movsd/movsb","cmpsb","cmpsd/cmpsw", | ||
77 | "testBai","testVai","stosb","stosd/stosw", | ||
78 | "lodsb","lodsd/lodsw","scasb","scasd/scasw", | ||
79 | --Bx | ||
80 | "movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi", | ||
81 | "movVRi","movVRi","movVRi","movVRi","movVRi","movVRi","movVRi","movVRi", | ||
82 | --Cx | ||
83 | "shift!Bmu","shift!Vmu","retBw","ret","lesVrm","ldsVrm","movBmi","movVmi", | ||
84 | "enterBwu","leave","retfBw","retf","int3","intBu","into","iret/iretw", | ||
85 | --Dx | ||
86 | "shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb", | ||
87 | "fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7", | ||
88 | --Ex | ||
89 | "loopneBj","loopeBj","loopBj","jecxz/jcxzBj","inBau","inVau","outBua","outVua", | ||
90 | "callDj","jmpDj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda", | ||
91 | --Fx | ||
92 | "lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm", | ||
93 | "clc","stc","cli","sti","cld","std","inc!Bm","inc!Vm", | ||
94 | } | ||
95 | assert(#map_opc1 == 255) | ||
96 | |||
97 | -- Map for 2nd opcode byte (0f xx). True CISC hell. Hey, I told you. | ||
98 | -- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne | ||
99 | local map_opc2 = { | ||
100 | --0x | ||
101 | [0]="sldt!Dmp","sgdt!Dmp","larVrm","lslVrm",nil,"syscall","clts","sysret", | ||
102 | "invd","wbinvd",nil,"ud1",nil,"prefetch!Bm","femms","3dnowMrmu", | ||
103 | --1x | ||
104 | "movupsXrm|movssXrm|movupdXrm|movsdXrm", | ||
105 | "movupsXmr|movssXmr|movupdXmr|movsdXmr", | ||
106 | "movhlpsXrm|movsldupXrm|movlpdXrm|movddupXrm", -- TODO: movlpsXrMm (mem case). | ||
107 | "movlpsXmr||movlpdXmr", | ||
108 | "unpcklpsXrm||unpcklpdXrm", | ||
109 | "unpckhpsXrm||unpckhpdXrm", | ||
110 | "movlhpsXrm|movshdupXrm|movhpdXrm", -- TODO: movhpsXrMm (mem case). | ||
111 | "movhpsXmr||movhpdXmr", | ||
112 | "prefetcht!Bm","hintnopBm","hintnopBm","hintnopBm", | ||
113 | "hintnopBm","hintnopBm","hintnopBm","hintnopBm", | ||
114 | --2x | ||
115 | "movDmx","movDmy","movDxm","movDym","movDmz",nil,"movDzm",nil, | ||
116 | "movapsXrm||movapdXrm", | ||
117 | "movapsXmr||movapdXmr", | ||
118 | "cvtpi2psXrMm|cvtsi2ssXrDm|cvtpi2pdXrMm|cvtsi2sdXrDm", | ||
119 | "movntpsXmr||movntpdXmr", | ||
120 | "cvttps2piMrXm|cvttss2siDrXm|cvttpd2piMrXm|cvttsd2siDrXm", | ||
121 | "cvtps2piMrXm|cvtss2siDrXm|cvtpd2piMrXm|cvtsd2siDrXm", | ||
122 | "ucomissXrm||ucomisdXrm", | ||
123 | "comissXrm||comisdXrm", | ||
124 | --3x | ||
125 | "wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,nil, | ||
126 | "ssse3*38",nil,"ssse3*3a",nil,nil,nil,nil,nil, | ||
127 | --4x | ||
128 | "cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm", | ||
129 | "cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm", | ||
130 | "cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm", | ||
131 | "cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm", | ||
132 | --5x | ||
133 | "movmskpsDrXm||movmskpdDrXm","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm", | ||
134 | "rsqrtpsXrm|rsqrtssXrm","rcppsXrm|rcpssXrm", | ||
135 | "andpsXrm||andpdXrm","andnpsXrm||andnpdXrm", | ||
136 | "orpsXrm||orpdXrm","xorpsXrm||xorpdXrm", | ||
137 | "addpsXrm|addssXrm|addpdXrm|addsdXrm","mulpsXrm|mulssXrm|mulpdXrm|mulsdXrm", | ||
138 | "cvtps2pdXrm|cvtss2sdXrm|cvtpd2psXrm|cvtsd2ssXrm", | ||
139 | "cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm", | ||
140 | "subpsXrm|subssXrm|subpdXrm|subsdXrm","minpsXrm|minssXrm|minpdXrm|minsdXrm", | ||
141 | "divpsXrm|divssXrm|divpdXrm|divsdXrm","maxpsXrm|maxssXrm|maxpdXrm|maxsdXrm", | ||
142 | --6x | ||
143 | "punpcklbwMrm||punpcklbqXrm","punpcklwdPrm","punpckldqPrm","packsswbPrm", | ||
144 | "pcmpgtbPrm","pcmpgtwPrm","pcmpgtdPrm","packuswbPrm", | ||
145 | "punpckhbwPrm","punpckhwdPrm","punpckhdqPrm","packssdwPrm", | ||
146 | "||punpcklqdqXrm","||punpckhqdqXrm", | ||
147 | "movdPrDm","movqMrm|movdquXrm|movdqaXrm", | ||
148 | --7x | ||
149 | "pshufwPrmu","pshiftw!Pmu","pshiftd!Pmu","pshiftq!Mmu||pshiftdq!Xmu", | ||
150 | "pcmpeqbPrm","pcmpeqwPrm","pcmpeqdPrm","emms|", | ||
151 | nil,nil,nil,nil, | ||
152 | "||haddpdXrm|haddpsXrm","||hsubpdXrm|hsubpsXrm", | ||
153 | "movdDmMr|movqXrm|movdDmXr","movqMmr|movdquXmr|movdqaXmr", | ||
154 | --8x | ||
155 | "joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj", | ||
156 | "jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj", | ||
157 | --9x | ||
158 | "setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm", | ||
159 | "setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm", | ||
160 | --Ax | ||
161 | "push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil, | ||
162 | "push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm", | ||
163 | --Bx | ||
164 | "cmpxchgBmr","cmpxchgVmr","lssVrm","btrVmr", | ||
165 | "lfsVrm","lgsVrm","movzxVrBm","movzxDrWm", | ||
166 | nil,"ud2","bt!Vmu","btcVmr", | ||
167 | "bsfVrm","bsrVrm","movsxVrBm","movsxDrWm", | ||
168 | --Cx | ||
169 | "xaddBmr","xaddVmr", | ||
170 | "cmppsXrmu|cmpssXrmu|cmppdXrmu|cmpsdXrmu","movntiDmr|", | ||
171 | "pinsrwPrWmu","pextrwDrPmu", | ||
172 | "shufpsXrmu||shufpdXrmu","cmpxchg!Dmp", | ||
173 | "bswapDR","bswapDR","bswapDR","bswapDR","bswapDR","bswapDR","bswapDR","bswapDR", | ||
174 | --Dx | ||
175 | "||addsubpdXrm|addsubpsXrm","psrlwPrm","psrldPrm","psrlqPrm", | ||
176 | "paddqPrm","pmullwPrm", | ||
177 | "|movq2dqXrMm|movqXmr|movdq2qMrXm","pmovmskbDrPm", | ||
178 | "psubusbPrm","psubuswPrm","pminubPrm","pandPrm", | ||
179 | "paddusbPrm","padduswPrm","pmaxubPrm","pandnPrm", | ||
180 | --Ex | ||
181 | "pavgbPrm","psrawPrm","psradPrm","pavgwPrm", | ||
182 | "pmulhuwPrm","pmulhwPrm", | ||
183 | "|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","movntqMmr||movntdqXmr", | ||
184 | "psubsbPrm","psubswPrm","pminswPrm","porPrm", | ||
185 | "paddsbPrm","paddswPrm","pmaxswPrm","pxorPrm", | ||
186 | --Fx | ||
187 | "|||lddquXrm","psllwPrm","pslldPrm","psllqPrm", | ||
188 | "pmuludqPrm","pmaddwdPrm","psadbwPrm","maskmovqMrm||maskmovdquXrm", | ||
189 | "psubbPrm","psubwPrm","psubdPrm","psubqPrm", | ||
190 | "paddbPrm","paddwPrm","padddPrm","ud", | ||
191 | } | ||
192 | assert(map_opc2[255] == "ud") | ||
193 | |||
194 | -- Map for SSSE3 opcodes. | ||
195 | local map_ssse3 = { | ||
196 | ["38"] = { -- [66] 0f 38 xx | ||
197 | --0x | ||
198 | [0]="pshufbPrm","phaddwPrm","phadddPrm","phaddswPrm", | ||
199 | "pmaddubswPrm","phsubwPrm","phsubdPrm","phsubswPrm", | ||
200 | "psignbPrm","psignwPrm","psigndPrm","pmulhrswPrm", | ||
201 | nil,nil,nil,nil, | ||
202 | --1x | ||
203 | nil,nil,nil,nil,nil,nil,nil,nil, | ||
204 | nil,nil,nil,nil,"pabsbPrm","pabswPrm","pabsdPrm",nil, | ||
205 | }, | ||
206 | ["3a"] = { -- [66] 0f 3a xx | ||
207 | [0x0f] = "palignrPrmu", | ||
208 | }, | ||
209 | } | ||
210 | |||
211 | -- Map for FP opcodes. And you thought stack machines are simple? | ||
212 | local map_opcfp = { | ||
213 | -- D8-DF 00-BF: opcodes with a memory operand. | ||
214 | -- D8 | ||
215 | [0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm", | ||
216 | "fldFm",nil,"fstFm","fstpFm","fldenvDmp","fldcwWm","fnstenvDmp","fnstcwWm", | ||
217 | -- DA | ||
218 | "fiaddDm","fimulDm","ficomDm","ficompDm", | ||
219 | "fisubDm","fisubrDm","fidivDm","fidivrDm", | ||
220 | -- DB | ||
221 | "fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp", | ||
222 | -- DC | ||
223 | "faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm", | ||
224 | -- DD | ||
225 | "fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm", | ||
226 | -- DE | ||
227 | "fiaddWm","fimulWm","ficomWm","ficompWm", | ||
228 | "fisubWm","fisubrWm","fidivWm","fidivrWm", | ||
229 | -- DF | ||
230 | "fildWm","fisttpWm","fistWm","fistpWm", | ||
231 | "fbld twordFmp","fildQm","fbstp twordFmp","fistpQm", | ||
232 | -- xx C0-FF: opcodes with a pseudo-register operand. | ||
233 | -- D8 | ||
234 | "faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf", | ||
235 | -- D9 | ||
236 | "fldFf","fxchFf",{"fnop"},nil, | ||
237 | {"fchs","fabs",nil,nil,"ftst","fxam"}, | ||
238 | {"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"}, | ||
239 | {"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"}, | ||
240 | {"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"}, | ||
241 | -- DA | ||
242 | "fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil, | ||
243 | -- DB | ||
244 | "fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf", | ||
245 | {nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil, | ||
246 | -- DC | ||
247 | "fadd toFf","fmul toFf",nil,nil, | ||
248 | "fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf", | ||
249 | -- DD | ||
250 | "ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil, | ||
251 | -- DE | ||
252 | "faddpFf","fmulpFf",nil,{nil,"fcompp"}, | ||
253 | "fsubrpFf","fsubpFf","fdivrpFf","fdivpFf", | ||
254 | -- DF | ||
255 | nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil, | ||
256 | } | ||
257 | assert(map_opcfp[126] == "fcomipFf") | ||
258 | |||
259 | -- Map for opcode groups. The subkey is sp from the ModRM byte. | ||
260 | local map_opcgroup = { | ||
261 | arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" }, | ||
262 | shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" }, | ||
263 | testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" }, | ||
264 | testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" }, | ||
265 | inc = { "inc", "dec", "callDmp", "call farDmp", | ||
266 | "jmpDmp", "jmp farDmp", "push" }, | ||
267 | sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" }, | ||
268 | sgdt = { "sgdt", "sidt", "lgdt", "lidt", "smsw", nil, "lmsw", "invlpg" }, | ||
269 | bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" }, | ||
270 | cmpxchg = { nil, "cmpxchg8b" }, | ||
271 | pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" }, | ||
272 | pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" }, | ||
273 | pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" }, | ||
274 | pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" }, | ||
275 | fxsave = { "fxsave", "fxrstor", "ldmxcsr", "stmxcsr", | ||
276 | nil, "lfenceDp", "mfenceDp", "sfenceDp" }, -- TODO: clflush. | ||
277 | prefetch = { "prefetch", "prefetchw" }, | ||
278 | prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" }, | ||
279 | } | ||
280 | |||
281 | ------------------------------------------------------------------------------ | ||
282 | |||
283 | -- Maps for register names. | ||
284 | local map_aregs = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi" } | ||
285 | local map_regs = { | ||
286 | B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh" }, | ||
287 | W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di" }, | ||
288 | D = map_aregs, | ||
289 | M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, | ||
290 | X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7" }, | ||
291 | } | ||
292 | local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" } | ||
293 | |||
294 | -- Maps for size names. | ||
295 | local map_sz2n = { | ||
296 | B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16, | ||
297 | } | ||
298 | local map_sz2prefix = { | ||
299 | B = "byte", W = "word", D = "dword", | ||
300 | Q = "qword", -- No associated reg in 32 bit mode. | ||
301 | F = "dword", G = "qword", -- No need for sizes/register names for these two. | ||
302 | M = "qword", X = "xword", | ||
303 | } | ||
304 | |||
305 | ------------------------------------------------------------------------------ | ||
306 | |||
307 | -- Output a nicely formatted line with an opcode and operands. | ||
308 | local function putop(ctx, text, operands) | ||
309 | local code, pos, hex = ctx.code, ctx.pos, "" | ||
310 | for i=ctx.start,pos-1 do | ||
311 | hex = hex..format("%02X", byte(code, i, i)) | ||
312 | end | ||
313 | if #hex > 16 then hex = sub(hex, 1, 16).."." end | ||
314 | if operands then text = text.." "..operands end | ||
315 | if ctx.o16 then text = "o16 "..text; ctx.o16 = false end | ||
316 | if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end | ||
317 | if ctx.seg then | ||
318 | local text2, n = gsub(text, "%[", "["..ctx.seg..":") | ||
319 | if n == 0 then text = ctx.seg.." "..text else text = text2 end | ||
320 | ctx.seg = false | ||
321 | end | ||
322 | if ctx.lock then text = "lock "..text; ctx.lock = false end | ||
323 | local imm = ctx.imm | ||
324 | if imm then | ||
325 | local sym = ctx.symtab[imm] | ||
326 | if sym then text = text.."\t->"..sym end | ||
327 | end | ||
328 | ctx.out(format("%08x %-18s%s\n", ctx.addr+ctx.start, hex, text)) | ||
329 | ctx.mrm = false | ||
330 | ctx.start = pos | ||
331 | ctx.imm = nil | ||
332 | end | ||
333 | |||
334 | -- Fallback for incomplete opcodes at the end. | ||
335 | local function incomplete(ctx) | ||
336 | ctx.pos = ctx.stop+1 | ||
337 | ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false | ||
338 | return putop(ctx, "(incomplete)") | ||
339 | end | ||
340 | |||
341 | -- Fallback for unknown opcodes. | ||
342 | local function unknown(ctx) | ||
343 | ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false | ||
344 | return putop(ctx, "(unknown)") | ||
345 | end | ||
346 | |||
347 | -- Return an immediate of the specified size. | ||
348 | local function getimm(ctx, pos, n) | ||
349 | if pos+n-1 > ctx.stop then return incomplete(ctx) end | ||
350 | local code = ctx.code | ||
351 | if n == 1 then | ||
352 | local b1 = byte(code, pos, pos) | ||
353 | return b1 | ||
354 | elseif n == 2 then | ||
355 | local b1, b2 = byte(code, pos, pos+1) | ||
356 | return b1+b2*256 | ||
357 | else | ||
358 | local b1, b2, b3, b4 = byte(code, pos, pos+3) | ||
359 | local imm = b1+b2*256+b3*65536+b4*16777216 | ||
360 | ctx.imm = imm | ||
361 | return imm | ||
362 | end | ||
363 | end | ||
364 | |||
365 | -- Process pattern string and generate the operands. | ||
366 | local function putpat(ctx, name, pat) | ||
367 | local operands, regs, sz, mode, sp, rm, sc, rx, disp, sdisp | ||
368 | local code, pos, stop = ctx.code, ctx.pos, ctx.stop | ||
369 | |||
370 | -- Chars used: 1DFGMPQRVWXacdfgijmoprsuwxyz | ||
371 | for p in gmatch(pat, ".") do | ||
372 | local x = nil | ||
373 | if p == "V" then | ||
374 | sz = ctx.o16 and "W" or "D"; ctx.o16 = false | ||
375 | regs = map_regs[sz] | ||
376 | elseif match(p, "[BWDQFGMX]") then | ||
377 | sz = p | ||
378 | regs = map_regs[sz] | ||
379 | elseif p == "P" then | ||
380 | sz = ctx.o16 and "X" or "M"; ctx.o16 = false | ||
381 | regs = map_regs[sz] | ||
382 | elseif p == "s" then | ||
383 | local imm = getimm(ctx, pos, 1); if not imm then return end | ||
384 | x = imm <= 127 and format("byte +0x%02x", imm) | ||
385 | or format("byte -0x%02x", 256-imm) | ||
386 | pos = pos+1 | ||
387 | elseif p == "u" then | ||
388 | local imm = getimm(ctx, pos, 1); if not imm then return end | ||
389 | x = format("0x%02x", imm) | ||
390 | pos = pos+1 | ||
391 | elseif p == "w" then | ||
392 | local imm = getimm(ctx, pos, 2); if not imm then return end | ||
393 | x = format("0x%x", imm) | ||
394 | pos = pos+2 | ||
395 | elseif p == "o" then -- [offset] | ||
396 | local imm = getimm(ctx, pos, 4); if not imm then return end | ||
397 | x = format("[0x%08x]", imm) | ||
398 | pos = pos+4 | ||
399 | elseif p == "i" then | ||
400 | local n = map_sz2n[sz] | ||
401 | local imm = getimm(ctx, pos, n); if not imm then return end | ||
402 | x = format(imm > 65535 and "0x%08x" or "0x%x", imm) | ||
403 | pos = pos+n | ||
404 | elseif p == "j" then | ||
405 | local n = map_sz2n[sz] | ||
406 | local imm = getimm(ctx, pos, n); if not imm then return end | ||
407 | if sz == "B" and imm > 127 then imm = imm-256 | ||
408 | elseif imm > 2147483647 then imm = imm-4294967296 end | ||
409 | pos = pos+n | ||
410 | imm = imm + pos + ctx.addr | ||
411 | ctx.imm = imm | ||
412 | x = sz == "W" and format("word 0x%04x", imm%65536) | ||
413 | or format("0x%08x", imm) | ||
414 | elseif p == "R" then x = regs[byte(code, pos-1, pos-1)%8+1] | ||
415 | elseif p == "a" then x = regs[1] | ||
416 | elseif p == "c" then x = "cl" | ||
417 | elseif p == "d" then x = "dx" | ||
418 | elseif p == "1" then x = "1" | ||
419 | else | ||
420 | if not mode then | ||
421 | mode = ctx.mrm | ||
422 | if not mode then | ||
423 | if pos > stop then return incomplete(ctx) end | ||
424 | mode = byte(code, pos, pos) | ||
425 | pos = pos+1 | ||
426 | end | ||
427 | rm = mode%8; mode = (mode-rm)/8 | ||
428 | sp = mode%8; mode = (mode-sp)/8 | ||
429 | sdisp = "" | ||
430 | if mode < 3 then | ||
431 | if rm == 4 then | ||
432 | if pos > stop then return incomplete(ctx) end | ||
433 | sc = byte(code, pos, pos) | ||
434 | pos = pos+1 | ||
435 | rm = sc%8; sc = (sc-rm)/8 | ||
436 | rx = sc%8; sc = (sc-rx)/8 | ||
437 | if rx == 4 then rx = nil end | ||
438 | end | ||
439 | if mode > 0 or rm == 5 then | ||
440 | local dsz = mode | ||
441 | if dsz ~= 1 then dsz = 4 end | ||
442 | disp = getimm(ctx, pos, dsz); if not disp then return end | ||
443 | sdisp = (dsz == 4 or disp <= 127) and | ||
444 | format(disp > 65535 and "+0x%08x" or "+0x%x", disp) or | ||
445 | format("-0x%x", 256-disp) | ||
446 | pos = pos+dsz | ||
447 | end | ||
448 | end | ||
449 | end | ||
450 | if p == "m" then | ||
451 | if mode == 3 then x = regs[rm+1] | ||
452 | else | ||
453 | local srm, srx = map_aregs[rm+1], "" | ||
454 | if rx then | ||
455 | srm = srm.."+" | ||
456 | srx = map_aregs[rx+1] | ||
457 | if sc > 0 then srx = srx.."*"..(2^sc) end | ||
458 | end | ||
459 | if mode == 0 and rm == 5 then | ||
460 | srm = "" | ||
461 | sdisp = format("%s0x%08x", rx and "+" or "", disp) | ||
462 | end | ||
463 | x = format("[%s%s%s]", srm, srx, sdisp) | ||
464 | end | ||
465 | if mode < 3 and | ||
466 | (not match(pat, "[aRrgp]") or | ||
467 | name == "movzx" or name == "movsx") then -- Yuck. | ||
468 | x = map_sz2prefix[sz].." "..x | ||
469 | end | ||
470 | elseif p == "r" then x = regs[sp+1] | ||
471 | elseif p == "g" then x = map_segregs[sp+1] | ||
472 | elseif p == "p" then -- Suppress prefix. | ||
473 | elseif p == "f" then x = "st"..rm | ||
474 | elseif p == "x" then x = "CR"..sp | ||
475 | elseif p == "y" then x = "DR"..sp | ||
476 | elseif p == "z" then x = "TR"..sp | ||
477 | else | ||
478 | error("bad pattern `"..pat.."'") | ||
479 | end | ||
480 | end | ||
481 | if x then operands = operands and operands..","..x or x end | ||
482 | end | ||
483 | ctx.pos = pos | ||
484 | return putop(ctx, name, operands) | ||
485 | end | ||
486 | |||
487 | -- Forward declaration. | ||
488 | local map_act | ||
489 | |||
490 | -- Get a pattern from an opcode map and dispatch to handler. | ||
491 | local function opcdispatch(ctx, opcmap) | ||
492 | local pos = ctx.pos | ||
493 | local opat = opcmap[byte(ctx.code, pos, pos)] | ||
494 | if not opat then return unknown(ctx) end | ||
495 | if match(opat, "%|") then -- MMX/SSE variants depending on prefix. | ||
496 | local p | ||
497 | if ctx.rep then p = ctx.rep=="rep" and "%|([^%|]*)" or "%|.-%|.-%|([^%|]*)" | ||
498 | elseif ctx.o16 then p = "%|.-%|([^%|]*)" | ||
499 | else p = "^[^%|]*" end | ||
500 | opat = match(opat, p) | ||
501 | if not opat or opat == "" then return unknown(ctx) end | ||
502 | ctx.rep = false; ctx.o16 = false | ||
503 | end | ||
504 | local name, pat, act = match(opat, "^([a-z0-9 ]*)((.?).*)") | ||
505 | ctx.pos = pos + 1 | ||
506 | return map_act[act](ctx, name, pat) | ||
507 | end | ||
508 | |||
509 | -- Map for action codes. The key is the first char after the name. | ||
510 | map_act = { | ||
511 | -- Simple opcodes without operands. | ||
512 | [""] = function(ctx, name, pat) | ||
513 | return putop(ctx, name) | ||
514 | end, | ||
515 | |||
516 | -- Operand size chars fall right through. | ||
517 | B = putpat, W = putpat, D = putpat, V = putpat, | ||
518 | F = putpat, G = putpat, | ||
519 | M = putpat, X = putpat, P = putpat, | ||
520 | |||
521 | -- Collect prefixes. | ||
522 | [":"] = function(ctx, name, pat) | ||
523 | ctx[pat == ":" and name or sub(pat, 2)] = name | ||
524 | end, | ||
525 | |||
526 | -- Select alternate opcode name when prefixed with o16. | ||
527 | ["/"] = function(ctx, name, pat) | ||
528 | local wname, rpat = match(pat, "^/([a-z0-9 ]+)(.*)") | ||
529 | if ctx.o16 then name = wname; ctx.o16 = false end | ||
530 | return putpat(ctx, name, rpat) | ||
531 | end, | ||
532 | |||
533 | -- Chain to special handler specified by name. | ||
534 | ["*"] = function(ctx, name, pat) | ||
535 | return map_act[name](ctx, name, sub(pat, 2)) | ||
536 | end, | ||
537 | |||
538 | -- Use named subtable for opcode group. | ||
539 | ["!"] = function(ctx, name, pat) | ||
540 | |||
541 | local pos = ctx.pos | ||
542 | if pos > ctx.stop then return incomplete(ctx) end | ||
543 | local mrm = byte(ctx.code, pos, pos) | ||
544 | ctx.pos = pos+1 | ||
545 | ctx.mrm = mrm | ||
546 | |||
547 | local opat = map_opcgroup[name][((mrm-(mrm%8))/8)%8+1] | ||
548 | if not opat then return unknown(ctx) end | ||
549 | local name, pat2 = match(opat, "^([a-z0-9 ]*)(.*)") | ||
550 | return putpat(ctx, name, pat2 ~= "" and pat2 or sub(pat, 2)) | ||
551 | end, | ||
552 | |||
553 | -- Two-byte opcode dispatch. | ||
554 | opc2 = function(ctx, name, pat) | ||
555 | return opcdispatch(ctx, map_opc2) | ||
556 | end, | ||
557 | |||
558 | -- SSSE3 dispatch. | ||
559 | ssse3 = function(ctx, name, pat) | ||
560 | return opcdispatch(ctx, map_ssse3[pat]) | ||
561 | end, | ||
562 | |||
563 | -- Floating point opcode dispatch. | ||
564 | fp = function(ctx, name, pat) | ||
565 | |||
566 | local pos = ctx.pos | ||
567 | if pos > ctx.stop then return incomplete(ctx) end | ||
568 | local mrm = byte(ctx.code, pos, pos) | ||
569 | ctx.pos = pos+1 | ||
570 | ctx.mrm = mrm | ||
571 | |||
572 | local rm = mrm%8 | ||
573 | local idx = pat*8 + ((mrm-rm)/8)%8 | ||
574 | if mrm >= 192 then idx = idx + 64 end | ||
575 | local opat = map_opcfp[idx] | ||
576 | if type(opat) == "table" then opat = opat[rm+1] end | ||
577 | if not opat then return unknown(ctx) end | ||
578 | local name, pat2 = match(opat, "^([a-z0-9 ]*)(.*)") | ||
579 | return putpat(ctx, name, pat2) | ||
580 | end, | ||
581 | } | ||
582 | |||
583 | ------------------------------------------------------------------------------ | ||
584 | |||
585 | -- Disassemble a block of code. | ||
586 | local function disass_block(ctx, ofs, len) | ||
587 | if not ofs then ofs = 0 end | ||
588 | local stop = len and ofs+len or #ctx.code | ||
589 | ofs = ofs + 1 | ||
590 | ctx.start = ofs | ||
591 | ctx.pos = ofs | ||
592 | ctx.stop = stop | ||
593 | ctx.imm = nil | ||
594 | ctx.mrm = false | ||
595 | ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false | ||
596 | while ctx.pos <= stop do opcdispatch(ctx, map_opc1) end | ||
597 | if ctx.pos ~= ctx.start then incomplete(ctx) end | ||
598 | end | ||
599 | |||
600 | -- Extended API: create a disassembler context. Then call ctx:disass(ofs, len). | ||
601 | local function create_(code, addr, out) | ||
602 | local ctx = {} | ||
603 | ctx.code = code | ||
604 | ctx.addr = (addr or 0) - 1 | ||
605 | ctx.out = out or io.write | ||
606 | ctx.symtab = {} | ||
607 | ctx.disass = disass_block | ||
608 | return ctx | ||
609 | end | ||
610 | |||
611 | -- Simple API: disassemble code (a string) at address and output via out. | ||
612 | local function disass_(code, addr, out) | ||
613 | create_(code, addr, out):disass() | ||
614 | end | ||
615 | |||
616 | |||
617 | -- Public module functions. | ||
618 | module(...) | ||
619 | |||
620 | create = create_ | ||
621 | disass = disass_ | ||
622 | |||