summaryrefslogtreecommitdiff
path: root/lib/dis_x86.lua
diff options
context:
space:
mode:
Diffstat (limited to 'lib/dis_x86.lua')
-rw-r--r--lib/dis_x86.lua824
1 files changed, 824 insertions, 0 deletions
diff --git a/lib/dis_x86.lua b/lib/dis_x86.lua
new file mode 100644
index 00000000..8f127bee
--- /dev/null
+++ b/lib/dis_x86.lua
@@ -0,0 +1,824 @@
1----------------------------------------------------------------------------
2-- LuaJIT x86/x64 disassembler module.
3--
4-- Copyright (C) 2005-2009 Mike Pall. All rights reserved.
5-- Released under the MIT/X license. See Copyright Notice in luajit.h
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 Intel or AMD CPU implements in
17-- 32 bit and 64 bit mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3,
18-- SSE4.1, SSE4.2, SSE4a and even privileged and hypervisor (VMX/SVM)
19-- instructions.
20--
21-- Notes:
22-- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported.
23-- * No attempt at optimization has been made -- it's fast enough for my needs.
24-- * The public API may change when more architectures are added.
25------------------------------------------------------------------------------
26
27local type = type
28local sub, byte, format = string.sub, string.byte, string.format
29local match, gmatch, gsub = string.match, string.gmatch, string.gsub
30local lower, rep = string.lower, string.rep
31
32-- Map for 1st opcode byte in 32 bit mode. Ugly? Well ... read on.
33local map_opc1_32 = {
34--0x
35[0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es",
36"orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*",
37--1x
38"adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss",
39"sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds",
40--2x
41"andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa",
42"subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das",
43--3x
44"xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa",
45"cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas",
46--4x
47"incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR",
48"decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR",
49--5x
50"pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR",
51"popUR","popUR","popUR","popUR","popUR","popUR","popUR","popUR",
52--6x
53"sz*pushaw,pusha","sz*popaw,popa","boundVrm","arplWmr",
54"fs:seg","gs:seg","o16:","a16",
55"pushUi","imulVrmi","pushBs","imulVrms",
56"insb","insVS","outsb","outsVS",
57--7x
58"joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj",
59"jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj",
60--8x
61"arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms",
62"testBmr","testVmr","xchgBrm","xchgVrm",
63"movBmr","movVmr","movBrm","movVrm",
64"movVmg","leaVrm","movWgm","popUm",
65--9x
66"nop*xchgVaR|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR",
67"xchgVaR","xchgVaR","xchgVaR","xchgVaR",
68"sz*cbw,cwde,cdqe","sz*cwd,cdq,cqo","call farViw","wait",
69"sz*pushfw,pushf","sz*popfw,popf","sahf","lahf",
70--Ax
71"movBao","movVao","movBoa","movVoa",
72"movsb","movsVS","cmpsb","cmpsVS",
73"testBai","testVai","stosb","stosVS",
74"lodsb","lodsVS","scasb","scasVS",
75--Bx
76"movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi",
77"movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI",
78--Cx
79"shift!Bmu","shift!Vmu","retBw","ret","$lesVrm","$ldsVrm","movBmi","movVmi",
80"enterBwu","leave","retfBw","retf","int3","intBu","into","iretVS",
81--Dx
82"shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb",
83"fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7",
84--Ex
85"loopneBj","loopeBj","loopBj","sz*jcxzBj,jecxzBj,jrcxzBj",
86"inBau","inVau","outBua","outVua",
87"callVj","jmpVj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda",
88--Fx
89"lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm",
90"clc","stc","cli","sti","cld","std","incb!Bm","incd!Vm",
91}
92assert(#map_opc1_32 == 255)
93
94-- Map for 1st opcode byte in 64 bit mode (overrides only).
95local map_opc1_64 = setmetatable({
96 [0x06]=false, [0x07]=false, [0x0e]=false,
97 [0x16]=false, [0x17]=false, [0x1e]=false, [0x1f]=false,
98 [0x27]=false, [0x2f]=false, [0x37]=false, [0x3f]=false,
99 [0x60]=false, [0x61]=false, [0x62]=false, [0x63]="movsxdVrDmt", [0x67]="a32:",
100 [0x40]="rex*", [0x41]="rex*b", [0x42]="rex*x", [0x43]="rex*xb",
101 [0x44]="rex*r", [0x45]="rex*rb", [0x46]="rex*rx", [0x47]="rex*rxb",
102 [0x48]="rex*w", [0x49]="rex*wb", [0x4a]="rex*wx", [0x4b]="rex*wxb",
103 [0x4c]="rex*wr", [0x4d]="rex*wrb", [0x4e]="rex*wrx", [0x4f]="rex*wrxb",
104 [0x82]=false, [0x9a]=false, [0xc4]=false, [0xc5]=false, [0xce]=false,
105 [0xd4]=false, [0xd5]=false, [0xd6]=false, [0xea]=false,
106}, { __index = map_opc1_32 })
107
108-- Map for 2nd opcode byte (0F xx). True CISC hell. Hey, I told you.
109-- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne, -|F3|66|F2
110local map_opc2 = {
111--0x
112[0]="sldt!Dmp","sgdt!Ump","larVrm","lslVrm",nil,"syscall","clts","sysret",
113"invd","wbinvd",nil,"ud1",nil,"$prefetch!Bm","femms","3dnowMrmu",
114--1x
115"movupsXrm|movssXrm|movupdXrm|movsdXrm",
116"movupsXmr|movssXmr|movupdXmr|movsdXmr",
117"movhlpsXrm$movlpsXrm|movsldupXrm|movlpdXrm|movddupXrm",
118"movlpsXmr||movlpdXmr",
119"unpcklpsXrm||unpcklpdXrm",
120"unpckhpsXrm||unpckhpdXrm",
121"movlhpsXrm$movhpsXrm|movshdupXrm|movhpdXrm",
122"movhpsXmr||movhpdXmr",
123"$prefetcht!Bm","hintnopVm","hintnopVm","hintnopVm",
124"hintnopVm","hintnopVm","hintnopVm","hintnopVm",
125--2x
126"movUmx$","movUmy$","movUxm$","movUym$","movUmz$",nil,"movUzm$",nil,
127"movapsXrm||movapdXrm",
128"movapsXmr||movapdXmr",
129"cvtpi2psXrMm|cvtsi2ssXrVm|cvtpi2pdXrMm|cvtsi2sdXrVm",
130"movntpsXmr|movntssXmr|movntpdXmr|movntsdXmr",
131"cvttps2piMrXm|cvttss2siVrXm|cvttpd2piMrXm|cvttsd2siVrXm",
132"cvtps2piMrXm|cvtss2siVrXm|cvtpd2piMrXm|cvtsd2siVrXm",
133"ucomissXrm||ucomisdXrm",
134"comissXrm||comisdXrm",
135--3x
136"wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,"getsec",
137"opc3*38",nil,"opc3*3a",nil,nil,nil,nil,nil,
138--4x
139"cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm",
140"cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm",
141"cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm",
142"cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm",
143--5x
144"movmskpsVrXm$||movmskpdVrXm$","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm",
145"rsqrtpsXrm|rsqrtssXrm","rcppsXrm|rcpssXrm",
146"andpsXrm||andpdXrm","andnpsXrm||andnpdXrm",
147"orpsXrm||orpdXrm","xorpsXrm||xorpdXrm",
148"addpsXrm|addssXrm|addpdXrm|addsdXrm","mulpsXrm|mulssXrm|mulpdXrm|mulsdXrm",
149"cvtps2pdXrm|cvtss2sdXrm|cvtpd2psXrm|cvtsd2ssXrm",
150"cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm",
151"subpsXrm|subssXrm|subpdXrm|subsdXrm","minpsXrm|minssXrm|minpdXrm|minsdXrm",
152"divpsXrm|divssXrm|divpdXrm|divsdXrm","maxpsXrm|maxssXrm|maxpdXrm|maxsdXrm",
153--6x
154"punpcklbwPrm","punpcklwdPrm","punpckldqPrm","packsswbPrm",
155"pcmpgtbPrm","pcmpgtwPrm","pcmpgtdPrm","packuswbPrm",
156"punpckhbwPrm","punpckhwdPrm","punpckhdqPrm","packssdwPrm",
157"||punpcklqdqXrm","||punpckhqdqXrm",
158"movPrVSm","movqMrm|movdquXrm|movdqaXrm",
159--7x
160"pshufwMrmu|pshufhwXrmu|pshufdXrmu|pshuflwXrmu","pshiftw!Pmu",
161"pshiftd!Pmu","pshiftq!Mmu||pshiftdq!Xmu",
162"pcmpeqbPrm","pcmpeqwPrm","pcmpeqdPrm","emms|",
163"vmreadUmr||extrqXmuu$|insertqXrmuu$","vmwriteUrm||extrqXrm$|insertqXrm$",
164nil,nil,
165"||haddpdXrm|haddpsXrm","||hsubpdXrm|hsubpsXrm",
166"movVSmMr|movqXrm|movVSmXr","movqMmr|movdquXmr|movdqaXmr",
167--8x
168"joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj",
169"jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj",
170--9x
171"setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm",
172"setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm",
173--Ax
174"push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil,
175"push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm",
176--Bx
177"cmpxchgBmr","cmpxchgVmr","$lssVrm","btrVmr",
178"$lfsVrm","$lgsVrm","movzxVrBmt","movzxVrWmt",
179"|popcntVrm","ud2Dp","bt!Vmu","btcVmr",
180"bsfVrm","bsrVrm|lzcntVrm|bsrWrm","movsxVrBmt","movsxVrWmt",
181--Cx
182"xaddBmr","xaddVmr",
183"cmppsXrmu|cmpssXrmu|cmppdXrmu|cmpsdXrmu","$movntiVmr|",
184"pinsrwPrWmu","pextrwDrPmu",
185"shufpsXrmu||shufpdXrmu","$cmpxchg!Qmp",
186"bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR",
187--Dx
188"||addsubpdXrm|addsubpsXrm","psrlwPrm","psrldPrm","psrlqPrm",
189"paddqPrm","pmullwPrm",
190"|movq2dqXrMm|movqXmr|movdq2qMrXm$","pmovmskbVrMm||pmovmskbVrXm",
191"psubusbPrm","psubuswPrm","pminubPrm","pandPrm",
192"paddusbPrm","padduswPrm","pmaxubPrm","pandnPrm",
193--Ex
194"pavgbPrm","psrawPrm","psradPrm","pavgwPrm",
195"pmulhuwPrm","pmulhwPrm",
196"|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","$movntqMmr||$movntdqXmr",
197"psubsbPrm","psubswPrm","pminswPrm","porPrm",
198"paddsbPrm","paddswPrm","pmaxswPrm","pxorPrm",
199--Fx
200"|||lddquXrm","psllwPrm","pslldPrm","psllqPrm",
201"pmuludqPrm","pmaddwdPrm","psadbwPrm","maskmovqMrm||maskmovdquXrm$",
202"psubbPrm","psubwPrm","psubdPrm","psubqPrm",
203"paddbPrm","paddwPrm","padddPrm","ud",
204}
205assert(map_opc2[255] == "ud")
206
207-- Map for three-byte opcodes. Can't wait for their next invention.
208local map_opc3 = {
209["38"] = { -- [66] 0f 38 xx
210--0x
211[0]="pshufbPrm","phaddwPrm","phadddPrm","phaddswPrm",
212"pmaddubswPrm","phsubwPrm","phsubdPrm","phsubswPrm",
213"psignbPrm","psignwPrm","psigndPrm","pmulhrswPrm",
214nil,nil,nil,nil,
215--1x
216"||pblendvbXrma",nil,nil,nil,
217"||blendvpsXrma","||blendvpdXrma",nil,"||ptestXrm",
218nil,nil,nil,nil,
219"pabsbPrm","pabswPrm","pabsdPrm",nil,
220--2x
221"||pmovsxbwXrm","||pmovsxbdXrm","||pmovsxbqXrm","||pmovsxwdXrm",
222"||pmovsxwqXrm","||pmovsxdqXrm",nil,nil,
223"||pmuldqXrm","||pcmpeqqXrm","||$movntdqaXrm","||packusdwXrm",
224nil,nil,nil,nil,
225--3x
226"||pmovzxbwXrm","||pmovzxbdXrm","||pmovzxbqXrm","||pmovzxwdXrm",
227"||pmovzxwqXrm","||pmovzxdqXrm",nil,"||pcmpgtqXrm",
228"||pminsbXrm","||pminsdXrm","||pminuwXrm","||pminudXrm",
229"||pmaxsbXrm","||pmaxsdXrm","||pmaxuwXrm","||pmaxudXrm",
230--4x
231"||pmulddXrm","||phminposuwXrm",
232--Fx
233[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
234},
235
236["3a"] = { -- [66] 0f 3a xx
237--0x
238[0x00]=nil,nil,nil,nil,nil,nil,nil,nil,
239"||roundpsXrmu","||roundpdXrmu","||roundssXrmu","||roundsdXrmu",
240"||blendpsXrmu","||blendpdXrmu","||pblendwXrmu","palignrPrmu",
241--1x
242nil,nil,nil,nil,
243"||pextrbVmXru","||pextrwVmXru","||pextrVmSXru","||extractpsVmXru",
244nil,nil,nil,nil,nil,nil,nil,nil,
245--2x
246"||pinsrbXrVmu","||insertpsXrmu","||pinsrXrVmuS",nil,
247--4x
248[0x40] = "||dppsXrmu",
249[0x41] = "||dppdXrmu",
250[0x42] = "||mpsadbwXrmu",
251--6x
252[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
253[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
254},
255}
256
257-- Map for VMX/SVM opcodes 0F 01 C0-FF (sgdt group with register operands).
258local map_opcvm = {
259[0xc1]="vmcall",[0xc2]="vmlaunch",[0xc3]="vmresume",[0xc4]="vmxoff",
260[0xc8]="monitor",[0xc9]="mwait",
261[0xd8]="vmrun",[0xd9]="vmmcall",[0xda]="vmload",[0xdb]="vmsave",
262[0xdc]="stgi",[0xdd]="clgi",[0xde]="skinit",[0xdf]="invlpga",
263[0xf8]="swapgs",[0xf9]="rdtscp",
264}
265
266-- Map for FP opcodes. And you thought stack machines are simple?
267local map_opcfp = {
268-- D8-DF 00-BF: opcodes with a memory operand.
269-- D8
270[0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm",
271"fldFm",nil,"fstFm","fstpFm","fldenvVm","fldcwWm","fnstenvVm","fnstcwWm",
272-- DA
273"fiaddDm","fimulDm","ficomDm","ficompDm",
274"fisubDm","fisubrDm","fidivDm","fidivrDm",
275-- DB
276"fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp",
277-- DC
278"faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm",
279-- DD
280"fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm",
281-- DE
282"fiaddWm","fimulWm","ficomWm","ficompWm",
283"fisubWm","fisubrWm","fidivWm","fidivrWm",
284-- DF
285"fildWm","fisttpWm","fistWm","fistpWm",
286"fbld twordFmp","fildQm","fbstp twordFmp","fistpQm",
287-- xx C0-FF: opcodes with a pseudo-register operand.
288-- D8
289"faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf",
290-- D9
291"fldFf","fxchFf",{"fnop"},nil,
292{"fchs","fabs",nil,nil,"ftst","fxam"},
293{"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"},
294{"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"},
295{"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"},
296-- DA
297"fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil,
298-- DB
299"fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf",
300{nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil,
301-- DC
302"fadd toFf","fmul toFf",nil,nil,
303"fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf",
304-- DD
305"ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil,
306-- DE
307"faddpFf","fmulpFf",nil,{nil,"fcompp"},
308"fsubrpFf","fsubpFf","fdivrpFf","fdivpFf",
309-- DF
310nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil,
311}
312assert(map_opcfp[126] == "fcomipFf")
313
314-- Map for opcode groups. The subkey is sp from the ModRM byte.
315local map_opcgroup = {
316 arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" },
317 shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" },
318 testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" },
319 testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" },
320 incb = { "inc", "dec" },
321 incd = { "inc", "dec", "callDmp", "$call farDmp",
322 "jmpDmp", "$jmp farDmp", "pushUm" },
323 sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" },
324 sgdt = { "vm*$sgdt", "vm*$sidt", "$lgdt", "vm*$lidt",
325 "smsw", nil, "lmsw", "vm*$invlpg" },
326 bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" },
327 cmpxchg = { nil, "sz*,cmpxchg8bQmp,cmpxchg16bXmp", nil, nil,
328 nil, nil, "vmptrld|vmxon|vmclear", "vmptrst" },
329 pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" },
330 pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" },
331 pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" },
332 pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" },
333 fxsave = { "$fxsave", "$fxrstor", "$ldmxcsr", "$stmxcsr",
334 nil, "lfenceDp$", "mfenceDp$", "sfenceDp$clflush" },
335 prefetch = { "prefetch", "prefetchw" },
336 prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" },
337}
338
339------------------------------------------------------------------------------
340
341-- Maps for register names.
342local map_regs = {
343 B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh",
344 "r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
345 B64 = { "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
346 "r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
347 W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di",
348 "r8w", "r9w", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w" },
349 D = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi",
350 "r8d", "r9d", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d" },
351 Q = { "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
352 "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15" },
353 M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
354 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, -- No x64 ext!
355 X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
356 "xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15" },
357}
358local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" }
359
360-- Maps for size names.
361local map_sz2n = {
362 B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16,
363}
364local map_sz2prefix = {
365 B = "byte", W = "word", D = "dword",
366 Q = "qword",
367 M = "qword", X = "xword",
368 F = "dword", G = "qword", -- No need for sizes/register names for these two.
369}
370
371------------------------------------------------------------------------------
372
373-- Output a nicely formatted line with an opcode and operands.
374local function putop(ctx, text, operands)
375 local code, pos, hex = ctx.code, ctx.pos, ""
376 local hmax = ctx.hexdump
377 if hmax > 0 then
378 for i=ctx.start,pos-1 do
379 hex = hex..format("%02X", byte(code, i, i))
380 end
381 if #hex > hmax then hex = sub(hex, 1, hmax)..". "
382 else hex = hex..rep(" ", hmax-#hex+2) end
383 end
384 if operands then text = text.." "..operands end
385 if ctx.o16 then text = "o16 "..text; ctx.o16 = false end
386 if ctx.a32 then text = "a32 "..text; ctx.a32 = false end
387 if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end
388 if ctx.rex then
389 local t = (ctx.rexw and "w" or "")..(ctx.rexr and "r" or "")..
390 (ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")
391 if t ~= "" then text = "rex."..t.." "..text end
392 ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
393 ctx.rex = false
394 end
395 if ctx.seg then
396 local text2, n = gsub(text, "%[", "["..ctx.seg..":")
397 if n == 0 then text = ctx.seg.." "..text else text = text2 end
398 ctx.seg = false
399 end
400 if ctx.lock then text = "lock "..text; ctx.lock = false end
401 local imm = ctx.imm
402 if imm then
403 local sym = ctx.symtab[imm]
404 if sym then text = text.."\t->"..sym end
405 end
406 ctx.out(format("%08x %s%s\n", ctx.addr+ctx.start, hex, text))
407 ctx.mrm = false
408 ctx.start = pos
409 ctx.imm = nil
410end
411
412-- Clear all prefix flags.
413local function clearprefixes(ctx)
414 ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false
415 ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
416 ctx.rex = false; ctx.a32 = false
417end
418
419-- Fallback for incomplete opcodes at the end.
420local function incomplete(ctx)
421 ctx.pos = ctx.stop+1
422 clearprefixes(ctx)
423 return putop(ctx, "(incomplete)")
424end
425
426-- Fallback for unknown opcodes.
427local function unknown(ctx)
428 clearprefixes(ctx)
429 return putop(ctx, "(unknown)")
430end
431
432-- Return an immediate of the specified size.
433local function getimm(ctx, pos, n)
434 if pos+n-1 > ctx.stop then return incomplete(ctx) end
435 local code = ctx.code
436 if n == 1 then
437 local b1 = byte(code, pos, pos)
438 return b1
439 elseif n == 2 then
440 local b1, b2 = byte(code, pos, pos+1)
441 return b1+b2*256
442 else
443 local b1, b2, b3, b4 = byte(code, pos, pos+3)
444 local imm = b1+b2*256+b3*65536+b4*16777216
445 ctx.imm = imm
446 return imm
447 end
448end
449
450-- Process pattern string and generate the operands.
451local function putpat(ctx, name, pat)
452 local operands, regs, sz, mode, sp, rm, sc, rx, sdisp
453 local code, pos, stop = ctx.code, ctx.pos, ctx.stop
454
455 -- Chars used: 1DFGIMPQRSTUVWXacdfgijmoprstuwxyz
456 for p in gmatch(pat, ".") do
457 local x = nil
458 if p == "V" or p == "U" then
459 if ctx.rexw then sz = "Q"; ctx.rexw = false
460 elseif ctx.o16 then sz = "W"; ctx.o16 = false
461 elseif p == "U" and ctx.x64 then sz = "Q"
462 else sz = "D" end
463 regs = map_regs[sz]
464 elseif p == "T" then
465 if ctx.rexw then sz = "Q"; ctx.rexw = false else sz = "D" end
466 regs = map_regs[sz]
467 elseif p == "B" then
468 sz = "B"
469 regs = ctx.rex and map_regs.B64 or map_regs.B
470 elseif match(p, "[WDQMXFG]") then
471 sz = p
472 regs = map_regs[sz]
473 elseif p == "P" then
474 sz = ctx.o16 and "X" or "M"; ctx.o16 = false
475 regs = map_regs[sz]
476 elseif p == "S" then
477 name = name..lower(sz)
478 elseif p == "s" then
479 local imm = getimm(ctx, pos, 1); if not imm then return end
480 x = imm <= 127 and format("+0x%02x", imm)
481 or format("-0x%02x", 256-imm)
482 pos = pos+1
483 elseif p == "u" then
484 local imm = getimm(ctx, pos, 1); if not imm then return end
485 x = format("0x%02x", imm)
486 pos = pos+1
487 elseif p == "w" then
488 local imm = getimm(ctx, pos, 2); if not imm then return end
489 x = format("0x%x", imm)
490 pos = pos+2
491 elseif p == "o" then -- [offset]
492 if ctx.x64 then
493 local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
494 local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
495 x = format("[0x%08x%08x]", imm2, imm1)
496 pos = pos+8
497 else
498 local imm = getimm(ctx, pos, 4); if not imm then return end
499 x = format("[0x%08x]", imm)
500 pos = pos+4
501 end
502 elseif p == "i" or p == "I" then
503 local n = map_sz2n[sz]
504 if n == 8 and ctx.x64 and p == "I" then
505 local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
506 local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
507 x = format("0x%08x%08x", imm2, imm1)
508 else
509 if n == 8 then n = 4 end
510 local imm = getimm(ctx, pos, n); if not imm then return end
511 if sz == "Q" and (imm < 0 or imm > 0x7fffffff) then
512 imm = (0xffffffff+1)-imm
513 x = format(imm > 65535 and "-0x%08x" or "-0x%x", imm)
514 else
515 x = format(imm > 65535 and "0x%08x" or "0x%x", imm)
516 end
517 end
518 pos = pos+n
519 elseif p == "j" then
520 local n = map_sz2n[sz]
521 if n == 8 then n = 4 end
522 local imm = getimm(ctx, pos, n); if not imm then return end
523 if sz == "B" and imm > 127 then imm = imm-256
524 elseif imm > 2147483647 then imm = imm-4294967296 end
525 pos = pos+n
526 imm = imm + pos + ctx.addr
527 if imm > 4294967295 and not ctx.x64 then imm = imm-4294967296 end
528 ctx.imm = imm
529 if sz == "W" then
530 x = format("word 0x%04x", imm%65536)
531 elseif ctx.x64 then
532 local lo = imm % 0x1000000
533 x = format("0x%02x%06x", (imm-lo) / 0x1000000, lo)
534 else
535 x = format("0x%08x", imm)
536 end
537 elseif p == "R" then
538 local r = byte(code, pos-1, pos-1)%8
539 if ctx.rexb then r = r + 8; ctx.rexb = false end
540 x = regs[r+1]
541 elseif p == "a" then x = regs[1]
542 elseif p == "c" then x = "cl"
543 elseif p == "d" then x = "dx"
544 elseif p == "1" then x = "1"
545 else
546 if not mode then
547 mode = ctx.mrm
548 if not mode then
549 if pos > stop then return incomplete(ctx) end
550 mode = byte(code, pos, pos)
551 pos = pos+1
552 end
553 rm = mode%8; mode = (mode-rm)/8
554 sp = mode%8; mode = (mode-sp)/8
555 sdisp = ""
556 if mode < 3 then
557 if rm == 4 then
558 if pos > stop then return incomplete(ctx) end
559 sc = byte(code, pos, pos)
560 pos = pos+1
561 rm = sc%8; sc = (sc-rm)/8
562 rx = sc%8; sc = (sc-rx)/8
563 if ctx.rexx then rx = rx + 8; ctx.rexx = false end
564 if rx == 4 then rx = nil end
565 end
566 if mode > 0 or rm == 5 then
567 local dsz = mode
568 if dsz ~= 1 then dsz = 4 end
569 local disp = getimm(ctx, pos, dsz); if not disp then return end
570 if mode == 0 then rm = nil end
571 if rm or rx or (not sc and ctx.x64 and not ctx.a32) then
572 if dsz == 1 and disp > 127 then
573 sdisp = format("-0x%x", 256-disp)
574 elseif disp >= 0 and disp <= 0x7fffffff then
575 sdisp = format("+0x%x", disp)
576 else
577 sdisp = format("-0x%x", (0xffffffff+1)-disp)
578 end
579 else
580 sdisp = format(ctx.x64 and not ctx.a32 and
581 not (disp >= 0 and disp <= 0x7fffffff)
582 and "0xffffffff%08x" or "0x%08x", disp)
583 end
584 pos = pos+dsz
585 end
586 end
587 if rm and ctx.rexb then rm = rm + 8; ctx.rexb = false end
588 if ctx.rexr then sp = sp + 8; ctx.rexr = false end
589 end
590 if p == "m" then
591 if mode == 3 then x = regs[rm+1]
592 else
593 local aregs = ctx.a32 and map_regs.D or ctx.aregs
594 local srm, srx = "", ""
595 if rm then srm = aregs[rm+1]
596 elseif not sc and ctx.x64 and not ctx.a32 then srm = "rip" end
597 ctx.a32 = false
598 if rx then
599 if rm then srm = srm.."+" end
600 srx = aregs[rx+1]
601 if sc > 0 then srx = srx.."*"..(2^sc) end
602 end
603 x = format("[%s%s%s]", srm, srx, sdisp)
604 end
605 if mode < 3 and
606 (not match(pat, "[aRrgp]") or match(pat, "t")) then -- Yuck.
607 x = map_sz2prefix[sz].." "..x
608 end
609 elseif p == "r" then x = regs[sp+1]
610 elseif p == "g" then x = map_segregs[sp+1]
611 elseif p == "p" then -- Suppress prefix.
612 elseif p == "f" then x = "st"..rm
613 elseif p == "x" then
614 if sp == 0 and ctx.lock and not ctx.x64 then
615 x = "CR8"; ctx.lock = false
616 else
617 x = "CR"..sp
618 end
619 elseif p == "y" then x = "DR"..sp
620 elseif p == "z" then x = "TR"..sp
621 elseif p == "t" then
622 else
623 error("bad pattern `"..pat.."'")
624 end
625 end
626 if x then operands = operands and operands..", "..x or x end
627 end
628 ctx.pos = pos
629 return putop(ctx, name, operands)
630end
631
632-- Forward declaration.
633local map_act
634
635-- Fetch and cache MRM byte.
636local function getmrm(ctx)
637 local mrm = ctx.mrm
638 if not mrm then
639 local pos = ctx.pos
640 if pos > ctx.stop then return nil end
641 mrm = byte(ctx.code, pos, pos)
642 ctx.pos = pos+1
643 ctx.mrm = mrm
644 end
645 return mrm
646end
647
648-- Dispatch to handler depending on pattern.
649local function dispatch(ctx, opat, patgrp)
650 if not opat then return unknown(ctx) end
651 if match(opat, "%|") then -- MMX/SSE variants depending on prefix.
652 local p
653 if ctx.rep then
654 p = ctx.rep=="rep" and "%|([^%|]*)" or "%|[^%|]*%|[^%|]*%|([^%|]*)"
655 ctx.rep = false
656 elseif ctx.o16 then p = "%|[^%|]*%|([^%|]*)"; ctx.o16 = false
657 else p = "^[^%|]*" end
658 opat = match(opat, p)
659 if not opat then return unknown(ctx) end
660-- ctx.rep = false; ctx.o16 = false
661 --XXX fails for 66 f2 0f 38 f1 06 crc32 eax,WORD PTR [esi]
662 --XXX remove in branches?
663 end
664 if match(opat, "%$") then -- reg$mem variants.
665 local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
666 opat = match(opat, mrm >= 192 and "^[^%$]*" or "%$(.*)")
667 if opat == "" then return unknown(ctx) end
668 end
669 if opat == "" then return unknown(ctx) end
670 local name, pat = match(opat, "^([a-z0-9 ]*)(.*)")
671 if pat == "" and patgrp then pat = patgrp end
672 return map_act[sub(pat, 1, 1)](ctx, name, pat)
673end
674
675-- Get a pattern from an opcode map and dispatch to handler.
676local function dispatchmap(ctx, opcmap)
677 local pos = ctx.pos
678 local opat = opcmap[byte(ctx.code, pos, pos)]
679 pos = pos + 1
680 ctx.pos = pos
681 return dispatch(ctx, opat)
682end
683
684-- Map for action codes. The key is the first char after the name.
685map_act = {
686 -- Simple opcodes without operands.
687 [""] = function(ctx, name, pat)
688 return putop(ctx, name)
689 end,
690
691 -- Operand size chars fall right through.
692 B = putpat, W = putpat, D = putpat, Q = putpat,
693 V = putpat, U = putpat, T = putpat,
694 M = putpat, X = putpat, P = putpat,
695 F = putpat, G = putpat,
696
697 -- Collect prefixes.
698 [":"] = function(ctx, name, pat)
699 ctx[pat == ":" and name or sub(pat, 2)] = name
700 if ctx.pos - ctx.start > 5 then return unknown(ctx) end -- Limit #prefixes.
701 end,
702
703 -- Chain to special handler specified by name.
704 ["*"] = function(ctx, name, pat)
705 return map_act[name](ctx, name, sub(pat, 2))
706 end,
707
708 -- Use named subtable for opcode group.
709 ["!"] = function(ctx, name, pat)
710 local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
711 return dispatch(ctx, map_opcgroup[name][((mrm-(mrm%8))/8)%8+1], sub(pat, 2))
712 end,
713
714 -- o16,o32[,o64] variants.
715 sz = function(ctx, name, pat)
716 if ctx.o16 then ctx.o16 = false
717 else
718 pat = match(pat, ",(.*)")
719 if ctx.rexw then
720 local p = match(pat, ",(.*)")
721 if p then pat = p; ctx.rexw = false end
722 end
723 end
724 pat = match(pat, "^[^,]*")
725 return dispatch(ctx, pat)
726 end,
727
728 -- Two-byte opcode dispatch.
729 opc2 = function(ctx, name, pat)
730 return dispatchmap(ctx, map_opc2)
731 end,
732
733 -- Three-byte opcode dispatch.
734 opc3 = function(ctx, name, pat)
735 return dispatchmap(ctx, map_opc3[pat])
736 end,
737
738 -- VMX/SVM dispatch.
739 vm = function(ctx, name, pat)
740 return dispatch(ctx, map_opcvm[ctx.mrm])
741 end,
742
743 -- Floating point opcode dispatch.
744 fp = function(ctx, name, pat)
745 local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
746 local rm = mrm%8
747 local idx = pat*8 + ((mrm-rm)/8)%8
748 if mrm >= 192 then idx = idx + 64 end
749 local opat = map_opcfp[idx]
750 if type(opat) == "table" then opat = opat[rm+1] end
751 return dispatch(ctx, opat)
752 end,
753
754 -- REX prefix.
755 rex = function(ctx, name, pat)
756 if ctx.rex then return unknown(ctx) end -- Only 1 REX prefix allowed.
757 for p in gmatch(pat, ".") do ctx["rex"..p] = true end
758 ctx.rex = true
759 end,
760
761 -- Special case for nop with REX prefix.
762 nop = function(ctx, name, pat)
763 return dispatch(ctx, ctx.rex and pat or "nop")
764 end,
765}
766
767------------------------------------------------------------------------------
768
769-- Disassemble a block of code.
770local function disass_block(ctx, ofs, len)
771 if not ofs then ofs = 0 end
772 local stop = len and ofs+len or #ctx.code
773 ofs = ofs + 1
774 ctx.start = ofs
775 ctx.pos = ofs
776 ctx.stop = stop
777 ctx.imm = nil
778 ctx.mrm = false
779 clearprefixes(ctx)
780 while ctx.pos <= stop do dispatchmap(ctx, ctx.map1) end
781 if ctx.pos ~= ctx.start then incomplete(ctx) end
782end
783
784-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
785local function create_(code, addr, out)
786 local ctx = {}
787 ctx.code = code
788 ctx.addr = (addr or 0) - 1
789 ctx.out = out or io.write
790 ctx.symtab = {}
791 ctx.disass = disass_block
792 ctx.hexdump = 16
793 ctx.x64 = false
794 ctx.map1 = map_opc1_32
795 ctx.aregs = map_regs.D
796 return ctx
797end
798
799local function create64_(code, addr, out)
800 local ctx = create_(code, addr, out)
801 ctx.x64 = true
802 ctx.map1 = map_opc1_64
803 ctx.aregs = map_regs.Q
804 return ctx
805end
806
807-- Simple API: disassemble code (a string) at address and output via out.
808local function disass_(code, addr, out)
809 create_(code, addr, out):disass()
810end
811
812local function disass64_(code, addr, out)
813 create64_(code, addr, out):disass()
814end
815
816
817-- Public module functions.
818module(...)
819
820create = create_
821create64 = create64_
822disass = disass_
823disass64 = disass64_
824
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-16 20:24:54 +0000