lexers/asm.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- NASM Assembly LPeg lexer.
   3
   4 local l = require('lexer')
   5 local token, word_match = l.token, l.word_match
   6 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   7
   8 local M = {_NAME = 'asm'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local comment = token(l.COMMENT, ';' * l.nonnewline^0)
  15
  16 -- Strings.
  17 local sq_str = l.delimited_range("'", true)
  18 local dq_str = l.delimited_range('"', true)
  19 local string = token(l.STRING, sq_str + dq_str)
  20
  21 -- Numbers.
  22 local number = token(l.NUMBER, l.float + l.integer * S('hqb')^-1)
  23
  24 -- Preprocessor.
  25 local preproc_word = word_match{
  26   'arg', 'assign', 'clear', 'define', 'defstr', 'deftok', 'depend', 'elif',
  27   'elifctx', 'elifdef', 'elifempty', 'elifenv', 'elifid', 'elifidn', 'elifidni',
  28   'elifmacro', 'elifn', 'elifnctx', 'elifndef', 'elifnempty', 'elifnenv',
  29   'elifnid', 'elifnidn', 'elifnidni', 'elifnmacro', 'elifnnum', 'elifnstr',
  30   'elifntoken', 'elifnum', 'elifstr', 'eliftoken', 'else', 'endif', 'endmacro',
  31   'endrep', 'endwhile', 'error', 'exitmacro', 'exitrep', 'exitwhile', 'fatal',
  32   'final', 'idefine', 'idefstr', 'ideftok', 'if', 'ifctx', 'ifdef', 'ifempty',
  33   'ifenv', 'ifid', 'ifidn', 'ifidni', 'ifmacro', 'ifn', 'ifnctx', 'ifndef',
  34   'ifnempty', 'ifnenv', 'ifnid', 'ifnidn', 'ifnidni', 'ifnmacro', 'ifnnum',
  35   'ifnstr', 'ifntoken', 'ifnum', 'ifstr', 'iftoken', 'imacro', 'include',
  36   'ixdefine', 'line', 'local', 'macro', 'pathsearch', 'pop', 'push', 'rep',
  37   'repl', 'rmacro', 'rotate', 'stacksize', 'strcat', 'strlen', 'substr',
  38   'undef', 'unmacro', 'use', 'warning', 'while', 'xdefine',
  39 }
  40 local preproc_symbol = '??' + S('!$+?') + '%' * -l.space + R('09')^1
  41 local preproc = token(l.PREPROCESSOR, '%' * (preproc_word + preproc_symbol))
  42
  43 -- Keywords.
  44 local keyword = token(l.KEYWORD, word_match({
  45   -- Preprocessor macros.
  46   'struc', 'endstruc', 'istruc', 'at', 'iend', 'align', 'alignb', 'sectalign',
  47   '.nolist',
  48   -- Preprocessor Packages.
  49   --'altreg', 'smartalign', 'fp', 'ifunc'
  50   -- Directives.
  51   'absolute', 'bits', 'class', 'common', 'common', 'cpu', 'default', 'export',
  52   'extern', 'float', 'global', 'group', 'import', 'osabi', 'overlay', 'private',
  53   'public', '__SECT__', 'section', 'segment', 'stack', 'use16', 'use32',
  54   'use64',
  55   -- Section Names.
  56   '.bss', '.comment', '.data', '.lbss', '.ldata', '.lrodata', '.rdata',
  57   '.rodata', '.tbss', '.tdata', '.text',
  58   -- Section Qualifiers.
  59   'alloc', 'bss', 'code', 'exec', 'data', 'noalloc', 'nobits', 'noexec',
  60   'nowrite', 'progbits', 'rdata', 'tls', 'write',
  61   -- Operators.
  62   'abs', 'rel', 'seg', 'wrt', 'strict',
  63   '__utf16__', '__utf16be__', '__utf16le__', '__utf32__', '__utf32be__',
  64   '__utf32le__',
  65 }, '.'))
  66
  67 -- Instructions.
  68 -- awk '{print $1}'|uniq|tr '[:upper:]' '[:lower:]'|
  69 -- lua -e "for l in io.lines() do print(\"'\"..l..\"',\") end"|fmt -w 78
  70 local instruction = token('instruction', word_match{
  71   -- Special Instructions.
  72   'db', 'dd', 'do', 'dq', 'dt', 'dw', 'dy', 'resb', 'resd', 'reso', 'resq',
  73   'rest', 'resw', 'resy',
  74   -- Conventional Instructions.
  75   'aaa', 'aad', 'aam', 'aas', 'adc', 'add', 'and', 'arpl', 'bb0_reset',
  76   'bb1_reset', 'bound', 'bsf', 'bsr', 'bswap', 'bt', 'btc', 'btr', 'bts',
  77   'call', 'cbw', 'cdq', 'cdqe', 'clc', 'cld', 'cli', 'clts', 'cmc', 'cmp',
  78   'cmpsb', 'cmpsd', 'cmpsq', 'cmpsw', 'cmpxchg', 'cmpxchg486', 'cmpxchg8b',
  79   'cmpxchg16b', 'cpuid', 'cpu_read', 'cpu_write', 'cqo', 'cwd', 'cwde', 'daa',
  80   'das', 'dec', 'div', 'dmint', 'emms', 'enter', 'equ', 'f2xm1', 'fabs',
  81   'fadd', 'faddp', 'fbld', 'fbstp', 'fchs', 'fclex', 'fcmovb', 'fcmovbe',
  82   'fcmove', 'fcmovnb', 'fcmovnbe', 'fcmovne', 'fcmovnu', 'fcmovu', 'fcom',
  83   'fcomi', 'fcomip', 'fcomp', 'fcompp', 'fcos', 'fdecstp', 'fdisi', 'fdiv',
  84   'fdivp', 'fdivr', 'fdivrp', 'femms', 'feni', 'ffree', 'ffreep', 'fiadd',
  85   'ficom', 'ficomp', 'fidiv', 'fidivr', 'fild', 'fimul', 'fincstp', 'finit',
  86   'fist', 'fistp', 'fisttp', 'fisub', 'fisubr', 'fld', 'fld1', 'fldcw',
  87   'fldenv', 'fldl2e', 'fldl2t', 'fldlg2', 'fldln2', 'fldpi', 'fldz', 'fmul',
  88   'fmulp', 'fnclex', 'fndisi', 'fneni', 'fninit', 'fnop', 'fnsave', 'fnstcw',
  89   'fnstenv', 'fnstsw', 'fpatan', 'fprem', 'fprem1', 'fptan', 'frndint',
  90   'frstor', 'fsave', 'fscale', 'fsetpm', 'fsin', 'fsincos', 'fsqrt',
  91   'fst', 'fstcw', 'fstenv', 'fstp', 'fstsw', 'fsub', 'fsubp', 'fsubr',
  92   'fsubrp', 'ftst', 'fucom', 'fucomi', 'fucomip', 'fucomp', 'fucompp',
  93   'fxam', 'fxch', 'fxtract', 'fyl2x', 'fyl2xp1', 'hlt', 'ibts', 'icebp',
  94   'idiv', 'imul', 'in', 'inc', 'incbin', 'insb', 'insd', 'insw', 'int',
  95   'int01', 'int1', 'int03', 'int3', 'into', 'invd', 'invpcid', 'invlpg',
  96   'invlpga', 'iret', 'iretd', 'iretq', 'iretw', 'jcxz', 'jecxz', 'jrcxz',
  97   'jmp', 'jmpe', 'lahf', 'lar', 'lds', 'lea', 'leave', 'les', 'lfence',
  98   'lfs', 'lgdt', 'lgs', 'lidt', 'lldt', 'lmsw', 'loadall', 'loadall286',
  99   'lodsb', 'lodsd', 'lodsq', 'lodsw', 'loop', 'loope', 'loopne', 'loopnz',
 100   'loopz', 'lsl', 'lss', 'ltr', 'mfence', 'monitor', 'mov', 'movd', 'movq',
 101   'movsb', 'movsd', 'movsq', 'movsw', 'movsx', 'movsxd', 'movsx', 'movzx',
 102   'mul', 'mwait', 'neg', 'nop', 'not', 'or', 'out', 'outsb', 'outsd', 'outsw',
 103   'packssdw', 'packsswb', 'packuswb', 'paddb', 'paddd', 'paddsb', 'paddsiw',
 104   'paddsw', 'paddusb', 'paddusw', 'paddw', 'pand', 'pandn', 'pause', 'paveb',
 105   'pavgusb', 'pcmpeqb', 'pcmpeqd', 'pcmpeqw', 'pcmpgtb', 'pcmpgtd', 'pcmpgtw',
 106   'pdistib', 'pf2id', 'pfacc', 'pfadd', 'pfcmpeq', 'pfcmpge', 'pfcmpgt',
 107   'pfmax', 'pfmin', 'pfmul', 'pfrcp', 'pfrcpit1', 'pfrcpit2', 'pfrsqit1',
 108   'pfrsqrt', 'pfsub', 'pfsubr', 'pi2fd', 'pmachriw', 'pmaddwd', 'pmagw',
 109   'pmulhriw', 'pmulhrwa', 'pmulhrwc', 'pmulhw', 'pmullw', 'pmvgezb', 'pmvlzb',
 110   'pmvnzb', 'pmvzb', 'pop', 'popa', 'popad', 'popaw', 'popf', 'popfd',
 111   'popfq', 'popfw', 'por', 'prefetch', 'prefetchw', 'pslld', 'psllq',
 112   'psllw', 'psrad', 'psraw', 'psrld', 'psrlq', 'psrlw', 'psubb', 'psubd',
 113   'psubsb', 'psubsiw', 'psubsw', 'psubusb', 'psubusw', 'psubw', 'punpckhbw',
 114   'punpckhdq', 'punpckhwd', 'punpcklbw', 'punpckldq', 'punpcklwd', 'push',
 115   'pusha', 'pushad', 'pushaw', 'pushf', 'pushfd', 'pushfq', 'pushfw', 'pxor',
 116   'rcl', 'rcr', 'rdshr', 'rdmsr', 'rdpmc', 'rdtsc', 'rdtscp', 'ret', 'retf',
 117   'retn', 'rol', 'ror', 'rdm', 'rsdc', 'rsldt', 'rsm', 'rsts', 'sahf', 'sal',
 118   'salc', 'sar', 'sbb', 'scasb', 'scasd', 'scasq', 'scasw', 'sfence', 'sgdt',
 119   'shl', 'shld', 'shr', 'shrd', 'sidt', 'sldt', 'skinit', 'smi', 'smint',
 120   'smintold', 'smsw', 'stc', 'std', 'sti', 'stosb', 'stosd', 'stosq', 'stosw',
 121   'str', 'sub', 'svdc', 'svldt', 'svts', 'swapgs', 'syscall', 'sysenter',
 122   'sysexit', 'sysret', 'test', 'ud0', 'ud1', 'ud2b', 'ud2', 'ud2a', 'umov',
 123   'verr', 'verw', 'fwait', 'wbinvd', 'wrshr', 'wrmsr', 'xadd', 'xbts',
 124   'xchg', 'xlatb', 'xlat', 'xor', 'cmovcc', 'jcc', 'setcc',
 125   -- Katmai Streaming SIMD instructions (SSE -- a.k.a. KNI, XMM, MMX2).
 126   'addps', 'addss', 'andnps', 'andps', 'cmpeqps', 'cmpeqss', 'cmpleps',
 127   'cmpless', 'cmpltps', 'cmpltss', 'cmpneqps', 'cmpneqss', 'cmpnleps',
 128   'cmpnless', 'cmpnltps', 'cmpnltss', 'cmpordps', 'cmpordss', 'cmpunordps',
 129   'cmpunordss', 'cmpps', 'cmpss', 'comiss', 'cvtpi2ps', 'cvtps2pi', 'cvtsi2ss',
 130   'cvtss2si', 'cvttps2pi', 'cvttss2si', 'divps', 'divss', 'ldmxcsr', 'maxps',
 131   'maxss', 'minps', 'minss', 'movaps', 'movhps', 'movlhps', 'movlps',
 132   'movhlps', 'movmskps', 'movntps', 'movss', 'movups', 'mulps', 'mulss',
 133   'orps', 'rcpps', 'rcpss', 'rsqrtps', 'rsqrtss', 'shufps', 'sqrtps', 'sqrtss',
 134   'stmxcsr', 'subps', 'subss', 'ucomiss', 'unpckhps', 'unpcklps', 'xorps',
 135   -- Introduced in Deschutes but necessary for SSE support.
 136   'fxrstor', 'fxrstor64', 'fxsave', 'fxsave64',
 137   -- XSAVE group (AVX and extended state).
 138   'xgetbv', 'xsetbv', 'xsave', 'xsave64', 'xsaveopt', 'xsaveopt64', 'xrstor',
 139   'xrstor64',
 140   -- Generic memory operations.
 141   'prefetchnta', 'prefetcht0', 'prefetcht1', 'prefetcht2', 'sfence',
 142   -- New MMX instructions introduced in Katmai.
 143   'maskmovq', 'movntq', 'pavgb', 'pavgw', 'pextrw', 'pinsrw', 'pmaxsw',
 144   'pmaxub', 'pminsw', 'pminub', 'pmovmskb', 'pmulhuw', 'psadbw', 'pshufw',
 145   -- AMD Enhanced 3DNow! (Athlon) instructions.
 146   'pf2iw', 'pfnacc', 'pfpnacc', 'pi2fw', 'pswapd',
 147   -- Willamette SSE2 Cacheability Instructions.
 148   'maskmovdqu', 'clflush', 'movntdq', 'movnti', 'movntpd', 'lfence', 'mfence',
 149   -- Willamette MMX instructions (SSE2 SIMD Integer Instructions).
 150   'movd', 'movdqa', 'movdqu', 'movdq2q', 'movq', 'movq2dq', 'packsswb',
 151   'packssdw', 'packuswb', 'paddb', 'paddw', 'paddd', 'paddq', 'paddsb',
 152   'paddsw', 'paddusb', 'paddusw', 'pand', 'pandn', 'pavgb', 'pavgw', 'pcmpeqb',
 153   'pcmpeqw', 'pcmpeqd', 'pcmpgtb', 'pcmpgtw', 'pcmpgtd', 'pextrw', 'pinsrw',
 154   'pmaddwd', 'pmaxsw', 'pmaxub', 'pminsw', 'pminub', 'pmovmskb', 'pmulhuw',
 155   'pmulhw', 'pmullw', 'pmuludq', 'por', 'psadbw', 'pshufd', 'pshufhw',
 156   'pshuflw', 'pslldq', 'psllw', 'pslld', 'psllq', 'psraw', 'psrad', 'psrldq',
 157   'psrlw', 'psrld', 'psrlq', 'psubb', 'psubw', 'psubd', 'psubq', 'psubsb',
 158   'psubsw', 'psubusb', 'psubusw', 'punpckhbw', 'punpckhwd', 'punpckhdq',
 159   'punpckhqdq', 'punpcklbw', 'punpcklwd', 'punpckldq', 'punpcklqdq', 'pxor',
 160   -- Willamette Streaming SIMD instructions (SSE2).
 161   'addpd', 'addsd', 'andnpd', 'andpd', 'cmpeqpd', 'cmpeqsd', 'cmplepd',
 162   'cmplesd', 'cmpltpd', 'cmpltsd', 'cmpneqpd', 'cmpneqsd', 'cmpnlepd',
 163   'cmpnlesd', 'cmpnltpd', 'cmpnltsd', 'cmpordpd', 'cmpordsd', 'cmpunordpd',
 164   'cmpunordsd', 'cmppd', 'cmpsd', 'comisd', 'cvtdq2pd', 'cvtdq2ps',
 165   'cvtpd2dq', 'cvtpd2pi', 'cvtpd2ps', 'cvtpi2pd', 'cvtps2dq', 'cvtps2pd',
 166   'cvtsd2si', 'cvtsd2ss', 'cvtsi2sd', 'cvtss2sd', 'cvttpd2pi', 'cvttpd2dq',
 167   'cvttps2dq', 'cvttsd2si', 'divpd', 'divsd', 'maxpd', 'maxsd', 'minpd',
 168   'minsd', 'movapd', 'movhpd', 'movlpd', 'movmskpd', 'movsd', 'movupd',
 169   'mulpd', 'mulsd', 'orpd', 'shufpd', 'sqrtpd', 'sqrtsd', 'subpd', 'subsd',
 170   'ucomisd', 'unpckhpd', 'unpcklpd', 'xorpd',
 171   -- Prescott New Instructions (SSE3).
 172   'addsubpd', 'addsubps', 'haddpd', 'haddps', 'hsubpd', 'hsubps', 'lddqu',
 173   'movddup', 'movshdup', 'movsldup',
 174   -- VMX/SVM Instructions.
 175   'clgi', 'stgi', 'vmcall', 'vmclear', 'vmfunc', 'vmlaunch', 'vmload',
 176   'vmmcall', 'vmptrld', 'vmptrst', 'vmread', 'vmresume', 'vmrun', 'vmsave',
 177   'vmwrite', 'vmxoff', 'vmxon',
 178   -- Extended Page Tables VMX instructions.
 179   'invept', 'invvpid',
 180   -- Tejas New Instructions (SSSE3).
 181   'pabsb', 'pabsw', 'pabsd', 'palignr', 'phaddw', 'phaddd', 'phaddsw',
 182   'phsubw', 'phsubd', 'phsubsw', 'pmaddubsw', 'pmulhrsw', 'pshufb', 'psignb',
 183   'psignw', 'psignd',
 184   -- AMD SSE4A.
 185   'extrq', 'insertq', 'movntsd', 'movntss',
 186   -- New instructions in Barcelona.
 187   'lzcnt',
 188   -- Penryn New Instructions (SSE4.1).
 189   'blendpd', 'blendps', 'blendvpd', 'blendvps', 'dppd', 'dpps', 'extractps',
 190   'insertps', 'movntdqa', 'mpsadbw', 'packusdw', 'pblendvb', 'pblendw',
 191   'pcmpeqq', 'pextrb', 'pextrd', 'pextrq', 'pextrw', 'phminposuw', 'pinsrb',
 192   'pinsrd', 'pinsrq', 'pmaxsb', 'pmaxsd', 'pmaxud', 'pmaxuw', 'pminsb',
 193   'pminsd', 'pminud', 'pminuw', 'pmovsxbw', 'pmovsxbd', 'pmovsxbq', 'pmovsxwd',
 194   'pmovsxwq', 'pmovsxdq', 'pmovzxbw', 'pmovzxbd', 'pmovzxbq', 'pmovzxwd',
 195   'pmovzxwq', 'pmovzxdq', 'pmuldq', 'pmulld', 'ptest', 'roundpd', 'roundps',
 196   'roundsd', 'roundss',
 197   -- Nehalem New Instructions (SSE4.2).
 198   'crc32', 'pcmpestri', 'pcmpestrm', 'pcmpistri', 'pcmpistrm', 'pcmpgtq',
 199   'popcnt',
 200   -- Intel SMX.
 201   'getsec',
 202   -- Geode (Cyrix) 3DNow! additions.
 203   'pfrcpv', 'pfrsqrtv',
 204   -- Intel new instructions in ???.
 205   'movbe',
 206   -- Intel AES instructions.
 207   'aesenc', 'aesenclast', 'aesdec', 'aesdeclast', 'aesimc', 'aeskeygenassist',
 208   -- Intel AVX AES instructions.
 209   'vaesenc', 'vaesenclast', 'vaesdec', 'vaesdeclast', 'vaesimc',
 210   'vaeskeygenassist',
 211   -- Intel AVX instructions.
 212   'vaddpd', 'vaddps', 'vaddsd', 'vaddss', 'vaddsubpd', 'vaddsubps',
 213   'vandpd', 'vandps', 'vandnpd', 'vandnps', 'vblendpd', 'vblendps',
 214   'vblendvpd', 'vblendvps', 'vbroadcastss', 'vbroadcastsd', 'vbroadcastf128',
 215   'vcmpeq_ospd', 'vcmpeqpd', 'vcmplt_ospd', 'vcmpltpd', 'vcmple_ospd',
 216   'vcmplepd', 'vcmpunord_qpd', 'vcmpunordpd', 'vcmpneq_uqpd', 'vcmpneqpd',
 217   'vcmpnlt_uspd', 'vcmpnltpd', 'vcmpnle_uspd', 'vcmpnlepd', 'vcmpord_qpd',
 218   'vcmpordpd', 'vcmpeq_uqpd', 'vcmpnge_uspd', 'vcmpngepd', 'vcmpngt_uspd',
 219   'vcmpngtpd', 'vcmpfalse_oqpd', 'vcmpfalsepd', 'vcmpneq_oqpd', 'vcmpge_ospd',
 220   'vcmpgepd', 'vcmpgt_ospd', 'vcmpgtpd', 'vcmptrue_uqpd', 'vcmptruepd',
 221   'vcmpeq_ospd', 'vcmplt_oqpd', 'vcmple_oqpd', 'vcmpunord_spd', 'vcmpneq_uspd',
 222   'vcmpnlt_uqpd', 'vcmpnle_uqpd', 'vcmpord_spd', 'vcmpeq_uspd', 'vcmpnge_uqpd',
 223   'vcmpngt_uqpd', 'vcmpfalse_ospd', 'vcmpneq_ospd', 'vcmpge_oqpd',
 224   'vcmpgt_oqpd', 'vcmptrue_uspd', 'vcmppd', 'vcmpeq_osps', 'vcmpeqps',
 225   'vcmplt_osps', 'vcmpltps', 'vcmple_osps', 'vcmpleps', 'vcmpunord_qps',
 226   'vcmpunordps', 'vcmpneq_uqps', 'vcmpneqps', 'vcmpnlt_usps', 'vcmpnltps',
 227   'vcmpnle_usps', 'vcmpnleps', 'vcmpord_qps', 'vcmpordps', 'vcmpeq_uqps',
 228   'vcmpnge_usps', 'vcmpngeps', 'vcmpngt_usps', 'vcmpngtps', 'vcmpfalse_oqps',
 229   'vcmpfalseps', 'vcmpneq_oqps', 'vcmpge_osps', 'vcmpgeps', 'vcmpgt_osps',
 230   'vcmpgtps', 'vcmptrue_uqps', 'vcmptrueps', 'vcmpeq_osps', 'vcmplt_oqps',
 231   'vcmple_oqps', 'vcmpunord_sps', 'vcmpneq_usps', 'vcmpnlt_uqps',
 232   'vcmpnle_uqps', 'vcmpord_sps', 'vcmpeq_usps', 'vcmpnge_uqps',
 233   'vcmpngt_uqps', 'vcmpfalse_osps', 'vcmpneq_osps', 'vcmpge_oqps',
 234   'vcmpgt_oqps', 'vcmptrue_usps', 'vcmpps', 'vcmpeq_ossd', 'vcmpeqsd',
 235   'vcmplt_ossd', 'vcmpltsd', 'vcmple_ossd', 'vcmplesd', 'vcmpunord_qsd',
 236   'vcmpunordsd', 'vcmpneq_uqsd', 'vcmpneqsd', 'vcmpnlt_ussd', 'vcmpnltsd',
 237   'vcmpnle_ussd', 'vcmpnlesd', 'vcmpord_qsd', 'vcmpordsd', 'vcmpeq_uqsd',
 238   'vcmpnge_ussd', 'vcmpngesd', 'vcmpngt_ussd', 'vcmpngtsd', 'vcmpfalse_oqsd',
 239   'vcmpfalsesd', 'vcmpneq_oqsd', 'vcmpge_ossd', 'vcmpgesd', 'vcmpgt_ossd',
 240   'vcmpgtsd', 'vcmptrue_uqsd', 'vcmptruesd', 'vcmpeq_ossd', 'vcmplt_oqsd',
 241   'vcmple_oqsd', 'vcmpunord_ssd', 'vcmpneq_ussd', 'vcmpnlt_uqsd',
 242   'vcmpnle_uqsd', 'vcmpord_ssd', 'vcmpeq_ussd', 'vcmpnge_uqsd',
 243   'vcmpngt_uqsd', 'vcmpfalse_ossd', 'vcmpneq_ossd', 'vcmpge_oqsd',
 244   'vcmpgt_oqsd', 'vcmptrue_ussd', 'vcmpsd', 'vcmpeq_osss', 'vcmpeqss',
 245   'vcmplt_osss', 'vcmpltss', 'vcmple_osss', 'vcmpless', 'vcmpunord_qss',
 246   'vcmpunordss', 'vcmpneq_uqss', 'vcmpneqss', 'vcmpnlt_usss', 'vcmpnltss',
 247   'vcmpnle_usss', 'vcmpnless', 'vcmpord_qss', 'vcmpordss', 'vcmpeq_uqss',
 248   'vcmpnge_usss', 'vcmpngess', 'vcmpngt_usss', 'vcmpngtss', 'vcmpfalse_oqss',
 249   'vcmpfalsess', 'vcmpneq_oqss', 'vcmpge_osss', 'vcmpgess', 'vcmpgt_osss',
 250   'vcmpgtss', 'vcmptrue_uqss', 'vcmptruess', 'vcmpeq_osss', 'vcmplt_oqss',
 251   'vcmple_oqss', 'vcmpunord_sss', 'vcmpneq_usss', 'vcmpnlt_uqss',
 252   'vcmpnle_uqss', 'vcmpord_sss', 'vcmpeq_usss', 'vcmpnge_uqss',
 253   'vcmpngt_uqss', 'vcmpfalse_osss', 'vcmpneq_osss', 'vcmpge_oqss',
 254   'vcmpgt_oqss', 'vcmptrue_usss', 'vcmpss', 'vcomisd', 'vcomiss',
 255   'vcvtdq2pd', 'vcvtdq2ps', 'vcvtpd2dq', 'vcvtpd2ps', 'vcvtps2dq',
 256   'vcvtps2pd', 'vcvtsd2si', 'vcvtsd2ss', 'vcvtsi2sd', 'vcvtsi2ss',
 257   'vcvtss2sd', 'vcvtss2si', 'vcvttpd2dq', 'vcvttps2dq', 'vcvttsd2si',
 258   'vcvttss2si', 'vdivpd', 'vdivps', 'vdivsd', 'vdivss', 'vdppd', 'vdpps',
 259   'vextractf128', 'vextractps', 'vhaddpd', 'vhaddps', 'vhsubpd', 'vhsubps',
 260   'vinsertf128', 'vinsertps', 'vlddqu', 'vldqqu', 'vlddqu', 'vldmxcsr',
 261   'vmaskmovdqu', 'vmaskmovps', 'vmaskmovpd', 'vmaxpd', 'vmaxps', 'vmaxsd',
 262   'vmaxss', 'vminpd', 'vminps', 'vminsd', 'vminss', 'vmovapd', 'vmovaps',
 263   'vmovd', 'vmovq', 'vmovddup', 'vmovdqa', 'vmovqqa', 'vmovdqa', 'vmovdqu',
 264   'vmovqqu', 'vmovdqu', 'vmovhlps', 'vmovhpd', 'vmovhps', 'vmovlhps',
 265   'vmovlpd', 'vmovlps', 'vmovmskpd', 'vmovmskps', 'vmovntdq', 'vmovntqq',
 266   'vmovntdq', 'vmovntdqa', 'vmovntpd', 'vmovntps', 'vmovsd', 'vmovshdup',
 267   'vmovsldup', 'vmovss', 'vmovupd', 'vmovups', 'vmpsadbw', 'vmulpd',
 268   'vmulps', 'vmulsd', 'vmulss', 'vorpd', 'vorps', 'vpabsb', 'vpabsw',
 269   'vpabsd', 'vpacksswb', 'vpackssdw', 'vpackuswb', 'vpackusdw', 'vpaddb',
 270   'vpaddw', 'vpaddd', 'vpaddq', 'vpaddsb', 'vpaddsw', 'vpaddusb', 'vpaddusw',
 271   'vpalignr', 'vpand', 'vpandn', 'vpavgb', 'vpavgw', 'vpblendvb', 'vpblendw',
 272   'vpcmpestri', 'vpcmpestrm', 'vpcmpistri', 'vpcmpistrm', 'vpcmpeqb',
 273   'vpcmpeqw', 'vpcmpeqd', 'vpcmpeqq', 'vpcmpgtb', 'vpcmpgtw', 'vpcmpgtd',
 274   'vpcmpgtq', 'vpermilpd', 'vpermilps', 'vperm2f128', 'vpextrb', 'vpextrw',
 275   'vpextrd', 'vpextrq', 'vphaddw', 'vphaddd', 'vphaddsw', 'vphminposuw',
 276   'vphsubw', 'vphsubd', 'vphsubsw', 'vpinsrb', 'vpinsrw', 'vpinsrd',
 277   'vpinsrq', 'vpmaddwd', 'vpmaddubsw', 'vpmaxsb', 'vpmaxsw', 'vpmaxsd',
 278   'vpmaxub', 'vpmaxuw', 'vpmaxud', 'vpminsb', 'vpminsw', 'vpminsd', 'vpminub',
 279   'vpminuw', 'vpminud', 'vpmovmskb', 'vpmovsxbw', 'vpmovsxbd', 'vpmovsxbq',
 280   'vpmovsxwd', 'vpmovsxwq', 'vpmovsxdq', 'vpmovzxbw', 'vpmovzxbd', 'vpmovzxbq',
 281   'vpmovzxwd', 'vpmovzxwq', 'vpmovzxdq', 'vpmulhuw', 'vpmulhrsw', 'vpmulhw',
 282   'vpmullw', 'vpmulld', 'vpmuludq', 'vpmuldq', 'vpor', 'vpsadbw', 'vpshufb',
 283   'vpshufd', 'vpshufhw', 'vpshuflw', 'vpsignb', 'vpsignw', 'vpsignd',
 284   'vpslldq', 'vpsrldq', 'vpsllw', 'vpslld', 'vpsllq', 'vpsraw', 'vpsrad',
 285   'vpsrlw', 'vpsrld', 'vpsrlq', 'vptest', 'vpsubb', 'vpsubw', 'vpsubd',
 286   'vpsubq', 'vpsubsb', 'vpsubsw', 'vpsubusb', 'vpsubusw', 'vpunpckhbw',
 287   'vpunpckhwd', 'vpunpckhdq', 'vpunpckhqdq', 'vpunpcklbw', 'vpunpcklwd',
 288   'vpunpckldq', 'vpunpcklqdq', 'vpxor', 'vrcpps', 'vrcpss', 'vrsqrtps',
 289   'vrsqrtss', 'vroundpd', 'vroundps', 'vroundsd', 'vroundss', 'vshufpd',
 290   'vshufps', 'vsqrtpd', 'vsqrtps', 'vsqrtsd', 'vsqrtss', 'vstmxcsr', 'vsubpd',
 291   'vsubps', 'vsubsd', 'vsubss', 'vtestps', 'vtestpd', 'vucomisd', 'vucomiss',
 292   'vunpckhpd', 'vunpckhps', 'vunpcklpd', 'vunpcklps', 'vxorpd', 'vxorps',
 293   'vzeroall', 'vzeroupper',
 294   -- Intel Carry-Less Multiplication instructions (CLMUL).
 295   'pclmullqlqdq', 'pclmulhqlqdq', 'pclmullqhqdq', 'pclmulhqhqdq', 'pclmulqdq',
 296   -- Intel AVX Carry-Less Multiplication instructions (CLMUL).
 297   'vpclmullqlqdq', 'vpclmulhqlqdq', 'vpclmullqhqdq', 'vpclmulhqhqdq',
 298   'vpclmulqdq',
 299   -- Intel Fused Multiply-Add instructions (FMA).
 300   'vfmadd132ps', 'vfmadd132pd', 'vfmadd312ps', 'vfmadd312pd', 'vfmadd213ps',
 301   'vfmadd213pd', 'vfmadd123ps', 'vfmadd123pd', 'vfmadd231ps', 'vfmadd231pd',
 302   'vfmadd321ps', 'vfmadd321pd', 'vfmaddsub132ps', 'vfmaddsub132pd',
 303   'vfmaddsub312ps', 'vfmaddsub312pd', 'vfmaddsub213ps', 'vfmaddsub213pd',
 304   'vfmaddsub123ps', 'vfmaddsub123pd', 'vfmaddsub231ps', 'vfmaddsub231pd',
 305   'vfmaddsub321ps', 'vfmaddsub321pd', 'vfmsub132ps', 'vfmsub132pd',
 306   'vfmsub312ps', 'vfmsub312pd', 'vfmsub213ps', 'vfmsub213pd', 'vfmsub123ps',
 307   'vfmsub123pd', 'vfmsub231ps', 'vfmsub231pd', 'vfmsub321ps', 'vfmsub321pd',
 308   'vfmsubadd132ps', 'vfmsubadd132pd', 'vfmsubadd312ps', 'vfmsubadd312pd',
 309   'vfmsubadd213ps', 'vfmsubadd213pd', 'vfmsubadd123ps', 'vfmsubadd123pd',
 310   'vfmsubadd231ps', 'vfmsubadd231pd', 'vfmsubadd321ps', 'vfmsubadd321pd',
 311   'vfnmadd132ps', 'vfnmadd132pd', 'vfnmadd312ps', 'vfnmadd312pd',
 312   'vfnmadd213ps', 'vfnmadd213pd', 'vfnmadd123ps', 'vfnmadd123pd',
 313   'vfnmadd231ps', 'vfnmadd231pd', 'vfnmadd321ps', 'vfnmadd321pd',
 314   'vfnmsub132ps', 'vfnmsub132pd', 'vfnmsub312ps', 'vfnmsub312pd',
 315   'vfnmsub213ps', 'vfnmsub213pd', 'vfnmsub123ps', 'vfnmsub123pd',
 316   'vfnmsub231ps', 'vfnmsub231pd', 'vfnmsub321ps', 'vfnmsub321pd',
 317   'vfmadd132ss', 'vfmadd132sd', 'vfmadd312ss', 'vfmadd312sd', 'vfmadd213ss',
 318   'vfmadd213sd', 'vfmadd123ss', 'vfmadd123sd', 'vfmadd231ss', 'vfmadd231sd',
 319   'vfmadd321ss', 'vfmadd321sd', 'vfmsub132ss', 'vfmsub132sd', 'vfmsub312ss',
 320   'vfmsub312sd', 'vfmsub213ss', 'vfmsub213sd', 'vfmsub123ss', 'vfmsub123sd',
 321   'vfmsub231ss', 'vfmsub231sd', 'vfmsub321ss', 'vfmsub321sd', 'vfnmadd132ss',
 322   'vfnmadd132sd', 'vfnmadd312ss', 'vfnmadd312sd', 'vfnmadd213ss',
 323   'vfnmadd213sd', 'vfnmadd123ss', 'vfnmadd123sd', 'vfnmadd231ss',
 324   'vfnmadd231sd', 'vfnmadd321ss', 'vfnmadd321sd', 'vfnmsub132ss',
 325   'vfnmsub132sd', 'vfnmsub312ss', 'vfnmsub312sd', 'vfnmsub213ss',
 326   'vfnmsub213sd', 'vfnmsub123ss', 'vfnmsub123sd', 'vfnmsub231ss',
 327   'vfnmsub231sd', 'vfnmsub321ss', 'vfnmsub321sd',
 328   -- Intel post-32 nm processor instructions.
 329   'rdfsbase', 'rdgsbase', 'rdrand', 'wrfsbase', 'wrgsbase', 'vcvtph2ps',
 330   'vcvtps2ph', 'adcx', 'adox', 'rdseed', 'clac', 'stac',
 331   -- VIA (Centaur) security instructions.
 332   'xstore', 'xcryptecb', 'xcryptcbc', 'xcryptctr', 'xcryptcfb', 'xcryptofb',
 333   'montmul', 'xsha1', 'xsha256',
 334   -- AMD Lightweight Profiling (LWP) instructions.
 335   'llwpcb', 'slwpcb', 'lwpval', 'lwpins',
 336   -- AMD XOP and FMA4 instructions (SSE5).
 337   'vfmaddpd', 'vfmaddps', 'vfmaddsd', 'vfmaddss', 'vfmaddsubpd',
 338   'vfmaddsubps', 'vfmsubaddpd', 'vfmsubaddps', 'vfmsubpd', 'vfmsubps',
 339   'vfmsubsd', 'vfmsubss', 'vfnmaddpd', 'vfnmaddps', 'vfnmaddsd', 'vfnmaddss',
 340   'vfnmsubpd', 'vfnmsubps', 'vfnmsubsd', 'vfnmsubss', 'vfrczpd', 'vfrczps',
 341   'vfrczsd', 'vfrczss', 'vpcmov', 'vpcomb', 'vpcomd', 'vpcomq', 'vpcomub',
 342   'vpcomud', 'vpcomuq', 'vpcomuw', 'vpcomw', 'vphaddbd', 'vphaddbq',
 343   'vphaddbw', 'vphadddq', 'vphaddubd', 'vphaddubq', 'vphaddubw', 'vphaddudq',
 344   'vphadduwd', 'vphadduwq', 'vphaddwd', 'vphaddwq', 'vphsubbw', 'vphsubdq',
 345   'vphsubwd', 'vpmacsdd', 'vpmacsdqh', 'vpmacsdql', 'vpmacssdd', 'vpmacssdqh',
 346   'vpmacssdql', 'vpmacsswd', 'vpmacssww', 'vpmacswd', 'vpmacsww', 'vpmadcsswd',
 347   'vpmadcswd', 'vpperm', 'vprotb', 'vprotd', 'vprotq', 'vprotw', 'vpshab',
 348   'vpshad', 'vpshaq', 'vpshaw', 'vpshlb', 'vpshld', 'vpshlq', 'vpshlw',
 349   -- Intel AVX2 instructions.
 350   'vmpsadbw', 'vpabsb', 'vpabsw', 'vpabsd', 'vpacksswb', 'vpackssdw',
 351   'vpackusdw', 'vpackuswb', 'vpaddb', 'vpaddw', 'vpaddd', 'vpaddq',
 352   'vpaddsb', 'vpaddsw', 'vpaddusb', 'vpaddusw', 'vpalignr', 'vpand',
 353   'vpandn', 'vpavgb', 'vpavgw', 'vpblendvb', 'vpblendw', 'vpcmpeqb',
 354   'vpcmpeqw', 'vpcmpeqd', 'vpcmpeqq', 'vpcmpgtb', 'vpcmpgtw', 'vpcmpgtd',
 355   'vpcmpgtq', 'vphaddw', 'vphaddd', 'vphaddsw', 'vphsubw', 'vphsubd',
 356   'vphsubsw', 'vpmaddubsw', 'vpmaddwd', 'vpmaxsb', 'vpmaxsw', 'vpmaxsd',
 357   'vpmaxub', 'vpmaxuw', 'vpmaxud', 'vpminsb', 'vpminsw', 'vpminsd', 'vpminub',
 358   'vpminuw', 'vpminud', 'vpmovmskb', 'vpmovsxbw', 'vpmovsxbd', 'vpmovsxbq',
 359   'vpmovsxwd', 'vpmovsxwq', 'vpmovsxdq', 'vpmovzxbw', 'vpmovzxbd', 'vpmovzxbq',
 360   'vpmovzxwd', 'vpmovzxwq', 'vpmovzxdq', 'vpmuldq', 'vpmulhrsw', 'vpmulhuw',
 361   'vpmulhw', 'vpmullw', 'vpmulld', 'vpmuludq', 'vpor', 'vpsadbw', 'vpshufb',
 362   'vpshufd', 'vpshufhw', 'vpshuflw', 'vpsignb', 'vpsignw', 'vpsignd',
 363   'vpslldq', 'vpsllw', 'vpslld', 'vpsllq', 'vpsraw', 'vpsrad', 'vpsrldq',
 364   'vpsrlw', 'vpsrld', 'vpsrlq', 'vpsubb', 'vpsubw', 'vpsubd', 'vpsubq',
 365   'vpsubsb', 'vpsubsw', 'vpsubusb', 'vpsubusw', 'vpunpckhbw', 'vpunpckhwd',
 366   'vpunpckhdq', 'vpunpckhqdq', 'vpunpcklbw', 'vpunpcklwd', 'vpunpckldq',
 367   'vpunpcklqdq', 'vpxor', 'vmovntdqa', 'vbroadcastss', 'vbroadcastsd',
 368   'vbroadcasti128', 'vpblendd', 'vpbroadcastb', 'vpbroadcastw', 'vpbroadcastd',
 369   'vpbroadcastq', 'vpermd', 'vpermpd', 'vpermps', 'vpermq', 'vperm2i128',
 370   'vextracti128', 'vinserti128', 'vpmaskmovd', 'vpmaskmovq', 'vpmaskmovd',
 371   'vpmaskmovq', 'vpsllvd', 'vpsllvq', 'vpsllvd', 'vpsllvq', 'vpsravd',
 372   'vpsrlvd', 'vpsrlvq', 'vpsrlvd', 'vpsrlvq', 'vgatherdpd', 'vgatherqpd',
 373   'vgatherdpd', 'vgatherqpd', 'vgatherdps', 'vgatherqps', 'vgatherdps',
 374   'vgatherqps', 'vpgatherdd', 'vpgatherqd', 'vpgatherdd', 'vpgatherqd',
 375   'vpgatherdq', 'vpgatherqq', 'vpgatherdq', 'vpgatherqq',
 376   -- Transactional Synchronization Extensions (TSX).
 377   'xabort', 'xbegin', 'xend', 'xtest',
 378   -- Intel BMI1 and BMI2 instructions, AMD TBM instructions.
 379   'andn', 'bextr', 'blci', 'blcic', 'blsi', 'blsic', 'blcfill', 'blsfill',
 380   'blcmsk', 'blsmsk', 'blsr', 'blcs', 'bzhi', 'mulx', 'pdep', 'pext', 'rorx',
 381   'sarx', 'shlx', 'shrx', 'tzcnt', 'tzmsk', 't1mskc',
 382   -- Systematic names for the hinting nop instructions.
 383   'hint_nop0', 'hint_nop1', 'hint_nop2', 'hint_nop3', 'hint_nop4',
 384   'hint_nop5', 'hint_nop6', 'hint_nop7', 'hint_nop8', 'hint_nop9',
 385   'hint_nop10', 'hint_nop11', 'hint_nop12', 'hint_nop13', 'hint_nop14',
 386   'hint_nop15', 'hint_nop16', 'hint_nop17', 'hint_nop18', 'hint_nop19',
 387   'hint_nop20', 'hint_nop21', 'hint_nop22', 'hint_nop23', 'hint_nop24',
 388   'hint_nop25', 'hint_nop26', 'hint_nop27', 'hint_nop28', 'hint_nop29',
 389   'hint_nop30', 'hint_nop31', 'hint_nop32', 'hint_nop33', 'hint_nop34',
 390   'hint_nop35', 'hint_nop36', 'hint_nop37', 'hint_nop38', 'hint_nop39',
 391   'hint_nop40', 'hint_nop41', 'hint_nop42', 'hint_nop43', 'hint_nop44',
 392   'hint_nop45', 'hint_nop46', 'hint_nop47', 'hint_nop48', 'hint_nop49',
 393   'hint_nop50', 'hint_nop51', 'hint_nop52', 'hint_nop53', 'hint_nop54',
 394   'hint_nop55', 'hint_nop56', 'hint_nop57', 'hint_nop58', 'hint_nop59',
 395   'hint_nop60', 'hint_nop61', 'hint_nop62', 'hint_nop63',
 396 })
 397
 398 -- Types.
 399 local sizes = word_match{
 400   'byte', 'word', 'dword', 'qword', 'tword', 'oword', 'yword',
 401   'a16', 'a32', 'a64', 'o16', 'o32', 'o64' -- instructions
 402 }
 403 local wrt_types = '..' * word_match{
 404   'start', 'gotpc', 'gotoff', 'gottpoff', 'got', 'plt', 'sym', 'tlsie'
 405 }
 406 local type = token(l.TYPE, sizes + wrt_types)
 407
 408 -- Registers.
 409 local register = token('register', word_match{
 410   -- 32-bit registers.
 411   'ah', 'al', 'ax', 'bh', 'bl', 'bp', 'bx', 'ch', 'cl', 'cx', 'dh', 'di', 'dl',
 412   'dx', 'eax', 'ebx', 'ebx', 'ecx', 'edi', 'edx', 'esi', 'esp', 'fs', 'mm0',
 413   'mm1', 'mm2', 'mm3', 'mm4', 'mm5', 'mm6', 'mm7', 'si', 'st0', 'st1', 'st2',
 414   'st3', 'st4', 'st5', 'st6', 'st7', 'xmm0', 'xmm1', 'xmm2', 'xmm3', 'xmm4',
 415   'xmm5', 'xmm6', 'xmm7', 'ymm0', 'ymm1', 'ymm2', 'ymm3', 'ymm4', 'ymm5',
 416   'ymm6', 'ymm7',
 417   -- 64-bit registers.
 418   'bpl', 'dil', 'gs', 'r8', 'r8b', 'r8w', 'r9', 'r9b', 'r9w', 'r10', 'r10b',
 419   'r10w', 'r11', 'r11b', 'r11w', 'r12', 'r12b', 'r12w', 'r13', 'r13b', 'r13w',
 420   'r14', 'r14b', 'r14w', 'r15', 'r15b', 'r15w', 'rax', 'rbp', 'rbx', 'rcx',
 421   'rdi', 'rdx', 'rsi', 'rsp', 'sil', 'xmm8', 'xmm9', 'xmm10', 'xmm11', 'xmm12',
 422   'xmm13', 'xmm14', 'xmm15', 'ymm8', 'ymm9', 'ymm10', 'ymm11', 'ymm12', 'ymm13',
 423   'ymm14', 'ymm15'
 424 })
 425
 426 local word = (l.alpha + S('$._?')) * (l.alnum + S('$._?#@~'))^0
 427
 428 -- Labels.
 429 local label = token(l.LABEL, word * ':')
 430
 431 -- Identifiers.
 432 local identifier = token(l.IDENTIFIER, word)
 433
 434 -- Constants.
 435 local constants = word_match{
 436   '__float8__', '__float16__', '__float32__', '__float64__', '__float80m__',
 437   '__float80e__', '__float128l__', '__float128h__', '__Infinity__', '__QNaN__',
 438   '__NaN__', '__SNaN__'
 439 }
 440 local constant = token(l.CONSTANT, constants + '$' * P('$')^-1 * -identifier)
 441
 442 -- Operators.
 443 local operator = token(l.OPERATOR, S('+-/*%<>!=^&|~:,()[]'))
 444
 445 M._rules = {
 446   {'whitespace', ws},
 447   {'keyword', keyword},
 448   {'instruction', instruction},
 449   {'register', register},
 450   {'type', type},
 451   {'constant', constant},
 452   {'label', label},
 453   {'identifier', identifier},
 454   {'string', string},
 455   {'comment', comment},
 456   {'number', number},
 457   {'preproc', preproc},
 458   {'operator', operator},
 459 }
 460
 461 M._tokenstyles = {
 462   instruction = l.STYLE_FUNCTION,
 463   register = l.STYLE_CONSTANT,
 464 }
 465
 466 M._foldsymbols = {
 467   _patterns = {'%l+', '//'},
 468   [l.PREPROCESSOR] = {
 469     ['if'] = 1, endif = -1, macro = 1, endmacro = -1, rep = 1, endrep = -1,
 470     ['while'] = 1, endwhile = -1,
 471   },
 472   [l.KEYWORD] = {struc = 1, endstruc = -1},
 473   [l.COMMENT] = {['//'] = l.fold_line_comments('//')}
 474 }
 475
 476 return M