lib/CodeGen/MIRParser/MILexer.cpp

   1 //===- MILexer.cpp - Machine instructions lexer implementation ------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file implements the lexing of machine instructions.
  10 //
  11 //===----------------------------------------------------------------------===//
  12
  13 #include "MILexer.h"
  14 #include "llvm/ADT/APSInt.h"
  15 #include "llvm/ADT/None.h"
  16 #include "llvm/ADT/STLExtras.h"
  17 #include "llvm/ADT/StringExtras.h"
  18 #include "llvm/ADT/StringSwitch.h"
  19 #include "llvm/ADT/StringRef.h"
  20 #include "llvm/ADT/Twine.h"
  21 #include <algorithm>
  22 #include <cassert>
  23 #include <cctype>
  24 #include <string>
  25
  26 using namespace llvm;
  27
  28 namespace {
  29
  30 using ErrorCallbackType =
  31     function_ref<void(StringRef::iterator Loc, const Twine &)>;
  32
  33 /// This class provides a way to iterate and get characters from the source
  34 /// string.
  35 class Cursor {
  36   const char *Ptr = nullptr;
  37   const char *End = nullptr;
  38
  39 public:
  40   Cursor(NoneType) {}
  41
  42   explicit Cursor(StringRef Str) {
  43     Ptr = Str.data();
  44     End = Ptr + Str.size();
  45   }
  46
  47   bool isEOF() const { return Ptr == End; }
  48
  49   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
  50
  51   void advance(unsigned I = 1) { Ptr += I; }
  52
  53   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
  54
  55   StringRef upto(Cursor C) const {
  56     assert(C.Ptr >= Ptr && C.Ptr <= End);
  57     return StringRef(Ptr, C.Ptr - Ptr);
  58   }
  59
  60   StringRef::iterator location() const { return Ptr; }
  61
  62   operator bool() const { return Ptr != nullptr; }
  63 };
  64
  65 } // end anonymous namespace
  66
  67 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
  68   this->Kind = Kind;
  69   this->Range = Range;
  70   return *this;
  71 }
  72
  73 MIToken &MIToken::setStringValue(StringRef StrVal) {
  74   StringValue = StrVal;
  75   return *this;
  76 }
  77
  78 MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
  79   StringValueStorage = std::move(StrVal);
  80   StringValue = StringValueStorage;
  81   return *this;
  82 }
  83
  84 MIToken &MIToken::setIntegerValue(APSInt IntVal) {
  85   this->IntVal = std::move(IntVal);
  86   return *this;
  87 }
  88
  89 /// Skip the leading whitespace characters and return the updated cursor.
  90 static Cursor skipWhitespace(Cursor C) {
  91   while (isblank(C.peek()))
  92     C.advance();
  93   return C;
  94 }
  95
  96 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
  97
  98 /// Skip a line comment and return the updated cursor.
  99 static Cursor skipComment(Cursor C) {
 100   if (C.peek() != ';')
 101     return C;
 102   while (!isNewlineChar(C.peek()) && !C.isEOF())
 103     C.advance();
 104   return C;
 105 }
 106
 107 /// Return true if the given character satisfies the following regular
 108 /// expression: [-a-zA-Z$._0-9]
 109 static bool isIdentifierChar(char C) {
 110   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
 111          C == '$';
 112 }
 113
 114 /// Unescapes the given string value.
 115 ///
 116 /// Expects the string value to be quoted.
 117 static std::string unescapeQuotedString(StringRef Value) {
 118   assert(Value.front() == '"' && Value.back() == '"');
 119   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
 120
 121   std::string Str;
 122   Str.reserve(C.remaining().size());
 123   while (!C.isEOF()) {
 124     char Char = C.peek();
 125     if (Char == '\\') {
 126       if (C.peek(1) == '\\') {
 127         // Two '\' become one
 128         Str += '\\';
 129         C.advance(2);
 130         continue;
 131       }
 132       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
 133         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
 134         C.advance(3);
 135         continue;
 136       }
 137     }
 138     Str += Char;
 139     C.advance();
 140   }
 141   return Str;
 142 }
 143
 144 /// Lex a string constant using the following regular expression: \"[^\"]*\"
 145 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
 146   assert(C.peek() == '"');
 147   for (C.advance(); C.peek() != '"'; C.advance()) {
 148     if (C.isEOF() || isNewlineChar(C.peek())) {
 149       ErrorCallback(
 150           C.location(),
 151           "end of machine instruction reached before the closing '\"'");
 152       return None;
 153     }
 154   }
 155   C.advance();
 156   return C;
 157 }
 158
 159 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
 160                       unsigned PrefixLength, ErrorCallbackType ErrorCallback) {
 161   auto Range = C;
 162   C.advance(PrefixLength);
 163   if (C.peek() == '"') {
 164     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
 165       StringRef String = Range.upto(R);
 166       Token.reset(Type, String)
 167           .setOwnedStringValue(
 168               unescapeQuotedString(String.drop_front(PrefixLength)));
 169       return R;
 170     }
 171     Token.reset(MIToken::Error, Range.remaining());
 172     return Range;
 173   }
 174   while (isIdentifierChar(C.peek()))
 175     C.advance();
 176   Token.reset(Type, Range.upto(C))
 177       .setStringValue(Range.upto(C).drop_front(PrefixLength));
 178   return C;
 179 }
 180
 181 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
 182   return StringSwitch<MIToken::TokenKind>(Identifier)
 183       .Case("_", MIToken::underscore)
 184       .Case("implicit", MIToken::kw_implicit)
 185       .Case("implicit-def", MIToken::kw_implicit_define)
 186       .Case("def", MIToken::kw_def)
 187       .Case("dead", MIToken::kw_dead)
 188       .Case("killed", MIToken::kw_killed)
 189       .Case("undef", MIToken::kw_undef)
 190       .Case("internal", MIToken::kw_internal)
 191       .Case("early-clobber", MIToken::kw_early_clobber)
 192       .Case("debug-use", MIToken::kw_debug_use)
 193       .Case("renamable", MIToken::kw_renamable)
 194       .Case("tied-def", MIToken::kw_tied_def)
 195       .Case("frame-setup", MIToken::kw_frame_setup)
 196       .Case("frame-destroy", MIToken::kw_frame_destroy)
 197       .Case("nnan", MIToken::kw_nnan)
 198       .Case("ninf", MIToken::kw_ninf)
 199       .Case("nsz", MIToken::kw_nsz)
 200       .Case("arcp", MIToken::kw_arcp)
 201       .Case("contract", MIToken::kw_contract)
 202       .Case("afn", MIToken::kw_afn)
 203       .Case("reassoc", MIToken::kw_reassoc)
 204       .Case("nuw" , MIToken::kw_nuw)
 205       .Case("nsw" , MIToken::kw_nsw)
 206       .Case("exact" , MIToken::kw_exact)
 207       .Case("fpexcept", MIToken::kw_fpexcept)
 208       .Case("debug-location", MIToken::kw_debug_location)
 209       .Case("same_value", MIToken::kw_cfi_same_value)
 210       .Case("offset", MIToken::kw_cfi_offset)
 211       .Case("rel_offset", MIToken::kw_cfi_rel_offset)
 212       .Case("def_cfa_register", MIToken::kw_cfi_def_cfa_register)
 213       .Case("def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
 214       .Case("adjust_cfa_offset", MIToken::kw_cfi_adjust_cfa_offset)
 215       .Case("escape", MIToken::kw_cfi_escape)
 216       .Case("def_cfa", MIToken::kw_cfi_def_cfa)
 217       .Case("remember_state", MIToken::kw_cfi_remember_state)
 218       .Case("restore", MIToken::kw_cfi_restore)
 219       .Case("restore_state", MIToken::kw_cfi_restore_state)
 220       .Case("undefined", MIToken::kw_cfi_undefined)
 221       .Case("register", MIToken::kw_cfi_register)
 222       .Case("window_save", MIToken::kw_cfi_window_save)
 223       .Case("negate_ra_sign_state", MIToken::kw_cfi_aarch64_negate_ra_sign_state)
 224       .Case("blockaddress", MIToken::kw_blockaddress)
 225       .Case("intrinsic", MIToken::kw_intrinsic)
 226       .Case("target-index", MIToken::kw_target_index)
 227       .Case("half", MIToken::kw_half)
 228       .Case("float", MIToken::kw_float)
 229       .Case("double", MIToken::kw_double)
 230       .Case("x86_fp80", MIToken::kw_x86_fp80)
 231       .Case("fp128", MIToken::kw_fp128)
 232       .Case("ppc_fp128", MIToken::kw_ppc_fp128)
 233       .Case("target-flags", MIToken::kw_target_flags)
 234       .Case("volatile", MIToken::kw_volatile)
 235       .Case("non-temporal", MIToken::kw_non_temporal)
 236       .Case("dereferenceable", MIToken::kw_dereferenceable)
 237       .Case("invariant", MIToken::kw_invariant)
 238       .Case("align", MIToken::kw_align)
 239       .Case("addrspace", MIToken::kw_addrspace)
 240       .Case("stack", MIToken::kw_stack)
 241       .Case("got", MIToken::kw_got)
 242       .Case("jump-table", MIToken::kw_jump_table)
 243       .Case("constant-pool", MIToken::kw_constant_pool)
 244       .Case("call-entry", MIToken::kw_call_entry)
 245       .Case("liveout", MIToken::kw_liveout)
 246       .Case("address-taken", MIToken::kw_address_taken)
 247       .Case("landing-pad", MIToken::kw_landing_pad)
 248       .Case("liveins", MIToken::kw_liveins)
 249       .Case("successors", MIToken::kw_successors)
 250       .Case("floatpred", MIToken::kw_floatpred)
 251       .Case("intpred", MIToken::kw_intpred)
 252       .Case("shufflemask", MIToken::kw_shufflemask)
 253       .Case("pre-instr-symbol", MIToken::kw_pre_instr_symbol)
 254       .Case("post-instr-symbol", MIToken::kw_post_instr_symbol)
 255       .Case("unknown-size", MIToken::kw_unknown_size)
 256       .Default(MIToken::Identifier);
 257 }
 258
 259 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
 260   if (!isalpha(C.peek()) && C.peek() != '_')
 261     return None;
 262   auto Range = C;
 263   while (isIdentifierChar(C.peek()))
 264     C.advance();
 265   auto Identifier = Range.upto(C);
 266   Token.reset(getIdentifierKind(Identifier), Identifier)
 267       .setStringValue(Identifier);
 268   return C;
 269 }
 270
 271 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
 272                                         ErrorCallbackType ErrorCallback) {
 273   bool IsReference = C.remaining().startswith("%bb.");
 274   if (!IsReference && !C.remaining().startswith("bb."))
 275     return None;
 276   auto Range = C;
 277   unsigned PrefixLength = IsReference ? 4 : 3;
 278   C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
 279   if (!isdigit(C.peek())) {
 280     Token.reset(MIToken::Error, C.remaining());
 281     ErrorCallback(C.location(), "expected a number after '%bb.'");
 282     return C;
 283   }
 284   auto NumberRange = C;
 285   while (isdigit(C.peek()))
 286     C.advance();
 287   StringRef Number = NumberRange.upto(C);
 288   unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
 289   // TODO: The format bb.<id>.<irname> is supported only when it's not a
 290   // reference. Once we deprecate the format where the irname shows up, we
 291   // should only lex forward if it is a reference.
 292   if (C.peek() == '.') {
 293     C.advance(); // Skip '.'
 294     ++StringOffset;
 295     while (isIdentifierChar(C.peek()))
 296       C.advance();
 297   }
 298   Token.reset(IsReference ? MIToken::MachineBasicBlock
 299                           : MIToken::MachineBasicBlockLabel,
 300               Range.upto(C))
 301       .setIntegerValue(APSInt(Number))
 302       .setStringValue(Range.upto(C).drop_front(StringOffset));
 303   return C;
 304 }
 305
 306 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
 307                             MIToken::TokenKind Kind) {
 308   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 309     return None;
 310   auto Range = C;
 311   C.advance(Rule.size());
 312   auto NumberRange = C;
 313   while (isdigit(C.peek()))
 314     C.advance();
 315   Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
 316   return C;
 317 }
 318
 319 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
 320                                    MIToken::TokenKind Kind) {
 321   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 322     return None;
 323   auto Range = C;
 324   C.advance(Rule.size());
 325   auto NumberRange = C;
 326   while (isdigit(C.peek()))
 327     C.advance();
 328   StringRef Number = NumberRange.upto(C);
 329   unsigned StringOffset = Rule.size() + Number.size();
 330   if (C.peek() == '.') {
 331     C.advance();
 332     ++StringOffset;
 333     while (isIdentifierChar(C.peek()))
 334       C.advance();
 335   }
 336   Token.reset(Kind, Range.upto(C))
 337       .setIntegerValue(APSInt(Number))
 338       .setStringValue(Range.upto(C).drop_front(StringOffset));
 339   return C;
 340 }
 341
 342 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
 343   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
 344 }
 345
 346 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
 347   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
 348 }
 349
 350 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
 351   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
 352 }
 353
 354 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
 355   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
 356 }
 357
 358 static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token,
 359                                        ErrorCallbackType ErrorCallback) {
 360   const StringRef Rule = "%subreg.";
 361   if (!C.remaining().startswith(Rule))
 362     return None;
 363   return lexName(C, Token, MIToken::SubRegisterIndex, Rule.size(),
 364                  ErrorCallback);
 365 }
 366
 367 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
 368                               ErrorCallbackType ErrorCallback) {
 369   const StringRef Rule = "%ir-block.";
 370   if (!C.remaining().startswith(Rule))
 371     return None;
 372   if (isdigit(C.peek(Rule.size())))
 373     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
 374   return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
 375 }
 376
 377 static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
 378                               ErrorCallbackType ErrorCallback) {
 379   const StringRef Rule = "%ir.";
 380   if (!C.remaining().startswith(Rule))
 381     return None;
 382   if (isdigit(C.peek(Rule.size())))
 383     return maybeLexIndex(C, Token, Rule, MIToken::IRValue);
 384   return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
 385 }
 386
 387 static Cursor maybeLexStringConstant(Cursor C, MIToken &Token,
 388                                      ErrorCallbackType ErrorCallback) {
 389   if (C.peek() != '"')
 390     return None;
 391   return lexName(C, Token, MIToken::StringConstant, /*PrefixLength=*/0,
 392                  ErrorCallback);
 393 }
 394
 395 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
 396   auto Range = C;
 397   C.advance(); // Skip '%'
 398   auto NumberRange = C;
 399   while (isdigit(C.peek()))
 400     C.advance();
 401   Token.reset(MIToken::VirtualRegister, Range.upto(C))
 402       .setIntegerValue(APSInt(NumberRange.upto(C)));
 403   return C;
 404 }
 405
 406 /// Returns true for a character allowed in a register name.
 407 static bool isRegisterChar(char C) {
 408   return isIdentifierChar(C) && C != '.';
 409 }
 410
 411 static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) {
 412   Cursor Range = C;
 413   C.advance(); // Skip '%'
 414   while (isRegisterChar(C.peek()))
 415     C.advance();
 416   Token.reset(MIToken::NamedVirtualRegister, Range.upto(C))
 417       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
 418   return C;
 419 }
 420
 421 static Cursor maybeLexRegister(Cursor C, MIToken &Token,
 422                                ErrorCallbackType ErrorCallback) {
 423   if (C.peek() != '%' && C.peek() != '$')
 424     return None;
 425
 426   if (C.peek() == '%') {
 427     if (isdigit(C.peek(1)))
 428       return lexVirtualRegister(C, Token);
 429
 430     if (isRegisterChar(C.peek(1)))
 431       return lexNamedVirtualRegister(C, Token);
 432
 433     return None;
 434   }
 435
 436   assert(C.peek() == '$');
 437   auto Range = C;
 438   C.advance(); // Skip '$'
 439   while (isRegisterChar(C.peek()))
 440     C.advance();
 441   Token.reset(MIToken::NamedRegister, Range.upto(C))
 442       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '$'
 443   return C;
 444 }
 445
 446 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
 447                                   ErrorCallbackType ErrorCallback) {
 448   if (C.peek() != '@')
 449     return None;
 450   if (!isdigit(C.peek(1)))
 451     return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
 452                    ErrorCallback);
 453   auto Range = C;
 454   C.advance(1); // Skip the '@'
 455   auto NumberRange = C;
 456   while (isdigit(C.peek()))
 457     C.advance();
 458   Token.reset(MIToken::GlobalValue, Range.upto(C))
 459       .setIntegerValue(APSInt(NumberRange.upto(C)));
 460   return C;
 461 }
 462
 463 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
 464                                      ErrorCallbackType ErrorCallback) {
 465   if (C.peek() != '&')
 466     return None;
 467   return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1,
 468                  ErrorCallback);
 469 }
 470
 471 static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token,
 472                                ErrorCallbackType ErrorCallback) {
 473   const StringRef Rule = "<mcsymbol ";
 474   if (!C.remaining().startswith(Rule))
 475     return None;
 476   auto Start = C;
 477   C.advance(Rule.size());
 478
 479   // Try a simple unquoted name.
 480   if (C.peek() != '"') {
 481     while (isIdentifierChar(C.peek()))
 482       C.advance();
 483     StringRef String = Start.upto(C).drop_front(Rule.size());
 484     if (C.peek() != '>') {
 485       ErrorCallback(C.location(),
 486                     "expected the '<mcsymbol ...' to be closed by a '>'");
 487       Token.reset(MIToken::Error, Start.remaining());
 488       return Start;
 489     }
 490     C.advance();
 491
 492     Token.reset(MIToken::MCSymbol, Start.upto(C)).setStringValue(String);
 493     return C;
 494   }
 495
 496   // Otherwise lex out a quoted name.
 497   Cursor R = lexStringConstant(C, ErrorCallback);
 498   if (!R) {
 499     ErrorCallback(C.location(),
 500                   "unable to parse quoted string from opening quote");
 501     Token.reset(MIToken::Error, Start.remaining());
 502     return Start;
 503   }
 504   StringRef String = Start.upto(R).drop_front(Rule.size());
 505   if (R.peek() != '>') {
 506     ErrorCallback(R.location(),
 507                   "expected the '<mcsymbol ...' to be closed by a '>'");
 508     Token.reset(MIToken::Error, Start.remaining());
 509     return Start;
 510   }
 511   R.advance();
 512
 513   Token.reset(MIToken::MCSymbol, Start.upto(R))
 514       .setOwnedStringValue(unescapeQuotedString(String));
 515   return R;
 516 }
 517
 518 static bool isValidHexFloatingPointPrefix(char C) {
 519   return C == 'H' || C == 'K' || C == 'L' || C == 'M';
 520 }
 521
 522 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
 523   C.advance();
 524   // Skip over [0-9]*([eE][-+]?[0-9]+)?
 525   while (isdigit(C.peek()))
 526     C.advance();
 527   if ((C.peek() == 'e' || C.peek() == 'E') &&
 528       (isdigit(C.peek(1)) ||
 529        ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) {
 530     C.advance(2);
 531     while (isdigit(C.peek()))
 532       C.advance();
 533   }
 534   Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 535   return C;
 536 }
 537
 538 static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) {
 539   if (C.peek() != '0' || (C.peek(1) != 'x' && C.peek(1) != 'X'))
 540     return None;
 541   Cursor Range = C;
 542   C.advance(2);
 543   unsigned PrefLen = 2;
 544   if (isValidHexFloatingPointPrefix(C.peek())) {
 545     C.advance();
 546     PrefLen++;
 547   }
 548   while (isxdigit(C.peek()))
 549     C.advance();
 550   StringRef StrVal = Range.upto(C);
 551   if (StrVal.size() <= PrefLen)
 552     return None;
 553   if (PrefLen == 2)
 554     Token.reset(MIToken::HexLiteral, Range.upto(C));
 555   else // It must be 3, which means that there was a floating-point prefix.
 556     Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 557   return C;
 558 }
 559
 560 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
 561   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
 562     return None;
 563   auto Range = C;
 564   C.advance();
 565   while (isdigit(C.peek()))
 566     C.advance();
 567   if (C.peek() == '.')
 568     return lexFloatingPointLiteral(Range, C, Token);
 569   StringRef StrVal = Range.upto(C);
 570   Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal));
 571   return C;
 572 }
 573
 574 static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) {
 575   return StringSwitch<MIToken::TokenKind>(Identifier)
 576       .Case("!tbaa", MIToken::md_tbaa)
 577       .Case("!alias.scope", MIToken::md_alias_scope)
 578       .Case("!noalias", MIToken::md_noalias)
 579       .Case("!range", MIToken::md_range)
 580       .Case("!DIExpression", MIToken::md_diexpr)
 581       .Case("!DILocation", MIToken::md_dilocation)
 582       .Default(MIToken::Error);
 583 }
 584
 585 static Cursor maybeLexExlaim(Cursor C, MIToken &Token,
 586                              ErrorCallbackType ErrorCallback) {
 587   if (C.peek() != '!')
 588     return None;
 589   auto Range = C;
 590   C.advance(1);
 591   if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
 592     Token.reset(MIToken::exclaim, Range.upto(C));
 593     return C;
 594   }
 595   while (isIdentifierChar(C.peek()))
 596     C.advance();
 597   StringRef StrVal = Range.upto(C);
 598   Token.reset(getMetadataKeywordKind(StrVal), StrVal);
 599   if (Token.isError())
 600     ErrorCallback(Token.location(),
 601                   "use of unknown metadata keyword '" + StrVal + "'");
 602   return C;
 603 }
 604
 605 static MIToken::TokenKind symbolToken(char C) {
 606   switch (C) {
 607   case ',':
 608     return MIToken::comma;
 609   case '.':
 610     return MIToken::dot;
 611   case '=':
 612     return MIToken::equal;
 613   case ':':
 614     return MIToken::colon;
 615   case '(':
 616     return MIToken::lparen;
 617   case ')':
 618     return MIToken::rparen;
 619   case '{':
 620     return MIToken::lbrace;
 621   case '}':
 622     return MIToken::rbrace;
 623   case '+':
 624     return MIToken::plus;
 625   case '-':
 626     return MIToken::minus;
 627   case '<':
 628     return MIToken::less;
 629   case '>':
 630     return MIToken::greater;
 631   default:
 632     return MIToken::Error;
 633   }
 634 }
 635
 636 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
 637   MIToken::TokenKind Kind;
 638   unsigned Length = 1;
 639   if (C.peek() == ':' && C.peek(1) == ':') {
 640     Kind = MIToken::coloncolon;
 641     Length = 2;
 642   } else
 643     Kind = symbolToken(C.peek());
 644   if (Kind == MIToken::Error)
 645     return None;
 646   auto Range = C;
 647   C.advance(Length);
 648   Token.reset(Kind, Range.upto(C));
 649   return C;
 650 }
 651
 652 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
 653   if (!isNewlineChar(C.peek()))
 654     return None;
 655   auto Range = C;
 656   C.advance();
 657   Token.reset(MIToken::Newline, Range.upto(C));
 658   return C;
 659 }
 660
 661 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
 662                                      ErrorCallbackType ErrorCallback) {
 663   if (C.peek() != '`')
 664     return None;
 665   auto Range = C;
 666   C.advance();
 667   auto StrRange = C;
 668   while (C.peek() != '`') {
 669     if (C.isEOF() || isNewlineChar(C.peek())) {
 670       ErrorCallback(
 671           C.location(),
 672           "end of machine instruction reached before the closing '`'");
 673       Token.reset(MIToken::Error, Range.remaining());
 674       return C;
 675     }
 676     C.advance();
 677   }
 678   StringRef Value = StrRange.upto(C);
 679   C.advance();
 680   Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value);
 681   return C;
 682 }
 683
 684 StringRef llvm::lexMIToken(StringRef Source, MIToken &Token,
 685                            ErrorCallbackType ErrorCallback) {
 686   auto C = skipComment(skipWhitespace(Cursor(Source)));
 687   if (C.isEOF()) {
 688     Token.reset(MIToken::Eof, C.remaining());
 689     return C.remaining();
 690   }
 691
 692   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
 693     return R.remaining();
 694   if (Cursor R = maybeLexIdentifier(C, Token))
 695     return R.remaining();
 696   if (Cursor R = maybeLexJumpTableIndex(C, Token))
 697     return R.remaining();
 698   if (Cursor R = maybeLexStackObject(C, Token))
 699     return R.remaining();
 700   if (Cursor R = maybeLexFixedStackObject(C, Token))
 701     return R.remaining();
 702   if (Cursor R = maybeLexConstantPoolItem(C, Token))
 703     return R.remaining();
 704   if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback))
 705     return R.remaining();
 706   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
 707     return R.remaining();
 708   if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
 709     return R.remaining();
 710   if (Cursor R = maybeLexRegister(C, Token, ErrorCallback))
 711     return R.remaining();
 712   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
 713     return R.remaining();
 714   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
 715     return R.remaining();
 716   if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback))
 717     return R.remaining();
 718   if (Cursor R = maybeLexHexadecimalLiteral(C, Token))
 719     return R.remaining();
 720   if (Cursor R = maybeLexNumericalLiteral(C, Token))
 721     return R.remaining();
 722   if (Cursor R = maybeLexExlaim(C, Token, ErrorCallback))
 723     return R.remaining();
 724   if (Cursor R = maybeLexSymbol(C, Token))
 725     return R.remaining();
 726   if (Cursor R = maybeLexNewline(C, Token))
 727     return R.remaining();
 728   if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
 729     return R.remaining();
 730   if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback))
 731     return R.remaining();
 732
 733   Token.reset(MIToken::Error, C.remaining());
 734   ErrorCallback(C.location(),
 735                 Twine("unexpected character '") + Twine(C.peek()) + "'");
 736   return C.remaining();
 737 }