llvm/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/None.h"
  15 #include "llvm/ADT/StringExtras.h"
  16 #include "llvm/ADT/StringSwitch.h"
  17 #include "llvm/ADT/Twine.h"
  18 #include <algorithm>
  19 #include <cassert>
  20 #include <cctype>
  21 #include <string>
  22
  23 using namespace llvm;
  24
  25 namespace {
  26
  27 using ErrorCallbackType =
  28     function_ref<void(StringRef::iterator Loc, const Twine &)>;
  29
  30 /// This class provides a way to iterate and get characters from the source
  31 /// string.
  32 class Cursor {
  33   const char *Ptr = nullptr;
  34   const char *End = nullptr;
  35
  36 public:
  37   Cursor(NoneType) {}
  38
  39   explicit Cursor(StringRef Str) {
  40     Ptr = Str.data();
  41     End = Ptr + Str.size();
  42   }
  43
  44   bool isEOF() const { return Ptr == End; }
  45
  46   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
  47
  48   void advance(unsigned I = 1) { Ptr += I; }
  49
  50   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
  51
  52   StringRef upto(Cursor C) const {
  53     assert(C.Ptr >= Ptr && C.Ptr <= End);
  54     return StringRef(Ptr, C.Ptr - Ptr);
  55   }
  56
  57   StringRef::iterator location() const { return Ptr; }
  58
  59   operator bool() const { return Ptr != nullptr; }
  60 };
  61
  62 } // end anonymous namespace
  63
  64 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
  65   this->Kind = Kind;
  66   this->Range = Range;
  67   return *this;
  68 }
  69
  70 MIToken &MIToken::setStringValue(StringRef StrVal) {
  71   StringValue = StrVal;
  72   return *this;
  73 }
  74
  75 MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
  76   StringValueStorage = std::move(StrVal);
  77   StringValue = StringValueStorage;
  78   return *this;
  79 }
  80
  81 MIToken &MIToken::setIntegerValue(APSInt IntVal) {
  82   this->IntVal = std::move(IntVal);
  83   return *this;
  84 }
  85
  86 /// Skip the leading whitespace characters and return the updated cursor.
  87 static Cursor skipWhitespace(Cursor C) {
  88   while (isblank(C.peek()))
  89     C.advance();
  90   return C;
  91 }
  92
  93 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
  94
  95 /// Skip a line comment and return the updated cursor.
  96 static Cursor skipComment(Cursor C) {
  97   if (C.peek() != ';')
  98     return C;
  99   while (!isNewlineChar(C.peek()) && !C.isEOF())
 100     C.advance();
 101   return C;
 102 }
 103
 104 /// Machine operands can have comments, enclosed between /* and */.
 105 /// This eats up all tokens, including /* and */.
 106 static Cursor skipMachineOperandComment(Cursor C) {
 107   if (C.peek() != '/' || C.peek(1) != '*')
 108     return C;
 109
 110   while (C.peek() != '*' || C.peek(1) != '/')
 111     C.advance();
 112
 113   C.advance();
 114   C.advance();
 115   return C;
 116 }
 117
 118 /// Return true if the given character satisfies the following regular
 119 /// expression: [-a-zA-Z$._0-9]
 120 static bool isIdentifierChar(char C) {
 121   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
 122          C == '$';
 123 }
 124
 125 /// Unescapes the given string value.
 126 ///
 127 /// Expects the string value to be quoted.
 128 static std::string unescapeQuotedString(StringRef Value) {
 129   assert(Value.front() == '"' && Value.back() == '"');
 130   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
 131
 132   std::string Str;
 133   Str.reserve(C.remaining().size());
 134   while (!C.isEOF()) {
 135     char Char = C.peek();
 136     if (Char == '\\') {
 137       if (C.peek(1) == '\\') {
 138         // Two '\' become one
 139         Str += '\\';
 140         C.advance(2);
 141         continue;
 142       }
 143       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
 144         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
 145         C.advance(3);
 146         continue;
 147       }
 148     }
 149     Str += Char;
 150     C.advance();
 151   }
 152   return Str;
 153 }
 154
 155 /// Lex a string constant using the following regular expression: \"[^\"]*\"
 156 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
 157   assert(C.peek() == '"');
 158   for (C.advance(); C.peek() != '"'; C.advance()) {
 159     if (C.isEOF() || isNewlineChar(C.peek())) {
 160       ErrorCallback(
 161           C.location(),
 162           "end of machine instruction reached before the closing '\"'");
 163       return None;
 164     }
 165   }
 166   C.advance();
 167   return C;
 168 }
 169
 170 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
 171                       unsigned PrefixLength, ErrorCallbackType ErrorCallback) {
 172   auto Range = C;
 173   C.advance(PrefixLength);
 174   if (C.peek() == '"') {
 175     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
 176       StringRef String = Range.upto(R);
 177       Token.reset(Type, String)
 178           .setOwnedStringValue(
 179               unescapeQuotedString(String.drop_front(PrefixLength)));
 180       return R;
 181     }
 182     Token.reset(MIToken::Error, Range.remaining());
 183     return Range;
 184   }
 185   while (isIdentifierChar(C.peek()))
 186     C.advance();
 187   Token.reset(Type, Range.upto(C))
 188       .setStringValue(Range.upto(C).drop_front(PrefixLength));
 189   return C;
 190 }
 191
 192 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
 193   return StringSwitch<MIToken::TokenKind>(Identifier)
 194       .Case("_", MIToken::underscore)
 195       .Case("implicit", MIToken::kw_implicit)
 196       .Case("implicit-def", MIToken::kw_implicit_define)
 197       .Case("def", MIToken::kw_def)
 198       .Case("dead", MIToken::kw_dead)
 199       .Case("killed", MIToken::kw_killed)
 200       .Case("undef", MIToken::kw_undef)
 201       .Case("internal", MIToken::kw_internal)
 202       .Case("early-clobber", MIToken::kw_early_clobber)
 203       .Case("debug-use", MIToken::kw_debug_use)
 204       .Case("renamable", MIToken::kw_renamable)
 205       .Case("tied-def", MIToken::kw_tied_def)
 206       .Case("frame-setup", MIToken::kw_frame_setup)
 207       .Case("frame-destroy", MIToken::kw_frame_destroy)
 208       .Case("nnan", MIToken::kw_nnan)
 209       .Case("ninf", MIToken::kw_ninf)
 210       .Case("nsz", MIToken::kw_nsz)
 211       .Case("arcp", MIToken::kw_arcp)
 212       .Case("contract", MIToken::kw_contract)
 213       .Case("afn", MIToken::kw_afn)
 214       .Case("reassoc", MIToken::kw_reassoc)
 215       .Case("nuw", MIToken::kw_nuw)
 216       .Case("nsw", MIToken::kw_nsw)
 217       .Case("exact", MIToken::kw_exact)
 218       .Case("nofpexcept", MIToken::kw_nofpexcept)
 219       .Case("debug-location", MIToken::kw_debug_location)
 220       .Case("debug-instr-number", MIToken::kw_debug_instr_number)
 221       .Case("same_value", MIToken::kw_cfi_same_value)
 222       .Case("offset", MIToken::kw_cfi_offset)
 223       .Case("rel_offset", MIToken::kw_cfi_rel_offset)
 224       .Case("def_cfa_register", MIToken::kw_cfi_def_cfa_register)
 225       .Case("def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
 226       .Case("adjust_cfa_offset", MIToken::kw_cfi_adjust_cfa_offset)
 227       .Case("escape", MIToken::kw_cfi_escape)
 228       .Case("def_cfa", MIToken::kw_cfi_def_cfa)
 229       .Case("llvm_def_aspace_cfa", MIToken::kw_cfi_llvm_def_aspace_cfa)
 230       .Case("remember_state", MIToken::kw_cfi_remember_state)
 231       .Case("restore", MIToken::kw_cfi_restore)
 232       .Case("restore_state", MIToken::kw_cfi_restore_state)
 233       .Case("undefined", MIToken::kw_cfi_undefined)
 234       .Case("register", MIToken::kw_cfi_register)
 235       .Case("window_save", MIToken::kw_cfi_window_save)
 236       .Case("negate_ra_sign_state",
 237             MIToken::kw_cfi_aarch64_negate_ra_sign_state)
 238       .Case("blockaddress", MIToken::kw_blockaddress)
 239       .Case("intrinsic", MIToken::kw_intrinsic)
 240       .Case("target-index", MIToken::kw_target_index)
 241       .Case("half", MIToken::kw_half)
 242       .Case("float", MIToken::kw_float)
 243       .Case("double", MIToken::kw_double)
 244       .Case("x86_fp80", MIToken::kw_x86_fp80)
 245       .Case("fp128", MIToken::kw_fp128)
 246       .Case("ppc_fp128", MIToken::kw_ppc_fp128)
 247       .Case("target-flags", MIToken::kw_target_flags)
 248       .Case("volatile", MIToken::kw_volatile)
 249       .Case("non-temporal", MIToken::kw_non_temporal)
 250       .Case("dereferenceable", MIToken::kw_dereferenceable)
 251       .Case("invariant", MIToken::kw_invariant)
 252       .Case("align", MIToken::kw_align)
 253       .Case("basealign", MIToken::kw_align)
 254       .Case("addrspace", MIToken::kw_addrspace)
 255       .Case("stack", MIToken::kw_stack)
 256       .Case("got", MIToken::kw_got)
 257       .Case("jump-table", MIToken::kw_jump_table)
 258       .Case("constant-pool", MIToken::kw_constant_pool)
 259       .Case("call-entry", MIToken::kw_call_entry)
 260       .Case("custom", MIToken::kw_custom)
 261       .Case("liveout", MIToken::kw_liveout)
 262       .Case("address-taken", MIToken::kw_address_taken)
 263       .Case("landing-pad", MIToken::kw_landing_pad)
 264       .Case("ehfunclet-entry", MIToken::kw_ehfunclet_entry)
 265       .Case("liveins", MIToken::kw_liveins)
 266       .Case("successors", MIToken::kw_successors)
 267       .Case("floatpred", MIToken::kw_floatpred)
 268       .Case("intpred", MIToken::kw_intpred)
 269       .Case("shufflemask", MIToken::kw_shufflemask)
 270       .Case("pre-instr-symbol", MIToken::kw_pre_instr_symbol)
 271       .Case("post-instr-symbol", MIToken::kw_post_instr_symbol)
 272       .Case("heap-alloc-marker", MIToken::kw_heap_alloc_marker)
 273       .Case("bbsections", MIToken::kw_bbsections)
 274       .Case("unknown-size", MIToken::kw_unknown_size)
 275       .Case("unknown-address", MIToken::kw_unknown_address)
 276       .Case("distinct", MIToken::kw_distinct)
 277       .Default(MIToken::Identifier);
 278 }
 279
 280 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
 281   if (!isalpha(C.peek()) && C.peek() != '_')
 282     return None;
 283   auto Range = C;
 284   while (isIdentifierChar(C.peek()))
 285     C.advance();
 286   auto Identifier = Range.upto(C);
 287   Token.reset(getIdentifierKind(Identifier), Identifier)
 288       .setStringValue(Identifier);
 289   return C;
 290 }
 291
 292 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
 293                                         ErrorCallbackType ErrorCallback) {
 294   bool IsReference = C.remaining().startswith("%bb.");
 295   if (!IsReference && !C.remaining().startswith("bb."))
 296     return None;
 297   auto Range = C;
 298   unsigned PrefixLength = IsReference ? 4 : 3;
 299   C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
 300   if (!isdigit(C.peek())) {
 301     Token.reset(MIToken::Error, C.remaining());
 302     ErrorCallback(C.location(), "expected a number after '%bb.'");
 303     return C;
 304   }
 305   auto NumberRange = C;
 306   while (isdigit(C.peek()))
 307     C.advance();
 308   StringRef Number = NumberRange.upto(C);
 309   unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
 310   // TODO: The format bb.<id>.<irname> is supported only when it's not a
 311   // reference. Once we deprecate the format where the irname shows up, we
 312   // should only lex forward if it is a reference.
 313   if (C.peek() == '.') {
 314     C.advance(); // Skip '.'
 315     ++StringOffset;
 316     while (isIdentifierChar(C.peek()))
 317       C.advance();
 318   }
 319   Token.reset(IsReference ? MIToken::MachineBasicBlock
 320                           : MIToken::MachineBasicBlockLabel,
 321               Range.upto(C))
 322       .setIntegerValue(APSInt(Number))
 323       .setStringValue(Range.upto(C).drop_front(StringOffset));
 324   return C;
 325 }
 326
 327 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
 328                             MIToken::TokenKind Kind) {
 329   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 330     return None;
 331   auto Range = C;
 332   C.advance(Rule.size());
 333   auto NumberRange = C;
 334   while (isdigit(C.peek()))
 335     C.advance();
 336   Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
 337   return C;
 338 }
 339
 340 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
 341                                    MIToken::TokenKind Kind) {
 342   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 343     return None;
 344   auto Range = C;
 345   C.advance(Rule.size());
 346   auto NumberRange = C;
 347   while (isdigit(C.peek()))
 348     C.advance();
 349   StringRef Number = NumberRange.upto(C);
 350   unsigned StringOffset = Rule.size() + Number.size();
 351   if (C.peek() == '.') {
 352     C.advance();
 353     ++StringOffset;
 354     while (isIdentifierChar(C.peek()))
 355       C.advance();
 356   }
 357   Token.reset(Kind, Range.upto(C))
 358       .setIntegerValue(APSInt(Number))
 359       .setStringValue(Range.upto(C).drop_front(StringOffset));
 360   return C;
 361 }
 362
 363 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
 364   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
 365 }
 366
 367 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
 368   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
 369 }
 370
 371 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
 372   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
 373 }
 374
 375 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
 376   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
 377 }
 378
 379 static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token,
 380                                        ErrorCallbackType ErrorCallback) {
 381   const StringRef Rule = "%subreg.";
 382   if (!C.remaining().startswith(Rule))
 383     return None;
 384   return lexName(C, Token, MIToken::SubRegisterIndex, Rule.size(),
 385                  ErrorCallback);
 386 }
 387
 388 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
 389                               ErrorCallbackType ErrorCallback) {
 390   const StringRef Rule = "%ir-block.";
 391   if (!C.remaining().startswith(Rule))
 392     return None;
 393   if (isdigit(C.peek(Rule.size())))
 394     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
 395   return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
 396 }
 397
 398 static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
 399                               ErrorCallbackType ErrorCallback) {
 400   const StringRef Rule = "%ir.";
 401   if (!C.remaining().startswith(Rule))
 402     return None;
 403   if (isdigit(C.peek(Rule.size())))
 404     return maybeLexIndex(C, Token, Rule, MIToken::IRValue);
 405   return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
 406 }
 407
 408 static Cursor maybeLexStringConstant(Cursor C, MIToken &Token,
 409                                      ErrorCallbackType ErrorCallback) {
 410   if (C.peek() != '"')
 411     return None;
 412   return lexName(C, Token, MIToken::StringConstant, /*PrefixLength=*/0,
 413                  ErrorCallback);
 414 }
 415
 416 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
 417   auto Range = C;
 418   C.advance(); // Skip '%'
 419   auto NumberRange = C;
 420   while (isdigit(C.peek()))
 421     C.advance();
 422   Token.reset(MIToken::VirtualRegister, Range.upto(C))
 423       .setIntegerValue(APSInt(NumberRange.upto(C)));
 424   return C;
 425 }
 426
 427 /// Returns true for a character allowed in a register name.
 428 static bool isRegisterChar(char C) {
 429   return isIdentifierChar(C) && C != '.';
 430 }
 431
 432 static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) {
 433   Cursor Range = C;
 434   C.advance(); // Skip '%'
 435   while (isRegisterChar(C.peek()))
 436     C.advance();
 437   Token.reset(MIToken::NamedVirtualRegister, Range.upto(C))
 438       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
 439   return C;
 440 }
 441
 442 static Cursor maybeLexRegister(Cursor C, MIToken &Token,
 443                                ErrorCallbackType ErrorCallback) {
 444   if (C.peek() != '%' && C.peek() != '$')
 445     return None;
 446
 447   if (C.peek() == '%') {
 448     if (isdigit(C.peek(1)))
 449       return lexVirtualRegister(C, Token);
 450
 451     if (isRegisterChar(C.peek(1)))
 452       return lexNamedVirtualRegister(C, Token);
 453
 454     return None;
 455   }
 456
 457   assert(C.peek() == '$');
 458   auto Range = C;
 459   C.advance(); // Skip '$'
 460   while (isRegisterChar(C.peek()))
 461     C.advance();
 462   Token.reset(MIToken::NamedRegister, Range.upto(C))
 463       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '$'
 464   return C;
 465 }
 466
 467 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
 468                                   ErrorCallbackType ErrorCallback) {
 469   if (C.peek() != '@')
 470     return None;
 471   if (!isdigit(C.peek(1)))
 472     return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
 473                    ErrorCallback);
 474   auto Range = C;
 475   C.advance(1); // Skip the '@'
 476   auto NumberRange = C;
 477   while (isdigit(C.peek()))
 478     C.advance();
 479   Token.reset(MIToken::GlobalValue, Range.upto(C))
 480       .setIntegerValue(APSInt(NumberRange.upto(C)));
 481   return C;
 482 }
 483
 484 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
 485                                      ErrorCallbackType ErrorCallback) {
 486   if (C.peek() != '&')
 487     return None;
 488   return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1,
 489                  ErrorCallback);
 490 }
 491
 492 static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token,
 493                                ErrorCallbackType ErrorCallback) {
 494   const StringRef Rule = "<mcsymbol ";
 495   if (!C.remaining().startswith(Rule))
 496     return None;
 497   auto Start = C;
 498   C.advance(Rule.size());
 499
 500   // Try a simple unquoted name.
 501   if (C.peek() != '"') {
 502     while (isIdentifierChar(C.peek()))
 503       C.advance();
 504     StringRef String = Start.upto(C).drop_front(Rule.size());
 505     if (C.peek() != '>') {
 506       ErrorCallback(C.location(),
 507                     "expected the '<mcsymbol ...' to be closed by a '>'");
 508       Token.reset(MIToken::Error, Start.remaining());
 509       return Start;
 510     }
 511     C.advance();
 512
 513     Token.reset(MIToken::MCSymbol, Start.upto(C)).setStringValue(String);
 514     return C;
 515   }
 516
 517   // Otherwise lex out a quoted name.
 518   Cursor R = lexStringConstant(C, ErrorCallback);
 519   if (!R) {
 520     ErrorCallback(C.location(),
 521                   "unable to parse quoted string from opening quote");
 522     Token.reset(MIToken::Error, Start.remaining());
 523     return Start;
 524   }
 525   StringRef String = Start.upto(R).drop_front(Rule.size());
 526   if (R.peek() != '>') {
 527     ErrorCallback(R.location(),
 528                   "expected the '<mcsymbol ...' to be closed by a '>'");
 529     Token.reset(MIToken::Error, Start.remaining());
 530     return Start;
 531   }
 532   R.advance();
 533
 534   Token.reset(MIToken::MCSymbol, Start.upto(R))
 535       .setOwnedStringValue(unescapeQuotedString(String));
 536   return R;
 537 }
 538
 539 static bool isValidHexFloatingPointPrefix(char C) {
 540   return C == 'H' || C == 'K' || C == 'L' || C == 'M' || C == 'R';
 541 }
 542
 543 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
 544   C.advance();
 545   // Skip over [0-9]*([eE][-+]?[0-9]+)?
 546   while (isdigit(C.peek()))
 547     C.advance();
 548   if ((C.peek() == 'e' || C.peek() == 'E') &&
 549       (isdigit(C.peek(1)) ||
 550        ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) {
 551     C.advance(2);
 552     while (isdigit(C.peek()))
 553       C.advance();
 554   }
 555   Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 556   return C;
 557 }
 558
 559 static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) {
 560   if (C.peek() != '0' || (C.peek(1) != 'x' && C.peek(1) != 'X'))
 561     return None;
 562   Cursor Range = C;
 563   C.advance(2);
 564   unsigned PrefLen = 2;
 565   if (isValidHexFloatingPointPrefix(C.peek())) {
 566     C.advance();
 567     PrefLen++;
 568   }
 569   while (isxdigit(C.peek()))
 570     C.advance();
 571   StringRef StrVal = Range.upto(C);
 572   if (StrVal.size() <= PrefLen)
 573     return None;
 574   if (PrefLen == 2)
 575     Token.reset(MIToken::HexLiteral, Range.upto(C));
 576   else // It must be 3, which means that there was a floating-point prefix.
 577     Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 578   return C;
 579 }
 580
 581 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
 582   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
 583     return None;
 584   auto Range = C;
 585   C.advance();
 586   while (isdigit(C.peek()))
 587     C.advance();
 588   if (C.peek() == '.')
 589     return lexFloatingPointLiteral(Range, C, Token);
 590   StringRef StrVal = Range.upto(C);
 591   Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal));
 592   return C;
 593 }
 594
 595 static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) {
 596   return StringSwitch<MIToken::TokenKind>(Identifier)
 597       .Case("!tbaa", MIToken::md_tbaa)
 598       .Case("!alias.scope", MIToken::md_alias_scope)
 599       .Case("!noalias", MIToken::md_noalias)
 600       .Case("!range", MIToken::md_range)
 601       .Case("!DIExpression", MIToken::md_diexpr)
 602       .Case("!DILocation", MIToken::md_dilocation)
 603       .Default(MIToken::Error);
 604 }
 605
 606 static Cursor maybeLexExclaim(Cursor C, MIToken &Token,
 607                               ErrorCallbackType ErrorCallback) {
 608   if (C.peek() != '!')
 609     return None;
 610   auto Range = C;
 611   C.advance(1);
 612   if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
 613     Token.reset(MIToken::exclaim, Range.upto(C));
 614     return C;
 615   }
 616   while (isIdentifierChar(C.peek()))
 617     C.advance();
 618   StringRef StrVal = Range.upto(C);
 619   Token.reset(getMetadataKeywordKind(StrVal), StrVal);
 620   if (Token.isError())
 621     ErrorCallback(Token.location(),
 622                   "use of unknown metadata keyword '" + StrVal + "'");
 623   return C;
 624 }
 625
 626 static MIToken::TokenKind symbolToken(char C) {
 627   switch (C) {
 628   case ',':
 629     return MIToken::comma;
 630   case '.':
 631     return MIToken::dot;
 632   case '=':
 633     return MIToken::equal;
 634   case ':':
 635     return MIToken::colon;
 636   case '(':
 637     return MIToken::lparen;
 638   case ')':
 639     return MIToken::rparen;
 640   case '{':
 641     return MIToken::lbrace;
 642   case '}':
 643     return MIToken::rbrace;
 644   case '+':
 645     return MIToken::plus;
 646   case '-':
 647     return MIToken::minus;
 648   case '<':
 649     return MIToken::less;
 650   case '>':
 651     return MIToken::greater;
 652   default:
 653     return MIToken::Error;
 654   }
 655 }
 656
 657 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
 658   MIToken::TokenKind Kind;
 659   unsigned Length = 1;
 660   if (C.peek() == ':' && C.peek(1) == ':') {
 661     Kind = MIToken::coloncolon;
 662     Length = 2;
 663   } else
 664     Kind = symbolToken(C.peek());
 665   if (Kind == MIToken::Error)
 666     return None;
 667   auto Range = C;
 668   C.advance(Length);
 669   Token.reset(Kind, Range.upto(C));
 670   return C;
 671 }
 672
 673 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
 674   if (!isNewlineChar(C.peek()))
 675     return None;
 676   auto Range = C;
 677   C.advance();
 678   Token.reset(MIToken::Newline, Range.upto(C));
 679   return C;
 680 }
 681
 682 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
 683                                      ErrorCallbackType ErrorCallback) {
 684   if (C.peek() != '`')
 685     return None;
 686   auto Range = C;
 687   C.advance();
 688   auto StrRange = C;
 689   while (C.peek() != '`') {
 690     if (C.isEOF() || isNewlineChar(C.peek())) {
 691       ErrorCallback(
 692           C.location(),
 693           "end of machine instruction reached before the closing '`'");
 694       Token.reset(MIToken::Error, Range.remaining());
 695       return C;
 696     }
 697     C.advance();
 698   }
 699   StringRef Value = StrRange.upto(C);
 700   C.advance();
 701   Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value);
 702   return C;
 703 }
 704
 705 StringRef llvm::lexMIToken(StringRef Source, MIToken &Token,
 706                            ErrorCallbackType ErrorCallback) {
 707   auto C = skipComment(skipWhitespace(Cursor(Source)));
 708   if (C.isEOF()) {
 709     Token.reset(MIToken::Eof, C.remaining());
 710     return C.remaining();
 711   }
 712
 713   C = skipMachineOperandComment(C);
 714
 715   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
 716     return R.remaining();
 717   if (Cursor R = maybeLexIdentifier(C, Token))
 718     return R.remaining();
 719   if (Cursor R = maybeLexJumpTableIndex(C, Token))
 720     return R.remaining();
 721   if (Cursor R = maybeLexStackObject(C, Token))
 722     return R.remaining();
 723   if (Cursor R = maybeLexFixedStackObject(C, Token))
 724     return R.remaining();
 725   if (Cursor R = maybeLexConstantPoolItem(C, Token))
 726     return R.remaining();
 727   if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback))
 728     return R.remaining();
 729   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
 730     return R.remaining();
 731   if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
 732     return R.remaining();
 733   if (Cursor R = maybeLexRegister(C, Token, ErrorCallback))
 734     return R.remaining();
 735   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
 736     return R.remaining();
 737   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
 738     return R.remaining();
 739   if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback))
 740     return R.remaining();
 741   if (Cursor R = maybeLexHexadecimalLiteral(C, Token))
 742     return R.remaining();
 743   if (Cursor R = maybeLexNumericalLiteral(C, Token))
 744     return R.remaining();
 745   if (Cursor R = maybeLexExclaim(C, Token, ErrorCallback))
 746     return R.remaining();
 747   if (Cursor R = maybeLexSymbol(C, Token))
 748     return R.remaining();
 749   if (Cursor R = maybeLexNewline(C, Token))
 750     return R.remaining();
 751   if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
 752     return R.remaining();
 753   if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback))
 754     return R.remaining();
 755
 756   Token.reset(MIToken::Error, C.remaining());
 757   ErrorCallback(C.location(),
 758                 Twine("unexpected character '") + Twine(C.peek()) + "'");
 759   return C.remaining();
 760 }