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/StringExtras.h" 15 #include "llvm/ADT/StringSwitch.h" 16 #include "llvm/ADT/Twine.h" 17 #include <cassert> 18 #include <cctype> 19 #include <string> 20 21 using namespace llvm; 22 23 namespace { 24 25 using ErrorCallbackType = 26 function_ref<void(StringRef::iterator Loc, const Twine &)>; 27 28 /// This class provides a way to iterate and get characters from the source 29 /// string. 30 class Cursor { 31 const char *Ptr = nullptr; 32 const char *End = nullptr; 33 34 public: 35 Cursor(std::nullopt_t) {} 36 37 explicit Cursor(StringRef Str) { 38 Ptr = Str.data(); 39 End = Ptr + Str.size(); 40 } 41 42 bool isEOF() const { return Ptr == End; } 43 44 char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; } 45 46 void advance(unsigned I = 1) { Ptr += I; } 47 48 StringRef remaining() const { return StringRef(Ptr, End - Ptr); } 49 50 StringRef upto(Cursor C) const { 51 assert(C.Ptr >= Ptr && C.Ptr <= End); 52 return StringRef(Ptr, C.Ptr - Ptr); 53 } 54 55 StringRef::iterator location() const { return Ptr; } 56 57 operator bool() const { return Ptr != nullptr; } 58 }; 59 60 } // end anonymous namespace 61 62 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) { 63 this->Kind = Kind; 64 this->Range = Range; 65 return *this; 66 } 67 68 MIToken &MIToken::setStringValue(StringRef StrVal) { 69 StringValue = StrVal; 70 return *this; 71 } 72 73 MIToken &MIToken::setOwnedStringValue(std::string StrVal) { 74 StringValueStorage = std::move(StrVal); 75 StringValue = StringValueStorage; 76 return *this; 77 } 78 79 MIToken &MIToken::setIntegerValue(APSInt IntVal) { 80 this->IntVal = std::move(IntVal); 81 return *this; 82 } 83 84 /// Skip the leading whitespace characters and return the updated cursor. 85 static Cursor skipWhitespace(Cursor C) { 86 while (isblank(C.peek())) 87 C.advance(); 88 return C; 89 } 90 91 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; } 92 93 /// Skip a line comment and return the updated cursor. 94 static Cursor skipComment(Cursor C) { 95 if (C.peek() != ';') 96 return C; 97 while (!isNewlineChar(C.peek()) && !C.isEOF()) 98 C.advance(); 99 return C; 100 } 101 102 /// Machine operands can have comments, enclosed between /* and */. 103 /// This eats up all tokens, including /* and */. 104 static Cursor skipMachineOperandComment(Cursor C) { 105 if (C.peek() != '/' || C.peek(1) != '*') 106 return C; 107 108 while (C.peek() != '*' || C.peek(1) != '/') 109 C.advance(); 110 111 C.advance(); 112 C.advance(); 113 return C; 114 } 115 116 /// Return true if the given character satisfies the following regular 117 /// expression: [-a-zA-Z$._0-9] 118 static bool isIdentifierChar(char C) { 119 return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' || 120 C == '$'; 121 } 122 123 /// Unescapes the given string value. 124 /// 125 /// Expects the string value to be quoted. 126 static std::string unescapeQuotedString(StringRef Value) { 127 assert(Value.front() == '"' && Value.back() == '"'); 128 Cursor C = Cursor(Value.substr(1, Value.size() - 2)); 129 130 std::string Str; 131 Str.reserve(C.remaining().size()); 132 while (!C.isEOF()) { 133 char Char = C.peek(); 134 if (Char == '\\') { 135 if (C.peek(1) == '\\') { 136 // Two '\' become one 137 Str += '\\'; 138 C.advance(2); 139 continue; 140 } 141 if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) { 142 Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2)); 143 C.advance(3); 144 continue; 145 } 146 } 147 Str += Char; 148 C.advance(); 149 } 150 return Str; 151 } 152 153 /// Lex a string constant using the following regular expression: \"[^\"]*\" 154 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) { 155 assert(C.peek() == '"'); 156 for (C.advance(); C.peek() != '"'; C.advance()) { 157 if (C.isEOF() || isNewlineChar(C.peek())) { 158 ErrorCallback( 159 C.location(), 160 "end of machine instruction reached before the closing '\"'"); 161 return std::nullopt; 162 } 163 } 164 C.advance(); 165 return C; 166 } 167 168 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type, 169 unsigned PrefixLength, ErrorCallbackType ErrorCallback) { 170 auto Range = C; 171 C.advance(PrefixLength); 172 if (C.peek() == '"') { 173 if (Cursor R = lexStringConstant(C, ErrorCallback)) { 174 StringRef String = Range.upto(R); 175 Token.reset(Type, String) 176 .setOwnedStringValue( 177 unescapeQuotedString(String.drop_front(PrefixLength))); 178 return R; 179 } 180 Token.reset(MIToken::Error, Range.remaining()); 181 return Range; 182 } 183 while (isIdentifierChar(C.peek())) 184 C.advance(); 185 Token.reset(Type, Range.upto(C)) 186 .setStringValue(Range.upto(C).drop_front(PrefixLength)); 187 return C; 188 } 189 190 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) { 191 return StringSwitch<MIToken::TokenKind>(Identifier) 192 .Case("_", MIToken::underscore) 193 .Case("implicit", MIToken::kw_implicit) 194 .Case("implicit-def", MIToken::kw_implicit_define) 195 .Case("def", MIToken::kw_def) 196 .Case("dead", MIToken::kw_dead) 197 .Case("killed", MIToken::kw_killed) 198 .Case("undef", MIToken::kw_undef) 199 .Case("internal", MIToken::kw_internal) 200 .Case("early-clobber", MIToken::kw_early_clobber) 201 .Case("debug-use", MIToken::kw_debug_use) 202 .Case("renamable", MIToken::kw_renamable) 203 .Case("tied-def", MIToken::kw_tied_def) 204 .Case("frame-setup", MIToken::kw_frame_setup) 205 .Case("frame-destroy", MIToken::kw_frame_destroy) 206 .Case("nnan", MIToken::kw_nnan) 207 .Case("ninf", MIToken::kw_ninf) 208 .Case("nsz", MIToken::kw_nsz) 209 .Case("arcp", MIToken::kw_arcp) 210 .Case("contract", MIToken::kw_contract) 211 .Case("afn", MIToken::kw_afn) 212 .Case("reassoc", MIToken::kw_reassoc) 213 .Case("nuw", MIToken::kw_nuw) 214 .Case("nsw", MIToken::kw_nsw) 215 .Case("nusw", MIToken::kw_nusw) 216 .Case("exact", MIToken::kw_exact) 217 .Case("nneg", MIToken::kw_nneg) 218 .Case("disjoint", MIToken::kw_disjoint) 219 .Case("samesign", MIToken::kw_samesign) 220 .Case("nofpexcept", MIToken::kw_nofpexcept) 221 .Case("unpredictable", MIToken::kw_unpredictable) 222 .Case("debug-location", MIToken::kw_debug_location) 223 .Case("debug-instr-number", MIToken::kw_debug_instr_number) 224 .Case("dbg-instr-ref", MIToken::kw_dbg_instr_ref) 225 .Case("same_value", MIToken::kw_cfi_same_value) 226 .Case("offset", MIToken::kw_cfi_offset) 227 .Case("rel_offset", MIToken::kw_cfi_rel_offset) 228 .Case("def_cfa_register", MIToken::kw_cfi_def_cfa_register) 229 .Case("def_cfa_offset", MIToken::kw_cfi_def_cfa_offset) 230 .Case("adjust_cfa_offset", MIToken::kw_cfi_adjust_cfa_offset) 231 .Case("escape", MIToken::kw_cfi_escape) 232 .Case("def_cfa", MIToken::kw_cfi_def_cfa) 233 .Case("llvm_def_aspace_cfa", MIToken::kw_cfi_llvm_def_aspace_cfa) 234 .Case("remember_state", MIToken::kw_cfi_remember_state) 235 .Case("restore", MIToken::kw_cfi_restore) 236 .Case("restore_state", MIToken::kw_cfi_restore_state) 237 .Case("undefined", MIToken::kw_cfi_undefined) 238 .Case("register", MIToken::kw_cfi_register) 239 .Case("window_save", MIToken::kw_cfi_window_save) 240 .Case("negate_ra_sign_state", 241 MIToken::kw_cfi_aarch64_negate_ra_sign_state) 242 .Case("negate_ra_sign_state_with_pc", 243 MIToken::kw_cfi_aarch64_negate_ra_sign_state_with_pc) 244 .Case("blockaddress", MIToken::kw_blockaddress) 245 .Case("intrinsic", MIToken::kw_intrinsic) 246 .Case("target-index", MIToken::kw_target_index) 247 .Case("half", MIToken::kw_half) 248 .Case("bfloat", MIToken::kw_bfloat) 249 .Case("float", MIToken::kw_float) 250 .Case("double", MIToken::kw_double) 251 .Case("x86_fp80", MIToken::kw_x86_fp80) 252 .Case("fp128", MIToken::kw_fp128) 253 .Case("ppc_fp128", MIToken::kw_ppc_fp128) 254 .Case("target-flags", MIToken::kw_target_flags) 255 .Case("volatile", MIToken::kw_volatile) 256 .Case("non-temporal", MIToken::kw_non_temporal) 257 .Case("dereferenceable", MIToken::kw_dereferenceable) 258 .Case("invariant", MIToken::kw_invariant) 259 .Case("align", MIToken::kw_align) 260 .Case("basealign", MIToken::kw_basealign) 261 .Case("addrspace", MIToken::kw_addrspace) 262 .Case("stack", MIToken::kw_stack) 263 .Case("got", MIToken::kw_got) 264 .Case("jump-table", MIToken::kw_jump_table) 265 .Case("constant-pool", MIToken::kw_constant_pool) 266 .Case("call-entry", MIToken::kw_call_entry) 267 .Case("custom", MIToken::kw_custom) 268 .Case("liveout", MIToken::kw_liveout) 269 .Case("landing-pad", MIToken::kw_landing_pad) 270 .Case("inlineasm-br-indirect-target", 271 MIToken::kw_inlineasm_br_indirect_target) 272 .Case("ehfunclet-entry", MIToken::kw_ehfunclet_entry) 273 .Case("liveins", MIToken::kw_liveins) 274 .Case("successors", MIToken::kw_successors) 275 .Case("floatpred", MIToken::kw_floatpred) 276 .Case("intpred", MIToken::kw_intpred) 277 .Case("shufflemask", MIToken::kw_shufflemask) 278 .Case("pre-instr-symbol", MIToken::kw_pre_instr_symbol) 279 .Case("post-instr-symbol", MIToken::kw_post_instr_symbol) 280 .Case("heap-alloc-marker", MIToken::kw_heap_alloc_marker) 281 .Case("pcsections", MIToken::kw_pcsections) 282 .Case("cfi-type", MIToken::kw_cfi_type) 283 .Case("bbsections", MIToken::kw_bbsections) 284 .Case("bb_id", MIToken::kw_bb_id) 285 .Case("unknown-size", MIToken::kw_unknown_size) 286 .Case("unknown-address", MIToken::kw_unknown_address) 287 .Case("distinct", MIToken::kw_distinct) 288 .Case("ir-block-address-taken", MIToken::kw_ir_block_address_taken) 289 .Case("machine-block-address-taken", 290 MIToken::kw_machine_block_address_taken) 291 .Case("call-frame-size", MIToken::kw_call_frame_size) 292 .Case("noconvergent", MIToken::kw_noconvergent) 293 .Default(MIToken::Identifier); 294 } 295 296 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) { 297 if (!isalpha(C.peek()) && C.peek() != '_') 298 return std::nullopt; 299 auto Range = C; 300 while (isIdentifierChar(C.peek())) 301 C.advance(); 302 auto Identifier = Range.upto(C); 303 Token.reset(getIdentifierKind(Identifier), Identifier) 304 .setStringValue(Identifier); 305 return C; 306 } 307 308 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token, 309 ErrorCallbackType ErrorCallback) { 310 bool IsReference = C.remaining().starts_with("%bb."); 311 if (!IsReference && !C.remaining().starts_with("bb.")) 312 return std::nullopt; 313 auto Range = C; 314 unsigned PrefixLength = IsReference ? 4 : 3; 315 C.advance(PrefixLength); // Skip '%bb.' or 'bb.' 316 if (!isdigit(C.peek())) { 317 Token.reset(MIToken::Error, C.remaining()); 318 ErrorCallback(C.location(), "expected a number after '%bb.'"); 319 return C; 320 } 321 auto NumberRange = C; 322 while (isdigit(C.peek())) 323 C.advance(); 324 StringRef Number = NumberRange.upto(C); 325 unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>' 326 // TODO: The format bb.<id>.<irname> is supported only when it's not a 327 // reference. Once we deprecate the format where the irname shows up, we 328 // should only lex forward if it is a reference. 329 if (C.peek() == '.') { 330 C.advance(); // Skip '.' 331 ++StringOffset; 332 while (isIdentifierChar(C.peek())) 333 C.advance(); 334 } 335 Token.reset(IsReference ? MIToken::MachineBasicBlock 336 : MIToken::MachineBasicBlockLabel, 337 Range.upto(C)) 338 .setIntegerValue(APSInt(Number)) 339 .setStringValue(Range.upto(C).drop_front(StringOffset)); 340 return C; 341 } 342 343 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule, 344 MIToken::TokenKind Kind) { 345 if (!C.remaining().starts_with(Rule) || !isdigit(C.peek(Rule.size()))) 346 return std::nullopt; 347 auto Range = C; 348 C.advance(Rule.size()); 349 auto NumberRange = C; 350 while (isdigit(C.peek())) 351 C.advance(); 352 Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C))); 353 return C; 354 } 355 356 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule, 357 MIToken::TokenKind Kind) { 358 if (!C.remaining().starts_with(Rule) || !isdigit(C.peek(Rule.size()))) 359 return std::nullopt; 360 auto Range = C; 361 C.advance(Rule.size()); 362 auto NumberRange = C; 363 while (isdigit(C.peek())) 364 C.advance(); 365 StringRef Number = NumberRange.upto(C); 366 unsigned StringOffset = Rule.size() + Number.size(); 367 if (C.peek() == '.') { 368 C.advance(); 369 ++StringOffset; 370 while (isIdentifierChar(C.peek())) 371 C.advance(); 372 } 373 Token.reset(Kind, Range.upto(C)) 374 .setIntegerValue(APSInt(Number)) 375 .setStringValue(Range.upto(C).drop_front(StringOffset)); 376 return C; 377 } 378 379 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) { 380 return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex); 381 } 382 383 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) { 384 return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject); 385 } 386 387 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) { 388 return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject); 389 } 390 391 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) { 392 return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem); 393 } 394 395 static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token, 396 ErrorCallbackType ErrorCallback) { 397 const StringRef Rule = "%subreg."; 398 if (!C.remaining().starts_with(Rule)) 399 return std::nullopt; 400 return lexName(C, Token, MIToken::SubRegisterIndex, Rule.size(), 401 ErrorCallback); 402 } 403 404 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token, 405 ErrorCallbackType ErrorCallback) { 406 const StringRef Rule = "%ir-block."; 407 if (!C.remaining().starts_with(Rule)) 408 return std::nullopt; 409 if (isdigit(C.peek(Rule.size()))) 410 return maybeLexIndex(C, Token, Rule, MIToken::IRBlock); 411 return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback); 412 } 413 414 static Cursor maybeLexIRValue(Cursor C, MIToken &Token, 415 ErrorCallbackType ErrorCallback) { 416 const StringRef Rule = "%ir."; 417 if (!C.remaining().starts_with(Rule)) 418 return std::nullopt; 419 if (isdigit(C.peek(Rule.size()))) 420 return maybeLexIndex(C, Token, Rule, MIToken::IRValue); 421 return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback); 422 } 423 424 static Cursor maybeLexStringConstant(Cursor C, MIToken &Token, 425 ErrorCallbackType ErrorCallback) { 426 if (C.peek() != '"') 427 return std::nullopt; 428 return lexName(C, Token, MIToken::StringConstant, /*PrefixLength=*/0, 429 ErrorCallback); 430 } 431 432 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) { 433 auto Range = C; 434 C.advance(); // Skip '%' 435 auto NumberRange = C; 436 while (isdigit(C.peek())) 437 C.advance(); 438 Token.reset(MIToken::VirtualRegister, Range.upto(C)) 439 .setIntegerValue(APSInt(NumberRange.upto(C))); 440 return C; 441 } 442 443 /// Returns true for a character allowed in a register name. 444 static bool isRegisterChar(char C) { 445 return isIdentifierChar(C) && C != '.'; 446 } 447 448 static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) { 449 Cursor Range = C; 450 C.advance(); // Skip '%' 451 while (isRegisterChar(C.peek())) 452 C.advance(); 453 Token.reset(MIToken::NamedVirtualRegister, Range.upto(C)) 454 .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%' 455 return C; 456 } 457 458 static Cursor maybeLexRegister(Cursor C, MIToken &Token, 459 ErrorCallbackType ErrorCallback) { 460 if (C.peek() != '%' && C.peek() != '$') 461 return std::nullopt; 462 463 if (C.peek() == '%') { 464 if (isdigit(C.peek(1))) 465 return lexVirtualRegister(C, Token); 466 467 if (isRegisterChar(C.peek(1))) 468 return lexNamedVirtualRegister(C, Token); 469 470 return std::nullopt; 471 } 472 473 assert(C.peek() == '$'); 474 auto Range = C; 475 C.advance(); // Skip '$' 476 while (isRegisterChar(C.peek())) 477 C.advance(); 478 Token.reset(MIToken::NamedRegister, Range.upto(C)) 479 .setStringValue(Range.upto(C).drop_front(1)); // Drop the '$' 480 return C; 481 } 482 483 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token, 484 ErrorCallbackType ErrorCallback) { 485 if (C.peek() != '@') 486 return std::nullopt; 487 if (!isdigit(C.peek(1))) 488 return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1, 489 ErrorCallback); 490 auto Range = C; 491 C.advance(1); // Skip the '@' 492 auto NumberRange = C; 493 while (isdigit(C.peek())) 494 C.advance(); 495 Token.reset(MIToken::GlobalValue, Range.upto(C)) 496 .setIntegerValue(APSInt(NumberRange.upto(C))); 497 return C; 498 } 499 500 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token, 501 ErrorCallbackType ErrorCallback) { 502 if (C.peek() != '&') 503 return std::nullopt; 504 return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1, 505 ErrorCallback); 506 } 507 508 static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token, 509 ErrorCallbackType ErrorCallback) { 510 const StringRef Rule = "<mcsymbol "; 511 if (!C.remaining().starts_with(Rule)) 512 return std::nullopt; 513 auto Start = C; 514 C.advance(Rule.size()); 515 516 // Try a simple unquoted name. 517 if (C.peek() != '"') { 518 while (isIdentifierChar(C.peek())) 519 C.advance(); 520 StringRef String = Start.upto(C).drop_front(Rule.size()); 521 if (C.peek() != '>') { 522 ErrorCallback(C.location(), 523 "expected the '<mcsymbol ...' to be closed by a '>'"); 524 Token.reset(MIToken::Error, Start.remaining()); 525 return Start; 526 } 527 C.advance(); 528 529 Token.reset(MIToken::MCSymbol, Start.upto(C)).setStringValue(String); 530 return C; 531 } 532 533 // Otherwise lex out a quoted name. 534 Cursor R = lexStringConstant(C, ErrorCallback); 535 if (!R) { 536 ErrorCallback(C.location(), 537 "unable to parse quoted string from opening quote"); 538 Token.reset(MIToken::Error, Start.remaining()); 539 return Start; 540 } 541 StringRef String = Start.upto(R).drop_front(Rule.size()); 542 if (R.peek() != '>') { 543 ErrorCallback(R.location(), 544 "expected the '<mcsymbol ...' to be closed by a '>'"); 545 Token.reset(MIToken::Error, Start.remaining()); 546 return Start; 547 } 548 R.advance(); 549 550 Token.reset(MIToken::MCSymbol, Start.upto(R)) 551 .setOwnedStringValue(unescapeQuotedString(String)); 552 return R; 553 } 554 555 static bool isValidHexFloatingPointPrefix(char C) { 556 return C == 'H' || C == 'K' || C == 'L' || C == 'M' || C == 'R'; 557 } 558 559 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) { 560 C.advance(); 561 // Skip over [0-9]*([eE][-+]?[0-9]+)? 562 while (isdigit(C.peek())) 563 C.advance(); 564 if ((C.peek() == 'e' || C.peek() == 'E') && 565 (isdigit(C.peek(1)) || 566 ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) { 567 C.advance(2); 568 while (isdigit(C.peek())) 569 C.advance(); 570 } 571 Token.reset(MIToken::FloatingPointLiteral, Range.upto(C)); 572 return C; 573 } 574 575 static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) { 576 if (C.peek() != '0' || (C.peek(1) != 'x' && C.peek(1) != 'X')) 577 return std::nullopt; 578 Cursor Range = C; 579 C.advance(2); 580 unsigned PrefLen = 2; 581 if (isValidHexFloatingPointPrefix(C.peek())) { 582 C.advance(); 583 PrefLen++; 584 } 585 while (isxdigit(C.peek())) 586 C.advance(); 587 StringRef StrVal = Range.upto(C); 588 if (StrVal.size() <= PrefLen) 589 return std::nullopt; 590 if (PrefLen == 2) 591 Token.reset(MIToken::HexLiteral, Range.upto(C)); 592 else // It must be 3, which means that there was a floating-point prefix. 593 Token.reset(MIToken::FloatingPointLiteral, Range.upto(C)); 594 return C; 595 } 596 597 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) { 598 if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1)))) 599 return std::nullopt; 600 auto Range = C; 601 C.advance(); 602 while (isdigit(C.peek())) 603 C.advance(); 604 if (C.peek() == '.') 605 return lexFloatingPointLiteral(Range, C, Token); 606 StringRef StrVal = Range.upto(C); 607 Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal)); 608 return C; 609 } 610 611 static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) { 612 return StringSwitch<MIToken::TokenKind>(Identifier) 613 .Case("!tbaa", MIToken::md_tbaa) 614 .Case("!alias.scope", MIToken::md_alias_scope) 615 .Case("!noalias", MIToken::md_noalias) 616 .Case("!range", MIToken::md_range) 617 .Case("!DIExpression", MIToken::md_diexpr) 618 .Case("!DILocation", MIToken::md_dilocation) 619 .Default(MIToken::Error); 620 } 621 622 static Cursor maybeLexExclaim(Cursor C, MIToken &Token, 623 ErrorCallbackType ErrorCallback) { 624 if (C.peek() != '!') 625 return std::nullopt; 626 auto Range = C; 627 C.advance(1); 628 if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) { 629 Token.reset(MIToken::exclaim, Range.upto(C)); 630 return C; 631 } 632 while (isIdentifierChar(C.peek())) 633 C.advance(); 634 StringRef StrVal = Range.upto(C); 635 Token.reset(getMetadataKeywordKind(StrVal), StrVal); 636 if (Token.isError()) 637 ErrorCallback(Token.location(), 638 "use of unknown metadata keyword '" + StrVal + "'"); 639 return C; 640 } 641 642 static MIToken::TokenKind symbolToken(char C) { 643 switch (C) { 644 case ',': 645 return MIToken::comma; 646 case '.': 647 return MIToken::dot; 648 case '=': 649 return MIToken::equal; 650 case ':': 651 return MIToken::colon; 652 case '(': 653 return MIToken::lparen; 654 case ')': 655 return MIToken::rparen; 656 case '{': 657 return MIToken::lbrace; 658 case '}': 659 return MIToken::rbrace; 660 case '+': 661 return MIToken::plus; 662 case '-': 663 return MIToken::minus; 664 case '<': 665 return MIToken::less; 666 case '>': 667 return MIToken::greater; 668 default: 669 return MIToken::Error; 670 } 671 } 672 673 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) { 674 MIToken::TokenKind Kind; 675 unsigned Length = 1; 676 if (C.peek() == ':' && C.peek(1) == ':') { 677 Kind = MIToken::coloncolon; 678 Length = 2; 679 } else 680 Kind = symbolToken(C.peek()); 681 if (Kind == MIToken::Error) 682 return std::nullopt; 683 auto Range = C; 684 C.advance(Length); 685 Token.reset(Kind, Range.upto(C)); 686 return C; 687 } 688 689 static Cursor maybeLexNewline(Cursor C, MIToken &Token) { 690 if (!isNewlineChar(C.peek())) 691 return std::nullopt; 692 auto Range = C; 693 C.advance(); 694 Token.reset(MIToken::Newline, Range.upto(C)); 695 return C; 696 } 697 698 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token, 699 ErrorCallbackType ErrorCallback) { 700 if (C.peek() != '`') 701 return std::nullopt; 702 auto Range = C; 703 C.advance(); 704 auto StrRange = C; 705 while (C.peek() != '`') { 706 if (C.isEOF() || isNewlineChar(C.peek())) { 707 ErrorCallback( 708 C.location(), 709 "end of machine instruction reached before the closing '`'"); 710 Token.reset(MIToken::Error, Range.remaining()); 711 return C; 712 } 713 C.advance(); 714 } 715 StringRef Value = StrRange.upto(C); 716 C.advance(); 717 Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value); 718 return C; 719 } 720 721 StringRef llvm::lexMIToken(StringRef Source, MIToken &Token, 722 ErrorCallbackType ErrorCallback) { 723 auto C = skipComment(skipWhitespace(Cursor(Source))); 724 if (C.isEOF()) { 725 Token.reset(MIToken::Eof, C.remaining()); 726 return C.remaining(); 727 } 728 729 C = skipWhitespace(skipMachineOperandComment(C)); 730 731 if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback)) 732 return R.remaining(); 733 if (Cursor R = maybeLexIdentifier(C, Token)) 734 return R.remaining(); 735 if (Cursor R = maybeLexJumpTableIndex(C, Token)) 736 return R.remaining(); 737 if (Cursor R = maybeLexStackObject(C, Token)) 738 return R.remaining(); 739 if (Cursor R = maybeLexFixedStackObject(C, Token)) 740 return R.remaining(); 741 if (Cursor R = maybeLexConstantPoolItem(C, Token)) 742 return R.remaining(); 743 if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback)) 744 return R.remaining(); 745 if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback)) 746 return R.remaining(); 747 if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback)) 748 return R.remaining(); 749 if (Cursor R = maybeLexRegister(C, Token, ErrorCallback)) 750 return R.remaining(); 751 if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback)) 752 return R.remaining(); 753 if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback)) 754 return R.remaining(); 755 if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback)) 756 return R.remaining(); 757 if (Cursor R = maybeLexHexadecimalLiteral(C, Token)) 758 return R.remaining(); 759 if (Cursor R = maybeLexNumericalLiteral(C, Token)) 760 return R.remaining(); 761 if (Cursor R = maybeLexExclaim(C, Token, ErrorCallback)) 762 return R.remaining(); 763 if (Cursor R = maybeLexSymbol(C, Token)) 764 return R.remaining(); 765 if (Cursor R = maybeLexNewline(C, Token)) 766 return R.remaining(); 767 if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback)) 768 return R.remaining(); 769 if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback)) 770 return R.remaining(); 771 772 Token.reset(MIToken::Error, C.remaining()); 773 ErrorCallback(C.location(), 774 Twine("unexpected character '") + Twine(C.peek()) + "'"); 775 return C.remaining(); 776 } 777