1 //===- AsmParser.cpp - Parser for Assembly Files --------------------------===// 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 class implements a parser for assembly files similar to gas syntax. 10 // 11 //===----------------------------------------------------------------------===// 12 13 #include "llvm/ADT/APFloat.h" 14 #include "llvm/ADT/APInt.h" 15 #include "llvm/ADT/ArrayRef.h" 16 #include "llvm/ADT/None.h" 17 #include "llvm/ADT/STLExtras.h" 18 #include "llvm/ADT/SmallSet.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/SmallVector.h" 21 #include "llvm/ADT/StringExtras.h" 22 #include "llvm/ADT/StringMap.h" 23 #include "llvm/ADT/StringRef.h" 24 #include "llvm/ADT/Twine.h" 25 #include "llvm/BinaryFormat/Dwarf.h" 26 #include "llvm/DebugInfo/CodeView/SymbolRecord.h" 27 #include "llvm/MC/MCAsmInfo.h" 28 #include "llvm/MC/MCCodeView.h" 29 #include "llvm/MC/MCContext.h" 30 #include "llvm/MC/MCDirectives.h" 31 #include "llvm/MC/MCDwarf.h" 32 #include "llvm/MC/MCExpr.h" 33 #include "llvm/MC/MCInstPrinter.h" 34 #include "llvm/MC/MCInstrDesc.h" 35 #include "llvm/MC/MCInstrInfo.h" 36 #include "llvm/MC/MCObjectFileInfo.h" 37 #include "llvm/MC/MCParser/AsmCond.h" 38 #include "llvm/MC/MCParser/AsmLexer.h" 39 #include "llvm/MC/MCParser/MCAsmLexer.h" 40 #include "llvm/MC/MCParser/MCAsmParser.h" 41 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 42 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 43 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 44 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 45 #include "llvm/MC/MCRegisterInfo.h" 46 #include "llvm/MC/MCSection.h" 47 #include "llvm/MC/MCStreamer.h" 48 #include "llvm/MC/MCSymbol.h" 49 #include "llvm/MC/MCTargetOptions.h" 50 #include "llvm/MC/MCValue.h" 51 #include "llvm/Support/Casting.h" 52 #include "llvm/Support/CommandLine.h" 53 #include "llvm/Support/ErrorHandling.h" 54 #include "llvm/Support/MD5.h" 55 #include "llvm/Support/MathExtras.h" 56 #include "llvm/Support/MemoryBuffer.h" 57 #include "llvm/Support/SMLoc.h" 58 #include "llvm/Support/SourceMgr.h" 59 #include "llvm/Support/raw_ostream.h" 60 #include <algorithm> 61 #include <cassert> 62 #include <cctype> 63 #include <climits> 64 #include <cstddef> 65 #include <cstdint> 66 #include <deque> 67 #include <memory> 68 #include <sstream> 69 #include <string> 70 #include <tuple> 71 #include <utility> 72 #include <vector> 73 74 using namespace llvm; 75 76 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 77 78 extern cl::opt<unsigned> AsmMacroMaxNestingDepth; 79 80 namespace { 81 82 /// Helper types for tracking macro definitions. 83 typedef std::vector<AsmToken> MCAsmMacroArgument; 84 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 85 86 /// Helper class for storing information about an active macro 87 /// instantiation. 88 struct MacroInstantiation { 89 /// The location of the instantiation. 90 SMLoc InstantiationLoc; 91 92 /// The buffer where parsing should resume upon instantiation completion. 93 unsigned ExitBuffer; 94 95 /// The location where parsing should resume upon instantiation completion. 96 SMLoc ExitLoc; 97 98 /// The depth of TheCondStack at the start of the instantiation. 99 size_t CondStackDepth; 100 }; 101 102 struct ParseStatementInfo { 103 /// The parsed operands from the last parsed statement. 104 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 105 106 /// The opcode from the last parsed instruction. 107 unsigned Opcode = ~0U; 108 109 /// Was there an error parsing the inline assembly? 110 bool ParseError = false; 111 112 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 113 114 ParseStatementInfo() = delete; 115 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 116 : AsmRewrites(rewrites) {} 117 }; 118 119 /// The concrete assembly parser instance. 120 class AsmParser : public MCAsmParser { 121 private: 122 AsmLexer Lexer; 123 MCContext &Ctx; 124 MCStreamer &Out; 125 const MCAsmInfo &MAI; 126 SourceMgr &SrcMgr; 127 SourceMgr::DiagHandlerTy SavedDiagHandler; 128 void *SavedDiagContext; 129 std::unique_ptr<MCAsmParserExtension> PlatformParser; 130 SMLoc StartTokLoc; 131 132 /// This is the current buffer index we're lexing from as managed by the 133 /// SourceMgr object. 134 unsigned CurBuffer; 135 136 AsmCond TheCondState; 137 std::vector<AsmCond> TheCondStack; 138 139 /// maps directive names to handler methods in parser 140 /// extensions. Extensions register themselves in this map by calling 141 /// addDirectiveHandler. 142 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 143 144 /// Stack of active macro instantiations. 145 std::vector<MacroInstantiation*> ActiveMacros; 146 147 /// List of bodies of anonymous macros. 148 std::deque<MCAsmMacro> MacroLikeBodies; 149 150 /// Boolean tracking whether macro substitution is enabled. 151 unsigned MacrosEnabledFlag : 1; 152 153 /// Keeps track of how many .macro's have been instantiated. 154 unsigned NumOfMacroInstantiations; 155 156 /// The values from the last parsed cpp hash file line comment if any. 157 struct CppHashInfoTy { 158 StringRef Filename; 159 int64_t LineNumber; 160 SMLoc Loc; 161 unsigned Buf; 162 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 163 }; 164 CppHashInfoTy CppHashInfo; 165 166 /// The filename from the first cpp hash file line comment, if any. 167 StringRef FirstCppHashFilename; 168 169 /// List of forward directional labels for diagnosis at the end. 170 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 171 172 SmallSet<StringRef, 2> LTODiscardSymbols; 173 174 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 175 unsigned AssemblerDialect = ~0U; 176 177 /// is Darwin compatibility enabled? 178 bool IsDarwin = false; 179 180 /// Are we parsing ms-style inline assembly? 181 bool ParsingMSInlineAsm = false; 182 183 /// Did we already inform the user about inconsistent MD5 usage? 184 bool ReportedInconsistentMD5 = false; 185 186 // Is alt macro mode enabled. 187 bool AltMacroMode = false; 188 189 public: 190 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 191 const MCAsmInfo &MAI, unsigned CB); 192 AsmParser(const AsmParser &) = delete; 193 AsmParser &operator=(const AsmParser &) = delete; 194 ~AsmParser() override; 195 196 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 197 198 void addDirectiveHandler(StringRef Directive, 199 ExtensionDirectiveHandler Handler) override { 200 ExtensionDirectiveMap[Directive] = Handler; 201 } 202 203 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 204 DirectiveKindMap[Directive.lower()] = DirectiveKindMap[Alias.lower()]; 205 } 206 207 /// @name MCAsmParser Interface 208 /// { 209 210 SourceMgr &getSourceManager() override { return SrcMgr; } 211 MCAsmLexer &getLexer() override { return Lexer; } 212 MCContext &getContext() override { return Ctx; } 213 MCStreamer &getStreamer() override { return Out; } 214 215 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 216 217 unsigned getAssemblerDialect() override { 218 if (AssemblerDialect == ~0U) 219 return MAI.getAssemblerDialect(); 220 else 221 return AssemblerDialect; 222 } 223 void setAssemblerDialect(unsigned i) override { 224 AssemblerDialect = i; 225 } 226 227 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 228 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 229 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 230 231 const AsmToken &Lex() override; 232 233 void setParsingMSInlineAsm(bool V) override { 234 ParsingMSInlineAsm = V; 235 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 236 // hex integer literals. 237 Lexer.setLexMasmIntegers(V); 238 } 239 bool isParsingMSInlineAsm() override { return ParsingMSInlineAsm; } 240 241 bool discardLTOSymbol(StringRef Name) const override { 242 return LTODiscardSymbols.contains(Name); 243 } 244 245 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 246 unsigned &NumOutputs, unsigned &NumInputs, 247 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 248 SmallVectorImpl<std::string> &Constraints, 249 SmallVectorImpl<std::string> &Clobbers, 250 const MCInstrInfo *MII, const MCInstPrinter *IP, 251 MCAsmParserSemaCallback &SI) override; 252 253 bool parseExpression(const MCExpr *&Res); 254 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 255 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 256 AsmTypeInfo *TypeInfo) override; 257 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 258 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 259 SMLoc &EndLoc) override; 260 bool parseAbsoluteExpression(int64_t &Res) override; 261 262 /// Parse a floating point expression using the float \p Semantics 263 /// and set \p Res to the value. 264 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 265 266 /// Parse an identifier or string (as a quoted identifier) 267 /// and set \p Res to the identifier contents. 268 bool parseIdentifier(StringRef &Res) override; 269 void eatToEndOfStatement() override; 270 271 bool checkForValidSection() override; 272 273 /// } 274 275 private: 276 bool parseStatement(ParseStatementInfo &Info, 277 MCAsmParserSemaCallback *SI); 278 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 279 bool parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo = true); 280 281 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 282 ArrayRef<MCAsmMacroParameter> Parameters); 283 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 284 ArrayRef<MCAsmMacroParameter> Parameters, 285 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 286 SMLoc L); 287 288 /// Are macros enabled in the parser? 289 bool areMacrosEnabled() {return MacrosEnabledFlag;} 290 291 /// Control a flag in the parser that enables or disables macros. 292 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 293 294 /// Are we inside a macro instantiation? 295 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 296 297 /// Handle entry to macro instantiation. 298 /// 299 /// \param M The macro. 300 /// \param NameLoc Instantiation location. 301 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 302 303 /// Handle exit from macro instantiation. 304 void handleMacroExit(); 305 306 /// Extract AsmTokens for a macro argument. 307 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 308 309 /// Parse all macro arguments for a given macro. 310 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 311 312 void printMacroInstantiations(); 313 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 314 SMRange Range = None) const { 315 ArrayRef<SMRange> Ranges(Range); 316 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 317 } 318 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 319 320 /// Should we emit DWARF describing this assembler source? (Returns false if 321 /// the source has .file directives, which means we don't want to generate 322 /// info describing the assembler source itself.) 323 bool enabledGenDwarfForAssembly(); 324 325 /// Enter the specified file. This returns true on failure. 326 bool enterIncludeFile(const std::string &Filename); 327 328 /// Process the specified file for the .incbin directive. 329 /// This returns true on failure. 330 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 331 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 332 333 /// Reset the current lexer position to that given by \p Loc. The 334 /// current token is not set; clients should ensure Lex() is called 335 /// subsequently. 336 /// 337 /// \param InBuffer If not 0, should be the known buffer id that contains the 338 /// location. 339 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 340 341 /// Parse up to the end of statement and a return the contents from the 342 /// current token until the end of the statement; the current token on exit 343 /// will be either the EndOfStatement or EOF. 344 StringRef parseStringToEndOfStatement() override; 345 346 /// Parse until the end of a statement or a comma is encountered, 347 /// return the contents from the current token up to the end or comma. 348 StringRef parseStringToComma(); 349 350 bool parseAssignment(StringRef Name, bool allow_redef, 351 bool NoDeadStrip = false); 352 353 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 354 MCBinaryExpr::Opcode &Kind); 355 356 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 357 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 358 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 359 360 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 361 362 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 363 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 364 365 // Generic (target and platform independent) directive parsing. 366 enum DirectiveKind { 367 DK_NO_DIRECTIVE, // Placeholder 368 DK_SET, 369 DK_EQU, 370 DK_EQUIV, 371 DK_ASCII, 372 DK_ASCIZ, 373 DK_STRING, 374 DK_BYTE, 375 DK_SHORT, 376 DK_RELOC, 377 DK_VALUE, 378 DK_2BYTE, 379 DK_LONG, 380 DK_INT, 381 DK_4BYTE, 382 DK_QUAD, 383 DK_8BYTE, 384 DK_OCTA, 385 DK_DC, 386 DK_DC_A, 387 DK_DC_B, 388 DK_DC_D, 389 DK_DC_L, 390 DK_DC_S, 391 DK_DC_W, 392 DK_DC_X, 393 DK_DCB, 394 DK_DCB_B, 395 DK_DCB_D, 396 DK_DCB_L, 397 DK_DCB_S, 398 DK_DCB_W, 399 DK_DCB_X, 400 DK_DS, 401 DK_DS_B, 402 DK_DS_D, 403 DK_DS_L, 404 DK_DS_P, 405 DK_DS_S, 406 DK_DS_W, 407 DK_DS_X, 408 DK_SINGLE, 409 DK_FLOAT, 410 DK_DOUBLE, 411 DK_ALIGN, 412 DK_ALIGN32, 413 DK_BALIGN, 414 DK_BALIGNW, 415 DK_BALIGNL, 416 DK_P2ALIGN, 417 DK_P2ALIGNW, 418 DK_P2ALIGNL, 419 DK_ORG, 420 DK_FILL, 421 DK_ENDR, 422 DK_BUNDLE_ALIGN_MODE, 423 DK_BUNDLE_LOCK, 424 DK_BUNDLE_UNLOCK, 425 DK_ZERO, 426 DK_EXTERN, 427 DK_GLOBL, 428 DK_GLOBAL, 429 DK_LAZY_REFERENCE, 430 DK_NO_DEAD_STRIP, 431 DK_SYMBOL_RESOLVER, 432 DK_PRIVATE_EXTERN, 433 DK_REFERENCE, 434 DK_WEAK_DEFINITION, 435 DK_WEAK_REFERENCE, 436 DK_WEAK_DEF_CAN_BE_HIDDEN, 437 DK_COLD, 438 DK_COMM, 439 DK_COMMON, 440 DK_LCOMM, 441 DK_ABORT, 442 DK_INCLUDE, 443 DK_INCBIN, 444 DK_CODE16, 445 DK_CODE16GCC, 446 DK_REPT, 447 DK_IRP, 448 DK_IRPC, 449 DK_IF, 450 DK_IFEQ, 451 DK_IFGE, 452 DK_IFGT, 453 DK_IFLE, 454 DK_IFLT, 455 DK_IFNE, 456 DK_IFB, 457 DK_IFNB, 458 DK_IFC, 459 DK_IFEQS, 460 DK_IFNC, 461 DK_IFNES, 462 DK_IFDEF, 463 DK_IFNDEF, 464 DK_IFNOTDEF, 465 DK_ELSEIF, 466 DK_ELSE, 467 DK_ENDIF, 468 DK_SPACE, 469 DK_SKIP, 470 DK_FILE, 471 DK_LINE, 472 DK_LOC, 473 DK_STABS, 474 DK_CV_FILE, 475 DK_CV_FUNC_ID, 476 DK_CV_INLINE_SITE_ID, 477 DK_CV_LOC, 478 DK_CV_LINETABLE, 479 DK_CV_INLINE_LINETABLE, 480 DK_CV_DEF_RANGE, 481 DK_CV_STRINGTABLE, 482 DK_CV_STRING, 483 DK_CV_FILECHECKSUMS, 484 DK_CV_FILECHECKSUM_OFFSET, 485 DK_CV_FPO_DATA, 486 DK_CFI_SECTIONS, 487 DK_CFI_STARTPROC, 488 DK_CFI_ENDPROC, 489 DK_CFI_DEF_CFA, 490 DK_CFI_DEF_CFA_OFFSET, 491 DK_CFI_ADJUST_CFA_OFFSET, 492 DK_CFI_DEF_CFA_REGISTER, 493 DK_CFI_OFFSET, 494 DK_CFI_REL_OFFSET, 495 DK_CFI_PERSONALITY, 496 DK_CFI_LSDA, 497 DK_CFI_REMEMBER_STATE, 498 DK_CFI_RESTORE_STATE, 499 DK_CFI_SAME_VALUE, 500 DK_CFI_RESTORE, 501 DK_CFI_ESCAPE, 502 DK_CFI_RETURN_COLUMN, 503 DK_CFI_SIGNAL_FRAME, 504 DK_CFI_UNDEFINED, 505 DK_CFI_REGISTER, 506 DK_CFI_WINDOW_SAVE, 507 DK_CFI_B_KEY_FRAME, 508 DK_MACROS_ON, 509 DK_MACROS_OFF, 510 DK_ALTMACRO, 511 DK_NOALTMACRO, 512 DK_MACRO, 513 DK_EXITM, 514 DK_ENDM, 515 DK_ENDMACRO, 516 DK_PURGEM, 517 DK_SLEB128, 518 DK_ULEB128, 519 DK_ERR, 520 DK_ERROR, 521 DK_WARNING, 522 DK_PRINT, 523 DK_ADDRSIG, 524 DK_ADDRSIG_SYM, 525 DK_PSEUDO_PROBE, 526 DK_LTO_DISCARD, 527 DK_END 528 }; 529 530 /// Maps directive name --> DirectiveKind enum, for 531 /// directives parsed by this class. 532 StringMap<DirectiveKind> DirectiveKindMap; 533 534 // Codeview def_range type parsing. 535 enum CVDefRangeType { 536 CVDR_DEFRANGE = 0, // Placeholder 537 CVDR_DEFRANGE_REGISTER, 538 CVDR_DEFRANGE_FRAMEPOINTER_REL, 539 CVDR_DEFRANGE_SUBFIELD_REGISTER, 540 CVDR_DEFRANGE_REGISTER_REL 541 }; 542 543 /// Maps Codeview def_range types --> CVDefRangeType enum, for 544 /// Codeview def_range types parsed by this class. 545 StringMap<CVDefRangeType> CVDefRangeTypeMap; 546 547 // ".ascii", ".asciz", ".string" 548 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 549 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 550 bool parseDirectiveValue(StringRef IDVal, 551 unsigned Size); // ".byte", ".long", ... 552 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 553 bool parseDirectiveRealValue(StringRef IDVal, 554 const fltSemantics &); // ".single", ... 555 bool parseDirectiveFill(); // ".fill" 556 bool parseDirectiveZero(); // ".zero" 557 // ".set", ".equ", ".equiv" 558 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 559 bool parseDirectiveOrg(); // ".org" 560 // ".align{,32}", ".p2align{,w,l}" 561 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 562 563 // ".file", ".line", ".loc", ".stabs" 564 bool parseDirectiveFile(SMLoc DirectiveLoc); 565 bool parseDirectiveLine(); 566 bool parseDirectiveLoc(); 567 bool parseDirectiveStabs(); 568 569 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 570 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 571 bool parseDirectiveCVFile(); 572 bool parseDirectiveCVFuncId(); 573 bool parseDirectiveCVInlineSiteId(); 574 bool parseDirectiveCVLoc(); 575 bool parseDirectiveCVLinetable(); 576 bool parseDirectiveCVInlineLinetable(); 577 bool parseDirectiveCVDefRange(); 578 bool parseDirectiveCVString(); 579 bool parseDirectiveCVStringTable(); 580 bool parseDirectiveCVFileChecksums(); 581 bool parseDirectiveCVFileChecksumOffset(); 582 bool parseDirectiveCVFPOData(); 583 584 // .cfi directives 585 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 586 bool parseDirectiveCFIWindowSave(); 587 bool parseDirectiveCFISections(); 588 bool parseDirectiveCFIStartProc(); 589 bool parseDirectiveCFIEndProc(); 590 bool parseDirectiveCFIDefCfaOffset(); 591 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 592 bool parseDirectiveCFIAdjustCfaOffset(); 593 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 594 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 595 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 596 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 597 bool parseDirectiveCFIRememberState(); 598 bool parseDirectiveCFIRestoreState(); 599 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 600 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 601 bool parseDirectiveCFIEscape(); 602 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 603 bool parseDirectiveCFISignalFrame(); 604 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 605 606 // macro directives 607 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 608 bool parseDirectiveExitMacro(StringRef Directive); 609 bool parseDirectiveEndMacro(StringRef Directive); 610 bool parseDirectiveMacro(SMLoc DirectiveLoc); 611 bool parseDirectiveMacrosOnOff(StringRef Directive); 612 // alternate macro mode directives 613 bool parseDirectiveAltmacro(StringRef Directive); 614 // ".bundle_align_mode" 615 bool parseDirectiveBundleAlignMode(); 616 // ".bundle_lock" 617 bool parseDirectiveBundleLock(); 618 // ".bundle_unlock" 619 bool parseDirectiveBundleUnlock(); 620 621 // ".space", ".skip" 622 bool parseDirectiveSpace(StringRef IDVal); 623 624 // ".dcb" 625 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 626 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 627 // ".ds" 628 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 629 630 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 631 bool parseDirectiveLEB128(bool Signed); 632 633 /// Parse a directive like ".globl" which 634 /// accepts a single symbol (which should be a label or an external). 635 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 636 637 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 638 639 bool parseDirectiveAbort(); // ".abort" 640 bool parseDirectiveInclude(); // ".include" 641 bool parseDirectiveIncbin(); // ".incbin" 642 643 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 644 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 645 // ".ifb" or ".ifnb", depending on ExpectBlank. 646 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 647 // ".ifc" or ".ifnc", depending on ExpectEqual. 648 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 649 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 650 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 651 // ".ifdef" or ".ifndef", depending on expect_defined 652 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 653 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 654 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 655 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 656 bool parseEscapedString(std::string &Data) override; 657 bool parseAngleBracketString(std::string &Data) override; 658 659 const MCExpr *applyModifierToExpr(const MCExpr *E, 660 MCSymbolRefExpr::VariantKind Variant); 661 662 // Macro-like directives 663 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 664 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 665 raw_svector_ostream &OS); 666 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 667 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 668 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 669 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 670 671 // "_emit" or "__emit" 672 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 673 size_t Len); 674 675 // "align" 676 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 677 678 // "end" 679 bool parseDirectiveEnd(SMLoc DirectiveLoc); 680 681 // ".err" or ".error" 682 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 683 684 // ".warning" 685 bool parseDirectiveWarning(SMLoc DirectiveLoc); 686 687 // .print <double-quotes-string> 688 bool parseDirectivePrint(SMLoc DirectiveLoc); 689 690 // .pseudoprobe 691 bool parseDirectivePseudoProbe(); 692 693 // ".lto_discard" 694 bool parseDirectiveLTODiscard(); 695 696 // Directives to support address-significance tables. 697 bool parseDirectiveAddrsig(); 698 bool parseDirectiveAddrsigSym(); 699 700 void initializeDirectiveKindMap(); 701 void initializeCVDefRangeTypeMap(); 702 }; 703 704 } // end anonymous namespace 705 706 namespace llvm { 707 708 extern MCAsmParserExtension *createDarwinAsmParser(); 709 extern MCAsmParserExtension *createELFAsmParser(); 710 extern MCAsmParserExtension *createCOFFAsmParser(); 711 extern MCAsmParserExtension *createWasmAsmParser(); 712 713 } // end namespace llvm 714 715 enum { DEFAULT_ADDRSPACE = 0 }; 716 717 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 718 const MCAsmInfo &MAI, unsigned CB = 0) 719 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 720 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 721 HadError = false; 722 // Save the old handler. 723 SavedDiagHandler = SrcMgr.getDiagHandler(); 724 SavedDiagContext = SrcMgr.getDiagContext(); 725 // Set our own handler which calls the saved handler. 726 SrcMgr.setDiagHandler(DiagHandler, this); 727 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 728 // Make MCStreamer aware of the StartTokLoc for locations in diagnostics. 729 Out.setStartTokLocPtr(&StartTokLoc); 730 731 // Initialize the platform / file format parser. 732 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 733 case MCObjectFileInfo::IsCOFF: 734 PlatformParser.reset(createCOFFAsmParser()); 735 break; 736 case MCObjectFileInfo::IsMachO: 737 PlatformParser.reset(createDarwinAsmParser()); 738 IsDarwin = true; 739 break; 740 case MCObjectFileInfo::IsELF: 741 PlatformParser.reset(createELFAsmParser()); 742 break; 743 case MCObjectFileInfo::IsWasm: 744 PlatformParser.reset(createWasmAsmParser()); 745 break; 746 case MCObjectFileInfo::IsXCOFF: 747 report_fatal_error( 748 "Need to implement createXCOFFAsmParser for XCOFF format."); 749 break; 750 } 751 752 PlatformParser->Initialize(*this); 753 initializeDirectiveKindMap(); 754 initializeCVDefRangeTypeMap(); 755 756 NumOfMacroInstantiations = 0; 757 } 758 759 AsmParser::~AsmParser() { 760 assert((HadError || ActiveMacros.empty()) && 761 "Unexpected active macro instantiation!"); 762 763 // Remove MCStreamer's reference to the parser SMLoc. 764 Out.setStartTokLocPtr(nullptr); 765 // Restore the saved diagnostics handler and context for use during 766 // finalization. 767 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 768 } 769 770 void AsmParser::printMacroInstantiations() { 771 // Print the active macro instantiation stack. 772 for (std::vector<MacroInstantiation *>::const_reverse_iterator 773 it = ActiveMacros.rbegin(), 774 ie = ActiveMacros.rend(); 775 it != ie; ++it) 776 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 777 "while in macro instantiation"); 778 } 779 780 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 781 printPendingErrors(); 782 printMessage(L, SourceMgr::DK_Note, Msg, Range); 783 printMacroInstantiations(); 784 } 785 786 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 787 if(getTargetParser().getTargetOptions().MCNoWarn) 788 return false; 789 if (getTargetParser().getTargetOptions().MCFatalWarnings) 790 return Error(L, Msg, Range); 791 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 792 printMacroInstantiations(); 793 return false; 794 } 795 796 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 797 HadError = true; 798 printMessage(L, SourceMgr::DK_Error, Msg, Range); 799 printMacroInstantiations(); 800 return true; 801 } 802 803 bool AsmParser::enterIncludeFile(const std::string &Filename) { 804 std::string IncludedFile; 805 unsigned NewBuf = 806 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 807 if (!NewBuf) 808 return true; 809 810 CurBuffer = NewBuf; 811 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 812 return false; 813 } 814 815 /// Process the specified .incbin file by searching for it in the include paths 816 /// then just emitting the byte contents of the file to the streamer. This 817 /// returns true on failure. 818 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 819 const MCExpr *Count, SMLoc Loc) { 820 std::string IncludedFile; 821 unsigned NewBuf = 822 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 823 if (!NewBuf) 824 return true; 825 826 // Pick up the bytes from the file and emit them. 827 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 828 Bytes = Bytes.drop_front(Skip); 829 if (Count) { 830 int64_t Res; 831 if (!Count->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 832 return Error(Loc, "expected absolute expression"); 833 if (Res < 0) 834 return Warning(Loc, "negative count has no effect"); 835 Bytes = Bytes.take_front(Res); 836 } 837 getStreamer().emitBytes(Bytes); 838 return false; 839 } 840 841 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 842 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 843 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 844 Loc.getPointer()); 845 } 846 847 const AsmToken &AsmParser::Lex() { 848 if (Lexer.getTok().is(AsmToken::Error)) 849 Error(Lexer.getErrLoc(), Lexer.getErr()); 850 851 // if it's a end of statement with a comment in it 852 if (getTok().is(AsmToken::EndOfStatement)) { 853 // if this is a line comment output it. 854 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 855 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 856 Out.addExplicitComment(Twine(getTok().getString())); 857 } 858 859 const AsmToken *tok = &Lexer.Lex(); 860 861 // Parse comments here to be deferred until end of next statement. 862 while (tok->is(AsmToken::Comment)) { 863 if (MAI.preserveAsmComments()) 864 Out.addExplicitComment(Twine(tok->getString())); 865 tok = &Lexer.Lex(); 866 } 867 868 if (tok->is(AsmToken::Eof)) { 869 // If this is the end of an included file, pop the parent file off the 870 // include stack. 871 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 872 if (ParentIncludeLoc != SMLoc()) { 873 jumpToLoc(ParentIncludeLoc); 874 return Lex(); 875 } 876 } 877 878 return *tok; 879 } 880 881 bool AsmParser::enabledGenDwarfForAssembly() { 882 // Check whether the user specified -g. 883 if (!getContext().getGenDwarfForAssembly()) 884 return false; 885 // If we haven't encountered any .file directives (which would imply that 886 // the assembler source was produced with debug info already) then emit one 887 // describing the assembler source file itself. 888 if (getContext().getGenDwarfFileNumber() == 0) { 889 // Use the first #line directive for this, if any. It's preprocessed, so 890 // there is no checksum, and of course no source directive. 891 if (!FirstCppHashFilename.empty()) 892 getContext().setMCLineTableRootFile(/*CUID=*/0, 893 getContext().getCompilationDir(), 894 FirstCppHashFilename, 895 /*Cksum=*/None, /*Source=*/None); 896 const MCDwarfFile &RootFile = 897 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 898 getContext().setGenDwarfFileNumber(getStreamer().emitDwarfFileDirective( 899 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 900 RootFile.Checksum, RootFile.Source)); 901 } 902 return true; 903 } 904 905 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 906 LTODiscardSymbols.clear(); 907 908 // Create the initial section, if requested. 909 if (!NoInitialTextSection) 910 Out.InitSections(false); 911 912 // Prime the lexer. 913 Lex(); 914 915 HadError = false; 916 AsmCond StartingCondState = TheCondState; 917 SmallVector<AsmRewrite, 4> AsmStrRewrites; 918 919 // If we are generating dwarf for assembly source files save the initial text 920 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 921 // emitting any actual debug info yet and haven't had a chance to parse any 922 // embedded .file directives.) 923 if (getContext().getGenDwarfForAssembly()) { 924 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 925 if (!Sec->getBeginSymbol()) { 926 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 927 getStreamer().emitLabel(SectionStartSym); 928 Sec->setBeginSymbol(SectionStartSym); 929 } 930 bool InsertResult = getContext().addGenDwarfSection(Sec); 931 assert(InsertResult && ".text section should not have debug info yet"); 932 (void)InsertResult; 933 } 934 935 getTargetParser().onBeginOfFile(); 936 937 // While we have input, parse each statement. 938 while (Lexer.isNot(AsmToken::Eof)) { 939 ParseStatementInfo Info(&AsmStrRewrites); 940 bool Parsed = parseStatement(Info, nullptr); 941 942 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 943 // for printing ErrMsg via Lex() only if no (presumably better) parser error 944 // exists. 945 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 946 Lex(); 947 } 948 949 // parseStatement returned true so may need to emit an error. 950 printPendingErrors(); 951 952 // Skipping to the next line if needed. 953 if (Parsed && !getLexer().isAtStartOfStatement()) 954 eatToEndOfStatement(); 955 } 956 957 getTargetParser().onEndOfFile(); 958 printPendingErrors(); 959 960 // All errors should have been emitted. 961 assert(!hasPendingError() && "unexpected error from parseStatement"); 962 963 getTargetParser().flushPendingInstructions(getStreamer()); 964 965 if (TheCondState.TheCond != StartingCondState.TheCond || 966 TheCondState.Ignore != StartingCondState.Ignore) 967 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 968 // Check to see there are no empty DwarfFile slots. 969 const auto &LineTables = getContext().getMCDwarfLineTables(); 970 if (!LineTables.empty()) { 971 unsigned Index = 0; 972 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 973 if (File.Name.empty() && Index != 0) 974 printError(getTok().getLoc(), "unassigned file number: " + 975 Twine(Index) + 976 " for .file directives"); 977 ++Index; 978 } 979 } 980 981 // Check to see that all assembler local symbols were actually defined. 982 // Targets that don't do subsections via symbols may not want this, though, 983 // so conservatively exclude them. Only do this if we're finalizing, though, 984 // as otherwise we won't necessarilly have seen everything yet. 985 if (!NoFinalize) { 986 if (MAI.hasSubsectionsViaSymbols()) { 987 for (const auto &TableEntry : getContext().getSymbols()) { 988 MCSymbol *Sym = TableEntry.getValue(); 989 // Variable symbols may not be marked as defined, so check those 990 // explicitly. If we know it's a variable, we have a definition for 991 // the purposes of this check. 992 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 993 // FIXME: We would really like to refer back to where the symbol was 994 // first referenced for a source location. We need to add something 995 // to track that. Currently, we just point to the end of the file. 996 printError(getTok().getLoc(), "assembler local symbol '" + 997 Sym->getName() + "' not defined"); 998 } 999 } 1000 1001 // Temporary symbols like the ones for directional jumps don't go in the 1002 // symbol table. They also need to be diagnosed in all (final) cases. 1003 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 1004 if (std::get<2>(LocSym)->isUndefined()) { 1005 // Reset the state of any "# line file" directives we've seen to the 1006 // context as it was at the diagnostic site. 1007 CppHashInfo = std::get<1>(LocSym); 1008 printError(std::get<0>(LocSym), "directional label undefined"); 1009 } 1010 } 1011 } 1012 1013 // Finalize the output stream if there are no errors and if the client wants 1014 // us to. 1015 if (!HadError && !NoFinalize) 1016 Out.Finish(Lexer.getLoc()); 1017 1018 return HadError || getContext().hadError(); 1019 } 1020 1021 bool AsmParser::checkForValidSection() { 1022 if (!ParsingMSInlineAsm && !getStreamer().getCurrentSectionOnly()) { 1023 Out.InitSections(false); 1024 return Error(getTok().getLoc(), 1025 "expected section directive before assembly directive"); 1026 } 1027 return false; 1028 } 1029 1030 /// Throw away the rest of the line for testing purposes. 1031 void AsmParser::eatToEndOfStatement() { 1032 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1033 Lexer.Lex(); 1034 1035 // Eat EOL. 1036 if (Lexer.is(AsmToken::EndOfStatement)) 1037 Lexer.Lex(); 1038 } 1039 1040 StringRef AsmParser::parseStringToEndOfStatement() { 1041 const char *Start = getTok().getLoc().getPointer(); 1042 1043 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 1044 Lexer.Lex(); 1045 1046 const char *End = getTok().getLoc().getPointer(); 1047 return StringRef(Start, End - Start); 1048 } 1049 1050 StringRef AsmParser::parseStringToComma() { 1051 const char *Start = getTok().getLoc().getPointer(); 1052 1053 while (Lexer.isNot(AsmToken::EndOfStatement) && 1054 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 1055 Lexer.Lex(); 1056 1057 const char *End = getTok().getLoc().getPointer(); 1058 return StringRef(Start, End - Start); 1059 } 1060 1061 /// Parse a paren expression and return it. 1062 /// NOTE: This assumes the leading '(' has already been consumed. 1063 /// 1064 /// parenexpr ::= expr) 1065 /// 1066 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1067 if (parseExpression(Res)) 1068 return true; 1069 if (Lexer.isNot(AsmToken::RParen)) 1070 return TokError("expected ')' in parentheses expression"); 1071 EndLoc = Lexer.getTok().getEndLoc(); 1072 Lex(); 1073 return false; 1074 } 1075 1076 /// Parse a bracket expression and return it. 1077 /// NOTE: This assumes the leading '[' has already been consumed. 1078 /// 1079 /// bracketexpr ::= expr] 1080 /// 1081 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 1082 if (parseExpression(Res)) 1083 return true; 1084 EndLoc = getTok().getEndLoc(); 1085 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 1086 return true; 1087 return false; 1088 } 1089 1090 /// Parse a primary expression and return it. 1091 /// primaryexpr ::= (parenexpr 1092 /// primaryexpr ::= symbol 1093 /// primaryexpr ::= number 1094 /// primaryexpr ::= '.' 1095 /// primaryexpr ::= ~,+,- primaryexpr 1096 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc, 1097 AsmTypeInfo *TypeInfo) { 1098 SMLoc FirstTokenLoc = getLexer().getLoc(); 1099 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 1100 switch (FirstTokenKind) { 1101 default: 1102 return TokError("unknown token in expression"); 1103 // If we have an error assume that we've already handled it. 1104 case AsmToken::Error: 1105 return true; 1106 case AsmToken::Exclaim: 1107 Lex(); // Eat the operator. 1108 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1109 return true; 1110 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1111 return false; 1112 case AsmToken::Dollar: 1113 case AsmToken::At: 1114 case AsmToken::String: 1115 case AsmToken::Identifier: { 1116 StringRef Identifier; 1117 if (parseIdentifier(Identifier)) { 1118 // We may have failed but $ may be a valid token. 1119 if (getTok().is(AsmToken::Dollar)) { 1120 if (Lexer.getMAI().getDollarIsPC()) { 1121 Lex(); 1122 // This is a '$' reference, which references the current PC. Emit a 1123 // temporary label to the streamer and refer to it. 1124 MCSymbol *Sym = Ctx.createTempSymbol(); 1125 Out.emitLabel(Sym); 1126 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1127 getContext()); 1128 EndLoc = FirstTokenLoc; 1129 return false; 1130 } 1131 return Error(FirstTokenLoc, "invalid token in expression"); 1132 } 1133 } 1134 // Parse symbol variant 1135 std::pair<StringRef, StringRef> Split; 1136 if (!MAI.useParensForSymbolVariant()) { 1137 if (FirstTokenKind == AsmToken::String) { 1138 if (Lexer.is(AsmToken::At)) { 1139 Lex(); // eat @ 1140 SMLoc AtLoc = getLexer().getLoc(); 1141 StringRef VName; 1142 if (parseIdentifier(VName)) 1143 return Error(AtLoc, "expected symbol variant after '@'"); 1144 1145 Split = std::make_pair(Identifier, VName); 1146 } 1147 } else { 1148 Split = Identifier.split('@'); 1149 } 1150 } else if (Lexer.is(AsmToken::LParen)) { 1151 Lex(); // eat '('. 1152 StringRef VName; 1153 parseIdentifier(VName); 1154 // eat ')'. 1155 if (parseToken(AsmToken::RParen, 1156 "unexpected token in variant, expected ')'")) 1157 return true; 1158 Split = std::make_pair(Identifier, VName); 1159 } 1160 1161 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1162 1163 // This is a symbol reference. 1164 StringRef SymbolName = Identifier; 1165 if (SymbolName.empty()) 1166 return Error(getLexer().getLoc(), "expected a symbol reference"); 1167 1168 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1169 1170 // Lookup the symbol variant if used. 1171 if (!Split.second.empty()) { 1172 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1173 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1174 SymbolName = Split.first; 1175 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1176 Variant = MCSymbolRefExpr::VK_None; 1177 } else { 1178 return Error(SMLoc::getFromPointer(Split.second.begin()), 1179 "invalid variant '" + Split.second + "'"); 1180 } 1181 } 1182 1183 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1184 if (!Sym) 1185 Sym = getContext().getOrCreateSymbol(SymbolName); 1186 1187 // If this is an absolute variable reference, substitute it now to preserve 1188 // semantics in the face of reassignment. 1189 if (Sym->isVariable()) { 1190 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1191 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1192 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1193 DoInline = TV->inlineAssignedExpr(); 1194 if (DoInline) { 1195 if (Variant) 1196 return Error(EndLoc, "unexpected modifier on variable reference"); 1197 Res = Sym->getVariableValue(/*SetUsed*/ false); 1198 return false; 1199 } 1200 } 1201 1202 // Otherwise create a symbol ref. 1203 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1204 return false; 1205 } 1206 case AsmToken::BigNum: 1207 return TokError("literal value out of range for directive"); 1208 case AsmToken::Integer: { 1209 SMLoc Loc = getTok().getLoc(); 1210 int64_t IntVal = getTok().getIntVal(); 1211 Res = MCConstantExpr::create(IntVal, getContext()); 1212 EndLoc = Lexer.getTok().getEndLoc(); 1213 Lex(); // Eat token. 1214 // Look for 'b' or 'f' following an Integer as a directional label 1215 if (Lexer.getKind() == AsmToken::Identifier) { 1216 StringRef IDVal = getTok().getString(); 1217 // Lookup the symbol variant if used. 1218 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1219 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1220 if (Split.first.size() != IDVal.size()) { 1221 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1222 if (Variant == MCSymbolRefExpr::VK_Invalid) 1223 return TokError("invalid variant '" + Split.second + "'"); 1224 IDVal = Split.first; 1225 } 1226 if (IDVal == "f" || IDVal == "b") { 1227 MCSymbol *Sym = 1228 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1229 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1230 if (IDVal == "b" && Sym->isUndefined()) 1231 return Error(Loc, "directional label undefined"); 1232 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1233 EndLoc = Lexer.getTok().getEndLoc(); 1234 Lex(); // Eat identifier. 1235 } 1236 } 1237 return false; 1238 } 1239 case AsmToken::Real: { 1240 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1241 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1242 Res = MCConstantExpr::create(IntVal, getContext()); 1243 EndLoc = Lexer.getTok().getEndLoc(); 1244 Lex(); // Eat token. 1245 return false; 1246 } 1247 case AsmToken::Dot: { 1248 if (!MAI.getDotIsPC()) 1249 return TokError("cannot use . as current PC"); 1250 1251 // This is a '.' reference, which references the current PC. Emit a 1252 // temporary label to the streamer and refer to it. 1253 MCSymbol *Sym = Ctx.createTempSymbol(); 1254 Out.emitLabel(Sym); 1255 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1256 EndLoc = Lexer.getTok().getEndLoc(); 1257 Lex(); // Eat identifier. 1258 return false; 1259 } 1260 case AsmToken::LParen: 1261 Lex(); // Eat the '('. 1262 return parseParenExpr(Res, EndLoc); 1263 case AsmToken::LBrac: 1264 if (!PlatformParser->HasBracketExpressions()) 1265 return TokError("brackets expression not supported on this target"); 1266 Lex(); // Eat the '['. 1267 return parseBracketExpr(Res, EndLoc); 1268 case AsmToken::Minus: 1269 Lex(); // Eat the operator. 1270 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1271 return true; 1272 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1273 return false; 1274 case AsmToken::Plus: 1275 Lex(); // Eat the operator. 1276 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1277 return true; 1278 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1279 return false; 1280 case AsmToken::Tilde: 1281 Lex(); // Eat the operator. 1282 if (parsePrimaryExpr(Res, EndLoc, TypeInfo)) 1283 return true; 1284 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1285 return false; 1286 // MIPS unary expression operators. The lexer won't generate these tokens if 1287 // MCAsmInfo::HasMipsExpressions is false for the target. 1288 case AsmToken::PercentCall16: 1289 case AsmToken::PercentCall_Hi: 1290 case AsmToken::PercentCall_Lo: 1291 case AsmToken::PercentDtprel_Hi: 1292 case AsmToken::PercentDtprel_Lo: 1293 case AsmToken::PercentGot: 1294 case AsmToken::PercentGot_Disp: 1295 case AsmToken::PercentGot_Hi: 1296 case AsmToken::PercentGot_Lo: 1297 case AsmToken::PercentGot_Ofst: 1298 case AsmToken::PercentGot_Page: 1299 case AsmToken::PercentGottprel: 1300 case AsmToken::PercentGp_Rel: 1301 case AsmToken::PercentHi: 1302 case AsmToken::PercentHigher: 1303 case AsmToken::PercentHighest: 1304 case AsmToken::PercentLo: 1305 case AsmToken::PercentNeg: 1306 case AsmToken::PercentPcrel_Hi: 1307 case AsmToken::PercentPcrel_Lo: 1308 case AsmToken::PercentTlsgd: 1309 case AsmToken::PercentTlsldm: 1310 case AsmToken::PercentTprel_Hi: 1311 case AsmToken::PercentTprel_Lo: 1312 Lex(); // Eat the operator. 1313 if (Lexer.isNot(AsmToken::LParen)) 1314 return TokError("expected '(' after operator"); 1315 Lex(); // Eat the operator. 1316 if (parseExpression(Res, EndLoc)) 1317 return true; 1318 if (Lexer.isNot(AsmToken::RParen)) 1319 return TokError("expected ')'"); 1320 Lex(); // Eat the operator. 1321 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1322 return !Res; 1323 } 1324 } 1325 1326 bool AsmParser::parseExpression(const MCExpr *&Res) { 1327 SMLoc EndLoc; 1328 return parseExpression(Res, EndLoc); 1329 } 1330 1331 const MCExpr * 1332 AsmParser::applyModifierToExpr(const MCExpr *E, 1333 MCSymbolRefExpr::VariantKind Variant) { 1334 // Ask the target implementation about this expression first. 1335 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1336 if (NewE) 1337 return NewE; 1338 // Recurse over the given expression, rebuilding it to apply the given variant 1339 // if there is exactly one symbol. 1340 switch (E->getKind()) { 1341 case MCExpr::Target: 1342 case MCExpr::Constant: 1343 return nullptr; 1344 1345 case MCExpr::SymbolRef: { 1346 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1347 1348 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1349 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1350 "' (already modified)"); 1351 return E; 1352 } 1353 1354 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1355 } 1356 1357 case MCExpr::Unary: { 1358 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1359 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1360 if (!Sub) 1361 return nullptr; 1362 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1363 } 1364 1365 case MCExpr::Binary: { 1366 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1367 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1368 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1369 1370 if (!LHS && !RHS) 1371 return nullptr; 1372 1373 if (!LHS) 1374 LHS = BE->getLHS(); 1375 if (!RHS) 1376 RHS = BE->getRHS(); 1377 1378 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1379 } 1380 } 1381 1382 llvm_unreachable("Invalid expression kind!"); 1383 } 1384 1385 /// This function checks if the next token is <string> type or arithmetic. 1386 /// string that begin with character '<' must end with character '>'. 1387 /// otherwise it is arithmetics. 1388 /// If the function returns a 'true' value, 1389 /// the End argument will be filled with the last location pointed to the '>' 1390 /// character. 1391 1392 /// There is a gap between the AltMacro's documentation and the single quote 1393 /// implementation. GCC does not fully support this feature and so we will not 1394 /// support it. 1395 /// TODO: Adding single quote as a string. 1396 static bool isAngleBracketString(SMLoc &StrLoc, SMLoc &EndLoc) { 1397 assert((StrLoc.getPointer() != nullptr) && 1398 "Argument to the function cannot be a NULL value"); 1399 const char *CharPtr = StrLoc.getPointer(); 1400 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1401 (*CharPtr != '\0')) { 1402 if (*CharPtr == '!') 1403 CharPtr++; 1404 CharPtr++; 1405 } 1406 if (*CharPtr == '>') { 1407 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1408 return true; 1409 } 1410 return false; 1411 } 1412 1413 /// creating a string without the escape characters '!'. 1414 static std::string angleBracketString(StringRef AltMacroStr) { 1415 std::string Res; 1416 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1417 if (AltMacroStr[Pos] == '!') 1418 Pos++; 1419 Res += AltMacroStr[Pos]; 1420 } 1421 return Res; 1422 } 1423 1424 /// Parse an expression and return it. 1425 /// 1426 /// expr ::= expr &&,|| expr -> lowest. 1427 /// expr ::= expr |,^,&,! expr 1428 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1429 /// expr ::= expr <<,>> expr 1430 /// expr ::= expr +,- expr 1431 /// expr ::= expr *,/,% expr -> highest. 1432 /// expr ::= primaryexpr 1433 /// 1434 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1435 // Parse the expression. 1436 Res = nullptr; 1437 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1438 parseBinOpRHS(1, Res, EndLoc)) 1439 return true; 1440 1441 // As a special case, we support 'a op b @ modifier' by rewriting the 1442 // expression to include the modifier. This is inefficient, but in general we 1443 // expect users to use 'a@modifier op b'. 1444 if (Lexer.getKind() == AsmToken::At) { 1445 Lex(); 1446 1447 if (Lexer.isNot(AsmToken::Identifier)) 1448 return TokError("unexpected symbol modifier following '@'"); 1449 1450 MCSymbolRefExpr::VariantKind Variant = 1451 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1452 if (Variant == MCSymbolRefExpr::VK_Invalid) 1453 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1454 1455 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1456 if (!ModifiedRes) { 1457 return TokError("invalid modifier '" + getTok().getIdentifier() + 1458 "' (no symbols present)"); 1459 } 1460 1461 Res = ModifiedRes; 1462 Lex(); 1463 } 1464 1465 // Try to constant fold it up front, if possible. Do not exploit 1466 // assembler here. 1467 int64_t Value; 1468 if (Res->evaluateAsAbsolute(Value)) 1469 Res = MCConstantExpr::create(Value, getContext()); 1470 1471 return false; 1472 } 1473 1474 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1475 Res = nullptr; 1476 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1477 } 1478 1479 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1480 SMLoc &EndLoc) { 1481 if (parseParenExpr(Res, EndLoc)) 1482 return true; 1483 1484 for (; ParenDepth > 0; --ParenDepth) { 1485 if (parseBinOpRHS(1, Res, EndLoc)) 1486 return true; 1487 1488 // We don't Lex() the last RParen. 1489 // This is the same behavior as parseParenExpression(). 1490 if (ParenDepth - 1 > 0) { 1491 EndLoc = getTok().getEndLoc(); 1492 if (parseToken(AsmToken::RParen, 1493 "expected ')' in parentheses expression")) 1494 return true; 1495 } 1496 } 1497 return false; 1498 } 1499 1500 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1501 const MCExpr *Expr; 1502 1503 SMLoc StartLoc = Lexer.getLoc(); 1504 if (parseExpression(Expr)) 1505 return true; 1506 1507 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1508 return Error(StartLoc, "expected absolute expression"); 1509 1510 return false; 1511 } 1512 1513 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1514 MCBinaryExpr::Opcode &Kind, 1515 bool ShouldUseLogicalShr) { 1516 switch (K) { 1517 default: 1518 return 0; // not a binop. 1519 1520 // Lowest Precedence: &&, || 1521 case AsmToken::AmpAmp: 1522 Kind = MCBinaryExpr::LAnd; 1523 return 1; 1524 case AsmToken::PipePipe: 1525 Kind = MCBinaryExpr::LOr; 1526 return 1; 1527 1528 // Low Precedence: |, &, ^ 1529 case AsmToken::Pipe: 1530 Kind = MCBinaryExpr::Or; 1531 return 2; 1532 case AsmToken::Caret: 1533 Kind = MCBinaryExpr::Xor; 1534 return 2; 1535 case AsmToken::Amp: 1536 Kind = MCBinaryExpr::And; 1537 return 2; 1538 1539 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1540 case AsmToken::EqualEqual: 1541 Kind = MCBinaryExpr::EQ; 1542 return 3; 1543 case AsmToken::ExclaimEqual: 1544 case AsmToken::LessGreater: 1545 Kind = MCBinaryExpr::NE; 1546 return 3; 1547 case AsmToken::Less: 1548 Kind = MCBinaryExpr::LT; 1549 return 3; 1550 case AsmToken::LessEqual: 1551 Kind = MCBinaryExpr::LTE; 1552 return 3; 1553 case AsmToken::Greater: 1554 Kind = MCBinaryExpr::GT; 1555 return 3; 1556 case AsmToken::GreaterEqual: 1557 Kind = MCBinaryExpr::GTE; 1558 return 3; 1559 1560 // Intermediate Precedence: <<, >> 1561 case AsmToken::LessLess: 1562 Kind = MCBinaryExpr::Shl; 1563 return 4; 1564 case AsmToken::GreaterGreater: 1565 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1566 return 4; 1567 1568 // High Intermediate Precedence: +, - 1569 case AsmToken::Plus: 1570 Kind = MCBinaryExpr::Add; 1571 return 5; 1572 case AsmToken::Minus: 1573 Kind = MCBinaryExpr::Sub; 1574 return 5; 1575 1576 // Highest Precedence: *, /, % 1577 case AsmToken::Star: 1578 Kind = MCBinaryExpr::Mul; 1579 return 6; 1580 case AsmToken::Slash: 1581 Kind = MCBinaryExpr::Div; 1582 return 6; 1583 case AsmToken::Percent: 1584 Kind = MCBinaryExpr::Mod; 1585 return 6; 1586 } 1587 } 1588 1589 static unsigned getGNUBinOpPrecedence(const MCAsmInfo &MAI, 1590 AsmToken::TokenKind K, 1591 MCBinaryExpr::Opcode &Kind, 1592 bool ShouldUseLogicalShr) { 1593 switch (K) { 1594 default: 1595 return 0; // not a binop. 1596 1597 // Lowest Precedence: &&, || 1598 case AsmToken::AmpAmp: 1599 Kind = MCBinaryExpr::LAnd; 1600 return 2; 1601 case AsmToken::PipePipe: 1602 Kind = MCBinaryExpr::LOr; 1603 return 1; 1604 1605 // Low Precedence: ==, !=, <>, <, <=, >, >= 1606 case AsmToken::EqualEqual: 1607 Kind = MCBinaryExpr::EQ; 1608 return 3; 1609 case AsmToken::ExclaimEqual: 1610 case AsmToken::LessGreater: 1611 Kind = MCBinaryExpr::NE; 1612 return 3; 1613 case AsmToken::Less: 1614 Kind = MCBinaryExpr::LT; 1615 return 3; 1616 case AsmToken::LessEqual: 1617 Kind = MCBinaryExpr::LTE; 1618 return 3; 1619 case AsmToken::Greater: 1620 Kind = MCBinaryExpr::GT; 1621 return 3; 1622 case AsmToken::GreaterEqual: 1623 Kind = MCBinaryExpr::GTE; 1624 return 3; 1625 1626 // Low Intermediate Precedence: +, - 1627 case AsmToken::Plus: 1628 Kind = MCBinaryExpr::Add; 1629 return 4; 1630 case AsmToken::Minus: 1631 Kind = MCBinaryExpr::Sub; 1632 return 4; 1633 1634 // High Intermediate Precedence: |, !, &, ^ 1635 // 1636 case AsmToken::Pipe: 1637 Kind = MCBinaryExpr::Or; 1638 return 5; 1639 case AsmToken::Exclaim: 1640 // Hack to support ARM compatible aliases (implied 'sp' operand in 'srs*' 1641 // instructions like 'srsda #31!') and not parse ! as an infix operator. 1642 if (MAI.getCommentString() == "@") 1643 return 0; 1644 Kind = MCBinaryExpr::OrNot; 1645 return 5; 1646 case AsmToken::Caret: 1647 Kind = MCBinaryExpr::Xor; 1648 return 5; 1649 case AsmToken::Amp: 1650 Kind = MCBinaryExpr::And; 1651 return 5; 1652 1653 // Highest Precedence: *, /, %, <<, >> 1654 case AsmToken::Star: 1655 Kind = MCBinaryExpr::Mul; 1656 return 6; 1657 case AsmToken::Slash: 1658 Kind = MCBinaryExpr::Div; 1659 return 6; 1660 case AsmToken::Percent: 1661 Kind = MCBinaryExpr::Mod; 1662 return 6; 1663 case AsmToken::LessLess: 1664 Kind = MCBinaryExpr::Shl; 1665 return 6; 1666 case AsmToken::GreaterGreater: 1667 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1668 return 6; 1669 } 1670 } 1671 1672 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1673 MCBinaryExpr::Opcode &Kind) { 1674 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1675 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1676 : getGNUBinOpPrecedence(MAI, K, Kind, ShouldUseLogicalShr); 1677 } 1678 1679 /// Parse all binary operators with precedence >= 'Precedence'. 1680 /// Res contains the LHS of the expression on input. 1681 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1682 SMLoc &EndLoc) { 1683 SMLoc StartLoc = Lexer.getLoc(); 1684 while (true) { 1685 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1686 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1687 1688 // If the next token is lower precedence than we are allowed to eat, return 1689 // successfully with what we ate already. 1690 if (TokPrec < Precedence) 1691 return false; 1692 1693 Lex(); 1694 1695 // Eat the next primary expression. 1696 const MCExpr *RHS; 1697 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1698 return true; 1699 1700 // If BinOp binds less tightly with RHS than the operator after RHS, let 1701 // the pending operator take RHS as its LHS. 1702 MCBinaryExpr::Opcode Dummy; 1703 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1704 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1705 return true; 1706 1707 // Merge LHS and RHS according to operator. 1708 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1709 } 1710 } 1711 1712 /// ParseStatement: 1713 /// ::= EndOfStatement 1714 /// ::= Label* Directive ...Operands... EndOfStatement 1715 /// ::= Label* Identifier OperandList* EndOfStatement 1716 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1717 MCAsmParserSemaCallback *SI) { 1718 assert(!hasPendingError() && "parseStatement started with pending error"); 1719 // Eat initial spaces and comments 1720 while (Lexer.is(AsmToken::Space)) 1721 Lex(); 1722 if (Lexer.is(AsmToken::EndOfStatement)) { 1723 // if this is a line comment we can drop it safely 1724 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1725 getTok().getString().front() == '\n') 1726 Out.AddBlankLine(); 1727 Lex(); 1728 return false; 1729 } 1730 // Statements always start with an identifier. 1731 AsmToken ID = getTok(); 1732 SMLoc IDLoc = ID.getLoc(); 1733 StringRef IDVal; 1734 int64_t LocalLabelVal = -1; 1735 StartTokLoc = ID.getLoc(); 1736 if (Lexer.is(AsmToken::HashDirective)) 1737 return parseCppHashLineFilenameComment(IDLoc, 1738 !isInsideMacroInstantiation()); 1739 1740 // Allow an integer followed by a ':' as a directional local label. 1741 if (Lexer.is(AsmToken::Integer)) { 1742 LocalLabelVal = getTok().getIntVal(); 1743 if (LocalLabelVal < 0) { 1744 if (!TheCondState.Ignore) { 1745 Lex(); // always eat a token 1746 return Error(IDLoc, "unexpected token at start of statement"); 1747 } 1748 IDVal = ""; 1749 } else { 1750 IDVal = getTok().getString(); 1751 Lex(); // Consume the integer token to be used as an identifier token. 1752 if (Lexer.getKind() != AsmToken::Colon) { 1753 if (!TheCondState.Ignore) { 1754 Lex(); // always eat a token 1755 return Error(IDLoc, "unexpected token at start of statement"); 1756 } 1757 } 1758 } 1759 } else if (Lexer.is(AsmToken::Dot)) { 1760 // Treat '.' as a valid identifier in this context. 1761 Lex(); 1762 IDVal = "."; 1763 } else if (Lexer.is(AsmToken::LCurly)) { 1764 // Treat '{' as a valid identifier in this context. 1765 Lex(); 1766 IDVal = "{"; 1767 1768 } else if (Lexer.is(AsmToken::RCurly)) { 1769 // Treat '}' as a valid identifier in this context. 1770 Lex(); 1771 IDVal = "}"; 1772 } else if (Lexer.is(AsmToken::Star) && 1773 getTargetParser().starIsStartOfStatement()) { 1774 // Accept '*' as a valid start of statement. 1775 Lex(); 1776 IDVal = "*"; 1777 } else if (parseIdentifier(IDVal)) { 1778 if (!TheCondState.Ignore) { 1779 Lex(); // always eat a token 1780 return Error(IDLoc, "unexpected token at start of statement"); 1781 } 1782 IDVal = ""; 1783 } 1784 1785 // Handle conditional assembly here before checking for skipping. We 1786 // have to do this so that .endif isn't skipped in a ".if 0" block for 1787 // example. 1788 StringMap<DirectiveKind>::const_iterator DirKindIt = 1789 DirectiveKindMap.find(IDVal.lower()); 1790 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1791 ? DK_NO_DIRECTIVE 1792 : DirKindIt->getValue(); 1793 switch (DirKind) { 1794 default: 1795 break; 1796 case DK_IF: 1797 case DK_IFEQ: 1798 case DK_IFGE: 1799 case DK_IFGT: 1800 case DK_IFLE: 1801 case DK_IFLT: 1802 case DK_IFNE: 1803 return parseDirectiveIf(IDLoc, DirKind); 1804 case DK_IFB: 1805 return parseDirectiveIfb(IDLoc, true); 1806 case DK_IFNB: 1807 return parseDirectiveIfb(IDLoc, false); 1808 case DK_IFC: 1809 return parseDirectiveIfc(IDLoc, true); 1810 case DK_IFEQS: 1811 return parseDirectiveIfeqs(IDLoc, true); 1812 case DK_IFNC: 1813 return parseDirectiveIfc(IDLoc, false); 1814 case DK_IFNES: 1815 return parseDirectiveIfeqs(IDLoc, false); 1816 case DK_IFDEF: 1817 return parseDirectiveIfdef(IDLoc, true); 1818 case DK_IFNDEF: 1819 case DK_IFNOTDEF: 1820 return parseDirectiveIfdef(IDLoc, false); 1821 case DK_ELSEIF: 1822 return parseDirectiveElseIf(IDLoc); 1823 case DK_ELSE: 1824 return parseDirectiveElse(IDLoc); 1825 case DK_ENDIF: 1826 return parseDirectiveEndIf(IDLoc); 1827 } 1828 1829 // Ignore the statement if in the middle of inactive conditional 1830 // (e.g. ".if 0"). 1831 if (TheCondState.Ignore) { 1832 eatToEndOfStatement(); 1833 return false; 1834 } 1835 1836 // FIXME: Recurse on local labels? 1837 1838 // See what kind of statement we have. 1839 switch (Lexer.getKind()) { 1840 case AsmToken::Colon: { 1841 if (!getTargetParser().isLabel(ID)) 1842 break; 1843 if (checkForValidSection()) 1844 return true; 1845 1846 // identifier ':' -> Label. 1847 Lex(); 1848 1849 // Diagnose attempt to use '.' as a label. 1850 if (IDVal == ".") 1851 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1852 1853 // Diagnose attempt to use a variable as a label. 1854 // 1855 // FIXME: Diagnostics. Note the location of the definition as a label. 1856 // FIXME: This doesn't diagnose assignment to a symbol which has been 1857 // implicitly marked as external. 1858 MCSymbol *Sym; 1859 if (LocalLabelVal == -1) { 1860 if (ParsingMSInlineAsm && SI) { 1861 StringRef RewrittenLabel = 1862 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1863 assert(!RewrittenLabel.empty() && 1864 "We should have an internal name here."); 1865 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1866 RewrittenLabel); 1867 IDVal = RewrittenLabel; 1868 } 1869 Sym = getContext().getOrCreateSymbol(IDVal); 1870 } else 1871 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1872 // End of Labels should be treated as end of line for lexing 1873 // purposes but that information is not available to the Lexer who 1874 // does not understand Labels. This may cause us to see a Hash 1875 // here instead of a preprocessor line comment. 1876 if (getTok().is(AsmToken::Hash)) { 1877 StringRef CommentStr = parseStringToEndOfStatement(); 1878 Lexer.Lex(); 1879 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1880 } 1881 1882 // Consume any end of statement token, if present, to avoid spurious 1883 // AddBlankLine calls(). 1884 if (getTok().is(AsmToken::EndOfStatement)) { 1885 Lex(); 1886 } 1887 1888 if (discardLTOSymbol(IDVal)) 1889 return false; 1890 1891 getTargetParser().doBeforeLabelEmit(Sym); 1892 1893 // Emit the label. 1894 if (!getTargetParser().isParsingMSInlineAsm()) 1895 Out.emitLabel(Sym, IDLoc); 1896 1897 // If we are generating dwarf for assembly source files then gather the 1898 // info to make a dwarf label entry for this label if needed. 1899 if (enabledGenDwarfForAssembly()) 1900 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1901 IDLoc); 1902 1903 getTargetParser().onLabelParsed(Sym); 1904 1905 return false; 1906 } 1907 1908 case AsmToken::Equal: 1909 if (!getTargetParser().equalIsAsmAssignment()) 1910 break; 1911 // identifier '=' ... -> assignment statement 1912 Lex(); 1913 1914 return parseAssignment(IDVal, true); 1915 1916 default: // Normal instruction or directive. 1917 break; 1918 } 1919 1920 // If macros are enabled, check to see if this is a macro instantiation. 1921 if (areMacrosEnabled()) 1922 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1923 return handleMacroEntry(M, IDLoc); 1924 } 1925 1926 // Otherwise, we have a normal instruction or directive. 1927 1928 // Directives start with "." 1929 if (IDVal.startswith(".") && IDVal != ".") { 1930 // There are several entities interested in parsing directives: 1931 // 1932 // 1. The target-specific assembly parser. Some directives are target 1933 // specific or may potentially behave differently on certain targets. 1934 // 2. Asm parser extensions. For example, platform-specific parsers 1935 // (like the ELF parser) register themselves as extensions. 1936 // 3. The generic directive parser implemented by this class. These are 1937 // all the directives that behave in a target and platform independent 1938 // manner, or at least have a default behavior that's shared between 1939 // all targets and platforms. 1940 1941 getTargetParser().flushPendingInstructions(getStreamer()); 1942 1943 SMLoc StartTokLoc = getTok().getLoc(); 1944 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1945 1946 if (hasPendingError()) 1947 return true; 1948 // Currently the return value should be true if we are 1949 // uninterested but as this is at odds with the standard parsing 1950 // convention (return true = error) we have instances of a parsed 1951 // directive that fails returning true as an error. Catch these 1952 // cases as best as possible errors here. 1953 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1954 return true; 1955 // Return if we did some parsing or believe we succeeded. 1956 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1957 return false; 1958 1959 // Next, check the extension directive map to see if any extension has 1960 // registered itself to parse this directive. 1961 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1962 ExtensionDirectiveMap.lookup(IDVal); 1963 if (Handler.first) 1964 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1965 1966 // Finally, if no one else is interested in this directive, it must be 1967 // generic and familiar to this class. 1968 switch (DirKind) { 1969 default: 1970 break; 1971 case DK_SET: 1972 case DK_EQU: 1973 return parseDirectiveSet(IDVal, true); 1974 case DK_EQUIV: 1975 return parseDirectiveSet(IDVal, false); 1976 case DK_ASCII: 1977 return parseDirectiveAscii(IDVal, false); 1978 case DK_ASCIZ: 1979 case DK_STRING: 1980 return parseDirectiveAscii(IDVal, true); 1981 case DK_BYTE: 1982 case DK_DC_B: 1983 return parseDirectiveValue(IDVal, 1); 1984 case DK_DC: 1985 case DK_DC_W: 1986 case DK_SHORT: 1987 case DK_VALUE: 1988 case DK_2BYTE: 1989 return parseDirectiveValue(IDVal, 2); 1990 case DK_LONG: 1991 case DK_INT: 1992 case DK_4BYTE: 1993 case DK_DC_L: 1994 return parseDirectiveValue(IDVal, 4); 1995 case DK_QUAD: 1996 case DK_8BYTE: 1997 return parseDirectiveValue(IDVal, 8); 1998 case DK_DC_A: 1999 return parseDirectiveValue( 2000 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 2001 case DK_OCTA: 2002 return parseDirectiveOctaValue(IDVal); 2003 case DK_SINGLE: 2004 case DK_FLOAT: 2005 case DK_DC_S: 2006 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 2007 case DK_DOUBLE: 2008 case DK_DC_D: 2009 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 2010 case DK_ALIGN: { 2011 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 2012 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 2013 } 2014 case DK_ALIGN32: { 2015 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 2016 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 2017 } 2018 case DK_BALIGN: 2019 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 2020 case DK_BALIGNW: 2021 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 2022 case DK_BALIGNL: 2023 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 2024 case DK_P2ALIGN: 2025 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 2026 case DK_P2ALIGNW: 2027 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 2028 case DK_P2ALIGNL: 2029 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 2030 case DK_ORG: 2031 return parseDirectiveOrg(); 2032 case DK_FILL: 2033 return parseDirectiveFill(); 2034 case DK_ZERO: 2035 return parseDirectiveZero(); 2036 case DK_EXTERN: 2037 eatToEndOfStatement(); // .extern is the default, ignore it. 2038 return false; 2039 case DK_GLOBL: 2040 case DK_GLOBAL: 2041 return parseDirectiveSymbolAttribute(MCSA_Global); 2042 case DK_LAZY_REFERENCE: 2043 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 2044 case DK_NO_DEAD_STRIP: 2045 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 2046 case DK_SYMBOL_RESOLVER: 2047 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 2048 case DK_PRIVATE_EXTERN: 2049 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 2050 case DK_REFERENCE: 2051 return parseDirectiveSymbolAttribute(MCSA_Reference); 2052 case DK_WEAK_DEFINITION: 2053 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 2054 case DK_WEAK_REFERENCE: 2055 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 2056 case DK_WEAK_DEF_CAN_BE_HIDDEN: 2057 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 2058 case DK_COLD: 2059 return parseDirectiveSymbolAttribute(MCSA_Cold); 2060 case DK_COMM: 2061 case DK_COMMON: 2062 return parseDirectiveComm(/*IsLocal=*/false); 2063 case DK_LCOMM: 2064 return parseDirectiveComm(/*IsLocal=*/true); 2065 case DK_ABORT: 2066 return parseDirectiveAbort(); 2067 case DK_INCLUDE: 2068 return parseDirectiveInclude(); 2069 case DK_INCBIN: 2070 return parseDirectiveIncbin(); 2071 case DK_CODE16: 2072 case DK_CODE16GCC: 2073 return TokError(Twine(IDVal) + 2074 " not currently supported for this target"); 2075 case DK_REPT: 2076 return parseDirectiveRept(IDLoc, IDVal); 2077 case DK_IRP: 2078 return parseDirectiveIrp(IDLoc); 2079 case DK_IRPC: 2080 return parseDirectiveIrpc(IDLoc); 2081 case DK_ENDR: 2082 return parseDirectiveEndr(IDLoc); 2083 case DK_BUNDLE_ALIGN_MODE: 2084 return parseDirectiveBundleAlignMode(); 2085 case DK_BUNDLE_LOCK: 2086 return parseDirectiveBundleLock(); 2087 case DK_BUNDLE_UNLOCK: 2088 return parseDirectiveBundleUnlock(); 2089 case DK_SLEB128: 2090 return parseDirectiveLEB128(true); 2091 case DK_ULEB128: 2092 return parseDirectiveLEB128(false); 2093 case DK_SPACE: 2094 case DK_SKIP: 2095 return parseDirectiveSpace(IDVal); 2096 case DK_FILE: 2097 return parseDirectiveFile(IDLoc); 2098 case DK_LINE: 2099 return parseDirectiveLine(); 2100 case DK_LOC: 2101 return parseDirectiveLoc(); 2102 case DK_STABS: 2103 return parseDirectiveStabs(); 2104 case DK_CV_FILE: 2105 return parseDirectiveCVFile(); 2106 case DK_CV_FUNC_ID: 2107 return parseDirectiveCVFuncId(); 2108 case DK_CV_INLINE_SITE_ID: 2109 return parseDirectiveCVInlineSiteId(); 2110 case DK_CV_LOC: 2111 return parseDirectiveCVLoc(); 2112 case DK_CV_LINETABLE: 2113 return parseDirectiveCVLinetable(); 2114 case DK_CV_INLINE_LINETABLE: 2115 return parseDirectiveCVInlineLinetable(); 2116 case DK_CV_DEF_RANGE: 2117 return parseDirectiveCVDefRange(); 2118 case DK_CV_STRING: 2119 return parseDirectiveCVString(); 2120 case DK_CV_STRINGTABLE: 2121 return parseDirectiveCVStringTable(); 2122 case DK_CV_FILECHECKSUMS: 2123 return parseDirectiveCVFileChecksums(); 2124 case DK_CV_FILECHECKSUM_OFFSET: 2125 return parseDirectiveCVFileChecksumOffset(); 2126 case DK_CV_FPO_DATA: 2127 return parseDirectiveCVFPOData(); 2128 case DK_CFI_SECTIONS: 2129 return parseDirectiveCFISections(); 2130 case DK_CFI_STARTPROC: 2131 return parseDirectiveCFIStartProc(); 2132 case DK_CFI_ENDPROC: 2133 return parseDirectiveCFIEndProc(); 2134 case DK_CFI_DEF_CFA: 2135 return parseDirectiveCFIDefCfa(IDLoc); 2136 case DK_CFI_DEF_CFA_OFFSET: 2137 return parseDirectiveCFIDefCfaOffset(); 2138 case DK_CFI_ADJUST_CFA_OFFSET: 2139 return parseDirectiveCFIAdjustCfaOffset(); 2140 case DK_CFI_DEF_CFA_REGISTER: 2141 return parseDirectiveCFIDefCfaRegister(IDLoc); 2142 case DK_CFI_OFFSET: 2143 return parseDirectiveCFIOffset(IDLoc); 2144 case DK_CFI_REL_OFFSET: 2145 return parseDirectiveCFIRelOffset(IDLoc); 2146 case DK_CFI_PERSONALITY: 2147 return parseDirectiveCFIPersonalityOrLsda(true); 2148 case DK_CFI_LSDA: 2149 return parseDirectiveCFIPersonalityOrLsda(false); 2150 case DK_CFI_REMEMBER_STATE: 2151 return parseDirectiveCFIRememberState(); 2152 case DK_CFI_RESTORE_STATE: 2153 return parseDirectiveCFIRestoreState(); 2154 case DK_CFI_SAME_VALUE: 2155 return parseDirectiveCFISameValue(IDLoc); 2156 case DK_CFI_RESTORE: 2157 return parseDirectiveCFIRestore(IDLoc); 2158 case DK_CFI_ESCAPE: 2159 return parseDirectiveCFIEscape(); 2160 case DK_CFI_RETURN_COLUMN: 2161 return parseDirectiveCFIReturnColumn(IDLoc); 2162 case DK_CFI_SIGNAL_FRAME: 2163 return parseDirectiveCFISignalFrame(); 2164 case DK_CFI_UNDEFINED: 2165 return parseDirectiveCFIUndefined(IDLoc); 2166 case DK_CFI_REGISTER: 2167 return parseDirectiveCFIRegister(IDLoc); 2168 case DK_CFI_WINDOW_SAVE: 2169 return parseDirectiveCFIWindowSave(); 2170 case DK_MACROS_ON: 2171 case DK_MACROS_OFF: 2172 return parseDirectiveMacrosOnOff(IDVal); 2173 case DK_MACRO: 2174 return parseDirectiveMacro(IDLoc); 2175 case DK_ALTMACRO: 2176 case DK_NOALTMACRO: 2177 return parseDirectiveAltmacro(IDVal); 2178 case DK_EXITM: 2179 return parseDirectiveExitMacro(IDVal); 2180 case DK_ENDM: 2181 case DK_ENDMACRO: 2182 return parseDirectiveEndMacro(IDVal); 2183 case DK_PURGEM: 2184 return parseDirectivePurgeMacro(IDLoc); 2185 case DK_END: 2186 return parseDirectiveEnd(IDLoc); 2187 case DK_ERR: 2188 return parseDirectiveError(IDLoc, false); 2189 case DK_ERROR: 2190 return parseDirectiveError(IDLoc, true); 2191 case DK_WARNING: 2192 return parseDirectiveWarning(IDLoc); 2193 case DK_RELOC: 2194 return parseDirectiveReloc(IDLoc); 2195 case DK_DCB: 2196 case DK_DCB_W: 2197 return parseDirectiveDCB(IDVal, 2); 2198 case DK_DCB_B: 2199 return parseDirectiveDCB(IDVal, 1); 2200 case DK_DCB_D: 2201 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 2202 case DK_DCB_L: 2203 return parseDirectiveDCB(IDVal, 4); 2204 case DK_DCB_S: 2205 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 2206 case DK_DC_X: 2207 case DK_DCB_X: 2208 return TokError(Twine(IDVal) + 2209 " not currently supported for this target"); 2210 case DK_DS: 2211 case DK_DS_W: 2212 return parseDirectiveDS(IDVal, 2); 2213 case DK_DS_B: 2214 return parseDirectiveDS(IDVal, 1); 2215 case DK_DS_D: 2216 return parseDirectiveDS(IDVal, 8); 2217 case DK_DS_L: 2218 case DK_DS_S: 2219 return parseDirectiveDS(IDVal, 4); 2220 case DK_DS_P: 2221 case DK_DS_X: 2222 return parseDirectiveDS(IDVal, 12); 2223 case DK_PRINT: 2224 return parseDirectivePrint(IDLoc); 2225 case DK_ADDRSIG: 2226 return parseDirectiveAddrsig(); 2227 case DK_ADDRSIG_SYM: 2228 return parseDirectiveAddrsigSym(); 2229 case DK_PSEUDO_PROBE: 2230 return parseDirectivePseudoProbe(); 2231 case DK_LTO_DISCARD: 2232 return parseDirectiveLTODiscard(); 2233 } 2234 2235 return Error(IDLoc, "unknown directive"); 2236 } 2237 2238 // __asm _emit or __asm __emit 2239 if (ParsingMSInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2240 IDVal == "_EMIT" || IDVal == "__EMIT")) 2241 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2242 2243 // __asm align 2244 if (ParsingMSInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2245 return parseDirectiveMSAlign(IDLoc, Info); 2246 2247 if (ParsingMSInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2248 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2249 if (checkForValidSection()) 2250 return true; 2251 2252 // Canonicalize the opcode to lower case. 2253 std::string OpcodeStr = IDVal.lower(); 2254 ParseInstructionInfo IInfo(Info.AsmRewrites); 2255 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2256 Info.ParsedOperands); 2257 Info.ParseError = ParseHadError; 2258 2259 // Dump the parsed representation, if requested. 2260 if (getShowParsedOperands()) { 2261 SmallString<256> Str; 2262 raw_svector_ostream OS(Str); 2263 OS << "parsed instruction: ["; 2264 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2265 if (i != 0) 2266 OS << ", "; 2267 Info.ParsedOperands[i]->print(OS); 2268 } 2269 OS << "]"; 2270 2271 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2272 } 2273 2274 // Fail even if ParseInstruction erroneously returns false. 2275 if (hasPendingError() || ParseHadError) 2276 return true; 2277 2278 // If we are generating dwarf for the current section then generate a .loc 2279 // directive for the instruction. 2280 if (!ParseHadError && enabledGenDwarfForAssembly() && 2281 getContext().getGenDwarfSectionSyms().count( 2282 getStreamer().getCurrentSectionOnly())) { 2283 unsigned Line; 2284 if (ActiveMacros.empty()) 2285 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2286 else 2287 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2288 ActiveMacros.front()->ExitBuffer); 2289 2290 // If we previously parsed a cpp hash file line comment then make sure the 2291 // current Dwarf File is for the CppHashFilename if not then emit the 2292 // Dwarf File table for it and adjust the line number for the .loc. 2293 if (!CppHashInfo.Filename.empty()) { 2294 unsigned FileNumber = getStreamer().emitDwarfFileDirective( 2295 0, StringRef(), CppHashInfo.Filename); 2296 getContext().setGenDwarfFileNumber(FileNumber); 2297 2298 unsigned CppHashLocLineNo = 2299 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2300 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2301 } 2302 2303 getStreamer().emitDwarfLocDirective( 2304 getContext().getGenDwarfFileNumber(), Line, 0, 2305 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2306 StringRef()); 2307 } 2308 2309 // If parsing succeeded, match the instruction. 2310 if (!ParseHadError) { 2311 uint64_t ErrorInfo; 2312 if (getTargetParser().MatchAndEmitInstruction( 2313 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2314 getTargetParser().isParsingMSInlineAsm())) 2315 return true; 2316 } 2317 return false; 2318 } 2319 2320 // Parse and erase curly braces marking block start/end 2321 bool 2322 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2323 // Identify curly brace marking block start/end 2324 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2325 return false; 2326 2327 SMLoc StartLoc = Lexer.getLoc(); 2328 Lex(); // Eat the brace 2329 if (Lexer.is(AsmToken::EndOfStatement)) 2330 Lex(); // Eat EndOfStatement following the brace 2331 2332 // Erase the block start/end brace from the output asm string 2333 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2334 StartLoc.getPointer()); 2335 return true; 2336 } 2337 2338 /// parseCppHashLineFilenameComment as this: 2339 /// ::= # number "filename" 2340 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L, bool SaveLocInfo) { 2341 Lex(); // Eat the hash token. 2342 // Lexer only ever emits HashDirective if it fully formed if it's 2343 // done the checking already so this is an internal error. 2344 assert(getTok().is(AsmToken::Integer) && 2345 "Lexing Cpp line comment: Expected Integer"); 2346 int64_t LineNumber = getTok().getIntVal(); 2347 Lex(); 2348 assert(getTok().is(AsmToken::String) && 2349 "Lexing Cpp line comment: Expected String"); 2350 StringRef Filename = getTok().getString(); 2351 Lex(); 2352 2353 if (!SaveLocInfo) 2354 return false; 2355 2356 // Get rid of the enclosing quotes. 2357 Filename = Filename.substr(1, Filename.size() - 2); 2358 2359 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2360 // and possibly DWARF file info. 2361 CppHashInfo.Loc = L; 2362 CppHashInfo.Filename = Filename; 2363 CppHashInfo.LineNumber = LineNumber; 2364 CppHashInfo.Buf = CurBuffer; 2365 if (FirstCppHashFilename.empty()) 2366 FirstCppHashFilename = Filename; 2367 return false; 2368 } 2369 2370 /// will use the last parsed cpp hash line filename comment 2371 /// for the Filename and LineNo if any in the diagnostic. 2372 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2373 auto *Parser = static_cast<AsmParser *>(Context); 2374 raw_ostream &OS = errs(); 2375 2376 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2377 SMLoc DiagLoc = Diag.getLoc(); 2378 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2379 unsigned CppHashBuf = 2380 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2381 2382 // Like SourceMgr::printMessage() we need to print the include stack if any 2383 // before printing the message. 2384 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2385 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2386 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2387 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2388 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2389 } 2390 2391 // If we have not parsed a cpp hash line filename comment or the source 2392 // manager changed or buffer changed (like in a nested include) then just 2393 // print the normal diagnostic using its Filename and LineNo. 2394 if (!Parser->CppHashInfo.LineNumber || DiagBuf != CppHashBuf) { 2395 if (Parser->SavedDiagHandler) 2396 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2397 else 2398 Parser->getContext().diagnose(Diag); 2399 return; 2400 } 2401 2402 // Use the CppHashFilename and calculate a line number based on the 2403 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2404 // for the diagnostic. 2405 const std::string &Filename = std::string(Parser->CppHashInfo.Filename); 2406 2407 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2408 int CppHashLocLineNo = 2409 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2410 int LineNo = 2411 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2412 2413 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2414 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2415 Diag.getLineContents(), Diag.getRanges()); 2416 2417 if (Parser->SavedDiagHandler) 2418 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2419 else 2420 Parser->getContext().diagnose(NewDiag); 2421 } 2422 2423 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2424 // difference being that that function accepts '@' as part of identifiers and 2425 // we can't do that. AsmLexer.cpp should probably be changed to handle 2426 // '@' as a special case when needed. 2427 static bool isIdentifierChar(char c) { 2428 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2429 c == '.'; 2430 } 2431 2432 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2433 ArrayRef<MCAsmMacroParameter> Parameters, 2434 ArrayRef<MCAsmMacroArgument> A, 2435 bool EnableAtPseudoVariable, SMLoc L) { 2436 unsigned NParameters = Parameters.size(); 2437 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2438 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2439 return Error(L, "Wrong number of arguments"); 2440 2441 // A macro without parameters is handled differently on Darwin: 2442 // gas accepts no arguments and does no substitutions 2443 while (!Body.empty()) { 2444 // Scan for the next substitution. 2445 std::size_t End = Body.size(), Pos = 0; 2446 for (; Pos != End; ++Pos) { 2447 // Check for a substitution or escape. 2448 if (IsDarwin && !NParameters) { 2449 // This macro has no parameters, look for $0, $1, etc. 2450 if (Body[Pos] != '$' || Pos + 1 == End) 2451 continue; 2452 2453 char Next = Body[Pos + 1]; 2454 if (Next == '$' || Next == 'n' || 2455 isdigit(static_cast<unsigned char>(Next))) 2456 break; 2457 } else { 2458 // This macro has parameters, look for \foo, \bar, etc. 2459 if (Body[Pos] == '\\' && Pos + 1 != End) 2460 break; 2461 } 2462 } 2463 2464 // Add the prefix. 2465 OS << Body.slice(0, Pos); 2466 2467 // Check if we reached the end. 2468 if (Pos == End) 2469 break; 2470 2471 if (IsDarwin && !NParameters) { 2472 switch (Body[Pos + 1]) { 2473 // $$ => $ 2474 case '$': 2475 OS << '$'; 2476 break; 2477 2478 // $n => number of arguments 2479 case 'n': 2480 OS << A.size(); 2481 break; 2482 2483 // $[0-9] => argument 2484 default: { 2485 // Missing arguments are ignored. 2486 unsigned Index = Body[Pos + 1] - '0'; 2487 if (Index >= A.size()) 2488 break; 2489 2490 // Otherwise substitute with the token values, with spaces eliminated. 2491 for (const AsmToken &Token : A[Index]) 2492 OS << Token.getString(); 2493 break; 2494 } 2495 } 2496 Pos += 2; 2497 } else { 2498 unsigned I = Pos + 1; 2499 2500 // Check for the \@ pseudo-variable. 2501 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2502 ++I; 2503 else 2504 while (isIdentifierChar(Body[I]) && I + 1 != End) 2505 ++I; 2506 2507 const char *Begin = Body.data() + Pos + 1; 2508 StringRef Argument(Begin, I - (Pos + 1)); 2509 unsigned Index = 0; 2510 2511 if (Argument == "@") { 2512 OS << NumOfMacroInstantiations; 2513 Pos += 2; 2514 } else { 2515 for (; Index < NParameters; ++Index) 2516 if (Parameters[Index].Name == Argument) 2517 break; 2518 2519 if (Index == NParameters) { 2520 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2521 Pos += 3; 2522 else { 2523 OS << '\\' << Argument; 2524 Pos = I; 2525 } 2526 } else { 2527 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2528 for (const AsmToken &Token : A[Index]) 2529 // For altmacro mode, you can write '%expr'. 2530 // The prefix '%' evaluates the expression 'expr' 2531 // and uses the result as a string (e.g. replace %(1+2) with the 2532 // string "3"). 2533 // Here, we identify the integer token which is the result of the 2534 // absolute expression evaluation and replace it with its string 2535 // representation. 2536 if (AltMacroMode && Token.getString().front() == '%' && 2537 Token.is(AsmToken::Integer)) 2538 // Emit an integer value to the buffer. 2539 OS << Token.getIntVal(); 2540 // Only Token that was validated as a string and begins with '<' 2541 // is considered altMacroString!!! 2542 else if (AltMacroMode && Token.getString().front() == '<' && 2543 Token.is(AsmToken::String)) { 2544 OS << angleBracketString(Token.getStringContents()); 2545 } 2546 // We expect no quotes around the string's contents when 2547 // parsing for varargs. 2548 else if (Token.isNot(AsmToken::String) || VarargParameter) 2549 OS << Token.getString(); 2550 else 2551 OS << Token.getStringContents(); 2552 2553 Pos += 1 + Argument.size(); 2554 } 2555 } 2556 } 2557 // Update the scan point. 2558 Body = Body.substr(Pos); 2559 } 2560 2561 return false; 2562 } 2563 2564 static bool isOperator(AsmToken::TokenKind kind) { 2565 switch (kind) { 2566 default: 2567 return false; 2568 case AsmToken::Plus: 2569 case AsmToken::Minus: 2570 case AsmToken::Tilde: 2571 case AsmToken::Slash: 2572 case AsmToken::Star: 2573 case AsmToken::Dot: 2574 case AsmToken::Equal: 2575 case AsmToken::EqualEqual: 2576 case AsmToken::Pipe: 2577 case AsmToken::PipePipe: 2578 case AsmToken::Caret: 2579 case AsmToken::Amp: 2580 case AsmToken::AmpAmp: 2581 case AsmToken::Exclaim: 2582 case AsmToken::ExclaimEqual: 2583 case AsmToken::Less: 2584 case AsmToken::LessEqual: 2585 case AsmToken::LessLess: 2586 case AsmToken::LessGreater: 2587 case AsmToken::Greater: 2588 case AsmToken::GreaterEqual: 2589 case AsmToken::GreaterGreater: 2590 return true; 2591 } 2592 } 2593 2594 namespace { 2595 2596 class AsmLexerSkipSpaceRAII { 2597 public: 2598 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2599 Lexer.setSkipSpace(SkipSpace); 2600 } 2601 2602 ~AsmLexerSkipSpaceRAII() { 2603 Lexer.setSkipSpace(true); 2604 } 2605 2606 private: 2607 AsmLexer &Lexer; 2608 }; 2609 2610 } // end anonymous namespace 2611 2612 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2613 2614 if (Vararg) { 2615 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2616 StringRef Str = parseStringToEndOfStatement(); 2617 MA.emplace_back(AsmToken::String, Str); 2618 } 2619 return false; 2620 } 2621 2622 unsigned ParenLevel = 0; 2623 2624 // Darwin doesn't use spaces to delmit arguments. 2625 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2626 2627 bool SpaceEaten; 2628 2629 while (true) { 2630 SpaceEaten = false; 2631 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2632 return TokError("unexpected token in macro instantiation"); 2633 2634 if (ParenLevel == 0) { 2635 2636 if (Lexer.is(AsmToken::Comma)) 2637 break; 2638 2639 if (Lexer.is(AsmToken::Space)) { 2640 SpaceEaten = true; 2641 Lexer.Lex(); // Eat spaces 2642 } 2643 2644 // Spaces can delimit parameters, but could also be part an expression. 2645 // If the token after a space is an operator, add the token and the next 2646 // one into this argument 2647 if (!IsDarwin) { 2648 if (isOperator(Lexer.getKind())) { 2649 MA.push_back(getTok()); 2650 Lexer.Lex(); 2651 2652 // Whitespace after an operator can be ignored. 2653 if (Lexer.is(AsmToken::Space)) 2654 Lexer.Lex(); 2655 2656 continue; 2657 } 2658 } 2659 if (SpaceEaten) 2660 break; 2661 } 2662 2663 // handleMacroEntry relies on not advancing the lexer here 2664 // to be able to fill in the remaining default parameter values 2665 if (Lexer.is(AsmToken::EndOfStatement)) 2666 break; 2667 2668 // Adjust the current parentheses level. 2669 if (Lexer.is(AsmToken::LParen)) 2670 ++ParenLevel; 2671 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2672 --ParenLevel; 2673 2674 // Append the token to the current argument list. 2675 MA.push_back(getTok()); 2676 Lexer.Lex(); 2677 } 2678 2679 if (ParenLevel != 0) 2680 return TokError("unbalanced parentheses in macro argument"); 2681 return false; 2682 } 2683 2684 // Parse the macro instantiation arguments. 2685 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2686 MCAsmMacroArguments &A) { 2687 const unsigned NParameters = M ? M->Parameters.size() : 0; 2688 bool NamedParametersFound = false; 2689 SmallVector<SMLoc, 4> FALocs; 2690 2691 A.resize(NParameters); 2692 FALocs.resize(NParameters); 2693 2694 // Parse two kinds of macro invocations: 2695 // - macros defined without any parameters accept an arbitrary number of them 2696 // - macros defined with parameters accept at most that many of them 2697 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2698 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2699 ++Parameter) { 2700 SMLoc IDLoc = Lexer.getLoc(); 2701 MCAsmMacroParameter FA; 2702 2703 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2704 if (parseIdentifier(FA.Name)) 2705 return Error(IDLoc, "invalid argument identifier for formal argument"); 2706 2707 if (Lexer.isNot(AsmToken::Equal)) 2708 return TokError("expected '=' after formal parameter identifier"); 2709 2710 Lex(); 2711 2712 NamedParametersFound = true; 2713 } 2714 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2715 2716 if (NamedParametersFound && FA.Name.empty()) 2717 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2718 2719 SMLoc StrLoc = Lexer.getLoc(); 2720 SMLoc EndLoc; 2721 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2722 const MCExpr *AbsoluteExp; 2723 int64_t Value; 2724 /// Eat '%' 2725 Lex(); 2726 if (parseExpression(AbsoluteExp, EndLoc)) 2727 return false; 2728 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2729 getStreamer().getAssemblerPtr())) 2730 return Error(StrLoc, "expected absolute expression"); 2731 const char *StrChar = StrLoc.getPointer(); 2732 const char *EndChar = EndLoc.getPointer(); 2733 AsmToken newToken(AsmToken::Integer, 2734 StringRef(StrChar, EndChar - StrChar), Value); 2735 FA.Value.push_back(newToken); 2736 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2737 isAngleBracketString(StrLoc, EndLoc)) { 2738 const char *StrChar = StrLoc.getPointer(); 2739 const char *EndChar = EndLoc.getPointer(); 2740 jumpToLoc(EndLoc, CurBuffer); 2741 /// Eat from '<' to '>' 2742 Lex(); 2743 AsmToken newToken(AsmToken::String, 2744 StringRef(StrChar, EndChar - StrChar)); 2745 FA.Value.push_back(newToken); 2746 } else if(parseMacroArgument(FA.Value, Vararg)) 2747 return true; 2748 2749 unsigned PI = Parameter; 2750 if (!FA.Name.empty()) { 2751 unsigned FAI = 0; 2752 for (FAI = 0; FAI < NParameters; ++FAI) 2753 if (M->Parameters[FAI].Name == FA.Name) 2754 break; 2755 2756 if (FAI >= NParameters) { 2757 assert(M && "expected macro to be defined"); 2758 return Error(IDLoc, "parameter named '" + FA.Name + 2759 "' does not exist for macro '" + M->Name + "'"); 2760 } 2761 PI = FAI; 2762 } 2763 2764 if (!FA.Value.empty()) { 2765 if (A.size() <= PI) 2766 A.resize(PI + 1); 2767 A[PI] = FA.Value; 2768 2769 if (FALocs.size() <= PI) 2770 FALocs.resize(PI + 1); 2771 2772 FALocs[PI] = Lexer.getLoc(); 2773 } 2774 2775 // At the end of the statement, fill in remaining arguments that have 2776 // default values. If there aren't any, then the next argument is 2777 // required but missing 2778 if (Lexer.is(AsmToken::EndOfStatement)) { 2779 bool Failure = false; 2780 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2781 if (A[FAI].empty()) { 2782 if (M->Parameters[FAI].Required) { 2783 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2784 "missing value for required parameter " 2785 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2786 Failure = true; 2787 } 2788 2789 if (!M->Parameters[FAI].Value.empty()) 2790 A[FAI] = M->Parameters[FAI].Value; 2791 } 2792 } 2793 return Failure; 2794 } 2795 2796 if (Lexer.is(AsmToken::Comma)) 2797 Lex(); 2798 } 2799 2800 return TokError("too many positional arguments"); 2801 } 2802 2803 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2804 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2805 // eliminate this, although we should protect against infinite loops. 2806 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2807 if (ActiveMacros.size() == MaxNestingDepth) { 2808 std::ostringstream MaxNestingDepthError; 2809 MaxNestingDepthError << "macros cannot be nested more than " 2810 << MaxNestingDepth << " levels deep." 2811 << " Use -asm-macro-max-nesting-depth to increase " 2812 "this limit."; 2813 return TokError(MaxNestingDepthError.str()); 2814 } 2815 2816 MCAsmMacroArguments A; 2817 if (parseMacroArguments(M, A)) 2818 return true; 2819 2820 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2821 // to hold the macro body with substitutions. 2822 SmallString<256> Buf; 2823 StringRef Body = M->Body; 2824 raw_svector_ostream OS(Buf); 2825 2826 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2827 return true; 2828 2829 // We include the .endmacro in the buffer as our cue to exit the macro 2830 // instantiation. 2831 OS << ".endmacro\n"; 2832 2833 std::unique_ptr<MemoryBuffer> Instantiation = 2834 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2835 2836 // Create the macro instantiation object and add to the current macro 2837 // instantiation stack. 2838 MacroInstantiation *MI = new MacroInstantiation{ 2839 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 2840 ActiveMacros.push_back(MI); 2841 2842 ++NumOfMacroInstantiations; 2843 2844 // Jump to the macro instantiation and prime the lexer. 2845 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2846 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2847 Lex(); 2848 2849 return false; 2850 } 2851 2852 void AsmParser::handleMacroExit() { 2853 // Jump to the EndOfStatement we should return to, and consume it. 2854 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2855 Lex(); 2856 2857 // Pop the instantiation entry. 2858 delete ActiveMacros.back(); 2859 ActiveMacros.pop_back(); 2860 } 2861 2862 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2863 bool NoDeadStrip) { 2864 MCSymbol *Sym; 2865 const MCExpr *Value; 2866 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2867 Value)) 2868 return true; 2869 2870 if (!Sym) { 2871 // In the case where we parse an expression starting with a '.', we will 2872 // not generate an error, nor will we create a symbol. In this case we 2873 // should just return out. 2874 return false; 2875 } 2876 2877 if (discardLTOSymbol(Name)) 2878 return false; 2879 2880 // Do the assignment. 2881 Out.emitAssignment(Sym, Value); 2882 if (NoDeadStrip) 2883 Out.emitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2884 2885 return false; 2886 } 2887 2888 /// parseIdentifier: 2889 /// ::= identifier 2890 /// ::= string 2891 bool AsmParser::parseIdentifier(StringRef &Res) { 2892 // The assembler has relaxed rules for accepting identifiers, in particular we 2893 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2894 // separate tokens. At this level, we have already lexed so we cannot (currently) 2895 // handle this as a context dependent token, instead we detect adjacent tokens 2896 // and return the combined identifier. 2897 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2898 SMLoc PrefixLoc = getLexer().getLoc(); 2899 2900 // Consume the prefix character, and check for a following identifier. 2901 2902 AsmToken Buf[1]; 2903 Lexer.peekTokens(Buf, false); 2904 2905 if (Buf[0].isNot(AsmToken::Identifier) && Buf[0].isNot(AsmToken::Integer)) 2906 return true; 2907 2908 // We have a '$' or '@' followed by an identifier or integer token, make 2909 // sure they are adjacent. 2910 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2911 return true; 2912 2913 // eat $ or @ 2914 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2915 // Construct the joined identifier and consume the token. 2916 Res = StringRef(PrefixLoc.getPointer(), getTok().getString().size() + 1); 2917 Lex(); // Parser Lex to maintain invariants. 2918 return false; 2919 } 2920 2921 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2922 return true; 2923 2924 Res = getTok().getIdentifier(); 2925 2926 Lex(); // Consume the identifier token. 2927 2928 return false; 2929 } 2930 2931 /// parseDirectiveSet: 2932 /// ::= .equ identifier ',' expression 2933 /// ::= .equiv identifier ',' expression 2934 /// ::= .set identifier ',' expression 2935 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2936 StringRef Name; 2937 if (check(parseIdentifier(Name), "expected identifier") || 2938 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2939 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2940 return false; 2941 } 2942 2943 bool AsmParser::parseEscapedString(std::string &Data) { 2944 if (check(getTok().isNot(AsmToken::String), "expected string")) 2945 return true; 2946 2947 Data = ""; 2948 StringRef Str = getTok().getStringContents(); 2949 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2950 if (Str[i] != '\\') { 2951 Data += Str[i]; 2952 continue; 2953 } 2954 2955 // Recognize escaped characters. Note that this escape semantics currently 2956 // loosely follows Darwin 'as'. 2957 ++i; 2958 if (i == e) 2959 return TokError("unexpected backslash at end of string"); 2960 2961 // Recognize hex sequences similarly to GNU 'as'. 2962 if (Str[i] == 'x' || Str[i] == 'X') { 2963 size_t length = Str.size(); 2964 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2965 return TokError("invalid hexadecimal escape sequence"); 2966 2967 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2968 // then truncates to the lower 16 bits. Seems reasonable. 2969 unsigned Value = 0; 2970 while (i + 1 < length && isHexDigit(Str[i + 1])) 2971 Value = Value * 16 + hexDigitValue(Str[++i]); 2972 2973 Data += (unsigned char)(Value & 0xFF); 2974 continue; 2975 } 2976 2977 // Recognize octal sequences. 2978 if ((unsigned)(Str[i] - '0') <= 7) { 2979 // Consume up to three octal characters. 2980 unsigned Value = Str[i] - '0'; 2981 2982 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2983 ++i; 2984 Value = Value * 8 + (Str[i] - '0'); 2985 2986 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2987 ++i; 2988 Value = Value * 8 + (Str[i] - '0'); 2989 } 2990 } 2991 2992 if (Value > 255) 2993 return TokError("invalid octal escape sequence (out of range)"); 2994 2995 Data += (unsigned char)Value; 2996 continue; 2997 } 2998 2999 // Otherwise recognize individual escapes. 3000 switch (Str[i]) { 3001 default: 3002 // Just reject invalid escape sequences for now. 3003 return TokError("invalid escape sequence (unrecognized character)"); 3004 3005 case 'b': Data += '\b'; break; 3006 case 'f': Data += '\f'; break; 3007 case 'n': Data += '\n'; break; 3008 case 'r': Data += '\r'; break; 3009 case 't': Data += '\t'; break; 3010 case '"': Data += '"'; break; 3011 case '\\': Data += '\\'; break; 3012 } 3013 } 3014 3015 Lex(); 3016 return false; 3017 } 3018 3019 bool AsmParser::parseAngleBracketString(std::string &Data) { 3020 SMLoc EndLoc, StartLoc = getTok().getLoc(); 3021 if (isAngleBracketString(StartLoc, EndLoc)) { 3022 const char *StartChar = StartLoc.getPointer() + 1; 3023 const char *EndChar = EndLoc.getPointer() - 1; 3024 jumpToLoc(EndLoc, CurBuffer); 3025 /// Eat from '<' to '>' 3026 Lex(); 3027 3028 Data = angleBracketString(StringRef(StartChar, EndChar - StartChar)); 3029 return false; 3030 } 3031 return true; 3032 } 3033 3034 /// parseDirectiveAscii: 3035 // ::= .ascii [ "string"+ ( , "string"+ )* ] 3036 /// ::= ( .asciz | .string ) [ "string" ( , "string" )* ] 3037 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 3038 auto parseOp = [&]() -> bool { 3039 std::string Data; 3040 if (checkForValidSection()) 3041 return true; 3042 // Only support spaces as separators for .ascii directive for now. See the 3043 // discusssion at https://reviews.llvm.org/D91460 for more details. 3044 do { 3045 if (parseEscapedString(Data)) 3046 return true; 3047 getStreamer().emitBytes(Data); 3048 } while (!ZeroTerminated && getTok().is(AsmToken::String)); 3049 if (ZeroTerminated) 3050 getStreamer().emitBytes(StringRef("\0", 1)); 3051 return false; 3052 }; 3053 3054 if (parseMany(parseOp)) 3055 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3056 return false; 3057 } 3058 3059 /// parseDirectiveReloc 3060 /// ::= .reloc expression , identifier [ , expression ] 3061 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 3062 const MCExpr *Offset; 3063 const MCExpr *Expr = nullptr; 3064 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 3065 3066 if (parseExpression(Offset)) 3067 return true; 3068 if (parseToken(AsmToken::Comma, "expected comma") || 3069 check(getTok().isNot(AsmToken::Identifier), "expected relocation name")) 3070 return true; 3071 3072 SMLoc NameLoc = Lexer.getTok().getLoc(); 3073 StringRef Name = Lexer.getTok().getIdentifier(); 3074 Lex(); 3075 3076 if (Lexer.is(AsmToken::Comma)) { 3077 Lex(); 3078 SMLoc ExprLoc = Lexer.getLoc(); 3079 if (parseExpression(Expr)) 3080 return true; 3081 3082 MCValue Value; 3083 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 3084 return Error(ExprLoc, "expression must be relocatable"); 3085 } 3086 3087 if (parseEOL()) 3088 return true; 3089 3090 const MCTargetAsmParser &MCT = getTargetParser(); 3091 const MCSubtargetInfo &STI = MCT.getSTI(); 3092 if (Optional<std::pair<bool, std::string>> Err = 3093 getStreamer().emitRelocDirective(*Offset, Name, Expr, DirectiveLoc, 3094 STI)) 3095 return Error(Err->first ? NameLoc : OffsetLoc, Err->second); 3096 3097 return false; 3098 } 3099 3100 /// parseDirectiveValue 3101 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 3102 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 3103 auto parseOp = [&]() -> bool { 3104 const MCExpr *Value; 3105 SMLoc ExprLoc = getLexer().getLoc(); 3106 if (checkForValidSection() || parseExpression(Value)) 3107 return true; 3108 // Special case constant expressions to match code generator. 3109 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3110 assert(Size <= 8 && "Invalid size"); 3111 uint64_t IntValue = MCE->getValue(); 3112 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 3113 return Error(ExprLoc, "out of range literal value"); 3114 getStreamer().emitIntValue(IntValue, Size); 3115 } else 3116 getStreamer().emitValue(Value, Size, ExprLoc); 3117 return false; 3118 }; 3119 3120 if (parseMany(parseOp)) 3121 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3122 return false; 3123 } 3124 3125 static bool parseHexOcta(AsmParser &Asm, uint64_t &hi, uint64_t &lo) { 3126 if (Asm.getTok().isNot(AsmToken::Integer) && 3127 Asm.getTok().isNot(AsmToken::BigNum)) 3128 return Asm.TokError("unknown token in expression"); 3129 SMLoc ExprLoc = Asm.getTok().getLoc(); 3130 APInt IntValue = Asm.getTok().getAPIntVal(); 3131 Asm.Lex(); 3132 if (!IntValue.isIntN(128)) 3133 return Asm.Error(ExprLoc, "out of range literal value"); 3134 if (!IntValue.isIntN(64)) { 3135 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 3136 lo = IntValue.getLoBits(64).getZExtValue(); 3137 } else { 3138 hi = 0; 3139 lo = IntValue.getZExtValue(); 3140 } 3141 return false; 3142 } 3143 3144 /// ParseDirectiveOctaValue 3145 /// ::= .octa [ hexconstant (, hexconstant)* ] 3146 3147 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 3148 auto parseOp = [&]() -> bool { 3149 if (checkForValidSection()) 3150 return true; 3151 uint64_t hi, lo; 3152 if (parseHexOcta(*this, hi, lo)) 3153 return true; 3154 if (MAI.isLittleEndian()) { 3155 getStreamer().emitInt64(lo); 3156 getStreamer().emitInt64(hi); 3157 } else { 3158 getStreamer().emitInt64(hi); 3159 getStreamer().emitInt64(lo); 3160 } 3161 return false; 3162 }; 3163 3164 if (parseMany(parseOp)) 3165 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3166 return false; 3167 } 3168 3169 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 3170 // We don't truly support arithmetic on floating point expressions, so we 3171 // have to manually parse unary prefixes. 3172 bool IsNeg = false; 3173 if (getLexer().is(AsmToken::Minus)) { 3174 Lexer.Lex(); 3175 IsNeg = true; 3176 } else if (getLexer().is(AsmToken::Plus)) 3177 Lexer.Lex(); 3178 3179 if (Lexer.is(AsmToken::Error)) 3180 return TokError(Lexer.getErr()); 3181 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 3182 Lexer.isNot(AsmToken::Identifier)) 3183 return TokError("unexpected token in directive"); 3184 3185 // Convert to an APFloat. 3186 APFloat Value(Semantics); 3187 StringRef IDVal = getTok().getString(); 3188 if (getLexer().is(AsmToken::Identifier)) { 3189 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 3190 Value = APFloat::getInf(Semantics); 3191 else if (!IDVal.compare_lower("nan")) 3192 Value = APFloat::getNaN(Semantics, false, ~0); 3193 else 3194 return TokError("invalid floating point literal"); 3195 } else if (errorToBool( 3196 Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) 3197 .takeError())) 3198 return TokError("invalid floating point literal"); 3199 if (IsNeg) 3200 Value.changeSign(); 3201 3202 // Consume the numeric token. 3203 Lex(); 3204 3205 Res = Value.bitcastToAPInt(); 3206 3207 return false; 3208 } 3209 3210 /// parseDirectiveRealValue 3211 /// ::= (.single | .double) [ expression (, expression)* ] 3212 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 3213 const fltSemantics &Semantics) { 3214 auto parseOp = [&]() -> bool { 3215 APInt AsInt; 3216 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 3217 return true; 3218 getStreamer().emitIntValue(AsInt.getLimitedValue(), 3219 AsInt.getBitWidth() / 8); 3220 return false; 3221 }; 3222 3223 if (parseMany(parseOp)) 3224 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 3225 return false; 3226 } 3227 3228 /// parseDirectiveZero 3229 /// ::= .zero expression 3230 bool AsmParser::parseDirectiveZero() { 3231 SMLoc NumBytesLoc = Lexer.getLoc(); 3232 const MCExpr *NumBytes; 3233 if (checkForValidSection() || parseExpression(NumBytes)) 3234 return true; 3235 3236 int64_t Val = 0; 3237 if (getLexer().is(AsmToken::Comma)) { 3238 Lex(); 3239 if (parseAbsoluteExpression(Val)) 3240 return true; 3241 } 3242 3243 if (parseEOL()) 3244 return true; 3245 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3246 3247 return false; 3248 } 3249 3250 /// parseDirectiveFill 3251 /// ::= .fill expression [ , expression [ , expression ] ] 3252 bool AsmParser::parseDirectiveFill() { 3253 SMLoc NumValuesLoc = Lexer.getLoc(); 3254 const MCExpr *NumValues; 3255 if (checkForValidSection() || parseExpression(NumValues)) 3256 return true; 3257 3258 int64_t FillSize = 1; 3259 int64_t FillExpr = 0; 3260 3261 SMLoc SizeLoc, ExprLoc; 3262 3263 if (parseOptionalToken(AsmToken::Comma)) { 3264 SizeLoc = getTok().getLoc(); 3265 if (parseAbsoluteExpression(FillSize)) 3266 return true; 3267 if (parseOptionalToken(AsmToken::Comma)) { 3268 ExprLoc = getTok().getLoc(); 3269 if (parseAbsoluteExpression(FillExpr)) 3270 return true; 3271 } 3272 } 3273 if (parseEOL()) 3274 return true; 3275 3276 if (FillSize < 0) { 3277 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3278 return false; 3279 } 3280 if (FillSize > 8) { 3281 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3282 FillSize = 8; 3283 } 3284 3285 if (!isUInt<32>(FillExpr) && FillSize > 4) 3286 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3287 3288 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3289 3290 return false; 3291 } 3292 3293 /// parseDirectiveOrg 3294 /// ::= .org expression [ , expression ] 3295 bool AsmParser::parseDirectiveOrg() { 3296 const MCExpr *Offset; 3297 SMLoc OffsetLoc = Lexer.getLoc(); 3298 if (checkForValidSection() || parseExpression(Offset)) 3299 return true; 3300 3301 // Parse optional fill expression. 3302 int64_t FillExpr = 0; 3303 if (parseOptionalToken(AsmToken::Comma)) 3304 if (parseAbsoluteExpression(FillExpr)) 3305 return addErrorSuffix(" in '.org' directive"); 3306 if (parseEOL()) 3307 return addErrorSuffix(" in '.org' directive"); 3308 3309 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3310 return false; 3311 } 3312 3313 /// parseDirectiveAlign 3314 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3315 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3316 SMLoc AlignmentLoc = getLexer().getLoc(); 3317 int64_t Alignment; 3318 SMLoc MaxBytesLoc; 3319 bool HasFillExpr = false; 3320 int64_t FillExpr = 0; 3321 int64_t MaxBytesToFill = 0; 3322 3323 auto parseAlign = [&]() -> bool { 3324 if (parseAbsoluteExpression(Alignment)) 3325 return true; 3326 if (parseOptionalToken(AsmToken::Comma)) { 3327 // The fill expression can be omitted while specifying a maximum number of 3328 // alignment bytes, e.g: 3329 // .align 3,,4 3330 if (getTok().isNot(AsmToken::Comma)) { 3331 HasFillExpr = true; 3332 if (parseAbsoluteExpression(FillExpr)) 3333 return true; 3334 } 3335 if (parseOptionalToken(AsmToken::Comma)) 3336 if (parseTokenLoc(MaxBytesLoc) || 3337 parseAbsoluteExpression(MaxBytesToFill)) 3338 return true; 3339 } 3340 return parseEOL(); 3341 }; 3342 3343 if (checkForValidSection()) 3344 return addErrorSuffix(" in directive"); 3345 // Ignore empty '.p2align' directives for GNU-as compatibility 3346 if (IsPow2 && (ValueSize == 1) && getTok().is(AsmToken::EndOfStatement)) { 3347 Warning(AlignmentLoc, "p2align directive with no operand(s) is ignored"); 3348 return parseEOL(); 3349 } 3350 if (parseAlign()) 3351 return addErrorSuffix(" in directive"); 3352 3353 // Always emit an alignment here even if we thrown an error. 3354 bool ReturnVal = false; 3355 3356 // Compute alignment in bytes. 3357 if (IsPow2) { 3358 // FIXME: Diagnose overflow. 3359 if (Alignment >= 32) { 3360 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3361 Alignment = 31; 3362 } 3363 3364 Alignment = 1ULL << Alignment; 3365 } else { 3366 // Reject alignments that aren't either a power of two or zero, 3367 // for gas compatibility. Alignment of zero is silently rounded 3368 // up to one. 3369 if (Alignment == 0) 3370 Alignment = 1; 3371 if (!isPowerOf2_64(Alignment)) 3372 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3373 if (!isUInt<32>(Alignment)) 3374 ReturnVal |= Error(AlignmentLoc, "alignment must be smaller than 2**32"); 3375 } 3376 3377 // Diagnose non-sensical max bytes to align. 3378 if (MaxBytesLoc.isValid()) { 3379 if (MaxBytesToFill < 1) { 3380 ReturnVal |= Error(MaxBytesLoc, 3381 "alignment directive can never be satisfied in this " 3382 "many bytes, ignoring maximum bytes expression"); 3383 MaxBytesToFill = 0; 3384 } 3385 3386 if (MaxBytesToFill >= Alignment) { 3387 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3388 "has no effect"); 3389 MaxBytesToFill = 0; 3390 } 3391 } 3392 3393 // Check whether we should use optimal code alignment for this .align 3394 // directive. 3395 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3396 assert(Section && "must have section to emit alignment"); 3397 bool UseCodeAlign = Section->UseCodeAlign(); 3398 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3399 ValueSize == 1 && UseCodeAlign) { 3400 getStreamer().emitCodeAlignment(Alignment, MaxBytesToFill); 3401 } else { 3402 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3403 getStreamer().emitValueToAlignment(Alignment, FillExpr, ValueSize, 3404 MaxBytesToFill); 3405 } 3406 3407 return ReturnVal; 3408 } 3409 3410 /// parseDirectiveFile 3411 /// ::= .file filename 3412 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3413 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3414 // FIXME: I'm not sure what this is. 3415 int64_t FileNumber = -1; 3416 if (getLexer().is(AsmToken::Integer)) { 3417 FileNumber = getTok().getIntVal(); 3418 Lex(); 3419 3420 if (FileNumber < 0) 3421 return TokError("negative file number"); 3422 } 3423 3424 std::string Path; 3425 3426 // Usually the directory and filename together, otherwise just the directory. 3427 // Allow the strings to have escaped octal character sequence. 3428 if (check(getTok().isNot(AsmToken::String), 3429 "unexpected token in '.file' directive") || 3430 parseEscapedString(Path)) 3431 return true; 3432 3433 StringRef Directory; 3434 StringRef Filename; 3435 std::string FilenameData; 3436 if (getLexer().is(AsmToken::String)) { 3437 if (check(FileNumber == -1, 3438 "explicit path specified, but no file number") || 3439 parseEscapedString(FilenameData)) 3440 return true; 3441 Filename = FilenameData; 3442 Directory = Path; 3443 } else { 3444 Filename = Path; 3445 } 3446 3447 uint64_t MD5Hi, MD5Lo; 3448 bool HasMD5 = false; 3449 3450 Optional<StringRef> Source; 3451 bool HasSource = false; 3452 std::string SourceString; 3453 3454 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3455 StringRef Keyword; 3456 if (check(getTok().isNot(AsmToken::Identifier), 3457 "unexpected token in '.file' directive") || 3458 parseIdentifier(Keyword)) 3459 return true; 3460 if (Keyword == "md5") { 3461 HasMD5 = true; 3462 if (check(FileNumber == -1, 3463 "MD5 checksum specified, but no file number") || 3464 parseHexOcta(*this, MD5Hi, MD5Lo)) 3465 return true; 3466 } else if (Keyword == "source") { 3467 HasSource = true; 3468 if (check(FileNumber == -1, 3469 "source specified, but no file number") || 3470 check(getTok().isNot(AsmToken::String), 3471 "unexpected token in '.file' directive") || 3472 parseEscapedString(SourceString)) 3473 return true; 3474 } else { 3475 return TokError("unexpected token in '.file' directive"); 3476 } 3477 } 3478 3479 if (FileNumber == -1) { 3480 // Ignore the directive if there is no number and the target doesn't support 3481 // numberless .file directives. This allows some portability of assembler 3482 // between different object file formats. 3483 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3484 getStreamer().emitFileDirective(Filename); 3485 } else { 3486 // In case there is a -g option as well as debug info from directive .file, 3487 // we turn off the -g option, directly use the existing debug info instead. 3488 // Throw away any implicit file table for the assembler source. 3489 if (Ctx.getGenDwarfForAssembly()) { 3490 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3491 Ctx.setGenDwarfForAssembly(false); 3492 } 3493 3494 Optional<MD5::MD5Result> CKMem; 3495 if (HasMD5) { 3496 MD5::MD5Result Sum; 3497 for (unsigned i = 0; i != 8; ++i) { 3498 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3499 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3500 } 3501 CKMem = Sum; 3502 } 3503 if (HasSource) { 3504 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3505 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3506 Source = StringRef(SourceBuf, SourceString.size()); 3507 } 3508 if (FileNumber == 0) { 3509 // Upgrade to Version 5 for assembly actions like clang -c a.s. 3510 if (Ctx.getDwarfVersion() < 5) 3511 Ctx.setDwarfVersion(5); 3512 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3513 } else { 3514 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3515 FileNumber, Directory, Filename, CKMem, Source); 3516 if (!FileNumOrErr) 3517 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3518 } 3519 // Alert the user if there are some .file directives with MD5 and some not. 3520 // But only do that once. 3521 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3522 ReportedInconsistentMD5 = true; 3523 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3524 } 3525 } 3526 3527 return false; 3528 } 3529 3530 /// parseDirectiveLine 3531 /// ::= .line [number] 3532 bool AsmParser::parseDirectiveLine() { 3533 int64_t LineNumber; 3534 if (getLexer().is(AsmToken::Integer)) { 3535 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3536 return true; 3537 (void)LineNumber; 3538 // FIXME: Do something with the .line. 3539 } 3540 return parseEOL(); 3541 } 3542 3543 /// parseDirectiveLoc 3544 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3545 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3546 /// The first number is a file number, must have been previously assigned with 3547 /// a .file directive, the second number is the line number and optionally the 3548 /// third number is a column position (zero if not specified). The remaining 3549 /// optional items are .loc sub-directives. 3550 bool AsmParser::parseDirectiveLoc() { 3551 int64_t FileNumber = 0, LineNumber = 0; 3552 SMLoc Loc = getTok().getLoc(); 3553 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3554 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3555 "file number less than one in '.loc' directive") || 3556 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3557 "unassigned file number in '.loc' directive")) 3558 return true; 3559 3560 // optional 3561 if (getLexer().is(AsmToken::Integer)) { 3562 LineNumber = getTok().getIntVal(); 3563 if (LineNumber < 0) 3564 return TokError("line number less than zero in '.loc' directive"); 3565 Lex(); 3566 } 3567 3568 int64_t ColumnPos = 0; 3569 if (getLexer().is(AsmToken::Integer)) { 3570 ColumnPos = getTok().getIntVal(); 3571 if (ColumnPos < 0) 3572 return TokError("column position less than zero in '.loc' directive"); 3573 Lex(); 3574 } 3575 3576 auto PrevFlags = getContext().getCurrentDwarfLoc().getFlags(); 3577 unsigned Flags = PrevFlags & DWARF2_FLAG_IS_STMT; 3578 unsigned Isa = 0; 3579 int64_t Discriminator = 0; 3580 3581 auto parseLocOp = [&]() -> bool { 3582 StringRef Name; 3583 SMLoc Loc = getTok().getLoc(); 3584 if (parseIdentifier(Name)) 3585 return TokError("unexpected token in '.loc' directive"); 3586 3587 if (Name == "basic_block") 3588 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3589 else if (Name == "prologue_end") 3590 Flags |= DWARF2_FLAG_PROLOGUE_END; 3591 else if (Name == "epilogue_begin") 3592 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3593 else if (Name == "is_stmt") { 3594 Loc = getTok().getLoc(); 3595 const MCExpr *Value; 3596 if (parseExpression(Value)) 3597 return true; 3598 // The expression must be the constant 0 or 1. 3599 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3600 int Value = MCE->getValue(); 3601 if (Value == 0) 3602 Flags &= ~DWARF2_FLAG_IS_STMT; 3603 else if (Value == 1) 3604 Flags |= DWARF2_FLAG_IS_STMT; 3605 else 3606 return Error(Loc, "is_stmt value not 0 or 1"); 3607 } else { 3608 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3609 } 3610 } else if (Name == "isa") { 3611 Loc = getTok().getLoc(); 3612 const MCExpr *Value; 3613 if (parseExpression(Value)) 3614 return true; 3615 // The expression must be a constant greater or equal to 0. 3616 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3617 int Value = MCE->getValue(); 3618 if (Value < 0) 3619 return Error(Loc, "isa number less than zero"); 3620 Isa = Value; 3621 } else { 3622 return Error(Loc, "isa number not a constant value"); 3623 } 3624 } else if (Name == "discriminator") { 3625 if (parseAbsoluteExpression(Discriminator)) 3626 return true; 3627 } else { 3628 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3629 } 3630 return false; 3631 }; 3632 3633 if (parseMany(parseLocOp, false /*hasComma*/)) 3634 return true; 3635 3636 getStreamer().emitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3637 Isa, Discriminator, StringRef()); 3638 3639 return false; 3640 } 3641 3642 /// parseDirectiveStabs 3643 /// ::= .stabs string, number, number, number 3644 bool AsmParser::parseDirectiveStabs() { 3645 return TokError("unsupported directive '.stabs'"); 3646 } 3647 3648 /// parseDirectiveCVFile 3649 /// ::= .cv_file number filename [checksum] [checksumkind] 3650 bool AsmParser::parseDirectiveCVFile() { 3651 SMLoc FileNumberLoc = getTok().getLoc(); 3652 int64_t FileNumber; 3653 std::string Filename; 3654 std::string Checksum; 3655 int64_t ChecksumKind = 0; 3656 3657 if (parseIntToken(FileNumber, 3658 "expected file number in '.cv_file' directive") || 3659 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3660 check(getTok().isNot(AsmToken::String), 3661 "unexpected token in '.cv_file' directive") || 3662 parseEscapedString(Filename)) 3663 return true; 3664 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3665 if (check(getTok().isNot(AsmToken::String), 3666 "unexpected token in '.cv_file' directive") || 3667 parseEscapedString(Checksum) || 3668 parseIntToken(ChecksumKind, 3669 "expected checksum kind in '.cv_file' directive") || 3670 parseToken(AsmToken::EndOfStatement, 3671 "unexpected token in '.cv_file' directive")) 3672 return true; 3673 } 3674 3675 Checksum = fromHex(Checksum); 3676 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3677 memcpy(CKMem, Checksum.data(), Checksum.size()); 3678 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3679 Checksum.size()); 3680 3681 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3682 static_cast<uint8_t>(ChecksumKind))) 3683 return Error(FileNumberLoc, "file number already allocated"); 3684 3685 return false; 3686 } 3687 3688 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3689 StringRef DirectiveName) { 3690 SMLoc Loc; 3691 return parseTokenLoc(Loc) || 3692 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3693 "' directive") || 3694 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3695 "expected function id within range [0, UINT_MAX)"); 3696 } 3697 3698 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3699 SMLoc Loc; 3700 return parseTokenLoc(Loc) || 3701 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3702 "' directive") || 3703 check(FileNumber < 1, Loc, "file number less than one in '" + 3704 DirectiveName + "' directive") || 3705 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3706 "unassigned file number in '" + DirectiveName + "' directive"); 3707 } 3708 3709 /// parseDirectiveCVFuncId 3710 /// ::= .cv_func_id FunctionId 3711 /// 3712 /// Introduces a function ID that can be used with .cv_loc. 3713 bool AsmParser::parseDirectiveCVFuncId() { 3714 SMLoc FunctionIdLoc = getTok().getLoc(); 3715 int64_t FunctionId; 3716 3717 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3718 parseToken(AsmToken::EndOfStatement, 3719 "unexpected token in '.cv_func_id' directive")) 3720 return true; 3721 3722 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3723 return Error(FunctionIdLoc, "function id already allocated"); 3724 3725 return false; 3726 } 3727 3728 /// parseDirectiveCVInlineSiteId 3729 /// ::= .cv_inline_site_id FunctionId 3730 /// "within" IAFunc 3731 /// "inlined_at" IAFile IALine [IACol] 3732 /// 3733 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3734 /// at" source location information for use in the line table of the caller, 3735 /// whether the caller is a real function or another inlined call site. 3736 bool AsmParser::parseDirectiveCVInlineSiteId() { 3737 SMLoc FunctionIdLoc = getTok().getLoc(); 3738 int64_t FunctionId; 3739 int64_t IAFunc; 3740 int64_t IAFile; 3741 int64_t IALine; 3742 int64_t IACol = 0; 3743 3744 // FunctionId 3745 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3746 return true; 3747 3748 // "within" 3749 if (check((getLexer().isNot(AsmToken::Identifier) || 3750 getTok().getIdentifier() != "within"), 3751 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3752 return true; 3753 Lex(); 3754 3755 // IAFunc 3756 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3757 return true; 3758 3759 // "inlined_at" 3760 if (check((getLexer().isNot(AsmToken::Identifier) || 3761 getTok().getIdentifier() != "inlined_at"), 3762 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3763 "directive") ) 3764 return true; 3765 Lex(); 3766 3767 // IAFile IALine 3768 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3769 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3770 return true; 3771 3772 // [IACol] 3773 if (getLexer().is(AsmToken::Integer)) { 3774 IACol = getTok().getIntVal(); 3775 Lex(); 3776 } 3777 3778 if (parseToken(AsmToken::EndOfStatement, 3779 "unexpected token in '.cv_inline_site_id' directive")) 3780 return true; 3781 3782 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3783 IALine, IACol, FunctionIdLoc)) 3784 return Error(FunctionIdLoc, "function id already allocated"); 3785 3786 return false; 3787 } 3788 3789 /// parseDirectiveCVLoc 3790 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3791 /// [is_stmt VALUE] 3792 /// The first number is a file number, must have been previously assigned with 3793 /// a .file directive, the second number is the line number and optionally the 3794 /// third number is a column position (zero if not specified). The remaining 3795 /// optional items are .loc sub-directives. 3796 bool AsmParser::parseDirectiveCVLoc() { 3797 SMLoc DirectiveLoc = getTok().getLoc(); 3798 int64_t FunctionId, FileNumber; 3799 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3800 parseCVFileId(FileNumber, ".cv_loc")) 3801 return true; 3802 3803 int64_t LineNumber = 0; 3804 if (getLexer().is(AsmToken::Integer)) { 3805 LineNumber = getTok().getIntVal(); 3806 if (LineNumber < 0) 3807 return TokError("line number less than zero in '.cv_loc' directive"); 3808 Lex(); 3809 } 3810 3811 int64_t ColumnPos = 0; 3812 if (getLexer().is(AsmToken::Integer)) { 3813 ColumnPos = getTok().getIntVal(); 3814 if (ColumnPos < 0) 3815 return TokError("column position less than zero in '.cv_loc' directive"); 3816 Lex(); 3817 } 3818 3819 bool PrologueEnd = false; 3820 uint64_t IsStmt = 0; 3821 3822 auto parseOp = [&]() -> bool { 3823 StringRef Name; 3824 SMLoc Loc = getTok().getLoc(); 3825 if (parseIdentifier(Name)) 3826 return TokError("unexpected token in '.cv_loc' directive"); 3827 if (Name == "prologue_end") 3828 PrologueEnd = true; 3829 else if (Name == "is_stmt") { 3830 Loc = getTok().getLoc(); 3831 const MCExpr *Value; 3832 if (parseExpression(Value)) 3833 return true; 3834 // The expression must be the constant 0 or 1. 3835 IsStmt = ~0ULL; 3836 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3837 IsStmt = MCE->getValue(); 3838 3839 if (IsStmt > 1) 3840 return Error(Loc, "is_stmt value not 0 or 1"); 3841 } else { 3842 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3843 } 3844 return false; 3845 }; 3846 3847 if (parseMany(parseOp, false /*hasComma*/)) 3848 return true; 3849 3850 getStreamer().emitCVLocDirective(FunctionId, FileNumber, LineNumber, 3851 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3852 DirectiveLoc); 3853 return false; 3854 } 3855 3856 /// parseDirectiveCVLinetable 3857 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3858 bool AsmParser::parseDirectiveCVLinetable() { 3859 int64_t FunctionId; 3860 StringRef FnStartName, FnEndName; 3861 SMLoc Loc = getTok().getLoc(); 3862 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3863 parseToken(AsmToken::Comma, 3864 "unexpected token in '.cv_linetable' directive") || 3865 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3866 "expected identifier in directive") || 3867 parseToken(AsmToken::Comma, 3868 "unexpected token in '.cv_linetable' directive") || 3869 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3870 "expected identifier in directive")) 3871 return true; 3872 3873 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3874 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3875 3876 getStreamer().emitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3877 return false; 3878 } 3879 3880 /// parseDirectiveCVInlineLinetable 3881 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3882 bool AsmParser::parseDirectiveCVInlineLinetable() { 3883 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3884 StringRef FnStartName, FnEndName; 3885 SMLoc Loc = getTok().getLoc(); 3886 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3887 parseTokenLoc(Loc) || 3888 parseIntToken( 3889 SourceFileId, 3890 "expected SourceField in '.cv_inline_linetable' directive") || 3891 check(SourceFileId <= 0, Loc, 3892 "File id less than zero in '.cv_inline_linetable' directive") || 3893 parseTokenLoc(Loc) || 3894 parseIntToken( 3895 SourceLineNum, 3896 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3897 check(SourceLineNum < 0, Loc, 3898 "Line number less than zero in '.cv_inline_linetable' directive") || 3899 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3900 "expected identifier in directive") || 3901 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3902 "expected identifier in directive")) 3903 return true; 3904 3905 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3906 return true; 3907 3908 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3909 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3910 getStreamer().emitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3911 SourceLineNum, FnStartSym, 3912 FnEndSym); 3913 return false; 3914 } 3915 3916 void AsmParser::initializeCVDefRangeTypeMap() { 3917 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3918 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3919 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3920 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3921 } 3922 3923 /// parseDirectiveCVDefRange 3924 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3925 bool AsmParser::parseDirectiveCVDefRange() { 3926 SMLoc Loc; 3927 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3928 while (getLexer().is(AsmToken::Identifier)) { 3929 Loc = getLexer().getLoc(); 3930 StringRef GapStartName; 3931 if (parseIdentifier(GapStartName)) 3932 return Error(Loc, "expected identifier in directive"); 3933 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3934 3935 Loc = getLexer().getLoc(); 3936 StringRef GapEndName; 3937 if (parseIdentifier(GapEndName)) 3938 return Error(Loc, "expected identifier in directive"); 3939 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3940 3941 Ranges.push_back({GapStartSym, GapEndSym}); 3942 } 3943 3944 StringRef CVDefRangeTypeStr; 3945 if (parseToken( 3946 AsmToken::Comma, 3947 "expected comma before def_range type in .cv_def_range directive") || 3948 parseIdentifier(CVDefRangeTypeStr)) 3949 return Error(Loc, "expected def_range type in directive"); 3950 3951 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3952 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3953 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3954 ? CVDR_DEFRANGE 3955 : CVTypeIt->getValue(); 3956 switch (CVDRType) { 3957 case CVDR_DEFRANGE_REGISTER: { 3958 int64_t DRRegister; 3959 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3960 ".cv_def_range directive") || 3961 parseAbsoluteExpression(DRRegister)) 3962 return Error(Loc, "expected register number"); 3963 3964 codeview::DefRangeRegisterHeader DRHdr; 3965 DRHdr.Register = DRRegister; 3966 DRHdr.MayHaveNoName = 0; 3967 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3968 break; 3969 } 3970 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3971 int64_t DROffset; 3972 if (parseToken(AsmToken::Comma, 3973 "expected comma before offset in .cv_def_range directive") || 3974 parseAbsoluteExpression(DROffset)) 3975 return Error(Loc, "expected offset value"); 3976 3977 codeview::DefRangeFramePointerRelHeader DRHdr; 3978 DRHdr.Offset = DROffset; 3979 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3980 break; 3981 } 3982 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3983 int64_t DRRegister; 3984 int64_t DROffsetInParent; 3985 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3986 ".cv_def_range directive") || 3987 parseAbsoluteExpression(DRRegister)) 3988 return Error(Loc, "expected register number"); 3989 if (parseToken(AsmToken::Comma, 3990 "expected comma before offset in .cv_def_range directive") || 3991 parseAbsoluteExpression(DROffsetInParent)) 3992 return Error(Loc, "expected offset value"); 3993 3994 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3995 DRHdr.Register = DRRegister; 3996 DRHdr.MayHaveNoName = 0; 3997 DRHdr.OffsetInParent = DROffsetInParent; 3998 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 3999 break; 4000 } 4001 case CVDR_DEFRANGE_REGISTER_REL: { 4002 int64_t DRRegister; 4003 int64_t DRFlags; 4004 int64_t DRBasePointerOffset; 4005 if (parseToken(AsmToken::Comma, "expected comma before register number in " 4006 ".cv_def_range directive") || 4007 parseAbsoluteExpression(DRRegister)) 4008 return Error(Loc, "expected register value"); 4009 if (parseToken( 4010 AsmToken::Comma, 4011 "expected comma before flag value in .cv_def_range directive") || 4012 parseAbsoluteExpression(DRFlags)) 4013 return Error(Loc, "expected flag value"); 4014 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 4015 "in .cv_def_range directive") || 4016 parseAbsoluteExpression(DRBasePointerOffset)) 4017 return Error(Loc, "expected base pointer offset value"); 4018 4019 codeview::DefRangeRegisterRelHeader DRHdr; 4020 DRHdr.Register = DRRegister; 4021 DRHdr.Flags = DRFlags; 4022 DRHdr.BasePointerOffset = DRBasePointerOffset; 4023 getStreamer().emitCVDefRangeDirective(Ranges, DRHdr); 4024 break; 4025 } 4026 default: 4027 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 4028 } 4029 return true; 4030 } 4031 4032 /// parseDirectiveCVString 4033 /// ::= .cv_stringtable "string" 4034 bool AsmParser::parseDirectiveCVString() { 4035 std::string Data; 4036 if (checkForValidSection() || parseEscapedString(Data)) 4037 return addErrorSuffix(" in '.cv_string' directive"); 4038 4039 // Put the string in the table and emit the offset. 4040 std::pair<StringRef, unsigned> Insertion = 4041 getCVContext().addToStringTable(Data); 4042 getStreamer().emitInt32(Insertion.second); 4043 return false; 4044 } 4045 4046 /// parseDirectiveCVStringTable 4047 /// ::= .cv_stringtable 4048 bool AsmParser::parseDirectiveCVStringTable() { 4049 getStreamer().emitCVStringTableDirective(); 4050 return false; 4051 } 4052 4053 /// parseDirectiveCVFileChecksums 4054 /// ::= .cv_filechecksums 4055 bool AsmParser::parseDirectiveCVFileChecksums() { 4056 getStreamer().emitCVFileChecksumsDirective(); 4057 return false; 4058 } 4059 4060 /// parseDirectiveCVFileChecksumOffset 4061 /// ::= .cv_filechecksumoffset fileno 4062 bool AsmParser::parseDirectiveCVFileChecksumOffset() { 4063 int64_t FileNo; 4064 if (parseIntToken(FileNo, "expected identifier in directive")) 4065 return true; 4066 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 4067 return true; 4068 getStreamer().emitCVFileChecksumOffsetDirective(FileNo); 4069 return false; 4070 } 4071 4072 /// parseDirectiveCVFPOData 4073 /// ::= .cv_fpo_data procsym 4074 bool AsmParser::parseDirectiveCVFPOData() { 4075 SMLoc DirLoc = getLexer().getLoc(); 4076 StringRef ProcName; 4077 if (parseIdentifier(ProcName)) 4078 return TokError("expected symbol name"); 4079 if (parseEOL("unexpected tokens")) 4080 return addErrorSuffix(" in '.cv_fpo_data' directive"); 4081 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 4082 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 4083 return false; 4084 } 4085 4086 /// parseDirectiveCFISections 4087 /// ::= .cfi_sections section [, section] 4088 bool AsmParser::parseDirectiveCFISections() { 4089 StringRef Name; 4090 bool EH = false; 4091 bool Debug = false; 4092 4093 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4094 for (;;) { 4095 if (parseIdentifier(Name)) 4096 return TokError("expected .eh_frame or .debug_frame"); 4097 if (Name == ".eh_frame") 4098 EH = true; 4099 else if (Name == ".debug_frame") 4100 Debug = true; 4101 if (parseOptionalToken(AsmToken::EndOfStatement)) 4102 break; 4103 if (parseToken(AsmToken::Comma)) 4104 return true; 4105 } 4106 } 4107 getStreamer().emitCFISections(EH, Debug); 4108 return false; 4109 } 4110 4111 /// parseDirectiveCFIStartProc 4112 /// ::= .cfi_startproc [simple] 4113 bool AsmParser::parseDirectiveCFIStartProc() { 4114 StringRef Simple; 4115 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4116 if (check(parseIdentifier(Simple) || Simple != "simple", 4117 "unexpected token") || 4118 parseEOL()) 4119 return addErrorSuffix(" in '.cfi_startproc' directive"); 4120 } 4121 4122 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 4123 // being produced if this directive is emitted as part of preprocessor macro 4124 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 4125 // Tools like llvm-mc on the other hand are not affected by it, and report 4126 // correct context information. 4127 getStreamer().emitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 4128 return false; 4129 } 4130 4131 /// parseDirectiveCFIEndProc 4132 /// ::= .cfi_endproc 4133 bool AsmParser::parseDirectiveCFIEndProc() { 4134 if (parseEOL()) 4135 return addErrorSuffix(" in '.cfi_endproc' directive"); 4136 getStreamer().emitCFIEndProc(); 4137 return false; 4138 } 4139 4140 /// parse register name or number. 4141 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 4142 SMLoc DirectiveLoc) { 4143 unsigned RegNo; 4144 4145 if (getLexer().isNot(AsmToken::Integer)) { 4146 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 4147 return true; 4148 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 4149 } else 4150 return parseAbsoluteExpression(Register); 4151 4152 return false; 4153 } 4154 4155 /// parseDirectiveCFIDefCfa 4156 /// ::= .cfi_def_cfa register, offset 4157 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 4158 int64_t Register = 0, Offset = 0; 4159 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4160 parseToken(AsmToken::Comma, "unexpected token in directive") || 4161 parseAbsoluteExpression(Offset) || parseEOL()) 4162 return true; 4163 4164 getStreamer().emitCFIDefCfa(Register, Offset); 4165 return false; 4166 } 4167 4168 /// parseDirectiveCFIDefCfaOffset 4169 /// ::= .cfi_def_cfa_offset offset 4170 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 4171 int64_t Offset = 0; 4172 if (parseAbsoluteExpression(Offset) || parseEOL()) 4173 return true; 4174 4175 getStreamer().emitCFIDefCfaOffset(Offset); 4176 return false; 4177 } 4178 4179 /// parseDirectiveCFIRegister 4180 /// ::= .cfi_register register, register 4181 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 4182 int64_t Register1 = 0, Register2 = 0; 4183 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 4184 parseToken(AsmToken::Comma, "unexpected token in directive") || 4185 parseRegisterOrRegisterNumber(Register2, DirectiveLoc) || parseEOL()) 4186 return true; 4187 4188 getStreamer().emitCFIRegister(Register1, Register2); 4189 return false; 4190 } 4191 4192 /// parseDirectiveCFIWindowSave 4193 /// ::= .cfi_window_save 4194 bool AsmParser::parseDirectiveCFIWindowSave() { 4195 if (parseEOL()) 4196 return true; 4197 getStreamer().emitCFIWindowSave(); 4198 return false; 4199 } 4200 4201 /// parseDirectiveCFIAdjustCfaOffset 4202 /// ::= .cfi_adjust_cfa_offset adjustment 4203 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 4204 int64_t Adjustment = 0; 4205 if (parseAbsoluteExpression(Adjustment) || parseEOL()) 4206 return true; 4207 4208 getStreamer().emitCFIAdjustCfaOffset(Adjustment); 4209 return false; 4210 } 4211 4212 /// parseDirectiveCFIDefCfaRegister 4213 /// ::= .cfi_def_cfa_register register 4214 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 4215 int64_t Register = 0; 4216 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4217 return true; 4218 4219 getStreamer().emitCFIDefCfaRegister(Register); 4220 return false; 4221 } 4222 4223 /// parseDirectiveCFIOffset 4224 /// ::= .cfi_offset register, offset 4225 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 4226 int64_t Register = 0; 4227 int64_t Offset = 0; 4228 4229 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4230 parseToken(AsmToken::Comma, "unexpected token in directive") || 4231 parseAbsoluteExpression(Offset) || parseEOL()) 4232 return true; 4233 4234 getStreamer().emitCFIOffset(Register, Offset); 4235 return false; 4236 } 4237 4238 /// parseDirectiveCFIRelOffset 4239 /// ::= .cfi_rel_offset register, offset 4240 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 4241 int64_t Register = 0, Offset = 0; 4242 4243 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 4244 parseToken(AsmToken::Comma, "unexpected token in directive") || 4245 parseAbsoluteExpression(Offset) || parseEOL()) 4246 return true; 4247 4248 getStreamer().emitCFIRelOffset(Register, Offset); 4249 return false; 4250 } 4251 4252 static bool isValidEncoding(int64_t Encoding) { 4253 if (Encoding & ~0xff) 4254 return false; 4255 4256 if (Encoding == dwarf::DW_EH_PE_omit) 4257 return true; 4258 4259 const unsigned Format = Encoding & 0xf; 4260 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 4261 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 4262 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 4263 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 4264 return false; 4265 4266 const unsigned Application = Encoding & 0x70; 4267 if (Application != dwarf::DW_EH_PE_absptr && 4268 Application != dwarf::DW_EH_PE_pcrel) 4269 return false; 4270 4271 return true; 4272 } 4273 4274 /// parseDirectiveCFIPersonalityOrLsda 4275 /// IsPersonality true for cfi_personality, false for cfi_lsda 4276 /// ::= .cfi_personality encoding, [symbol_name] 4277 /// ::= .cfi_lsda encoding, [symbol_name] 4278 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 4279 int64_t Encoding = 0; 4280 if (parseAbsoluteExpression(Encoding)) 4281 return true; 4282 if (Encoding == dwarf::DW_EH_PE_omit) 4283 return false; 4284 4285 StringRef Name; 4286 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 4287 parseToken(AsmToken::Comma, "unexpected token in directive") || 4288 check(parseIdentifier(Name), "expected identifier in directive") || 4289 parseEOL()) 4290 return true; 4291 4292 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4293 4294 if (IsPersonality) 4295 getStreamer().emitCFIPersonality(Sym, Encoding); 4296 else 4297 getStreamer().emitCFILsda(Sym, Encoding); 4298 return false; 4299 } 4300 4301 /// parseDirectiveCFIRememberState 4302 /// ::= .cfi_remember_state 4303 bool AsmParser::parseDirectiveCFIRememberState() { 4304 if (parseEOL()) 4305 return true; 4306 getStreamer().emitCFIRememberState(); 4307 return false; 4308 } 4309 4310 /// parseDirectiveCFIRestoreState 4311 /// ::= .cfi_remember_state 4312 bool AsmParser::parseDirectiveCFIRestoreState() { 4313 if (parseEOL()) 4314 return true; 4315 getStreamer().emitCFIRestoreState(); 4316 return false; 4317 } 4318 4319 /// parseDirectiveCFISameValue 4320 /// ::= .cfi_same_value register 4321 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 4322 int64_t Register = 0; 4323 4324 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4325 return true; 4326 4327 getStreamer().emitCFISameValue(Register); 4328 return false; 4329 } 4330 4331 /// parseDirectiveCFIRestore 4332 /// ::= .cfi_restore register 4333 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 4334 int64_t Register = 0; 4335 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4336 return true; 4337 4338 getStreamer().emitCFIRestore(Register); 4339 return false; 4340 } 4341 4342 /// parseDirectiveCFIEscape 4343 /// ::= .cfi_escape expression[,...] 4344 bool AsmParser::parseDirectiveCFIEscape() { 4345 std::string Values; 4346 int64_t CurrValue; 4347 if (parseAbsoluteExpression(CurrValue)) 4348 return true; 4349 4350 Values.push_back((uint8_t)CurrValue); 4351 4352 while (getLexer().is(AsmToken::Comma)) { 4353 Lex(); 4354 4355 if (parseAbsoluteExpression(CurrValue)) 4356 return true; 4357 4358 Values.push_back((uint8_t)CurrValue); 4359 } 4360 4361 getStreamer().emitCFIEscape(Values); 4362 return false; 4363 } 4364 4365 /// parseDirectiveCFIReturnColumn 4366 /// ::= .cfi_return_column register 4367 bool AsmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4368 int64_t Register = 0; 4369 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4370 return true; 4371 getStreamer().emitCFIReturnColumn(Register); 4372 return false; 4373 } 4374 4375 /// parseDirectiveCFISignalFrame 4376 /// ::= .cfi_signal_frame 4377 bool AsmParser::parseDirectiveCFISignalFrame() { 4378 if (parseEOL()) 4379 return true; 4380 4381 getStreamer().emitCFISignalFrame(); 4382 return false; 4383 } 4384 4385 /// parseDirectiveCFIUndefined 4386 /// ::= .cfi_undefined register 4387 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4388 int64_t Register = 0; 4389 4390 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || parseEOL()) 4391 return true; 4392 4393 getStreamer().emitCFIUndefined(Register); 4394 return false; 4395 } 4396 4397 /// parseDirectiveAltmacro 4398 /// ::= .altmacro 4399 /// ::= .noaltmacro 4400 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 4401 if (parseEOL()) 4402 return true; 4403 AltMacroMode = (Directive == ".altmacro"); 4404 return false; 4405 } 4406 4407 /// parseDirectiveMacrosOnOff 4408 /// ::= .macros_on 4409 /// ::= .macros_off 4410 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 4411 if (parseEOL()) 4412 return true; 4413 setMacrosEnabled(Directive == ".macros_on"); 4414 return false; 4415 } 4416 4417 /// parseDirectiveMacro 4418 /// ::= .macro name[,] [parameters] 4419 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4420 StringRef Name; 4421 if (parseIdentifier(Name)) 4422 return TokError("expected identifier in '.macro' directive"); 4423 4424 if (getLexer().is(AsmToken::Comma)) 4425 Lex(); 4426 4427 MCAsmMacroParameters Parameters; 4428 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4429 4430 if (!Parameters.empty() && Parameters.back().Vararg) 4431 return Error(Lexer.getLoc(), "vararg parameter '" + 4432 Parameters.back().Name + 4433 "' should be the last parameter"); 4434 4435 MCAsmMacroParameter Parameter; 4436 if (parseIdentifier(Parameter.Name)) 4437 return TokError("expected identifier in '.macro' directive"); 4438 4439 // Emit an error if two (or more) named parameters share the same name 4440 for (const MCAsmMacroParameter& CurrParam : Parameters) 4441 if (CurrParam.Name.equals(Parameter.Name)) 4442 return TokError("macro '" + Name + "' has multiple parameters" 4443 " named '" + Parameter.Name + "'"); 4444 4445 if (Lexer.is(AsmToken::Colon)) { 4446 Lex(); // consume ':' 4447 4448 SMLoc QualLoc; 4449 StringRef Qualifier; 4450 4451 QualLoc = Lexer.getLoc(); 4452 if (parseIdentifier(Qualifier)) 4453 return Error(QualLoc, "missing parameter qualifier for " 4454 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4455 4456 if (Qualifier == "req") 4457 Parameter.Required = true; 4458 else if (Qualifier == "vararg") 4459 Parameter.Vararg = true; 4460 else 4461 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4462 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4463 } 4464 4465 if (getLexer().is(AsmToken::Equal)) { 4466 Lex(); 4467 4468 SMLoc ParamLoc; 4469 4470 ParamLoc = Lexer.getLoc(); 4471 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4472 return true; 4473 4474 if (Parameter.Required) 4475 Warning(ParamLoc, "pointless default value for required parameter " 4476 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4477 } 4478 4479 Parameters.push_back(std::move(Parameter)); 4480 4481 if (getLexer().is(AsmToken::Comma)) 4482 Lex(); 4483 } 4484 4485 // Eat just the end of statement. 4486 Lexer.Lex(); 4487 4488 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4489 AsmToken EndToken, StartToken = getTok(); 4490 unsigned MacroDepth = 0; 4491 // Lex the macro definition. 4492 while (true) { 4493 // Ignore Lexing errors in macros. 4494 while (Lexer.is(AsmToken::Error)) { 4495 Lexer.Lex(); 4496 } 4497 4498 // Check whether we have reached the end of the file. 4499 if (getLexer().is(AsmToken::Eof)) 4500 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4501 4502 // Otherwise, check whether we have reach the .endmacro or the start of a 4503 // preprocessor line marker. 4504 if (getLexer().is(AsmToken::Identifier)) { 4505 if (getTok().getIdentifier() == ".endm" || 4506 getTok().getIdentifier() == ".endmacro") { 4507 if (MacroDepth == 0) { // Outermost macro. 4508 EndToken = getTok(); 4509 Lexer.Lex(); 4510 if (getLexer().isNot(AsmToken::EndOfStatement)) 4511 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4512 "' directive"); 4513 break; 4514 } else { 4515 // Otherwise we just found the end of an inner macro. 4516 --MacroDepth; 4517 } 4518 } else if (getTok().getIdentifier() == ".macro") { 4519 // We allow nested macros. Those aren't instantiated until the outermost 4520 // macro is expanded so just ignore them for now. 4521 ++MacroDepth; 4522 } 4523 } else if (Lexer.is(AsmToken::HashDirective)) { 4524 (void)parseCppHashLineFilenameComment(getLexer().getLoc()); 4525 } 4526 4527 // Otherwise, scan til the end of the statement. 4528 eatToEndOfStatement(); 4529 } 4530 4531 if (getContext().lookupMacro(Name)) { 4532 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4533 } 4534 4535 const char *BodyStart = StartToken.getLoc().getPointer(); 4536 const char *BodyEnd = EndToken.getLoc().getPointer(); 4537 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4538 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4539 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4540 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4541 Macro.dump()); 4542 getContext().defineMacro(Name, std::move(Macro)); 4543 return false; 4544 } 4545 4546 /// checkForBadMacro 4547 /// 4548 /// With the support added for named parameters there may be code out there that 4549 /// is transitioning from positional parameters. In versions of gas that did 4550 /// not support named parameters they would be ignored on the macro definition. 4551 /// But to support both styles of parameters this is not possible so if a macro 4552 /// definition has named parameters but does not use them and has what appears 4553 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4554 /// warning that the positional parameter found in body which have no effect. 4555 /// Hoping the developer will either remove the named parameters from the macro 4556 /// definition so the positional parameters get used if that was what was 4557 /// intended or change the macro to use the named parameters. It is possible 4558 /// this warning will trigger when the none of the named parameters are used 4559 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4560 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4561 StringRef Body, 4562 ArrayRef<MCAsmMacroParameter> Parameters) { 4563 // If this macro is not defined with named parameters the warning we are 4564 // checking for here doesn't apply. 4565 unsigned NParameters = Parameters.size(); 4566 if (NParameters == 0) 4567 return; 4568 4569 bool NamedParametersFound = false; 4570 bool PositionalParametersFound = false; 4571 4572 // Look at the body of the macro for use of both the named parameters and what 4573 // are likely to be positional parameters. This is what expandMacro() is 4574 // doing when it finds the parameters in the body. 4575 while (!Body.empty()) { 4576 // Scan for the next possible parameter. 4577 std::size_t End = Body.size(), Pos = 0; 4578 for (; Pos != End; ++Pos) { 4579 // Check for a substitution or escape. 4580 // This macro is defined with parameters, look for \foo, \bar, etc. 4581 if (Body[Pos] == '\\' && Pos + 1 != End) 4582 break; 4583 4584 // This macro should have parameters, but look for $0, $1, ..., $n too. 4585 if (Body[Pos] != '$' || Pos + 1 == End) 4586 continue; 4587 char Next = Body[Pos + 1]; 4588 if (Next == '$' || Next == 'n' || 4589 isdigit(static_cast<unsigned char>(Next))) 4590 break; 4591 } 4592 4593 // Check if we reached the end. 4594 if (Pos == End) 4595 break; 4596 4597 if (Body[Pos] == '$') { 4598 switch (Body[Pos + 1]) { 4599 // $$ => $ 4600 case '$': 4601 break; 4602 4603 // $n => number of arguments 4604 case 'n': 4605 PositionalParametersFound = true; 4606 break; 4607 4608 // $[0-9] => argument 4609 default: { 4610 PositionalParametersFound = true; 4611 break; 4612 } 4613 } 4614 Pos += 2; 4615 } else { 4616 unsigned I = Pos + 1; 4617 while (isIdentifierChar(Body[I]) && I + 1 != End) 4618 ++I; 4619 4620 const char *Begin = Body.data() + Pos + 1; 4621 StringRef Argument(Begin, I - (Pos + 1)); 4622 unsigned Index = 0; 4623 for (; Index < NParameters; ++Index) 4624 if (Parameters[Index].Name == Argument) 4625 break; 4626 4627 if (Index == NParameters) { 4628 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4629 Pos += 3; 4630 else { 4631 Pos = I; 4632 } 4633 } else { 4634 NamedParametersFound = true; 4635 Pos += 1 + Argument.size(); 4636 } 4637 } 4638 // Update the scan point. 4639 Body = Body.substr(Pos); 4640 } 4641 4642 if (!NamedParametersFound && PositionalParametersFound) 4643 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4644 "used in macro body, possible positional parameter " 4645 "found in body which will have no effect"); 4646 } 4647 4648 /// parseDirectiveExitMacro 4649 /// ::= .exitm 4650 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4651 if (parseEOL()) 4652 return true; 4653 4654 if (!isInsideMacroInstantiation()) 4655 return TokError("unexpected '" + Directive + "' in file, " 4656 "no current macro definition"); 4657 4658 // Exit all conditionals that are active in the current macro. 4659 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4660 TheCondState = TheCondStack.back(); 4661 TheCondStack.pop_back(); 4662 } 4663 4664 handleMacroExit(); 4665 return false; 4666 } 4667 4668 /// parseDirectiveEndMacro 4669 /// ::= .endm 4670 /// ::= .endmacro 4671 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4672 if (getLexer().isNot(AsmToken::EndOfStatement)) 4673 return TokError("unexpected token in '" + Directive + "' directive"); 4674 4675 // If we are inside a macro instantiation, terminate the current 4676 // instantiation. 4677 if (isInsideMacroInstantiation()) { 4678 handleMacroExit(); 4679 return false; 4680 } 4681 4682 // Otherwise, this .endmacro is a stray entry in the file; well formed 4683 // .endmacro directives are handled during the macro definition parsing. 4684 return TokError("unexpected '" + Directive + "' in file, " 4685 "no current macro definition"); 4686 } 4687 4688 /// parseDirectivePurgeMacro 4689 /// ::= .purgem name 4690 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4691 StringRef Name; 4692 SMLoc Loc; 4693 if (parseTokenLoc(Loc) || 4694 check(parseIdentifier(Name), Loc, 4695 "expected identifier in '.purgem' directive") || 4696 parseEOL()) 4697 return true; 4698 4699 if (!getContext().lookupMacro(Name)) 4700 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4701 4702 getContext().undefineMacro(Name); 4703 DEBUG_WITH_TYPE("asm-macros", dbgs() 4704 << "Un-defining macro: " << Name << "\n"); 4705 return false; 4706 } 4707 4708 /// parseDirectiveBundleAlignMode 4709 /// ::= {.bundle_align_mode} expression 4710 bool AsmParser::parseDirectiveBundleAlignMode() { 4711 // Expect a single argument: an expression that evaluates to a constant 4712 // in the inclusive range 0-30. 4713 SMLoc ExprLoc = getLexer().getLoc(); 4714 int64_t AlignSizePow2; 4715 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4716 parseEOL() || 4717 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4718 "invalid bundle alignment size (expected between 0 and 30)")) 4719 return true; 4720 4721 // Because of AlignSizePow2's verified range we can safely truncate it to 4722 // unsigned. 4723 getStreamer().emitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4724 return false; 4725 } 4726 4727 /// parseDirectiveBundleLock 4728 /// ::= {.bundle_lock} [align_to_end] 4729 bool AsmParser::parseDirectiveBundleLock() { 4730 if (checkForValidSection()) 4731 return true; 4732 bool AlignToEnd = false; 4733 4734 StringRef Option; 4735 SMLoc Loc = getTok().getLoc(); 4736 const char *kInvalidOptionError = 4737 "invalid option for '.bundle_lock' directive"; 4738 4739 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4740 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4741 check(Option != "align_to_end", Loc, kInvalidOptionError) || parseEOL()) 4742 return true; 4743 AlignToEnd = true; 4744 } 4745 4746 getStreamer().emitBundleLock(AlignToEnd); 4747 return false; 4748 } 4749 4750 /// parseDirectiveBundleLock 4751 /// ::= {.bundle_lock} 4752 bool AsmParser::parseDirectiveBundleUnlock() { 4753 if (checkForValidSection() || parseEOL()) 4754 return true; 4755 4756 getStreamer().emitBundleUnlock(); 4757 return false; 4758 } 4759 4760 /// parseDirectiveSpace 4761 /// ::= (.skip | .space) expression [ , expression ] 4762 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4763 SMLoc NumBytesLoc = Lexer.getLoc(); 4764 const MCExpr *NumBytes; 4765 if (checkForValidSection() || parseExpression(NumBytes)) 4766 return true; 4767 4768 int64_t FillExpr = 0; 4769 if (parseOptionalToken(AsmToken::Comma)) 4770 if (parseAbsoluteExpression(FillExpr)) 4771 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4772 if (parseEOL()) 4773 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4774 4775 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4776 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4777 4778 return false; 4779 } 4780 4781 /// parseDirectiveDCB 4782 /// ::= .dcb.{b, l, w} expression, expression 4783 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4784 SMLoc NumValuesLoc = Lexer.getLoc(); 4785 int64_t NumValues; 4786 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4787 return true; 4788 4789 if (NumValues < 0) { 4790 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4791 return false; 4792 } 4793 4794 if (parseToken(AsmToken::Comma, 4795 "unexpected token in '" + Twine(IDVal) + "' directive")) 4796 return true; 4797 4798 const MCExpr *Value; 4799 SMLoc ExprLoc = getLexer().getLoc(); 4800 if (parseExpression(Value)) 4801 return true; 4802 4803 // Special case constant expressions to match code generator. 4804 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4805 assert(Size <= 8 && "Invalid size"); 4806 uint64_t IntValue = MCE->getValue(); 4807 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4808 return Error(ExprLoc, "literal value out of range for directive"); 4809 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4810 getStreamer().emitIntValue(IntValue, Size); 4811 } else { 4812 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4813 getStreamer().emitValue(Value, Size, ExprLoc); 4814 } 4815 4816 return parseEOL(); 4817 } 4818 4819 /// parseDirectiveRealDCB 4820 /// ::= .dcb.{d, s} expression, expression 4821 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4822 SMLoc NumValuesLoc = Lexer.getLoc(); 4823 int64_t NumValues; 4824 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4825 return true; 4826 4827 if (NumValues < 0) { 4828 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4829 return false; 4830 } 4831 4832 if (parseToken(AsmToken::Comma, 4833 "unexpected token in '" + Twine(IDVal) + "' directive")) 4834 return true; 4835 4836 APInt AsInt; 4837 if (parseRealValue(Semantics, AsInt) || parseEOL()) 4838 return true; 4839 4840 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4841 getStreamer().emitIntValue(AsInt.getLimitedValue(), 4842 AsInt.getBitWidth() / 8); 4843 4844 return false; 4845 } 4846 4847 /// parseDirectiveDS 4848 /// ::= .ds.{b, d, l, p, s, w, x} expression 4849 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4850 SMLoc NumValuesLoc = Lexer.getLoc(); 4851 int64_t NumValues; 4852 if (checkForValidSection() || parseAbsoluteExpression(NumValues) || 4853 parseEOL()) 4854 return true; 4855 4856 if (NumValues < 0) { 4857 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4858 return false; 4859 } 4860 4861 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4862 getStreamer().emitFill(Size, 0); 4863 4864 return false; 4865 } 4866 4867 /// parseDirectiveLEB128 4868 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4869 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4870 if (checkForValidSection()) 4871 return true; 4872 4873 auto parseOp = [&]() -> bool { 4874 const MCExpr *Value; 4875 if (parseExpression(Value)) 4876 return true; 4877 if (Signed) 4878 getStreamer().emitSLEB128Value(Value); 4879 else 4880 getStreamer().emitULEB128Value(Value); 4881 return false; 4882 }; 4883 4884 if (parseMany(parseOp)) 4885 return addErrorSuffix(" in directive"); 4886 4887 return false; 4888 } 4889 4890 /// parseDirectiveSymbolAttribute 4891 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4892 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4893 auto parseOp = [&]() -> bool { 4894 StringRef Name; 4895 SMLoc Loc = getTok().getLoc(); 4896 if (parseIdentifier(Name)) 4897 return Error(Loc, "expected identifier"); 4898 4899 if (discardLTOSymbol(Name)) 4900 return false; 4901 4902 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4903 4904 // Assembler local symbols don't make any sense here. Complain loudly. 4905 if (Sym->isTemporary()) 4906 return Error(Loc, "non-local symbol required"); 4907 4908 if (!getStreamer().emitSymbolAttribute(Sym, Attr)) 4909 return Error(Loc, "unable to emit symbol attribute"); 4910 return false; 4911 }; 4912 4913 if (parseMany(parseOp)) 4914 return addErrorSuffix(" in directive"); 4915 return false; 4916 } 4917 4918 /// parseDirectiveComm 4919 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4920 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4921 if (checkForValidSection()) 4922 return true; 4923 4924 SMLoc IDLoc = getLexer().getLoc(); 4925 StringRef Name; 4926 if (parseIdentifier(Name)) 4927 return TokError("expected identifier in directive"); 4928 4929 // Handle the identifier as the key symbol. 4930 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4931 4932 if (getLexer().isNot(AsmToken::Comma)) 4933 return TokError("unexpected token in directive"); 4934 Lex(); 4935 4936 int64_t Size; 4937 SMLoc SizeLoc = getLexer().getLoc(); 4938 if (parseAbsoluteExpression(Size)) 4939 return true; 4940 4941 int64_t Pow2Alignment = 0; 4942 SMLoc Pow2AlignmentLoc; 4943 if (getLexer().is(AsmToken::Comma)) { 4944 Lex(); 4945 Pow2AlignmentLoc = getLexer().getLoc(); 4946 if (parseAbsoluteExpression(Pow2Alignment)) 4947 return true; 4948 4949 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4950 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4951 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4952 4953 // If this target takes alignments in bytes (not log) validate and convert. 4954 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4955 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4956 if (!isPowerOf2_64(Pow2Alignment)) 4957 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4958 Pow2Alignment = Log2_64(Pow2Alignment); 4959 } 4960 } 4961 4962 if (parseEOL()) 4963 return true; 4964 4965 // NOTE: a size of zero for a .comm should create a undefined symbol 4966 // but a size of .lcomm creates a bss symbol of size zero. 4967 if (Size < 0) 4968 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4969 "be less than zero"); 4970 4971 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4972 // may internally end up wanting an alignment in bytes. 4973 // FIXME: Diagnose overflow. 4974 if (Pow2Alignment < 0) 4975 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4976 "alignment, can't be less than zero"); 4977 4978 Sym->redefineIfPossible(); 4979 if (!Sym->isUndefined()) 4980 return Error(IDLoc, "invalid symbol redefinition"); 4981 4982 // Create the Symbol as a common or local common with Size and Pow2Alignment 4983 if (IsLocal) { 4984 getStreamer().emitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4985 return false; 4986 } 4987 4988 getStreamer().emitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4989 return false; 4990 } 4991 4992 /// parseDirectiveAbort 4993 /// ::= .abort [... message ...] 4994 bool AsmParser::parseDirectiveAbort() { 4995 // FIXME: Use loc from directive. 4996 SMLoc Loc = getLexer().getLoc(); 4997 4998 StringRef Str = parseStringToEndOfStatement(); 4999 if (parseEOL()) 5000 return true; 5001 5002 if (Str.empty()) 5003 return Error(Loc, ".abort detected. Assembly stopping."); 5004 else 5005 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 5006 // FIXME: Actually abort assembly here. 5007 5008 return false; 5009 } 5010 5011 /// parseDirectiveInclude 5012 /// ::= .include "filename" 5013 bool AsmParser::parseDirectiveInclude() { 5014 // Allow the strings to have escaped octal character sequence. 5015 std::string Filename; 5016 SMLoc IncludeLoc = getTok().getLoc(); 5017 5018 if (check(getTok().isNot(AsmToken::String), 5019 "expected string in '.include' directive") || 5020 parseEscapedString(Filename) || 5021 check(getTok().isNot(AsmToken::EndOfStatement), 5022 "unexpected token in '.include' directive") || 5023 // Attempt to switch the lexer to the included file before consuming the 5024 // end of statement to avoid losing it when we switch. 5025 check(enterIncludeFile(Filename), IncludeLoc, 5026 "Could not find include file '" + Filename + "'")) 5027 return true; 5028 5029 return false; 5030 } 5031 5032 /// parseDirectiveIncbin 5033 /// ::= .incbin "filename" [ , skip [ , count ] ] 5034 bool AsmParser::parseDirectiveIncbin() { 5035 // Allow the strings to have escaped octal character sequence. 5036 std::string Filename; 5037 SMLoc IncbinLoc = getTok().getLoc(); 5038 if (check(getTok().isNot(AsmToken::String), 5039 "expected string in '.incbin' directive") || 5040 parseEscapedString(Filename)) 5041 return true; 5042 5043 int64_t Skip = 0; 5044 const MCExpr *Count = nullptr; 5045 SMLoc SkipLoc, CountLoc; 5046 if (parseOptionalToken(AsmToken::Comma)) { 5047 // The skip expression can be omitted while specifying the count, e.g: 5048 // .incbin "filename",,4 5049 if (getTok().isNot(AsmToken::Comma)) { 5050 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 5051 return true; 5052 } 5053 if (parseOptionalToken(AsmToken::Comma)) { 5054 CountLoc = getTok().getLoc(); 5055 if (parseExpression(Count)) 5056 return true; 5057 } 5058 } 5059 5060 if (parseEOL()) 5061 return true; 5062 5063 if (check(Skip < 0, SkipLoc, "skip is negative")) 5064 return true; 5065 5066 // Attempt to process the included file. 5067 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 5068 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 5069 return false; 5070 } 5071 5072 /// parseDirectiveIf 5073 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 5074 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 5075 TheCondStack.push_back(TheCondState); 5076 TheCondState.TheCond = AsmCond::IfCond; 5077 if (TheCondState.Ignore) { 5078 eatToEndOfStatement(); 5079 } else { 5080 int64_t ExprValue; 5081 if (parseAbsoluteExpression(ExprValue) || parseEOL()) 5082 return true; 5083 5084 switch (DirKind) { 5085 default: 5086 llvm_unreachable("unsupported directive"); 5087 case DK_IF: 5088 case DK_IFNE: 5089 break; 5090 case DK_IFEQ: 5091 ExprValue = ExprValue == 0; 5092 break; 5093 case DK_IFGE: 5094 ExprValue = ExprValue >= 0; 5095 break; 5096 case DK_IFGT: 5097 ExprValue = ExprValue > 0; 5098 break; 5099 case DK_IFLE: 5100 ExprValue = ExprValue <= 0; 5101 break; 5102 case DK_IFLT: 5103 ExprValue = ExprValue < 0; 5104 break; 5105 } 5106 5107 TheCondState.CondMet = ExprValue; 5108 TheCondState.Ignore = !TheCondState.CondMet; 5109 } 5110 5111 return false; 5112 } 5113 5114 /// parseDirectiveIfb 5115 /// ::= .ifb string 5116 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 5117 TheCondStack.push_back(TheCondState); 5118 TheCondState.TheCond = AsmCond::IfCond; 5119 5120 if (TheCondState.Ignore) { 5121 eatToEndOfStatement(); 5122 } else { 5123 StringRef Str = parseStringToEndOfStatement(); 5124 5125 if (parseEOL()) 5126 return true; 5127 5128 TheCondState.CondMet = ExpectBlank == Str.empty(); 5129 TheCondState.Ignore = !TheCondState.CondMet; 5130 } 5131 5132 return false; 5133 } 5134 5135 /// parseDirectiveIfc 5136 /// ::= .ifc string1, string2 5137 /// ::= .ifnc string1, string2 5138 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 5139 TheCondStack.push_back(TheCondState); 5140 TheCondState.TheCond = AsmCond::IfCond; 5141 5142 if (TheCondState.Ignore) { 5143 eatToEndOfStatement(); 5144 } else { 5145 StringRef Str1 = parseStringToComma(); 5146 5147 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 5148 return true; 5149 5150 StringRef Str2 = parseStringToEndOfStatement(); 5151 5152 if (parseEOL()) 5153 return true; 5154 5155 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 5156 TheCondState.Ignore = !TheCondState.CondMet; 5157 } 5158 5159 return false; 5160 } 5161 5162 /// parseDirectiveIfeqs 5163 /// ::= .ifeqs string1, string2 5164 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 5165 if (Lexer.isNot(AsmToken::String)) { 5166 if (ExpectEqual) 5167 return TokError("expected string parameter for '.ifeqs' directive"); 5168 return TokError("expected string parameter for '.ifnes' directive"); 5169 } 5170 5171 StringRef String1 = getTok().getStringContents(); 5172 Lex(); 5173 5174 if (Lexer.isNot(AsmToken::Comma)) { 5175 if (ExpectEqual) 5176 return TokError( 5177 "expected comma after first string for '.ifeqs' directive"); 5178 return TokError("expected comma after first string for '.ifnes' directive"); 5179 } 5180 5181 Lex(); 5182 5183 if (Lexer.isNot(AsmToken::String)) { 5184 if (ExpectEqual) 5185 return TokError("expected string parameter for '.ifeqs' directive"); 5186 return TokError("expected string parameter for '.ifnes' directive"); 5187 } 5188 5189 StringRef String2 = getTok().getStringContents(); 5190 Lex(); 5191 5192 TheCondStack.push_back(TheCondState); 5193 TheCondState.TheCond = AsmCond::IfCond; 5194 TheCondState.CondMet = ExpectEqual == (String1 == String2); 5195 TheCondState.Ignore = !TheCondState.CondMet; 5196 5197 return false; 5198 } 5199 5200 /// parseDirectiveIfdef 5201 /// ::= .ifdef symbol 5202 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 5203 StringRef Name; 5204 TheCondStack.push_back(TheCondState); 5205 TheCondState.TheCond = AsmCond::IfCond; 5206 5207 if (TheCondState.Ignore) { 5208 eatToEndOfStatement(); 5209 } else { 5210 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 5211 parseEOL()) 5212 return true; 5213 5214 MCSymbol *Sym = getContext().lookupSymbol(Name); 5215 5216 if (expect_defined) 5217 TheCondState.CondMet = (Sym && !Sym->isUndefined(false)); 5218 else 5219 TheCondState.CondMet = (!Sym || Sym->isUndefined(false)); 5220 TheCondState.Ignore = !TheCondState.CondMet; 5221 } 5222 5223 return false; 5224 } 5225 5226 /// parseDirectiveElseIf 5227 /// ::= .elseif expression 5228 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 5229 if (TheCondState.TheCond != AsmCond::IfCond && 5230 TheCondState.TheCond != AsmCond::ElseIfCond) 5231 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 5232 " .if or an .elseif"); 5233 TheCondState.TheCond = AsmCond::ElseIfCond; 5234 5235 bool LastIgnoreState = false; 5236 if (!TheCondStack.empty()) 5237 LastIgnoreState = TheCondStack.back().Ignore; 5238 if (LastIgnoreState || TheCondState.CondMet) { 5239 TheCondState.Ignore = true; 5240 eatToEndOfStatement(); 5241 } else { 5242 int64_t ExprValue; 5243 if (parseAbsoluteExpression(ExprValue)) 5244 return true; 5245 5246 if (parseEOL()) 5247 return true; 5248 5249 TheCondState.CondMet = ExprValue; 5250 TheCondState.Ignore = !TheCondState.CondMet; 5251 } 5252 5253 return false; 5254 } 5255 5256 /// parseDirectiveElse 5257 /// ::= .else 5258 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 5259 if (parseEOL()) 5260 return true; 5261 5262 if (TheCondState.TheCond != AsmCond::IfCond && 5263 TheCondState.TheCond != AsmCond::ElseIfCond) 5264 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 5265 " an .if or an .elseif"); 5266 TheCondState.TheCond = AsmCond::ElseCond; 5267 bool LastIgnoreState = false; 5268 if (!TheCondStack.empty()) 5269 LastIgnoreState = TheCondStack.back().Ignore; 5270 if (LastIgnoreState || TheCondState.CondMet) 5271 TheCondState.Ignore = true; 5272 else 5273 TheCondState.Ignore = false; 5274 5275 return false; 5276 } 5277 5278 /// parseDirectiveEnd 5279 /// ::= .end 5280 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 5281 if (parseEOL()) 5282 return true; 5283 5284 while (Lexer.isNot(AsmToken::Eof)) 5285 Lexer.Lex(); 5286 5287 return false; 5288 } 5289 5290 /// parseDirectiveError 5291 /// ::= .err 5292 /// ::= .error [string] 5293 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 5294 if (!TheCondStack.empty()) { 5295 if (TheCondStack.back().Ignore) { 5296 eatToEndOfStatement(); 5297 return false; 5298 } 5299 } 5300 5301 if (!WithMessage) 5302 return Error(L, ".err encountered"); 5303 5304 StringRef Message = ".error directive invoked in source file"; 5305 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5306 if (Lexer.isNot(AsmToken::String)) 5307 return TokError(".error argument must be a string"); 5308 5309 Message = getTok().getStringContents(); 5310 Lex(); 5311 } 5312 5313 return Error(L, Message); 5314 } 5315 5316 /// parseDirectiveWarning 5317 /// ::= .warning [string] 5318 bool AsmParser::parseDirectiveWarning(SMLoc L) { 5319 if (!TheCondStack.empty()) { 5320 if (TheCondStack.back().Ignore) { 5321 eatToEndOfStatement(); 5322 return false; 5323 } 5324 } 5325 5326 StringRef Message = ".warning directive invoked in source file"; 5327 5328 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 5329 if (Lexer.isNot(AsmToken::String)) 5330 return TokError(".warning argument must be a string"); 5331 5332 Message = getTok().getStringContents(); 5333 Lex(); 5334 if (parseEOL()) 5335 return true; 5336 } 5337 5338 return Warning(L, Message); 5339 } 5340 5341 /// parseDirectiveEndIf 5342 /// ::= .endif 5343 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5344 if (parseEOL()) 5345 return true; 5346 5347 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5348 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5349 "an .if or .else"); 5350 if (!TheCondStack.empty()) { 5351 TheCondState = TheCondStack.back(); 5352 TheCondStack.pop_back(); 5353 } 5354 5355 return false; 5356 } 5357 5358 void AsmParser::initializeDirectiveKindMap() { 5359 /* Lookup will be done with the directive 5360 * converted to lower case, so all these 5361 * keys should be lower case. 5362 * (target specific directives are handled 5363 * elsewhere) 5364 */ 5365 DirectiveKindMap[".set"] = DK_SET; 5366 DirectiveKindMap[".equ"] = DK_EQU; 5367 DirectiveKindMap[".equiv"] = DK_EQUIV; 5368 DirectiveKindMap[".ascii"] = DK_ASCII; 5369 DirectiveKindMap[".asciz"] = DK_ASCIZ; 5370 DirectiveKindMap[".string"] = DK_STRING; 5371 DirectiveKindMap[".byte"] = DK_BYTE; 5372 DirectiveKindMap[".short"] = DK_SHORT; 5373 DirectiveKindMap[".value"] = DK_VALUE; 5374 DirectiveKindMap[".2byte"] = DK_2BYTE; 5375 DirectiveKindMap[".long"] = DK_LONG; 5376 DirectiveKindMap[".int"] = DK_INT; 5377 DirectiveKindMap[".4byte"] = DK_4BYTE; 5378 DirectiveKindMap[".quad"] = DK_QUAD; 5379 DirectiveKindMap[".8byte"] = DK_8BYTE; 5380 DirectiveKindMap[".octa"] = DK_OCTA; 5381 DirectiveKindMap[".single"] = DK_SINGLE; 5382 DirectiveKindMap[".float"] = DK_FLOAT; 5383 DirectiveKindMap[".double"] = DK_DOUBLE; 5384 DirectiveKindMap[".align"] = DK_ALIGN; 5385 DirectiveKindMap[".align32"] = DK_ALIGN32; 5386 DirectiveKindMap[".balign"] = DK_BALIGN; 5387 DirectiveKindMap[".balignw"] = DK_BALIGNW; 5388 DirectiveKindMap[".balignl"] = DK_BALIGNL; 5389 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 5390 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 5391 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 5392 DirectiveKindMap[".org"] = DK_ORG; 5393 DirectiveKindMap[".fill"] = DK_FILL; 5394 DirectiveKindMap[".zero"] = DK_ZERO; 5395 DirectiveKindMap[".extern"] = DK_EXTERN; 5396 DirectiveKindMap[".globl"] = DK_GLOBL; 5397 DirectiveKindMap[".global"] = DK_GLOBAL; 5398 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 5399 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 5400 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 5401 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 5402 DirectiveKindMap[".reference"] = DK_REFERENCE; 5403 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 5404 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 5405 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 5406 DirectiveKindMap[".cold"] = DK_COLD; 5407 DirectiveKindMap[".comm"] = DK_COMM; 5408 DirectiveKindMap[".common"] = DK_COMMON; 5409 DirectiveKindMap[".lcomm"] = DK_LCOMM; 5410 DirectiveKindMap[".abort"] = DK_ABORT; 5411 DirectiveKindMap[".include"] = DK_INCLUDE; 5412 DirectiveKindMap[".incbin"] = DK_INCBIN; 5413 DirectiveKindMap[".code16"] = DK_CODE16; 5414 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 5415 DirectiveKindMap[".rept"] = DK_REPT; 5416 DirectiveKindMap[".rep"] = DK_REPT; 5417 DirectiveKindMap[".irp"] = DK_IRP; 5418 DirectiveKindMap[".irpc"] = DK_IRPC; 5419 DirectiveKindMap[".endr"] = DK_ENDR; 5420 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 5421 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 5422 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 5423 DirectiveKindMap[".if"] = DK_IF; 5424 DirectiveKindMap[".ifeq"] = DK_IFEQ; 5425 DirectiveKindMap[".ifge"] = DK_IFGE; 5426 DirectiveKindMap[".ifgt"] = DK_IFGT; 5427 DirectiveKindMap[".ifle"] = DK_IFLE; 5428 DirectiveKindMap[".iflt"] = DK_IFLT; 5429 DirectiveKindMap[".ifne"] = DK_IFNE; 5430 DirectiveKindMap[".ifb"] = DK_IFB; 5431 DirectiveKindMap[".ifnb"] = DK_IFNB; 5432 DirectiveKindMap[".ifc"] = DK_IFC; 5433 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 5434 DirectiveKindMap[".ifnc"] = DK_IFNC; 5435 DirectiveKindMap[".ifnes"] = DK_IFNES; 5436 DirectiveKindMap[".ifdef"] = DK_IFDEF; 5437 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 5438 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 5439 DirectiveKindMap[".elseif"] = DK_ELSEIF; 5440 DirectiveKindMap[".else"] = DK_ELSE; 5441 DirectiveKindMap[".end"] = DK_END; 5442 DirectiveKindMap[".endif"] = DK_ENDIF; 5443 DirectiveKindMap[".skip"] = DK_SKIP; 5444 DirectiveKindMap[".space"] = DK_SPACE; 5445 DirectiveKindMap[".file"] = DK_FILE; 5446 DirectiveKindMap[".line"] = DK_LINE; 5447 DirectiveKindMap[".loc"] = DK_LOC; 5448 DirectiveKindMap[".stabs"] = DK_STABS; 5449 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5450 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5451 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5452 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5453 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5454 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5455 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5456 DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5457 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5458 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5459 DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5460 DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5461 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5462 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5463 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5464 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5465 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5466 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5467 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5468 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5469 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5470 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5471 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5472 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5473 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5474 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5475 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5476 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5477 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5478 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5479 DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5480 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5481 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5482 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5483 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5484 DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5485 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5486 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5487 DirectiveKindMap[".macro"] = DK_MACRO; 5488 DirectiveKindMap[".exitm"] = DK_EXITM; 5489 DirectiveKindMap[".endm"] = DK_ENDM; 5490 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5491 DirectiveKindMap[".purgem"] = DK_PURGEM; 5492 DirectiveKindMap[".err"] = DK_ERR; 5493 DirectiveKindMap[".error"] = DK_ERROR; 5494 DirectiveKindMap[".warning"] = DK_WARNING; 5495 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5496 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5497 DirectiveKindMap[".reloc"] = DK_RELOC; 5498 DirectiveKindMap[".dc"] = DK_DC; 5499 DirectiveKindMap[".dc.a"] = DK_DC_A; 5500 DirectiveKindMap[".dc.b"] = DK_DC_B; 5501 DirectiveKindMap[".dc.d"] = DK_DC_D; 5502 DirectiveKindMap[".dc.l"] = DK_DC_L; 5503 DirectiveKindMap[".dc.s"] = DK_DC_S; 5504 DirectiveKindMap[".dc.w"] = DK_DC_W; 5505 DirectiveKindMap[".dc.x"] = DK_DC_X; 5506 DirectiveKindMap[".dcb"] = DK_DCB; 5507 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5508 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5509 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5510 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5511 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5512 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5513 DirectiveKindMap[".ds"] = DK_DS; 5514 DirectiveKindMap[".ds.b"] = DK_DS_B; 5515 DirectiveKindMap[".ds.d"] = DK_DS_D; 5516 DirectiveKindMap[".ds.l"] = DK_DS_L; 5517 DirectiveKindMap[".ds.p"] = DK_DS_P; 5518 DirectiveKindMap[".ds.s"] = DK_DS_S; 5519 DirectiveKindMap[".ds.w"] = DK_DS_W; 5520 DirectiveKindMap[".ds.x"] = DK_DS_X; 5521 DirectiveKindMap[".print"] = DK_PRINT; 5522 DirectiveKindMap[".addrsig"] = DK_ADDRSIG; 5523 DirectiveKindMap[".addrsig_sym"] = DK_ADDRSIG_SYM; 5524 DirectiveKindMap[".pseudoprobe"] = DK_PSEUDO_PROBE; 5525 DirectiveKindMap[".lto_discard"] = DK_LTO_DISCARD; 5526 } 5527 5528 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5529 AsmToken EndToken, StartToken = getTok(); 5530 5531 unsigned NestLevel = 0; 5532 while (true) { 5533 // Check whether we have reached the end of the file. 5534 if (getLexer().is(AsmToken::Eof)) { 5535 printError(DirectiveLoc, "no matching '.endr' in definition"); 5536 return nullptr; 5537 } 5538 5539 if (Lexer.is(AsmToken::Identifier) && 5540 (getTok().getIdentifier() == ".rep" || 5541 getTok().getIdentifier() == ".rept" || 5542 getTok().getIdentifier() == ".irp" || 5543 getTok().getIdentifier() == ".irpc")) { 5544 ++NestLevel; 5545 } 5546 5547 // Otherwise, check whether we have reached the .endr. 5548 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5549 if (NestLevel == 0) { 5550 EndToken = getTok(); 5551 Lex(); 5552 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5553 printError(getTok().getLoc(), 5554 "unexpected token in '.endr' directive"); 5555 return nullptr; 5556 } 5557 break; 5558 } 5559 --NestLevel; 5560 } 5561 5562 // Otherwise, scan till the end of the statement. 5563 eatToEndOfStatement(); 5564 } 5565 5566 const char *BodyStart = StartToken.getLoc().getPointer(); 5567 const char *BodyEnd = EndToken.getLoc().getPointer(); 5568 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5569 5570 // We Are Anonymous. 5571 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5572 return &MacroLikeBodies.back(); 5573 } 5574 5575 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5576 raw_svector_ostream &OS) { 5577 OS << ".endr\n"; 5578 5579 std::unique_ptr<MemoryBuffer> Instantiation = 5580 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5581 5582 // Create the macro instantiation object and add to the current macro 5583 // instantiation stack. 5584 MacroInstantiation *MI = new MacroInstantiation{ 5585 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 5586 ActiveMacros.push_back(MI); 5587 5588 // Jump to the macro instantiation and prime the lexer. 5589 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5590 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5591 Lex(); 5592 } 5593 5594 /// parseDirectiveRept 5595 /// ::= .rep | .rept count 5596 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5597 const MCExpr *CountExpr; 5598 SMLoc CountLoc = getTok().getLoc(); 5599 if (parseExpression(CountExpr)) 5600 return true; 5601 5602 int64_t Count; 5603 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5604 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5605 } 5606 5607 if (check(Count < 0, CountLoc, "Count is negative") || parseEOL()) 5608 return true; 5609 5610 // Lex the rept definition. 5611 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5612 if (!M) 5613 return true; 5614 5615 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5616 // to hold the macro body with substitutions. 5617 SmallString<256> Buf; 5618 raw_svector_ostream OS(Buf); 5619 while (Count--) { 5620 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5621 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5622 return true; 5623 } 5624 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5625 5626 return false; 5627 } 5628 5629 /// parseDirectiveIrp 5630 /// ::= .irp symbol,values 5631 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5632 MCAsmMacroParameter Parameter; 5633 MCAsmMacroArguments A; 5634 if (check(parseIdentifier(Parameter.Name), 5635 "expected identifier in '.irp' directive") || 5636 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5637 parseMacroArguments(nullptr, A) || parseEOL()) 5638 return true; 5639 5640 // Lex the irp definition. 5641 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5642 if (!M) 5643 return true; 5644 5645 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5646 // to hold the macro body with substitutions. 5647 SmallString<256> Buf; 5648 raw_svector_ostream OS(Buf); 5649 5650 for (const MCAsmMacroArgument &Arg : A) { 5651 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5652 // This is undocumented, but GAS seems to support it. 5653 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5654 return true; 5655 } 5656 5657 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5658 5659 return false; 5660 } 5661 5662 /// parseDirectiveIrpc 5663 /// ::= .irpc symbol,values 5664 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5665 MCAsmMacroParameter Parameter; 5666 MCAsmMacroArguments A; 5667 5668 if (check(parseIdentifier(Parameter.Name), 5669 "expected identifier in '.irpc' directive") || 5670 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5671 parseMacroArguments(nullptr, A)) 5672 return true; 5673 5674 if (A.size() != 1 || A.front().size() != 1) 5675 return TokError("unexpected token in '.irpc' directive"); 5676 if (parseEOL()) 5677 return true; 5678 5679 // Lex the irpc definition. 5680 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5681 if (!M) 5682 return true; 5683 5684 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5685 // to hold the macro body with substitutions. 5686 SmallString<256> Buf; 5687 raw_svector_ostream OS(Buf); 5688 5689 StringRef Values = A.front().front().getString(); 5690 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5691 MCAsmMacroArgument Arg; 5692 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5693 5694 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5695 // This is undocumented, but GAS seems to support it. 5696 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5697 return true; 5698 } 5699 5700 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5701 5702 return false; 5703 } 5704 5705 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5706 if (ActiveMacros.empty()) 5707 return TokError("unmatched '.endr' directive"); 5708 5709 // The only .repl that should get here are the ones created by 5710 // instantiateMacroLikeBody. 5711 assert(getLexer().is(AsmToken::EndOfStatement)); 5712 5713 handleMacroExit(); 5714 return false; 5715 } 5716 5717 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5718 size_t Len) { 5719 const MCExpr *Value; 5720 SMLoc ExprLoc = getLexer().getLoc(); 5721 if (parseExpression(Value)) 5722 return true; 5723 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5724 if (!MCE) 5725 return Error(ExprLoc, "unexpected expression in _emit"); 5726 uint64_t IntValue = MCE->getValue(); 5727 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5728 return Error(ExprLoc, "literal value out of range for directive"); 5729 5730 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5731 return false; 5732 } 5733 5734 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5735 const MCExpr *Value; 5736 SMLoc ExprLoc = getLexer().getLoc(); 5737 if (parseExpression(Value)) 5738 return true; 5739 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5740 if (!MCE) 5741 return Error(ExprLoc, "unexpected expression in align"); 5742 uint64_t IntValue = MCE->getValue(); 5743 if (!isPowerOf2_64(IntValue)) 5744 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5745 5746 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5747 return false; 5748 } 5749 5750 bool AsmParser::parseDirectivePrint(SMLoc DirectiveLoc) { 5751 const AsmToken StrTok = getTok(); 5752 Lex(); 5753 if (StrTok.isNot(AsmToken::String) || StrTok.getString().front() != '"') 5754 return Error(DirectiveLoc, "expected double quoted string after .print"); 5755 if (parseEOL()) 5756 return true; 5757 llvm::outs() << StrTok.getStringContents() << '\n'; 5758 return false; 5759 } 5760 5761 bool AsmParser::parseDirectiveAddrsig() { 5762 if (parseEOL()) 5763 return true; 5764 getStreamer().emitAddrsig(); 5765 return false; 5766 } 5767 5768 bool AsmParser::parseDirectiveAddrsigSym() { 5769 StringRef Name; 5770 if (check(parseIdentifier(Name), "expected identifier") || parseEOL()) 5771 return true; 5772 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 5773 getStreamer().emitAddrsigSym(Sym); 5774 return false; 5775 } 5776 5777 bool AsmParser::parseDirectivePseudoProbe() { 5778 int64_t Guid; 5779 int64_t Index; 5780 int64_t Type; 5781 int64_t Attr; 5782 5783 if (getLexer().is(AsmToken::Integer)) { 5784 if (parseIntToken(Guid, "unexpected token in '.pseudoprobe' directive")) 5785 return true; 5786 } 5787 5788 if (getLexer().is(AsmToken::Integer)) { 5789 if (parseIntToken(Index, "unexpected token in '.pseudoprobe' directive")) 5790 return true; 5791 } 5792 5793 if (getLexer().is(AsmToken::Integer)) { 5794 if (parseIntToken(Type, "unexpected token in '.pseudoprobe' directive")) 5795 return true; 5796 } 5797 5798 if (getLexer().is(AsmToken::Integer)) { 5799 if (parseIntToken(Attr, "unexpected token in '.pseudoprobe' directive")) 5800 return true; 5801 } 5802 5803 // Parse inline stack like @ GUID:11:12 @ GUID:1:11 @ GUID:3:21 5804 MCPseudoProbeInlineStack InlineStack; 5805 5806 while (getLexer().is(AsmToken::At)) { 5807 // eat @ 5808 Lex(); 5809 5810 int64_t CallerGuid = 0; 5811 if (getLexer().is(AsmToken::Integer)) { 5812 if (parseIntToken(CallerGuid, 5813 "unexpected token in '.pseudoprobe' directive")) 5814 return true; 5815 } 5816 5817 // eat colon 5818 if (getLexer().is(AsmToken::Colon)) 5819 Lex(); 5820 5821 int64_t CallerProbeId = 0; 5822 if (getLexer().is(AsmToken::Integer)) { 5823 if (parseIntToken(CallerProbeId, 5824 "unexpected token in '.pseudoprobe' directive")) 5825 return true; 5826 } 5827 5828 InlineSite Site(CallerGuid, CallerProbeId); 5829 InlineStack.push_back(Site); 5830 } 5831 5832 if (parseEOL()) 5833 return true; 5834 5835 getStreamer().emitPseudoProbe(Guid, Index, Type, Attr, InlineStack); 5836 return false; 5837 } 5838 5839 /// parseDirectiveLTODiscard 5840 /// ::= ".lto_discard" [ identifier ( , identifier )* ] 5841 /// The LTO library emits this directive to discard non-prevailing symbols. 5842 /// We ignore symbol assignments and attribute changes for the specified 5843 /// symbols. 5844 bool AsmParser::parseDirectiveLTODiscard() { 5845 auto ParseOp = [&]() -> bool { 5846 StringRef Name; 5847 SMLoc Loc = getTok().getLoc(); 5848 if (parseIdentifier(Name)) 5849 return Error(Loc, "expected identifier"); 5850 LTODiscardSymbols.insert(Name); 5851 return false; 5852 }; 5853 5854 LTODiscardSymbols.clear(); 5855 if (parseMany(ParseOp)) 5856 return addErrorSuffix(" in directive"); 5857 return false; 5858 } 5859 5860 // We are comparing pointers, but the pointers are relative to a single string. 5861 // Thus, this should always be deterministic. 5862 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5863 const AsmRewrite *AsmRewriteB) { 5864 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5865 return -1; 5866 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5867 return 1; 5868 5869 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5870 // rewrite to the same location. Make sure the SizeDirective rewrite is 5871 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5872 // ensures the sort algorithm is stable. 5873 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5874 AsmRewritePrecedence[AsmRewriteB->Kind]) 5875 return -1; 5876 5877 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5878 AsmRewritePrecedence[AsmRewriteB->Kind]) 5879 return 1; 5880 llvm_unreachable("Unstable rewrite sort."); 5881 } 5882 5883 bool AsmParser::parseMSInlineAsm( 5884 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5885 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5886 SmallVectorImpl<std::string> &Constraints, 5887 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5888 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5889 SmallVector<void *, 4> InputDecls; 5890 SmallVector<void *, 4> OutputDecls; 5891 SmallVector<bool, 4> InputDeclsAddressOf; 5892 SmallVector<bool, 4> OutputDeclsAddressOf; 5893 SmallVector<std::string, 4> InputConstraints; 5894 SmallVector<std::string, 4> OutputConstraints; 5895 SmallVector<unsigned, 4> ClobberRegs; 5896 5897 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5898 5899 // Prime the lexer. 5900 Lex(); 5901 5902 // While we have input, parse each statement. 5903 unsigned InputIdx = 0; 5904 unsigned OutputIdx = 0; 5905 while (getLexer().isNot(AsmToken::Eof)) { 5906 // Parse curly braces marking block start/end 5907 if (parseCurlyBlockScope(AsmStrRewrites)) 5908 continue; 5909 5910 ParseStatementInfo Info(&AsmStrRewrites); 5911 bool StatementErr = parseStatement(Info, &SI); 5912 5913 if (StatementErr || Info.ParseError) { 5914 // Emit pending errors if any exist. 5915 printPendingErrors(); 5916 return true; 5917 } 5918 5919 // No pending error should exist here. 5920 assert(!hasPendingError() && "unexpected error from parseStatement"); 5921 5922 if (Info.Opcode == ~0U) 5923 continue; 5924 5925 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5926 5927 // Build the list of clobbers, outputs and inputs. 5928 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5929 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5930 5931 // Register operand. 5932 if (Operand.isReg() && !Operand.needAddressOf() && 5933 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5934 unsigned NumDefs = Desc.getNumDefs(); 5935 // Clobber. 5936 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5937 ClobberRegs.push_back(Operand.getReg()); 5938 continue; 5939 } 5940 5941 // Expr/Input or Output. 5942 StringRef SymName = Operand.getSymName(); 5943 if (SymName.empty()) 5944 continue; 5945 5946 void *OpDecl = Operand.getOpDecl(); 5947 if (!OpDecl) 5948 continue; 5949 5950 StringRef Constraint = Operand.getConstraint(); 5951 if (Operand.isImm()) { 5952 // Offset as immediate 5953 if (Operand.isOffsetOfLocal()) 5954 Constraint = "r"; 5955 else 5956 Constraint = "i"; 5957 } 5958 5959 bool isOutput = (i == 1) && Desc.mayStore(); 5960 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5961 if (isOutput) { 5962 ++InputIdx; 5963 OutputDecls.push_back(OpDecl); 5964 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5965 OutputConstraints.push_back(("=" + Constraint).str()); 5966 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5967 } else { 5968 InputDecls.push_back(OpDecl); 5969 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5970 InputConstraints.push_back(Constraint.str()); 5971 if (Desc.OpInfo[i - 1].isBranchTarget()) 5972 AsmStrRewrites.emplace_back(AOK_CallInput, Start, SymName.size()); 5973 else 5974 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5975 } 5976 } 5977 5978 // Consider implicit defs to be clobbers. Think of cpuid and push. 5979 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5980 Desc.getNumImplicitDefs()); 5981 llvm::append_range(ClobberRegs, ImpDefs); 5982 } 5983 5984 // Set the number of Outputs and Inputs. 5985 NumOutputs = OutputDecls.size(); 5986 NumInputs = InputDecls.size(); 5987 5988 // Set the unique clobbers. 5989 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5990 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5991 ClobberRegs.end()); 5992 Clobbers.assign(ClobberRegs.size(), std::string()); 5993 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5994 raw_string_ostream OS(Clobbers[I]); 5995 IP->printRegName(OS, ClobberRegs[I]); 5996 } 5997 5998 // Merge the various outputs and inputs. Output are expected first. 5999 if (NumOutputs || NumInputs) { 6000 unsigned NumExprs = NumOutputs + NumInputs; 6001 OpDecls.resize(NumExprs); 6002 Constraints.resize(NumExprs); 6003 for (unsigned i = 0; i < NumOutputs; ++i) { 6004 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 6005 Constraints[i] = OutputConstraints[i]; 6006 } 6007 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 6008 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 6009 Constraints[j] = InputConstraints[i]; 6010 } 6011 } 6012 6013 // Build the IR assembly string. 6014 std::string AsmStringIR; 6015 raw_string_ostream OS(AsmStringIR); 6016 StringRef ASMString = 6017 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 6018 const char *AsmStart = ASMString.begin(); 6019 const char *AsmEnd = ASMString.end(); 6020 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 6021 for (auto it = AsmStrRewrites.begin(); it != AsmStrRewrites.end(); ++it) { 6022 const AsmRewrite &AR = *it; 6023 // Check if this has already been covered by another rewrite... 6024 if (AR.Done) 6025 continue; 6026 AsmRewriteKind Kind = AR.Kind; 6027 6028 const char *Loc = AR.Loc.getPointer(); 6029 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 6030 6031 // Emit everything up to the immediate/expression. 6032 if (unsigned Len = Loc - AsmStart) 6033 OS << StringRef(AsmStart, Len); 6034 6035 // Skip the original expression. 6036 if (Kind == AOK_Skip) { 6037 AsmStart = Loc + AR.Len; 6038 continue; 6039 } 6040 6041 unsigned AdditionalSkip = 0; 6042 // Rewrite expressions in $N notation. 6043 switch (Kind) { 6044 default: 6045 break; 6046 case AOK_IntelExpr: 6047 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 6048 if (AR.IntelExp.NeedBracs) 6049 OS << "["; 6050 if (AR.IntelExp.hasBaseReg()) 6051 OS << AR.IntelExp.BaseReg; 6052 if (AR.IntelExp.hasIndexReg()) 6053 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 6054 << AR.IntelExp.IndexReg; 6055 if (AR.IntelExp.Scale > 1) 6056 OS << " * $$" << AR.IntelExp.Scale; 6057 if (AR.IntelExp.hasOffset()) { 6058 if (AR.IntelExp.hasRegs()) 6059 OS << " + "; 6060 // Fuse this rewrite with a rewrite of the offset name, if present. 6061 StringRef OffsetName = AR.IntelExp.OffsetName; 6062 SMLoc OffsetLoc = SMLoc::getFromPointer(AR.IntelExp.OffsetName.data()); 6063 size_t OffsetLen = OffsetName.size(); 6064 auto rewrite_it = std::find_if( 6065 it, AsmStrRewrites.end(), [&](const AsmRewrite &FusingAR) { 6066 return FusingAR.Loc == OffsetLoc && FusingAR.Len == OffsetLen && 6067 (FusingAR.Kind == AOK_Input || 6068 FusingAR.Kind == AOK_CallInput); 6069 }); 6070 if (rewrite_it == AsmStrRewrites.end()) { 6071 OS << "offset " << OffsetName; 6072 } else if (rewrite_it->Kind == AOK_CallInput) { 6073 OS << "${" << InputIdx++ << ":P}"; 6074 rewrite_it->Done = true; 6075 } else { 6076 OS << '$' << InputIdx++; 6077 rewrite_it->Done = true; 6078 } 6079 } 6080 if (AR.IntelExp.Imm || AR.IntelExp.emitImm()) 6081 OS << (AR.IntelExp.emitImm() ? "$$" : " + $$") << AR.IntelExp.Imm; 6082 if (AR.IntelExp.NeedBracs) 6083 OS << "]"; 6084 break; 6085 case AOK_Label: 6086 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 6087 break; 6088 case AOK_Input: 6089 OS << '$' << InputIdx++; 6090 break; 6091 case AOK_CallInput: 6092 OS << "${" << InputIdx++ << ":P}"; 6093 break; 6094 case AOK_Output: 6095 OS << '$' << OutputIdx++; 6096 break; 6097 case AOK_SizeDirective: 6098 switch (AR.Val) { 6099 default: break; 6100 case 8: OS << "byte ptr "; break; 6101 case 16: OS << "word ptr "; break; 6102 case 32: OS << "dword ptr "; break; 6103 case 64: OS << "qword ptr "; break; 6104 case 80: OS << "xword ptr "; break; 6105 case 128: OS << "xmmword ptr "; break; 6106 case 256: OS << "ymmword ptr "; break; 6107 } 6108 break; 6109 case AOK_Emit: 6110 OS << ".byte"; 6111 break; 6112 case AOK_Align: { 6113 // MS alignment directives are measured in bytes. If the native assembler 6114 // measures alignment in bytes, we can pass it straight through. 6115 OS << ".align"; 6116 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 6117 break; 6118 6119 // Alignment is in log2 form, so print that instead and skip the original 6120 // immediate. 6121 unsigned Val = AR.Val; 6122 OS << ' ' << Val; 6123 assert(Val < 10 && "Expected alignment less then 2^10."); 6124 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 6125 break; 6126 } 6127 case AOK_EVEN: 6128 OS << ".even"; 6129 break; 6130 case AOK_EndOfStatement: 6131 OS << "\n\t"; 6132 break; 6133 } 6134 6135 // Skip the original expression. 6136 AsmStart = Loc + AR.Len + AdditionalSkip; 6137 } 6138 6139 // Emit the remainder of the asm string. 6140 if (AsmStart != AsmEnd) 6141 OS << StringRef(AsmStart, AsmEnd - AsmStart); 6142 6143 AsmString = OS.str(); 6144 return false; 6145 } 6146 6147 namespace llvm { 6148 namespace MCParserUtils { 6149 6150 /// Returns whether the given symbol is used anywhere in the given expression, 6151 /// or subexpressions. 6152 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 6153 switch (Value->getKind()) { 6154 case MCExpr::Binary: { 6155 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 6156 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 6157 isSymbolUsedInExpression(Sym, BE->getRHS()); 6158 } 6159 case MCExpr::Target: 6160 case MCExpr::Constant: 6161 return false; 6162 case MCExpr::SymbolRef: { 6163 const MCSymbol &S = 6164 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 6165 if (S.isVariable()) 6166 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 6167 return &S == Sym; 6168 } 6169 case MCExpr::Unary: 6170 return isSymbolUsedInExpression( 6171 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 6172 } 6173 6174 llvm_unreachable("Unknown expr kind!"); 6175 } 6176 6177 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 6178 MCAsmParser &Parser, MCSymbol *&Sym, 6179 const MCExpr *&Value) { 6180 6181 // FIXME: Use better location, we should use proper tokens. 6182 SMLoc EqualLoc = Parser.getTok().getLoc(); 6183 if (Parser.parseExpression(Value)) 6184 return Parser.TokError("missing expression"); 6185 6186 // Note: we don't count b as used in "a = b". This is to allow 6187 // a = b 6188 // b = c 6189 6190 if (Parser.parseEOL()) 6191 return true; 6192 6193 // Validate that the LHS is allowed to be a variable (either it has not been 6194 // used as a symbol, or it is an absolute symbol). 6195 Sym = Parser.getContext().lookupSymbol(Name); 6196 if (Sym) { 6197 // Diagnose assignment to a label. 6198 // 6199 // FIXME: Diagnostics. Note the location of the definition as a label. 6200 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 6201 if (isSymbolUsedInExpression(Sym, Value)) 6202 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 6203 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 6204 !Sym->isVariable()) 6205 ; // Allow redefinitions of undefined symbols only used in directives. 6206 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 6207 ; // Allow redefinitions of variables that haven't yet been used. 6208 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 6209 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 6210 else if (!Sym->isVariable()) 6211 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 6212 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 6213 return Parser.Error(EqualLoc, 6214 "invalid reassignment of non-absolute variable '" + 6215 Name + "'"); 6216 } else if (Name == ".") { 6217 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 6218 return false; 6219 } else 6220 Sym = Parser.getContext().getOrCreateSymbol(Name); 6221 6222 Sym->setRedefinable(allow_redef); 6223 6224 return false; 6225 } 6226 6227 } // end namespace MCParserUtils 6228 } // end namespace llvm 6229 6230 /// Create an MCAsmParser instance. 6231 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 6232 MCStreamer &Out, const MCAsmInfo &MAI, 6233 unsigned CB) { 6234 return new AsmParser(SM, C, Out, MAI, CB); 6235 } 6236