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