1 //===- AsmParser.cpp - Parser for Assembly Files --------------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // This class implements the parser for assembly files. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "llvm/ADT/APFloat.h" 15 #include "llvm/ADT/STLExtras.h" 16 #include "llvm/ADT/SmallString.h" 17 #include "llvm/ADT/StringMap.h" 18 #include "llvm/ADT/Twine.h" 19 #include "llvm/MC/MCAsmInfo.h" 20 #include "llvm/MC/MCContext.h" 21 #include "llvm/MC/MCDwarf.h" 22 #include "llvm/MC/MCExpr.h" 23 #include "llvm/MC/MCInstPrinter.h" 24 #include "llvm/MC/MCInstrInfo.h" 25 #include "llvm/MC/MCObjectFileInfo.h" 26 #include "llvm/MC/MCParser/AsmCond.h" 27 #include "llvm/MC/MCParser/AsmLexer.h" 28 #include "llvm/MC/MCParser/MCAsmParser.h" 29 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 30 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 31 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 32 #include "llvm/MC/MCRegisterInfo.h" 33 #include "llvm/MC/MCSectionMachO.h" 34 #include "llvm/MC/MCStreamer.h" 35 #include "llvm/MC/MCSymbol.h" 36 #include "llvm/MC/MCValue.h" 37 #include "llvm/Support/ErrorHandling.h" 38 #include "llvm/Support/MathExtras.h" 39 #include "llvm/Support/MemoryBuffer.h" 40 #include "llvm/Support/SourceMgr.h" 41 #include "llvm/Support/raw_ostream.h" 42 #include <cctype> 43 #include <deque> 44 #include <string> 45 #include <vector> 46 using namespace llvm; 47 48 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() {} 49 50 namespace { 51 /// \brief Helper types for tracking macro definitions. 52 typedef std::vector<AsmToken> MCAsmMacroArgument; 53 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 54 55 struct MCAsmMacroParameter { 56 StringRef Name; 57 MCAsmMacroArgument Value; 58 bool Required; 59 bool Vararg; 60 61 MCAsmMacroParameter() : Required(false), Vararg(false) {} 62 }; 63 64 typedef std::vector<MCAsmMacroParameter> MCAsmMacroParameters; 65 66 struct MCAsmMacro { 67 StringRef Name; 68 StringRef Body; 69 MCAsmMacroParameters Parameters; 70 71 public: 72 MCAsmMacro(StringRef N, StringRef B, MCAsmMacroParameters P) 73 : Name(N), Body(B), Parameters(std::move(P)) {} 74 }; 75 76 /// \brief Helper class for storing information about an active macro 77 /// instantiation. 78 struct MacroInstantiation { 79 /// The location of the instantiation. 80 SMLoc InstantiationLoc; 81 82 /// The buffer where parsing should resume upon instantiation completion. 83 int ExitBuffer; 84 85 /// The location where parsing should resume upon instantiation completion. 86 SMLoc ExitLoc; 87 88 /// The depth of TheCondStack at the start of the instantiation. 89 size_t CondStackDepth; 90 91 public: 92 MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth); 93 }; 94 95 struct ParseStatementInfo { 96 /// \brief The parsed operands from the last parsed statement. 97 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 98 99 /// \brief The opcode from the last parsed instruction. 100 unsigned Opcode; 101 102 /// \brief Was there an error parsing the inline assembly? 103 bool ParseError; 104 105 SmallVectorImpl<AsmRewrite> *AsmRewrites; 106 107 ParseStatementInfo() : Opcode(~0U), ParseError(false), AsmRewrites(nullptr) {} 108 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 109 : Opcode(~0), ParseError(false), AsmRewrites(rewrites) {} 110 }; 111 112 /// \brief The concrete assembly parser instance. 113 class AsmParser : public MCAsmParser { 114 AsmParser(const AsmParser &) = delete; 115 void operator=(const AsmParser &) = delete; 116 private: 117 AsmLexer Lexer; 118 MCContext &Ctx; 119 MCStreamer &Out; 120 const MCAsmInfo &MAI; 121 SourceMgr &SrcMgr; 122 SourceMgr::DiagHandlerTy SavedDiagHandler; 123 void *SavedDiagContext; 124 std::unique_ptr<MCAsmParserExtension> PlatformParser; 125 126 /// This is the current buffer index we're lexing from as managed by the 127 /// SourceMgr object. 128 unsigned CurBuffer; 129 130 AsmCond TheCondState; 131 std::vector<AsmCond> TheCondStack; 132 133 /// \brief maps directive names to handler methods in parser 134 /// extensions. Extensions register themselves in this map by calling 135 /// addDirectiveHandler. 136 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 137 138 /// \brief Map of currently defined macros. 139 StringMap<MCAsmMacro> MacroMap; 140 141 /// \brief Stack of active macro instantiations. 142 std::vector<MacroInstantiation*> ActiveMacros; 143 144 /// \brief List of bodies of anonymous macros. 145 std::deque<MCAsmMacro> MacroLikeBodies; 146 147 /// Boolean tracking whether macro substitution is enabled. 148 unsigned MacrosEnabledFlag : 1; 149 150 /// \brief Keeps track of how many .macro's have been instantiated. 151 unsigned NumOfMacroInstantiations; 152 153 /// Flag tracking whether any errors have been encountered. 154 unsigned HadError : 1; 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 = 0; 160 SMLoc Loc; 161 unsigned Buf = 0; 162 }; 163 CppHashInfoTy CppHashInfo; 164 165 /// \brief List of forward directional labels for diagnosis at the end. 166 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 167 168 /// When generating dwarf for assembly source files we need to calculate the 169 /// logical line number based on the last parsed cpp hash file line comment 170 /// and current line. Since this is slow and messes up the SourceMgr's 171 /// cache we save the last info we queried with SrcMgr.FindLineNumber(). 172 SMLoc LastQueryIDLoc; 173 unsigned LastQueryBuffer; 174 unsigned LastQueryLine; 175 176 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 177 unsigned AssemblerDialect; 178 179 /// \brief is Darwin compatibility enabled? 180 bool IsDarwin; 181 182 /// \brief Are we parsing ms-style inline assembly? 183 bool ParsingInlineAsm; 184 185 public: 186 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 187 const MCAsmInfo &MAI); 188 ~AsmParser() override; 189 190 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 191 192 void addDirectiveHandler(StringRef Directive, 193 ExtensionDirectiveHandler Handler) override { 194 ExtensionDirectiveMap[Directive] = Handler; 195 } 196 197 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 198 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 199 } 200 201 public: 202 /// @name MCAsmParser Interface 203 /// { 204 205 SourceMgr &getSourceManager() override { return SrcMgr; } 206 MCAsmLexer &getLexer() override { return Lexer; } 207 MCContext &getContext() override { return Ctx; } 208 MCStreamer &getStreamer() override { return Out; } 209 unsigned getAssemblerDialect() override { 210 if (AssemblerDialect == ~0U) 211 return MAI.getAssemblerDialect(); 212 else 213 return AssemblerDialect; 214 } 215 void setAssemblerDialect(unsigned i) override { 216 AssemblerDialect = i; 217 } 218 219 void Note(SMLoc L, const Twine &Msg, 220 ArrayRef<SMRange> Ranges = None) override; 221 bool Warning(SMLoc L, const Twine &Msg, 222 ArrayRef<SMRange> Ranges = None) override; 223 bool Error(SMLoc L, const Twine &Msg, 224 ArrayRef<SMRange> Ranges = None) override; 225 226 const AsmToken &Lex() override; 227 228 void setParsingInlineAsm(bool V) override { ParsingInlineAsm = V; } 229 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 230 231 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 232 unsigned &NumOutputs, unsigned &NumInputs, 233 SmallVectorImpl<std::pair<void *,bool> > &OpDecls, 234 SmallVectorImpl<std::string> &Constraints, 235 SmallVectorImpl<std::string> &Clobbers, 236 const MCInstrInfo *MII, const MCInstPrinter *IP, 237 MCAsmParserSemaCallback &SI) override; 238 239 bool parseExpression(const MCExpr *&Res); 240 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 241 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 242 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 243 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 244 SMLoc &EndLoc) override; 245 bool parseAbsoluteExpression(int64_t &Res) override; 246 247 /// \brief Parse an identifier or string (as a quoted identifier) 248 /// and set \p Res to the identifier contents. 249 bool parseIdentifier(StringRef &Res) override; 250 void eatToEndOfStatement() override; 251 252 void checkForValidSection() override; 253 254 bool getTokenLoc(SMLoc &Loc) { 255 Loc = getTok().getLoc(); 256 return false; 257 } 258 259 /// parseToken - If current token has the specified kind, eat it and 260 /// return success. Otherwise, emit the specified error and return failure. 261 bool parseToken(AsmToken::TokenKind T, const Twine &ErrMsg) { 262 if (getTok().getKind() != T) 263 return TokError(ErrMsg); 264 Lex(); 265 return false; 266 } 267 268 bool parseIntToken(int64_t &V, const Twine &ErrMsg) { 269 if (getTok().getKind() != AsmToken::Integer) 270 return TokError(ErrMsg); 271 V = getTok().getIntVal(); 272 Lex(); 273 return false; 274 } 275 276 /// } 277 278 private: 279 280 bool parseStatement(ParseStatementInfo &Info, 281 MCAsmParserSemaCallback *SI); 282 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 283 bool parseCppHashLineFilenameComment(SMLoc L); 284 285 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 286 ArrayRef<MCAsmMacroParameter> Parameters); 287 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 288 ArrayRef<MCAsmMacroParameter> Parameters, 289 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 290 SMLoc L); 291 292 /// \brief Are macros enabled in the parser? 293 bool areMacrosEnabled() {return MacrosEnabledFlag;} 294 295 /// \brief Control a flag in the parser that enables or disables macros. 296 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 297 298 /// \brief Lookup a previously defined macro. 299 /// \param Name Macro name. 300 /// \returns Pointer to macro. NULL if no such macro was defined. 301 const MCAsmMacro* lookupMacro(StringRef Name); 302 303 /// \brief Define a new macro with the given name and information. 304 void defineMacro(StringRef Name, MCAsmMacro Macro); 305 306 /// \brief Undefine a macro. If no such macro was defined, it's a no-op. 307 void undefineMacro(StringRef Name); 308 309 /// \brief Are we inside a macro instantiation? 310 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 311 312 /// \brief Handle entry to macro instantiation. 313 /// 314 /// \param M The macro. 315 /// \param NameLoc Instantiation location. 316 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 317 318 /// \brief Handle exit from macro instantiation. 319 void handleMacroExit(); 320 321 /// \brief Extract AsmTokens for a macro argument. 322 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 323 324 /// \brief Parse all macro arguments for a given macro. 325 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 326 327 void printMacroInstantiations(); 328 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 329 ArrayRef<SMRange> Ranges = None) const { 330 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 331 } 332 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 333 334 bool check(bool P, SMLoc Loc, const Twine &Msg) { 335 if (P) 336 return Error(Loc, Msg); 337 return false; 338 } 339 340 bool check(bool P, const Twine &Msg) { 341 if (P) 342 return TokError(Msg); 343 return false; 344 } 345 346 /// \brief Enter the specified file. This returns true on failure. 347 bool enterIncludeFile(const std::string &Filename); 348 349 /// \brief Process the specified file for the .incbin directive. 350 /// This returns true on failure. 351 bool processIncbinFile(const std::string &Filename); 352 353 /// \brief Reset the current lexer position to that given by \p Loc. The 354 /// current token is not set; clients should ensure Lex() is called 355 /// subsequently. 356 /// 357 /// \param InBuffer If not 0, should be the known buffer id that contains the 358 /// location. 359 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 360 361 /// \brief Parse up to the end of statement and a return the contents from the 362 /// current token until the end of the statement; the current token on exit 363 /// will be either the EndOfStatement or EOF. 364 StringRef parseStringToEndOfStatement() override; 365 366 /// \brief Parse until the end of a statement or a comma is encountered, 367 /// return the contents from the current token up to the end or comma. 368 StringRef parseStringToComma(); 369 370 bool parseAssignment(StringRef Name, bool allow_redef, 371 bool NoDeadStrip = false); 372 373 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 374 MCBinaryExpr::Opcode &Kind); 375 376 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 377 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 378 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 379 380 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 381 382 // Generic (target and platform independent) directive parsing. 383 enum DirectiveKind { 384 DK_NO_DIRECTIVE, // Placeholder 385 DK_SET, DK_EQU, DK_EQUIV, DK_ASCII, DK_ASCIZ, DK_STRING, DK_BYTE, DK_SHORT, 386 DK_RELOC, 387 DK_VALUE, DK_2BYTE, DK_LONG, DK_INT, DK_4BYTE, DK_QUAD, DK_8BYTE, DK_OCTA, 388 DK_SINGLE, DK_FLOAT, DK_DOUBLE, DK_ALIGN, DK_ALIGN32, DK_BALIGN, DK_BALIGNW, 389 DK_BALIGNL, DK_P2ALIGN, DK_P2ALIGNW, DK_P2ALIGNL, DK_ORG, DK_FILL, DK_ENDR, 390 DK_BUNDLE_ALIGN_MODE, DK_BUNDLE_LOCK, DK_BUNDLE_UNLOCK, 391 DK_ZERO, DK_EXTERN, DK_GLOBL, DK_GLOBAL, 392 DK_LAZY_REFERENCE, DK_NO_DEAD_STRIP, DK_SYMBOL_RESOLVER, 393 DK_PRIVATE_EXTERN, DK_REFERENCE, DK_WEAK_DEFINITION, DK_WEAK_REFERENCE, 394 DK_WEAK_DEF_CAN_BE_HIDDEN, DK_COMM, DK_COMMON, DK_LCOMM, DK_ABORT, 395 DK_INCLUDE, DK_INCBIN, DK_CODE16, DK_CODE16GCC, DK_REPT, DK_IRP, DK_IRPC, 396 DK_IF, DK_IFEQ, DK_IFGE, DK_IFGT, DK_IFLE, DK_IFLT, DK_IFNE, DK_IFB, 397 DK_IFNB, DK_IFC, DK_IFEQS, DK_IFNC, DK_IFNES, DK_IFDEF, DK_IFNDEF, 398 DK_IFNOTDEF, DK_ELSEIF, DK_ELSE, DK_ENDIF, 399 DK_SPACE, DK_SKIP, DK_FILE, DK_LINE, DK_LOC, DK_STABS, 400 DK_CV_FILE, DK_CV_LOC, DK_CV_LINETABLE, DK_CV_INLINE_LINETABLE, 401 DK_CV_DEF_RANGE, DK_CV_STRINGTABLE, DK_CV_FILECHECKSUMS, 402 DK_CFI_SECTIONS, DK_CFI_STARTPROC, DK_CFI_ENDPROC, DK_CFI_DEF_CFA, 403 DK_CFI_DEF_CFA_OFFSET, DK_CFI_ADJUST_CFA_OFFSET, DK_CFI_DEF_CFA_REGISTER, 404 DK_CFI_OFFSET, DK_CFI_REL_OFFSET, DK_CFI_PERSONALITY, DK_CFI_LSDA, 405 DK_CFI_REMEMBER_STATE, DK_CFI_RESTORE_STATE, DK_CFI_SAME_VALUE, 406 DK_CFI_RESTORE, DK_CFI_ESCAPE, DK_CFI_SIGNAL_FRAME, DK_CFI_UNDEFINED, 407 DK_CFI_REGISTER, DK_CFI_WINDOW_SAVE, 408 DK_MACROS_ON, DK_MACROS_OFF, 409 DK_MACRO, DK_EXITM, DK_ENDM, DK_ENDMACRO, DK_PURGEM, 410 DK_SLEB128, DK_ULEB128, 411 DK_ERR, DK_ERROR, DK_WARNING, 412 DK_END 413 }; 414 415 /// \brief Maps directive name --> DirectiveKind enum, for 416 /// directives parsed by this class. 417 StringMap<DirectiveKind> DirectiveKindMap; 418 419 // ".ascii", ".asciz", ".string" 420 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 421 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 422 bool parseDirectiveValue(unsigned Size); // ".byte", ".long", ... 423 bool parseDirectiveOctaValue(); // ".octa" 424 bool parseDirectiveRealValue(const fltSemantics &); // ".single", ... 425 bool parseDirectiveFill(); // ".fill" 426 bool parseDirectiveZero(); // ".zero" 427 // ".set", ".equ", ".equiv" 428 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 429 bool parseDirectiveOrg(); // ".org" 430 // ".align{,32}", ".p2align{,w,l}" 431 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 432 433 // ".file", ".line", ".loc", ".stabs" 434 bool parseDirectiveFile(SMLoc DirectiveLoc); 435 bool parseDirectiveLine(); 436 bool parseDirectiveLoc(); 437 bool parseDirectiveStabs(); 438 439 // ".cv_file", ".cv_loc", ".cv_linetable", "cv_inline_linetable", 440 // ".cv_def_range" 441 bool parseDirectiveCVFile(); 442 bool parseDirectiveCVLoc(); 443 bool parseDirectiveCVLinetable(); 444 bool parseDirectiveCVInlineLinetable(); 445 bool parseDirectiveCVDefRange(); 446 bool parseDirectiveCVStringTable(); 447 bool parseDirectiveCVFileChecksums(); 448 449 // .cfi directives 450 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 451 bool parseDirectiveCFIWindowSave(); 452 bool parseDirectiveCFISections(); 453 bool parseDirectiveCFIStartProc(); 454 bool parseDirectiveCFIEndProc(); 455 bool parseDirectiveCFIDefCfaOffset(); 456 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 457 bool parseDirectiveCFIAdjustCfaOffset(); 458 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 459 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 460 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 461 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 462 bool parseDirectiveCFIRememberState(); 463 bool parseDirectiveCFIRestoreState(); 464 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 465 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 466 bool parseDirectiveCFIEscape(); 467 bool parseDirectiveCFISignalFrame(); 468 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 469 470 // macro directives 471 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 472 bool parseDirectiveExitMacro(StringRef Directive); 473 bool parseDirectiveEndMacro(StringRef Directive); 474 bool parseDirectiveMacro(SMLoc DirectiveLoc); 475 bool parseDirectiveMacrosOnOff(StringRef Directive); 476 477 // ".bundle_align_mode" 478 bool parseDirectiveBundleAlignMode(); 479 // ".bundle_lock" 480 bool parseDirectiveBundleLock(); 481 // ".bundle_unlock" 482 bool parseDirectiveBundleUnlock(); 483 484 // ".space", ".skip" 485 bool parseDirectiveSpace(StringRef IDVal); 486 487 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 488 bool parseDirectiveLEB128(bool Signed); 489 490 /// \brief Parse a directive like ".globl" which 491 /// accepts a single symbol (which should be a label or an external). 492 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 493 494 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 495 496 bool parseDirectiveAbort(); // ".abort" 497 bool parseDirectiveInclude(); // ".include" 498 bool parseDirectiveIncbin(); // ".incbin" 499 500 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 501 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 502 // ".ifb" or ".ifnb", depending on ExpectBlank. 503 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 504 // ".ifc" or ".ifnc", depending on ExpectEqual. 505 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 506 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 507 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 508 // ".ifdef" or ".ifndef", depending on expect_defined 509 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 510 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 511 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 512 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 513 bool parseEscapedString(std::string &Data) override; 514 515 const MCExpr *applyModifierToExpr(const MCExpr *E, 516 MCSymbolRefExpr::VariantKind Variant); 517 518 // Macro-like directives 519 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 520 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 521 raw_svector_ostream &OS); 522 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 523 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 524 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 525 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 526 527 // "_emit" or "__emit" 528 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 529 size_t Len); 530 531 // "align" 532 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 533 534 // "end" 535 bool parseDirectiveEnd(SMLoc DirectiveLoc); 536 537 // ".err" or ".error" 538 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 539 540 // ".warning" 541 bool parseDirectiveWarning(SMLoc DirectiveLoc); 542 543 void initializeDirectiveKindMap(); 544 }; 545 } 546 547 namespace llvm { 548 549 extern MCAsmParserExtension *createDarwinAsmParser(); 550 extern MCAsmParserExtension *createELFAsmParser(); 551 extern MCAsmParserExtension *createCOFFAsmParser(); 552 553 } 554 555 enum { DEFAULT_ADDRSPACE = 0 }; 556 557 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 558 const MCAsmInfo &MAI) 559 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 560 PlatformParser(nullptr), CurBuffer(SM.getMainFileID()), 561 MacrosEnabledFlag(true), HadError(false), CppHashInfo(), 562 AssemblerDialect(~0U), IsDarwin(false), ParsingInlineAsm(false) { 563 // Save the old handler. 564 SavedDiagHandler = SrcMgr.getDiagHandler(); 565 SavedDiagContext = SrcMgr.getDiagContext(); 566 // Set our own handler which calls the saved handler. 567 SrcMgr.setDiagHandler(DiagHandler, this); 568 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 569 570 // Initialize the platform / file format parser. 571 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 572 case MCObjectFileInfo::IsCOFF: 573 PlatformParser.reset(createCOFFAsmParser()); 574 break; 575 case MCObjectFileInfo::IsMachO: 576 PlatformParser.reset(createDarwinAsmParser()); 577 IsDarwin = true; 578 break; 579 case MCObjectFileInfo::IsELF: 580 PlatformParser.reset(createELFAsmParser()); 581 break; 582 } 583 584 PlatformParser->Initialize(*this); 585 initializeDirectiveKindMap(); 586 587 NumOfMacroInstantiations = 0; 588 } 589 590 AsmParser::~AsmParser() { 591 assert((HadError || ActiveMacros.empty()) && 592 "Unexpected active macro instantiation!"); 593 } 594 595 void AsmParser::printMacroInstantiations() { 596 // Print the active macro instantiation stack. 597 for (std::vector<MacroInstantiation *>::const_reverse_iterator 598 it = ActiveMacros.rbegin(), 599 ie = ActiveMacros.rend(); 600 it != ie; ++it) 601 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 602 "while in macro instantiation"); 603 } 604 605 void AsmParser::Note(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 606 printMessage(L, SourceMgr::DK_Note, Msg, Ranges); 607 printMacroInstantiations(); 608 } 609 610 bool AsmParser::Warning(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 611 if(getTargetParser().getTargetOptions().MCNoWarn) 612 return false; 613 if (getTargetParser().getTargetOptions().MCFatalWarnings) 614 return Error(L, Msg, Ranges); 615 printMessage(L, SourceMgr::DK_Warning, Msg, Ranges); 616 printMacroInstantiations(); 617 return false; 618 } 619 620 bool AsmParser::Error(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 621 HadError = true; 622 printMessage(L, SourceMgr::DK_Error, Msg, Ranges); 623 printMacroInstantiations(); 624 return true; 625 } 626 627 bool AsmParser::enterIncludeFile(const std::string &Filename) { 628 std::string IncludedFile; 629 unsigned NewBuf = 630 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 631 if (!NewBuf) 632 return true; 633 634 CurBuffer = NewBuf; 635 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 636 return false; 637 } 638 639 /// Process the specified .incbin file by searching for it in the include paths 640 /// then just emitting the byte contents of the file to the streamer. This 641 /// returns true on failure. 642 bool AsmParser::processIncbinFile(const std::string &Filename) { 643 std::string IncludedFile; 644 unsigned NewBuf = 645 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 646 if (!NewBuf) 647 return true; 648 649 // Pick up the bytes from the file and emit them. 650 getStreamer().EmitBytes(SrcMgr.getMemoryBuffer(NewBuf)->getBuffer()); 651 return false; 652 } 653 654 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 655 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 656 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 657 Loc.getPointer()); 658 } 659 660 const AsmToken &AsmParser::Lex() { 661 if (Lexer.getTok().is(AsmToken::Error)) 662 Error(Lexer.getErrLoc(), Lexer.getErr()); 663 664 // if it's a end of statement with a comment in it 665 if (getTok().is(AsmToken::EndOfStatement)) { 666 // if this is a line comment output it. 667 if (getTok().getString().front() != '\n' && 668 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 669 Out.addExplicitComment(Twine(getTok().getString())); 670 } 671 672 const AsmToken *tok = &Lexer.Lex(); 673 674 // Parse comments here to be deferred until end of next statement. 675 while (tok->is(AsmToken::Comment)) { 676 if (MAI.preserveAsmComments()) 677 Out.addExplicitComment(Twine(tok->getString())); 678 tok = &Lexer.Lex(); 679 } 680 681 if (tok->is(AsmToken::Eof)) { 682 // If this is the end of an included file, pop the parent file off the 683 // include stack. 684 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 685 if (ParentIncludeLoc != SMLoc()) { 686 jumpToLoc(ParentIncludeLoc); 687 return Lex(); 688 } 689 } 690 691 692 return *tok; 693 } 694 695 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 696 // Create the initial section, if requested. 697 if (!NoInitialTextSection) 698 Out.InitSections(false); 699 700 // Prime the lexer. 701 Lex(); 702 703 HadError = false; 704 AsmCond StartingCondState = TheCondState; 705 706 // If we are generating dwarf for assembly source files save the initial text 707 // section and generate a .file directive. 708 if (getContext().getGenDwarfForAssembly()) { 709 MCSection *Sec = getStreamer().getCurrentSection().first; 710 if (!Sec->getBeginSymbol()) { 711 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 712 getStreamer().EmitLabel(SectionStartSym); 713 Sec->setBeginSymbol(SectionStartSym); 714 } 715 bool InsertResult = getContext().addGenDwarfSection(Sec); 716 assert(InsertResult && ".text section should not have debug info yet"); 717 (void)InsertResult; 718 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 719 0, StringRef(), getContext().getMainFileName())); 720 } 721 722 // While we have input, parse each statement. 723 while (Lexer.isNot(AsmToken::Eof)) { 724 ParseStatementInfo Info; 725 if (!parseStatement(Info, nullptr)) 726 continue; 727 728 // If we've failed, but on a Error Token, but did not consume it in 729 // favor of a better message, emit it now. 730 if (Lexer.getTok().is(AsmToken::Error)) { 731 Lex(); 732 } 733 734 // We had an error, validate that one was emitted and recover by skipping to 735 // the next line. 736 assert(HadError && "Parse statement returned an error, but none emitted!"); 737 eatToEndOfStatement(); 738 } 739 740 if (TheCondState.TheCond != StartingCondState.TheCond || 741 TheCondState.Ignore != StartingCondState.Ignore) 742 return TokError("unmatched .ifs or .elses"); 743 744 // Check to see there are no empty DwarfFile slots. 745 const auto &LineTables = getContext().getMCDwarfLineTables(); 746 if (!LineTables.empty()) { 747 unsigned Index = 0; 748 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 749 if (File.Name.empty() && Index != 0) 750 TokError("unassigned file number: " + Twine(Index) + 751 " for .file directives"); 752 ++Index; 753 } 754 } 755 756 // Check to see that all assembler local symbols were actually defined. 757 // Targets that don't do subsections via symbols may not want this, though, 758 // so conservatively exclude them. Only do this if we're finalizing, though, 759 // as otherwise we won't necessarilly have seen everything yet. 760 if (!NoFinalize) { 761 if (MAI.hasSubsectionsViaSymbols()) { 762 for (const auto &TableEntry : getContext().getSymbols()) { 763 MCSymbol *Sym = TableEntry.getValue(); 764 // Variable symbols may not be marked as defined, so check those 765 // explicitly. If we know it's a variable, we have a definition for 766 // the purposes of this check. 767 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 768 // FIXME: We would really like to refer back to where the symbol was 769 // first referenced for a source location. We need to add something 770 // to track that. Currently, we just point to the end of the file. 771 HadError |= 772 Error(getTok().getLoc(), "assembler local symbol '" + 773 Sym->getName() + "' not defined"); 774 } 775 } 776 777 // Temporary symbols like the ones for directional jumps don't go in the 778 // symbol table. They also need to be diagnosed in all (final) cases. 779 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 780 if (std::get<2>(LocSym)->isUndefined()) { 781 // Reset the state of any "# line file" directives we've seen to the 782 // context as it was at the diagnostic site. 783 CppHashInfo = std::get<1>(LocSym); 784 HadError |= Error(std::get<0>(LocSym), "directional label undefined"); 785 } 786 } 787 } 788 789 // Finalize the output stream if there are no errors and if the client wants 790 // us to. 791 if (!HadError && !NoFinalize) 792 Out.Finish(); 793 794 return HadError || getContext().hadError(); 795 } 796 797 void AsmParser::checkForValidSection() { 798 if (!ParsingInlineAsm && !getStreamer().getCurrentSection().first) { 799 TokError("expected section directive before assembly directive"); 800 Out.InitSections(false); 801 } 802 } 803 804 /// \brief Throw away the rest of the line for testing purposes. 805 void AsmParser::eatToEndOfStatement() { 806 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 807 Lexer.Lex(); 808 809 // Eat EOL. 810 if (Lexer.is(AsmToken::EndOfStatement)) 811 Lexer.Lex(); 812 } 813 814 StringRef AsmParser::parseStringToEndOfStatement() { 815 const char *Start = getTok().getLoc().getPointer(); 816 817 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 818 Lexer.Lex(); 819 820 const char *End = getTok().getLoc().getPointer(); 821 return StringRef(Start, End - Start); 822 } 823 824 StringRef AsmParser::parseStringToComma() { 825 const char *Start = getTok().getLoc().getPointer(); 826 827 while (Lexer.isNot(AsmToken::EndOfStatement) && 828 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 829 Lexer.Lex(); 830 831 const char *End = getTok().getLoc().getPointer(); 832 return StringRef(Start, End - Start); 833 } 834 835 /// \brief Parse a paren expression and return it. 836 /// NOTE: This assumes the leading '(' has already been consumed. 837 /// 838 /// parenexpr ::= expr) 839 /// 840 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 841 if (parseExpression(Res)) 842 return true; 843 if (Lexer.isNot(AsmToken::RParen)) 844 return TokError("expected ')' in parentheses expression"); 845 EndLoc = Lexer.getTok().getEndLoc(); 846 Lex(); 847 return false; 848 } 849 850 /// \brief Parse a bracket expression and return it. 851 /// NOTE: This assumes the leading '[' has already been consumed. 852 /// 853 /// bracketexpr ::= expr] 854 /// 855 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 856 if (parseExpression(Res)) 857 return true; 858 EndLoc = getTok().getEndLoc(); 859 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 860 return true; 861 return false; 862 } 863 864 /// \brief Parse a primary expression and return it. 865 /// primaryexpr ::= (parenexpr 866 /// primaryexpr ::= symbol 867 /// primaryexpr ::= number 868 /// primaryexpr ::= '.' 869 /// primaryexpr ::= ~,+,- primaryexpr 870 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 871 SMLoc FirstTokenLoc = getLexer().getLoc(); 872 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 873 switch (FirstTokenKind) { 874 default: 875 return TokError("unknown token in expression"); 876 // If we have an error assume that we've already handled it. 877 case AsmToken::Error: 878 return true; 879 case AsmToken::Exclaim: 880 Lex(); // Eat the operator. 881 if (parsePrimaryExpr(Res, EndLoc)) 882 return true; 883 Res = MCUnaryExpr::createLNot(Res, getContext()); 884 return false; 885 case AsmToken::Dollar: 886 case AsmToken::At: 887 case AsmToken::String: 888 case AsmToken::Identifier: { 889 StringRef Identifier; 890 if (parseIdentifier(Identifier)) { 891 if (FirstTokenKind == AsmToken::Dollar) { 892 if (Lexer.getMAI().getDollarIsPC()) { 893 // This is a '$' reference, which references the current PC. Emit a 894 // temporary label to the streamer and refer to it. 895 MCSymbol *Sym = Ctx.createTempSymbol(); 896 Out.EmitLabel(Sym); 897 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 898 getContext()); 899 EndLoc = FirstTokenLoc; 900 return false; 901 } 902 return Error(FirstTokenLoc, "invalid token in expression"); 903 } 904 } 905 // Parse symbol variant 906 std::pair<StringRef, StringRef> Split; 907 if (!MAI.useParensForSymbolVariant()) { 908 if (FirstTokenKind == AsmToken::String) { 909 if (Lexer.is(AsmToken::At)) { 910 Lex(); // eat @ 911 SMLoc AtLoc = getLexer().getLoc(); 912 StringRef VName; 913 if (parseIdentifier(VName)) 914 return Error(AtLoc, "expected symbol variant after '@'"); 915 916 Split = std::make_pair(Identifier, VName); 917 } 918 } else { 919 Split = Identifier.split('@'); 920 } 921 } else if (Lexer.is(AsmToken::LParen)) { 922 Lex(); // eat '('. 923 StringRef VName; 924 parseIdentifier(VName); 925 // eat ')'. 926 if (parseToken(AsmToken::RParen, 927 "unexpected token in variant, expected ')'")) 928 return true; 929 Split = std::make_pair(Identifier, VName); 930 } 931 932 EndLoc = SMLoc::getFromPointer(Identifier.end()); 933 934 // This is a symbol reference. 935 StringRef SymbolName = Identifier; 936 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 937 938 // Lookup the symbol variant if used. 939 if (Split.second.size()) { 940 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 941 if (Variant != MCSymbolRefExpr::VK_Invalid) { 942 SymbolName = Split.first; 943 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 944 Variant = MCSymbolRefExpr::VK_None; 945 } else { 946 return Error(SMLoc::getFromPointer(Split.second.begin()), 947 "invalid variant '" + Split.second + "'"); 948 } 949 } 950 951 MCSymbol *Sym = getContext().getOrCreateSymbol(SymbolName); 952 953 // If this is an absolute variable reference, substitute it now to preserve 954 // semantics in the face of reassignment. 955 if (Sym->isVariable() && 956 isa<MCConstantExpr>(Sym->getVariableValue(/*SetUsed*/ false))) { 957 if (Variant) 958 return Error(EndLoc, "unexpected modifier on variable reference"); 959 960 Res = Sym->getVariableValue(/*SetUsed*/ false); 961 return false; 962 } 963 964 // Otherwise create a symbol ref. 965 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 966 return false; 967 } 968 case AsmToken::BigNum: 969 return TokError("literal value out of range for directive"); 970 case AsmToken::Integer: { 971 SMLoc Loc = getTok().getLoc(); 972 int64_t IntVal = getTok().getIntVal(); 973 Res = MCConstantExpr::create(IntVal, getContext()); 974 EndLoc = Lexer.getTok().getEndLoc(); 975 Lex(); // Eat token. 976 // Look for 'b' or 'f' following an Integer as a directional label 977 if (Lexer.getKind() == AsmToken::Identifier) { 978 StringRef IDVal = getTok().getString(); 979 // Lookup the symbol variant if used. 980 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 981 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 982 if (Split.first.size() != IDVal.size()) { 983 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 984 if (Variant == MCSymbolRefExpr::VK_Invalid) 985 return TokError("invalid variant '" + Split.second + "'"); 986 IDVal = Split.first; 987 } 988 if (IDVal == "f" || IDVal == "b") { 989 MCSymbol *Sym = 990 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 991 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 992 if (IDVal == "b" && Sym->isUndefined()) 993 return Error(Loc, "directional label undefined"); 994 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 995 EndLoc = Lexer.getTok().getEndLoc(); 996 Lex(); // Eat identifier. 997 } 998 } 999 return false; 1000 } 1001 case AsmToken::Real: { 1002 APFloat RealVal(APFloat::IEEEdouble, getTok().getString()); 1003 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1004 Res = MCConstantExpr::create(IntVal, getContext()); 1005 EndLoc = Lexer.getTok().getEndLoc(); 1006 Lex(); // Eat token. 1007 return false; 1008 } 1009 case AsmToken::Dot: { 1010 // This is a '.' reference, which references the current PC. Emit a 1011 // temporary label to the streamer and refer to it. 1012 MCSymbol *Sym = Ctx.createTempSymbol(); 1013 Out.EmitLabel(Sym); 1014 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1015 EndLoc = Lexer.getTok().getEndLoc(); 1016 Lex(); // Eat identifier. 1017 return false; 1018 } 1019 case AsmToken::LParen: 1020 Lex(); // Eat the '('. 1021 return parseParenExpr(Res, EndLoc); 1022 case AsmToken::LBrac: 1023 if (!PlatformParser->HasBracketExpressions()) 1024 return TokError("brackets expression not supported on this target"); 1025 Lex(); // Eat the '['. 1026 return parseBracketExpr(Res, EndLoc); 1027 case AsmToken::Minus: 1028 Lex(); // Eat the operator. 1029 if (parsePrimaryExpr(Res, EndLoc)) 1030 return true; 1031 Res = MCUnaryExpr::createMinus(Res, getContext()); 1032 return false; 1033 case AsmToken::Plus: 1034 Lex(); // Eat the operator. 1035 if (parsePrimaryExpr(Res, EndLoc)) 1036 return true; 1037 Res = MCUnaryExpr::createPlus(Res, getContext()); 1038 return false; 1039 case AsmToken::Tilde: 1040 Lex(); // Eat the operator. 1041 if (parsePrimaryExpr(Res, EndLoc)) 1042 return true; 1043 Res = MCUnaryExpr::createNot(Res, getContext()); 1044 return false; 1045 } 1046 } 1047 1048 bool AsmParser::parseExpression(const MCExpr *&Res) { 1049 SMLoc EndLoc; 1050 return parseExpression(Res, EndLoc); 1051 } 1052 1053 const MCExpr * 1054 AsmParser::applyModifierToExpr(const MCExpr *E, 1055 MCSymbolRefExpr::VariantKind Variant) { 1056 // Ask the target implementation about this expression first. 1057 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1058 if (NewE) 1059 return NewE; 1060 // Recurse over the given expression, rebuilding it to apply the given variant 1061 // if there is exactly one symbol. 1062 switch (E->getKind()) { 1063 case MCExpr::Target: 1064 case MCExpr::Constant: 1065 return nullptr; 1066 1067 case MCExpr::SymbolRef: { 1068 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1069 1070 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1071 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1072 "' (already modified)"); 1073 return E; 1074 } 1075 1076 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1077 } 1078 1079 case MCExpr::Unary: { 1080 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1081 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1082 if (!Sub) 1083 return nullptr; 1084 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1085 } 1086 1087 case MCExpr::Binary: { 1088 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1089 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1090 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1091 1092 if (!LHS && !RHS) 1093 return nullptr; 1094 1095 if (!LHS) 1096 LHS = BE->getLHS(); 1097 if (!RHS) 1098 RHS = BE->getRHS(); 1099 1100 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1101 } 1102 } 1103 1104 llvm_unreachable("Invalid expression kind!"); 1105 } 1106 1107 /// \brief Parse an expression and return it. 1108 /// 1109 /// expr ::= expr &&,|| expr -> lowest. 1110 /// expr ::= expr |,^,&,! expr 1111 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1112 /// expr ::= expr <<,>> expr 1113 /// expr ::= expr +,- expr 1114 /// expr ::= expr *,/,% expr -> highest. 1115 /// expr ::= primaryexpr 1116 /// 1117 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1118 // Parse the expression. 1119 Res = nullptr; 1120 if (parsePrimaryExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc)) 1121 return true; 1122 1123 // As a special case, we support 'a op b @ modifier' by rewriting the 1124 // expression to include the modifier. This is inefficient, but in general we 1125 // expect users to use 'a@modifier op b'. 1126 if (Lexer.getKind() == AsmToken::At) { 1127 Lex(); 1128 1129 if (Lexer.isNot(AsmToken::Identifier)) 1130 return TokError("unexpected symbol modifier following '@'"); 1131 1132 MCSymbolRefExpr::VariantKind Variant = 1133 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1134 if (Variant == MCSymbolRefExpr::VK_Invalid) 1135 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1136 1137 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1138 if (!ModifiedRes) { 1139 return TokError("invalid modifier '" + getTok().getIdentifier() + 1140 "' (no symbols present)"); 1141 } 1142 1143 Res = ModifiedRes; 1144 Lex(); 1145 } 1146 1147 // Try to constant fold it up front, if possible. 1148 int64_t Value; 1149 if (Res->evaluateAsAbsolute(Value)) 1150 Res = MCConstantExpr::create(Value, getContext()); 1151 1152 return false; 1153 } 1154 1155 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1156 Res = nullptr; 1157 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1158 } 1159 1160 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1161 SMLoc &EndLoc) { 1162 if (parseParenExpr(Res, EndLoc)) 1163 return true; 1164 1165 for (; ParenDepth > 0; --ParenDepth) { 1166 if (parseBinOpRHS(1, Res, EndLoc)) 1167 return true; 1168 1169 // We don't Lex() the last RParen. 1170 // This is the same behavior as parseParenExpression(). 1171 if (ParenDepth - 1 > 0) { 1172 EndLoc = getTok().getEndLoc(); 1173 if (parseToken(AsmToken::RParen, 1174 "expected ')' in parentheses expression")) 1175 return true; 1176 } 1177 } 1178 return false; 1179 } 1180 1181 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1182 const MCExpr *Expr; 1183 1184 SMLoc StartLoc = Lexer.getLoc(); 1185 if (parseExpression(Expr)) 1186 return true; 1187 1188 if (!Expr->evaluateAsAbsolute(Res)) 1189 return Error(StartLoc, "expected absolute expression"); 1190 1191 return false; 1192 } 1193 1194 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1195 MCBinaryExpr::Opcode &Kind, 1196 bool ShouldUseLogicalShr) { 1197 switch (K) { 1198 default: 1199 return 0; // not a binop. 1200 1201 // Lowest Precedence: &&, || 1202 case AsmToken::AmpAmp: 1203 Kind = MCBinaryExpr::LAnd; 1204 return 1; 1205 case AsmToken::PipePipe: 1206 Kind = MCBinaryExpr::LOr; 1207 return 1; 1208 1209 // Low Precedence: |, &, ^ 1210 // 1211 // FIXME: gas seems to support '!' as an infix operator? 1212 case AsmToken::Pipe: 1213 Kind = MCBinaryExpr::Or; 1214 return 2; 1215 case AsmToken::Caret: 1216 Kind = MCBinaryExpr::Xor; 1217 return 2; 1218 case AsmToken::Amp: 1219 Kind = MCBinaryExpr::And; 1220 return 2; 1221 1222 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1223 case AsmToken::EqualEqual: 1224 Kind = MCBinaryExpr::EQ; 1225 return 3; 1226 case AsmToken::ExclaimEqual: 1227 case AsmToken::LessGreater: 1228 Kind = MCBinaryExpr::NE; 1229 return 3; 1230 case AsmToken::Less: 1231 Kind = MCBinaryExpr::LT; 1232 return 3; 1233 case AsmToken::LessEqual: 1234 Kind = MCBinaryExpr::LTE; 1235 return 3; 1236 case AsmToken::Greater: 1237 Kind = MCBinaryExpr::GT; 1238 return 3; 1239 case AsmToken::GreaterEqual: 1240 Kind = MCBinaryExpr::GTE; 1241 return 3; 1242 1243 // Intermediate Precedence: <<, >> 1244 case AsmToken::LessLess: 1245 Kind = MCBinaryExpr::Shl; 1246 return 4; 1247 case AsmToken::GreaterGreater: 1248 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1249 return 4; 1250 1251 // High Intermediate Precedence: +, - 1252 case AsmToken::Plus: 1253 Kind = MCBinaryExpr::Add; 1254 return 5; 1255 case AsmToken::Minus: 1256 Kind = MCBinaryExpr::Sub; 1257 return 5; 1258 1259 // Highest Precedence: *, /, % 1260 case AsmToken::Star: 1261 Kind = MCBinaryExpr::Mul; 1262 return 6; 1263 case AsmToken::Slash: 1264 Kind = MCBinaryExpr::Div; 1265 return 6; 1266 case AsmToken::Percent: 1267 Kind = MCBinaryExpr::Mod; 1268 return 6; 1269 } 1270 } 1271 1272 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1273 MCBinaryExpr::Opcode &Kind, 1274 bool ShouldUseLogicalShr) { 1275 switch (K) { 1276 default: 1277 return 0; // not a binop. 1278 1279 // Lowest Precedence: &&, || 1280 case AsmToken::AmpAmp: 1281 Kind = MCBinaryExpr::LAnd; 1282 return 2; 1283 case AsmToken::PipePipe: 1284 Kind = MCBinaryExpr::LOr; 1285 return 1; 1286 1287 // Low Precedence: ==, !=, <>, <, <=, >, >= 1288 case AsmToken::EqualEqual: 1289 Kind = MCBinaryExpr::EQ; 1290 return 3; 1291 case AsmToken::ExclaimEqual: 1292 case AsmToken::LessGreater: 1293 Kind = MCBinaryExpr::NE; 1294 return 3; 1295 case AsmToken::Less: 1296 Kind = MCBinaryExpr::LT; 1297 return 3; 1298 case AsmToken::LessEqual: 1299 Kind = MCBinaryExpr::LTE; 1300 return 3; 1301 case AsmToken::Greater: 1302 Kind = MCBinaryExpr::GT; 1303 return 3; 1304 case AsmToken::GreaterEqual: 1305 Kind = MCBinaryExpr::GTE; 1306 return 3; 1307 1308 // Low Intermediate Precedence: +, - 1309 case AsmToken::Plus: 1310 Kind = MCBinaryExpr::Add; 1311 return 4; 1312 case AsmToken::Minus: 1313 Kind = MCBinaryExpr::Sub; 1314 return 4; 1315 1316 // High Intermediate Precedence: |, &, ^ 1317 // 1318 // FIXME: gas seems to support '!' as an infix operator? 1319 case AsmToken::Pipe: 1320 Kind = MCBinaryExpr::Or; 1321 return 5; 1322 case AsmToken::Caret: 1323 Kind = MCBinaryExpr::Xor; 1324 return 5; 1325 case AsmToken::Amp: 1326 Kind = MCBinaryExpr::And; 1327 return 5; 1328 1329 // Highest Precedence: *, /, %, <<, >> 1330 case AsmToken::Star: 1331 Kind = MCBinaryExpr::Mul; 1332 return 6; 1333 case AsmToken::Slash: 1334 Kind = MCBinaryExpr::Div; 1335 return 6; 1336 case AsmToken::Percent: 1337 Kind = MCBinaryExpr::Mod; 1338 return 6; 1339 case AsmToken::LessLess: 1340 Kind = MCBinaryExpr::Shl; 1341 return 6; 1342 case AsmToken::GreaterGreater: 1343 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1344 return 6; 1345 } 1346 } 1347 1348 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1349 MCBinaryExpr::Opcode &Kind) { 1350 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1351 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1352 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1353 } 1354 1355 /// \brief Parse all binary operators with precedence >= 'Precedence'. 1356 /// Res contains the LHS of the expression on input. 1357 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1358 SMLoc &EndLoc) { 1359 while (1) { 1360 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1361 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1362 1363 // If the next token is lower precedence than we are allowed to eat, return 1364 // successfully with what we ate already. 1365 if (TokPrec < Precedence) 1366 return false; 1367 1368 Lex(); 1369 1370 // Eat the next primary expression. 1371 const MCExpr *RHS; 1372 if (parsePrimaryExpr(RHS, EndLoc)) 1373 return true; 1374 1375 // If BinOp binds less tightly with RHS than the operator after RHS, let 1376 // the pending operator take RHS as its LHS. 1377 MCBinaryExpr::Opcode Dummy; 1378 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1379 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1380 return true; 1381 1382 // Merge LHS and RHS according to operator. 1383 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext()); 1384 } 1385 } 1386 1387 /// ParseStatement: 1388 /// ::= EndOfStatement 1389 /// ::= Label* Directive ...Operands... EndOfStatement 1390 /// ::= Label* Identifier OperandList* EndOfStatement 1391 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1392 MCAsmParserSemaCallback *SI) { 1393 // Eat initial spaces and comments 1394 while (Lexer.is(AsmToken::Space)) 1395 Lex(); 1396 if (Lexer.is(AsmToken::EndOfStatement)) { 1397 // if this is a line comment we can drop it safely 1398 if (getTok().getString().front() == '\r' || 1399 getTok().getString().front() == '\n') 1400 Out.AddBlankLine(); 1401 Lex(); 1402 return false; 1403 } 1404 // Statements always start with an identifier. 1405 AsmToken ID = getTok(); 1406 SMLoc IDLoc = ID.getLoc(); 1407 StringRef IDVal; 1408 int64_t LocalLabelVal = -1; 1409 if (Lexer.is(AsmToken::HashDirective)) 1410 return parseCppHashLineFilenameComment(IDLoc); 1411 // Allow an integer followed by a ':' as a directional local label. 1412 if (Lexer.is(AsmToken::Integer)) { 1413 LocalLabelVal = getTok().getIntVal(); 1414 if (LocalLabelVal < 0) { 1415 if (!TheCondState.Ignore) 1416 return TokError("unexpected token at start of statement"); 1417 IDVal = ""; 1418 } else { 1419 IDVal = getTok().getString(); 1420 Lex(); // Consume the integer token to be used as an identifier token. 1421 if (Lexer.getKind() != AsmToken::Colon) { 1422 if (!TheCondState.Ignore) 1423 return TokError("unexpected token at start of statement"); 1424 } 1425 } 1426 } else if (Lexer.is(AsmToken::Dot)) { 1427 // Treat '.' as a valid identifier in this context. 1428 Lex(); 1429 IDVal = "."; 1430 } else if (Lexer.is(AsmToken::LCurly)) { 1431 // Treat '{' as a valid identifier in this context. 1432 Lex(); 1433 IDVal = "{"; 1434 1435 } else if (Lexer.is(AsmToken::RCurly)) { 1436 // Treat '}' as a valid identifier in this context. 1437 Lex(); 1438 IDVal = "}"; 1439 } else if (parseIdentifier(IDVal)) { 1440 if (!TheCondState.Ignore) 1441 return TokError("unexpected token at start of statement"); 1442 IDVal = ""; 1443 } 1444 1445 // Handle conditional assembly here before checking for skipping. We 1446 // have to do this so that .endif isn't skipped in a ".if 0" block for 1447 // example. 1448 StringMap<DirectiveKind>::const_iterator DirKindIt = 1449 DirectiveKindMap.find(IDVal); 1450 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1451 ? DK_NO_DIRECTIVE 1452 : DirKindIt->getValue(); 1453 switch (DirKind) { 1454 default: 1455 break; 1456 case DK_IF: 1457 case DK_IFEQ: 1458 case DK_IFGE: 1459 case DK_IFGT: 1460 case DK_IFLE: 1461 case DK_IFLT: 1462 case DK_IFNE: 1463 return parseDirectiveIf(IDLoc, DirKind); 1464 case DK_IFB: 1465 return parseDirectiveIfb(IDLoc, true); 1466 case DK_IFNB: 1467 return parseDirectiveIfb(IDLoc, false); 1468 case DK_IFC: 1469 return parseDirectiveIfc(IDLoc, true); 1470 case DK_IFEQS: 1471 return parseDirectiveIfeqs(IDLoc, true); 1472 case DK_IFNC: 1473 return parseDirectiveIfc(IDLoc, false); 1474 case DK_IFNES: 1475 return parseDirectiveIfeqs(IDLoc, false); 1476 case DK_IFDEF: 1477 return parseDirectiveIfdef(IDLoc, true); 1478 case DK_IFNDEF: 1479 case DK_IFNOTDEF: 1480 return parseDirectiveIfdef(IDLoc, false); 1481 case DK_ELSEIF: 1482 return parseDirectiveElseIf(IDLoc); 1483 case DK_ELSE: 1484 return parseDirectiveElse(IDLoc); 1485 case DK_ENDIF: 1486 return parseDirectiveEndIf(IDLoc); 1487 } 1488 1489 // Ignore the statement if in the middle of inactive conditional 1490 // (e.g. ".if 0"). 1491 if (TheCondState.Ignore) { 1492 eatToEndOfStatement(); 1493 return false; 1494 } 1495 1496 // FIXME: Recurse on local labels? 1497 1498 // See what kind of statement we have. 1499 switch (Lexer.getKind()) { 1500 case AsmToken::Colon: { 1501 if (!getTargetParser().isLabel(ID)) 1502 break; 1503 checkForValidSection(); 1504 1505 // identifier ':' -> Label. 1506 Lex(); 1507 1508 // Diagnose attempt to use '.' as a label. 1509 if (IDVal == ".") 1510 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1511 1512 // Diagnose attempt to use a variable as a label. 1513 // 1514 // FIXME: Diagnostics. Note the location of the definition as a label. 1515 // FIXME: This doesn't diagnose assignment to a symbol which has been 1516 // implicitly marked as external. 1517 MCSymbol *Sym; 1518 if (LocalLabelVal == -1) { 1519 if (ParsingInlineAsm && SI) { 1520 StringRef RewrittenLabel = 1521 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1522 assert(RewrittenLabel.size() && 1523 "We should have an internal name here."); 1524 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1525 RewrittenLabel); 1526 IDVal = RewrittenLabel; 1527 } 1528 Sym = getContext().getOrCreateSymbol(IDVal); 1529 } else 1530 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1531 1532 Sym->redefineIfPossible(); 1533 1534 if (!Sym->isUndefined() || Sym->isVariable()) 1535 return Error(IDLoc, "invalid symbol redefinition"); 1536 1537 // Consume any end of statement token, if present, to avoid spurious 1538 // AddBlankLine calls(). 1539 if (getTok().is(AsmToken::EndOfStatement)) { 1540 Lex(); 1541 } 1542 1543 // Emit the label. 1544 if (!ParsingInlineAsm) 1545 Out.EmitLabel(Sym); 1546 1547 // If we are generating dwarf for assembly source files then gather the 1548 // info to make a dwarf label entry for this label if needed. 1549 if (getContext().getGenDwarfForAssembly()) 1550 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1551 IDLoc); 1552 1553 getTargetParser().onLabelParsed(Sym); 1554 1555 1556 1557 return false; 1558 } 1559 1560 case AsmToken::Equal: 1561 if (!getTargetParser().equalIsAsmAssignment()) 1562 break; 1563 // identifier '=' ... -> assignment statement 1564 Lex(); 1565 1566 return parseAssignment(IDVal, true); 1567 1568 default: // Normal instruction or directive. 1569 break; 1570 } 1571 1572 // If macros are enabled, check to see if this is a macro instantiation. 1573 if (areMacrosEnabled()) 1574 if (const MCAsmMacro *M = lookupMacro(IDVal)) { 1575 return handleMacroEntry(M, IDLoc); 1576 } 1577 1578 // Otherwise, we have a normal instruction or directive. 1579 1580 // Directives start with "." 1581 if (IDVal[0] == '.' && IDVal != ".") { 1582 // There are several entities interested in parsing directives: 1583 // 1584 // 1. The target-specific assembly parser. Some directives are target 1585 // specific or may potentially behave differently on certain targets. 1586 // 2. Asm parser extensions. For example, platform-specific parsers 1587 // (like the ELF parser) register themselves as extensions. 1588 // 3. The generic directive parser implemented by this class. These are 1589 // all the directives that behave in a target and platform independent 1590 // manner, or at least have a default behavior that's shared between 1591 // all targets and platforms. 1592 1593 // First query the target-specific parser. It will return 'true' if it 1594 // isn't interested in this directive. 1595 if (!getTargetParser().ParseDirective(ID)) 1596 return false; 1597 1598 // Next, check the extension directive map to see if any extension has 1599 // registered itself to parse this directive. 1600 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1601 ExtensionDirectiveMap.lookup(IDVal); 1602 if (Handler.first) 1603 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1604 1605 // Finally, if no one else is interested in this directive, it must be 1606 // generic and familiar to this class. 1607 switch (DirKind) { 1608 default: 1609 break; 1610 case DK_SET: 1611 case DK_EQU: 1612 return parseDirectiveSet(IDVal, true); 1613 case DK_EQUIV: 1614 return parseDirectiveSet(IDVal, false); 1615 case DK_ASCII: 1616 return parseDirectiveAscii(IDVal, false); 1617 case DK_ASCIZ: 1618 case DK_STRING: 1619 return parseDirectiveAscii(IDVal, true); 1620 case DK_BYTE: 1621 return parseDirectiveValue(1); 1622 case DK_SHORT: 1623 case DK_VALUE: 1624 case DK_2BYTE: 1625 return parseDirectiveValue(2); 1626 case DK_LONG: 1627 case DK_INT: 1628 case DK_4BYTE: 1629 return parseDirectiveValue(4); 1630 case DK_QUAD: 1631 case DK_8BYTE: 1632 return parseDirectiveValue(8); 1633 case DK_OCTA: 1634 return parseDirectiveOctaValue(); 1635 case DK_SINGLE: 1636 case DK_FLOAT: 1637 return parseDirectiveRealValue(APFloat::IEEEsingle); 1638 case DK_DOUBLE: 1639 return parseDirectiveRealValue(APFloat::IEEEdouble); 1640 case DK_ALIGN: { 1641 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1642 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1643 } 1644 case DK_ALIGN32: { 1645 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1646 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1647 } 1648 case DK_BALIGN: 1649 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1650 case DK_BALIGNW: 1651 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1652 case DK_BALIGNL: 1653 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1654 case DK_P2ALIGN: 1655 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1656 case DK_P2ALIGNW: 1657 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1658 case DK_P2ALIGNL: 1659 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1660 case DK_ORG: 1661 return parseDirectiveOrg(); 1662 case DK_FILL: 1663 return parseDirectiveFill(); 1664 case DK_ZERO: 1665 return parseDirectiveZero(); 1666 case DK_EXTERN: 1667 eatToEndOfStatement(); // .extern is the default, ignore it. 1668 return false; 1669 case DK_GLOBL: 1670 case DK_GLOBAL: 1671 return parseDirectiveSymbolAttribute(MCSA_Global); 1672 case DK_LAZY_REFERENCE: 1673 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1674 case DK_NO_DEAD_STRIP: 1675 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1676 case DK_SYMBOL_RESOLVER: 1677 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1678 case DK_PRIVATE_EXTERN: 1679 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1680 case DK_REFERENCE: 1681 return parseDirectiveSymbolAttribute(MCSA_Reference); 1682 case DK_WEAK_DEFINITION: 1683 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1684 case DK_WEAK_REFERENCE: 1685 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1686 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1687 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1688 case DK_COMM: 1689 case DK_COMMON: 1690 return parseDirectiveComm(/*IsLocal=*/false); 1691 case DK_LCOMM: 1692 return parseDirectiveComm(/*IsLocal=*/true); 1693 case DK_ABORT: 1694 return parseDirectiveAbort(); 1695 case DK_INCLUDE: 1696 return parseDirectiveInclude(); 1697 case DK_INCBIN: 1698 return parseDirectiveIncbin(); 1699 case DK_CODE16: 1700 case DK_CODE16GCC: 1701 return TokError(Twine(IDVal) + 1702 " not currently supported for this target"); 1703 case DK_REPT: 1704 return parseDirectiveRept(IDLoc, IDVal); 1705 case DK_IRP: 1706 return parseDirectiveIrp(IDLoc); 1707 case DK_IRPC: 1708 return parseDirectiveIrpc(IDLoc); 1709 case DK_ENDR: 1710 return parseDirectiveEndr(IDLoc); 1711 case DK_BUNDLE_ALIGN_MODE: 1712 return parseDirectiveBundleAlignMode(); 1713 case DK_BUNDLE_LOCK: 1714 return parseDirectiveBundleLock(); 1715 case DK_BUNDLE_UNLOCK: 1716 return parseDirectiveBundleUnlock(); 1717 case DK_SLEB128: 1718 return parseDirectiveLEB128(true); 1719 case DK_ULEB128: 1720 return parseDirectiveLEB128(false); 1721 case DK_SPACE: 1722 case DK_SKIP: 1723 return parseDirectiveSpace(IDVal); 1724 case DK_FILE: 1725 return parseDirectiveFile(IDLoc); 1726 case DK_LINE: 1727 return parseDirectiveLine(); 1728 case DK_LOC: 1729 return parseDirectiveLoc(); 1730 case DK_STABS: 1731 return parseDirectiveStabs(); 1732 case DK_CV_FILE: 1733 return parseDirectiveCVFile(); 1734 case DK_CV_LOC: 1735 return parseDirectiveCVLoc(); 1736 case DK_CV_LINETABLE: 1737 return parseDirectiveCVLinetable(); 1738 case DK_CV_INLINE_LINETABLE: 1739 return parseDirectiveCVInlineLinetable(); 1740 case DK_CV_DEF_RANGE: 1741 return parseDirectiveCVDefRange(); 1742 case DK_CV_STRINGTABLE: 1743 return parseDirectiveCVStringTable(); 1744 case DK_CV_FILECHECKSUMS: 1745 return parseDirectiveCVFileChecksums(); 1746 case DK_CFI_SECTIONS: 1747 return parseDirectiveCFISections(); 1748 case DK_CFI_STARTPROC: 1749 return parseDirectiveCFIStartProc(); 1750 case DK_CFI_ENDPROC: 1751 return parseDirectiveCFIEndProc(); 1752 case DK_CFI_DEF_CFA: 1753 return parseDirectiveCFIDefCfa(IDLoc); 1754 case DK_CFI_DEF_CFA_OFFSET: 1755 return parseDirectiveCFIDefCfaOffset(); 1756 case DK_CFI_ADJUST_CFA_OFFSET: 1757 return parseDirectiveCFIAdjustCfaOffset(); 1758 case DK_CFI_DEF_CFA_REGISTER: 1759 return parseDirectiveCFIDefCfaRegister(IDLoc); 1760 case DK_CFI_OFFSET: 1761 return parseDirectiveCFIOffset(IDLoc); 1762 case DK_CFI_REL_OFFSET: 1763 return parseDirectiveCFIRelOffset(IDLoc); 1764 case DK_CFI_PERSONALITY: 1765 return parseDirectiveCFIPersonalityOrLsda(true); 1766 case DK_CFI_LSDA: 1767 return parseDirectiveCFIPersonalityOrLsda(false); 1768 case DK_CFI_REMEMBER_STATE: 1769 return parseDirectiveCFIRememberState(); 1770 case DK_CFI_RESTORE_STATE: 1771 return parseDirectiveCFIRestoreState(); 1772 case DK_CFI_SAME_VALUE: 1773 return parseDirectiveCFISameValue(IDLoc); 1774 case DK_CFI_RESTORE: 1775 return parseDirectiveCFIRestore(IDLoc); 1776 case DK_CFI_ESCAPE: 1777 return parseDirectiveCFIEscape(); 1778 case DK_CFI_SIGNAL_FRAME: 1779 return parseDirectiveCFISignalFrame(); 1780 case DK_CFI_UNDEFINED: 1781 return parseDirectiveCFIUndefined(IDLoc); 1782 case DK_CFI_REGISTER: 1783 return parseDirectiveCFIRegister(IDLoc); 1784 case DK_CFI_WINDOW_SAVE: 1785 return parseDirectiveCFIWindowSave(); 1786 case DK_MACROS_ON: 1787 case DK_MACROS_OFF: 1788 return parseDirectiveMacrosOnOff(IDVal); 1789 case DK_MACRO: 1790 return parseDirectiveMacro(IDLoc); 1791 case DK_EXITM: 1792 return parseDirectiveExitMacro(IDVal); 1793 case DK_ENDM: 1794 case DK_ENDMACRO: 1795 return parseDirectiveEndMacro(IDVal); 1796 case DK_PURGEM: 1797 return parseDirectivePurgeMacro(IDLoc); 1798 case DK_END: 1799 return parseDirectiveEnd(IDLoc); 1800 case DK_ERR: 1801 return parseDirectiveError(IDLoc, false); 1802 case DK_ERROR: 1803 return parseDirectiveError(IDLoc, true); 1804 case DK_WARNING: 1805 return parseDirectiveWarning(IDLoc); 1806 case DK_RELOC: 1807 return parseDirectiveReloc(IDLoc); 1808 } 1809 1810 return Error(IDLoc, "unknown directive"); 1811 } 1812 1813 // __asm _emit or __asm __emit 1814 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 1815 IDVal == "_EMIT" || IDVal == "__EMIT")) 1816 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 1817 1818 // __asm align 1819 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 1820 return parseDirectiveMSAlign(IDLoc, Info); 1821 1822 if (ParsingInlineAsm && (IDVal == "even")) 1823 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 1824 checkForValidSection(); 1825 1826 // Canonicalize the opcode to lower case. 1827 std::string OpcodeStr = IDVal.lower(); 1828 ParseInstructionInfo IInfo(Info.AsmRewrites); 1829 bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 1830 Info.ParsedOperands); 1831 Info.ParseError = HadError; 1832 1833 // Dump the parsed representation, if requested. 1834 if (getShowParsedOperands()) { 1835 SmallString<256> Str; 1836 raw_svector_ostream OS(Str); 1837 OS << "parsed instruction: ["; 1838 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 1839 if (i != 0) 1840 OS << ", "; 1841 Info.ParsedOperands[i]->print(OS); 1842 } 1843 OS << "]"; 1844 1845 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 1846 } 1847 1848 // If we are generating dwarf for the current section then generate a .loc 1849 // directive for the instruction. 1850 if (!HadError && getContext().getGenDwarfForAssembly() && 1851 getContext().getGenDwarfSectionSyms().count( 1852 getStreamer().getCurrentSection().first)) { 1853 unsigned Line; 1854 if (ActiveMacros.empty()) 1855 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 1856 else 1857 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 1858 ActiveMacros.front()->ExitBuffer); 1859 1860 // If we previously parsed a cpp hash file line comment then make sure the 1861 // current Dwarf File is for the CppHashFilename if not then emit the 1862 // Dwarf File table for it and adjust the line number for the .loc. 1863 if (CppHashInfo.Filename.size()) { 1864 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 1865 0, StringRef(), CppHashInfo.Filename); 1866 getContext().setGenDwarfFileNumber(FileNumber); 1867 1868 // Since SrcMgr.FindLineNumber() is slow and messes up the SourceMgr's 1869 // cache with the different Loc from the call above we save the last 1870 // info we queried here with SrcMgr.FindLineNumber(). 1871 unsigned CppHashLocLineNo; 1872 if (LastQueryIDLoc == CppHashInfo.Loc && 1873 LastQueryBuffer == CppHashInfo.Buf) 1874 CppHashLocLineNo = LastQueryLine; 1875 else { 1876 CppHashLocLineNo = 1877 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 1878 LastQueryLine = CppHashLocLineNo; 1879 LastQueryIDLoc = CppHashInfo.Loc; 1880 LastQueryBuffer = CppHashInfo.Buf; 1881 } 1882 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 1883 } 1884 1885 getStreamer().EmitDwarfLocDirective( 1886 getContext().getGenDwarfFileNumber(), Line, 0, 1887 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 1888 StringRef()); 1889 } 1890 1891 // If parsing succeeded, match the instruction. 1892 if (!HadError) { 1893 uint64_t ErrorInfo; 1894 getTargetParser().MatchAndEmitInstruction(IDLoc, Info.Opcode, 1895 Info.ParsedOperands, Out, 1896 ErrorInfo, ParsingInlineAsm); 1897 } 1898 1899 // Don't skip the rest of the line, the instruction parser is responsible for 1900 // that. 1901 return false; 1902 } 1903 1904 // Parse and erase curly braces marking block start/end 1905 bool 1906 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 1907 // Identify curly brace marking block start/end 1908 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 1909 return false; 1910 1911 SMLoc StartLoc = Lexer.getLoc(); 1912 Lex(); // Eat the brace 1913 if (Lexer.is(AsmToken::EndOfStatement)) 1914 Lex(); // Eat EndOfStatement following the brace 1915 1916 // Erase the block start/end brace from the output asm string 1917 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 1918 StartLoc.getPointer()); 1919 return true; 1920 } 1921 1922 /// parseCppHashLineFilenameComment as this: 1923 /// ::= # number "filename" 1924 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) { 1925 Lex(); // Eat the hash token. 1926 // Lexer only ever emits HashDirective if it fully formed if it's 1927 // done the checking already so this is an internal error. 1928 assert(getTok().is(AsmToken::Integer) && 1929 "Lexing Cpp line comment: Expected Integer"); 1930 int64_t LineNumber = getTok().getIntVal(); 1931 Lex(); 1932 assert(getTok().is(AsmToken::String) && 1933 "Lexing Cpp line comment: Expected String"); 1934 StringRef Filename = getTok().getString(); 1935 Lex(); 1936 // Get rid of the enclosing quotes. 1937 Filename = Filename.substr(1, Filename.size() - 2); 1938 1939 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 1940 CppHashInfo.Loc = L; 1941 CppHashInfo.Filename = Filename; 1942 CppHashInfo.LineNumber = LineNumber; 1943 CppHashInfo.Buf = CurBuffer; 1944 return false; 1945 } 1946 1947 /// \brief will use the last parsed cpp hash line filename comment 1948 /// for the Filename and LineNo if any in the diagnostic. 1949 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 1950 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 1951 raw_ostream &OS = errs(); 1952 1953 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 1954 SMLoc DiagLoc = Diag.getLoc(); 1955 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1956 unsigned CppHashBuf = 1957 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 1958 1959 // Like SourceMgr::printMessage() we need to print the include stack if any 1960 // before printing the message. 1961 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1962 if (!Parser->SavedDiagHandler && DiagCurBuffer && 1963 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 1964 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 1965 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 1966 } 1967 1968 // If we have not parsed a cpp hash line filename comment or the source 1969 // manager changed or buffer changed (like in a nested include) then just 1970 // print the normal diagnostic using its Filename and LineNo. 1971 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 1972 DiagBuf != CppHashBuf) { 1973 if (Parser->SavedDiagHandler) 1974 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 1975 else 1976 Diag.print(nullptr, OS); 1977 return; 1978 } 1979 1980 // Use the CppHashFilename and calculate a line number based on the 1981 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 1982 // for the diagnostic. 1983 const std::string &Filename = Parser->CppHashInfo.Filename; 1984 1985 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 1986 int CppHashLocLineNo = 1987 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 1988 int LineNo = 1989 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 1990 1991 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 1992 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 1993 Diag.getLineContents(), Diag.getRanges()); 1994 1995 if (Parser->SavedDiagHandler) 1996 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 1997 else 1998 NewDiag.print(nullptr, OS); 1999 } 2000 2001 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2002 // difference being that that function accepts '@' as part of identifiers and 2003 // we can't do that. AsmLexer.cpp should probably be changed to handle 2004 // '@' as a special case when needed. 2005 static bool isIdentifierChar(char c) { 2006 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2007 c == '.'; 2008 } 2009 2010 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2011 ArrayRef<MCAsmMacroParameter> Parameters, 2012 ArrayRef<MCAsmMacroArgument> A, 2013 bool EnableAtPseudoVariable, SMLoc L) { 2014 unsigned NParameters = Parameters.size(); 2015 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2016 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2017 return Error(L, "Wrong number of arguments"); 2018 2019 // A macro without parameters is handled differently on Darwin: 2020 // gas accepts no arguments and does no substitutions 2021 while (!Body.empty()) { 2022 // Scan for the next substitution. 2023 std::size_t End = Body.size(), Pos = 0; 2024 for (; Pos != End; ++Pos) { 2025 // Check for a substitution or escape. 2026 if (IsDarwin && !NParameters) { 2027 // This macro has no parameters, look for $0, $1, etc. 2028 if (Body[Pos] != '$' || Pos + 1 == End) 2029 continue; 2030 2031 char Next = Body[Pos + 1]; 2032 if (Next == '$' || Next == 'n' || 2033 isdigit(static_cast<unsigned char>(Next))) 2034 break; 2035 } else { 2036 // This macro has parameters, look for \foo, \bar, etc. 2037 if (Body[Pos] == '\\' && Pos + 1 != End) 2038 break; 2039 } 2040 } 2041 2042 // Add the prefix. 2043 OS << Body.slice(0, Pos); 2044 2045 // Check if we reached the end. 2046 if (Pos == End) 2047 break; 2048 2049 if (IsDarwin && !NParameters) { 2050 switch (Body[Pos + 1]) { 2051 // $$ => $ 2052 case '$': 2053 OS << '$'; 2054 break; 2055 2056 // $n => number of arguments 2057 case 'n': 2058 OS << A.size(); 2059 break; 2060 2061 // $[0-9] => argument 2062 default: { 2063 // Missing arguments are ignored. 2064 unsigned Index = Body[Pos + 1] - '0'; 2065 if (Index >= A.size()) 2066 break; 2067 2068 // Otherwise substitute with the token values, with spaces eliminated. 2069 for (const AsmToken &Token : A[Index]) 2070 OS << Token.getString(); 2071 break; 2072 } 2073 } 2074 Pos += 2; 2075 } else { 2076 unsigned I = Pos + 1; 2077 2078 // Check for the \@ pseudo-variable. 2079 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2080 ++I; 2081 else 2082 while (isIdentifierChar(Body[I]) && I + 1 != End) 2083 ++I; 2084 2085 const char *Begin = Body.data() + Pos + 1; 2086 StringRef Argument(Begin, I - (Pos + 1)); 2087 unsigned Index = 0; 2088 2089 if (Argument == "@") { 2090 OS << NumOfMacroInstantiations; 2091 Pos += 2; 2092 } else { 2093 for (; Index < NParameters; ++Index) 2094 if (Parameters[Index].Name == Argument) 2095 break; 2096 2097 if (Index == NParameters) { 2098 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2099 Pos += 3; 2100 else { 2101 OS << '\\' << Argument; 2102 Pos = I; 2103 } 2104 } else { 2105 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2106 for (const AsmToken &Token : A[Index]) 2107 // We expect no quotes around the string's contents when 2108 // parsing for varargs. 2109 if (Token.getKind() != AsmToken::String || VarargParameter) 2110 OS << Token.getString(); 2111 else 2112 OS << Token.getStringContents(); 2113 2114 Pos += 1 + Argument.size(); 2115 } 2116 } 2117 } 2118 // Update the scan point. 2119 Body = Body.substr(Pos); 2120 } 2121 2122 return false; 2123 } 2124 2125 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 2126 size_t CondStackDepth) 2127 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 2128 CondStackDepth(CondStackDepth) {} 2129 2130 static bool isOperator(AsmToken::TokenKind kind) { 2131 switch (kind) { 2132 default: 2133 return false; 2134 case AsmToken::Plus: 2135 case AsmToken::Minus: 2136 case AsmToken::Tilde: 2137 case AsmToken::Slash: 2138 case AsmToken::Star: 2139 case AsmToken::Dot: 2140 case AsmToken::Equal: 2141 case AsmToken::EqualEqual: 2142 case AsmToken::Pipe: 2143 case AsmToken::PipePipe: 2144 case AsmToken::Caret: 2145 case AsmToken::Amp: 2146 case AsmToken::AmpAmp: 2147 case AsmToken::Exclaim: 2148 case AsmToken::ExclaimEqual: 2149 case AsmToken::Less: 2150 case AsmToken::LessEqual: 2151 case AsmToken::LessLess: 2152 case AsmToken::LessGreater: 2153 case AsmToken::Greater: 2154 case AsmToken::GreaterEqual: 2155 case AsmToken::GreaterGreater: 2156 return true; 2157 } 2158 } 2159 2160 namespace { 2161 class AsmLexerSkipSpaceRAII { 2162 public: 2163 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2164 Lexer.setSkipSpace(SkipSpace); 2165 } 2166 2167 ~AsmLexerSkipSpaceRAII() { 2168 Lexer.setSkipSpace(true); 2169 } 2170 2171 private: 2172 AsmLexer &Lexer; 2173 }; 2174 } 2175 2176 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2177 2178 if (Vararg) { 2179 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2180 StringRef Str = parseStringToEndOfStatement(); 2181 MA.emplace_back(AsmToken::String, Str); 2182 } 2183 return false; 2184 } 2185 2186 unsigned ParenLevel = 0; 2187 2188 // Darwin doesn't use spaces to delmit arguments. 2189 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2190 2191 bool SpaceEaten; 2192 2193 for (;;) { 2194 SpaceEaten = false; 2195 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2196 return TokError("unexpected token in macro instantiation"); 2197 2198 if (ParenLevel == 0) { 2199 2200 if (Lexer.is(AsmToken::Comma)) 2201 break; 2202 2203 if (Lexer.is(AsmToken::Space)) { 2204 SpaceEaten = true; 2205 Lexer.Lex(); // Eat spaces 2206 } 2207 2208 // Spaces can delimit parameters, but could also be part an expression. 2209 // If the token after a space is an operator, add the token and the next 2210 // one into this argument 2211 if (!IsDarwin) { 2212 if (isOperator(Lexer.getKind())) { 2213 MA.push_back(getTok()); 2214 Lexer.Lex(); 2215 2216 // Whitespace after an operator can be ignored. 2217 if (Lexer.is(AsmToken::Space)) 2218 Lexer.Lex(); 2219 2220 continue; 2221 } 2222 } 2223 if (SpaceEaten) 2224 break; 2225 } 2226 2227 // handleMacroEntry relies on not advancing the lexer here 2228 // to be able to fill in the remaining default parameter values 2229 if (Lexer.is(AsmToken::EndOfStatement)) 2230 break; 2231 2232 // Adjust the current parentheses level. 2233 if (Lexer.is(AsmToken::LParen)) 2234 ++ParenLevel; 2235 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2236 --ParenLevel; 2237 2238 // Append the token to the current argument list. 2239 MA.push_back(getTok()); 2240 Lexer.Lex(); 2241 } 2242 2243 if (ParenLevel != 0) 2244 return TokError("unbalanced parentheses in macro argument"); 2245 return false; 2246 } 2247 2248 // Parse the macro instantiation arguments. 2249 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2250 MCAsmMacroArguments &A) { 2251 const unsigned NParameters = M ? M->Parameters.size() : 0; 2252 bool NamedParametersFound = false; 2253 SmallVector<SMLoc, 4> FALocs; 2254 2255 A.resize(NParameters); 2256 FALocs.resize(NParameters); 2257 2258 // Parse two kinds of macro invocations: 2259 // - macros defined without any parameters accept an arbitrary number of them 2260 // - macros defined with parameters accept at most that many of them 2261 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2262 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2263 ++Parameter) { 2264 SMLoc IDLoc = Lexer.getLoc(); 2265 MCAsmMacroParameter FA; 2266 2267 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2268 if (parseIdentifier(FA.Name)) { 2269 Error(IDLoc, "invalid argument identifier for formal argument"); 2270 eatToEndOfStatement(); 2271 return true; 2272 } 2273 2274 if (Lexer.isNot(AsmToken::Equal)) { 2275 TokError("expected '=' after formal parameter identifier"); 2276 eatToEndOfStatement(); 2277 return true; 2278 } 2279 Lex(); 2280 2281 NamedParametersFound = true; 2282 } 2283 2284 if (NamedParametersFound && FA.Name.empty()) { 2285 Error(IDLoc, "cannot mix positional and keyword arguments"); 2286 eatToEndOfStatement(); 2287 return true; 2288 } 2289 2290 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2291 if (parseMacroArgument(FA.Value, Vararg)) 2292 return true; 2293 2294 unsigned PI = Parameter; 2295 if (!FA.Name.empty()) { 2296 unsigned FAI = 0; 2297 for (FAI = 0; FAI < NParameters; ++FAI) 2298 if (M->Parameters[FAI].Name == FA.Name) 2299 break; 2300 2301 if (FAI >= NParameters) { 2302 assert(M && "expected macro to be defined"); 2303 Error(IDLoc, 2304 "parameter named '" + FA.Name + "' does not exist for macro '" + 2305 M->Name + "'"); 2306 return true; 2307 } 2308 PI = FAI; 2309 } 2310 2311 if (!FA.Value.empty()) { 2312 if (A.size() <= PI) 2313 A.resize(PI + 1); 2314 A[PI] = FA.Value; 2315 2316 if (FALocs.size() <= PI) 2317 FALocs.resize(PI + 1); 2318 2319 FALocs[PI] = Lexer.getLoc(); 2320 } 2321 2322 // At the end of the statement, fill in remaining arguments that have 2323 // default values. If there aren't any, then the next argument is 2324 // required but missing 2325 if (Lexer.is(AsmToken::EndOfStatement)) { 2326 bool Failure = false; 2327 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2328 if (A[FAI].empty()) { 2329 if (M->Parameters[FAI].Required) { 2330 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2331 "missing value for required parameter " 2332 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2333 Failure = true; 2334 } 2335 2336 if (!M->Parameters[FAI].Value.empty()) 2337 A[FAI] = M->Parameters[FAI].Value; 2338 } 2339 } 2340 return Failure; 2341 } 2342 2343 if (Lexer.is(AsmToken::Comma)) 2344 Lex(); 2345 } 2346 2347 return TokError("too many positional arguments"); 2348 } 2349 2350 const MCAsmMacro *AsmParser::lookupMacro(StringRef Name) { 2351 StringMap<MCAsmMacro>::iterator I = MacroMap.find(Name); 2352 return (I == MacroMap.end()) ? nullptr : &I->getValue(); 2353 } 2354 2355 void AsmParser::defineMacro(StringRef Name, MCAsmMacro Macro) { 2356 MacroMap.insert(std::make_pair(Name, std::move(Macro))); 2357 } 2358 2359 void AsmParser::undefineMacro(StringRef Name) { MacroMap.erase(Name); } 2360 2361 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2362 // Arbitrarily limit macro nesting depth, to match 'as'. We can eliminate 2363 // this, although we should protect against infinite loops. 2364 if (ActiveMacros.size() == 20) 2365 return TokError("macros cannot be nested more than 20 levels deep"); 2366 2367 MCAsmMacroArguments A; 2368 if (parseMacroArguments(M, A)) 2369 return true; 2370 2371 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2372 // to hold the macro body with substitutions. 2373 SmallString<256> Buf; 2374 StringRef Body = M->Body; 2375 raw_svector_ostream OS(Buf); 2376 2377 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2378 return true; 2379 2380 // We include the .endmacro in the buffer as our cue to exit the macro 2381 // instantiation. 2382 OS << ".endmacro\n"; 2383 2384 std::unique_ptr<MemoryBuffer> Instantiation = 2385 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2386 2387 // Create the macro instantiation object and add to the current macro 2388 // instantiation stack. 2389 MacroInstantiation *MI = new MacroInstantiation( 2390 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2391 ActiveMacros.push_back(MI); 2392 2393 ++NumOfMacroInstantiations; 2394 2395 // Jump to the macro instantiation and prime the lexer. 2396 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2397 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2398 Lex(); 2399 2400 return false; 2401 } 2402 2403 void AsmParser::handleMacroExit() { 2404 // Jump to the EndOfStatement we should return to, and consume it. 2405 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2406 Lex(); 2407 2408 // Pop the instantiation entry. 2409 delete ActiveMacros.back(); 2410 ActiveMacros.pop_back(); 2411 } 2412 2413 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2414 bool NoDeadStrip) { 2415 MCSymbol *Sym; 2416 const MCExpr *Value; 2417 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2418 Value)) 2419 return true; 2420 2421 if (!Sym) { 2422 // In the case where we parse an expression starting with a '.', we will 2423 // not generate an error, nor will we create a symbol. In this case we 2424 // should just return out. 2425 return false; 2426 } 2427 2428 // Do the assignment. 2429 Out.EmitAssignment(Sym, Value); 2430 if (NoDeadStrip) 2431 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2432 2433 return false; 2434 } 2435 2436 /// parseIdentifier: 2437 /// ::= identifier 2438 /// ::= string 2439 bool AsmParser::parseIdentifier(StringRef &Res) { 2440 // The assembler has relaxed rules for accepting identifiers, in particular we 2441 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2442 // separate tokens. At this level, we have already lexed so we cannot (currently) 2443 // handle this as a context dependent token, instead we detect adjacent tokens 2444 // and return the combined identifier. 2445 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2446 SMLoc PrefixLoc = getLexer().getLoc(); 2447 2448 // Consume the prefix character, and check for a following identifier. 2449 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2450 if (Lexer.isNot(AsmToken::Identifier)) 2451 return true; 2452 2453 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2454 if (PrefixLoc.getPointer() + 1 != getTok().getLoc().getPointer()) 2455 return true; 2456 2457 // Construct the joined identifier and consume the token. 2458 Res = 2459 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2460 Lex(); // Parser Lex to maintain invariants. 2461 return false; 2462 } 2463 2464 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2465 return true; 2466 2467 Res = getTok().getIdentifier(); 2468 2469 Lex(); // Consume the identifier token. 2470 2471 return false; 2472 } 2473 2474 /// parseDirectiveSet: 2475 /// ::= .equ identifier ',' expression 2476 /// ::= .equiv identifier ',' expression 2477 /// ::= .set identifier ',' expression 2478 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2479 StringRef Name; 2480 2481 if (check(parseIdentifier(Name), 2482 "expected identifier after '" + Twine(IDVal) + "'") || 2483 parseToken(AsmToken::Comma, "unexpected token in '" + Twine(IDVal) + "'")) 2484 return true; 2485 2486 return parseAssignment(Name, allow_redef, true); 2487 } 2488 2489 bool AsmParser::parseEscapedString(std::string &Data) { 2490 assert(getLexer().is(AsmToken::String) && "Unexpected current token!"); 2491 2492 Data = ""; 2493 StringRef Str = getTok().getStringContents(); 2494 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2495 if (Str[i] != '\\') { 2496 Data += Str[i]; 2497 continue; 2498 } 2499 2500 // Recognize escaped characters. Note that this escape semantics currently 2501 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2502 ++i; 2503 if (i == e) 2504 return TokError("unexpected backslash at end of string"); 2505 2506 // Recognize octal sequences. 2507 if ((unsigned)(Str[i] - '0') <= 7) { 2508 // Consume up to three octal characters. 2509 unsigned Value = Str[i] - '0'; 2510 2511 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2512 ++i; 2513 Value = Value * 8 + (Str[i] - '0'); 2514 2515 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2516 ++i; 2517 Value = Value * 8 + (Str[i] - '0'); 2518 } 2519 } 2520 2521 if (Value > 255) 2522 return TokError("invalid octal escape sequence (out of range)"); 2523 2524 Data += (unsigned char)Value; 2525 continue; 2526 } 2527 2528 // Otherwise recognize individual escapes. 2529 switch (Str[i]) { 2530 default: 2531 // Just reject invalid escape sequences for now. 2532 return TokError("invalid escape sequence (unrecognized character)"); 2533 2534 case 'b': Data += '\b'; break; 2535 case 'f': Data += '\f'; break; 2536 case 'n': Data += '\n'; break; 2537 case 'r': Data += '\r'; break; 2538 case 't': Data += '\t'; break; 2539 case '"': Data += '"'; break; 2540 case '\\': Data += '\\'; break; 2541 } 2542 } 2543 2544 Lex(); 2545 return false; 2546 } 2547 2548 /// parseDirectiveAscii: 2549 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2550 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2551 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2552 checkForValidSection(); 2553 2554 for (;;) { 2555 std::string Data; 2556 if (check(getTok().isNot(AsmToken::String), 2557 "expected string in '" + Twine(IDVal) + "' directive") || 2558 parseEscapedString(Data)) 2559 return true; 2560 2561 getStreamer().EmitBytes(Data); 2562 if (ZeroTerminated) 2563 getStreamer().EmitBytes(StringRef("\0", 1)); 2564 2565 if (getLexer().is(AsmToken::EndOfStatement)) 2566 break; 2567 2568 if (parseToken(AsmToken::Comma, 2569 "unexpected token in '" + Twine(IDVal) + "' directive")) 2570 return true; 2571 } 2572 } 2573 2574 Lex(); 2575 return false; 2576 } 2577 2578 /// parseDirectiveReloc 2579 /// ::= .reloc expression , identifier [ , expression ] 2580 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 2581 const MCExpr *Offset; 2582 const MCExpr *Expr = nullptr; 2583 2584 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 2585 if (parseExpression(Offset)) 2586 return true; 2587 2588 // We can only deal with constant expressions at the moment. 2589 int64_t OffsetValue; 2590 if (check(!Offset->evaluateAsAbsolute(OffsetValue), OffsetLoc, 2591 "expression is not a constant value") || 2592 check(OffsetValue < 0, OffsetLoc, "expression is negative") || 2593 parseToken(AsmToken::Comma, "expected comma") || 2594 check(getTok().isNot(AsmToken::Identifier), "expected relocation name")) 2595 return true; 2596 2597 SMLoc NameLoc = Lexer.getTok().getLoc(); 2598 StringRef Name = Lexer.getTok().getIdentifier(); 2599 Lex(); 2600 2601 if (Lexer.is(AsmToken::Comma)) { 2602 Lex(); 2603 SMLoc ExprLoc = Lexer.getLoc(); 2604 if (parseExpression(Expr)) 2605 return true; 2606 2607 MCValue Value; 2608 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 2609 return Error(ExprLoc, "expression must be relocatable"); 2610 } 2611 2612 if (parseToken(AsmToken::EndOfStatement, 2613 "unexpected token in .reloc directive") || 2614 check(getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc), 2615 NameLoc, "unknown relocation name")) 2616 return true; 2617 return false; 2618 } 2619 2620 /// parseDirectiveValue 2621 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 2622 bool AsmParser::parseDirectiveValue(unsigned Size) { 2623 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2624 checkForValidSection(); 2625 2626 for (;;) { 2627 const MCExpr *Value; 2628 SMLoc ExprLoc = getLexer().getLoc(); 2629 if (parseExpression(Value)) 2630 return true; 2631 2632 // Special case constant expressions to match code generator. 2633 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2634 assert(Size <= 8 && "Invalid size"); 2635 uint64_t IntValue = MCE->getValue(); 2636 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2637 return Error(ExprLoc, "literal value out of range for directive"); 2638 getStreamer().EmitIntValue(IntValue, Size); 2639 } else 2640 getStreamer().EmitValue(Value, Size, ExprLoc); 2641 2642 if (getLexer().is(AsmToken::EndOfStatement)) 2643 break; 2644 2645 // FIXME: Improve diagnostic. 2646 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 2647 return true; 2648 } 2649 } 2650 2651 Lex(); 2652 return false; 2653 } 2654 2655 /// ParseDirectiveOctaValue 2656 /// ::= .octa [ hexconstant (, hexconstant)* ] 2657 bool AsmParser::parseDirectiveOctaValue() { 2658 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2659 checkForValidSection(); 2660 2661 for (;;) { 2662 if (getTok().is(AsmToken::Error)) 2663 return true; 2664 if (getTok().isNot(AsmToken::Integer) && getTok().isNot(AsmToken::BigNum)) 2665 return TokError("unknown token in expression"); 2666 2667 SMLoc ExprLoc = getLexer().getLoc(); 2668 APInt IntValue = getTok().getAPIntVal(); 2669 Lex(); 2670 2671 uint64_t hi, lo; 2672 if (IntValue.isIntN(64)) { 2673 hi = 0; 2674 lo = IntValue.getZExtValue(); 2675 } else if (IntValue.isIntN(128)) { 2676 // It might actually have more than 128 bits, but the top ones are zero. 2677 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2678 lo = IntValue.getLoBits(64).getZExtValue(); 2679 } else 2680 return Error(ExprLoc, "literal value out of range for directive"); 2681 2682 if (MAI.isLittleEndian()) { 2683 getStreamer().EmitIntValue(lo, 8); 2684 getStreamer().EmitIntValue(hi, 8); 2685 } else { 2686 getStreamer().EmitIntValue(hi, 8); 2687 getStreamer().EmitIntValue(lo, 8); 2688 } 2689 2690 if (getLexer().is(AsmToken::EndOfStatement)) 2691 break; 2692 2693 // FIXME: Improve diagnostic. 2694 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 2695 return true; 2696 } 2697 } 2698 2699 Lex(); 2700 return false; 2701 } 2702 2703 /// parseDirectiveRealValue 2704 /// ::= (.single | .double) [ expression (, expression)* ] 2705 bool AsmParser::parseDirectiveRealValue(const fltSemantics &Semantics) { 2706 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2707 checkForValidSection(); 2708 2709 for (;;) { 2710 // We don't truly support arithmetic on floating point expressions, so we 2711 // have to manually parse unary prefixes. 2712 bool IsNeg = false; 2713 if (getLexer().is(AsmToken::Minus)) { 2714 Lexer.Lex(); 2715 IsNeg = true; 2716 } else if (getLexer().is(AsmToken::Plus)) 2717 Lexer.Lex(); 2718 2719 if (Lexer.is(AsmToken::Error)) 2720 return TokError(Lexer.getErr()); 2721 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 2722 Lexer.isNot(AsmToken::Identifier)) 2723 return TokError("unexpected token in directive"); 2724 2725 // Convert to an APFloat. 2726 APFloat Value(Semantics); 2727 StringRef IDVal = getTok().getString(); 2728 if (getLexer().is(AsmToken::Identifier)) { 2729 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2730 Value = APFloat::getInf(Semantics); 2731 else if (!IDVal.compare_lower("nan")) 2732 Value = APFloat::getNaN(Semantics, false, ~0); 2733 else 2734 return TokError("invalid floating point literal"); 2735 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 2736 APFloat::opInvalidOp) 2737 return TokError("invalid floating point literal"); 2738 if (IsNeg) 2739 Value.changeSign(); 2740 2741 // Consume the numeric token. 2742 Lex(); 2743 2744 // Emit the value as an integer. 2745 APInt AsInt = Value.bitcastToAPInt(); 2746 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 2747 AsInt.getBitWidth() / 8); 2748 2749 if (Lexer.is(AsmToken::EndOfStatement)) 2750 break; 2751 2752 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 2753 return true; 2754 } 2755 } 2756 2757 Lex(); 2758 return false; 2759 } 2760 2761 /// parseDirectiveZero 2762 /// ::= .zero expression 2763 bool AsmParser::parseDirectiveZero() { 2764 checkForValidSection(); 2765 2766 SMLoc NumBytesLoc = Lexer.getLoc(); 2767 const MCExpr *NumBytes; 2768 if (parseExpression(NumBytes)) 2769 return true; 2770 2771 int64_t Val = 0; 2772 if (getLexer().is(AsmToken::Comma)) { 2773 Lex(); 2774 if (parseAbsoluteExpression(Val)) 2775 return true; 2776 } 2777 2778 if (parseToken(AsmToken::EndOfStatement, 2779 "unexpected token in '.zero' directive")) 2780 return true; 2781 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 2782 2783 return false; 2784 } 2785 2786 /// parseDirectiveFill 2787 /// ::= .fill expression [ , expression [ , expression ] ] 2788 bool AsmParser::parseDirectiveFill() { 2789 checkForValidSection(); 2790 2791 SMLoc NumValuesLoc = Lexer.getLoc(); 2792 const MCExpr *NumValues; 2793 if (parseExpression(NumValues)) 2794 return true; 2795 2796 int64_t FillSize = 1; 2797 int64_t FillExpr = 0; 2798 2799 SMLoc SizeLoc, ExprLoc; 2800 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2801 2802 if (parseToken(AsmToken::Comma, "unexpected token in '.fill' directive") || 2803 getTokenLoc(SizeLoc) || parseAbsoluteExpression(FillSize)) 2804 return true; 2805 2806 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2807 if (parseToken(AsmToken::Comma, 2808 "unexpected token in '.fill' directive") || 2809 getTokenLoc(ExprLoc) || parseAbsoluteExpression(FillExpr) || 2810 parseToken(AsmToken::EndOfStatement, 2811 "unexpected token in '.fill' directive")) 2812 return true; 2813 } 2814 } 2815 2816 if (FillSize < 0) { 2817 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 2818 return false; 2819 } 2820 if (FillSize > 8) { 2821 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 2822 FillSize = 8; 2823 } 2824 2825 if (!isUInt<32>(FillExpr) && FillSize > 4) 2826 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 2827 2828 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 2829 2830 return false; 2831 } 2832 2833 /// parseDirectiveOrg 2834 /// ::= .org expression [ , expression ] 2835 bool AsmParser::parseDirectiveOrg() { 2836 checkForValidSection(); 2837 2838 const MCExpr *Offset; 2839 if (parseExpression(Offset)) 2840 return true; 2841 2842 // Parse optional fill expression. 2843 int64_t FillExpr = 0; 2844 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2845 if (parseToken(AsmToken::Comma, "unexpected token in '.org' directive") || 2846 parseAbsoluteExpression(FillExpr)) 2847 return true; 2848 } 2849 2850 if (parseToken(AsmToken::EndOfStatement, 2851 "unexpected token in '.org' directive")) 2852 return true; 2853 2854 getStreamer().emitValueToOffset(Offset, FillExpr); 2855 return false; 2856 } 2857 2858 /// parseDirectiveAlign 2859 /// ::= {.align, ...} expression [ , expression [ , expression ]] 2860 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 2861 checkForValidSection(); 2862 2863 SMLoc AlignmentLoc = getLexer().getLoc(); 2864 int64_t Alignment; 2865 if (parseAbsoluteExpression(Alignment)) 2866 return true; 2867 2868 SMLoc MaxBytesLoc; 2869 bool HasFillExpr = false; 2870 int64_t FillExpr = 0; 2871 int64_t MaxBytesToFill = 0; 2872 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2873 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 2874 return true; 2875 2876 // The fill expression can be omitted while specifying a maximum number of 2877 // alignment bytes, e.g: 2878 // .align 3,,4 2879 if (getTok().isNot(AsmToken::Comma)) { 2880 HasFillExpr = true; 2881 if (parseAbsoluteExpression(FillExpr)) 2882 return true; 2883 } 2884 2885 if (getTok().isNot(AsmToken::EndOfStatement)) { 2886 if (parseToken(AsmToken::Comma, "unexpected token in directive") || 2887 getTokenLoc(MaxBytesLoc) || parseAbsoluteExpression(MaxBytesToFill)) 2888 return true; 2889 } 2890 } 2891 2892 if (parseToken(AsmToken::EndOfStatement, "unexpected token in directive")) 2893 return true; 2894 2895 if (!HasFillExpr) 2896 FillExpr = 0; 2897 2898 // Compute alignment in bytes. 2899 if (IsPow2) { 2900 // FIXME: Diagnose overflow. 2901 if (Alignment >= 32) { 2902 Error(AlignmentLoc, "invalid alignment value"); 2903 Alignment = 31; 2904 } 2905 2906 Alignment = 1ULL << Alignment; 2907 } else { 2908 // Reject alignments that aren't either a power of two or zero, 2909 // for gas compatibility. Alignment of zero is silently rounded 2910 // up to one. 2911 if (Alignment == 0) 2912 Alignment = 1; 2913 if (!isPowerOf2_64(Alignment)) 2914 Error(AlignmentLoc, "alignment must be a power of 2"); 2915 } 2916 2917 // Diagnose non-sensical max bytes to align. 2918 if (MaxBytesLoc.isValid()) { 2919 if (MaxBytesToFill < 1) { 2920 Error(MaxBytesLoc, "alignment directive can never be satisfied in this " 2921 "many bytes, ignoring maximum bytes expression"); 2922 MaxBytesToFill = 0; 2923 } 2924 2925 if (MaxBytesToFill >= Alignment) { 2926 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 2927 "has no effect"); 2928 MaxBytesToFill = 0; 2929 } 2930 } 2931 2932 // Check whether we should use optimal code alignment for this .align 2933 // directive. 2934 const MCSection *Section = getStreamer().getCurrentSection().first; 2935 assert(Section && "must have section to emit alignment"); 2936 bool UseCodeAlign = Section->UseCodeAlign(); 2937 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 2938 ValueSize == 1 && UseCodeAlign) { 2939 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 2940 } else { 2941 // FIXME: Target specific behavior about how the "extra" bytes are filled. 2942 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 2943 MaxBytesToFill); 2944 } 2945 2946 return false; 2947 } 2948 2949 /// parseDirectiveFile 2950 /// ::= .file [number] filename 2951 /// ::= .file number directory filename 2952 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 2953 // FIXME: I'm not sure what this is. 2954 int64_t FileNumber = -1; 2955 SMLoc FileNumberLoc = getLexer().getLoc(); 2956 if (getLexer().is(AsmToken::Integer)) { 2957 FileNumber = getTok().getIntVal(); 2958 Lex(); 2959 2960 if (FileNumber < 1) 2961 return TokError("file number less than one"); 2962 } 2963 2964 std::string Path = getTok().getString(); 2965 2966 // Usually the directory and filename together, otherwise just the directory. 2967 // Allow the strings to have escaped octal character sequence. 2968 if (check(getTok().isNot(AsmToken::String), 2969 "unexpected token in '.file' directive") || 2970 parseEscapedString(Path)) 2971 return true; 2972 2973 StringRef Directory; 2974 StringRef Filename; 2975 std::string FilenameData; 2976 if (getLexer().is(AsmToken::String)) { 2977 if (check(FileNumber == -1, 2978 "explicit path specified, but no file number") || 2979 parseEscapedString(FilenameData)) 2980 return true; 2981 Filename = FilenameData; 2982 Directory = Path; 2983 } else { 2984 Filename = Path; 2985 } 2986 2987 if (parseToken(AsmToken::EndOfStatement, 2988 "unexpected token in '.file' directive")) 2989 return true; 2990 2991 if (FileNumber == -1) 2992 getStreamer().EmitFileDirective(Filename); 2993 else { 2994 // If there is -g option as well as debug info from directive file, 2995 // we turn off -g option, directly use the existing debug info instead. 2996 if (getContext().getGenDwarfForAssembly()) 2997 getContext().setGenDwarfForAssembly(false); 2998 else if (getStreamer().EmitDwarfFileDirective(FileNumber, Directory, Filename) == 2999 0) 3000 Error(FileNumberLoc, "file number already allocated"); 3001 } 3002 3003 return false; 3004 } 3005 3006 /// parseDirectiveLine 3007 /// ::= .line [number] 3008 bool AsmParser::parseDirectiveLine() { 3009 int64_t LineNumber; 3010 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3011 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3012 return true; 3013 (void)LineNumber; 3014 // FIXME: Do something with the .line. 3015 } 3016 if (parseToken(AsmToken::EndOfStatement, 3017 "unexpected token in '.line' directive")) 3018 return true; 3019 3020 return false; 3021 } 3022 3023 /// parseDirectiveLoc 3024 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3025 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3026 /// The first number is a file number, must have been previously assigned with 3027 /// a .file directive, the second number is the line number and optionally the 3028 /// third number is a column position (zero if not specified). The remaining 3029 /// optional items are .loc sub-directives. 3030 bool AsmParser::parseDirectiveLoc() { 3031 int64_t FileNumber = 0, LineNumber = 0; 3032 SMLoc Loc = getTok().getLoc(); 3033 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3034 check(FileNumber < 1, Loc, 3035 "file number less than one in '.loc' directive") || 3036 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3037 "unassigned file number in '.loc' directive")) 3038 return true; 3039 3040 // optional 3041 if (getLexer().is(AsmToken::Integer)) { 3042 LineNumber = getTok().getIntVal(); 3043 if (LineNumber < 0) 3044 return TokError("line number less than zero in '.loc' directive"); 3045 Lex(); 3046 } 3047 3048 int64_t ColumnPos = 0; 3049 if (getLexer().is(AsmToken::Integer)) { 3050 ColumnPos = getTok().getIntVal(); 3051 if (ColumnPos < 0) 3052 return TokError("column position less than zero in '.loc' directive"); 3053 Lex(); 3054 } 3055 3056 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3057 unsigned Isa = 0; 3058 int64_t Discriminator = 0; 3059 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3060 for (;;) { 3061 if (getLexer().is(AsmToken::EndOfStatement)) 3062 break; 3063 3064 StringRef Name; 3065 SMLoc Loc = getTok().getLoc(); 3066 if (parseIdentifier(Name)) 3067 return TokError("unexpected token in '.loc' directive"); 3068 3069 if (Name == "basic_block") 3070 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3071 else if (Name == "prologue_end") 3072 Flags |= DWARF2_FLAG_PROLOGUE_END; 3073 else if (Name == "epilogue_begin") 3074 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3075 else if (Name == "is_stmt") { 3076 Loc = getTok().getLoc(); 3077 const MCExpr *Value; 3078 if (parseExpression(Value)) 3079 return true; 3080 // The expression must be the constant 0 or 1. 3081 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3082 int Value = MCE->getValue(); 3083 if (Value == 0) 3084 Flags &= ~DWARF2_FLAG_IS_STMT; 3085 else if (Value == 1) 3086 Flags |= DWARF2_FLAG_IS_STMT; 3087 else 3088 return Error(Loc, "is_stmt value not 0 or 1"); 3089 } else { 3090 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3091 } 3092 } else if (Name == "isa") { 3093 Loc = getTok().getLoc(); 3094 const MCExpr *Value; 3095 if (parseExpression(Value)) 3096 return true; 3097 // The expression must be a constant greater or equal to 0. 3098 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3099 int Value = MCE->getValue(); 3100 if (Value < 0) 3101 return Error(Loc, "isa number less than zero"); 3102 Isa = Value; 3103 } else { 3104 return Error(Loc, "isa number not a constant value"); 3105 } 3106 } else if (Name == "discriminator") { 3107 if (parseAbsoluteExpression(Discriminator)) 3108 return true; 3109 } else { 3110 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3111 } 3112 3113 if (getLexer().is(AsmToken::EndOfStatement)) 3114 break; 3115 } 3116 } 3117 Lex(); 3118 3119 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3120 Isa, Discriminator, StringRef()); 3121 3122 return false; 3123 } 3124 3125 /// parseDirectiveStabs 3126 /// ::= .stabs string, number, number, number 3127 bool AsmParser::parseDirectiveStabs() { 3128 return TokError("unsupported directive '.stabs'"); 3129 } 3130 3131 /// parseDirectiveCVFile 3132 /// ::= .cv_file number filename 3133 bool AsmParser::parseDirectiveCVFile() { 3134 SMLoc FileNumberLoc = getTok().getLoc(); 3135 int64_t FileNumber; 3136 std::string Filename; 3137 3138 if (parseIntToken(FileNumber, 3139 "expected file number in '.cv_file' directive") || 3140 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3141 check(getTok().isNot(AsmToken::String), 3142 "unexpected token in '.cv_file' directive") || 3143 // Usually directory and filename are together, otherwise just 3144 // directory. Allow the strings to have escaped octal character sequence. 3145 parseEscapedString(Filename) || 3146 parseToken(AsmToken::EndOfStatement, 3147 "unexpected token in '.cv_file' directive") || 3148 check(getStreamer().EmitCVFileDirective(FileNumber, Filename) == 0, 3149 FileNumberLoc, "file number already allocated")) 3150 return true; 3151 3152 return false; 3153 } 3154 3155 /// parseDirectiveCVLoc 3156 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3157 /// [is_stmt VALUE] 3158 /// The first number is a file number, must have been previously assigned with 3159 /// a .file directive, the second number is the line number and optionally the 3160 /// third number is a column position (zero if not specified). The remaining 3161 /// optional items are .loc sub-directives. 3162 bool AsmParser::parseDirectiveCVLoc() { 3163 SMLoc Loc; 3164 int64_t FunctionId, FileNumber; 3165 if (getTokenLoc(Loc) || 3166 parseIntToken(FunctionId, "unexpected token in '.cv_loc' directive") || 3167 check(FunctionId < 0, Loc, 3168 "function id less than zero in '.cv_loc' directive") || 3169 getTokenLoc(Loc) || 3170 parseIntToken(FileNumber, "expected integer in '.cv_loc' directive") || 3171 check(FileNumber < 1, Loc, 3172 "file number less than one in '.cv_loc' directive") || 3173 check(!getContext().isValidCVFileNumber(FileNumber), Loc, 3174 "unassigned file number in '.cv_loc' directive")) 3175 return true; 3176 3177 int64_t LineNumber = 0; 3178 if (getLexer().is(AsmToken::Integer)) { 3179 LineNumber = getTok().getIntVal(); 3180 if (LineNumber < 0) 3181 return TokError("line number less than zero in '.cv_loc' directive"); 3182 Lex(); 3183 } 3184 3185 int64_t ColumnPos = 0; 3186 if (getLexer().is(AsmToken::Integer)) { 3187 ColumnPos = getTok().getIntVal(); 3188 if (ColumnPos < 0) 3189 return TokError("column position less than zero in '.cv_loc' directive"); 3190 Lex(); 3191 } 3192 3193 bool PrologueEnd = false; 3194 uint64_t IsStmt = 0; 3195 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3196 StringRef Name; 3197 SMLoc Loc = getTok().getLoc(); 3198 if (parseIdentifier(Name)) 3199 return TokError("unexpected token in '.cv_loc' directive"); 3200 3201 if (Name == "prologue_end") 3202 PrologueEnd = true; 3203 else if (Name == "is_stmt") { 3204 Loc = getTok().getLoc(); 3205 const MCExpr *Value; 3206 if (parseExpression(Value)) 3207 return true; 3208 // The expression must be the constant 0 or 1. 3209 IsStmt = ~0ULL; 3210 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3211 IsStmt = MCE->getValue(); 3212 3213 if (IsStmt > 1) 3214 return Error(Loc, "is_stmt value not 0 or 1"); 3215 } else { 3216 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3217 } 3218 } 3219 Lex(); 3220 3221 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3222 ColumnPos, PrologueEnd, IsStmt, StringRef()); 3223 return false; 3224 } 3225 3226 /// parseDirectiveCVLinetable 3227 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3228 bool AsmParser::parseDirectiveCVLinetable() { 3229 int64_t FunctionId; 3230 StringRef FnStartName, FnEndName; 3231 SMLoc Loc = getTok().getLoc(); 3232 if (parseIntToken(FunctionId, 3233 "expected Integer in '.cv_linetable' directive") || 3234 check(FunctionId < 0, Loc, 3235 "function id less than zero in '.cv_linetable' directive") || 3236 parseToken(AsmToken::Comma, 3237 "unexpected token in '.cv_linetable' directive") || 3238 getTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3239 "expected identifier in directive") || 3240 parseToken(AsmToken::Comma, 3241 "unexpected token in '.cv_linetable' directive") || 3242 getTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3243 "expected identifier in directive")) 3244 return true; 3245 3246 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3247 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3248 3249 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3250 return false; 3251 } 3252 3253 /// parseDirectiveCVInlineLinetable 3254 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3255 /// ("contains" SecondaryFunctionId+)? 3256 bool AsmParser::parseDirectiveCVInlineLinetable() { 3257 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3258 StringRef FnStartName, FnEndName; 3259 SMLoc Loc = getTok().getLoc(); 3260 if (parseIntToken( 3261 PrimaryFunctionId, 3262 "expected PrimaryFunctionId in '.cv_inline_linetable' directive") || 3263 check(PrimaryFunctionId < 0, Loc, 3264 "function id less than zero in '.cv_inline_linetable' directive") || 3265 getTokenLoc(Loc) || 3266 parseIntToken( 3267 SourceFileId, 3268 "expected SourceField in '.cv_inline_linetable' directive") || 3269 check(SourceFileId <= 0, Loc, 3270 "File id less than zero in '.cv_inline_linetable' directive") || 3271 getTokenLoc(Loc) || 3272 parseIntToken( 3273 SourceLineNum, 3274 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3275 check(SourceLineNum < 0, Loc, 3276 "Line number less than zero in '.cv_inline_linetable' directive") || 3277 getTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3278 "expected identifier in directive") || 3279 getTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3280 "expected identifier in directive")) 3281 return true; 3282 3283 SmallVector<unsigned, 8> SecondaryFunctionIds; 3284 if (getLexer().is(AsmToken::Identifier)) { 3285 if (getTok().getIdentifier() != "contains") 3286 return TokError( 3287 "unexpected identifier in '.cv_inline_linetable' directive"); 3288 Lex(); 3289 3290 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3291 int64_t SecondaryFunctionId = getTok().getIntVal(); 3292 if (SecondaryFunctionId < 0) 3293 return TokError( 3294 "function id less than zero in '.cv_inline_linetable' directive"); 3295 Lex(); 3296 3297 SecondaryFunctionIds.push_back(SecondaryFunctionId); 3298 } 3299 } 3300 3301 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3302 return true; 3303 3304 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3305 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3306 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3307 SourceLineNum, FnStartSym, 3308 FnEndSym, SecondaryFunctionIds); 3309 return false; 3310 } 3311 3312 /// parseDirectiveCVDefRange 3313 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3314 bool AsmParser::parseDirectiveCVDefRange() { 3315 SMLoc Loc; 3316 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3317 while (getLexer().is(AsmToken::Identifier)) { 3318 Loc = getLexer().getLoc(); 3319 StringRef GapStartName; 3320 if (parseIdentifier(GapStartName)) 3321 return Error(Loc, "expected identifier in directive"); 3322 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3323 3324 Loc = getLexer().getLoc(); 3325 StringRef GapEndName; 3326 if (parseIdentifier(GapEndName)) 3327 return Error(Loc, "expected identifier in directive"); 3328 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3329 3330 Ranges.push_back({GapStartSym, GapEndSym}); 3331 } 3332 3333 std::string FixedSizePortion; 3334 if (parseToken(AsmToken::Comma, "unexpected token in directive") || 3335 parseEscapedString(FixedSizePortion)) 3336 return true; 3337 3338 getStreamer().EmitCVDefRangeDirective(Ranges, FixedSizePortion); 3339 return false; 3340 } 3341 3342 /// parseDirectiveCVStringTable 3343 /// ::= .cv_stringtable 3344 bool AsmParser::parseDirectiveCVStringTable() { 3345 getStreamer().EmitCVStringTableDirective(); 3346 return false; 3347 } 3348 3349 /// parseDirectiveCVFileChecksums 3350 /// ::= .cv_filechecksums 3351 bool AsmParser::parseDirectiveCVFileChecksums() { 3352 getStreamer().EmitCVFileChecksumsDirective(); 3353 return false; 3354 } 3355 3356 /// parseDirectiveCFISections 3357 /// ::= .cfi_sections section [, section] 3358 bool AsmParser::parseDirectiveCFISections() { 3359 StringRef Name; 3360 bool EH = false; 3361 bool Debug = false; 3362 3363 if (parseIdentifier(Name)) 3364 return TokError("Expected an identifier"); 3365 3366 if (Name == ".eh_frame") 3367 EH = true; 3368 else if (Name == ".debug_frame") 3369 Debug = true; 3370 3371 if (getLexer().is(AsmToken::Comma)) { 3372 Lex(); 3373 3374 if (parseIdentifier(Name)) 3375 return TokError("Expected an identifier"); 3376 3377 if (Name == ".eh_frame") 3378 EH = true; 3379 else if (Name == ".debug_frame") 3380 Debug = true; 3381 } 3382 3383 getStreamer().EmitCFISections(EH, Debug); 3384 return false; 3385 } 3386 3387 /// parseDirectiveCFIStartProc 3388 /// ::= .cfi_startproc [simple] 3389 bool AsmParser::parseDirectiveCFIStartProc() { 3390 StringRef Simple; 3391 if (getLexer().isNot(AsmToken::EndOfStatement)) 3392 if (parseIdentifier(Simple) || Simple != "simple") 3393 return TokError("unexpected token in .cfi_startproc directive"); 3394 3395 if (parseToken(AsmToken::EndOfStatement, "Expected end of statement")) 3396 return true; 3397 3398 getStreamer().EmitCFIStartProc(!Simple.empty()); 3399 return false; 3400 } 3401 3402 /// parseDirectiveCFIEndProc 3403 /// ::= .cfi_endproc 3404 bool AsmParser::parseDirectiveCFIEndProc() { 3405 getStreamer().EmitCFIEndProc(); 3406 return false; 3407 } 3408 3409 /// \brief parse register name or number. 3410 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 3411 SMLoc DirectiveLoc) { 3412 unsigned RegNo; 3413 3414 if (getLexer().isNot(AsmToken::Integer)) { 3415 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 3416 return true; 3417 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 3418 } else 3419 return parseAbsoluteExpression(Register); 3420 3421 return false; 3422 } 3423 3424 /// parseDirectiveCFIDefCfa 3425 /// ::= .cfi_def_cfa register, offset 3426 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3427 int64_t Register = 0, Offset = 0; 3428 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3429 parseToken(AsmToken::Comma, "unexpected token in directive") || 3430 parseAbsoluteExpression(Offset)) 3431 return true; 3432 3433 getStreamer().EmitCFIDefCfa(Register, Offset); 3434 return false; 3435 } 3436 3437 /// parseDirectiveCFIDefCfaOffset 3438 /// ::= .cfi_def_cfa_offset offset 3439 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3440 int64_t Offset = 0; 3441 if (parseAbsoluteExpression(Offset)) 3442 return true; 3443 3444 getStreamer().EmitCFIDefCfaOffset(Offset); 3445 return false; 3446 } 3447 3448 /// parseDirectiveCFIRegister 3449 /// ::= .cfi_register register, register 3450 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3451 int64_t Register1 = 0, Register2 = 0; 3452 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 3453 parseToken(AsmToken::Comma, "unexpected token in directive") || 3454 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3455 return true; 3456 3457 getStreamer().EmitCFIRegister(Register1, Register2); 3458 return false; 3459 } 3460 3461 /// parseDirectiveCFIWindowSave 3462 /// ::= .cfi_window_save 3463 bool AsmParser::parseDirectiveCFIWindowSave() { 3464 getStreamer().EmitCFIWindowSave(); 3465 return false; 3466 } 3467 3468 /// parseDirectiveCFIAdjustCfaOffset 3469 /// ::= .cfi_adjust_cfa_offset adjustment 3470 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 3471 int64_t Adjustment = 0; 3472 if (parseAbsoluteExpression(Adjustment)) 3473 return true; 3474 3475 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 3476 return false; 3477 } 3478 3479 /// parseDirectiveCFIDefCfaRegister 3480 /// ::= .cfi_def_cfa_register register 3481 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3482 int64_t Register = 0; 3483 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3484 return true; 3485 3486 getStreamer().EmitCFIDefCfaRegister(Register); 3487 return false; 3488 } 3489 3490 /// parseDirectiveCFIOffset 3491 /// ::= .cfi_offset register, offset 3492 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3493 int64_t Register = 0; 3494 int64_t Offset = 0; 3495 3496 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3497 parseToken(AsmToken::Comma, "unexpected token in directive") || 3498 parseAbsoluteExpression(Offset)) 3499 return true; 3500 3501 getStreamer().EmitCFIOffset(Register, Offset); 3502 return false; 3503 } 3504 3505 /// parseDirectiveCFIRelOffset 3506 /// ::= .cfi_rel_offset register, offset 3507 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3508 int64_t Register = 0, Offset = 0; 3509 3510 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3511 parseToken(AsmToken::Comma, "unexpected token in directive") || 3512 parseAbsoluteExpression(Offset)) 3513 return true; 3514 3515 getStreamer().EmitCFIRelOffset(Register, Offset); 3516 return false; 3517 } 3518 3519 static bool isValidEncoding(int64_t Encoding) { 3520 if (Encoding & ~0xff) 3521 return false; 3522 3523 if (Encoding == dwarf::DW_EH_PE_omit) 3524 return true; 3525 3526 const unsigned Format = Encoding & 0xf; 3527 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3528 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3529 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3530 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3531 return false; 3532 3533 const unsigned Application = Encoding & 0x70; 3534 if (Application != dwarf::DW_EH_PE_absptr && 3535 Application != dwarf::DW_EH_PE_pcrel) 3536 return false; 3537 3538 return true; 3539 } 3540 3541 /// parseDirectiveCFIPersonalityOrLsda 3542 /// IsPersonality true for cfi_personality, false for cfi_lsda 3543 /// ::= .cfi_personality encoding, [symbol_name] 3544 /// ::= .cfi_lsda encoding, [symbol_name] 3545 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3546 int64_t Encoding = 0; 3547 if (parseAbsoluteExpression(Encoding)) 3548 return true; 3549 if (Encoding == dwarf::DW_EH_PE_omit) 3550 return false; 3551 3552 StringRef Name; 3553 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 3554 parseToken(AsmToken::Comma, "unexpected token in directive") || 3555 check(parseIdentifier(Name), "expected identifier in directive")) 3556 return true; 3557 3558 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 3559 3560 if (IsPersonality) 3561 getStreamer().EmitCFIPersonality(Sym, Encoding); 3562 else 3563 getStreamer().EmitCFILsda(Sym, Encoding); 3564 return false; 3565 } 3566 3567 /// parseDirectiveCFIRememberState 3568 /// ::= .cfi_remember_state 3569 bool AsmParser::parseDirectiveCFIRememberState() { 3570 getStreamer().EmitCFIRememberState(); 3571 return false; 3572 } 3573 3574 /// parseDirectiveCFIRestoreState 3575 /// ::= .cfi_remember_state 3576 bool AsmParser::parseDirectiveCFIRestoreState() { 3577 getStreamer().EmitCFIRestoreState(); 3578 return false; 3579 } 3580 3581 /// parseDirectiveCFISameValue 3582 /// ::= .cfi_same_value register 3583 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3584 int64_t Register = 0; 3585 3586 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3587 return true; 3588 3589 getStreamer().EmitCFISameValue(Register); 3590 return false; 3591 } 3592 3593 /// parseDirectiveCFIRestore 3594 /// ::= .cfi_restore register 3595 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3596 int64_t Register = 0; 3597 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3598 return true; 3599 3600 getStreamer().EmitCFIRestore(Register); 3601 return false; 3602 } 3603 3604 /// parseDirectiveCFIEscape 3605 /// ::= .cfi_escape expression[,...] 3606 bool AsmParser::parseDirectiveCFIEscape() { 3607 std::string Values; 3608 int64_t CurrValue; 3609 if (parseAbsoluteExpression(CurrValue)) 3610 return true; 3611 3612 Values.push_back((uint8_t)CurrValue); 3613 3614 while (getLexer().is(AsmToken::Comma)) { 3615 Lex(); 3616 3617 if (parseAbsoluteExpression(CurrValue)) 3618 return true; 3619 3620 Values.push_back((uint8_t)CurrValue); 3621 } 3622 3623 getStreamer().EmitCFIEscape(Values); 3624 return false; 3625 } 3626 3627 /// parseDirectiveCFISignalFrame 3628 /// ::= .cfi_signal_frame 3629 bool AsmParser::parseDirectiveCFISignalFrame() { 3630 if (parseToken(AsmToken::EndOfStatement, 3631 "unexpected token in '.cfi_signal_frame'")) 3632 return true; 3633 3634 getStreamer().EmitCFISignalFrame(); 3635 return false; 3636 } 3637 3638 /// parseDirectiveCFIUndefined 3639 /// ::= .cfi_undefined register 3640 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 3641 int64_t Register = 0; 3642 3643 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3644 return true; 3645 3646 getStreamer().EmitCFIUndefined(Register); 3647 return false; 3648 } 3649 3650 /// parseDirectiveMacrosOnOff 3651 /// ::= .macros_on 3652 /// ::= .macros_off 3653 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 3654 if (parseToken(AsmToken::EndOfStatement, 3655 "unexpected token in '" + Directive + "' directive")) 3656 return true; 3657 3658 setMacrosEnabled(Directive == ".macros_on"); 3659 return false; 3660 } 3661 3662 /// parseDirectiveMacro 3663 /// ::= .macro name[,] [parameters] 3664 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 3665 StringRef Name; 3666 if (parseIdentifier(Name)) 3667 return TokError("expected identifier in '.macro' directive"); 3668 3669 if (getLexer().is(AsmToken::Comma)) 3670 Lex(); 3671 3672 MCAsmMacroParameters Parameters; 3673 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3674 3675 if (!Parameters.empty() && Parameters.back().Vararg) 3676 return Error(Lexer.getLoc(), 3677 "Vararg parameter '" + Parameters.back().Name + 3678 "' should be last one in the list of parameters."); 3679 3680 MCAsmMacroParameter Parameter; 3681 if (parseIdentifier(Parameter.Name)) 3682 return TokError("expected identifier in '.macro' directive"); 3683 3684 if (Lexer.is(AsmToken::Colon)) { 3685 Lex(); // consume ':' 3686 3687 SMLoc QualLoc; 3688 StringRef Qualifier; 3689 3690 QualLoc = Lexer.getLoc(); 3691 if (parseIdentifier(Qualifier)) 3692 return Error(QualLoc, "missing parameter qualifier for " 3693 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3694 3695 if (Qualifier == "req") 3696 Parameter.Required = true; 3697 else if (Qualifier == "vararg") 3698 Parameter.Vararg = true; 3699 else 3700 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 3701 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 3702 } 3703 3704 if (getLexer().is(AsmToken::Equal)) { 3705 Lex(); 3706 3707 SMLoc ParamLoc; 3708 3709 ParamLoc = Lexer.getLoc(); 3710 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 3711 return true; 3712 3713 if (Parameter.Required) 3714 Warning(ParamLoc, "pointless default value for required parameter " 3715 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3716 } 3717 3718 Parameters.push_back(std::move(Parameter)); 3719 3720 if (getLexer().is(AsmToken::Comma)) 3721 Lex(); 3722 } 3723 3724 // Eat just the end of statement. 3725 Lexer.Lex(); 3726 3727 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 3728 AsmToken EndToken, StartToken = getTok(); 3729 unsigned MacroDepth = 0; 3730 // Lex the macro definition. 3731 for (;;) { 3732 // Ignore Lexing errors in macros. 3733 while (Lexer.is(AsmToken::Error)) { 3734 Lexer.Lex(); 3735 } 3736 3737 // Check whether we have reached the end of the file. 3738 if (getLexer().is(AsmToken::Eof)) 3739 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 3740 3741 // Otherwise, check whether we have reach the .endmacro. 3742 if (getLexer().is(AsmToken::Identifier)) { 3743 if (getTok().getIdentifier() == ".endm" || 3744 getTok().getIdentifier() == ".endmacro") { 3745 if (MacroDepth == 0) { // Outermost macro. 3746 EndToken = getTok(); 3747 Lexer.Lex(); 3748 if (getLexer().isNot(AsmToken::EndOfStatement)) 3749 return TokError("unexpected token in '" + EndToken.getIdentifier() + 3750 "' directive"); 3751 break; 3752 } else { 3753 // Otherwise we just found the end of an inner macro. 3754 --MacroDepth; 3755 } 3756 } else if (getTok().getIdentifier() == ".macro") { 3757 // We allow nested macros. Those aren't instantiated until the outermost 3758 // macro is expanded so just ignore them for now. 3759 ++MacroDepth; 3760 } 3761 } 3762 3763 // Otherwise, scan til the end of the statement. 3764 eatToEndOfStatement(); 3765 } 3766 3767 if (lookupMacro(Name)) { 3768 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 3769 } 3770 3771 const char *BodyStart = StartToken.getLoc().getPointer(); 3772 const char *BodyEnd = EndToken.getLoc().getPointer(); 3773 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 3774 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 3775 defineMacro(Name, MCAsmMacro(Name, Body, std::move(Parameters))); 3776 return false; 3777 } 3778 3779 /// checkForBadMacro 3780 /// 3781 /// With the support added for named parameters there may be code out there that 3782 /// is transitioning from positional parameters. In versions of gas that did 3783 /// not support named parameters they would be ignored on the macro definition. 3784 /// But to support both styles of parameters this is not possible so if a macro 3785 /// definition has named parameters but does not use them and has what appears 3786 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 3787 /// warning that the positional parameter found in body which have no effect. 3788 /// Hoping the developer will either remove the named parameters from the macro 3789 /// definition so the positional parameters get used if that was what was 3790 /// intended or change the macro to use the named parameters. It is possible 3791 /// this warning will trigger when the none of the named parameters are used 3792 /// and the strings like $1 are infact to simply to be passed trough unchanged. 3793 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 3794 StringRef Body, 3795 ArrayRef<MCAsmMacroParameter> Parameters) { 3796 // If this macro is not defined with named parameters the warning we are 3797 // checking for here doesn't apply. 3798 unsigned NParameters = Parameters.size(); 3799 if (NParameters == 0) 3800 return; 3801 3802 bool NamedParametersFound = false; 3803 bool PositionalParametersFound = false; 3804 3805 // Look at the body of the macro for use of both the named parameters and what 3806 // are likely to be positional parameters. This is what expandMacro() is 3807 // doing when it finds the parameters in the body. 3808 while (!Body.empty()) { 3809 // Scan for the next possible parameter. 3810 std::size_t End = Body.size(), Pos = 0; 3811 for (; Pos != End; ++Pos) { 3812 // Check for a substitution or escape. 3813 // This macro is defined with parameters, look for \foo, \bar, etc. 3814 if (Body[Pos] == '\\' && Pos + 1 != End) 3815 break; 3816 3817 // This macro should have parameters, but look for $0, $1, ..., $n too. 3818 if (Body[Pos] != '$' || Pos + 1 == End) 3819 continue; 3820 char Next = Body[Pos + 1]; 3821 if (Next == '$' || Next == 'n' || 3822 isdigit(static_cast<unsigned char>(Next))) 3823 break; 3824 } 3825 3826 // Check if we reached the end. 3827 if (Pos == End) 3828 break; 3829 3830 if (Body[Pos] == '$') { 3831 switch (Body[Pos + 1]) { 3832 // $$ => $ 3833 case '$': 3834 break; 3835 3836 // $n => number of arguments 3837 case 'n': 3838 PositionalParametersFound = true; 3839 break; 3840 3841 // $[0-9] => argument 3842 default: { 3843 PositionalParametersFound = true; 3844 break; 3845 } 3846 } 3847 Pos += 2; 3848 } else { 3849 unsigned I = Pos + 1; 3850 while (isIdentifierChar(Body[I]) && I + 1 != End) 3851 ++I; 3852 3853 const char *Begin = Body.data() + Pos + 1; 3854 StringRef Argument(Begin, I - (Pos + 1)); 3855 unsigned Index = 0; 3856 for (; Index < NParameters; ++Index) 3857 if (Parameters[Index].Name == Argument) 3858 break; 3859 3860 if (Index == NParameters) { 3861 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 3862 Pos += 3; 3863 else { 3864 Pos = I; 3865 } 3866 } else { 3867 NamedParametersFound = true; 3868 Pos += 1 + Argument.size(); 3869 } 3870 } 3871 // Update the scan point. 3872 Body = Body.substr(Pos); 3873 } 3874 3875 if (!NamedParametersFound && PositionalParametersFound) 3876 Warning(DirectiveLoc, "macro defined with named parameters which are not " 3877 "used in macro body, possible positional parameter " 3878 "found in body which will have no effect"); 3879 } 3880 3881 /// parseDirectiveExitMacro 3882 /// ::= .exitm 3883 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 3884 if (parseToken(AsmToken::EndOfStatement, 3885 "unexpected token in '" + Directive + "' directive")) 3886 return true; 3887 3888 if (!isInsideMacroInstantiation()) 3889 return TokError("unexpected '" + Directive + "' in file, " 3890 "no current macro definition"); 3891 3892 // Exit all conditionals that are active in the current macro. 3893 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 3894 TheCondState = TheCondStack.back(); 3895 TheCondStack.pop_back(); 3896 } 3897 3898 handleMacroExit(); 3899 return false; 3900 } 3901 3902 /// parseDirectiveEndMacro 3903 /// ::= .endm 3904 /// ::= .endmacro 3905 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 3906 if (getLexer().isNot(AsmToken::EndOfStatement)) 3907 return TokError("unexpected token in '" + Directive + "' directive"); 3908 3909 // If we are inside a macro instantiation, terminate the current 3910 // instantiation. 3911 if (isInsideMacroInstantiation()) { 3912 handleMacroExit(); 3913 return false; 3914 } 3915 3916 // Otherwise, this .endmacro is a stray entry in the file; well formed 3917 // .endmacro directives are handled during the macro definition parsing. 3918 return TokError("unexpected '" + Directive + "' in file, " 3919 "no current macro definition"); 3920 } 3921 3922 /// parseDirectivePurgeMacro 3923 /// ::= .purgem 3924 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 3925 StringRef Name; 3926 SMLoc Loc; 3927 if (getTokenLoc(Loc) || check(parseIdentifier(Name), Loc, 3928 "expected identifier in '.purgem' directive") || 3929 parseToken(AsmToken::EndOfStatement, 3930 "unexpected token in '.purgem' directive") || 3931 check(!lookupMacro(Name), DirectiveLoc, 3932 "macro '" + Name + "' is not defined")) 3933 return true; 3934 3935 undefineMacro(Name); 3936 return false; 3937 } 3938 3939 /// parseDirectiveBundleAlignMode 3940 /// ::= {.bundle_align_mode} expression 3941 bool AsmParser::parseDirectiveBundleAlignMode() { 3942 checkForValidSection(); 3943 3944 // Expect a single argument: an expression that evaluates to a constant 3945 // in the inclusive range 0-30. 3946 SMLoc ExprLoc = getLexer().getLoc(); 3947 int64_t AlignSizePow2; 3948 if (parseAbsoluteExpression(AlignSizePow2) || 3949 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 3950 "in '.bundle_align_mode' " 3951 "directive") || 3952 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 3953 "invalid bundle alignment size (expected between 0 and 30)")) 3954 return true; 3955 3956 // Because of AlignSizePow2's verified range we can safely truncate it to 3957 // unsigned. 3958 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 3959 return false; 3960 } 3961 3962 /// parseDirectiveBundleLock 3963 /// ::= {.bundle_lock} [align_to_end] 3964 bool AsmParser::parseDirectiveBundleLock() { 3965 checkForValidSection(); 3966 bool AlignToEnd = false; 3967 3968 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3969 StringRef Option; 3970 SMLoc Loc = getTok().getLoc(); 3971 const char *kInvalidOptionError = 3972 "invalid option for '.bundle_lock' directive"; 3973 3974 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 3975 check(Option != "align_to_end", Loc, kInvalidOptionError) || 3976 check(getTok().isNot(AsmToken::EndOfStatement), Loc, 3977 "unexpected token after '.bundle_lock' directive option")) 3978 return true; 3979 AlignToEnd = true; 3980 } 3981 3982 Lex(); 3983 3984 getStreamer().EmitBundleLock(AlignToEnd); 3985 return false; 3986 } 3987 3988 /// parseDirectiveBundleLock 3989 /// ::= {.bundle_lock} 3990 bool AsmParser::parseDirectiveBundleUnlock() { 3991 checkForValidSection(); 3992 3993 if (parseToken(AsmToken::EndOfStatement, 3994 "unexpected token in '.bundle_unlock' directive")) 3995 return true; 3996 3997 getStreamer().EmitBundleUnlock(); 3998 return false; 3999 } 4000 4001 /// parseDirectiveSpace 4002 /// ::= (.skip | .space) expression [ , expression ] 4003 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4004 checkForValidSection(); 4005 4006 SMLoc NumBytesLoc = Lexer.getLoc(); 4007 const MCExpr *NumBytes; 4008 if (parseExpression(NumBytes)) 4009 return true; 4010 4011 int64_t FillExpr = 0; 4012 if (getLexer().isNot(AsmToken::EndOfStatement)) { 4013 4014 if (parseToken(AsmToken::Comma, 4015 "unexpected token in '" + Twine(IDVal) + "' directive") || 4016 parseAbsoluteExpression(FillExpr)) 4017 return true; 4018 } 4019 4020 if (parseToken(AsmToken::EndOfStatement, 4021 "unexpected token in '" + Twine(IDVal) + "' directive")) 4022 return true; 4023 4024 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4025 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4026 4027 return false; 4028 } 4029 4030 /// parseDirectiveLEB128 4031 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4032 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4033 checkForValidSection(); 4034 const MCExpr *Value; 4035 4036 for (;;) { 4037 if (parseExpression(Value)) 4038 return true; 4039 4040 if (Signed) 4041 getStreamer().EmitSLEB128Value(Value); 4042 else 4043 getStreamer().EmitULEB128Value(Value); 4044 4045 if (getLexer().is(AsmToken::EndOfStatement)) 4046 break; 4047 4048 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 4049 return true; 4050 } 4051 Lex(); 4052 4053 return false; 4054 } 4055 4056 /// parseDirectiveSymbolAttribute 4057 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4058 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4059 if (getLexer().isNot(AsmToken::EndOfStatement)) { 4060 for (;;) { 4061 StringRef Name; 4062 SMLoc Loc = getTok().getLoc(); 4063 4064 if (parseIdentifier(Name)) 4065 return Error(Loc, "expected identifier in directive"); 4066 4067 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4068 4069 // Assembler local symbols don't make any sense here. Complain loudly. 4070 if (Sym->isTemporary()) 4071 return Error(Loc, "non-local symbol required in directive"); 4072 4073 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 4074 return Error(Loc, "unable to emit symbol attribute"); 4075 4076 if (getLexer().is(AsmToken::EndOfStatement)) 4077 break; 4078 4079 if (parseToken(AsmToken::Comma, "unexpected token in directive")) 4080 return true; 4081 } 4082 } 4083 4084 Lex(); 4085 return false; 4086 } 4087 4088 /// parseDirectiveComm 4089 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4090 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4091 checkForValidSection(); 4092 4093 SMLoc IDLoc = getLexer().getLoc(); 4094 StringRef Name; 4095 if (parseIdentifier(Name)) 4096 return TokError("expected identifier in directive"); 4097 4098 // Handle the identifier as the key symbol. 4099 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4100 4101 if (getLexer().isNot(AsmToken::Comma)) 4102 return TokError("unexpected token in directive"); 4103 Lex(); 4104 4105 int64_t Size; 4106 SMLoc SizeLoc = getLexer().getLoc(); 4107 if (parseAbsoluteExpression(Size)) 4108 return true; 4109 4110 int64_t Pow2Alignment = 0; 4111 SMLoc Pow2AlignmentLoc; 4112 if (getLexer().is(AsmToken::Comma)) { 4113 Lex(); 4114 Pow2AlignmentLoc = getLexer().getLoc(); 4115 if (parseAbsoluteExpression(Pow2Alignment)) 4116 return true; 4117 4118 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4119 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4120 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4121 4122 // If this target takes alignments in bytes (not log) validate and convert. 4123 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4124 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4125 if (!isPowerOf2_64(Pow2Alignment)) 4126 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4127 Pow2Alignment = Log2_64(Pow2Alignment); 4128 } 4129 } 4130 4131 if (getLexer().isNot(AsmToken::EndOfStatement)) 4132 return TokError("unexpected token in '.comm' or '.lcomm' directive"); 4133 4134 Lex(); 4135 4136 // NOTE: a size of zero for a .comm should create a undefined symbol 4137 // but a size of .lcomm creates a bss symbol of size zero. 4138 if (Size < 0) 4139 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4140 "be less than zero"); 4141 4142 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4143 // may internally end up wanting an alignment in bytes. 4144 // FIXME: Diagnose overflow. 4145 if (Pow2Alignment < 0) 4146 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4147 "alignment, can't be less than zero"); 4148 4149 if (!Sym->isUndefined()) 4150 return Error(IDLoc, "invalid symbol redefinition"); 4151 4152 // Create the Symbol as a common or local common with Size and Pow2Alignment 4153 if (IsLocal) { 4154 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4155 return false; 4156 } 4157 4158 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4159 return false; 4160 } 4161 4162 /// parseDirectiveAbort 4163 /// ::= .abort [... message ...] 4164 bool AsmParser::parseDirectiveAbort() { 4165 // FIXME: Use loc from directive. 4166 SMLoc Loc = getLexer().getLoc(); 4167 4168 StringRef Str = parseStringToEndOfStatement(); 4169 if (parseToken(AsmToken::EndOfStatement, 4170 "unexpected token in '.abort' directive")) 4171 return true; 4172 4173 if (Str.empty()) 4174 Error(Loc, ".abort detected. Assembly stopping."); 4175 else 4176 Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4177 // FIXME: Actually abort assembly here. 4178 4179 return false; 4180 } 4181 4182 /// parseDirectiveInclude 4183 /// ::= .include "filename" 4184 bool AsmParser::parseDirectiveInclude() { 4185 // Allow the strings to have escaped octal character sequence. 4186 std::string Filename; 4187 SMLoc IncludeLoc = getTok().getLoc(); 4188 4189 if (check(getTok().isNot(AsmToken::String), 4190 "expected string in '.include' directive") || 4191 parseEscapedString(Filename) || 4192 check(getTok().isNot(AsmToken::EndOfStatement), 4193 "unexpected token in '.include' directive") || 4194 // Attempt to switch the lexer to the included file before consuming the 4195 // end of statement to avoid losing it when we switch. 4196 check(enterIncludeFile(Filename), IncludeLoc, 4197 "Could not find include file '" + Filename + "'")) 4198 return true; 4199 4200 return false; 4201 } 4202 4203 /// parseDirectiveIncbin 4204 /// ::= .incbin "filename" 4205 bool AsmParser::parseDirectiveIncbin() { 4206 // Allow the strings to have escaped octal character sequence. 4207 std::string Filename; 4208 SMLoc IncbinLoc = getTok().getLoc(); 4209 if (check(getTok().isNot(AsmToken::String), 4210 "expected string in '.incbin' directive") || 4211 parseEscapedString(Filename) || 4212 parseToken(AsmToken::EndOfStatement, 4213 "unexpected token in '.incbin' directive") || 4214 // Attempt to process the included file. 4215 check(processIncbinFile(Filename), IncbinLoc, 4216 "Could not find incbin file '" + Filename + "'")) 4217 return true; 4218 return false; 4219 } 4220 4221 /// parseDirectiveIf 4222 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 4223 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 4224 TheCondStack.push_back(TheCondState); 4225 TheCondState.TheCond = AsmCond::IfCond; 4226 if (TheCondState.Ignore) { 4227 eatToEndOfStatement(); 4228 } else { 4229 int64_t ExprValue; 4230 if (parseAbsoluteExpression(ExprValue) || 4231 parseToken(AsmToken::EndOfStatement, 4232 "unexpected token in '.if' directive")) 4233 return true; 4234 4235 switch (DirKind) { 4236 default: 4237 llvm_unreachable("unsupported directive"); 4238 case DK_IF: 4239 case DK_IFNE: 4240 break; 4241 case DK_IFEQ: 4242 ExprValue = ExprValue == 0; 4243 break; 4244 case DK_IFGE: 4245 ExprValue = ExprValue >= 0; 4246 break; 4247 case DK_IFGT: 4248 ExprValue = ExprValue > 0; 4249 break; 4250 case DK_IFLE: 4251 ExprValue = ExprValue <= 0; 4252 break; 4253 case DK_IFLT: 4254 ExprValue = ExprValue < 0; 4255 break; 4256 } 4257 4258 TheCondState.CondMet = ExprValue; 4259 TheCondState.Ignore = !TheCondState.CondMet; 4260 } 4261 4262 return false; 4263 } 4264 4265 /// parseDirectiveIfb 4266 /// ::= .ifb string 4267 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4268 TheCondStack.push_back(TheCondState); 4269 TheCondState.TheCond = AsmCond::IfCond; 4270 4271 if (TheCondState.Ignore) { 4272 eatToEndOfStatement(); 4273 } else { 4274 StringRef Str = parseStringToEndOfStatement(); 4275 4276 if (parseToken(AsmToken::EndOfStatement, 4277 "unexpected token in '.ifb' directive")) 4278 return true; 4279 4280 TheCondState.CondMet = ExpectBlank == Str.empty(); 4281 TheCondState.Ignore = !TheCondState.CondMet; 4282 } 4283 4284 return false; 4285 } 4286 4287 /// parseDirectiveIfc 4288 /// ::= .ifc string1, string2 4289 /// ::= .ifnc string1, string2 4290 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 4291 TheCondStack.push_back(TheCondState); 4292 TheCondState.TheCond = AsmCond::IfCond; 4293 4294 if (TheCondState.Ignore) { 4295 eatToEndOfStatement(); 4296 } else { 4297 StringRef Str1 = parseStringToComma(); 4298 4299 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 4300 return true; 4301 4302 StringRef Str2 = parseStringToEndOfStatement(); 4303 4304 if (parseToken(AsmToken::EndOfStatement, 4305 "unexpected token in '.ifc' directive")) 4306 return true; 4307 4308 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 4309 TheCondState.Ignore = !TheCondState.CondMet; 4310 } 4311 4312 return false; 4313 } 4314 4315 /// parseDirectiveIfeqs 4316 /// ::= .ifeqs string1, string2 4317 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 4318 if (Lexer.isNot(AsmToken::String)) { 4319 if (ExpectEqual) 4320 TokError("expected string parameter for '.ifeqs' directive"); 4321 else 4322 TokError("expected string parameter for '.ifnes' directive"); 4323 eatToEndOfStatement(); 4324 return true; 4325 } 4326 4327 StringRef String1 = getTok().getStringContents(); 4328 Lex(); 4329 4330 if (Lexer.isNot(AsmToken::Comma)) { 4331 if (ExpectEqual) 4332 TokError("expected comma after first string for '.ifeqs' directive"); 4333 else 4334 TokError("expected comma after first string for '.ifnes' directive"); 4335 eatToEndOfStatement(); 4336 return true; 4337 } 4338 4339 Lex(); 4340 4341 if (Lexer.isNot(AsmToken::String)) { 4342 if (ExpectEqual) 4343 TokError("expected string parameter for '.ifeqs' directive"); 4344 else 4345 TokError("expected string parameter for '.ifnes' directive"); 4346 eatToEndOfStatement(); 4347 return true; 4348 } 4349 4350 StringRef String2 = getTok().getStringContents(); 4351 Lex(); 4352 4353 TheCondStack.push_back(TheCondState); 4354 TheCondState.TheCond = AsmCond::IfCond; 4355 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4356 TheCondState.Ignore = !TheCondState.CondMet; 4357 4358 return false; 4359 } 4360 4361 /// parseDirectiveIfdef 4362 /// ::= .ifdef symbol 4363 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 4364 StringRef Name; 4365 TheCondStack.push_back(TheCondState); 4366 TheCondState.TheCond = AsmCond::IfCond; 4367 4368 if (TheCondState.Ignore) { 4369 eatToEndOfStatement(); 4370 } else { 4371 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 4372 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 4373 return true; 4374 4375 MCSymbol *Sym = getContext().lookupSymbol(Name); 4376 4377 if (expect_defined) 4378 TheCondState.CondMet = (Sym && !Sym->isUndefined()); 4379 else 4380 TheCondState.CondMet = (!Sym || Sym->isUndefined()); 4381 TheCondState.Ignore = !TheCondState.CondMet; 4382 } 4383 4384 return false; 4385 } 4386 4387 /// parseDirectiveElseIf 4388 /// ::= .elseif expression 4389 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 4390 if (TheCondState.TheCond != AsmCond::IfCond && 4391 TheCondState.TheCond != AsmCond::ElseIfCond) 4392 Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " 4393 " an .elseif"); 4394 TheCondState.TheCond = AsmCond::ElseIfCond; 4395 4396 bool LastIgnoreState = false; 4397 if (!TheCondStack.empty()) 4398 LastIgnoreState = TheCondStack.back().Ignore; 4399 if (LastIgnoreState || TheCondState.CondMet) { 4400 TheCondState.Ignore = true; 4401 eatToEndOfStatement(); 4402 } else { 4403 int64_t ExprValue; 4404 if (parseAbsoluteExpression(ExprValue)) 4405 return true; 4406 4407 if (getLexer().isNot(AsmToken::EndOfStatement)) 4408 return TokError("unexpected token in '.elseif' directive"); 4409 4410 Lex(); 4411 TheCondState.CondMet = ExprValue; 4412 TheCondState.Ignore = !TheCondState.CondMet; 4413 } 4414 4415 return false; 4416 } 4417 4418 /// parseDirectiveElse 4419 /// ::= .else 4420 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4421 if (parseToken(AsmToken::EndOfStatement, 4422 "unexpected token in '.else' directive")) 4423 return true; 4424 4425 if (TheCondState.TheCond != AsmCond::IfCond && 4426 TheCondState.TheCond != AsmCond::ElseIfCond) 4427 Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " 4428 ".elseif"); 4429 TheCondState.TheCond = AsmCond::ElseCond; 4430 bool LastIgnoreState = false; 4431 if (!TheCondStack.empty()) 4432 LastIgnoreState = TheCondStack.back().Ignore; 4433 if (LastIgnoreState || TheCondState.CondMet) 4434 TheCondState.Ignore = true; 4435 else 4436 TheCondState.Ignore = false; 4437 4438 return false; 4439 } 4440 4441 /// parseDirectiveEnd 4442 /// ::= .end 4443 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4444 if (parseToken(AsmToken::EndOfStatement, 4445 "unexpected token in '.end' directive")) 4446 return true; 4447 4448 while (Lexer.isNot(AsmToken::Eof)) 4449 Lex(); 4450 4451 return false; 4452 } 4453 4454 /// parseDirectiveError 4455 /// ::= .err 4456 /// ::= .error [string] 4457 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 4458 if (!TheCondStack.empty()) { 4459 if (TheCondStack.back().Ignore) { 4460 eatToEndOfStatement(); 4461 return false; 4462 } 4463 } 4464 4465 if (!WithMessage) 4466 return Error(L, ".err encountered"); 4467 4468 StringRef Message = ".error directive invoked in source file"; 4469 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4470 if (Lexer.isNot(AsmToken::String)) { 4471 TokError(".error argument must be a string"); 4472 eatToEndOfStatement(); 4473 return true; 4474 } 4475 4476 Message = getTok().getStringContents(); 4477 Lex(); 4478 } 4479 4480 Error(L, Message); 4481 return true; 4482 } 4483 4484 /// parseDirectiveWarning 4485 /// ::= .warning [string] 4486 bool AsmParser::parseDirectiveWarning(SMLoc L) { 4487 if (!TheCondStack.empty()) { 4488 if (TheCondStack.back().Ignore) { 4489 eatToEndOfStatement(); 4490 return false; 4491 } 4492 } 4493 4494 StringRef Message = ".warning directive invoked in source file"; 4495 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4496 if (Lexer.isNot(AsmToken::String)) { 4497 TokError(".warning argument must be a string"); 4498 eatToEndOfStatement(); 4499 return true; 4500 } 4501 4502 Message = getTok().getStringContents(); 4503 Lex(); 4504 } 4505 4506 Warning(L, Message); 4507 return false; 4508 } 4509 4510 /// parseDirectiveEndIf 4511 /// ::= .endif 4512 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 4513 if (parseToken(AsmToken::EndOfStatement, 4514 "unexpected token in '.endif' directive")) 4515 return true; 4516 4517 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 4518 Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " 4519 ".else"); 4520 if (!TheCondStack.empty()) { 4521 TheCondState = TheCondStack.back(); 4522 TheCondStack.pop_back(); 4523 } 4524 4525 return false; 4526 } 4527 4528 void AsmParser::initializeDirectiveKindMap() { 4529 DirectiveKindMap[".set"] = DK_SET; 4530 DirectiveKindMap[".equ"] = DK_EQU; 4531 DirectiveKindMap[".equiv"] = DK_EQUIV; 4532 DirectiveKindMap[".ascii"] = DK_ASCII; 4533 DirectiveKindMap[".asciz"] = DK_ASCIZ; 4534 DirectiveKindMap[".string"] = DK_STRING; 4535 DirectiveKindMap[".byte"] = DK_BYTE; 4536 DirectiveKindMap[".short"] = DK_SHORT; 4537 DirectiveKindMap[".value"] = DK_VALUE; 4538 DirectiveKindMap[".2byte"] = DK_2BYTE; 4539 DirectiveKindMap[".long"] = DK_LONG; 4540 DirectiveKindMap[".int"] = DK_INT; 4541 DirectiveKindMap[".4byte"] = DK_4BYTE; 4542 DirectiveKindMap[".quad"] = DK_QUAD; 4543 DirectiveKindMap[".8byte"] = DK_8BYTE; 4544 DirectiveKindMap[".octa"] = DK_OCTA; 4545 DirectiveKindMap[".single"] = DK_SINGLE; 4546 DirectiveKindMap[".float"] = DK_FLOAT; 4547 DirectiveKindMap[".double"] = DK_DOUBLE; 4548 DirectiveKindMap[".align"] = DK_ALIGN; 4549 DirectiveKindMap[".align32"] = DK_ALIGN32; 4550 DirectiveKindMap[".balign"] = DK_BALIGN; 4551 DirectiveKindMap[".balignw"] = DK_BALIGNW; 4552 DirectiveKindMap[".balignl"] = DK_BALIGNL; 4553 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 4554 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 4555 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 4556 DirectiveKindMap[".org"] = DK_ORG; 4557 DirectiveKindMap[".fill"] = DK_FILL; 4558 DirectiveKindMap[".zero"] = DK_ZERO; 4559 DirectiveKindMap[".extern"] = DK_EXTERN; 4560 DirectiveKindMap[".globl"] = DK_GLOBL; 4561 DirectiveKindMap[".global"] = DK_GLOBAL; 4562 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 4563 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 4564 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 4565 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 4566 DirectiveKindMap[".reference"] = DK_REFERENCE; 4567 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 4568 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 4569 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 4570 DirectiveKindMap[".comm"] = DK_COMM; 4571 DirectiveKindMap[".common"] = DK_COMMON; 4572 DirectiveKindMap[".lcomm"] = DK_LCOMM; 4573 DirectiveKindMap[".abort"] = DK_ABORT; 4574 DirectiveKindMap[".include"] = DK_INCLUDE; 4575 DirectiveKindMap[".incbin"] = DK_INCBIN; 4576 DirectiveKindMap[".code16"] = DK_CODE16; 4577 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 4578 DirectiveKindMap[".rept"] = DK_REPT; 4579 DirectiveKindMap[".rep"] = DK_REPT; 4580 DirectiveKindMap[".irp"] = DK_IRP; 4581 DirectiveKindMap[".irpc"] = DK_IRPC; 4582 DirectiveKindMap[".endr"] = DK_ENDR; 4583 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 4584 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 4585 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 4586 DirectiveKindMap[".if"] = DK_IF; 4587 DirectiveKindMap[".ifeq"] = DK_IFEQ; 4588 DirectiveKindMap[".ifge"] = DK_IFGE; 4589 DirectiveKindMap[".ifgt"] = DK_IFGT; 4590 DirectiveKindMap[".ifle"] = DK_IFLE; 4591 DirectiveKindMap[".iflt"] = DK_IFLT; 4592 DirectiveKindMap[".ifne"] = DK_IFNE; 4593 DirectiveKindMap[".ifb"] = DK_IFB; 4594 DirectiveKindMap[".ifnb"] = DK_IFNB; 4595 DirectiveKindMap[".ifc"] = DK_IFC; 4596 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 4597 DirectiveKindMap[".ifnc"] = DK_IFNC; 4598 DirectiveKindMap[".ifnes"] = DK_IFNES; 4599 DirectiveKindMap[".ifdef"] = DK_IFDEF; 4600 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 4601 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 4602 DirectiveKindMap[".elseif"] = DK_ELSEIF; 4603 DirectiveKindMap[".else"] = DK_ELSE; 4604 DirectiveKindMap[".end"] = DK_END; 4605 DirectiveKindMap[".endif"] = DK_ENDIF; 4606 DirectiveKindMap[".skip"] = DK_SKIP; 4607 DirectiveKindMap[".space"] = DK_SPACE; 4608 DirectiveKindMap[".file"] = DK_FILE; 4609 DirectiveKindMap[".line"] = DK_LINE; 4610 DirectiveKindMap[".loc"] = DK_LOC; 4611 DirectiveKindMap[".stabs"] = DK_STABS; 4612 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 4613 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 4614 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 4615 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 4616 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 4617 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 4618 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 4619 DirectiveKindMap[".sleb128"] = DK_SLEB128; 4620 DirectiveKindMap[".uleb128"] = DK_ULEB128; 4621 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 4622 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 4623 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 4624 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 4625 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 4626 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 4627 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 4628 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 4629 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 4630 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 4631 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 4632 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 4633 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 4634 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 4635 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 4636 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 4637 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 4638 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 4639 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 4640 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 4641 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 4642 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 4643 DirectiveKindMap[".macro"] = DK_MACRO; 4644 DirectiveKindMap[".exitm"] = DK_EXITM; 4645 DirectiveKindMap[".endm"] = DK_ENDM; 4646 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 4647 DirectiveKindMap[".purgem"] = DK_PURGEM; 4648 DirectiveKindMap[".err"] = DK_ERR; 4649 DirectiveKindMap[".error"] = DK_ERROR; 4650 DirectiveKindMap[".warning"] = DK_WARNING; 4651 DirectiveKindMap[".reloc"] = DK_RELOC; 4652 } 4653 4654 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 4655 AsmToken EndToken, StartToken = getTok(); 4656 4657 unsigned NestLevel = 0; 4658 for (;;) { 4659 // Check whether we have reached the end of the file. 4660 if (getLexer().is(AsmToken::Eof)) { 4661 Error(DirectiveLoc, "no matching '.endr' in definition"); 4662 return nullptr; 4663 } 4664 4665 if (Lexer.is(AsmToken::Identifier) && 4666 (getTok().getIdentifier() == ".rept" || 4667 getTok().getIdentifier() == ".irp" || 4668 getTok().getIdentifier() == ".irpc")) { 4669 ++NestLevel; 4670 } 4671 4672 // Otherwise, check whether we have reached the .endr. 4673 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 4674 if (NestLevel == 0) { 4675 EndToken = getTok(); 4676 Lex(); 4677 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4678 TokError("unexpected token in '.endr' directive"); 4679 return nullptr; 4680 } 4681 break; 4682 } 4683 --NestLevel; 4684 } 4685 4686 // Otherwise, scan till the end of the statement. 4687 eatToEndOfStatement(); 4688 } 4689 4690 const char *BodyStart = StartToken.getLoc().getPointer(); 4691 const char *BodyEnd = EndToken.getLoc().getPointer(); 4692 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4693 4694 // We Are Anonymous. 4695 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 4696 return &MacroLikeBodies.back(); 4697 } 4698 4699 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 4700 raw_svector_ostream &OS) { 4701 OS << ".endr\n"; 4702 4703 std::unique_ptr<MemoryBuffer> Instantiation = 4704 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 4705 4706 // Create the macro instantiation object and add to the current macro 4707 // instantiation stack. 4708 MacroInstantiation *MI = new MacroInstantiation( 4709 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 4710 ActiveMacros.push_back(MI); 4711 4712 // Jump to the macro instantiation and prime the lexer. 4713 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 4714 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 4715 Lex(); 4716 } 4717 4718 /// parseDirectiveRept 4719 /// ::= .rep | .rept count 4720 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 4721 const MCExpr *CountExpr; 4722 SMLoc CountLoc = getTok().getLoc(); 4723 if (parseExpression(CountExpr)) 4724 return true; 4725 4726 int64_t Count; 4727 if (!CountExpr->evaluateAsAbsolute(Count)) { 4728 eatToEndOfStatement(); 4729 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 4730 } 4731 4732 if (check(Count < 0, CountLoc, "Count is negative") || 4733 parseToken(AsmToken::EndOfStatement, 4734 "unexpected token in '" + Dir + "' directive")) 4735 return true; 4736 4737 // Lex the rept definition. 4738 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4739 if (!M) 4740 return true; 4741 4742 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4743 // to hold the macro body with substitutions. 4744 SmallString<256> Buf; 4745 raw_svector_ostream OS(Buf); 4746 while (Count--) { 4747 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 4748 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 4749 return true; 4750 } 4751 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4752 4753 return false; 4754 } 4755 4756 /// parseDirectiveIrp 4757 /// ::= .irp symbol,values 4758 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 4759 MCAsmMacroParameter Parameter; 4760 MCAsmMacroArguments A; 4761 if (check(parseIdentifier(Parameter.Name), 4762 "expected identifier in '.irp' directive") || 4763 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 4764 parseMacroArguments(nullptr, A) || 4765 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 4766 return true; 4767 4768 // Lex the irp definition. 4769 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4770 if (!M) 4771 return true; 4772 4773 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4774 // to hold the macro body with substitutions. 4775 SmallString<256> Buf; 4776 raw_svector_ostream OS(Buf); 4777 4778 for (const MCAsmMacroArgument &Arg : A) { 4779 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 4780 // This is undocumented, but GAS seems to support it. 4781 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 4782 return true; 4783 } 4784 4785 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4786 4787 return false; 4788 } 4789 4790 /// parseDirectiveIrpc 4791 /// ::= .irpc symbol,values 4792 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 4793 MCAsmMacroParameter Parameter; 4794 MCAsmMacroArguments A; 4795 4796 if (check(parseIdentifier(Parameter.Name), 4797 "expected identifier in '.irpc' directive") || 4798 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 4799 parseMacroArguments(nullptr, A)) 4800 return true; 4801 4802 if (A.size() != 1 || A.front().size() != 1) 4803 return TokError("unexpected token in '.irpc' directive"); 4804 4805 // Eat the end of statement. 4806 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 4807 return true; 4808 4809 // Lex the irpc definition. 4810 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4811 if (!M) 4812 return true; 4813 4814 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4815 // to hold the macro body with substitutions. 4816 SmallString<256> Buf; 4817 raw_svector_ostream OS(Buf); 4818 4819 StringRef Values = A.front().front().getString(); 4820 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 4821 MCAsmMacroArgument Arg; 4822 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 4823 4824 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 4825 // This is undocumented, but GAS seems to support it. 4826 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 4827 return true; 4828 } 4829 4830 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4831 4832 return false; 4833 } 4834 4835 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 4836 if (ActiveMacros.empty()) 4837 return TokError("unmatched '.endr' directive"); 4838 4839 // The only .repl that should get here are the ones created by 4840 // instantiateMacroLikeBody. 4841 assert(getLexer().is(AsmToken::EndOfStatement)); 4842 4843 handleMacroExit(); 4844 return false; 4845 } 4846 4847 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 4848 size_t Len) { 4849 const MCExpr *Value; 4850 SMLoc ExprLoc = getLexer().getLoc(); 4851 if (parseExpression(Value)) 4852 return true; 4853 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4854 if (!MCE) 4855 return Error(ExprLoc, "unexpected expression in _emit"); 4856 uint64_t IntValue = MCE->getValue(); 4857 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 4858 return Error(ExprLoc, "literal value out of range for directive"); 4859 4860 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 4861 return false; 4862 } 4863 4864 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 4865 const MCExpr *Value; 4866 SMLoc ExprLoc = getLexer().getLoc(); 4867 if (parseExpression(Value)) 4868 return true; 4869 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4870 if (!MCE) 4871 return Error(ExprLoc, "unexpected expression in align"); 4872 uint64_t IntValue = MCE->getValue(); 4873 if (!isPowerOf2_64(IntValue)) 4874 return Error(ExprLoc, "literal value not a power of two greater then zero"); 4875 4876 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 4877 return false; 4878 } 4879 4880 // We are comparing pointers, but the pointers are relative to a single string. 4881 // Thus, this should always be deterministic. 4882 static int rewritesSort(const AsmRewrite *AsmRewriteA, 4883 const AsmRewrite *AsmRewriteB) { 4884 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 4885 return -1; 4886 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 4887 return 1; 4888 4889 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 4890 // rewrite to the same location. Make sure the SizeDirective rewrite is 4891 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 4892 // ensures the sort algorithm is stable. 4893 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 4894 AsmRewritePrecedence[AsmRewriteB->Kind]) 4895 return -1; 4896 4897 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 4898 AsmRewritePrecedence[AsmRewriteB->Kind]) 4899 return 1; 4900 llvm_unreachable("Unstable rewrite sort."); 4901 } 4902 4903 bool AsmParser::parseMSInlineAsm( 4904 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 4905 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool> > &OpDecls, 4906 SmallVectorImpl<std::string> &Constraints, 4907 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 4908 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 4909 SmallVector<void *, 4> InputDecls; 4910 SmallVector<void *, 4> OutputDecls; 4911 SmallVector<bool, 4> InputDeclsAddressOf; 4912 SmallVector<bool, 4> OutputDeclsAddressOf; 4913 SmallVector<std::string, 4> InputConstraints; 4914 SmallVector<std::string, 4> OutputConstraints; 4915 SmallVector<unsigned, 4> ClobberRegs; 4916 4917 SmallVector<AsmRewrite, 4> AsmStrRewrites; 4918 4919 // Prime the lexer. 4920 Lex(); 4921 4922 // While we have input, parse each statement. 4923 unsigned InputIdx = 0; 4924 unsigned OutputIdx = 0; 4925 while (getLexer().isNot(AsmToken::Eof)) { 4926 // Parse curly braces marking block start/end 4927 if (parseCurlyBlockScope(AsmStrRewrites)) 4928 continue; 4929 4930 ParseStatementInfo Info(&AsmStrRewrites); 4931 if (parseStatement(Info, &SI)) 4932 return true; 4933 4934 if (Info.ParseError) 4935 return true; 4936 4937 if (Info.Opcode == ~0U) 4938 continue; 4939 4940 const MCInstrDesc &Desc = MII->get(Info.Opcode); 4941 4942 // Build the list of clobbers, outputs and inputs. 4943 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 4944 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 4945 4946 // Immediate. 4947 if (Operand.isImm()) 4948 continue; 4949 4950 // Register operand. 4951 if (Operand.isReg() && !Operand.needAddressOf() && 4952 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 4953 unsigned NumDefs = Desc.getNumDefs(); 4954 // Clobber. 4955 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 4956 ClobberRegs.push_back(Operand.getReg()); 4957 continue; 4958 } 4959 4960 // Expr/Input or Output. 4961 StringRef SymName = Operand.getSymName(); 4962 if (SymName.empty()) 4963 continue; 4964 4965 void *OpDecl = Operand.getOpDecl(); 4966 if (!OpDecl) 4967 continue; 4968 4969 bool isOutput = (i == 1) && Desc.mayStore(); 4970 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 4971 if (isOutput) { 4972 ++InputIdx; 4973 OutputDecls.push_back(OpDecl); 4974 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 4975 OutputConstraints.push_back(("=" + Operand.getConstraint()).str()); 4976 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 4977 } else { 4978 InputDecls.push_back(OpDecl); 4979 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 4980 InputConstraints.push_back(Operand.getConstraint().str()); 4981 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 4982 } 4983 } 4984 4985 // Consider implicit defs to be clobbers. Think of cpuid and push. 4986 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 4987 Desc.getNumImplicitDefs()); 4988 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 4989 } 4990 4991 // Set the number of Outputs and Inputs. 4992 NumOutputs = OutputDecls.size(); 4993 NumInputs = InputDecls.size(); 4994 4995 // Set the unique clobbers. 4996 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 4997 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 4998 ClobberRegs.end()); 4999 Clobbers.assign(ClobberRegs.size(), std::string()); 5000 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5001 raw_string_ostream OS(Clobbers[I]); 5002 IP->printRegName(OS, ClobberRegs[I]); 5003 } 5004 5005 // Merge the various outputs and inputs. Output are expected first. 5006 if (NumOutputs || NumInputs) { 5007 unsigned NumExprs = NumOutputs + NumInputs; 5008 OpDecls.resize(NumExprs); 5009 Constraints.resize(NumExprs); 5010 for (unsigned i = 0; i < NumOutputs; ++i) { 5011 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5012 Constraints[i] = OutputConstraints[i]; 5013 } 5014 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5015 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5016 Constraints[j] = InputConstraints[i]; 5017 } 5018 } 5019 5020 // Build the IR assembly string. 5021 std::string AsmStringIR; 5022 raw_string_ostream OS(AsmStringIR); 5023 StringRef ASMString = 5024 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5025 const char *AsmStart = ASMString.begin(); 5026 const char *AsmEnd = ASMString.end(); 5027 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5028 for (const AsmRewrite &AR : AsmStrRewrites) { 5029 AsmRewriteKind Kind = AR.Kind; 5030 if (Kind == AOK_Delete) 5031 continue; 5032 5033 const char *Loc = AR.Loc.getPointer(); 5034 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5035 5036 // Emit everything up to the immediate/expression. 5037 if (unsigned Len = Loc - AsmStart) 5038 OS << StringRef(AsmStart, Len); 5039 5040 // Skip the original expression. 5041 if (Kind == AOK_Skip) { 5042 AsmStart = Loc + AR.Len; 5043 continue; 5044 } 5045 5046 unsigned AdditionalSkip = 0; 5047 // Rewrite expressions in $N notation. 5048 switch (Kind) { 5049 default: 5050 break; 5051 case AOK_Imm: 5052 OS << "$$" << AR.Val; 5053 break; 5054 case AOK_ImmPrefix: 5055 OS << "$$"; 5056 break; 5057 case AOK_Label: 5058 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5059 break; 5060 case AOK_Input: 5061 OS << '$' << InputIdx++; 5062 break; 5063 case AOK_Output: 5064 OS << '$' << OutputIdx++; 5065 break; 5066 case AOK_SizeDirective: 5067 switch (AR.Val) { 5068 default: break; 5069 case 8: OS << "byte ptr "; break; 5070 case 16: OS << "word ptr "; break; 5071 case 32: OS << "dword ptr "; break; 5072 case 64: OS << "qword ptr "; break; 5073 case 80: OS << "xword ptr "; break; 5074 case 128: OS << "xmmword ptr "; break; 5075 case 256: OS << "ymmword ptr "; break; 5076 } 5077 break; 5078 case AOK_Emit: 5079 OS << ".byte"; 5080 break; 5081 case AOK_Align: { 5082 // MS alignment directives are measured in bytes. If the native assembler 5083 // measures alignment in bytes, we can pass it straight through. 5084 OS << ".align"; 5085 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5086 break; 5087 5088 // Alignment is in log2 form, so print that instead and skip the original 5089 // immediate. 5090 unsigned Val = AR.Val; 5091 OS << ' ' << Val; 5092 assert(Val < 10 && "Expected alignment less then 2^10."); 5093 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5094 break; 5095 } 5096 case AOK_EVEN: 5097 OS << ".even"; 5098 break; 5099 case AOK_DotOperator: 5100 // Insert the dot if the user omitted it. 5101 OS.flush(); 5102 if (AsmStringIR.back() != '.') 5103 OS << '.'; 5104 OS << AR.Val; 5105 break; 5106 case AOK_EndOfStatement: 5107 OS << "\n\t"; 5108 break; 5109 } 5110 5111 // Skip the original expression. 5112 AsmStart = Loc + AR.Len + AdditionalSkip; 5113 } 5114 5115 // Emit the remainder of the asm string. 5116 if (AsmStart != AsmEnd) 5117 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5118 5119 AsmString = OS.str(); 5120 return false; 5121 } 5122 5123 namespace llvm { 5124 namespace MCParserUtils { 5125 5126 /// Returns whether the given symbol is used anywhere in the given expression, 5127 /// or subexpressions. 5128 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 5129 switch (Value->getKind()) { 5130 case MCExpr::Binary: { 5131 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 5132 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 5133 isSymbolUsedInExpression(Sym, BE->getRHS()); 5134 } 5135 case MCExpr::Target: 5136 case MCExpr::Constant: 5137 return false; 5138 case MCExpr::SymbolRef: { 5139 const MCSymbol &S = 5140 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 5141 if (S.isVariable()) 5142 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 5143 return &S == Sym; 5144 } 5145 case MCExpr::Unary: 5146 return isSymbolUsedInExpression( 5147 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 5148 } 5149 5150 llvm_unreachable("Unknown expr kind!"); 5151 } 5152 5153 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 5154 MCAsmParser &Parser, MCSymbol *&Sym, 5155 const MCExpr *&Value) { 5156 5157 // FIXME: Use better location, we should use proper tokens. 5158 SMLoc EqualLoc = Parser.getTok().getLoc(); 5159 5160 if (Parser.parseExpression(Value)) { 5161 Parser.TokError("missing expression"); 5162 Parser.eatToEndOfStatement(); 5163 return true; 5164 } 5165 5166 // Note: we don't count b as used in "a = b". This is to allow 5167 // a = b 5168 // b = c 5169 5170 if (Parser.getTok().isNot(AsmToken::EndOfStatement)) 5171 return Parser.TokError("unexpected token in assignment"); 5172 5173 // Eat the end of statement marker. 5174 Parser.Lex(); 5175 5176 // Validate that the LHS is allowed to be a variable (either it has not been 5177 // used as a symbol, or it is an absolute symbol). 5178 Sym = Parser.getContext().lookupSymbol(Name); 5179 if (Sym) { 5180 // Diagnose assignment to a label. 5181 // 5182 // FIXME: Diagnostics. Note the location of the definition as a label. 5183 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 5184 if (isSymbolUsedInExpression(Sym, Value)) 5185 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 5186 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 5187 !Sym->isVariable()) 5188 ; // Allow redefinitions of undefined symbols only used in directives. 5189 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 5190 ; // Allow redefinitions of variables that haven't yet been used. 5191 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 5192 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 5193 else if (!Sym->isVariable()) 5194 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 5195 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 5196 return Parser.Error(EqualLoc, 5197 "invalid reassignment of non-absolute variable '" + 5198 Name + "'"); 5199 } else if (Name == ".") { 5200 Parser.getStreamer().emitValueToOffset(Value, 0); 5201 return false; 5202 } else 5203 Sym = Parser.getContext().getOrCreateSymbol(Name); 5204 5205 Sym->setRedefinable(allow_redef); 5206 5207 return false; 5208 } 5209 5210 } // namespace MCParserUtils 5211 } // namespace llvm 5212 5213 /// \brief Create an MCAsmParser instance. 5214 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 5215 MCStreamer &Out, const MCAsmInfo &MAI) { 5216 return new AsmParser(SM, C, Out, MAI); 5217 } 5218