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/MCParsedAsmOperand.h" 30 #include "llvm/MC/MCRegisterInfo.h" 31 #include "llvm/MC/MCSectionMachO.h" 32 #include "llvm/MC/MCStreamer.h" 33 #include "llvm/MC/MCSymbol.h" 34 #include "llvm/MC/MCTargetAsmParser.h" 35 #include "llvm/Support/CommandLine.h" 36 #include "llvm/Support/ErrorHandling.h" 37 #include "llvm/Support/MathExtras.h" 38 #include "llvm/Support/MemoryBuffer.h" 39 #include "llvm/Support/SourceMgr.h" 40 #include "llvm/Support/raw_ostream.h" 41 #include <cctype> 42 #include <deque> 43 #include <set> 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 &) LLVM_DELETED_FUNCTION; 115 void operator=(const AsmParser &) LLVM_DELETED_FUNCTION; 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 /// Flag tracking whether any errors have been encountered. 151 unsigned HadError : 1; 152 153 /// The values from the last parsed cpp hash file line comment if any. 154 StringRef CppHashFilename; 155 int64_t CppHashLineNumber; 156 SMLoc CppHashLoc; 157 unsigned CppHashBuf; 158 /// When generating dwarf for assembly source files we need to calculate the 159 /// logical line number based on the last parsed cpp hash file line comment 160 /// and current line. Since this is slow and messes up the SourceMgr's 161 /// cache we save the last info we queried with SrcMgr.FindLineNumber(). 162 SMLoc LastQueryIDLoc; 163 unsigned LastQueryBuffer; 164 unsigned LastQueryLine; 165 166 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 167 unsigned AssemblerDialect; 168 169 /// \brief is Darwin compatibility enabled? 170 bool IsDarwin; 171 172 /// \brief Are we parsing ms-style inline assembly? 173 bool ParsingInlineAsm; 174 175 public: 176 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 177 const MCAsmInfo &MAI); 178 virtual ~AsmParser(); 179 180 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 181 182 void addDirectiveHandler(StringRef Directive, 183 ExtensionDirectiveHandler Handler) override { 184 ExtensionDirectiveMap[Directive] = Handler; 185 } 186 187 public: 188 /// @name MCAsmParser Interface 189 /// { 190 191 SourceMgr &getSourceManager() override { return SrcMgr; } 192 MCAsmLexer &getLexer() override { return Lexer; } 193 MCContext &getContext() override { return Ctx; } 194 MCStreamer &getStreamer() override { return Out; } 195 unsigned getAssemblerDialect() override { 196 if (AssemblerDialect == ~0U) 197 return MAI.getAssemblerDialect(); 198 else 199 return AssemblerDialect; 200 } 201 void setAssemblerDialect(unsigned i) override { 202 AssemblerDialect = i; 203 } 204 205 void Note(SMLoc L, const Twine &Msg, 206 ArrayRef<SMRange> Ranges = None) override; 207 bool Warning(SMLoc L, const Twine &Msg, 208 ArrayRef<SMRange> Ranges = None) override; 209 bool Error(SMLoc L, const Twine &Msg, 210 ArrayRef<SMRange> Ranges = None) override; 211 212 const AsmToken &Lex() override; 213 214 void setParsingInlineAsm(bool V) override { ParsingInlineAsm = V; } 215 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 216 217 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 218 unsigned &NumOutputs, unsigned &NumInputs, 219 SmallVectorImpl<std::pair<void *,bool> > &OpDecls, 220 SmallVectorImpl<std::string> &Constraints, 221 SmallVectorImpl<std::string> &Clobbers, 222 const MCInstrInfo *MII, const MCInstPrinter *IP, 223 MCAsmParserSemaCallback &SI) override; 224 225 bool parseExpression(const MCExpr *&Res); 226 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 227 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 228 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 229 bool parseAbsoluteExpression(int64_t &Res) override; 230 231 /// \brief Parse an identifier or string (as a quoted identifier) 232 /// and set \p Res to the identifier contents. 233 bool parseIdentifier(StringRef &Res) override; 234 void eatToEndOfStatement() override; 235 236 void checkForValidSection() override; 237 /// } 238 239 private: 240 241 bool parseStatement(ParseStatementInfo &Info, 242 MCAsmParserSemaCallback *SI); 243 void eatToEndOfLine(); 244 bool parseCppHashLineFilenameComment(const SMLoc &L); 245 246 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 247 ArrayRef<MCAsmMacroParameter> Parameters); 248 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 249 ArrayRef<MCAsmMacroParameter> Parameters, 250 ArrayRef<MCAsmMacroArgument> A, 251 const SMLoc &L); 252 253 /// \brief Are macros enabled in the parser? 254 bool areMacrosEnabled() {return MacrosEnabledFlag;} 255 256 /// \brief Control a flag in the parser that enables or disables macros. 257 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 258 259 /// \brief Lookup a previously defined macro. 260 /// \param Name Macro name. 261 /// \returns Pointer to macro. NULL if no such macro was defined. 262 const MCAsmMacro* lookupMacro(StringRef Name); 263 264 /// \brief Define a new macro with the given name and information. 265 void defineMacro(StringRef Name, MCAsmMacro Macro); 266 267 /// \brief Undefine a macro. If no such macro was defined, it's a no-op. 268 void undefineMacro(StringRef Name); 269 270 /// \brief Are we inside a macro instantiation? 271 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 272 273 /// \brief Handle entry to macro instantiation. 274 /// 275 /// \param M The macro. 276 /// \param NameLoc Instantiation location. 277 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 278 279 /// \brief Handle exit from macro instantiation. 280 void handleMacroExit(); 281 282 /// \brief Extract AsmTokens for a macro argument. 283 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 284 285 /// \brief Parse all macro arguments for a given macro. 286 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 287 288 void printMacroInstantiations(); 289 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 290 ArrayRef<SMRange> Ranges = None) const { 291 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 292 } 293 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 294 295 /// \brief Enter the specified file. This returns true on failure. 296 bool enterIncludeFile(const std::string &Filename); 297 298 /// \brief Process the specified file for the .incbin directive. 299 /// This returns true on failure. 300 bool processIncbinFile(const std::string &Filename); 301 302 /// \brief Reset the current lexer position to that given by \p Loc. The 303 /// current token is not set; clients should ensure Lex() is called 304 /// subsequently. 305 /// 306 /// \param InBuffer If not 0, should be the known buffer id that contains the 307 /// location. 308 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 309 310 /// \brief Parse up to the end of statement and a return the contents from the 311 /// current token until the end of the statement; the current token on exit 312 /// will be either the EndOfStatement or EOF. 313 StringRef parseStringToEndOfStatement() override; 314 315 /// \brief Parse until the end of a statement or a comma is encountered, 316 /// return the contents from the current token up to the end or comma. 317 StringRef parseStringToComma(); 318 319 bool parseAssignment(StringRef Name, bool allow_redef, 320 bool NoDeadStrip = false); 321 322 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 323 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 324 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 325 326 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 327 328 // Generic (target and platform independent) directive parsing. 329 enum DirectiveKind { 330 DK_NO_DIRECTIVE, // Placeholder 331 DK_SET, DK_EQU, DK_EQUIV, DK_ASCII, DK_ASCIZ, DK_STRING, DK_BYTE, DK_SHORT, 332 DK_VALUE, DK_2BYTE, DK_LONG, DK_INT, DK_4BYTE, DK_QUAD, DK_8BYTE, DK_OCTA, 333 DK_SINGLE, DK_FLOAT, DK_DOUBLE, DK_ALIGN, DK_ALIGN32, DK_BALIGN, DK_BALIGNW, 334 DK_BALIGNL, DK_P2ALIGN, DK_P2ALIGNW, DK_P2ALIGNL, DK_ORG, DK_FILL, DK_ENDR, 335 DK_BUNDLE_ALIGN_MODE, DK_BUNDLE_LOCK, DK_BUNDLE_UNLOCK, 336 DK_ZERO, DK_EXTERN, DK_GLOBL, DK_GLOBAL, 337 DK_LAZY_REFERENCE, DK_NO_DEAD_STRIP, DK_SYMBOL_RESOLVER, DK_PRIVATE_EXTERN, 338 DK_REFERENCE, DK_WEAK_DEFINITION, DK_WEAK_REFERENCE, 339 DK_WEAK_DEF_CAN_BE_HIDDEN, DK_COMM, DK_COMMON, DK_LCOMM, DK_ABORT, 340 DK_INCLUDE, DK_INCBIN, DK_CODE16, DK_CODE16GCC, DK_REPT, DK_IRP, DK_IRPC, 341 DK_IF, DK_IFEQ, DK_IFGE, DK_IFGT, DK_IFLE, DK_IFLT, DK_IFNE, DK_IFB, 342 DK_IFNB, DK_IFC, DK_IFEQS, DK_IFNC, DK_IFDEF, DK_IFNDEF, DK_IFNOTDEF, 343 DK_ELSEIF, DK_ELSE, DK_ENDIF, 344 DK_SPACE, DK_SKIP, DK_FILE, DK_LINE, DK_LOC, DK_STABS, 345 DK_CFI_SECTIONS, DK_CFI_STARTPROC, DK_CFI_ENDPROC, DK_CFI_DEF_CFA, 346 DK_CFI_DEF_CFA_OFFSET, DK_CFI_ADJUST_CFA_OFFSET, DK_CFI_DEF_CFA_REGISTER, 347 DK_CFI_OFFSET, DK_CFI_REL_OFFSET, DK_CFI_PERSONALITY, DK_CFI_LSDA, 348 DK_CFI_REMEMBER_STATE, DK_CFI_RESTORE_STATE, DK_CFI_SAME_VALUE, 349 DK_CFI_RESTORE, DK_CFI_ESCAPE, DK_CFI_SIGNAL_FRAME, DK_CFI_UNDEFINED, 350 DK_CFI_REGISTER, DK_CFI_WINDOW_SAVE, 351 DK_MACROS_ON, DK_MACROS_OFF, 352 DK_MACRO, DK_EXITM, DK_ENDM, DK_ENDMACRO, DK_PURGEM, 353 DK_SLEB128, DK_ULEB128, 354 DK_ERR, DK_ERROR, DK_WARNING, 355 DK_END 356 }; 357 358 /// \brief Maps directive name --> DirectiveKind enum, for 359 /// directives parsed by this class. 360 StringMap<DirectiveKind> DirectiveKindMap; 361 362 // ".ascii", ".asciz", ".string" 363 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 364 bool parseDirectiveValue(unsigned Size); // ".byte", ".long", ... 365 bool parseDirectiveOctaValue(); // ".octa" 366 bool parseDirectiveRealValue(const fltSemantics &); // ".single", ... 367 bool parseDirectiveFill(); // ".fill" 368 bool parseDirectiveZero(); // ".zero" 369 // ".set", ".equ", ".equiv" 370 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 371 bool parseDirectiveOrg(); // ".org" 372 // ".align{,32}", ".p2align{,w,l}" 373 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 374 375 // ".file", ".line", ".loc", ".stabs" 376 bool parseDirectiveFile(SMLoc DirectiveLoc); 377 bool parseDirectiveLine(); 378 bool parseDirectiveLoc(); 379 bool parseDirectiveStabs(); 380 381 // .cfi directives 382 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 383 bool parseDirectiveCFIWindowSave(); 384 bool parseDirectiveCFISections(); 385 bool parseDirectiveCFIStartProc(); 386 bool parseDirectiveCFIEndProc(); 387 bool parseDirectiveCFIDefCfaOffset(); 388 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 389 bool parseDirectiveCFIAdjustCfaOffset(); 390 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 391 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 392 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 393 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 394 bool parseDirectiveCFIRememberState(); 395 bool parseDirectiveCFIRestoreState(); 396 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 397 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 398 bool parseDirectiveCFIEscape(); 399 bool parseDirectiveCFISignalFrame(); 400 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 401 402 // macro directives 403 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 404 bool parseDirectiveExitMacro(StringRef Directive); 405 bool parseDirectiveEndMacro(StringRef Directive); 406 bool parseDirectiveMacro(SMLoc DirectiveLoc); 407 bool parseDirectiveMacrosOnOff(StringRef Directive); 408 409 // ".bundle_align_mode" 410 bool parseDirectiveBundleAlignMode(); 411 // ".bundle_lock" 412 bool parseDirectiveBundleLock(); 413 // ".bundle_unlock" 414 bool parseDirectiveBundleUnlock(); 415 416 // ".space", ".skip" 417 bool parseDirectiveSpace(StringRef IDVal); 418 419 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 420 bool parseDirectiveLEB128(bool Signed); 421 422 /// \brief Parse a directive like ".globl" which 423 /// accepts a single symbol (which should be a label or an external). 424 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 425 426 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 427 428 bool parseDirectiveAbort(); // ".abort" 429 bool parseDirectiveInclude(); // ".include" 430 bool parseDirectiveIncbin(); // ".incbin" 431 432 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 433 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 434 // ".ifb" or ".ifnb", depending on ExpectBlank. 435 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 436 // ".ifc" or ".ifnc", depending on ExpectEqual. 437 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 438 // ".ifeqs" 439 bool parseDirectiveIfeqs(SMLoc DirectiveLoc); 440 // ".ifdef" or ".ifndef", depending on expect_defined 441 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 442 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 443 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 444 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 445 bool parseEscapedString(std::string &Data) override; 446 447 const MCExpr *applyModifierToExpr(const MCExpr *E, 448 MCSymbolRefExpr::VariantKind Variant); 449 450 // Macro-like directives 451 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 452 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 453 raw_svector_ostream &OS); 454 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 455 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 456 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 457 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 458 459 // "_emit" or "__emit" 460 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 461 size_t Len); 462 463 // "align" 464 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 465 466 // "end" 467 bool parseDirectiveEnd(SMLoc DirectiveLoc); 468 469 // ".err" or ".error" 470 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 471 472 // ".warning" 473 bool parseDirectiveWarning(SMLoc DirectiveLoc); 474 475 void initializeDirectiveKindMap(); 476 }; 477 } 478 479 namespace llvm { 480 481 extern MCAsmParserExtension *createDarwinAsmParser(); 482 extern MCAsmParserExtension *createELFAsmParser(); 483 extern MCAsmParserExtension *createCOFFAsmParser(); 484 485 } 486 487 enum { DEFAULT_ADDRSPACE = 0 }; 488 489 AsmParser::AsmParser(SourceMgr &_SM, MCContext &_Ctx, MCStreamer &_Out, 490 const MCAsmInfo &_MAI) 491 : Lexer(_MAI), Ctx(_Ctx), Out(_Out), MAI(_MAI), SrcMgr(_SM), 492 PlatformParser(nullptr), CurBuffer(_SM.getMainFileID()), 493 MacrosEnabledFlag(true), HadError(false), CppHashLineNumber(0), 494 AssemblerDialect(~0U), IsDarwin(false), ParsingInlineAsm(false) { 495 // Save the old handler. 496 SavedDiagHandler = SrcMgr.getDiagHandler(); 497 SavedDiagContext = SrcMgr.getDiagContext(); 498 // Set our own handler which calls the saved handler. 499 SrcMgr.setDiagHandler(DiagHandler, this); 500 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 501 502 // Initialize the platform / file format parser. 503 switch (_Ctx.getObjectFileInfo()->getObjectFileType()) { 504 case MCObjectFileInfo::IsCOFF: 505 PlatformParser.reset(createCOFFAsmParser()); 506 break; 507 case MCObjectFileInfo::IsMachO: 508 PlatformParser.reset(createDarwinAsmParser()); 509 IsDarwin = true; 510 break; 511 case MCObjectFileInfo::IsELF: 512 PlatformParser.reset(createELFAsmParser()); 513 break; 514 } 515 516 PlatformParser->Initialize(*this); 517 initializeDirectiveKindMap(); 518 } 519 520 AsmParser::~AsmParser() { 521 assert((HadError || ActiveMacros.empty()) && 522 "Unexpected active macro instantiation!"); 523 } 524 525 void AsmParser::printMacroInstantiations() { 526 // Print the active macro instantiation stack. 527 for (std::vector<MacroInstantiation *>::const_reverse_iterator 528 it = ActiveMacros.rbegin(), 529 ie = ActiveMacros.rend(); 530 it != ie; ++it) 531 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 532 "while in macro instantiation"); 533 } 534 535 void AsmParser::Note(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 536 printMessage(L, SourceMgr::DK_Note, Msg, Ranges); 537 printMacroInstantiations(); 538 } 539 540 bool AsmParser::Warning(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 541 if (getTargetParser().getTargetOptions().MCFatalWarnings) 542 return Error(L, Msg, Ranges); 543 printMessage(L, SourceMgr::DK_Warning, Msg, Ranges); 544 printMacroInstantiations(); 545 return false; 546 } 547 548 bool AsmParser::Error(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 549 HadError = true; 550 printMessage(L, SourceMgr::DK_Error, Msg, Ranges); 551 printMacroInstantiations(); 552 return true; 553 } 554 555 bool AsmParser::enterIncludeFile(const std::string &Filename) { 556 std::string IncludedFile; 557 unsigned NewBuf = 558 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 559 if (!NewBuf) 560 return true; 561 562 CurBuffer = NewBuf; 563 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 564 return false; 565 } 566 567 /// Process the specified .incbin file by searching for it in the include paths 568 /// then just emitting the byte contents of the file to the streamer. This 569 /// returns true on failure. 570 bool AsmParser::processIncbinFile(const std::string &Filename) { 571 std::string IncludedFile; 572 unsigned NewBuf = 573 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 574 if (!NewBuf) 575 return true; 576 577 // Pick up the bytes from the file and emit them. 578 getStreamer().EmitBytes(SrcMgr.getMemoryBuffer(NewBuf)->getBuffer()); 579 return false; 580 } 581 582 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 583 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 584 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 585 Loc.getPointer()); 586 } 587 588 const AsmToken &AsmParser::Lex() { 589 const AsmToken *tok = &Lexer.Lex(); 590 591 if (tok->is(AsmToken::Eof)) { 592 // If this is the end of an included file, pop the parent file off the 593 // include stack. 594 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 595 if (ParentIncludeLoc != SMLoc()) { 596 jumpToLoc(ParentIncludeLoc); 597 tok = &Lexer.Lex(); 598 } 599 } 600 601 if (tok->is(AsmToken::Error)) 602 Error(Lexer.getErrLoc(), Lexer.getErr()); 603 604 return *tok; 605 } 606 607 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 608 // Create the initial section, if requested. 609 if (!NoInitialTextSection) 610 Out.InitSections(false); 611 612 // Prime the lexer. 613 Lex(); 614 615 HadError = false; 616 AsmCond StartingCondState = TheCondState; 617 618 // If we are generating dwarf for assembly source files save the initial text 619 // section and generate a .file directive. 620 if (getContext().getGenDwarfForAssembly()) { 621 MCSymbol *SectionStartSym = getContext().CreateTempSymbol(); 622 getStreamer().EmitLabel(SectionStartSym); 623 auto InsertResult = getContext().addGenDwarfSection( 624 getStreamer().getCurrentSection().first); 625 assert(InsertResult.second && ".text section should not have debug info yet"); 626 InsertResult.first->second.first = SectionStartSym; 627 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 628 0, StringRef(), getContext().getMainFileName())); 629 } 630 631 // While we have input, parse each statement. 632 while (Lexer.isNot(AsmToken::Eof)) { 633 ParseStatementInfo Info; 634 if (!parseStatement(Info, nullptr)) 635 continue; 636 637 // We had an error, validate that one was emitted and recover by skipping to 638 // the next line. 639 assert(HadError && "Parse statement returned an error, but none emitted!"); 640 eatToEndOfStatement(); 641 } 642 643 if (TheCondState.TheCond != StartingCondState.TheCond || 644 TheCondState.Ignore != StartingCondState.Ignore) 645 return TokError("unmatched .ifs or .elses"); 646 647 // Check to see there are no empty DwarfFile slots. 648 const auto &LineTables = getContext().getMCDwarfLineTables(); 649 if (!LineTables.empty()) { 650 unsigned Index = 0; 651 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 652 if (File.Name.empty() && Index != 0) 653 TokError("unassigned file number: " + Twine(Index) + 654 " for .file directives"); 655 ++Index; 656 } 657 } 658 659 // Check to see that all assembler local symbols were actually defined. 660 // Targets that don't do subsections via symbols may not want this, though, 661 // so conservatively exclude them. Only do this if we're finalizing, though, 662 // as otherwise we won't necessarilly have seen everything yet. 663 if (!NoFinalize && MAI.hasSubsectionsViaSymbols()) { 664 const MCContext::SymbolTable &Symbols = getContext().getSymbols(); 665 for (MCContext::SymbolTable::const_iterator i = Symbols.begin(), 666 e = Symbols.end(); 667 i != e; ++i) { 668 MCSymbol *Sym = i->getValue(); 669 // Variable symbols may not be marked as defined, so check those 670 // explicitly. If we know it's a variable, we have a definition for 671 // the purposes of this check. 672 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 673 // FIXME: We would really like to refer back to where the symbol was 674 // first referenced for a source location. We need to add something 675 // to track that. Currently, we just point to the end of the file. 676 printMessage( 677 getLexer().getLoc(), SourceMgr::DK_Error, 678 "assembler local symbol '" + Sym->getName() + "' not defined"); 679 } 680 } 681 682 // Finalize the output stream if there are no errors and if the client wants 683 // us to. 684 if (!HadError && !NoFinalize) 685 Out.Finish(); 686 687 return HadError; 688 } 689 690 void AsmParser::checkForValidSection() { 691 if (!ParsingInlineAsm && !getStreamer().getCurrentSection().first) { 692 TokError("expected section directive before assembly directive"); 693 Out.InitSections(false); 694 } 695 } 696 697 /// \brief Throw away the rest of the line for testing purposes. 698 void AsmParser::eatToEndOfStatement() { 699 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 700 Lex(); 701 702 // Eat EOL. 703 if (Lexer.is(AsmToken::EndOfStatement)) 704 Lex(); 705 } 706 707 StringRef AsmParser::parseStringToEndOfStatement() { 708 const char *Start = getTok().getLoc().getPointer(); 709 710 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 711 Lex(); 712 713 const char *End = getTok().getLoc().getPointer(); 714 return StringRef(Start, End - Start); 715 } 716 717 StringRef AsmParser::parseStringToComma() { 718 const char *Start = getTok().getLoc().getPointer(); 719 720 while (Lexer.isNot(AsmToken::EndOfStatement) && 721 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 722 Lex(); 723 724 const char *End = getTok().getLoc().getPointer(); 725 return StringRef(Start, End - Start); 726 } 727 728 /// \brief Parse a paren expression and return it. 729 /// NOTE: This assumes the leading '(' has already been consumed. 730 /// 731 /// parenexpr ::= expr) 732 /// 733 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 734 if (parseExpression(Res)) 735 return true; 736 if (Lexer.isNot(AsmToken::RParen)) 737 return TokError("expected ')' in parentheses expression"); 738 EndLoc = Lexer.getTok().getEndLoc(); 739 Lex(); 740 return false; 741 } 742 743 /// \brief Parse a bracket expression and return it. 744 /// NOTE: This assumes the leading '[' has already been consumed. 745 /// 746 /// bracketexpr ::= expr] 747 /// 748 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 749 if (parseExpression(Res)) 750 return true; 751 if (Lexer.isNot(AsmToken::RBrac)) 752 return TokError("expected ']' in brackets expression"); 753 EndLoc = Lexer.getTok().getEndLoc(); 754 Lex(); 755 return false; 756 } 757 758 /// \brief Parse a primary expression and return it. 759 /// primaryexpr ::= (parenexpr 760 /// primaryexpr ::= symbol 761 /// primaryexpr ::= number 762 /// primaryexpr ::= '.' 763 /// primaryexpr ::= ~,+,- primaryexpr 764 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 765 SMLoc FirstTokenLoc = getLexer().getLoc(); 766 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 767 switch (FirstTokenKind) { 768 default: 769 return TokError("unknown token in expression"); 770 // If we have an error assume that we've already handled it. 771 case AsmToken::Error: 772 return true; 773 case AsmToken::Exclaim: 774 Lex(); // Eat the operator. 775 if (parsePrimaryExpr(Res, EndLoc)) 776 return true; 777 Res = MCUnaryExpr::CreateLNot(Res, getContext()); 778 return false; 779 case AsmToken::Dollar: 780 case AsmToken::At: 781 case AsmToken::String: 782 case AsmToken::Identifier: { 783 StringRef Identifier; 784 if (parseIdentifier(Identifier)) { 785 if (FirstTokenKind == AsmToken::Dollar) { 786 if (Lexer.getMAI().getDollarIsPC()) { 787 // This is a '$' reference, which references the current PC. Emit a 788 // temporary label to the streamer and refer to it. 789 MCSymbol *Sym = Ctx.CreateTempSymbol(); 790 Out.EmitLabel(Sym); 791 Res = MCSymbolRefExpr::Create(Sym, MCSymbolRefExpr::VK_None, 792 getContext()); 793 EndLoc = FirstTokenLoc; 794 return false; 795 } 796 return Error(FirstTokenLoc, "invalid token in expression"); 797 } 798 } 799 // Parse symbol variant 800 std::pair<StringRef, StringRef> Split; 801 if (!MAI.useParensForSymbolVariant()) { 802 if (FirstTokenKind == AsmToken::String) { 803 if (Lexer.is(AsmToken::At)) { 804 Lexer.Lex(); // eat @ 805 SMLoc AtLoc = getLexer().getLoc(); 806 StringRef VName; 807 if (parseIdentifier(VName)) 808 return Error(AtLoc, "expected symbol variant after '@'"); 809 810 Split = std::make_pair(Identifier, VName); 811 } 812 } else { 813 Split = Identifier.split('@'); 814 } 815 } else if (Lexer.is(AsmToken::LParen)) { 816 Lexer.Lex(); // eat ( 817 StringRef VName; 818 parseIdentifier(VName); 819 if (Lexer.isNot(AsmToken::RParen)) { 820 return Error(Lexer.getTok().getLoc(), 821 "unexpected token in variant, expected ')'"); 822 } 823 Lexer.Lex(); // eat ) 824 Split = std::make_pair(Identifier, VName); 825 } 826 827 EndLoc = SMLoc::getFromPointer(Identifier.end()); 828 829 // This is a symbol reference. 830 StringRef SymbolName = Identifier; 831 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 832 833 // Lookup the symbol variant if used. 834 if (Split.second.size()) { 835 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 836 if (Variant != MCSymbolRefExpr::VK_Invalid) { 837 SymbolName = Split.first; 838 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 839 Variant = MCSymbolRefExpr::VK_None; 840 } else { 841 return Error(SMLoc::getFromPointer(Split.second.begin()), 842 "invalid variant '" + Split.second + "'"); 843 } 844 } 845 846 MCSymbol *Sym = getContext().GetOrCreateSymbol(SymbolName); 847 848 // If this is an absolute variable reference, substitute it now to preserve 849 // semantics in the face of reassignment. 850 if (Sym->isVariable() && isa<MCConstantExpr>(Sym->getVariableValue())) { 851 if (Variant) 852 return Error(EndLoc, "unexpected modifier on variable reference"); 853 854 Res = Sym->getVariableValue(); 855 return false; 856 } 857 858 // Otherwise create a symbol ref. 859 Res = MCSymbolRefExpr::Create(Sym, Variant, getContext()); 860 return false; 861 } 862 case AsmToken::BigNum: 863 return TokError("literal value out of range for directive"); 864 case AsmToken::Integer: { 865 SMLoc Loc = getTok().getLoc(); 866 int64_t IntVal = getTok().getIntVal(); 867 Res = MCConstantExpr::Create(IntVal, getContext()); 868 EndLoc = Lexer.getTok().getEndLoc(); 869 Lex(); // Eat token. 870 // Look for 'b' or 'f' following an Integer as a directional label 871 if (Lexer.getKind() == AsmToken::Identifier) { 872 StringRef IDVal = getTok().getString(); 873 // Lookup the symbol variant if used. 874 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 875 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 876 if (Split.first.size() != IDVal.size()) { 877 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 878 if (Variant == MCSymbolRefExpr::VK_Invalid) 879 return TokError("invalid variant '" + Split.second + "'"); 880 IDVal = Split.first; 881 } 882 if (IDVal == "f" || IDVal == "b") { 883 MCSymbol *Sym = 884 Ctx.GetDirectionalLocalSymbol(IntVal, IDVal == "b"); 885 Res = MCSymbolRefExpr::Create(Sym, Variant, getContext()); 886 if (IDVal == "b" && Sym->isUndefined()) 887 return Error(Loc, "invalid reference to undefined symbol"); 888 EndLoc = Lexer.getTok().getEndLoc(); 889 Lex(); // Eat identifier. 890 } 891 } 892 return false; 893 } 894 case AsmToken::Real: { 895 APFloat RealVal(APFloat::IEEEdouble, getTok().getString()); 896 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 897 Res = MCConstantExpr::Create(IntVal, getContext()); 898 EndLoc = Lexer.getTok().getEndLoc(); 899 Lex(); // Eat token. 900 return false; 901 } 902 case AsmToken::Dot: { 903 // This is a '.' reference, which references the current PC. Emit a 904 // temporary label to the streamer and refer to it. 905 MCSymbol *Sym = Ctx.CreateTempSymbol(); 906 Out.EmitLabel(Sym); 907 Res = MCSymbolRefExpr::Create(Sym, MCSymbolRefExpr::VK_None, getContext()); 908 EndLoc = Lexer.getTok().getEndLoc(); 909 Lex(); // Eat identifier. 910 return false; 911 } 912 case AsmToken::LParen: 913 Lex(); // Eat the '('. 914 return parseParenExpr(Res, EndLoc); 915 case AsmToken::LBrac: 916 if (!PlatformParser->HasBracketExpressions()) 917 return TokError("brackets expression not supported on this target"); 918 Lex(); // Eat the '['. 919 return parseBracketExpr(Res, EndLoc); 920 case AsmToken::Minus: 921 Lex(); // Eat the operator. 922 if (parsePrimaryExpr(Res, EndLoc)) 923 return true; 924 Res = MCUnaryExpr::CreateMinus(Res, getContext()); 925 return false; 926 case AsmToken::Plus: 927 Lex(); // Eat the operator. 928 if (parsePrimaryExpr(Res, EndLoc)) 929 return true; 930 Res = MCUnaryExpr::CreatePlus(Res, getContext()); 931 return false; 932 case AsmToken::Tilde: 933 Lex(); // Eat the operator. 934 if (parsePrimaryExpr(Res, EndLoc)) 935 return true; 936 Res = MCUnaryExpr::CreateNot(Res, getContext()); 937 return false; 938 } 939 } 940 941 bool AsmParser::parseExpression(const MCExpr *&Res) { 942 SMLoc EndLoc; 943 return parseExpression(Res, EndLoc); 944 } 945 946 const MCExpr * 947 AsmParser::applyModifierToExpr(const MCExpr *E, 948 MCSymbolRefExpr::VariantKind Variant) { 949 // Ask the target implementation about this expression first. 950 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 951 if (NewE) 952 return NewE; 953 // Recurse over the given expression, rebuilding it to apply the given variant 954 // if there is exactly one symbol. 955 switch (E->getKind()) { 956 case MCExpr::Target: 957 case MCExpr::Constant: 958 return nullptr; 959 960 case MCExpr::SymbolRef: { 961 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 962 963 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 964 TokError("invalid variant on expression '" + getTok().getIdentifier() + 965 "' (already modified)"); 966 return E; 967 } 968 969 return MCSymbolRefExpr::Create(&SRE->getSymbol(), Variant, getContext()); 970 } 971 972 case MCExpr::Unary: { 973 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 974 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 975 if (!Sub) 976 return nullptr; 977 return MCUnaryExpr::Create(UE->getOpcode(), Sub, getContext()); 978 } 979 980 case MCExpr::Binary: { 981 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 982 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 983 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 984 985 if (!LHS && !RHS) 986 return nullptr; 987 988 if (!LHS) 989 LHS = BE->getLHS(); 990 if (!RHS) 991 RHS = BE->getRHS(); 992 993 return MCBinaryExpr::Create(BE->getOpcode(), LHS, RHS, getContext()); 994 } 995 } 996 997 llvm_unreachable("Invalid expression kind!"); 998 } 999 1000 /// \brief Parse an expression and return it. 1001 /// 1002 /// expr ::= expr &&,|| expr -> lowest. 1003 /// expr ::= expr |,^,&,! expr 1004 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1005 /// expr ::= expr <<,>> expr 1006 /// expr ::= expr +,- expr 1007 /// expr ::= expr *,/,% expr -> highest. 1008 /// expr ::= primaryexpr 1009 /// 1010 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1011 // Parse the expression. 1012 Res = nullptr; 1013 if (parsePrimaryExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc)) 1014 return true; 1015 1016 // As a special case, we support 'a op b @ modifier' by rewriting the 1017 // expression to include the modifier. This is inefficient, but in general we 1018 // expect users to use 'a@modifier op b'. 1019 if (Lexer.getKind() == AsmToken::At) { 1020 Lex(); 1021 1022 if (Lexer.isNot(AsmToken::Identifier)) 1023 return TokError("unexpected symbol modifier following '@'"); 1024 1025 MCSymbolRefExpr::VariantKind Variant = 1026 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1027 if (Variant == MCSymbolRefExpr::VK_Invalid) 1028 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1029 1030 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1031 if (!ModifiedRes) { 1032 return TokError("invalid modifier '" + getTok().getIdentifier() + 1033 "' (no symbols present)"); 1034 } 1035 1036 Res = ModifiedRes; 1037 Lex(); 1038 } 1039 1040 // Try to constant fold it up front, if possible. 1041 int64_t Value; 1042 if (Res->EvaluateAsAbsolute(Value)) 1043 Res = MCConstantExpr::Create(Value, getContext()); 1044 1045 return false; 1046 } 1047 1048 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1049 Res = nullptr; 1050 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1051 } 1052 1053 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1054 const MCExpr *Expr; 1055 1056 SMLoc StartLoc = Lexer.getLoc(); 1057 if (parseExpression(Expr)) 1058 return true; 1059 1060 if (!Expr->EvaluateAsAbsolute(Res)) 1061 return Error(StartLoc, "expected absolute expression"); 1062 1063 return false; 1064 } 1065 1066 static unsigned getBinOpPrecedence(AsmToken::TokenKind K, 1067 MCBinaryExpr::Opcode &Kind) { 1068 switch (K) { 1069 default: 1070 return 0; // not a binop. 1071 1072 // Lowest Precedence: &&, || 1073 case AsmToken::AmpAmp: 1074 Kind = MCBinaryExpr::LAnd; 1075 return 1; 1076 case AsmToken::PipePipe: 1077 Kind = MCBinaryExpr::LOr; 1078 return 1; 1079 1080 // Low Precedence: |, &, ^ 1081 // 1082 // FIXME: gas seems to support '!' as an infix operator? 1083 case AsmToken::Pipe: 1084 Kind = MCBinaryExpr::Or; 1085 return 2; 1086 case AsmToken::Caret: 1087 Kind = MCBinaryExpr::Xor; 1088 return 2; 1089 case AsmToken::Amp: 1090 Kind = MCBinaryExpr::And; 1091 return 2; 1092 1093 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1094 case AsmToken::EqualEqual: 1095 Kind = MCBinaryExpr::EQ; 1096 return 3; 1097 case AsmToken::ExclaimEqual: 1098 case AsmToken::LessGreater: 1099 Kind = MCBinaryExpr::NE; 1100 return 3; 1101 case AsmToken::Less: 1102 Kind = MCBinaryExpr::LT; 1103 return 3; 1104 case AsmToken::LessEqual: 1105 Kind = MCBinaryExpr::LTE; 1106 return 3; 1107 case AsmToken::Greater: 1108 Kind = MCBinaryExpr::GT; 1109 return 3; 1110 case AsmToken::GreaterEqual: 1111 Kind = MCBinaryExpr::GTE; 1112 return 3; 1113 1114 // Intermediate Precedence: <<, >> 1115 case AsmToken::LessLess: 1116 Kind = MCBinaryExpr::Shl; 1117 return 4; 1118 case AsmToken::GreaterGreater: 1119 Kind = MCBinaryExpr::Shr; 1120 return 4; 1121 1122 // High Intermediate Precedence: +, - 1123 case AsmToken::Plus: 1124 Kind = MCBinaryExpr::Add; 1125 return 5; 1126 case AsmToken::Minus: 1127 Kind = MCBinaryExpr::Sub; 1128 return 5; 1129 1130 // Highest Precedence: *, /, % 1131 case AsmToken::Star: 1132 Kind = MCBinaryExpr::Mul; 1133 return 6; 1134 case AsmToken::Slash: 1135 Kind = MCBinaryExpr::Div; 1136 return 6; 1137 case AsmToken::Percent: 1138 Kind = MCBinaryExpr::Mod; 1139 return 6; 1140 } 1141 } 1142 1143 /// \brief Parse all binary operators with precedence >= 'Precedence'. 1144 /// Res contains the LHS of the expression on input. 1145 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1146 SMLoc &EndLoc) { 1147 while (1) { 1148 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1149 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1150 1151 // If the next token is lower precedence than we are allowed to eat, return 1152 // successfully with what we ate already. 1153 if (TokPrec < Precedence) 1154 return false; 1155 1156 Lex(); 1157 1158 // Eat the next primary expression. 1159 const MCExpr *RHS; 1160 if (parsePrimaryExpr(RHS, EndLoc)) 1161 return true; 1162 1163 // If BinOp binds less tightly with RHS than the operator after RHS, let 1164 // the pending operator take RHS as its LHS. 1165 MCBinaryExpr::Opcode Dummy; 1166 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1167 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1168 return true; 1169 1170 // Merge LHS and RHS according to operator. 1171 Res = MCBinaryExpr::Create(Kind, Res, RHS, getContext()); 1172 } 1173 } 1174 1175 /// ParseStatement: 1176 /// ::= EndOfStatement 1177 /// ::= Label* Directive ...Operands... EndOfStatement 1178 /// ::= Label* Identifier OperandList* EndOfStatement 1179 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1180 MCAsmParserSemaCallback *SI) { 1181 if (Lexer.is(AsmToken::EndOfStatement)) { 1182 Out.AddBlankLine(); 1183 Lex(); 1184 return false; 1185 } 1186 1187 // Statements always start with an identifier or are a full line comment. 1188 AsmToken ID = getTok(); 1189 SMLoc IDLoc = ID.getLoc(); 1190 StringRef IDVal; 1191 int64_t LocalLabelVal = -1; 1192 // A full line comment is a '#' as the first token. 1193 if (Lexer.is(AsmToken::Hash)) 1194 return parseCppHashLineFilenameComment(IDLoc); 1195 1196 // Allow an integer followed by a ':' as a directional local label. 1197 if (Lexer.is(AsmToken::Integer)) { 1198 LocalLabelVal = getTok().getIntVal(); 1199 if (LocalLabelVal < 0) { 1200 if (!TheCondState.Ignore) 1201 return TokError("unexpected token at start of statement"); 1202 IDVal = ""; 1203 } else { 1204 IDVal = getTok().getString(); 1205 Lex(); // Consume the integer token to be used as an identifier token. 1206 if (Lexer.getKind() != AsmToken::Colon) { 1207 if (!TheCondState.Ignore) 1208 return TokError("unexpected token at start of statement"); 1209 } 1210 } 1211 } else if (Lexer.is(AsmToken::Dot)) { 1212 // Treat '.' as a valid identifier in this context. 1213 Lex(); 1214 IDVal = "."; 1215 } else if (parseIdentifier(IDVal)) { 1216 if (!TheCondState.Ignore) 1217 return TokError("unexpected token at start of statement"); 1218 IDVal = ""; 1219 } 1220 1221 // Handle conditional assembly here before checking for skipping. We 1222 // have to do this so that .endif isn't skipped in a ".if 0" block for 1223 // example. 1224 StringMap<DirectiveKind>::const_iterator DirKindIt = 1225 DirectiveKindMap.find(IDVal); 1226 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1227 ? DK_NO_DIRECTIVE 1228 : DirKindIt->getValue(); 1229 switch (DirKind) { 1230 default: 1231 break; 1232 case DK_IF: 1233 case DK_IFEQ: 1234 case DK_IFGE: 1235 case DK_IFGT: 1236 case DK_IFLE: 1237 case DK_IFLT: 1238 case DK_IFNE: 1239 return parseDirectiveIf(IDLoc, DirKind); 1240 case DK_IFB: 1241 return parseDirectiveIfb(IDLoc, true); 1242 case DK_IFNB: 1243 return parseDirectiveIfb(IDLoc, false); 1244 case DK_IFC: 1245 return parseDirectiveIfc(IDLoc, true); 1246 case DK_IFEQS: 1247 return parseDirectiveIfeqs(IDLoc); 1248 case DK_IFNC: 1249 return parseDirectiveIfc(IDLoc, false); 1250 case DK_IFDEF: 1251 return parseDirectiveIfdef(IDLoc, true); 1252 case DK_IFNDEF: 1253 case DK_IFNOTDEF: 1254 return parseDirectiveIfdef(IDLoc, false); 1255 case DK_ELSEIF: 1256 return parseDirectiveElseIf(IDLoc); 1257 case DK_ELSE: 1258 return parseDirectiveElse(IDLoc); 1259 case DK_ENDIF: 1260 return parseDirectiveEndIf(IDLoc); 1261 } 1262 1263 // Ignore the statement if in the middle of inactive conditional 1264 // (e.g. ".if 0"). 1265 if (TheCondState.Ignore) { 1266 eatToEndOfStatement(); 1267 return false; 1268 } 1269 1270 // FIXME: Recurse on local labels? 1271 1272 // See what kind of statement we have. 1273 switch (Lexer.getKind()) { 1274 case AsmToken::Colon: { 1275 checkForValidSection(); 1276 1277 // identifier ':' -> Label. 1278 Lex(); 1279 1280 // Diagnose attempt to use '.' as a label. 1281 if (IDVal == ".") 1282 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1283 1284 // Diagnose attempt to use a variable as a label. 1285 // 1286 // FIXME: Diagnostics. Note the location of the definition as a label. 1287 // FIXME: This doesn't diagnose assignment to a symbol which has been 1288 // implicitly marked as external. 1289 MCSymbol *Sym; 1290 if (LocalLabelVal == -1) { 1291 if (ParsingInlineAsm && SI) { 1292 StringRef RewrittenLabel = SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1293 assert(RewrittenLabel.size() && "We should have an internal name here."); 1294 Info.AsmRewrites->push_back(AsmRewrite(AOK_Label, IDLoc, 1295 IDVal.size(), RewrittenLabel)); 1296 IDVal = RewrittenLabel; 1297 } 1298 Sym = getContext().GetOrCreateSymbol(IDVal); 1299 } else 1300 Sym = Ctx.CreateDirectionalLocalSymbol(LocalLabelVal); 1301 if (!Sym->isUndefined() || Sym->isVariable()) 1302 return Error(IDLoc, "invalid symbol redefinition"); 1303 1304 // Emit the label. 1305 if (!ParsingInlineAsm) 1306 Out.EmitLabel(Sym); 1307 1308 // If we are generating dwarf for assembly source files then gather the 1309 // info to make a dwarf label entry for this label if needed. 1310 if (getContext().getGenDwarfForAssembly()) 1311 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1312 IDLoc); 1313 1314 getTargetParser().onLabelParsed(Sym); 1315 1316 // Consume any end of statement token, if present, to avoid spurious 1317 // AddBlankLine calls(). 1318 if (Lexer.is(AsmToken::EndOfStatement)) { 1319 Lex(); 1320 if (Lexer.is(AsmToken::Eof)) 1321 return false; 1322 } 1323 1324 return false; 1325 } 1326 1327 case AsmToken::Equal: 1328 // identifier '=' ... -> assignment statement 1329 Lex(); 1330 1331 return parseAssignment(IDVal, true); 1332 1333 default: // Normal instruction or directive. 1334 break; 1335 } 1336 1337 // If macros are enabled, check to see if this is a macro instantiation. 1338 if (areMacrosEnabled()) 1339 if (const MCAsmMacro *M = lookupMacro(IDVal)) { 1340 return handleMacroEntry(M, IDLoc); 1341 } 1342 1343 // Otherwise, we have a normal instruction or directive. 1344 1345 // Directives start with "." 1346 if (IDVal[0] == '.' && IDVal != ".") { 1347 // There are several entities interested in parsing directives: 1348 // 1349 // 1. The target-specific assembly parser. Some directives are target 1350 // specific or may potentially behave differently on certain targets. 1351 // 2. Asm parser extensions. For example, platform-specific parsers 1352 // (like the ELF parser) register themselves as extensions. 1353 // 3. The generic directive parser implemented by this class. These are 1354 // all the directives that behave in a target and platform independent 1355 // manner, or at least have a default behavior that's shared between 1356 // all targets and platforms. 1357 1358 // First query the target-specific parser. It will return 'true' if it 1359 // isn't interested in this directive. 1360 if (!getTargetParser().ParseDirective(ID)) 1361 return false; 1362 1363 // Next, check the extension directive map to see if any extension has 1364 // registered itself to parse this directive. 1365 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1366 ExtensionDirectiveMap.lookup(IDVal); 1367 if (Handler.first) 1368 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1369 1370 // Finally, if no one else is interested in this directive, it must be 1371 // generic and familiar to this class. 1372 switch (DirKind) { 1373 default: 1374 break; 1375 case DK_SET: 1376 case DK_EQU: 1377 return parseDirectiveSet(IDVal, true); 1378 case DK_EQUIV: 1379 return parseDirectiveSet(IDVal, false); 1380 case DK_ASCII: 1381 return parseDirectiveAscii(IDVal, false); 1382 case DK_ASCIZ: 1383 case DK_STRING: 1384 return parseDirectiveAscii(IDVal, true); 1385 case DK_BYTE: 1386 return parseDirectiveValue(1); 1387 case DK_SHORT: 1388 case DK_VALUE: 1389 case DK_2BYTE: 1390 return parseDirectiveValue(2); 1391 case DK_LONG: 1392 case DK_INT: 1393 case DK_4BYTE: 1394 return parseDirectiveValue(4); 1395 case DK_QUAD: 1396 case DK_8BYTE: 1397 return parseDirectiveValue(8); 1398 case DK_OCTA: 1399 return parseDirectiveOctaValue(); 1400 case DK_SINGLE: 1401 case DK_FLOAT: 1402 return parseDirectiveRealValue(APFloat::IEEEsingle); 1403 case DK_DOUBLE: 1404 return parseDirectiveRealValue(APFloat::IEEEdouble); 1405 case DK_ALIGN: { 1406 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1407 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1408 } 1409 case DK_ALIGN32: { 1410 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1411 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1412 } 1413 case DK_BALIGN: 1414 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1415 case DK_BALIGNW: 1416 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1417 case DK_BALIGNL: 1418 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1419 case DK_P2ALIGN: 1420 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1421 case DK_P2ALIGNW: 1422 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1423 case DK_P2ALIGNL: 1424 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1425 case DK_ORG: 1426 return parseDirectiveOrg(); 1427 case DK_FILL: 1428 return parseDirectiveFill(); 1429 case DK_ZERO: 1430 return parseDirectiveZero(); 1431 case DK_EXTERN: 1432 eatToEndOfStatement(); // .extern is the default, ignore it. 1433 return false; 1434 case DK_GLOBL: 1435 case DK_GLOBAL: 1436 return parseDirectiveSymbolAttribute(MCSA_Global); 1437 case DK_LAZY_REFERENCE: 1438 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1439 case DK_NO_DEAD_STRIP: 1440 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1441 case DK_SYMBOL_RESOLVER: 1442 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1443 case DK_PRIVATE_EXTERN: 1444 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1445 case DK_REFERENCE: 1446 return parseDirectiveSymbolAttribute(MCSA_Reference); 1447 case DK_WEAK_DEFINITION: 1448 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1449 case DK_WEAK_REFERENCE: 1450 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1451 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1452 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1453 case DK_COMM: 1454 case DK_COMMON: 1455 return parseDirectiveComm(/*IsLocal=*/false); 1456 case DK_LCOMM: 1457 return parseDirectiveComm(/*IsLocal=*/true); 1458 case DK_ABORT: 1459 return parseDirectiveAbort(); 1460 case DK_INCLUDE: 1461 return parseDirectiveInclude(); 1462 case DK_INCBIN: 1463 return parseDirectiveIncbin(); 1464 case DK_CODE16: 1465 case DK_CODE16GCC: 1466 return TokError(Twine(IDVal) + " not supported yet"); 1467 case DK_REPT: 1468 return parseDirectiveRept(IDLoc, IDVal); 1469 case DK_IRP: 1470 return parseDirectiveIrp(IDLoc); 1471 case DK_IRPC: 1472 return parseDirectiveIrpc(IDLoc); 1473 case DK_ENDR: 1474 return parseDirectiveEndr(IDLoc); 1475 case DK_BUNDLE_ALIGN_MODE: 1476 return parseDirectiveBundleAlignMode(); 1477 case DK_BUNDLE_LOCK: 1478 return parseDirectiveBundleLock(); 1479 case DK_BUNDLE_UNLOCK: 1480 return parseDirectiveBundleUnlock(); 1481 case DK_SLEB128: 1482 return parseDirectiveLEB128(true); 1483 case DK_ULEB128: 1484 return parseDirectiveLEB128(false); 1485 case DK_SPACE: 1486 case DK_SKIP: 1487 return parseDirectiveSpace(IDVal); 1488 case DK_FILE: 1489 return parseDirectiveFile(IDLoc); 1490 case DK_LINE: 1491 return parseDirectiveLine(); 1492 case DK_LOC: 1493 return parseDirectiveLoc(); 1494 case DK_STABS: 1495 return parseDirectiveStabs(); 1496 case DK_CFI_SECTIONS: 1497 return parseDirectiveCFISections(); 1498 case DK_CFI_STARTPROC: 1499 return parseDirectiveCFIStartProc(); 1500 case DK_CFI_ENDPROC: 1501 return parseDirectiveCFIEndProc(); 1502 case DK_CFI_DEF_CFA: 1503 return parseDirectiveCFIDefCfa(IDLoc); 1504 case DK_CFI_DEF_CFA_OFFSET: 1505 return parseDirectiveCFIDefCfaOffset(); 1506 case DK_CFI_ADJUST_CFA_OFFSET: 1507 return parseDirectiveCFIAdjustCfaOffset(); 1508 case DK_CFI_DEF_CFA_REGISTER: 1509 return parseDirectiveCFIDefCfaRegister(IDLoc); 1510 case DK_CFI_OFFSET: 1511 return parseDirectiveCFIOffset(IDLoc); 1512 case DK_CFI_REL_OFFSET: 1513 return parseDirectiveCFIRelOffset(IDLoc); 1514 case DK_CFI_PERSONALITY: 1515 return parseDirectiveCFIPersonalityOrLsda(true); 1516 case DK_CFI_LSDA: 1517 return parseDirectiveCFIPersonalityOrLsda(false); 1518 case DK_CFI_REMEMBER_STATE: 1519 return parseDirectiveCFIRememberState(); 1520 case DK_CFI_RESTORE_STATE: 1521 return parseDirectiveCFIRestoreState(); 1522 case DK_CFI_SAME_VALUE: 1523 return parseDirectiveCFISameValue(IDLoc); 1524 case DK_CFI_RESTORE: 1525 return parseDirectiveCFIRestore(IDLoc); 1526 case DK_CFI_ESCAPE: 1527 return parseDirectiveCFIEscape(); 1528 case DK_CFI_SIGNAL_FRAME: 1529 return parseDirectiveCFISignalFrame(); 1530 case DK_CFI_UNDEFINED: 1531 return parseDirectiveCFIUndefined(IDLoc); 1532 case DK_CFI_REGISTER: 1533 return parseDirectiveCFIRegister(IDLoc); 1534 case DK_CFI_WINDOW_SAVE: 1535 return parseDirectiveCFIWindowSave(); 1536 case DK_MACROS_ON: 1537 case DK_MACROS_OFF: 1538 return parseDirectiveMacrosOnOff(IDVal); 1539 case DK_MACRO: 1540 return parseDirectiveMacro(IDLoc); 1541 case DK_EXITM: 1542 return parseDirectiveExitMacro(IDVal); 1543 case DK_ENDM: 1544 case DK_ENDMACRO: 1545 return parseDirectiveEndMacro(IDVal); 1546 case DK_PURGEM: 1547 return parseDirectivePurgeMacro(IDLoc); 1548 case DK_END: 1549 return parseDirectiveEnd(IDLoc); 1550 case DK_ERR: 1551 return parseDirectiveError(IDLoc, false); 1552 case DK_ERROR: 1553 return parseDirectiveError(IDLoc, true); 1554 case DK_WARNING: 1555 return parseDirectiveWarning(IDLoc); 1556 } 1557 1558 return Error(IDLoc, "unknown directive"); 1559 } 1560 1561 // __asm _emit or __asm __emit 1562 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 1563 IDVal == "_EMIT" || IDVal == "__EMIT")) 1564 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 1565 1566 // __asm align 1567 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 1568 return parseDirectiveMSAlign(IDLoc, Info); 1569 1570 checkForValidSection(); 1571 1572 // Canonicalize the opcode to lower case. 1573 std::string OpcodeStr = IDVal.lower(); 1574 ParseInstructionInfo IInfo(Info.AsmRewrites); 1575 bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, IDLoc, 1576 Info.ParsedOperands); 1577 Info.ParseError = HadError; 1578 1579 // Dump the parsed representation, if requested. 1580 if (getShowParsedOperands()) { 1581 SmallString<256> Str; 1582 raw_svector_ostream OS(Str); 1583 OS << "parsed instruction: ["; 1584 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 1585 if (i != 0) 1586 OS << ", "; 1587 Info.ParsedOperands[i]->print(OS); 1588 } 1589 OS << "]"; 1590 1591 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 1592 } 1593 1594 // If we are generating dwarf for the current section then generate a .loc 1595 // directive for the instruction. 1596 if (!HadError && getContext().getGenDwarfForAssembly() && 1597 getContext().getGenDwarfSectionSyms().count( 1598 getStreamer().getCurrentSection().first)) { 1599 1600 unsigned Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 1601 1602 // If we previously parsed a cpp hash file line comment then make sure the 1603 // current Dwarf File is for the CppHashFilename if not then emit the 1604 // Dwarf File table for it and adjust the line number for the .loc. 1605 if (CppHashFilename.size() != 0) { 1606 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 1607 0, StringRef(), CppHashFilename); 1608 getContext().setGenDwarfFileNumber(FileNumber); 1609 1610 // Since SrcMgr.FindLineNumber() is slow and messes up the SourceMgr's 1611 // cache with the different Loc from the call above we save the last 1612 // info we queried here with SrcMgr.FindLineNumber(). 1613 unsigned CppHashLocLineNo; 1614 if (LastQueryIDLoc == CppHashLoc && LastQueryBuffer == CppHashBuf) 1615 CppHashLocLineNo = LastQueryLine; 1616 else { 1617 CppHashLocLineNo = SrcMgr.FindLineNumber(CppHashLoc, CppHashBuf); 1618 LastQueryLine = CppHashLocLineNo; 1619 LastQueryIDLoc = CppHashLoc; 1620 LastQueryBuffer = CppHashBuf; 1621 } 1622 Line = CppHashLineNumber - 1 + (Line - CppHashLocLineNo); 1623 } 1624 1625 getStreamer().EmitDwarfLocDirective( 1626 getContext().getGenDwarfFileNumber(), Line, 0, 1627 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 1628 StringRef()); 1629 } 1630 1631 // If parsing succeeded, match the instruction. 1632 if (!HadError) { 1633 uint64_t ErrorInfo; 1634 getTargetParser().MatchAndEmitInstruction(IDLoc, Info.Opcode, 1635 Info.ParsedOperands, Out, 1636 ErrorInfo, ParsingInlineAsm); 1637 } 1638 1639 // Don't skip the rest of the line, the instruction parser is responsible for 1640 // that. 1641 return false; 1642 } 1643 1644 /// eatToEndOfLine uses the Lexer to eat the characters to the end of the line 1645 /// since they may not be able to be tokenized to get to the end of line token. 1646 void AsmParser::eatToEndOfLine() { 1647 if (!Lexer.is(AsmToken::EndOfStatement)) 1648 Lexer.LexUntilEndOfLine(); 1649 // Eat EOL. 1650 Lex(); 1651 } 1652 1653 /// parseCppHashLineFilenameComment as this: 1654 /// ::= # number "filename" 1655 /// or just as a full line comment if it doesn't have a number and a string. 1656 bool AsmParser::parseCppHashLineFilenameComment(const SMLoc &L) { 1657 Lex(); // Eat the hash token. 1658 1659 if (getLexer().isNot(AsmToken::Integer)) { 1660 // Consume the line since in cases it is not a well-formed line directive, 1661 // as if were simply a full line comment. 1662 eatToEndOfLine(); 1663 return false; 1664 } 1665 1666 int64_t LineNumber = getTok().getIntVal(); 1667 Lex(); 1668 1669 if (getLexer().isNot(AsmToken::String)) { 1670 eatToEndOfLine(); 1671 return false; 1672 } 1673 1674 StringRef Filename = getTok().getString(); 1675 // Get rid of the enclosing quotes. 1676 Filename = Filename.substr(1, Filename.size() - 2); 1677 1678 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 1679 CppHashLoc = L; 1680 CppHashFilename = Filename; 1681 CppHashLineNumber = LineNumber; 1682 CppHashBuf = CurBuffer; 1683 1684 // Ignore any trailing characters, they're just comment. 1685 eatToEndOfLine(); 1686 return false; 1687 } 1688 1689 /// \brief will use the last parsed cpp hash line filename comment 1690 /// for the Filename and LineNo if any in the diagnostic. 1691 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 1692 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 1693 raw_ostream &OS = errs(); 1694 1695 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 1696 const SMLoc &DiagLoc = Diag.getLoc(); 1697 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1698 unsigned CppHashBuf = 1699 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashLoc); 1700 1701 // Like SourceMgr::printMessage() we need to print the include stack if any 1702 // before printing the message. 1703 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1704 if (!Parser->SavedDiagHandler && DiagCurBuffer && 1705 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 1706 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 1707 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 1708 } 1709 1710 // If we have not parsed a cpp hash line filename comment or the source 1711 // manager changed or buffer changed (like in a nested include) then just 1712 // print the normal diagnostic using its Filename and LineNo. 1713 if (!Parser->CppHashLineNumber || &DiagSrcMgr != &Parser->SrcMgr || 1714 DiagBuf != CppHashBuf) { 1715 if (Parser->SavedDiagHandler) 1716 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 1717 else 1718 Diag.print(nullptr, OS); 1719 return; 1720 } 1721 1722 // Use the CppHashFilename and calculate a line number based on the 1723 // CppHashLoc and CppHashLineNumber relative to this Diag's SMLoc for 1724 // the diagnostic. 1725 const std::string &Filename = Parser->CppHashFilename; 1726 1727 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 1728 int CppHashLocLineNo = 1729 Parser->SrcMgr.FindLineNumber(Parser->CppHashLoc, CppHashBuf); 1730 int LineNo = 1731 Parser->CppHashLineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 1732 1733 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 1734 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 1735 Diag.getLineContents(), Diag.getRanges()); 1736 1737 if (Parser->SavedDiagHandler) 1738 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 1739 else 1740 NewDiag.print(nullptr, OS); 1741 } 1742 1743 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 1744 // difference being that that function accepts '@' as part of identifiers and 1745 // we can't do that. AsmLexer.cpp should probably be changed to handle 1746 // '@' as a special case when needed. 1747 static bool isIdentifierChar(char c) { 1748 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 1749 c == '.'; 1750 } 1751 1752 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 1753 ArrayRef<MCAsmMacroParameter> Parameters, 1754 ArrayRef<MCAsmMacroArgument> A, const SMLoc &L) { 1755 unsigned NParameters = Parameters.size(); 1756 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 1757 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 1758 return Error(L, "Wrong number of arguments"); 1759 1760 // A macro without parameters is handled differently on Darwin: 1761 // gas accepts no arguments and does no substitutions 1762 while (!Body.empty()) { 1763 // Scan for the next substitution. 1764 std::size_t End = Body.size(), Pos = 0; 1765 for (; Pos != End; ++Pos) { 1766 // Check for a substitution or escape. 1767 if (IsDarwin && !NParameters) { 1768 // This macro has no parameters, look for $0, $1, etc. 1769 if (Body[Pos] != '$' || Pos + 1 == End) 1770 continue; 1771 1772 char Next = Body[Pos + 1]; 1773 if (Next == '$' || Next == 'n' || 1774 isdigit(static_cast<unsigned char>(Next))) 1775 break; 1776 } else { 1777 // This macro has parameters, look for \foo, \bar, etc. 1778 if (Body[Pos] == '\\' && Pos + 1 != End) 1779 break; 1780 } 1781 } 1782 1783 // Add the prefix. 1784 OS << Body.slice(0, Pos); 1785 1786 // Check if we reached the end. 1787 if (Pos == End) 1788 break; 1789 1790 if (IsDarwin && !NParameters) { 1791 switch (Body[Pos + 1]) { 1792 // $$ => $ 1793 case '$': 1794 OS << '$'; 1795 break; 1796 1797 // $n => number of arguments 1798 case 'n': 1799 OS << A.size(); 1800 break; 1801 1802 // $[0-9] => argument 1803 default: { 1804 // Missing arguments are ignored. 1805 unsigned Index = Body[Pos + 1] - '0'; 1806 if (Index >= A.size()) 1807 break; 1808 1809 // Otherwise substitute with the token values, with spaces eliminated. 1810 for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), 1811 ie = A[Index].end(); 1812 it != ie; ++it) 1813 OS << it->getString(); 1814 break; 1815 } 1816 } 1817 Pos += 2; 1818 } else { 1819 unsigned I = Pos + 1; 1820 while (isIdentifierChar(Body[I]) && I + 1 != End) 1821 ++I; 1822 1823 const char *Begin = Body.data() + Pos + 1; 1824 StringRef Argument(Begin, I - (Pos + 1)); 1825 unsigned Index = 0; 1826 for (; Index < NParameters; ++Index) 1827 if (Parameters[Index].Name == Argument) 1828 break; 1829 1830 if (Index == NParameters) { 1831 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 1832 Pos += 3; 1833 else { 1834 OS << '\\' << Argument; 1835 Pos = I; 1836 } 1837 } else { 1838 bool VarargParameter = HasVararg && Index == (NParameters - 1); 1839 for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), 1840 ie = A[Index].end(); 1841 it != ie; ++it) 1842 // We expect no quotes around the string's contents when 1843 // parsing for varargs. 1844 if (it->getKind() != AsmToken::String || VarargParameter) 1845 OS << it->getString(); 1846 else 1847 OS << it->getStringContents(); 1848 1849 Pos += 1 + Argument.size(); 1850 } 1851 } 1852 // Update the scan point. 1853 Body = Body.substr(Pos); 1854 } 1855 1856 return false; 1857 } 1858 1859 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 1860 size_t CondStackDepth) 1861 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 1862 CondStackDepth(CondStackDepth) {} 1863 1864 static bool isOperator(AsmToken::TokenKind kind) { 1865 switch (kind) { 1866 default: 1867 return false; 1868 case AsmToken::Plus: 1869 case AsmToken::Minus: 1870 case AsmToken::Tilde: 1871 case AsmToken::Slash: 1872 case AsmToken::Star: 1873 case AsmToken::Dot: 1874 case AsmToken::Equal: 1875 case AsmToken::EqualEqual: 1876 case AsmToken::Pipe: 1877 case AsmToken::PipePipe: 1878 case AsmToken::Caret: 1879 case AsmToken::Amp: 1880 case AsmToken::AmpAmp: 1881 case AsmToken::Exclaim: 1882 case AsmToken::ExclaimEqual: 1883 case AsmToken::Percent: 1884 case AsmToken::Less: 1885 case AsmToken::LessEqual: 1886 case AsmToken::LessLess: 1887 case AsmToken::LessGreater: 1888 case AsmToken::Greater: 1889 case AsmToken::GreaterEqual: 1890 case AsmToken::GreaterGreater: 1891 return true; 1892 } 1893 } 1894 1895 namespace { 1896 class AsmLexerSkipSpaceRAII { 1897 public: 1898 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 1899 Lexer.setSkipSpace(SkipSpace); 1900 } 1901 1902 ~AsmLexerSkipSpaceRAII() { 1903 Lexer.setSkipSpace(true); 1904 } 1905 1906 private: 1907 AsmLexer &Lexer; 1908 }; 1909 } 1910 1911 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 1912 1913 if (Vararg) { 1914 if (Lexer.isNot(AsmToken::EndOfStatement)) { 1915 StringRef Str = parseStringToEndOfStatement(); 1916 MA.push_back(AsmToken(AsmToken::String, Str)); 1917 } 1918 return false; 1919 } 1920 1921 unsigned ParenLevel = 0; 1922 unsigned AddTokens = 0; 1923 1924 // Darwin doesn't use spaces to delmit arguments. 1925 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 1926 1927 for (;;) { 1928 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 1929 return TokError("unexpected token in macro instantiation"); 1930 1931 if (ParenLevel == 0 && Lexer.is(AsmToken::Comma)) 1932 break; 1933 1934 if (Lexer.is(AsmToken::Space)) { 1935 Lex(); // Eat spaces 1936 1937 // Spaces can delimit parameters, but could also be part an expression. 1938 // If the token after a space is an operator, add the token and the next 1939 // one into this argument 1940 if (!IsDarwin) { 1941 if (isOperator(Lexer.getKind())) { 1942 // Check to see whether the token is used as an operator, 1943 // or part of an identifier 1944 const char *NextChar = getTok().getEndLoc().getPointer(); 1945 if (*NextChar == ' ') 1946 AddTokens = 2; 1947 } 1948 1949 if (!AddTokens && ParenLevel == 0) { 1950 break; 1951 } 1952 } 1953 } 1954 1955 // handleMacroEntry relies on not advancing the lexer here 1956 // to be able to fill in the remaining default parameter values 1957 if (Lexer.is(AsmToken::EndOfStatement)) 1958 break; 1959 1960 // Adjust the current parentheses level. 1961 if (Lexer.is(AsmToken::LParen)) 1962 ++ParenLevel; 1963 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 1964 --ParenLevel; 1965 1966 // Append the token to the current argument list. 1967 MA.push_back(getTok()); 1968 if (AddTokens) 1969 AddTokens--; 1970 Lex(); 1971 } 1972 1973 if (ParenLevel != 0) 1974 return TokError("unbalanced parentheses in macro argument"); 1975 return false; 1976 } 1977 1978 // Parse the macro instantiation arguments. 1979 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 1980 MCAsmMacroArguments &A) { 1981 const unsigned NParameters = M ? M->Parameters.size() : 0; 1982 bool NamedParametersFound = false; 1983 SmallVector<SMLoc, 4> FALocs; 1984 1985 A.resize(NParameters); 1986 FALocs.resize(NParameters); 1987 1988 // Parse two kinds of macro invocations: 1989 // - macros defined without any parameters accept an arbitrary number of them 1990 // - macros defined with parameters accept at most that many of them 1991 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 1992 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 1993 ++Parameter) { 1994 SMLoc IDLoc = Lexer.getLoc(); 1995 MCAsmMacroParameter FA; 1996 1997 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 1998 if (parseIdentifier(FA.Name)) { 1999 Error(IDLoc, "invalid argument identifier for formal argument"); 2000 eatToEndOfStatement(); 2001 return true; 2002 } 2003 2004 if (!Lexer.is(AsmToken::Equal)) { 2005 TokError("expected '=' after formal parameter identifier"); 2006 eatToEndOfStatement(); 2007 return true; 2008 } 2009 Lex(); 2010 2011 NamedParametersFound = true; 2012 } 2013 2014 if (NamedParametersFound && FA.Name.empty()) { 2015 Error(IDLoc, "cannot mix positional and keyword arguments"); 2016 eatToEndOfStatement(); 2017 return true; 2018 } 2019 2020 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2021 if (parseMacroArgument(FA.Value, Vararg)) 2022 return true; 2023 2024 unsigned PI = Parameter; 2025 if (!FA.Name.empty()) { 2026 unsigned FAI = 0; 2027 for (FAI = 0; FAI < NParameters; ++FAI) 2028 if (M->Parameters[FAI].Name == FA.Name) 2029 break; 2030 2031 if (FAI >= NParameters) { 2032 assert(M && "expected macro to be defined"); 2033 Error(IDLoc, 2034 "parameter named '" + FA.Name + "' does not exist for macro '" + 2035 M->Name + "'"); 2036 return true; 2037 } 2038 PI = FAI; 2039 } 2040 2041 if (!FA.Value.empty()) { 2042 if (A.size() <= PI) 2043 A.resize(PI + 1); 2044 A[PI] = FA.Value; 2045 2046 if (FALocs.size() <= PI) 2047 FALocs.resize(PI + 1); 2048 2049 FALocs[PI] = Lexer.getLoc(); 2050 } 2051 2052 // At the end of the statement, fill in remaining arguments that have 2053 // default values. If there aren't any, then the next argument is 2054 // required but missing 2055 if (Lexer.is(AsmToken::EndOfStatement)) { 2056 bool Failure = false; 2057 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2058 if (A[FAI].empty()) { 2059 if (M->Parameters[FAI].Required) { 2060 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2061 "missing value for required parameter " 2062 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2063 Failure = true; 2064 } 2065 2066 if (!M->Parameters[FAI].Value.empty()) 2067 A[FAI] = M->Parameters[FAI].Value; 2068 } 2069 } 2070 return Failure; 2071 } 2072 2073 if (Lexer.is(AsmToken::Comma)) 2074 Lex(); 2075 } 2076 2077 return TokError("too many positional arguments"); 2078 } 2079 2080 const MCAsmMacro *AsmParser::lookupMacro(StringRef Name) { 2081 StringMap<MCAsmMacro>::iterator I = MacroMap.find(Name); 2082 return (I == MacroMap.end()) ? nullptr : &I->getValue(); 2083 } 2084 2085 void AsmParser::defineMacro(StringRef Name, MCAsmMacro Macro) { 2086 MacroMap.insert(std::make_pair(Name, std::move(Macro))); 2087 } 2088 2089 void AsmParser::undefineMacro(StringRef Name) { MacroMap.erase(Name); } 2090 2091 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2092 // Arbitrarily limit macro nesting depth, to match 'as'. We can eliminate 2093 // this, although we should protect against infinite loops. 2094 if (ActiveMacros.size() == 20) 2095 return TokError("macros cannot be nested more than 20 levels deep"); 2096 2097 MCAsmMacroArguments A; 2098 if (parseMacroArguments(M, A)) 2099 return true; 2100 2101 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2102 // to hold the macro body with substitutions. 2103 SmallString<256> Buf; 2104 StringRef Body = M->Body; 2105 raw_svector_ostream OS(Buf); 2106 2107 if (expandMacro(OS, Body, M->Parameters, A, getTok().getLoc())) 2108 return true; 2109 2110 // We include the .endmacro in the buffer as our cue to exit the macro 2111 // instantiation. 2112 OS << ".endmacro\n"; 2113 2114 std::unique_ptr<MemoryBuffer> Instantiation = 2115 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2116 2117 // Create the macro instantiation object and add to the current macro 2118 // instantiation stack. 2119 MacroInstantiation *MI = new MacroInstantiation( 2120 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2121 ActiveMacros.push_back(MI); 2122 2123 // Jump to the macro instantiation and prime the lexer. 2124 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2125 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2126 Lex(); 2127 2128 return false; 2129 } 2130 2131 void AsmParser::handleMacroExit() { 2132 // Jump to the EndOfStatement we should return to, and consume it. 2133 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2134 Lex(); 2135 2136 // Pop the instantiation entry. 2137 delete ActiveMacros.back(); 2138 ActiveMacros.pop_back(); 2139 } 2140 2141 static bool isUsedIn(const MCSymbol *Sym, const MCExpr *Value) { 2142 switch (Value->getKind()) { 2143 case MCExpr::Binary: { 2144 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 2145 return isUsedIn(Sym, BE->getLHS()) || isUsedIn(Sym, BE->getRHS()); 2146 } 2147 case MCExpr::Target: 2148 case MCExpr::Constant: 2149 return false; 2150 case MCExpr::SymbolRef: { 2151 const MCSymbol &S = 2152 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 2153 if (S.isVariable()) 2154 return isUsedIn(Sym, S.getVariableValue()); 2155 return &S == Sym; 2156 } 2157 case MCExpr::Unary: 2158 return isUsedIn(Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 2159 } 2160 2161 llvm_unreachable("Unknown expr kind!"); 2162 } 2163 2164 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2165 bool NoDeadStrip) { 2166 // FIXME: Use better location, we should use proper tokens. 2167 SMLoc EqualLoc = Lexer.getLoc(); 2168 2169 const MCExpr *Value; 2170 if (parseExpression(Value)) 2171 return true; 2172 2173 // Note: we don't count b as used in "a = b". This is to allow 2174 // a = b 2175 // b = c 2176 2177 if (Lexer.isNot(AsmToken::EndOfStatement)) 2178 return TokError("unexpected token in assignment"); 2179 2180 // Eat the end of statement marker. 2181 Lex(); 2182 2183 // Validate that the LHS is allowed to be a variable (either it has not been 2184 // used as a symbol, or it is an absolute symbol). 2185 MCSymbol *Sym = getContext().LookupSymbol(Name); 2186 if (Sym) { 2187 // Diagnose assignment to a label. 2188 // 2189 // FIXME: Diagnostics. Note the location of the definition as a label. 2190 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 2191 if (isUsedIn(Sym, Value)) 2192 return Error(EqualLoc, "Recursive use of '" + Name + "'"); 2193 else if (Sym->isUndefined() && !Sym->isUsed() && !Sym->isVariable()) 2194 ; // Allow redefinitions of undefined symbols only used in directives. 2195 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 2196 ; // Allow redefinitions of variables that haven't yet been used. 2197 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 2198 return Error(EqualLoc, "redefinition of '" + Name + "'"); 2199 else if (!Sym->isVariable()) 2200 return Error(EqualLoc, "invalid assignment to '" + Name + "'"); 2201 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 2202 return Error(EqualLoc, "invalid reassignment of non-absolute variable '" + 2203 Name + "'"); 2204 2205 // Don't count these checks as uses. 2206 Sym->setUsed(false); 2207 } else if (Name == ".") { 2208 if (Out.EmitValueToOffset(Value, 0)) { 2209 Error(EqualLoc, "expected absolute expression"); 2210 eatToEndOfStatement(); 2211 } 2212 return false; 2213 } else 2214 Sym = getContext().GetOrCreateSymbol(Name); 2215 2216 // Do the assignment. 2217 Out.EmitAssignment(Sym, Value); 2218 if (NoDeadStrip) 2219 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2220 2221 return false; 2222 } 2223 2224 /// parseIdentifier: 2225 /// ::= identifier 2226 /// ::= string 2227 bool AsmParser::parseIdentifier(StringRef &Res) { 2228 // The assembler has relaxed rules for accepting identifiers, in particular we 2229 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2230 // separate tokens. At this level, we have already lexed so we cannot (currently) 2231 // handle this as a context dependent token, instead we detect adjacent tokens 2232 // and return the combined identifier. 2233 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2234 SMLoc PrefixLoc = getLexer().getLoc(); 2235 2236 // Consume the prefix character, and check for a following identifier. 2237 Lex(); 2238 if (Lexer.isNot(AsmToken::Identifier)) 2239 return true; 2240 2241 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2242 if (PrefixLoc.getPointer() + 1 != getTok().getLoc().getPointer()) 2243 return true; 2244 2245 // Construct the joined identifier and consume the token. 2246 Res = 2247 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2248 Lex(); 2249 return false; 2250 } 2251 2252 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2253 return true; 2254 2255 Res = getTok().getIdentifier(); 2256 2257 Lex(); // Consume the identifier token. 2258 2259 return false; 2260 } 2261 2262 /// parseDirectiveSet: 2263 /// ::= .equ identifier ',' expression 2264 /// ::= .equiv identifier ',' expression 2265 /// ::= .set identifier ',' expression 2266 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2267 StringRef Name; 2268 2269 if (parseIdentifier(Name)) 2270 return TokError("expected identifier after '" + Twine(IDVal) + "'"); 2271 2272 if (getLexer().isNot(AsmToken::Comma)) 2273 return TokError("unexpected token in '" + Twine(IDVal) + "'"); 2274 Lex(); 2275 2276 return parseAssignment(Name, allow_redef, true); 2277 } 2278 2279 bool AsmParser::parseEscapedString(std::string &Data) { 2280 assert(getLexer().is(AsmToken::String) && "Unexpected current token!"); 2281 2282 Data = ""; 2283 StringRef Str = getTok().getStringContents(); 2284 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2285 if (Str[i] != '\\') { 2286 Data += Str[i]; 2287 continue; 2288 } 2289 2290 // Recognize escaped characters. Note that this escape semantics currently 2291 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2292 ++i; 2293 if (i == e) 2294 return TokError("unexpected backslash at end of string"); 2295 2296 // Recognize octal sequences. 2297 if ((unsigned)(Str[i] - '0') <= 7) { 2298 // Consume up to three octal characters. 2299 unsigned Value = Str[i] - '0'; 2300 2301 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2302 ++i; 2303 Value = Value * 8 + (Str[i] - '0'); 2304 2305 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2306 ++i; 2307 Value = Value * 8 + (Str[i] - '0'); 2308 } 2309 } 2310 2311 if (Value > 255) 2312 return TokError("invalid octal escape sequence (out of range)"); 2313 2314 Data += (unsigned char)Value; 2315 continue; 2316 } 2317 2318 // Otherwise recognize individual escapes. 2319 switch (Str[i]) { 2320 default: 2321 // Just reject invalid escape sequences for now. 2322 return TokError("invalid escape sequence (unrecognized character)"); 2323 2324 case 'b': Data += '\b'; break; 2325 case 'f': Data += '\f'; break; 2326 case 'n': Data += '\n'; break; 2327 case 'r': Data += '\r'; break; 2328 case 't': Data += '\t'; break; 2329 case '"': Data += '"'; break; 2330 case '\\': Data += '\\'; break; 2331 } 2332 } 2333 2334 return false; 2335 } 2336 2337 /// parseDirectiveAscii: 2338 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2339 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2340 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2341 checkForValidSection(); 2342 2343 for (;;) { 2344 if (getLexer().isNot(AsmToken::String)) 2345 return TokError("expected string in '" + Twine(IDVal) + "' directive"); 2346 2347 std::string Data; 2348 if (parseEscapedString(Data)) 2349 return true; 2350 2351 getStreamer().EmitBytes(Data); 2352 if (ZeroTerminated) 2353 getStreamer().EmitBytes(StringRef("\0", 1)); 2354 2355 Lex(); 2356 2357 if (getLexer().is(AsmToken::EndOfStatement)) 2358 break; 2359 2360 if (getLexer().isNot(AsmToken::Comma)) 2361 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 2362 Lex(); 2363 } 2364 } 2365 2366 Lex(); 2367 return false; 2368 } 2369 2370 /// parseDirectiveValue 2371 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 2372 bool AsmParser::parseDirectiveValue(unsigned Size) { 2373 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2374 checkForValidSection(); 2375 2376 for (;;) { 2377 const MCExpr *Value; 2378 SMLoc ExprLoc = getLexer().getLoc(); 2379 if (parseExpression(Value)) 2380 return true; 2381 2382 // Special case constant expressions to match code generator. 2383 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2384 assert(Size <= 8 && "Invalid size"); 2385 uint64_t IntValue = MCE->getValue(); 2386 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2387 return Error(ExprLoc, "literal value out of range for directive"); 2388 getStreamer().EmitIntValue(IntValue, Size); 2389 } else 2390 getStreamer().EmitValue(Value, Size, ExprLoc); 2391 2392 if (getLexer().is(AsmToken::EndOfStatement)) 2393 break; 2394 2395 // FIXME: Improve diagnostic. 2396 if (getLexer().isNot(AsmToken::Comma)) 2397 return TokError("unexpected token in directive"); 2398 Lex(); 2399 } 2400 } 2401 2402 Lex(); 2403 return false; 2404 } 2405 2406 /// ParseDirectiveOctaValue 2407 /// ::= .octa [ hexconstant (, hexconstant)* ] 2408 bool AsmParser::parseDirectiveOctaValue() { 2409 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2410 checkForValidSection(); 2411 2412 for (;;) { 2413 if (Lexer.getKind() == AsmToken::Error) 2414 return true; 2415 if (Lexer.getKind() != AsmToken::Integer && 2416 Lexer.getKind() != AsmToken::BigNum) 2417 return TokError("unknown token in expression"); 2418 2419 SMLoc ExprLoc = getLexer().getLoc(); 2420 APInt IntValue = getTok().getAPIntVal(); 2421 Lex(); 2422 2423 uint64_t hi, lo; 2424 if (IntValue.isIntN(64)) { 2425 hi = 0; 2426 lo = IntValue.getZExtValue(); 2427 } else if (IntValue.isIntN(128)) { 2428 // It might actually have more than 128 bits, but the top ones are zero. 2429 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2430 lo = IntValue.getLoBits(64).getZExtValue(); 2431 } else 2432 return Error(ExprLoc, "literal value out of range for directive"); 2433 2434 if (MAI.isLittleEndian()) { 2435 getStreamer().EmitIntValue(lo, 8); 2436 getStreamer().EmitIntValue(hi, 8); 2437 } else { 2438 getStreamer().EmitIntValue(hi, 8); 2439 getStreamer().EmitIntValue(lo, 8); 2440 } 2441 2442 if (getLexer().is(AsmToken::EndOfStatement)) 2443 break; 2444 2445 // FIXME: Improve diagnostic. 2446 if (getLexer().isNot(AsmToken::Comma)) 2447 return TokError("unexpected token in directive"); 2448 Lex(); 2449 } 2450 } 2451 2452 Lex(); 2453 return false; 2454 } 2455 2456 /// parseDirectiveRealValue 2457 /// ::= (.single | .double) [ expression (, expression)* ] 2458 bool AsmParser::parseDirectiveRealValue(const fltSemantics &Semantics) { 2459 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2460 checkForValidSection(); 2461 2462 for (;;) { 2463 // We don't truly support arithmetic on floating point expressions, so we 2464 // have to manually parse unary prefixes. 2465 bool IsNeg = false; 2466 if (getLexer().is(AsmToken::Minus)) { 2467 Lex(); 2468 IsNeg = true; 2469 } else if (getLexer().is(AsmToken::Plus)) 2470 Lex(); 2471 2472 if (getLexer().isNot(AsmToken::Integer) && 2473 getLexer().isNot(AsmToken::Real) && 2474 getLexer().isNot(AsmToken::Identifier)) 2475 return TokError("unexpected token in directive"); 2476 2477 // Convert to an APFloat. 2478 APFloat Value(Semantics); 2479 StringRef IDVal = getTok().getString(); 2480 if (getLexer().is(AsmToken::Identifier)) { 2481 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2482 Value = APFloat::getInf(Semantics); 2483 else if (!IDVal.compare_lower("nan")) 2484 Value = APFloat::getNaN(Semantics, false, ~0); 2485 else 2486 return TokError("invalid floating point literal"); 2487 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 2488 APFloat::opInvalidOp) 2489 return TokError("invalid floating point literal"); 2490 if (IsNeg) 2491 Value.changeSign(); 2492 2493 // Consume the numeric token. 2494 Lex(); 2495 2496 // Emit the value as an integer. 2497 APInt AsInt = Value.bitcastToAPInt(); 2498 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 2499 AsInt.getBitWidth() / 8); 2500 2501 if (getLexer().is(AsmToken::EndOfStatement)) 2502 break; 2503 2504 if (getLexer().isNot(AsmToken::Comma)) 2505 return TokError("unexpected token in directive"); 2506 Lex(); 2507 } 2508 } 2509 2510 Lex(); 2511 return false; 2512 } 2513 2514 /// parseDirectiveZero 2515 /// ::= .zero expression 2516 bool AsmParser::parseDirectiveZero() { 2517 checkForValidSection(); 2518 2519 int64_t NumBytes; 2520 if (parseAbsoluteExpression(NumBytes)) 2521 return true; 2522 2523 int64_t Val = 0; 2524 if (getLexer().is(AsmToken::Comma)) { 2525 Lex(); 2526 if (parseAbsoluteExpression(Val)) 2527 return true; 2528 } 2529 2530 if (getLexer().isNot(AsmToken::EndOfStatement)) 2531 return TokError("unexpected token in '.zero' directive"); 2532 2533 Lex(); 2534 2535 getStreamer().EmitFill(NumBytes, Val); 2536 2537 return false; 2538 } 2539 2540 /// parseDirectiveFill 2541 /// ::= .fill expression [ , expression [ , expression ] ] 2542 bool AsmParser::parseDirectiveFill() { 2543 checkForValidSection(); 2544 2545 SMLoc RepeatLoc = getLexer().getLoc(); 2546 int64_t NumValues; 2547 if (parseAbsoluteExpression(NumValues)) 2548 return true; 2549 2550 if (NumValues < 0) { 2551 Warning(RepeatLoc, 2552 "'.fill' directive with negative repeat count has no effect"); 2553 NumValues = 0; 2554 } 2555 2556 int64_t FillSize = 1; 2557 int64_t FillExpr = 0; 2558 2559 SMLoc SizeLoc, ExprLoc; 2560 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2561 if (getLexer().isNot(AsmToken::Comma)) 2562 return TokError("unexpected token in '.fill' directive"); 2563 Lex(); 2564 2565 SizeLoc = getLexer().getLoc(); 2566 if (parseAbsoluteExpression(FillSize)) 2567 return true; 2568 2569 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2570 if (getLexer().isNot(AsmToken::Comma)) 2571 return TokError("unexpected token in '.fill' directive"); 2572 Lex(); 2573 2574 ExprLoc = getLexer().getLoc(); 2575 if (parseAbsoluteExpression(FillExpr)) 2576 return true; 2577 2578 if (getLexer().isNot(AsmToken::EndOfStatement)) 2579 return TokError("unexpected token in '.fill' directive"); 2580 2581 Lex(); 2582 } 2583 } 2584 2585 if (FillSize < 0) { 2586 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 2587 NumValues = 0; 2588 } 2589 if (FillSize > 8) { 2590 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 2591 FillSize = 8; 2592 } 2593 2594 if (!isUInt<32>(FillExpr) && FillSize > 4) 2595 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 2596 2597 if (NumValues > 0) { 2598 int64_t NonZeroFillSize = FillSize > 4 ? 4 : FillSize; 2599 FillExpr &= ~0ULL >> (64 - NonZeroFillSize * 8); 2600 for (uint64_t i = 0, e = NumValues; i != e; ++i) { 2601 getStreamer().EmitIntValue(FillExpr, NonZeroFillSize); 2602 if (NonZeroFillSize < FillSize) 2603 getStreamer().EmitIntValue(0, FillSize - NonZeroFillSize); 2604 } 2605 } 2606 2607 return false; 2608 } 2609 2610 /// parseDirectiveOrg 2611 /// ::= .org expression [ , expression ] 2612 bool AsmParser::parseDirectiveOrg() { 2613 checkForValidSection(); 2614 2615 const MCExpr *Offset; 2616 SMLoc Loc = getTok().getLoc(); 2617 if (parseExpression(Offset)) 2618 return true; 2619 2620 // Parse optional fill expression. 2621 int64_t FillExpr = 0; 2622 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2623 if (getLexer().isNot(AsmToken::Comma)) 2624 return TokError("unexpected token in '.org' directive"); 2625 Lex(); 2626 2627 if (parseAbsoluteExpression(FillExpr)) 2628 return true; 2629 2630 if (getLexer().isNot(AsmToken::EndOfStatement)) 2631 return TokError("unexpected token in '.org' directive"); 2632 } 2633 2634 Lex(); 2635 2636 // Only limited forms of relocatable expressions are accepted here, it 2637 // has to be relative to the current section. The streamer will return 2638 // 'true' if the expression wasn't evaluatable. 2639 if (getStreamer().EmitValueToOffset(Offset, FillExpr)) 2640 return Error(Loc, "expected assembly-time absolute expression"); 2641 2642 return false; 2643 } 2644 2645 /// parseDirectiveAlign 2646 /// ::= {.align, ...} expression [ , expression [ , expression ]] 2647 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 2648 checkForValidSection(); 2649 2650 SMLoc AlignmentLoc = getLexer().getLoc(); 2651 int64_t Alignment; 2652 if (parseAbsoluteExpression(Alignment)) 2653 return true; 2654 2655 SMLoc MaxBytesLoc; 2656 bool HasFillExpr = false; 2657 int64_t FillExpr = 0; 2658 int64_t MaxBytesToFill = 0; 2659 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2660 if (getLexer().isNot(AsmToken::Comma)) 2661 return TokError("unexpected token in directive"); 2662 Lex(); 2663 2664 // The fill expression can be omitted while specifying a maximum number of 2665 // alignment bytes, e.g: 2666 // .align 3,,4 2667 if (getLexer().isNot(AsmToken::Comma)) { 2668 HasFillExpr = true; 2669 if (parseAbsoluteExpression(FillExpr)) 2670 return true; 2671 } 2672 2673 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2674 if (getLexer().isNot(AsmToken::Comma)) 2675 return TokError("unexpected token in directive"); 2676 Lex(); 2677 2678 MaxBytesLoc = getLexer().getLoc(); 2679 if (parseAbsoluteExpression(MaxBytesToFill)) 2680 return true; 2681 2682 if (getLexer().isNot(AsmToken::EndOfStatement)) 2683 return TokError("unexpected token in directive"); 2684 } 2685 } 2686 2687 Lex(); 2688 2689 if (!HasFillExpr) 2690 FillExpr = 0; 2691 2692 // Compute alignment in bytes. 2693 if (IsPow2) { 2694 // FIXME: Diagnose overflow. 2695 if (Alignment >= 32) { 2696 Error(AlignmentLoc, "invalid alignment value"); 2697 Alignment = 31; 2698 } 2699 2700 Alignment = 1ULL << Alignment; 2701 } else { 2702 // Reject alignments that aren't a power of two, for gas compatibility. 2703 if (!isPowerOf2_64(Alignment)) 2704 Error(AlignmentLoc, "alignment must be a power of 2"); 2705 } 2706 2707 // Diagnose non-sensical max bytes to align. 2708 if (MaxBytesLoc.isValid()) { 2709 if (MaxBytesToFill < 1) { 2710 Error(MaxBytesLoc, "alignment directive can never be satisfied in this " 2711 "many bytes, ignoring maximum bytes expression"); 2712 MaxBytesToFill = 0; 2713 } 2714 2715 if (MaxBytesToFill >= Alignment) { 2716 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 2717 "has no effect"); 2718 MaxBytesToFill = 0; 2719 } 2720 } 2721 2722 // Check whether we should use optimal code alignment for this .align 2723 // directive. 2724 const MCSection *Section = getStreamer().getCurrentSection().first; 2725 assert(Section && "must have section to emit alignment"); 2726 bool UseCodeAlign = Section->UseCodeAlign(); 2727 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 2728 ValueSize == 1 && UseCodeAlign) { 2729 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 2730 } else { 2731 // FIXME: Target specific behavior about how the "extra" bytes are filled. 2732 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 2733 MaxBytesToFill); 2734 } 2735 2736 return false; 2737 } 2738 2739 /// parseDirectiveFile 2740 /// ::= .file [number] filename 2741 /// ::= .file number directory filename 2742 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 2743 // FIXME: I'm not sure what this is. 2744 int64_t FileNumber = -1; 2745 SMLoc FileNumberLoc = getLexer().getLoc(); 2746 if (getLexer().is(AsmToken::Integer)) { 2747 FileNumber = getTok().getIntVal(); 2748 Lex(); 2749 2750 if (FileNumber < 1) 2751 return TokError("file number less than one"); 2752 } 2753 2754 if (getLexer().isNot(AsmToken::String)) 2755 return TokError("unexpected token in '.file' directive"); 2756 2757 // Usually the directory and filename together, otherwise just the directory. 2758 // Allow the strings to have escaped octal character sequence. 2759 std::string Path = getTok().getString(); 2760 if (parseEscapedString(Path)) 2761 return true; 2762 Lex(); 2763 2764 StringRef Directory; 2765 StringRef Filename; 2766 std::string FilenameData; 2767 if (getLexer().is(AsmToken::String)) { 2768 if (FileNumber == -1) 2769 return TokError("explicit path specified, but no file number"); 2770 if (parseEscapedString(FilenameData)) 2771 return true; 2772 Filename = FilenameData; 2773 Directory = Path; 2774 Lex(); 2775 } else { 2776 Filename = Path; 2777 } 2778 2779 if (getLexer().isNot(AsmToken::EndOfStatement)) 2780 return TokError("unexpected token in '.file' directive"); 2781 2782 if (FileNumber == -1) 2783 getStreamer().EmitFileDirective(Filename); 2784 else { 2785 if (getContext().getGenDwarfForAssembly() == true) 2786 Error(DirectiveLoc, 2787 "input can't have .file dwarf directives when -g is " 2788 "used to generate dwarf debug info for assembly code"); 2789 2790 if (getStreamer().EmitDwarfFileDirective(FileNumber, Directory, Filename) == 2791 0) 2792 Error(FileNumberLoc, "file number already allocated"); 2793 } 2794 2795 return false; 2796 } 2797 2798 /// parseDirectiveLine 2799 /// ::= .line [number] 2800 bool AsmParser::parseDirectiveLine() { 2801 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2802 if (getLexer().isNot(AsmToken::Integer)) 2803 return TokError("unexpected token in '.line' directive"); 2804 2805 int64_t LineNumber = getTok().getIntVal(); 2806 (void)LineNumber; 2807 Lex(); 2808 2809 // FIXME: Do something with the .line. 2810 } 2811 2812 if (getLexer().isNot(AsmToken::EndOfStatement)) 2813 return TokError("unexpected token in '.line' directive"); 2814 2815 return false; 2816 } 2817 2818 /// parseDirectiveLoc 2819 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 2820 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 2821 /// The first number is a file number, must have been previously assigned with 2822 /// a .file directive, the second number is the line number and optionally the 2823 /// third number is a column position (zero if not specified). The remaining 2824 /// optional items are .loc sub-directives. 2825 bool AsmParser::parseDirectiveLoc() { 2826 if (getLexer().isNot(AsmToken::Integer)) 2827 return TokError("unexpected token in '.loc' directive"); 2828 int64_t FileNumber = getTok().getIntVal(); 2829 if (FileNumber < 1) 2830 return TokError("file number less than one in '.loc' directive"); 2831 if (!getContext().isValidDwarfFileNumber(FileNumber)) 2832 return TokError("unassigned file number in '.loc' directive"); 2833 Lex(); 2834 2835 int64_t LineNumber = 0; 2836 if (getLexer().is(AsmToken::Integer)) { 2837 LineNumber = getTok().getIntVal(); 2838 if (LineNumber < 0) 2839 return TokError("line number less than zero in '.loc' directive"); 2840 Lex(); 2841 } 2842 2843 int64_t ColumnPos = 0; 2844 if (getLexer().is(AsmToken::Integer)) { 2845 ColumnPos = getTok().getIntVal(); 2846 if (ColumnPos < 0) 2847 return TokError("column position less than zero in '.loc' directive"); 2848 Lex(); 2849 } 2850 2851 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 2852 unsigned Isa = 0; 2853 int64_t Discriminator = 0; 2854 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2855 for (;;) { 2856 if (getLexer().is(AsmToken::EndOfStatement)) 2857 break; 2858 2859 StringRef Name; 2860 SMLoc Loc = getTok().getLoc(); 2861 if (parseIdentifier(Name)) 2862 return TokError("unexpected token in '.loc' directive"); 2863 2864 if (Name == "basic_block") 2865 Flags |= DWARF2_FLAG_BASIC_BLOCK; 2866 else if (Name == "prologue_end") 2867 Flags |= DWARF2_FLAG_PROLOGUE_END; 2868 else if (Name == "epilogue_begin") 2869 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 2870 else if (Name == "is_stmt") { 2871 Loc = getTok().getLoc(); 2872 const MCExpr *Value; 2873 if (parseExpression(Value)) 2874 return true; 2875 // The expression must be the constant 0 or 1. 2876 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2877 int Value = MCE->getValue(); 2878 if (Value == 0) 2879 Flags &= ~DWARF2_FLAG_IS_STMT; 2880 else if (Value == 1) 2881 Flags |= DWARF2_FLAG_IS_STMT; 2882 else 2883 return Error(Loc, "is_stmt value not 0 or 1"); 2884 } else { 2885 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 2886 } 2887 } else if (Name == "isa") { 2888 Loc = getTok().getLoc(); 2889 const MCExpr *Value; 2890 if (parseExpression(Value)) 2891 return true; 2892 // The expression must be a constant greater or equal to 0. 2893 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2894 int Value = MCE->getValue(); 2895 if (Value < 0) 2896 return Error(Loc, "isa number less than zero"); 2897 Isa = Value; 2898 } else { 2899 return Error(Loc, "isa number not a constant value"); 2900 } 2901 } else if (Name == "discriminator") { 2902 if (parseAbsoluteExpression(Discriminator)) 2903 return true; 2904 } else { 2905 return Error(Loc, "unknown sub-directive in '.loc' directive"); 2906 } 2907 2908 if (getLexer().is(AsmToken::EndOfStatement)) 2909 break; 2910 } 2911 } 2912 2913 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 2914 Isa, Discriminator, StringRef()); 2915 2916 return false; 2917 } 2918 2919 /// parseDirectiveStabs 2920 /// ::= .stabs string, number, number, number 2921 bool AsmParser::parseDirectiveStabs() { 2922 return TokError("unsupported directive '.stabs'"); 2923 } 2924 2925 /// parseDirectiveCFISections 2926 /// ::= .cfi_sections section [, section] 2927 bool AsmParser::parseDirectiveCFISections() { 2928 StringRef Name; 2929 bool EH = false; 2930 bool Debug = false; 2931 2932 if (parseIdentifier(Name)) 2933 return TokError("Expected an identifier"); 2934 2935 if (Name == ".eh_frame") 2936 EH = true; 2937 else if (Name == ".debug_frame") 2938 Debug = true; 2939 2940 if (getLexer().is(AsmToken::Comma)) { 2941 Lex(); 2942 2943 if (parseIdentifier(Name)) 2944 return TokError("Expected an identifier"); 2945 2946 if (Name == ".eh_frame") 2947 EH = true; 2948 else if (Name == ".debug_frame") 2949 Debug = true; 2950 } 2951 2952 getStreamer().EmitCFISections(EH, Debug); 2953 return false; 2954 } 2955 2956 /// parseDirectiveCFIStartProc 2957 /// ::= .cfi_startproc [simple] 2958 bool AsmParser::parseDirectiveCFIStartProc() { 2959 StringRef Simple; 2960 if (getLexer().isNot(AsmToken::EndOfStatement)) 2961 if (parseIdentifier(Simple) || Simple != "simple") 2962 return TokError("unexpected token in .cfi_startproc directive"); 2963 2964 getStreamer().EmitCFIStartProc(!Simple.empty()); 2965 return false; 2966 } 2967 2968 /// parseDirectiveCFIEndProc 2969 /// ::= .cfi_endproc 2970 bool AsmParser::parseDirectiveCFIEndProc() { 2971 getStreamer().EmitCFIEndProc(); 2972 return false; 2973 } 2974 2975 /// \brief parse register name or number. 2976 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 2977 SMLoc DirectiveLoc) { 2978 unsigned RegNo; 2979 2980 if (getLexer().isNot(AsmToken::Integer)) { 2981 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 2982 return true; 2983 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 2984 } else 2985 return parseAbsoluteExpression(Register); 2986 2987 return false; 2988 } 2989 2990 /// parseDirectiveCFIDefCfa 2991 /// ::= .cfi_def_cfa register, offset 2992 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 2993 int64_t Register = 0; 2994 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 2995 return true; 2996 2997 if (getLexer().isNot(AsmToken::Comma)) 2998 return TokError("unexpected token in directive"); 2999 Lex(); 3000 3001 int64_t Offset = 0; 3002 if (parseAbsoluteExpression(Offset)) 3003 return true; 3004 3005 getStreamer().EmitCFIDefCfa(Register, Offset); 3006 return false; 3007 } 3008 3009 /// parseDirectiveCFIDefCfaOffset 3010 /// ::= .cfi_def_cfa_offset offset 3011 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3012 int64_t Offset = 0; 3013 if (parseAbsoluteExpression(Offset)) 3014 return true; 3015 3016 getStreamer().EmitCFIDefCfaOffset(Offset); 3017 return false; 3018 } 3019 3020 /// parseDirectiveCFIRegister 3021 /// ::= .cfi_register register, register 3022 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3023 int64_t Register1 = 0; 3024 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc)) 3025 return true; 3026 3027 if (getLexer().isNot(AsmToken::Comma)) 3028 return TokError("unexpected token in directive"); 3029 Lex(); 3030 3031 int64_t Register2 = 0; 3032 if (parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3033 return true; 3034 3035 getStreamer().EmitCFIRegister(Register1, Register2); 3036 return false; 3037 } 3038 3039 /// parseDirectiveCFIWindowSave 3040 /// ::= .cfi_window_save 3041 bool AsmParser::parseDirectiveCFIWindowSave() { 3042 getStreamer().EmitCFIWindowSave(); 3043 return false; 3044 } 3045 3046 /// parseDirectiveCFIAdjustCfaOffset 3047 /// ::= .cfi_adjust_cfa_offset adjustment 3048 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 3049 int64_t Adjustment = 0; 3050 if (parseAbsoluteExpression(Adjustment)) 3051 return true; 3052 3053 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 3054 return false; 3055 } 3056 3057 /// parseDirectiveCFIDefCfaRegister 3058 /// ::= .cfi_def_cfa_register register 3059 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3060 int64_t Register = 0; 3061 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3062 return true; 3063 3064 getStreamer().EmitCFIDefCfaRegister(Register); 3065 return false; 3066 } 3067 3068 /// parseDirectiveCFIOffset 3069 /// ::= .cfi_offset register, offset 3070 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3071 int64_t Register = 0; 3072 int64_t Offset = 0; 3073 3074 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3075 return true; 3076 3077 if (getLexer().isNot(AsmToken::Comma)) 3078 return TokError("unexpected token in directive"); 3079 Lex(); 3080 3081 if (parseAbsoluteExpression(Offset)) 3082 return true; 3083 3084 getStreamer().EmitCFIOffset(Register, Offset); 3085 return false; 3086 } 3087 3088 /// parseDirectiveCFIRelOffset 3089 /// ::= .cfi_rel_offset register, offset 3090 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3091 int64_t Register = 0; 3092 3093 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3094 return true; 3095 3096 if (getLexer().isNot(AsmToken::Comma)) 3097 return TokError("unexpected token in directive"); 3098 Lex(); 3099 3100 int64_t Offset = 0; 3101 if (parseAbsoluteExpression(Offset)) 3102 return true; 3103 3104 getStreamer().EmitCFIRelOffset(Register, Offset); 3105 return false; 3106 } 3107 3108 static bool isValidEncoding(int64_t Encoding) { 3109 if (Encoding & ~0xff) 3110 return false; 3111 3112 if (Encoding == dwarf::DW_EH_PE_omit) 3113 return true; 3114 3115 const unsigned Format = Encoding & 0xf; 3116 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3117 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3118 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3119 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3120 return false; 3121 3122 const unsigned Application = Encoding & 0x70; 3123 if (Application != dwarf::DW_EH_PE_absptr && 3124 Application != dwarf::DW_EH_PE_pcrel) 3125 return false; 3126 3127 return true; 3128 } 3129 3130 /// parseDirectiveCFIPersonalityOrLsda 3131 /// IsPersonality true for cfi_personality, false for cfi_lsda 3132 /// ::= .cfi_personality encoding, [symbol_name] 3133 /// ::= .cfi_lsda encoding, [symbol_name] 3134 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3135 int64_t Encoding = 0; 3136 if (parseAbsoluteExpression(Encoding)) 3137 return true; 3138 if (Encoding == dwarf::DW_EH_PE_omit) 3139 return false; 3140 3141 if (!isValidEncoding(Encoding)) 3142 return TokError("unsupported encoding."); 3143 3144 if (getLexer().isNot(AsmToken::Comma)) 3145 return TokError("unexpected token in directive"); 3146 Lex(); 3147 3148 StringRef Name; 3149 if (parseIdentifier(Name)) 3150 return TokError("expected identifier in directive"); 3151 3152 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3153 3154 if (IsPersonality) 3155 getStreamer().EmitCFIPersonality(Sym, Encoding); 3156 else 3157 getStreamer().EmitCFILsda(Sym, Encoding); 3158 return false; 3159 } 3160 3161 /// parseDirectiveCFIRememberState 3162 /// ::= .cfi_remember_state 3163 bool AsmParser::parseDirectiveCFIRememberState() { 3164 getStreamer().EmitCFIRememberState(); 3165 return false; 3166 } 3167 3168 /// parseDirectiveCFIRestoreState 3169 /// ::= .cfi_remember_state 3170 bool AsmParser::parseDirectiveCFIRestoreState() { 3171 getStreamer().EmitCFIRestoreState(); 3172 return false; 3173 } 3174 3175 /// parseDirectiveCFISameValue 3176 /// ::= .cfi_same_value register 3177 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3178 int64_t Register = 0; 3179 3180 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3181 return true; 3182 3183 getStreamer().EmitCFISameValue(Register); 3184 return false; 3185 } 3186 3187 /// parseDirectiveCFIRestore 3188 /// ::= .cfi_restore register 3189 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3190 int64_t Register = 0; 3191 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3192 return true; 3193 3194 getStreamer().EmitCFIRestore(Register); 3195 return false; 3196 } 3197 3198 /// parseDirectiveCFIEscape 3199 /// ::= .cfi_escape expression[,...] 3200 bool AsmParser::parseDirectiveCFIEscape() { 3201 std::string Values; 3202 int64_t CurrValue; 3203 if (parseAbsoluteExpression(CurrValue)) 3204 return true; 3205 3206 Values.push_back((uint8_t)CurrValue); 3207 3208 while (getLexer().is(AsmToken::Comma)) { 3209 Lex(); 3210 3211 if (parseAbsoluteExpression(CurrValue)) 3212 return true; 3213 3214 Values.push_back((uint8_t)CurrValue); 3215 } 3216 3217 getStreamer().EmitCFIEscape(Values); 3218 return false; 3219 } 3220 3221 /// parseDirectiveCFISignalFrame 3222 /// ::= .cfi_signal_frame 3223 bool AsmParser::parseDirectiveCFISignalFrame() { 3224 if (getLexer().isNot(AsmToken::EndOfStatement)) 3225 return Error(getLexer().getLoc(), 3226 "unexpected token in '.cfi_signal_frame'"); 3227 3228 getStreamer().EmitCFISignalFrame(); 3229 return false; 3230 } 3231 3232 /// parseDirectiveCFIUndefined 3233 /// ::= .cfi_undefined register 3234 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 3235 int64_t Register = 0; 3236 3237 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3238 return true; 3239 3240 getStreamer().EmitCFIUndefined(Register); 3241 return false; 3242 } 3243 3244 /// parseDirectiveMacrosOnOff 3245 /// ::= .macros_on 3246 /// ::= .macros_off 3247 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 3248 if (getLexer().isNot(AsmToken::EndOfStatement)) 3249 return Error(getLexer().getLoc(), 3250 "unexpected token in '" + Directive + "' directive"); 3251 3252 setMacrosEnabled(Directive == ".macros_on"); 3253 return false; 3254 } 3255 3256 /// parseDirectiveMacro 3257 /// ::= .macro name[,] [parameters] 3258 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 3259 StringRef Name; 3260 if (parseIdentifier(Name)) 3261 return TokError("expected identifier in '.macro' directive"); 3262 3263 if (getLexer().is(AsmToken::Comma)) 3264 Lex(); 3265 3266 MCAsmMacroParameters Parameters; 3267 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3268 3269 if (Parameters.size() && Parameters.back().Vararg) 3270 return Error(Lexer.getLoc(), 3271 "Vararg parameter '" + Parameters.back().Name + 3272 "' should be last one in the list of parameters."); 3273 3274 MCAsmMacroParameter Parameter; 3275 if (parseIdentifier(Parameter.Name)) 3276 return TokError("expected identifier in '.macro' directive"); 3277 3278 if (Lexer.is(AsmToken::Colon)) { 3279 Lex(); // consume ':' 3280 3281 SMLoc QualLoc; 3282 StringRef Qualifier; 3283 3284 QualLoc = Lexer.getLoc(); 3285 if (parseIdentifier(Qualifier)) 3286 return Error(QualLoc, "missing parameter qualifier for " 3287 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3288 3289 if (Qualifier == "req") 3290 Parameter.Required = true; 3291 else if (Qualifier == "vararg") 3292 Parameter.Vararg = true; 3293 else 3294 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 3295 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 3296 } 3297 3298 if (getLexer().is(AsmToken::Equal)) { 3299 Lex(); 3300 3301 SMLoc ParamLoc; 3302 3303 ParamLoc = Lexer.getLoc(); 3304 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 3305 return true; 3306 3307 if (Parameter.Required) 3308 Warning(ParamLoc, "pointless default value for required parameter " 3309 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3310 } 3311 3312 Parameters.push_back(std::move(Parameter)); 3313 3314 if (getLexer().is(AsmToken::Comma)) 3315 Lex(); 3316 } 3317 3318 // Eat the end of statement. 3319 Lex(); 3320 3321 AsmToken EndToken, StartToken = getTok(); 3322 unsigned MacroDepth = 0; 3323 3324 // Lex the macro definition. 3325 for (;;) { 3326 // Check whether we have reached the end of the file. 3327 if (getLexer().is(AsmToken::Eof)) 3328 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 3329 3330 // Otherwise, check whether we have reach the .endmacro. 3331 if (getLexer().is(AsmToken::Identifier)) { 3332 if (getTok().getIdentifier() == ".endm" || 3333 getTok().getIdentifier() == ".endmacro") { 3334 if (MacroDepth == 0) { // Outermost macro. 3335 EndToken = getTok(); 3336 Lex(); 3337 if (getLexer().isNot(AsmToken::EndOfStatement)) 3338 return TokError("unexpected token in '" + EndToken.getIdentifier() + 3339 "' directive"); 3340 break; 3341 } else { 3342 // Otherwise we just found the end of an inner macro. 3343 --MacroDepth; 3344 } 3345 } else if (getTok().getIdentifier() == ".macro") { 3346 // We allow nested macros. Those aren't instantiated until the outermost 3347 // macro is expanded so just ignore them for now. 3348 ++MacroDepth; 3349 } 3350 } 3351 3352 // Otherwise, scan til the end of the statement. 3353 eatToEndOfStatement(); 3354 } 3355 3356 if (lookupMacro(Name)) { 3357 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 3358 } 3359 3360 const char *BodyStart = StartToken.getLoc().getPointer(); 3361 const char *BodyEnd = EndToken.getLoc().getPointer(); 3362 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 3363 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 3364 defineMacro(Name, MCAsmMacro(Name, Body, std::move(Parameters))); 3365 return false; 3366 } 3367 3368 /// checkForBadMacro 3369 /// 3370 /// With the support added for named parameters there may be code out there that 3371 /// is transitioning from positional parameters. In versions of gas that did 3372 /// not support named parameters they would be ignored on the macro definition. 3373 /// But to support both styles of parameters this is not possible so if a macro 3374 /// definition has named parameters but does not use them and has what appears 3375 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 3376 /// warning that the positional parameter found in body which have no effect. 3377 /// Hoping the developer will either remove the named parameters from the macro 3378 /// definition so the positional parameters get used if that was what was 3379 /// intended or change the macro to use the named parameters. It is possible 3380 /// this warning will trigger when the none of the named parameters are used 3381 /// and the strings like $1 are infact to simply to be passed trough unchanged. 3382 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 3383 StringRef Body, 3384 ArrayRef<MCAsmMacroParameter> Parameters) { 3385 // If this macro is not defined with named parameters the warning we are 3386 // checking for here doesn't apply. 3387 unsigned NParameters = Parameters.size(); 3388 if (NParameters == 0) 3389 return; 3390 3391 bool NamedParametersFound = false; 3392 bool PositionalParametersFound = false; 3393 3394 // Look at the body of the macro for use of both the named parameters and what 3395 // are likely to be positional parameters. This is what expandMacro() is 3396 // doing when it finds the parameters in the body. 3397 while (!Body.empty()) { 3398 // Scan for the next possible parameter. 3399 std::size_t End = Body.size(), Pos = 0; 3400 for (; Pos != End; ++Pos) { 3401 // Check for a substitution or escape. 3402 // This macro is defined with parameters, look for \foo, \bar, etc. 3403 if (Body[Pos] == '\\' && Pos + 1 != End) 3404 break; 3405 3406 // This macro should have parameters, but look for $0, $1, ..., $n too. 3407 if (Body[Pos] != '$' || Pos + 1 == End) 3408 continue; 3409 char Next = Body[Pos + 1]; 3410 if (Next == '$' || Next == 'n' || 3411 isdigit(static_cast<unsigned char>(Next))) 3412 break; 3413 } 3414 3415 // Check if we reached the end. 3416 if (Pos == End) 3417 break; 3418 3419 if (Body[Pos] == '$') { 3420 switch (Body[Pos + 1]) { 3421 // $$ => $ 3422 case '$': 3423 break; 3424 3425 // $n => number of arguments 3426 case 'n': 3427 PositionalParametersFound = true; 3428 break; 3429 3430 // $[0-9] => argument 3431 default: { 3432 PositionalParametersFound = true; 3433 break; 3434 } 3435 } 3436 Pos += 2; 3437 } else { 3438 unsigned I = Pos + 1; 3439 while (isIdentifierChar(Body[I]) && I + 1 != End) 3440 ++I; 3441 3442 const char *Begin = Body.data() + Pos + 1; 3443 StringRef Argument(Begin, I - (Pos + 1)); 3444 unsigned Index = 0; 3445 for (; Index < NParameters; ++Index) 3446 if (Parameters[Index].Name == Argument) 3447 break; 3448 3449 if (Index == NParameters) { 3450 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 3451 Pos += 3; 3452 else { 3453 Pos = I; 3454 } 3455 } else { 3456 NamedParametersFound = true; 3457 Pos += 1 + Argument.size(); 3458 } 3459 } 3460 // Update the scan point. 3461 Body = Body.substr(Pos); 3462 } 3463 3464 if (!NamedParametersFound && PositionalParametersFound) 3465 Warning(DirectiveLoc, "macro defined with named parameters which are not " 3466 "used in macro body, possible positional parameter " 3467 "found in body which will have no effect"); 3468 } 3469 3470 /// parseDirectiveExitMacro 3471 /// ::= .exitm 3472 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 3473 if (getLexer().isNot(AsmToken::EndOfStatement)) 3474 return TokError("unexpected token in '" + Directive + "' directive"); 3475 3476 if (!isInsideMacroInstantiation()) 3477 return TokError("unexpected '" + Directive + "' in file, " 3478 "no current macro definition"); 3479 3480 // Exit all conditionals that are active in the current macro. 3481 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 3482 TheCondState = TheCondStack.back(); 3483 TheCondStack.pop_back(); 3484 } 3485 3486 handleMacroExit(); 3487 return false; 3488 } 3489 3490 /// parseDirectiveEndMacro 3491 /// ::= .endm 3492 /// ::= .endmacro 3493 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 3494 if (getLexer().isNot(AsmToken::EndOfStatement)) 3495 return TokError("unexpected token in '" + Directive + "' directive"); 3496 3497 // If we are inside a macro instantiation, terminate the current 3498 // instantiation. 3499 if (isInsideMacroInstantiation()) { 3500 handleMacroExit(); 3501 return false; 3502 } 3503 3504 // Otherwise, this .endmacro is a stray entry in the file; well formed 3505 // .endmacro directives are handled during the macro definition parsing. 3506 return TokError("unexpected '" + Directive + "' in file, " 3507 "no current macro definition"); 3508 } 3509 3510 /// parseDirectivePurgeMacro 3511 /// ::= .purgem 3512 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 3513 StringRef Name; 3514 if (parseIdentifier(Name)) 3515 return TokError("expected identifier in '.purgem' directive"); 3516 3517 if (getLexer().isNot(AsmToken::EndOfStatement)) 3518 return TokError("unexpected token in '.purgem' directive"); 3519 3520 if (!lookupMacro(Name)) 3521 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 3522 3523 undefineMacro(Name); 3524 return false; 3525 } 3526 3527 /// parseDirectiveBundleAlignMode 3528 /// ::= {.bundle_align_mode} expression 3529 bool AsmParser::parseDirectiveBundleAlignMode() { 3530 checkForValidSection(); 3531 3532 // Expect a single argument: an expression that evaluates to a constant 3533 // in the inclusive range 0-30. 3534 SMLoc ExprLoc = getLexer().getLoc(); 3535 int64_t AlignSizePow2; 3536 if (parseAbsoluteExpression(AlignSizePow2)) 3537 return true; 3538 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3539 return TokError("unexpected token after expression in" 3540 " '.bundle_align_mode' directive"); 3541 else if (AlignSizePow2 < 0 || AlignSizePow2 > 30) 3542 return Error(ExprLoc, 3543 "invalid bundle alignment size (expected between 0 and 30)"); 3544 3545 Lex(); 3546 3547 // Because of AlignSizePow2's verified range we can safely truncate it to 3548 // unsigned. 3549 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 3550 return false; 3551 } 3552 3553 /// parseDirectiveBundleLock 3554 /// ::= {.bundle_lock} [align_to_end] 3555 bool AsmParser::parseDirectiveBundleLock() { 3556 checkForValidSection(); 3557 bool AlignToEnd = false; 3558 3559 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3560 StringRef Option; 3561 SMLoc Loc = getTok().getLoc(); 3562 const char *kInvalidOptionError = 3563 "invalid option for '.bundle_lock' directive"; 3564 3565 if (parseIdentifier(Option)) 3566 return Error(Loc, kInvalidOptionError); 3567 3568 if (Option != "align_to_end") 3569 return Error(Loc, kInvalidOptionError); 3570 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3571 return Error(Loc, 3572 "unexpected token after '.bundle_lock' directive option"); 3573 AlignToEnd = true; 3574 } 3575 3576 Lex(); 3577 3578 getStreamer().EmitBundleLock(AlignToEnd); 3579 return false; 3580 } 3581 3582 /// parseDirectiveBundleLock 3583 /// ::= {.bundle_lock} 3584 bool AsmParser::parseDirectiveBundleUnlock() { 3585 checkForValidSection(); 3586 3587 if (getLexer().isNot(AsmToken::EndOfStatement)) 3588 return TokError("unexpected token in '.bundle_unlock' directive"); 3589 Lex(); 3590 3591 getStreamer().EmitBundleUnlock(); 3592 return false; 3593 } 3594 3595 /// parseDirectiveSpace 3596 /// ::= (.skip | .space) expression [ , expression ] 3597 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 3598 checkForValidSection(); 3599 3600 int64_t NumBytes; 3601 if (parseAbsoluteExpression(NumBytes)) 3602 return true; 3603 3604 int64_t FillExpr = 0; 3605 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3606 if (getLexer().isNot(AsmToken::Comma)) 3607 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 3608 Lex(); 3609 3610 if (parseAbsoluteExpression(FillExpr)) 3611 return true; 3612 3613 if (getLexer().isNot(AsmToken::EndOfStatement)) 3614 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 3615 } 3616 3617 Lex(); 3618 3619 if (NumBytes <= 0) 3620 return TokError("invalid number of bytes in '" + Twine(IDVal) + 3621 "' directive"); 3622 3623 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 3624 getStreamer().EmitFill(NumBytes, FillExpr); 3625 3626 return false; 3627 } 3628 3629 /// parseDirectiveLEB128 3630 /// ::= (.sleb128 | .uleb128) expression 3631 bool AsmParser::parseDirectiveLEB128(bool Signed) { 3632 checkForValidSection(); 3633 const MCExpr *Value; 3634 3635 if (parseExpression(Value)) 3636 return true; 3637 3638 if (getLexer().isNot(AsmToken::EndOfStatement)) 3639 return TokError("unexpected token in directive"); 3640 3641 if (Signed) 3642 getStreamer().EmitSLEB128Value(Value); 3643 else 3644 getStreamer().EmitULEB128Value(Value); 3645 3646 return false; 3647 } 3648 3649 /// parseDirectiveSymbolAttribute 3650 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 3651 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 3652 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3653 for (;;) { 3654 StringRef Name; 3655 SMLoc Loc = getTok().getLoc(); 3656 3657 if (parseIdentifier(Name)) 3658 return Error(Loc, "expected identifier in directive"); 3659 3660 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3661 3662 // Assembler local symbols don't make any sense here. Complain loudly. 3663 if (Sym->isTemporary()) 3664 return Error(Loc, "non-local symbol required in directive"); 3665 3666 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 3667 return Error(Loc, "unable to emit symbol attribute"); 3668 3669 if (getLexer().is(AsmToken::EndOfStatement)) 3670 break; 3671 3672 if (getLexer().isNot(AsmToken::Comma)) 3673 return TokError("unexpected token in directive"); 3674 Lex(); 3675 } 3676 } 3677 3678 Lex(); 3679 return false; 3680 } 3681 3682 /// parseDirectiveComm 3683 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 3684 bool AsmParser::parseDirectiveComm(bool IsLocal) { 3685 checkForValidSection(); 3686 3687 SMLoc IDLoc = getLexer().getLoc(); 3688 StringRef Name; 3689 if (parseIdentifier(Name)) 3690 return TokError("expected identifier in directive"); 3691 3692 // Handle the identifier as the key symbol. 3693 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3694 3695 if (getLexer().isNot(AsmToken::Comma)) 3696 return TokError("unexpected token in directive"); 3697 Lex(); 3698 3699 int64_t Size; 3700 SMLoc SizeLoc = getLexer().getLoc(); 3701 if (parseAbsoluteExpression(Size)) 3702 return true; 3703 3704 int64_t Pow2Alignment = 0; 3705 SMLoc Pow2AlignmentLoc; 3706 if (getLexer().is(AsmToken::Comma)) { 3707 Lex(); 3708 Pow2AlignmentLoc = getLexer().getLoc(); 3709 if (parseAbsoluteExpression(Pow2Alignment)) 3710 return true; 3711 3712 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 3713 if (IsLocal && LCOMM == LCOMM::NoAlignment) 3714 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 3715 3716 // If this target takes alignments in bytes (not log) validate and convert. 3717 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 3718 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 3719 if (!isPowerOf2_64(Pow2Alignment)) 3720 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 3721 Pow2Alignment = Log2_64(Pow2Alignment); 3722 } 3723 } 3724 3725 if (getLexer().isNot(AsmToken::EndOfStatement)) 3726 return TokError("unexpected token in '.comm' or '.lcomm' directive"); 3727 3728 Lex(); 3729 3730 // NOTE: a size of zero for a .comm should create a undefined symbol 3731 // but a size of .lcomm creates a bss symbol of size zero. 3732 if (Size < 0) 3733 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 3734 "be less than zero"); 3735 3736 // NOTE: The alignment in the directive is a power of 2 value, the assembler 3737 // may internally end up wanting an alignment in bytes. 3738 // FIXME: Diagnose overflow. 3739 if (Pow2Alignment < 0) 3740 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 3741 "alignment, can't be less than zero"); 3742 3743 if (!Sym->isUndefined()) 3744 return Error(IDLoc, "invalid symbol redefinition"); 3745 3746 // Create the Symbol as a common or local common with Size and Pow2Alignment 3747 if (IsLocal) { 3748 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 3749 return false; 3750 } 3751 3752 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 3753 return false; 3754 } 3755 3756 /// parseDirectiveAbort 3757 /// ::= .abort [... message ...] 3758 bool AsmParser::parseDirectiveAbort() { 3759 // FIXME: Use loc from directive. 3760 SMLoc Loc = getLexer().getLoc(); 3761 3762 StringRef Str = parseStringToEndOfStatement(); 3763 if (getLexer().isNot(AsmToken::EndOfStatement)) 3764 return TokError("unexpected token in '.abort' directive"); 3765 3766 Lex(); 3767 3768 if (Str.empty()) 3769 Error(Loc, ".abort detected. Assembly stopping."); 3770 else 3771 Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 3772 // FIXME: Actually abort assembly here. 3773 3774 return false; 3775 } 3776 3777 /// parseDirectiveInclude 3778 /// ::= .include "filename" 3779 bool AsmParser::parseDirectiveInclude() { 3780 if (getLexer().isNot(AsmToken::String)) 3781 return TokError("expected string in '.include' directive"); 3782 3783 // Allow the strings to have escaped octal character sequence. 3784 std::string Filename; 3785 if (parseEscapedString(Filename)) 3786 return true; 3787 SMLoc IncludeLoc = getLexer().getLoc(); 3788 Lex(); 3789 3790 if (getLexer().isNot(AsmToken::EndOfStatement)) 3791 return TokError("unexpected token in '.include' directive"); 3792 3793 // Attempt to switch the lexer to the included file before consuming the end 3794 // of statement to avoid losing it when we switch. 3795 if (enterIncludeFile(Filename)) { 3796 Error(IncludeLoc, "Could not find include file '" + Filename + "'"); 3797 return true; 3798 } 3799 3800 return false; 3801 } 3802 3803 /// parseDirectiveIncbin 3804 /// ::= .incbin "filename" 3805 bool AsmParser::parseDirectiveIncbin() { 3806 if (getLexer().isNot(AsmToken::String)) 3807 return TokError("expected string in '.incbin' directive"); 3808 3809 // Allow the strings to have escaped octal character sequence. 3810 std::string Filename; 3811 if (parseEscapedString(Filename)) 3812 return true; 3813 SMLoc IncbinLoc = getLexer().getLoc(); 3814 Lex(); 3815 3816 if (getLexer().isNot(AsmToken::EndOfStatement)) 3817 return TokError("unexpected token in '.incbin' directive"); 3818 3819 // Attempt to process the included file. 3820 if (processIncbinFile(Filename)) { 3821 Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 3822 return true; 3823 } 3824 3825 return false; 3826 } 3827 3828 /// parseDirectiveIf 3829 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 3830 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 3831 TheCondStack.push_back(TheCondState); 3832 TheCondState.TheCond = AsmCond::IfCond; 3833 if (TheCondState.Ignore) { 3834 eatToEndOfStatement(); 3835 } else { 3836 int64_t ExprValue; 3837 if (parseAbsoluteExpression(ExprValue)) 3838 return true; 3839 3840 if (getLexer().isNot(AsmToken::EndOfStatement)) 3841 return TokError("unexpected token in '.if' directive"); 3842 3843 Lex(); 3844 3845 switch (DirKind) { 3846 default: 3847 llvm_unreachable("unsupported directive"); 3848 case DK_IF: 3849 case DK_IFNE: 3850 break; 3851 case DK_IFEQ: 3852 ExprValue = ExprValue == 0; 3853 break; 3854 case DK_IFGE: 3855 ExprValue = ExprValue >= 0; 3856 break; 3857 case DK_IFGT: 3858 ExprValue = ExprValue > 0; 3859 break; 3860 case DK_IFLE: 3861 ExprValue = ExprValue <= 0; 3862 break; 3863 case DK_IFLT: 3864 ExprValue = ExprValue < 0; 3865 break; 3866 } 3867 3868 TheCondState.CondMet = ExprValue; 3869 TheCondState.Ignore = !TheCondState.CondMet; 3870 } 3871 3872 return false; 3873 } 3874 3875 /// parseDirectiveIfb 3876 /// ::= .ifb string 3877 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 3878 TheCondStack.push_back(TheCondState); 3879 TheCondState.TheCond = AsmCond::IfCond; 3880 3881 if (TheCondState.Ignore) { 3882 eatToEndOfStatement(); 3883 } else { 3884 StringRef Str = parseStringToEndOfStatement(); 3885 3886 if (getLexer().isNot(AsmToken::EndOfStatement)) 3887 return TokError("unexpected token in '.ifb' directive"); 3888 3889 Lex(); 3890 3891 TheCondState.CondMet = ExpectBlank == Str.empty(); 3892 TheCondState.Ignore = !TheCondState.CondMet; 3893 } 3894 3895 return false; 3896 } 3897 3898 /// parseDirectiveIfc 3899 /// ::= .ifc string1, string2 3900 /// ::= .ifnc string1, string2 3901 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 3902 TheCondStack.push_back(TheCondState); 3903 TheCondState.TheCond = AsmCond::IfCond; 3904 3905 if (TheCondState.Ignore) { 3906 eatToEndOfStatement(); 3907 } else { 3908 StringRef Str1 = parseStringToComma(); 3909 3910 if (getLexer().isNot(AsmToken::Comma)) 3911 return TokError("unexpected token in '.ifc' directive"); 3912 3913 Lex(); 3914 3915 StringRef Str2 = parseStringToEndOfStatement(); 3916 3917 if (getLexer().isNot(AsmToken::EndOfStatement)) 3918 return TokError("unexpected token in '.ifc' directive"); 3919 3920 Lex(); 3921 3922 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 3923 TheCondState.Ignore = !TheCondState.CondMet; 3924 } 3925 3926 return false; 3927 } 3928 3929 /// parseDirectiveIfeqs 3930 /// ::= .ifeqs string1, string2 3931 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc) { 3932 if (Lexer.isNot(AsmToken::String)) { 3933 TokError("expected string parameter for '.ifeqs' directive"); 3934 eatToEndOfStatement(); 3935 return true; 3936 } 3937 3938 StringRef String1 = getTok().getStringContents(); 3939 Lex(); 3940 3941 if (Lexer.isNot(AsmToken::Comma)) { 3942 TokError("expected comma after first string for '.ifeqs' directive"); 3943 eatToEndOfStatement(); 3944 return true; 3945 } 3946 3947 Lex(); 3948 3949 if (Lexer.isNot(AsmToken::String)) { 3950 TokError("expected string parameter for '.ifeqs' directive"); 3951 eatToEndOfStatement(); 3952 return true; 3953 } 3954 3955 StringRef String2 = getTok().getStringContents(); 3956 Lex(); 3957 3958 TheCondStack.push_back(TheCondState); 3959 TheCondState.TheCond = AsmCond::IfCond; 3960 TheCondState.CondMet = String1 == String2; 3961 TheCondState.Ignore = !TheCondState.CondMet; 3962 3963 return false; 3964 } 3965 3966 /// parseDirectiveIfdef 3967 /// ::= .ifdef symbol 3968 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 3969 StringRef Name; 3970 TheCondStack.push_back(TheCondState); 3971 TheCondState.TheCond = AsmCond::IfCond; 3972 3973 if (TheCondState.Ignore) { 3974 eatToEndOfStatement(); 3975 } else { 3976 if (parseIdentifier(Name)) 3977 return TokError("expected identifier after '.ifdef'"); 3978 3979 Lex(); 3980 3981 MCSymbol *Sym = getContext().LookupSymbol(Name); 3982 3983 if (expect_defined) 3984 TheCondState.CondMet = (Sym && !Sym->isUndefined()); 3985 else 3986 TheCondState.CondMet = (!Sym || Sym->isUndefined()); 3987 TheCondState.Ignore = !TheCondState.CondMet; 3988 } 3989 3990 return false; 3991 } 3992 3993 /// parseDirectiveElseIf 3994 /// ::= .elseif expression 3995 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 3996 if (TheCondState.TheCond != AsmCond::IfCond && 3997 TheCondState.TheCond != AsmCond::ElseIfCond) 3998 Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " 3999 " an .elseif"); 4000 TheCondState.TheCond = AsmCond::ElseIfCond; 4001 4002 bool LastIgnoreState = false; 4003 if (!TheCondStack.empty()) 4004 LastIgnoreState = TheCondStack.back().Ignore; 4005 if (LastIgnoreState || TheCondState.CondMet) { 4006 TheCondState.Ignore = true; 4007 eatToEndOfStatement(); 4008 } else { 4009 int64_t ExprValue; 4010 if (parseAbsoluteExpression(ExprValue)) 4011 return true; 4012 4013 if (getLexer().isNot(AsmToken::EndOfStatement)) 4014 return TokError("unexpected token in '.elseif' directive"); 4015 4016 Lex(); 4017 TheCondState.CondMet = ExprValue; 4018 TheCondState.Ignore = !TheCondState.CondMet; 4019 } 4020 4021 return false; 4022 } 4023 4024 /// parseDirectiveElse 4025 /// ::= .else 4026 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4027 if (getLexer().isNot(AsmToken::EndOfStatement)) 4028 return TokError("unexpected token in '.else' directive"); 4029 4030 Lex(); 4031 4032 if (TheCondState.TheCond != AsmCond::IfCond && 4033 TheCondState.TheCond != AsmCond::ElseIfCond) 4034 Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " 4035 ".elseif"); 4036 TheCondState.TheCond = AsmCond::ElseCond; 4037 bool LastIgnoreState = false; 4038 if (!TheCondStack.empty()) 4039 LastIgnoreState = TheCondStack.back().Ignore; 4040 if (LastIgnoreState || TheCondState.CondMet) 4041 TheCondState.Ignore = true; 4042 else 4043 TheCondState.Ignore = false; 4044 4045 return false; 4046 } 4047 4048 /// parseDirectiveEnd 4049 /// ::= .end 4050 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4051 if (getLexer().isNot(AsmToken::EndOfStatement)) 4052 return TokError("unexpected token in '.end' directive"); 4053 4054 Lex(); 4055 4056 while (Lexer.isNot(AsmToken::Eof)) 4057 Lex(); 4058 4059 return false; 4060 } 4061 4062 /// parseDirectiveError 4063 /// ::= .err 4064 /// ::= .error [string] 4065 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 4066 if (!TheCondStack.empty()) { 4067 if (TheCondStack.back().Ignore) { 4068 eatToEndOfStatement(); 4069 return false; 4070 } 4071 } 4072 4073 if (!WithMessage) 4074 return Error(L, ".err encountered"); 4075 4076 StringRef Message = ".error directive invoked in source file"; 4077 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4078 if (Lexer.isNot(AsmToken::String)) { 4079 TokError(".error argument must be a string"); 4080 eatToEndOfStatement(); 4081 return true; 4082 } 4083 4084 Message = getTok().getStringContents(); 4085 Lex(); 4086 } 4087 4088 Error(L, Message); 4089 return true; 4090 } 4091 4092 /// parseDirectiveWarning 4093 /// ::= .warning [string] 4094 bool AsmParser::parseDirectiveWarning(SMLoc L) { 4095 if (!TheCondStack.empty()) { 4096 if (TheCondStack.back().Ignore) { 4097 eatToEndOfStatement(); 4098 return false; 4099 } 4100 } 4101 4102 StringRef Message = ".warning directive invoked in source file"; 4103 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4104 if (Lexer.isNot(AsmToken::String)) { 4105 TokError(".warning argument must be a string"); 4106 eatToEndOfStatement(); 4107 return true; 4108 } 4109 4110 Message = getTok().getStringContents(); 4111 Lex(); 4112 } 4113 4114 Warning(L, Message); 4115 return false; 4116 } 4117 4118 /// parseDirectiveEndIf 4119 /// ::= .endif 4120 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 4121 if (getLexer().isNot(AsmToken::EndOfStatement)) 4122 return TokError("unexpected token in '.endif' directive"); 4123 4124 Lex(); 4125 4126 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 4127 Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " 4128 ".else"); 4129 if (!TheCondStack.empty()) { 4130 TheCondState = TheCondStack.back(); 4131 TheCondStack.pop_back(); 4132 } 4133 4134 return false; 4135 } 4136 4137 void AsmParser::initializeDirectiveKindMap() { 4138 DirectiveKindMap[".set"] = DK_SET; 4139 DirectiveKindMap[".equ"] = DK_EQU; 4140 DirectiveKindMap[".equiv"] = DK_EQUIV; 4141 DirectiveKindMap[".ascii"] = DK_ASCII; 4142 DirectiveKindMap[".asciz"] = DK_ASCIZ; 4143 DirectiveKindMap[".string"] = DK_STRING; 4144 DirectiveKindMap[".byte"] = DK_BYTE; 4145 DirectiveKindMap[".short"] = DK_SHORT; 4146 DirectiveKindMap[".value"] = DK_VALUE; 4147 DirectiveKindMap[".2byte"] = DK_2BYTE; 4148 DirectiveKindMap[".long"] = DK_LONG; 4149 DirectiveKindMap[".int"] = DK_INT; 4150 DirectiveKindMap[".4byte"] = DK_4BYTE; 4151 DirectiveKindMap[".quad"] = DK_QUAD; 4152 DirectiveKindMap[".8byte"] = DK_8BYTE; 4153 DirectiveKindMap[".octa"] = DK_OCTA; 4154 DirectiveKindMap[".single"] = DK_SINGLE; 4155 DirectiveKindMap[".float"] = DK_FLOAT; 4156 DirectiveKindMap[".double"] = DK_DOUBLE; 4157 DirectiveKindMap[".align"] = DK_ALIGN; 4158 DirectiveKindMap[".align32"] = DK_ALIGN32; 4159 DirectiveKindMap[".balign"] = DK_BALIGN; 4160 DirectiveKindMap[".balignw"] = DK_BALIGNW; 4161 DirectiveKindMap[".balignl"] = DK_BALIGNL; 4162 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 4163 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 4164 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 4165 DirectiveKindMap[".org"] = DK_ORG; 4166 DirectiveKindMap[".fill"] = DK_FILL; 4167 DirectiveKindMap[".zero"] = DK_ZERO; 4168 DirectiveKindMap[".extern"] = DK_EXTERN; 4169 DirectiveKindMap[".globl"] = DK_GLOBL; 4170 DirectiveKindMap[".global"] = DK_GLOBAL; 4171 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 4172 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 4173 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 4174 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 4175 DirectiveKindMap[".reference"] = DK_REFERENCE; 4176 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 4177 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 4178 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 4179 DirectiveKindMap[".comm"] = DK_COMM; 4180 DirectiveKindMap[".common"] = DK_COMMON; 4181 DirectiveKindMap[".lcomm"] = DK_LCOMM; 4182 DirectiveKindMap[".abort"] = DK_ABORT; 4183 DirectiveKindMap[".include"] = DK_INCLUDE; 4184 DirectiveKindMap[".incbin"] = DK_INCBIN; 4185 DirectiveKindMap[".code16"] = DK_CODE16; 4186 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 4187 DirectiveKindMap[".rept"] = DK_REPT; 4188 DirectiveKindMap[".rep"] = DK_REPT; 4189 DirectiveKindMap[".irp"] = DK_IRP; 4190 DirectiveKindMap[".irpc"] = DK_IRPC; 4191 DirectiveKindMap[".endr"] = DK_ENDR; 4192 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 4193 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 4194 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 4195 DirectiveKindMap[".if"] = DK_IF; 4196 DirectiveKindMap[".ifeq"] = DK_IFEQ; 4197 DirectiveKindMap[".ifge"] = DK_IFGE; 4198 DirectiveKindMap[".ifgt"] = DK_IFGT; 4199 DirectiveKindMap[".ifle"] = DK_IFLE; 4200 DirectiveKindMap[".iflt"] = DK_IFLT; 4201 DirectiveKindMap[".ifne"] = DK_IFNE; 4202 DirectiveKindMap[".ifb"] = DK_IFB; 4203 DirectiveKindMap[".ifnb"] = DK_IFNB; 4204 DirectiveKindMap[".ifc"] = DK_IFC; 4205 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 4206 DirectiveKindMap[".ifnc"] = DK_IFNC; 4207 DirectiveKindMap[".ifdef"] = DK_IFDEF; 4208 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 4209 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 4210 DirectiveKindMap[".elseif"] = DK_ELSEIF; 4211 DirectiveKindMap[".else"] = DK_ELSE; 4212 DirectiveKindMap[".end"] = DK_END; 4213 DirectiveKindMap[".endif"] = DK_ENDIF; 4214 DirectiveKindMap[".skip"] = DK_SKIP; 4215 DirectiveKindMap[".space"] = DK_SPACE; 4216 DirectiveKindMap[".file"] = DK_FILE; 4217 DirectiveKindMap[".line"] = DK_LINE; 4218 DirectiveKindMap[".loc"] = DK_LOC; 4219 DirectiveKindMap[".stabs"] = DK_STABS; 4220 DirectiveKindMap[".sleb128"] = DK_SLEB128; 4221 DirectiveKindMap[".uleb128"] = DK_ULEB128; 4222 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 4223 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 4224 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 4225 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 4226 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 4227 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 4228 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 4229 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 4230 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 4231 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 4232 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 4233 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 4234 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 4235 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 4236 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 4237 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 4238 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 4239 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 4240 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 4241 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 4242 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 4243 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 4244 DirectiveKindMap[".macro"] = DK_MACRO; 4245 DirectiveKindMap[".exitm"] = DK_EXITM; 4246 DirectiveKindMap[".endm"] = DK_ENDM; 4247 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 4248 DirectiveKindMap[".purgem"] = DK_PURGEM; 4249 DirectiveKindMap[".err"] = DK_ERR; 4250 DirectiveKindMap[".error"] = DK_ERROR; 4251 DirectiveKindMap[".warning"] = DK_WARNING; 4252 } 4253 4254 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 4255 AsmToken EndToken, StartToken = getTok(); 4256 4257 unsigned NestLevel = 0; 4258 for (;;) { 4259 // Check whether we have reached the end of the file. 4260 if (getLexer().is(AsmToken::Eof)) { 4261 Error(DirectiveLoc, "no matching '.endr' in definition"); 4262 return nullptr; 4263 } 4264 4265 if (Lexer.is(AsmToken::Identifier) && 4266 (getTok().getIdentifier() == ".rept")) { 4267 ++NestLevel; 4268 } 4269 4270 // Otherwise, check whether we have reached the .endr. 4271 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 4272 if (NestLevel == 0) { 4273 EndToken = getTok(); 4274 Lex(); 4275 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4276 TokError("unexpected token in '.endr' directive"); 4277 return nullptr; 4278 } 4279 break; 4280 } 4281 --NestLevel; 4282 } 4283 4284 // Otherwise, scan till the end of the statement. 4285 eatToEndOfStatement(); 4286 } 4287 4288 const char *BodyStart = StartToken.getLoc().getPointer(); 4289 const char *BodyEnd = EndToken.getLoc().getPointer(); 4290 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4291 4292 // We Are Anonymous. 4293 MacroLikeBodies.push_back( 4294 MCAsmMacro(StringRef(), Body, MCAsmMacroParameters())); 4295 return &MacroLikeBodies.back(); 4296 } 4297 4298 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 4299 raw_svector_ostream &OS) { 4300 OS << ".endr\n"; 4301 4302 std::unique_ptr<MemoryBuffer> Instantiation = 4303 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 4304 4305 // Create the macro instantiation object and add to the current macro 4306 // instantiation stack. 4307 MacroInstantiation *MI = new MacroInstantiation( 4308 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 4309 ActiveMacros.push_back(MI); 4310 4311 // Jump to the macro instantiation and prime the lexer. 4312 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 4313 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 4314 Lex(); 4315 } 4316 4317 /// parseDirectiveRept 4318 /// ::= .rep | .rept count 4319 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 4320 const MCExpr *CountExpr; 4321 SMLoc CountLoc = getTok().getLoc(); 4322 if (parseExpression(CountExpr)) 4323 return true; 4324 4325 int64_t Count; 4326 if (!CountExpr->EvaluateAsAbsolute(Count)) { 4327 eatToEndOfStatement(); 4328 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 4329 } 4330 4331 if (Count < 0) 4332 return Error(CountLoc, "Count is negative"); 4333 4334 if (Lexer.isNot(AsmToken::EndOfStatement)) 4335 return TokError("unexpected token in '" + Dir + "' directive"); 4336 4337 // Eat the end of statement. 4338 Lex(); 4339 4340 // Lex the rept definition. 4341 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4342 if (!M) 4343 return true; 4344 4345 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4346 // to hold the macro body with substitutions. 4347 SmallString<256> Buf; 4348 raw_svector_ostream OS(Buf); 4349 while (Count--) { 4350 if (expandMacro(OS, M->Body, None, None, getTok().getLoc())) 4351 return true; 4352 } 4353 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4354 4355 return false; 4356 } 4357 4358 /// parseDirectiveIrp 4359 /// ::= .irp symbol,values 4360 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 4361 MCAsmMacroParameter Parameter; 4362 4363 if (parseIdentifier(Parameter.Name)) 4364 return TokError("expected identifier in '.irp' directive"); 4365 4366 if (Lexer.isNot(AsmToken::Comma)) 4367 return TokError("expected comma in '.irp' directive"); 4368 4369 Lex(); 4370 4371 MCAsmMacroArguments A; 4372 if (parseMacroArguments(nullptr, A)) 4373 return true; 4374 4375 // Eat the end of statement. 4376 Lex(); 4377 4378 // Lex the irp definition. 4379 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4380 if (!M) 4381 return true; 4382 4383 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4384 // to hold the macro body with substitutions. 4385 SmallString<256> Buf; 4386 raw_svector_ostream OS(Buf); 4387 4388 for (MCAsmMacroArguments::iterator i = A.begin(), e = A.end(); i != e; ++i) { 4389 if (expandMacro(OS, M->Body, Parameter, *i, getTok().getLoc())) 4390 return true; 4391 } 4392 4393 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4394 4395 return false; 4396 } 4397 4398 /// parseDirectiveIrpc 4399 /// ::= .irpc symbol,values 4400 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 4401 MCAsmMacroParameter Parameter; 4402 4403 if (parseIdentifier(Parameter.Name)) 4404 return TokError("expected identifier in '.irpc' directive"); 4405 4406 if (Lexer.isNot(AsmToken::Comma)) 4407 return TokError("expected comma in '.irpc' directive"); 4408 4409 Lex(); 4410 4411 MCAsmMacroArguments A; 4412 if (parseMacroArguments(nullptr, A)) 4413 return true; 4414 4415 if (A.size() != 1 || A.front().size() != 1) 4416 return TokError("unexpected token in '.irpc' directive"); 4417 4418 // Eat the end of statement. 4419 Lex(); 4420 4421 // Lex the irpc definition. 4422 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4423 if (!M) 4424 return true; 4425 4426 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4427 // to hold the macro body with substitutions. 4428 SmallString<256> Buf; 4429 raw_svector_ostream OS(Buf); 4430 4431 StringRef Values = A.front().front().getString(); 4432 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 4433 MCAsmMacroArgument Arg; 4434 Arg.push_back(AsmToken(AsmToken::Identifier, Values.slice(I, I + 1))); 4435 4436 if (expandMacro(OS, M->Body, Parameter, Arg, getTok().getLoc())) 4437 return true; 4438 } 4439 4440 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4441 4442 return false; 4443 } 4444 4445 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 4446 if (ActiveMacros.empty()) 4447 return TokError("unmatched '.endr' directive"); 4448 4449 // The only .repl that should get here are the ones created by 4450 // instantiateMacroLikeBody. 4451 assert(getLexer().is(AsmToken::EndOfStatement)); 4452 4453 handleMacroExit(); 4454 return false; 4455 } 4456 4457 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 4458 size_t Len) { 4459 const MCExpr *Value; 4460 SMLoc ExprLoc = getLexer().getLoc(); 4461 if (parseExpression(Value)) 4462 return true; 4463 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4464 if (!MCE) 4465 return Error(ExprLoc, "unexpected expression in _emit"); 4466 uint64_t IntValue = MCE->getValue(); 4467 if (!isUIntN(8, IntValue) && !isIntN(8, IntValue)) 4468 return Error(ExprLoc, "literal value out of range for directive"); 4469 4470 Info.AsmRewrites->push_back(AsmRewrite(AOK_Emit, IDLoc, Len)); 4471 return false; 4472 } 4473 4474 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 4475 const MCExpr *Value; 4476 SMLoc ExprLoc = getLexer().getLoc(); 4477 if (parseExpression(Value)) 4478 return true; 4479 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4480 if (!MCE) 4481 return Error(ExprLoc, "unexpected expression in align"); 4482 uint64_t IntValue = MCE->getValue(); 4483 if (!isPowerOf2_64(IntValue)) 4484 return Error(ExprLoc, "literal value not a power of two greater then zero"); 4485 4486 Info.AsmRewrites->push_back( 4487 AsmRewrite(AOK_Align, IDLoc, 5, Log2_64(IntValue))); 4488 return false; 4489 } 4490 4491 // We are comparing pointers, but the pointers are relative to a single string. 4492 // Thus, this should always be deterministic. 4493 static int rewritesSort(const AsmRewrite *AsmRewriteA, 4494 const AsmRewrite *AsmRewriteB) { 4495 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 4496 return -1; 4497 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 4498 return 1; 4499 4500 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 4501 // rewrite to the same location. Make sure the SizeDirective rewrite is 4502 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 4503 // ensures the sort algorithm is stable. 4504 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 4505 AsmRewritePrecedence[AsmRewriteB->Kind]) 4506 return -1; 4507 4508 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 4509 AsmRewritePrecedence[AsmRewriteB->Kind]) 4510 return 1; 4511 llvm_unreachable("Unstable rewrite sort."); 4512 } 4513 4514 bool AsmParser::parseMSInlineAsm( 4515 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 4516 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool> > &OpDecls, 4517 SmallVectorImpl<std::string> &Constraints, 4518 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 4519 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 4520 SmallVector<void *, 4> InputDecls; 4521 SmallVector<void *, 4> OutputDecls; 4522 SmallVector<bool, 4> InputDeclsAddressOf; 4523 SmallVector<bool, 4> OutputDeclsAddressOf; 4524 SmallVector<std::string, 4> InputConstraints; 4525 SmallVector<std::string, 4> OutputConstraints; 4526 SmallVector<unsigned, 4> ClobberRegs; 4527 4528 SmallVector<AsmRewrite, 4> AsmStrRewrites; 4529 4530 // Prime the lexer. 4531 Lex(); 4532 4533 // While we have input, parse each statement. 4534 unsigned InputIdx = 0; 4535 unsigned OutputIdx = 0; 4536 while (getLexer().isNot(AsmToken::Eof)) { 4537 ParseStatementInfo Info(&AsmStrRewrites); 4538 if (parseStatement(Info, &SI)) 4539 return true; 4540 4541 if (Info.ParseError) 4542 return true; 4543 4544 if (Info.Opcode == ~0U) 4545 continue; 4546 4547 const MCInstrDesc &Desc = MII->get(Info.Opcode); 4548 4549 // Build the list of clobbers, outputs and inputs. 4550 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 4551 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 4552 4553 // Immediate. 4554 if (Operand.isImm()) 4555 continue; 4556 4557 // Register operand. 4558 if (Operand.isReg() && !Operand.needAddressOf() && 4559 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 4560 unsigned NumDefs = Desc.getNumDefs(); 4561 // Clobber. 4562 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 4563 ClobberRegs.push_back(Operand.getReg()); 4564 continue; 4565 } 4566 4567 // Expr/Input or Output. 4568 StringRef SymName = Operand.getSymName(); 4569 if (SymName.empty()) 4570 continue; 4571 4572 void *OpDecl = Operand.getOpDecl(); 4573 if (!OpDecl) 4574 continue; 4575 4576 bool isOutput = (i == 1) && Desc.mayStore(); 4577 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 4578 if (isOutput) { 4579 ++InputIdx; 4580 OutputDecls.push_back(OpDecl); 4581 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 4582 OutputConstraints.push_back('=' + Operand.getConstraint().str()); 4583 AsmStrRewrites.push_back(AsmRewrite(AOK_Output, Start, SymName.size())); 4584 } else { 4585 InputDecls.push_back(OpDecl); 4586 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 4587 InputConstraints.push_back(Operand.getConstraint().str()); 4588 AsmStrRewrites.push_back(AsmRewrite(AOK_Input, Start, SymName.size())); 4589 } 4590 } 4591 4592 // Consider implicit defs to be clobbers. Think of cpuid and push. 4593 ArrayRef<uint16_t> ImpDefs(Desc.getImplicitDefs(), 4594 Desc.getNumImplicitDefs()); 4595 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 4596 } 4597 4598 // Set the number of Outputs and Inputs. 4599 NumOutputs = OutputDecls.size(); 4600 NumInputs = InputDecls.size(); 4601 4602 // Set the unique clobbers. 4603 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 4604 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 4605 ClobberRegs.end()); 4606 Clobbers.assign(ClobberRegs.size(), std::string()); 4607 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 4608 raw_string_ostream OS(Clobbers[I]); 4609 IP->printRegName(OS, ClobberRegs[I]); 4610 } 4611 4612 // Merge the various outputs and inputs. Output are expected first. 4613 if (NumOutputs || NumInputs) { 4614 unsigned NumExprs = NumOutputs + NumInputs; 4615 OpDecls.resize(NumExprs); 4616 Constraints.resize(NumExprs); 4617 for (unsigned i = 0; i < NumOutputs; ++i) { 4618 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 4619 Constraints[i] = OutputConstraints[i]; 4620 } 4621 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 4622 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 4623 Constraints[j] = InputConstraints[i]; 4624 } 4625 } 4626 4627 // Build the IR assembly string. 4628 std::string AsmStringIR; 4629 raw_string_ostream OS(AsmStringIR); 4630 StringRef ASMString = 4631 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 4632 const char *AsmStart = ASMString.begin(); 4633 const char *AsmEnd = ASMString.end(); 4634 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 4635 for (const AsmRewrite &AR : AsmStrRewrites) { 4636 AsmRewriteKind Kind = AR.Kind; 4637 if (Kind == AOK_Delete) 4638 continue; 4639 4640 const char *Loc = AR.Loc.getPointer(); 4641 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 4642 4643 // Emit everything up to the immediate/expression. 4644 if (unsigned Len = Loc - AsmStart) 4645 OS << StringRef(AsmStart, Len); 4646 4647 // Skip the original expression. 4648 if (Kind == AOK_Skip) { 4649 AsmStart = Loc + AR.Len; 4650 continue; 4651 } 4652 4653 unsigned AdditionalSkip = 0; 4654 // Rewrite expressions in $N notation. 4655 switch (Kind) { 4656 default: 4657 break; 4658 case AOK_Imm: 4659 OS << "$$" << AR.Val; 4660 break; 4661 case AOK_ImmPrefix: 4662 OS << "$$"; 4663 break; 4664 case AOK_Label: 4665 OS << Ctx.getAsmInfo()->getPrivateGlobalPrefix() << AR.Label; 4666 break; 4667 case AOK_Input: 4668 OS << '$' << InputIdx++; 4669 break; 4670 case AOK_Output: 4671 OS << '$' << OutputIdx++; 4672 break; 4673 case AOK_SizeDirective: 4674 switch (AR.Val) { 4675 default: break; 4676 case 8: OS << "byte ptr "; break; 4677 case 16: OS << "word ptr "; break; 4678 case 32: OS << "dword ptr "; break; 4679 case 64: OS << "qword ptr "; break; 4680 case 80: OS << "xword ptr "; break; 4681 case 128: OS << "xmmword ptr "; break; 4682 case 256: OS << "ymmword ptr "; break; 4683 } 4684 break; 4685 case AOK_Emit: 4686 OS << ".byte"; 4687 break; 4688 case AOK_Align: { 4689 unsigned Val = AR.Val; 4690 OS << ".align " << Val; 4691 4692 // Skip the original immediate. 4693 assert(Val < 10 && "Expected alignment less then 2^10."); 4694 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 4695 break; 4696 } 4697 case AOK_DotOperator: 4698 // Insert the dot if the user omitted it. 4699 OS.flush(); 4700 if (AsmStringIR.back() != '.') 4701 OS << '.'; 4702 OS << AR.Val; 4703 break; 4704 } 4705 4706 // Skip the original expression. 4707 AsmStart = Loc + AR.Len + AdditionalSkip; 4708 } 4709 4710 // Emit the remainder of the asm string. 4711 if (AsmStart != AsmEnd) 4712 OS << StringRef(AsmStart, AsmEnd - AsmStart); 4713 4714 AsmString = OS.str(); 4715 return false; 4716 } 4717 4718 /// \brief Create an MCAsmParser instance. 4719 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 4720 MCStreamer &Out, const MCAsmInfo &MAI) { 4721 return new AsmParser(SM, C, Out, MAI); 4722 } 4723