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 1302 Sym->redefineIfPossible(); 1303 1304 if (!Sym->isUndefined() || Sym->isVariable()) 1305 return Error(IDLoc, "invalid symbol redefinition"); 1306 1307 // Emit the label. 1308 if (!ParsingInlineAsm) 1309 Out.EmitLabel(Sym); 1310 1311 // If we are generating dwarf for assembly source files then gather the 1312 // info to make a dwarf label entry for this label if needed. 1313 if (getContext().getGenDwarfForAssembly()) 1314 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1315 IDLoc); 1316 1317 getTargetParser().onLabelParsed(Sym); 1318 1319 // Consume any end of statement token, if present, to avoid spurious 1320 // AddBlankLine calls(). 1321 if (Lexer.is(AsmToken::EndOfStatement)) { 1322 Lex(); 1323 if (Lexer.is(AsmToken::Eof)) 1324 return false; 1325 } 1326 1327 return false; 1328 } 1329 1330 case AsmToken::Equal: 1331 // identifier '=' ... -> assignment statement 1332 Lex(); 1333 1334 return parseAssignment(IDVal, true); 1335 1336 default: // Normal instruction or directive. 1337 break; 1338 } 1339 1340 // If macros are enabled, check to see if this is a macro instantiation. 1341 if (areMacrosEnabled()) 1342 if (const MCAsmMacro *M = lookupMacro(IDVal)) { 1343 return handleMacroEntry(M, IDLoc); 1344 } 1345 1346 // Otherwise, we have a normal instruction or directive. 1347 1348 // Directives start with "." 1349 if (IDVal[0] == '.' && IDVal != ".") { 1350 // There are several entities interested in parsing directives: 1351 // 1352 // 1. The target-specific assembly parser. Some directives are target 1353 // specific or may potentially behave differently on certain targets. 1354 // 2. Asm parser extensions. For example, platform-specific parsers 1355 // (like the ELF parser) register themselves as extensions. 1356 // 3. The generic directive parser implemented by this class. These are 1357 // all the directives that behave in a target and platform independent 1358 // manner, or at least have a default behavior that's shared between 1359 // all targets and platforms. 1360 1361 // First query the target-specific parser. It will return 'true' if it 1362 // isn't interested in this directive. 1363 if (!getTargetParser().ParseDirective(ID)) 1364 return false; 1365 1366 // Next, check the extension directive map to see if any extension has 1367 // registered itself to parse this directive. 1368 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1369 ExtensionDirectiveMap.lookup(IDVal); 1370 if (Handler.first) 1371 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1372 1373 // Finally, if no one else is interested in this directive, it must be 1374 // generic and familiar to this class. 1375 switch (DirKind) { 1376 default: 1377 break; 1378 case DK_SET: 1379 case DK_EQU: 1380 return parseDirectiveSet(IDVal, true); 1381 case DK_EQUIV: 1382 return parseDirectiveSet(IDVal, false); 1383 case DK_ASCII: 1384 return parseDirectiveAscii(IDVal, false); 1385 case DK_ASCIZ: 1386 case DK_STRING: 1387 return parseDirectiveAscii(IDVal, true); 1388 case DK_BYTE: 1389 return parseDirectiveValue(1); 1390 case DK_SHORT: 1391 case DK_VALUE: 1392 case DK_2BYTE: 1393 return parseDirectiveValue(2); 1394 case DK_LONG: 1395 case DK_INT: 1396 case DK_4BYTE: 1397 return parseDirectiveValue(4); 1398 case DK_QUAD: 1399 case DK_8BYTE: 1400 return parseDirectiveValue(8); 1401 case DK_OCTA: 1402 return parseDirectiveOctaValue(); 1403 case DK_SINGLE: 1404 case DK_FLOAT: 1405 return parseDirectiveRealValue(APFloat::IEEEsingle); 1406 case DK_DOUBLE: 1407 return parseDirectiveRealValue(APFloat::IEEEdouble); 1408 case DK_ALIGN: { 1409 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1410 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1411 } 1412 case DK_ALIGN32: { 1413 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1414 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1415 } 1416 case DK_BALIGN: 1417 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1418 case DK_BALIGNW: 1419 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1420 case DK_BALIGNL: 1421 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1422 case DK_P2ALIGN: 1423 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1424 case DK_P2ALIGNW: 1425 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1426 case DK_P2ALIGNL: 1427 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1428 case DK_ORG: 1429 return parseDirectiveOrg(); 1430 case DK_FILL: 1431 return parseDirectiveFill(); 1432 case DK_ZERO: 1433 return parseDirectiveZero(); 1434 case DK_EXTERN: 1435 eatToEndOfStatement(); // .extern is the default, ignore it. 1436 return false; 1437 case DK_GLOBL: 1438 case DK_GLOBAL: 1439 return parseDirectiveSymbolAttribute(MCSA_Global); 1440 case DK_LAZY_REFERENCE: 1441 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1442 case DK_NO_DEAD_STRIP: 1443 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1444 case DK_SYMBOL_RESOLVER: 1445 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1446 case DK_PRIVATE_EXTERN: 1447 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1448 case DK_REFERENCE: 1449 return parseDirectiveSymbolAttribute(MCSA_Reference); 1450 case DK_WEAK_DEFINITION: 1451 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1452 case DK_WEAK_REFERENCE: 1453 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1454 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1455 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1456 case DK_COMM: 1457 case DK_COMMON: 1458 return parseDirectiveComm(/*IsLocal=*/false); 1459 case DK_LCOMM: 1460 return parseDirectiveComm(/*IsLocal=*/true); 1461 case DK_ABORT: 1462 return parseDirectiveAbort(); 1463 case DK_INCLUDE: 1464 return parseDirectiveInclude(); 1465 case DK_INCBIN: 1466 return parseDirectiveIncbin(); 1467 case DK_CODE16: 1468 case DK_CODE16GCC: 1469 return TokError(Twine(IDVal) + " not supported yet"); 1470 case DK_REPT: 1471 return parseDirectiveRept(IDLoc, IDVal); 1472 case DK_IRP: 1473 return parseDirectiveIrp(IDLoc); 1474 case DK_IRPC: 1475 return parseDirectiveIrpc(IDLoc); 1476 case DK_ENDR: 1477 return parseDirectiveEndr(IDLoc); 1478 case DK_BUNDLE_ALIGN_MODE: 1479 return parseDirectiveBundleAlignMode(); 1480 case DK_BUNDLE_LOCK: 1481 return parseDirectiveBundleLock(); 1482 case DK_BUNDLE_UNLOCK: 1483 return parseDirectiveBundleUnlock(); 1484 case DK_SLEB128: 1485 return parseDirectiveLEB128(true); 1486 case DK_ULEB128: 1487 return parseDirectiveLEB128(false); 1488 case DK_SPACE: 1489 case DK_SKIP: 1490 return parseDirectiveSpace(IDVal); 1491 case DK_FILE: 1492 return parseDirectiveFile(IDLoc); 1493 case DK_LINE: 1494 return parseDirectiveLine(); 1495 case DK_LOC: 1496 return parseDirectiveLoc(); 1497 case DK_STABS: 1498 return parseDirectiveStabs(); 1499 case DK_CFI_SECTIONS: 1500 return parseDirectiveCFISections(); 1501 case DK_CFI_STARTPROC: 1502 return parseDirectiveCFIStartProc(); 1503 case DK_CFI_ENDPROC: 1504 return parseDirectiveCFIEndProc(); 1505 case DK_CFI_DEF_CFA: 1506 return parseDirectiveCFIDefCfa(IDLoc); 1507 case DK_CFI_DEF_CFA_OFFSET: 1508 return parseDirectiveCFIDefCfaOffset(); 1509 case DK_CFI_ADJUST_CFA_OFFSET: 1510 return parseDirectiveCFIAdjustCfaOffset(); 1511 case DK_CFI_DEF_CFA_REGISTER: 1512 return parseDirectiveCFIDefCfaRegister(IDLoc); 1513 case DK_CFI_OFFSET: 1514 return parseDirectiveCFIOffset(IDLoc); 1515 case DK_CFI_REL_OFFSET: 1516 return parseDirectiveCFIRelOffset(IDLoc); 1517 case DK_CFI_PERSONALITY: 1518 return parseDirectiveCFIPersonalityOrLsda(true); 1519 case DK_CFI_LSDA: 1520 return parseDirectiveCFIPersonalityOrLsda(false); 1521 case DK_CFI_REMEMBER_STATE: 1522 return parseDirectiveCFIRememberState(); 1523 case DK_CFI_RESTORE_STATE: 1524 return parseDirectiveCFIRestoreState(); 1525 case DK_CFI_SAME_VALUE: 1526 return parseDirectiveCFISameValue(IDLoc); 1527 case DK_CFI_RESTORE: 1528 return parseDirectiveCFIRestore(IDLoc); 1529 case DK_CFI_ESCAPE: 1530 return parseDirectiveCFIEscape(); 1531 case DK_CFI_SIGNAL_FRAME: 1532 return parseDirectiveCFISignalFrame(); 1533 case DK_CFI_UNDEFINED: 1534 return parseDirectiveCFIUndefined(IDLoc); 1535 case DK_CFI_REGISTER: 1536 return parseDirectiveCFIRegister(IDLoc); 1537 case DK_CFI_WINDOW_SAVE: 1538 return parseDirectiveCFIWindowSave(); 1539 case DK_MACROS_ON: 1540 case DK_MACROS_OFF: 1541 return parseDirectiveMacrosOnOff(IDVal); 1542 case DK_MACRO: 1543 return parseDirectiveMacro(IDLoc); 1544 case DK_EXITM: 1545 return parseDirectiveExitMacro(IDVal); 1546 case DK_ENDM: 1547 case DK_ENDMACRO: 1548 return parseDirectiveEndMacro(IDVal); 1549 case DK_PURGEM: 1550 return parseDirectivePurgeMacro(IDLoc); 1551 case DK_END: 1552 return parseDirectiveEnd(IDLoc); 1553 case DK_ERR: 1554 return parseDirectiveError(IDLoc, false); 1555 case DK_ERROR: 1556 return parseDirectiveError(IDLoc, true); 1557 case DK_WARNING: 1558 return parseDirectiveWarning(IDLoc); 1559 } 1560 1561 return Error(IDLoc, "unknown directive"); 1562 } 1563 1564 // __asm _emit or __asm __emit 1565 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 1566 IDVal == "_EMIT" || IDVal == "__EMIT")) 1567 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 1568 1569 // __asm align 1570 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 1571 return parseDirectiveMSAlign(IDLoc, Info); 1572 1573 checkForValidSection(); 1574 1575 // Canonicalize the opcode to lower case. 1576 std::string OpcodeStr = IDVal.lower(); 1577 ParseInstructionInfo IInfo(Info.AsmRewrites); 1578 bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, IDLoc, 1579 Info.ParsedOperands); 1580 Info.ParseError = HadError; 1581 1582 // Dump the parsed representation, if requested. 1583 if (getShowParsedOperands()) { 1584 SmallString<256> Str; 1585 raw_svector_ostream OS(Str); 1586 OS << "parsed instruction: ["; 1587 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 1588 if (i != 0) 1589 OS << ", "; 1590 Info.ParsedOperands[i]->print(OS); 1591 } 1592 OS << "]"; 1593 1594 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 1595 } 1596 1597 // If we are generating dwarf for the current section then generate a .loc 1598 // directive for the instruction. 1599 if (!HadError && getContext().getGenDwarfForAssembly() && 1600 getContext().getGenDwarfSectionSyms().count( 1601 getStreamer().getCurrentSection().first)) { 1602 unsigned Line; 1603 if (ActiveMacros.empty()) 1604 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 1605 else 1606 Line = SrcMgr.FindLineNumber(ActiveMacros.back()->InstantiationLoc, 1607 ActiveMacros.back()->ExitBuffer); 1608 1609 // If we previously parsed a cpp hash file line comment then make sure the 1610 // current Dwarf File is for the CppHashFilename if not then emit the 1611 // Dwarf File table for it and adjust the line number for the .loc. 1612 if (CppHashFilename.size()) { 1613 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 1614 0, StringRef(), CppHashFilename); 1615 getContext().setGenDwarfFileNumber(FileNumber); 1616 1617 // Since SrcMgr.FindLineNumber() is slow and messes up the SourceMgr's 1618 // cache with the different Loc from the call above we save the last 1619 // info we queried here with SrcMgr.FindLineNumber(). 1620 unsigned CppHashLocLineNo; 1621 if (LastQueryIDLoc == CppHashLoc && LastQueryBuffer == CppHashBuf) 1622 CppHashLocLineNo = LastQueryLine; 1623 else { 1624 CppHashLocLineNo = SrcMgr.FindLineNumber(CppHashLoc, CppHashBuf); 1625 LastQueryLine = CppHashLocLineNo; 1626 LastQueryIDLoc = CppHashLoc; 1627 LastQueryBuffer = CppHashBuf; 1628 } 1629 Line = CppHashLineNumber - 1 + (Line - CppHashLocLineNo); 1630 } 1631 1632 getStreamer().EmitDwarfLocDirective( 1633 getContext().getGenDwarfFileNumber(), Line, 0, 1634 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 1635 StringRef()); 1636 } 1637 1638 // If parsing succeeded, match the instruction. 1639 if (!HadError) { 1640 uint64_t ErrorInfo; 1641 getTargetParser().MatchAndEmitInstruction(IDLoc, Info.Opcode, 1642 Info.ParsedOperands, Out, 1643 ErrorInfo, ParsingInlineAsm); 1644 } 1645 1646 // Don't skip the rest of the line, the instruction parser is responsible for 1647 // that. 1648 return false; 1649 } 1650 1651 /// eatToEndOfLine uses the Lexer to eat the characters to the end of the line 1652 /// since they may not be able to be tokenized to get to the end of line token. 1653 void AsmParser::eatToEndOfLine() { 1654 if (!Lexer.is(AsmToken::EndOfStatement)) 1655 Lexer.LexUntilEndOfLine(); 1656 // Eat EOL. 1657 Lex(); 1658 } 1659 1660 /// parseCppHashLineFilenameComment as this: 1661 /// ::= # number "filename" 1662 /// or just as a full line comment if it doesn't have a number and a string. 1663 bool AsmParser::parseCppHashLineFilenameComment(const SMLoc &L) { 1664 Lex(); // Eat the hash token. 1665 1666 if (getLexer().isNot(AsmToken::Integer)) { 1667 // Consume the line since in cases it is not a well-formed line directive, 1668 // as if were simply a full line comment. 1669 eatToEndOfLine(); 1670 return false; 1671 } 1672 1673 int64_t LineNumber = getTok().getIntVal(); 1674 Lex(); 1675 1676 if (getLexer().isNot(AsmToken::String)) { 1677 eatToEndOfLine(); 1678 return false; 1679 } 1680 1681 StringRef Filename = getTok().getString(); 1682 // Get rid of the enclosing quotes. 1683 Filename = Filename.substr(1, Filename.size() - 2); 1684 1685 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 1686 CppHashLoc = L; 1687 CppHashFilename = Filename; 1688 CppHashLineNumber = LineNumber; 1689 CppHashBuf = CurBuffer; 1690 1691 // Ignore any trailing characters, they're just comment. 1692 eatToEndOfLine(); 1693 return false; 1694 } 1695 1696 /// \brief will use the last parsed cpp hash line filename comment 1697 /// for the Filename and LineNo if any in the diagnostic. 1698 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 1699 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 1700 raw_ostream &OS = errs(); 1701 1702 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 1703 const SMLoc &DiagLoc = Diag.getLoc(); 1704 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1705 unsigned CppHashBuf = 1706 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashLoc); 1707 1708 // Like SourceMgr::printMessage() we need to print the include stack if any 1709 // before printing the message. 1710 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1711 if (!Parser->SavedDiagHandler && DiagCurBuffer && 1712 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 1713 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 1714 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 1715 } 1716 1717 // If we have not parsed a cpp hash line filename comment or the source 1718 // manager changed or buffer changed (like in a nested include) then just 1719 // print the normal diagnostic using its Filename and LineNo. 1720 if (!Parser->CppHashLineNumber || &DiagSrcMgr != &Parser->SrcMgr || 1721 DiagBuf != CppHashBuf) { 1722 if (Parser->SavedDiagHandler) 1723 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 1724 else 1725 Diag.print(nullptr, OS); 1726 return; 1727 } 1728 1729 // Use the CppHashFilename and calculate a line number based on the 1730 // CppHashLoc and CppHashLineNumber relative to this Diag's SMLoc for 1731 // the diagnostic. 1732 const std::string &Filename = Parser->CppHashFilename; 1733 1734 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 1735 int CppHashLocLineNo = 1736 Parser->SrcMgr.FindLineNumber(Parser->CppHashLoc, CppHashBuf); 1737 int LineNo = 1738 Parser->CppHashLineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 1739 1740 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 1741 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 1742 Diag.getLineContents(), Diag.getRanges()); 1743 1744 if (Parser->SavedDiagHandler) 1745 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 1746 else 1747 NewDiag.print(nullptr, OS); 1748 } 1749 1750 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 1751 // difference being that that function accepts '@' as part of identifiers and 1752 // we can't do that. AsmLexer.cpp should probably be changed to handle 1753 // '@' as a special case when needed. 1754 static bool isIdentifierChar(char c) { 1755 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 1756 c == '.'; 1757 } 1758 1759 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 1760 ArrayRef<MCAsmMacroParameter> Parameters, 1761 ArrayRef<MCAsmMacroArgument> A, const SMLoc &L) { 1762 unsigned NParameters = Parameters.size(); 1763 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 1764 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 1765 return Error(L, "Wrong number of arguments"); 1766 1767 // A macro without parameters is handled differently on Darwin: 1768 // gas accepts no arguments and does no substitutions 1769 while (!Body.empty()) { 1770 // Scan for the next substitution. 1771 std::size_t End = Body.size(), Pos = 0; 1772 for (; Pos != End; ++Pos) { 1773 // Check for a substitution or escape. 1774 if (IsDarwin && !NParameters) { 1775 // This macro has no parameters, look for $0, $1, etc. 1776 if (Body[Pos] != '$' || Pos + 1 == End) 1777 continue; 1778 1779 char Next = Body[Pos + 1]; 1780 if (Next == '$' || Next == 'n' || 1781 isdigit(static_cast<unsigned char>(Next))) 1782 break; 1783 } else { 1784 // This macro has parameters, look for \foo, \bar, etc. 1785 if (Body[Pos] == '\\' && Pos + 1 != End) 1786 break; 1787 } 1788 } 1789 1790 // Add the prefix. 1791 OS << Body.slice(0, Pos); 1792 1793 // Check if we reached the end. 1794 if (Pos == End) 1795 break; 1796 1797 if (IsDarwin && !NParameters) { 1798 switch (Body[Pos + 1]) { 1799 // $$ => $ 1800 case '$': 1801 OS << '$'; 1802 break; 1803 1804 // $n => number of arguments 1805 case 'n': 1806 OS << A.size(); 1807 break; 1808 1809 // $[0-9] => argument 1810 default: { 1811 // Missing arguments are ignored. 1812 unsigned Index = Body[Pos + 1] - '0'; 1813 if (Index >= A.size()) 1814 break; 1815 1816 // Otherwise substitute with the token values, with spaces eliminated. 1817 for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), 1818 ie = A[Index].end(); 1819 it != ie; ++it) 1820 OS << it->getString(); 1821 break; 1822 } 1823 } 1824 Pos += 2; 1825 } else { 1826 unsigned I = Pos + 1; 1827 while (isIdentifierChar(Body[I]) && I + 1 != End) 1828 ++I; 1829 1830 const char *Begin = Body.data() + Pos + 1; 1831 StringRef Argument(Begin, I - (Pos + 1)); 1832 unsigned Index = 0; 1833 for (; Index < NParameters; ++Index) 1834 if (Parameters[Index].Name == Argument) 1835 break; 1836 1837 if (Index == NParameters) { 1838 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 1839 Pos += 3; 1840 else { 1841 OS << '\\' << Argument; 1842 Pos = I; 1843 } 1844 } else { 1845 bool VarargParameter = HasVararg && Index == (NParameters - 1); 1846 for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), 1847 ie = A[Index].end(); 1848 it != ie; ++it) 1849 // We expect no quotes around the string's contents when 1850 // parsing for varargs. 1851 if (it->getKind() != AsmToken::String || VarargParameter) 1852 OS << it->getString(); 1853 else 1854 OS << it->getStringContents(); 1855 1856 Pos += 1 + Argument.size(); 1857 } 1858 } 1859 // Update the scan point. 1860 Body = Body.substr(Pos); 1861 } 1862 1863 return false; 1864 } 1865 1866 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 1867 size_t CondStackDepth) 1868 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 1869 CondStackDepth(CondStackDepth) {} 1870 1871 static bool isOperator(AsmToken::TokenKind kind) { 1872 switch (kind) { 1873 default: 1874 return false; 1875 case AsmToken::Plus: 1876 case AsmToken::Minus: 1877 case AsmToken::Tilde: 1878 case AsmToken::Slash: 1879 case AsmToken::Star: 1880 case AsmToken::Dot: 1881 case AsmToken::Equal: 1882 case AsmToken::EqualEqual: 1883 case AsmToken::Pipe: 1884 case AsmToken::PipePipe: 1885 case AsmToken::Caret: 1886 case AsmToken::Amp: 1887 case AsmToken::AmpAmp: 1888 case AsmToken::Exclaim: 1889 case AsmToken::ExclaimEqual: 1890 case AsmToken::Percent: 1891 case AsmToken::Less: 1892 case AsmToken::LessEqual: 1893 case AsmToken::LessLess: 1894 case AsmToken::LessGreater: 1895 case AsmToken::Greater: 1896 case AsmToken::GreaterEqual: 1897 case AsmToken::GreaterGreater: 1898 return true; 1899 } 1900 } 1901 1902 namespace { 1903 class AsmLexerSkipSpaceRAII { 1904 public: 1905 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 1906 Lexer.setSkipSpace(SkipSpace); 1907 } 1908 1909 ~AsmLexerSkipSpaceRAII() { 1910 Lexer.setSkipSpace(true); 1911 } 1912 1913 private: 1914 AsmLexer &Lexer; 1915 }; 1916 } 1917 1918 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 1919 1920 if (Vararg) { 1921 if (Lexer.isNot(AsmToken::EndOfStatement)) { 1922 StringRef Str = parseStringToEndOfStatement(); 1923 MA.push_back(AsmToken(AsmToken::String, Str)); 1924 } 1925 return false; 1926 } 1927 1928 unsigned ParenLevel = 0; 1929 unsigned AddTokens = 0; 1930 1931 // Darwin doesn't use spaces to delmit arguments. 1932 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 1933 1934 for (;;) { 1935 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 1936 return TokError("unexpected token in macro instantiation"); 1937 1938 if (ParenLevel == 0 && Lexer.is(AsmToken::Comma)) 1939 break; 1940 1941 if (Lexer.is(AsmToken::Space)) { 1942 Lex(); // Eat spaces 1943 1944 // Spaces can delimit parameters, but could also be part an expression. 1945 // If the token after a space is an operator, add the token and the next 1946 // one into this argument 1947 if (!IsDarwin) { 1948 if (isOperator(Lexer.getKind())) { 1949 // Check to see whether the token is used as an operator, 1950 // or part of an identifier 1951 const char *NextChar = getTok().getEndLoc().getPointer(); 1952 if (*NextChar == ' ') 1953 AddTokens = 2; 1954 } 1955 1956 if (!AddTokens && ParenLevel == 0) { 1957 break; 1958 } 1959 } 1960 } 1961 1962 // handleMacroEntry relies on not advancing the lexer here 1963 // to be able to fill in the remaining default parameter values 1964 if (Lexer.is(AsmToken::EndOfStatement)) 1965 break; 1966 1967 // Adjust the current parentheses level. 1968 if (Lexer.is(AsmToken::LParen)) 1969 ++ParenLevel; 1970 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 1971 --ParenLevel; 1972 1973 // Append the token to the current argument list. 1974 MA.push_back(getTok()); 1975 if (AddTokens) 1976 AddTokens--; 1977 Lex(); 1978 } 1979 1980 if (ParenLevel != 0) 1981 return TokError("unbalanced parentheses in macro argument"); 1982 return false; 1983 } 1984 1985 // Parse the macro instantiation arguments. 1986 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 1987 MCAsmMacroArguments &A) { 1988 const unsigned NParameters = M ? M->Parameters.size() : 0; 1989 bool NamedParametersFound = false; 1990 SmallVector<SMLoc, 4> FALocs; 1991 1992 A.resize(NParameters); 1993 FALocs.resize(NParameters); 1994 1995 // Parse two kinds of macro invocations: 1996 // - macros defined without any parameters accept an arbitrary number of them 1997 // - macros defined with parameters accept at most that many of them 1998 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 1999 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2000 ++Parameter) { 2001 SMLoc IDLoc = Lexer.getLoc(); 2002 MCAsmMacroParameter FA; 2003 2004 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2005 if (parseIdentifier(FA.Name)) { 2006 Error(IDLoc, "invalid argument identifier for formal argument"); 2007 eatToEndOfStatement(); 2008 return true; 2009 } 2010 2011 if (!Lexer.is(AsmToken::Equal)) { 2012 TokError("expected '=' after formal parameter identifier"); 2013 eatToEndOfStatement(); 2014 return true; 2015 } 2016 Lex(); 2017 2018 NamedParametersFound = true; 2019 } 2020 2021 if (NamedParametersFound && FA.Name.empty()) { 2022 Error(IDLoc, "cannot mix positional and keyword arguments"); 2023 eatToEndOfStatement(); 2024 return true; 2025 } 2026 2027 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2028 if (parseMacroArgument(FA.Value, Vararg)) 2029 return true; 2030 2031 unsigned PI = Parameter; 2032 if (!FA.Name.empty()) { 2033 unsigned FAI = 0; 2034 for (FAI = 0; FAI < NParameters; ++FAI) 2035 if (M->Parameters[FAI].Name == FA.Name) 2036 break; 2037 2038 if (FAI >= NParameters) { 2039 assert(M && "expected macro to be defined"); 2040 Error(IDLoc, 2041 "parameter named '" + FA.Name + "' does not exist for macro '" + 2042 M->Name + "'"); 2043 return true; 2044 } 2045 PI = FAI; 2046 } 2047 2048 if (!FA.Value.empty()) { 2049 if (A.size() <= PI) 2050 A.resize(PI + 1); 2051 A[PI] = FA.Value; 2052 2053 if (FALocs.size() <= PI) 2054 FALocs.resize(PI + 1); 2055 2056 FALocs[PI] = Lexer.getLoc(); 2057 } 2058 2059 // At the end of the statement, fill in remaining arguments that have 2060 // default values. If there aren't any, then the next argument is 2061 // required but missing 2062 if (Lexer.is(AsmToken::EndOfStatement)) { 2063 bool Failure = false; 2064 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2065 if (A[FAI].empty()) { 2066 if (M->Parameters[FAI].Required) { 2067 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2068 "missing value for required parameter " 2069 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2070 Failure = true; 2071 } 2072 2073 if (!M->Parameters[FAI].Value.empty()) 2074 A[FAI] = M->Parameters[FAI].Value; 2075 } 2076 } 2077 return Failure; 2078 } 2079 2080 if (Lexer.is(AsmToken::Comma)) 2081 Lex(); 2082 } 2083 2084 return TokError("too many positional arguments"); 2085 } 2086 2087 const MCAsmMacro *AsmParser::lookupMacro(StringRef Name) { 2088 StringMap<MCAsmMacro>::iterator I = MacroMap.find(Name); 2089 return (I == MacroMap.end()) ? nullptr : &I->getValue(); 2090 } 2091 2092 void AsmParser::defineMacro(StringRef Name, MCAsmMacro Macro) { 2093 MacroMap.insert(std::make_pair(Name, std::move(Macro))); 2094 } 2095 2096 void AsmParser::undefineMacro(StringRef Name) { MacroMap.erase(Name); } 2097 2098 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2099 // Arbitrarily limit macro nesting depth, to match 'as'. We can eliminate 2100 // this, although we should protect against infinite loops. 2101 if (ActiveMacros.size() == 20) 2102 return TokError("macros cannot be nested more than 20 levels deep"); 2103 2104 MCAsmMacroArguments A; 2105 if (parseMacroArguments(M, A)) 2106 return true; 2107 2108 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2109 // to hold the macro body with substitutions. 2110 SmallString<256> Buf; 2111 StringRef Body = M->Body; 2112 raw_svector_ostream OS(Buf); 2113 2114 if (expandMacro(OS, Body, M->Parameters, A, getTok().getLoc())) 2115 return true; 2116 2117 // We include the .endmacro in the buffer as our cue to exit the macro 2118 // instantiation. 2119 OS << ".endmacro\n"; 2120 2121 std::unique_ptr<MemoryBuffer> Instantiation = 2122 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2123 2124 // Create the macro instantiation object and add to the current macro 2125 // instantiation stack. 2126 MacroInstantiation *MI = new MacroInstantiation( 2127 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2128 ActiveMacros.push_back(MI); 2129 2130 // Jump to the macro instantiation and prime the lexer. 2131 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2132 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2133 Lex(); 2134 2135 return false; 2136 } 2137 2138 void AsmParser::handleMacroExit() { 2139 // Jump to the EndOfStatement we should return to, and consume it. 2140 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2141 Lex(); 2142 2143 // Pop the instantiation entry. 2144 delete ActiveMacros.back(); 2145 ActiveMacros.pop_back(); 2146 } 2147 2148 static bool isUsedIn(const MCSymbol *Sym, const MCExpr *Value) { 2149 switch (Value->getKind()) { 2150 case MCExpr::Binary: { 2151 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 2152 return isUsedIn(Sym, BE->getLHS()) || isUsedIn(Sym, BE->getRHS()); 2153 } 2154 case MCExpr::Target: 2155 case MCExpr::Constant: 2156 return false; 2157 case MCExpr::SymbolRef: { 2158 const MCSymbol &S = 2159 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 2160 if (S.isVariable()) 2161 return isUsedIn(Sym, S.getVariableValue()); 2162 return &S == Sym; 2163 } 2164 case MCExpr::Unary: 2165 return isUsedIn(Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 2166 } 2167 2168 llvm_unreachable("Unknown expr kind!"); 2169 } 2170 2171 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2172 bool NoDeadStrip) { 2173 // FIXME: Use better location, we should use proper tokens. 2174 SMLoc EqualLoc = Lexer.getLoc(); 2175 2176 const MCExpr *Value; 2177 if (parseExpression(Value)) 2178 return true; 2179 2180 // Note: we don't count b as used in "a = b". This is to allow 2181 // a = b 2182 // b = c 2183 2184 if (Lexer.isNot(AsmToken::EndOfStatement)) 2185 return TokError("unexpected token in assignment"); 2186 2187 // Eat the end of statement marker. 2188 Lex(); 2189 2190 // Validate that the LHS is allowed to be a variable (either it has not been 2191 // used as a symbol, or it is an absolute symbol). 2192 MCSymbol *Sym = getContext().LookupSymbol(Name); 2193 if (Sym) { 2194 // Diagnose assignment to a label. 2195 // 2196 // FIXME: Diagnostics. Note the location of the definition as a label. 2197 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 2198 if (isUsedIn(Sym, Value)) 2199 return Error(EqualLoc, "Recursive use of '" + Name + "'"); 2200 else if (Sym->isUndefined() && !Sym->isUsed() && !Sym->isVariable()) 2201 ; // Allow redefinitions of undefined symbols only used in directives. 2202 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 2203 ; // Allow redefinitions of variables that haven't yet been used. 2204 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 2205 return Error(EqualLoc, "redefinition of '" + Name + "'"); 2206 else if (!Sym->isVariable()) 2207 return Error(EqualLoc, "invalid assignment to '" + Name + "'"); 2208 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 2209 return Error(EqualLoc, "invalid reassignment of non-absolute variable '" + 2210 Name + "'"); 2211 2212 // Don't count these checks as uses. 2213 Sym->setUsed(false); 2214 } else if (Name == ".") { 2215 if (Out.EmitValueToOffset(Value, 0)) { 2216 Error(EqualLoc, "expected absolute expression"); 2217 eatToEndOfStatement(); 2218 } 2219 return false; 2220 } else 2221 Sym = getContext().GetOrCreateSymbol(Name); 2222 2223 Sym->setRedefinable(allow_redef); 2224 2225 // Do the assignment. 2226 Out.EmitAssignment(Sym, Value); 2227 if (NoDeadStrip) 2228 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2229 2230 return false; 2231 } 2232 2233 /// parseIdentifier: 2234 /// ::= identifier 2235 /// ::= string 2236 bool AsmParser::parseIdentifier(StringRef &Res) { 2237 // The assembler has relaxed rules for accepting identifiers, in particular we 2238 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2239 // separate tokens. At this level, we have already lexed so we cannot (currently) 2240 // handle this as a context dependent token, instead we detect adjacent tokens 2241 // and return the combined identifier. 2242 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2243 SMLoc PrefixLoc = getLexer().getLoc(); 2244 2245 // Consume the prefix character, and check for a following identifier. 2246 Lex(); 2247 if (Lexer.isNot(AsmToken::Identifier)) 2248 return true; 2249 2250 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2251 if (PrefixLoc.getPointer() + 1 != getTok().getLoc().getPointer()) 2252 return true; 2253 2254 // Construct the joined identifier and consume the token. 2255 Res = 2256 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2257 Lex(); 2258 return false; 2259 } 2260 2261 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2262 return true; 2263 2264 Res = getTok().getIdentifier(); 2265 2266 Lex(); // Consume the identifier token. 2267 2268 return false; 2269 } 2270 2271 /// parseDirectiveSet: 2272 /// ::= .equ identifier ',' expression 2273 /// ::= .equiv identifier ',' expression 2274 /// ::= .set identifier ',' expression 2275 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2276 StringRef Name; 2277 2278 if (parseIdentifier(Name)) 2279 return TokError("expected identifier after '" + Twine(IDVal) + "'"); 2280 2281 if (getLexer().isNot(AsmToken::Comma)) 2282 return TokError("unexpected token in '" + Twine(IDVal) + "'"); 2283 Lex(); 2284 2285 return parseAssignment(Name, allow_redef, true); 2286 } 2287 2288 bool AsmParser::parseEscapedString(std::string &Data) { 2289 assert(getLexer().is(AsmToken::String) && "Unexpected current token!"); 2290 2291 Data = ""; 2292 StringRef Str = getTok().getStringContents(); 2293 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2294 if (Str[i] != '\\') { 2295 Data += Str[i]; 2296 continue; 2297 } 2298 2299 // Recognize escaped characters. Note that this escape semantics currently 2300 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2301 ++i; 2302 if (i == e) 2303 return TokError("unexpected backslash at end of string"); 2304 2305 // Recognize octal sequences. 2306 if ((unsigned)(Str[i] - '0') <= 7) { 2307 // Consume up to three octal characters. 2308 unsigned Value = Str[i] - '0'; 2309 2310 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2311 ++i; 2312 Value = Value * 8 + (Str[i] - '0'); 2313 2314 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2315 ++i; 2316 Value = Value * 8 + (Str[i] - '0'); 2317 } 2318 } 2319 2320 if (Value > 255) 2321 return TokError("invalid octal escape sequence (out of range)"); 2322 2323 Data += (unsigned char)Value; 2324 continue; 2325 } 2326 2327 // Otherwise recognize individual escapes. 2328 switch (Str[i]) { 2329 default: 2330 // Just reject invalid escape sequences for now. 2331 return TokError("invalid escape sequence (unrecognized character)"); 2332 2333 case 'b': Data += '\b'; break; 2334 case 'f': Data += '\f'; break; 2335 case 'n': Data += '\n'; break; 2336 case 'r': Data += '\r'; break; 2337 case 't': Data += '\t'; break; 2338 case '"': Data += '"'; break; 2339 case '\\': Data += '\\'; break; 2340 } 2341 } 2342 2343 return false; 2344 } 2345 2346 /// parseDirectiveAscii: 2347 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2348 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2349 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2350 checkForValidSection(); 2351 2352 for (;;) { 2353 if (getLexer().isNot(AsmToken::String)) 2354 return TokError("expected string in '" + Twine(IDVal) + "' directive"); 2355 2356 std::string Data; 2357 if (parseEscapedString(Data)) 2358 return true; 2359 2360 getStreamer().EmitBytes(Data); 2361 if (ZeroTerminated) 2362 getStreamer().EmitBytes(StringRef("\0", 1)); 2363 2364 Lex(); 2365 2366 if (getLexer().is(AsmToken::EndOfStatement)) 2367 break; 2368 2369 if (getLexer().isNot(AsmToken::Comma)) 2370 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 2371 Lex(); 2372 } 2373 } 2374 2375 Lex(); 2376 return false; 2377 } 2378 2379 /// parseDirectiveValue 2380 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 2381 bool AsmParser::parseDirectiveValue(unsigned Size) { 2382 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2383 checkForValidSection(); 2384 2385 for (;;) { 2386 const MCExpr *Value; 2387 SMLoc ExprLoc = getLexer().getLoc(); 2388 if (parseExpression(Value)) 2389 return true; 2390 2391 // Special case constant expressions to match code generator. 2392 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2393 assert(Size <= 8 && "Invalid size"); 2394 uint64_t IntValue = MCE->getValue(); 2395 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2396 return Error(ExprLoc, "literal value out of range for directive"); 2397 getStreamer().EmitIntValue(IntValue, Size); 2398 } else 2399 getStreamer().EmitValue(Value, Size, ExprLoc); 2400 2401 if (getLexer().is(AsmToken::EndOfStatement)) 2402 break; 2403 2404 // FIXME: Improve diagnostic. 2405 if (getLexer().isNot(AsmToken::Comma)) 2406 return TokError("unexpected token in directive"); 2407 Lex(); 2408 } 2409 } 2410 2411 Lex(); 2412 return false; 2413 } 2414 2415 /// ParseDirectiveOctaValue 2416 /// ::= .octa [ hexconstant (, hexconstant)* ] 2417 bool AsmParser::parseDirectiveOctaValue() { 2418 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2419 checkForValidSection(); 2420 2421 for (;;) { 2422 if (Lexer.getKind() == AsmToken::Error) 2423 return true; 2424 if (Lexer.getKind() != AsmToken::Integer && 2425 Lexer.getKind() != AsmToken::BigNum) 2426 return TokError("unknown token in expression"); 2427 2428 SMLoc ExprLoc = getLexer().getLoc(); 2429 APInt IntValue = getTok().getAPIntVal(); 2430 Lex(); 2431 2432 uint64_t hi, lo; 2433 if (IntValue.isIntN(64)) { 2434 hi = 0; 2435 lo = IntValue.getZExtValue(); 2436 } else if (IntValue.isIntN(128)) { 2437 // It might actually have more than 128 bits, but the top ones are zero. 2438 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2439 lo = IntValue.getLoBits(64).getZExtValue(); 2440 } else 2441 return Error(ExprLoc, "literal value out of range for directive"); 2442 2443 if (MAI.isLittleEndian()) { 2444 getStreamer().EmitIntValue(lo, 8); 2445 getStreamer().EmitIntValue(hi, 8); 2446 } else { 2447 getStreamer().EmitIntValue(hi, 8); 2448 getStreamer().EmitIntValue(lo, 8); 2449 } 2450 2451 if (getLexer().is(AsmToken::EndOfStatement)) 2452 break; 2453 2454 // FIXME: Improve diagnostic. 2455 if (getLexer().isNot(AsmToken::Comma)) 2456 return TokError("unexpected token in directive"); 2457 Lex(); 2458 } 2459 } 2460 2461 Lex(); 2462 return false; 2463 } 2464 2465 /// parseDirectiveRealValue 2466 /// ::= (.single | .double) [ expression (, expression)* ] 2467 bool AsmParser::parseDirectiveRealValue(const fltSemantics &Semantics) { 2468 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2469 checkForValidSection(); 2470 2471 for (;;) { 2472 // We don't truly support arithmetic on floating point expressions, so we 2473 // have to manually parse unary prefixes. 2474 bool IsNeg = false; 2475 if (getLexer().is(AsmToken::Minus)) { 2476 Lex(); 2477 IsNeg = true; 2478 } else if (getLexer().is(AsmToken::Plus)) 2479 Lex(); 2480 2481 if (getLexer().isNot(AsmToken::Integer) && 2482 getLexer().isNot(AsmToken::Real) && 2483 getLexer().isNot(AsmToken::Identifier)) 2484 return TokError("unexpected token in directive"); 2485 2486 // Convert to an APFloat. 2487 APFloat Value(Semantics); 2488 StringRef IDVal = getTok().getString(); 2489 if (getLexer().is(AsmToken::Identifier)) { 2490 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2491 Value = APFloat::getInf(Semantics); 2492 else if (!IDVal.compare_lower("nan")) 2493 Value = APFloat::getNaN(Semantics, false, ~0); 2494 else 2495 return TokError("invalid floating point literal"); 2496 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 2497 APFloat::opInvalidOp) 2498 return TokError("invalid floating point literal"); 2499 if (IsNeg) 2500 Value.changeSign(); 2501 2502 // Consume the numeric token. 2503 Lex(); 2504 2505 // Emit the value as an integer. 2506 APInt AsInt = Value.bitcastToAPInt(); 2507 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 2508 AsInt.getBitWidth() / 8); 2509 2510 if (getLexer().is(AsmToken::EndOfStatement)) 2511 break; 2512 2513 if (getLexer().isNot(AsmToken::Comma)) 2514 return TokError("unexpected token in directive"); 2515 Lex(); 2516 } 2517 } 2518 2519 Lex(); 2520 return false; 2521 } 2522 2523 /// parseDirectiveZero 2524 /// ::= .zero expression 2525 bool AsmParser::parseDirectiveZero() { 2526 checkForValidSection(); 2527 2528 int64_t NumBytes; 2529 if (parseAbsoluteExpression(NumBytes)) 2530 return true; 2531 2532 int64_t Val = 0; 2533 if (getLexer().is(AsmToken::Comma)) { 2534 Lex(); 2535 if (parseAbsoluteExpression(Val)) 2536 return true; 2537 } 2538 2539 if (getLexer().isNot(AsmToken::EndOfStatement)) 2540 return TokError("unexpected token in '.zero' directive"); 2541 2542 Lex(); 2543 2544 getStreamer().EmitFill(NumBytes, Val); 2545 2546 return false; 2547 } 2548 2549 /// parseDirectiveFill 2550 /// ::= .fill expression [ , expression [ , expression ] ] 2551 bool AsmParser::parseDirectiveFill() { 2552 checkForValidSection(); 2553 2554 SMLoc RepeatLoc = getLexer().getLoc(); 2555 int64_t NumValues; 2556 if (parseAbsoluteExpression(NumValues)) 2557 return true; 2558 2559 if (NumValues < 0) { 2560 Warning(RepeatLoc, 2561 "'.fill' directive with negative repeat count has no effect"); 2562 NumValues = 0; 2563 } 2564 2565 int64_t FillSize = 1; 2566 int64_t FillExpr = 0; 2567 2568 SMLoc SizeLoc, ExprLoc; 2569 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2570 if (getLexer().isNot(AsmToken::Comma)) 2571 return TokError("unexpected token in '.fill' directive"); 2572 Lex(); 2573 2574 SizeLoc = getLexer().getLoc(); 2575 if (parseAbsoluteExpression(FillSize)) 2576 return true; 2577 2578 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2579 if (getLexer().isNot(AsmToken::Comma)) 2580 return TokError("unexpected token in '.fill' directive"); 2581 Lex(); 2582 2583 ExprLoc = getLexer().getLoc(); 2584 if (parseAbsoluteExpression(FillExpr)) 2585 return true; 2586 2587 if (getLexer().isNot(AsmToken::EndOfStatement)) 2588 return TokError("unexpected token in '.fill' directive"); 2589 2590 Lex(); 2591 } 2592 } 2593 2594 if (FillSize < 0) { 2595 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 2596 NumValues = 0; 2597 } 2598 if (FillSize > 8) { 2599 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 2600 FillSize = 8; 2601 } 2602 2603 if (!isUInt<32>(FillExpr) && FillSize > 4) 2604 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 2605 2606 if (NumValues > 0) { 2607 int64_t NonZeroFillSize = FillSize > 4 ? 4 : FillSize; 2608 FillExpr &= ~0ULL >> (64 - NonZeroFillSize * 8); 2609 for (uint64_t i = 0, e = NumValues; i != e; ++i) { 2610 getStreamer().EmitIntValue(FillExpr, NonZeroFillSize); 2611 if (NonZeroFillSize < FillSize) 2612 getStreamer().EmitIntValue(0, FillSize - NonZeroFillSize); 2613 } 2614 } 2615 2616 return false; 2617 } 2618 2619 /// parseDirectiveOrg 2620 /// ::= .org expression [ , expression ] 2621 bool AsmParser::parseDirectiveOrg() { 2622 checkForValidSection(); 2623 2624 const MCExpr *Offset; 2625 SMLoc Loc = getTok().getLoc(); 2626 if (parseExpression(Offset)) 2627 return true; 2628 2629 // Parse optional fill expression. 2630 int64_t FillExpr = 0; 2631 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2632 if (getLexer().isNot(AsmToken::Comma)) 2633 return TokError("unexpected token in '.org' directive"); 2634 Lex(); 2635 2636 if (parseAbsoluteExpression(FillExpr)) 2637 return true; 2638 2639 if (getLexer().isNot(AsmToken::EndOfStatement)) 2640 return TokError("unexpected token in '.org' directive"); 2641 } 2642 2643 Lex(); 2644 2645 // Only limited forms of relocatable expressions are accepted here, it 2646 // has to be relative to the current section. The streamer will return 2647 // 'true' if the expression wasn't evaluatable. 2648 if (getStreamer().EmitValueToOffset(Offset, FillExpr)) 2649 return Error(Loc, "expected assembly-time absolute expression"); 2650 2651 return false; 2652 } 2653 2654 /// parseDirectiveAlign 2655 /// ::= {.align, ...} expression [ , expression [ , expression ]] 2656 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 2657 checkForValidSection(); 2658 2659 SMLoc AlignmentLoc = getLexer().getLoc(); 2660 int64_t Alignment; 2661 if (parseAbsoluteExpression(Alignment)) 2662 return true; 2663 2664 SMLoc MaxBytesLoc; 2665 bool HasFillExpr = false; 2666 int64_t FillExpr = 0; 2667 int64_t MaxBytesToFill = 0; 2668 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2669 if (getLexer().isNot(AsmToken::Comma)) 2670 return TokError("unexpected token in directive"); 2671 Lex(); 2672 2673 // The fill expression can be omitted while specifying a maximum number of 2674 // alignment bytes, e.g: 2675 // .align 3,,4 2676 if (getLexer().isNot(AsmToken::Comma)) { 2677 HasFillExpr = true; 2678 if (parseAbsoluteExpression(FillExpr)) 2679 return true; 2680 } 2681 2682 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2683 if (getLexer().isNot(AsmToken::Comma)) 2684 return TokError("unexpected token in directive"); 2685 Lex(); 2686 2687 MaxBytesLoc = getLexer().getLoc(); 2688 if (parseAbsoluteExpression(MaxBytesToFill)) 2689 return true; 2690 2691 if (getLexer().isNot(AsmToken::EndOfStatement)) 2692 return TokError("unexpected token in directive"); 2693 } 2694 } 2695 2696 Lex(); 2697 2698 if (!HasFillExpr) 2699 FillExpr = 0; 2700 2701 // Compute alignment in bytes. 2702 if (IsPow2) { 2703 // FIXME: Diagnose overflow. 2704 if (Alignment >= 32) { 2705 Error(AlignmentLoc, "invalid alignment value"); 2706 Alignment = 31; 2707 } 2708 2709 Alignment = 1ULL << Alignment; 2710 } else { 2711 // Reject alignments that aren't a power of two, for gas compatibility. 2712 if (!isPowerOf2_64(Alignment)) 2713 Error(AlignmentLoc, "alignment must be a power of 2"); 2714 } 2715 2716 // Diagnose non-sensical max bytes to align. 2717 if (MaxBytesLoc.isValid()) { 2718 if (MaxBytesToFill < 1) { 2719 Error(MaxBytesLoc, "alignment directive can never be satisfied in this " 2720 "many bytes, ignoring maximum bytes expression"); 2721 MaxBytesToFill = 0; 2722 } 2723 2724 if (MaxBytesToFill >= Alignment) { 2725 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 2726 "has no effect"); 2727 MaxBytesToFill = 0; 2728 } 2729 } 2730 2731 // Check whether we should use optimal code alignment for this .align 2732 // directive. 2733 const MCSection *Section = getStreamer().getCurrentSection().first; 2734 assert(Section && "must have section to emit alignment"); 2735 bool UseCodeAlign = Section->UseCodeAlign(); 2736 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 2737 ValueSize == 1 && UseCodeAlign) { 2738 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 2739 } else { 2740 // FIXME: Target specific behavior about how the "extra" bytes are filled. 2741 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 2742 MaxBytesToFill); 2743 } 2744 2745 return false; 2746 } 2747 2748 /// parseDirectiveFile 2749 /// ::= .file [number] filename 2750 /// ::= .file number directory filename 2751 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 2752 // FIXME: I'm not sure what this is. 2753 int64_t FileNumber = -1; 2754 SMLoc FileNumberLoc = getLexer().getLoc(); 2755 if (getLexer().is(AsmToken::Integer)) { 2756 FileNumber = getTok().getIntVal(); 2757 Lex(); 2758 2759 if (FileNumber < 1) 2760 return TokError("file number less than one"); 2761 } 2762 2763 if (getLexer().isNot(AsmToken::String)) 2764 return TokError("unexpected token in '.file' directive"); 2765 2766 // Usually the directory and filename together, otherwise just the directory. 2767 // Allow the strings to have escaped octal character sequence. 2768 std::string Path = getTok().getString(); 2769 if (parseEscapedString(Path)) 2770 return true; 2771 Lex(); 2772 2773 StringRef Directory; 2774 StringRef Filename; 2775 std::string FilenameData; 2776 if (getLexer().is(AsmToken::String)) { 2777 if (FileNumber == -1) 2778 return TokError("explicit path specified, but no file number"); 2779 if (parseEscapedString(FilenameData)) 2780 return true; 2781 Filename = FilenameData; 2782 Directory = Path; 2783 Lex(); 2784 } else { 2785 Filename = Path; 2786 } 2787 2788 if (getLexer().isNot(AsmToken::EndOfStatement)) 2789 return TokError("unexpected token in '.file' directive"); 2790 2791 if (FileNumber == -1) 2792 getStreamer().EmitFileDirective(Filename); 2793 else { 2794 if (getContext().getGenDwarfForAssembly() == true) 2795 Error(DirectiveLoc, 2796 "input can't have .file dwarf directives when -g is " 2797 "used to generate dwarf debug info for assembly code"); 2798 2799 if (getStreamer().EmitDwarfFileDirective(FileNumber, Directory, Filename) == 2800 0) 2801 Error(FileNumberLoc, "file number already allocated"); 2802 } 2803 2804 return false; 2805 } 2806 2807 /// parseDirectiveLine 2808 /// ::= .line [number] 2809 bool AsmParser::parseDirectiveLine() { 2810 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2811 if (getLexer().isNot(AsmToken::Integer)) 2812 return TokError("unexpected token in '.line' directive"); 2813 2814 int64_t LineNumber = getTok().getIntVal(); 2815 (void)LineNumber; 2816 Lex(); 2817 2818 // FIXME: Do something with the .line. 2819 } 2820 2821 if (getLexer().isNot(AsmToken::EndOfStatement)) 2822 return TokError("unexpected token in '.line' directive"); 2823 2824 return false; 2825 } 2826 2827 /// parseDirectiveLoc 2828 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 2829 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 2830 /// The first number is a file number, must have been previously assigned with 2831 /// a .file directive, the second number is the line number and optionally the 2832 /// third number is a column position (zero if not specified). The remaining 2833 /// optional items are .loc sub-directives. 2834 bool AsmParser::parseDirectiveLoc() { 2835 if (getLexer().isNot(AsmToken::Integer)) 2836 return TokError("unexpected token in '.loc' directive"); 2837 int64_t FileNumber = getTok().getIntVal(); 2838 if (FileNumber < 1) 2839 return TokError("file number less than one in '.loc' directive"); 2840 if (!getContext().isValidDwarfFileNumber(FileNumber)) 2841 return TokError("unassigned file number in '.loc' directive"); 2842 Lex(); 2843 2844 int64_t LineNumber = 0; 2845 if (getLexer().is(AsmToken::Integer)) { 2846 LineNumber = getTok().getIntVal(); 2847 if (LineNumber < 0) 2848 return TokError("line number less than zero in '.loc' directive"); 2849 Lex(); 2850 } 2851 2852 int64_t ColumnPos = 0; 2853 if (getLexer().is(AsmToken::Integer)) { 2854 ColumnPos = getTok().getIntVal(); 2855 if (ColumnPos < 0) 2856 return TokError("column position less than zero in '.loc' directive"); 2857 Lex(); 2858 } 2859 2860 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 2861 unsigned Isa = 0; 2862 int64_t Discriminator = 0; 2863 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2864 for (;;) { 2865 if (getLexer().is(AsmToken::EndOfStatement)) 2866 break; 2867 2868 StringRef Name; 2869 SMLoc Loc = getTok().getLoc(); 2870 if (parseIdentifier(Name)) 2871 return TokError("unexpected token in '.loc' directive"); 2872 2873 if (Name == "basic_block") 2874 Flags |= DWARF2_FLAG_BASIC_BLOCK; 2875 else if (Name == "prologue_end") 2876 Flags |= DWARF2_FLAG_PROLOGUE_END; 2877 else if (Name == "epilogue_begin") 2878 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 2879 else if (Name == "is_stmt") { 2880 Loc = getTok().getLoc(); 2881 const MCExpr *Value; 2882 if (parseExpression(Value)) 2883 return true; 2884 // The expression must be the constant 0 or 1. 2885 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2886 int Value = MCE->getValue(); 2887 if (Value == 0) 2888 Flags &= ~DWARF2_FLAG_IS_STMT; 2889 else if (Value == 1) 2890 Flags |= DWARF2_FLAG_IS_STMT; 2891 else 2892 return Error(Loc, "is_stmt value not 0 or 1"); 2893 } else { 2894 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 2895 } 2896 } else if (Name == "isa") { 2897 Loc = getTok().getLoc(); 2898 const MCExpr *Value; 2899 if (parseExpression(Value)) 2900 return true; 2901 // The expression must be a constant greater or equal to 0. 2902 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2903 int Value = MCE->getValue(); 2904 if (Value < 0) 2905 return Error(Loc, "isa number less than zero"); 2906 Isa = Value; 2907 } else { 2908 return Error(Loc, "isa number not a constant value"); 2909 } 2910 } else if (Name == "discriminator") { 2911 if (parseAbsoluteExpression(Discriminator)) 2912 return true; 2913 } else { 2914 return Error(Loc, "unknown sub-directive in '.loc' directive"); 2915 } 2916 2917 if (getLexer().is(AsmToken::EndOfStatement)) 2918 break; 2919 } 2920 } 2921 2922 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 2923 Isa, Discriminator, StringRef()); 2924 2925 return false; 2926 } 2927 2928 /// parseDirectiveStabs 2929 /// ::= .stabs string, number, number, number 2930 bool AsmParser::parseDirectiveStabs() { 2931 return TokError("unsupported directive '.stabs'"); 2932 } 2933 2934 /// parseDirectiveCFISections 2935 /// ::= .cfi_sections section [, section] 2936 bool AsmParser::parseDirectiveCFISections() { 2937 StringRef Name; 2938 bool EH = false; 2939 bool Debug = false; 2940 2941 if (parseIdentifier(Name)) 2942 return TokError("Expected an identifier"); 2943 2944 if (Name == ".eh_frame") 2945 EH = true; 2946 else if (Name == ".debug_frame") 2947 Debug = true; 2948 2949 if (getLexer().is(AsmToken::Comma)) { 2950 Lex(); 2951 2952 if (parseIdentifier(Name)) 2953 return TokError("Expected an identifier"); 2954 2955 if (Name == ".eh_frame") 2956 EH = true; 2957 else if (Name == ".debug_frame") 2958 Debug = true; 2959 } 2960 2961 getStreamer().EmitCFISections(EH, Debug); 2962 return false; 2963 } 2964 2965 /// parseDirectiveCFIStartProc 2966 /// ::= .cfi_startproc [simple] 2967 bool AsmParser::parseDirectiveCFIStartProc() { 2968 StringRef Simple; 2969 if (getLexer().isNot(AsmToken::EndOfStatement)) 2970 if (parseIdentifier(Simple) || Simple != "simple") 2971 return TokError("unexpected token in .cfi_startproc directive"); 2972 2973 getStreamer().EmitCFIStartProc(!Simple.empty()); 2974 return false; 2975 } 2976 2977 /// parseDirectiveCFIEndProc 2978 /// ::= .cfi_endproc 2979 bool AsmParser::parseDirectiveCFIEndProc() { 2980 getStreamer().EmitCFIEndProc(); 2981 return false; 2982 } 2983 2984 /// \brief parse register name or number. 2985 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 2986 SMLoc DirectiveLoc) { 2987 unsigned RegNo; 2988 2989 if (getLexer().isNot(AsmToken::Integer)) { 2990 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 2991 return true; 2992 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 2993 } else 2994 return parseAbsoluteExpression(Register); 2995 2996 return false; 2997 } 2998 2999 /// parseDirectiveCFIDefCfa 3000 /// ::= .cfi_def_cfa register, offset 3001 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3002 int64_t Register = 0; 3003 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3004 return true; 3005 3006 if (getLexer().isNot(AsmToken::Comma)) 3007 return TokError("unexpected token in directive"); 3008 Lex(); 3009 3010 int64_t Offset = 0; 3011 if (parseAbsoluteExpression(Offset)) 3012 return true; 3013 3014 getStreamer().EmitCFIDefCfa(Register, Offset); 3015 return false; 3016 } 3017 3018 /// parseDirectiveCFIDefCfaOffset 3019 /// ::= .cfi_def_cfa_offset offset 3020 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3021 int64_t Offset = 0; 3022 if (parseAbsoluteExpression(Offset)) 3023 return true; 3024 3025 getStreamer().EmitCFIDefCfaOffset(Offset); 3026 return false; 3027 } 3028 3029 /// parseDirectiveCFIRegister 3030 /// ::= .cfi_register register, register 3031 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3032 int64_t Register1 = 0; 3033 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc)) 3034 return true; 3035 3036 if (getLexer().isNot(AsmToken::Comma)) 3037 return TokError("unexpected token in directive"); 3038 Lex(); 3039 3040 int64_t Register2 = 0; 3041 if (parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3042 return true; 3043 3044 getStreamer().EmitCFIRegister(Register1, Register2); 3045 return false; 3046 } 3047 3048 /// parseDirectiveCFIWindowSave 3049 /// ::= .cfi_window_save 3050 bool AsmParser::parseDirectiveCFIWindowSave() { 3051 getStreamer().EmitCFIWindowSave(); 3052 return false; 3053 } 3054 3055 /// parseDirectiveCFIAdjustCfaOffset 3056 /// ::= .cfi_adjust_cfa_offset adjustment 3057 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 3058 int64_t Adjustment = 0; 3059 if (parseAbsoluteExpression(Adjustment)) 3060 return true; 3061 3062 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 3063 return false; 3064 } 3065 3066 /// parseDirectiveCFIDefCfaRegister 3067 /// ::= .cfi_def_cfa_register register 3068 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3069 int64_t Register = 0; 3070 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3071 return true; 3072 3073 getStreamer().EmitCFIDefCfaRegister(Register); 3074 return false; 3075 } 3076 3077 /// parseDirectiveCFIOffset 3078 /// ::= .cfi_offset register, offset 3079 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3080 int64_t Register = 0; 3081 int64_t Offset = 0; 3082 3083 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3084 return true; 3085 3086 if (getLexer().isNot(AsmToken::Comma)) 3087 return TokError("unexpected token in directive"); 3088 Lex(); 3089 3090 if (parseAbsoluteExpression(Offset)) 3091 return true; 3092 3093 getStreamer().EmitCFIOffset(Register, Offset); 3094 return false; 3095 } 3096 3097 /// parseDirectiveCFIRelOffset 3098 /// ::= .cfi_rel_offset register, offset 3099 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3100 int64_t Register = 0; 3101 3102 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3103 return true; 3104 3105 if (getLexer().isNot(AsmToken::Comma)) 3106 return TokError("unexpected token in directive"); 3107 Lex(); 3108 3109 int64_t Offset = 0; 3110 if (parseAbsoluteExpression(Offset)) 3111 return true; 3112 3113 getStreamer().EmitCFIRelOffset(Register, Offset); 3114 return false; 3115 } 3116 3117 static bool isValidEncoding(int64_t Encoding) { 3118 if (Encoding & ~0xff) 3119 return false; 3120 3121 if (Encoding == dwarf::DW_EH_PE_omit) 3122 return true; 3123 3124 const unsigned Format = Encoding & 0xf; 3125 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3126 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3127 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3128 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3129 return false; 3130 3131 const unsigned Application = Encoding & 0x70; 3132 if (Application != dwarf::DW_EH_PE_absptr && 3133 Application != dwarf::DW_EH_PE_pcrel) 3134 return false; 3135 3136 return true; 3137 } 3138 3139 /// parseDirectiveCFIPersonalityOrLsda 3140 /// IsPersonality true for cfi_personality, false for cfi_lsda 3141 /// ::= .cfi_personality encoding, [symbol_name] 3142 /// ::= .cfi_lsda encoding, [symbol_name] 3143 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3144 int64_t Encoding = 0; 3145 if (parseAbsoluteExpression(Encoding)) 3146 return true; 3147 if (Encoding == dwarf::DW_EH_PE_omit) 3148 return false; 3149 3150 if (!isValidEncoding(Encoding)) 3151 return TokError("unsupported encoding."); 3152 3153 if (getLexer().isNot(AsmToken::Comma)) 3154 return TokError("unexpected token in directive"); 3155 Lex(); 3156 3157 StringRef Name; 3158 if (parseIdentifier(Name)) 3159 return TokError("expected identifier in directive"); 3160 3161 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3162 3163 if (IsPersonality) 3164 getStreamer().EmitCFIPersonality(Sym, Encoding); 3165 else 3166 getStreamer().EmitCFILsda(Sym, Encoding); 3167 return false; 3168 } 3169 3170 /// parseDirectiveCFIRememberState 3171 /// ::= .cfi_remember_state 3172 bool AsmParser::parseDirectiveCFIRememberState() { 3173 getStreamer().EmitCFIRememberState(); 3174 return false; 3175 } 3176 3177 /// parseDirectiveCFIRestoreState 3178 /// ::= .cfi_remember_state 3179 bool AsmParser::parseDirectiveCFIRestoreState() { 3180 getStreamer().EmitCFIRestoreState(); 3181 return false; 3182 } 3183 3184 /// parseDirectiveCFISameValue 3185 /// ::= .cfi_same_value register 3186 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3187 int64_t Register = 0; 3188 3189 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3190 return true; 3191 3192 getStreamer().EmitCFISameValue(Register); 3193 return false; 3194 } 3195 3196 /// parseDirectiveCFIRestore 3197 /// ::= .cfi_restore register 3198 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3199 int64_t Register = 0; 3200 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3201 return true; 3202 3203 getStreamer().EmitCFIRestore(Register); 3204 return false; 3205 } 3206 3207 /// parseDirectiveCFIEscape 3208 /// ::= .cfi_escape expression[,...] 3209 bool AsmParser::parseDirectiveCFIEscape() { 3210 std::string Values; 3211 int64_t CurrValue; 3212 if (parseAbsoluteExpression(CurrValue)) 3213 return true; 3214 3215 Values.push_back((uint8_t)CurrValue); 3216 3217 while (getLexer().is(AsmToken::Comma)) { 3218 Lex(); 3219 3220 if (parseAbsoluteExpression(CurrValue)) 3221 return true; 3222 3223 Values.push_back((uint8_t)CurrValue); 3224 } 3225 3226 getStreamer().EmitCFIEscape(Values); 3227 return false; 3228 } 3229 3230 /// parseDirectiveCFISignalFrame 3231 /// ::= .cfi_signal_frame 3232 bool AsmParser::parseDirectiveCFISignalFrame() { 3233 if (getLexer().isNot(AsmToken::EndOfStatement)) 3234 return Error(getLexer().getLoc(), 3235 "unexpected token in '.cfi_signal_frame'"); 3236 3237 getStreamer().EmitCFISignalFrame(); 3238 return false; 3239 } 3240 3241 /// parseDirectiveCFIUndefined 3242 /// ::= .cfi_undefined register 3243 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 3244 int64_t Register = 0; 3245 3246 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3247 return true; 3248 3249 getStreamer().EmitCFIUndefined(Register); 3250 return false; 3251 } 3252 3253 /// parseDirectiveMacrosOnOff 3254 /// ::= .macros_on 3255 /// ::= .macros_off 3256 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 3257 if (getLexer().isNot(AsmToken::EndOfStatement)) 3258 return Error(getLexer().getLoc(), 3259 "unexpected token in '" + Directive + "' directive"); 3260 3261 setMacrosEnabled(Directive == ".macros_on"); 3262 return false; 3263 } 3264 3265 /// parseDirectiveMacro 3266 /// ::= .macro name[,] [parameters] 3267 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 3268 StringRef Name; 3269 if (parseIdentifier(Name)) 3270 return TokError("expected identifier in '.macro' directive"); 3271 3272 if (getLexer().is(AsmToken::Comma)) 3273 Lex(); 3274 3275 MCAsmMacroParameters Parameters; 3276 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3277 3278 if (Parameters.size() && Parameters.back().Vararg) 3279 return Error(Lexer.getLoc(), 3280 "Vararg parameter '" + Parameters.back().Name + 3281 "' should be last one in the list of parameters."); 3282 3283 MCAsmMacroParameter Parameter; 3284 if (parseIdentifier(Parameter.Name)) 3285 return TokError("expected identifier in '.macro' directive"); 3286 3287 if (Lexer.is(AsmToken::Colon)) { 3288 Lex(); // consume ':' 3289 3290 SMLoc QualLoc; 3291 StringRef Qualifier; 3292 3293 QualLoc = Lexer.getLoc(); 3294 if (parseIdentifier(Qualifier)) 3295 return Error(QualLoc, "missing parameter qualifier for " 3296 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3297 3298 if (Qualifier == "req") 3299 Parameter.Required = true; 3300 else if (Qualifier == "vararg") 3301 Parameter.Vararg = true; 3302 else 3303 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 3304 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 3305 } 3306 3307 if (getLexer().is(AsmToken::Equal)) { 3308 Lex(); 3309 3310 SMLoc ParamLoc; 3311 3312 ParamLoc = Lexer.getLoc(); 3313 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 3314 return true; 3315 3316 if (Parameter.Required) 3317 Warning(ParamLoc, "pointless default value for required parameter " 3318 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3319 } 3320 3321 Parameters.push_back(std::move(Parameter)); 3322 3323 if (getLexer().is(AsmToken::Comma)) 3324 Lex(); 3325 } 3326 3327 // Eat the end of statement. 3328 Lex(); 3329 3330 AsmToken EndToken, StartToken = getTok(); 3331 unsigned MacroDepth = 0; 3332 3333 // Lex the macro definition. 3334 for (;;) { 3335 // Check whether we have reached the end of the file. 3336 if (getLexer().is(AsmToken::Eof)) 3337 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 3338 3339 // Otherwise, check whether we have reach the .endmacro. 3340 if (getLexer().is(AsmToken::Identifier)) { 3341 if (getTok().getIdentifier() == ".endm" || 3342 getTok().getIdentifier() == ".endmacro") { 3343 if (MacroDepth == 0) { // Outermost macro. 3344 EndToken = getTok(); 3345 Lex(); 3346 if (getLexer().isNot(AsmToken::EndOfStatement)) 3347 return TokError("unexpected token in '" + EndToken.getIdentifier() + 3348 "' directive"); 3349 break; 3350 } else { 3351 // Otherwise we just found the end of an inner macro. 3352 --MacroDepth; 3353 } 3354 } else if (getTok().getIdentifier() == ".macro") { 3355 // We allow nested macros. Those aren't instantiated until the outermost 3356 // macro is expanded so just ignore them for now. 3357 ++MacroDepth; 3358 } 3359 } 3360 3361 // Otherwise, scan til the end of the statement. 3362 eatToEndOfStatement(); 3363 } 3364 3365 if (lookupMacro(Name)) { 3366 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 3367 } 3368 3369 const char *BodyStart = StartToken.getLoc().getPointer(); 3370 const char *BodyEnd = EndToken.getLoc().getPointer(); 3371 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 3372 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 3373 defineMacro(Name, MCAsmMacro(Name, Body, std::move(Parameters))); 3374 return false; 3375 } 3376 3377 /// checkForBadMacro 3378 /// 3379 /// With the support added for named parameters there may be code out there that 3380 /// is transitioning from positional parameters. In versions of gas that did 3381 /// not support named parameters they would be ignored on the macro definition. 3382 /// But to support both styles of parameters this is not possible so if a macro 3383 /// definition has named parameters but does not use them and has what appears 3384 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 3385 /// warning that the positional parameter found in body which have no effect. 3386 /// Hoping the developer will either remove the named parameters from the macro 3387 /// definition so the positional parameters get used if that was what was 3388 /// intended or change the macro to use the named parameters. It is possible 3389 /// this warning will trigger when the none of the named parameters are used 3390 /// and the strings like $1 are infact to simply to be passed trough unchanged. 3391 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 3392 StringRef Body, 3393 ArrayRef<MCAsmMacroParameter> Parameters) { 3394 // If this macro is not defined with named parameters the warning we are 3395 // checking for here doesn't apply. 3396 unsigned NParameters = Parameters.size(); 3397 if (NParameters == 0) 3398 return; 3399 3400 bool NamedParametersFound = false; 3401 bool PositionalParametersFound = false; 3402 3403 // Look at the body of the macro for use of both the named parameters and what 3404 // are likely to be positional parameters. This is what expandMacro() is 3405 // doing when it finds the parameters in the body. 3406 while (!Body.empty()) { 3407 // Scan for the next possible parameter. 3408 std::size_t End = Body.size(), Pos = 0; 3409 for (; Pos != End; ++Pos) { 3410 // Check for a substitution or escape. 3411 // This macro is defined with parameters, look for \foo, \bar, etc. 3412 if (Body[Pos] == '\\' && Pos + 1 != End) 3413 break; 3414 3415 // This macro should have parameters, but look for $0, $1, ..., $n too. 3416 if (Body[Pos] != '$' || Pos + 1 == End) 3417 continue; 3418 char Next = Body[Pos + 1]; 3419 if (Next == '$' || Next == 'n' || 3420 isdigit(static_cast<unsigned char>(Next))) 3421 break; 3422 } 3423 3424 // Check if we reached the end. 3425 if (Pos == End) 3426 break; 3427 3428 if (Body[Pos] == '$') { 3429 switch (Body[Pos + 1]) { 3430 // $$ => $ 3431 case '$': 3432 break; 3433 3434 // $n => number of arguments 3435 case 'n': 3436 PositionalParametersFound = true; 3437 break; 3438 3439 // $[0-9] => argument 3440 default: { 3441 PositionalParametersFound = true; 3442 break; 3443 } 3444 } 3445 Pos += 2; 3446 } else { 3447 unsigned I = Pos + 1; 3448 while (isIdentifierChar(Body[I]) && I + 1 != End) 3449 ++I; 3450 3451 const char *Begin = Body.data() + Pos + 1; 3452 StringRef Argument(Begin, I - (Pos + 1)); 3453 unsigned Index = 0; 3454 for (; Index < NParameters; ++Index) 3455 if (Parameters[Index].Name == Argument) 3456 break; 3457 3458 if (Index == NParameters) { 3459 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 3460 Pos += 3; 3461 else { 3462 Pos = I; 3463 } 3464 } else { 3465 NamedParametersFound = true; 3466 Pos += 1 + Argument.size(); 3467 } 3468 } 3469 // Update the scan point. 3470 Body = Body.substr(Pos); 3471 } 3472 3473 if (!NamedParametersFound && PositionalParametersFound) 3474 Warning(DirectiveLoc, "macro defined with named parameters which are not " 3475 "used in macro body, possible positional parameter " 3476 "found in body which will have no effect"); 3477 } 3478 3479 /// parseDirectiveExitMacro 3480 /// ::= .exitm 3481 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 3482 if (getLexer().isNot(AsmToken::EndOfStatement)) 3483 return TokError("unexpected token in '" + Directive + "' directive"); 3484 3485 if (!isInsideMacroInstantiation()) 3486 return TokError("unexpected '" + Directive + "' in file, " 3487 "no current macro definition"); 3488 3489 // Exit all conditionals that are active in the current macro. 3490 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 3491 TheCondState = TheCondStack.back(); 3492 TheCondStack.pop_back(); 3493 } 3494 3495 handleMacroExit(); 3496 return false; 3497 } 3498 3499 /// parseDirectiveEndMacro 3500 /// ::= .endm 3501 /// ::= .endmacro 3502 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 3503 if (getLexer().isNot(AsmToken::EndOfStatement)) 3504 return TokError("unexpected token in '" + Directive + "' directive"); 3505 3506 // If we are inside a macro instantiation, terminate the current 3507 // instantiation. 3508 if (isInsideMacroInstantiation()) { 3509 handleMacroExit(); 3510 return false; 3511 } 3512 3513 // Otherwise, this .endmacro is a stray entry in the file; well formed 3514 // .endmacro directives are handled during the macro definition parsing. 3515 return TokError("unexpected '" + Directive + "' in file, " 3516 "no current macro definition"); 3517 } 3518 3519 /// parseDirectivePurgeMacro 3520 /// ::= .purgem 3521 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 3522 StringRef Name; 3523 if (parseIdentifier(Name)) 3524 return TokError("expected identifier in '.purgem' directive"); 3525 3526 if (getLexer().isNot(AsmToken::EndOfStatement)) 3527 return TokError("unexpected token in '.purgem' directive"); 3528 3529 if (!lookupMacro(Name)) 3530 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 3531 3532 undefineMacro(Name); 3533 return false; 3534 } 3535 3536 /// parseDirectiveBundleAlignMode 3537 /// ::= {.bundle_align_mode} expression 3538 bool AsmParser::parseDirectiveBundleAlignMode() { 3539 checkForValidSection(); 3540 3541 // Expect a single argument: an expression that evaluates to a constant 3542 // in the inclusive range 0-30. 3543 SMLoc ExprLoc = getLexer().getLoc(); 3544 int64_t AlignSizePow2; 3545 if (parseAbsoluteExpression(AlignSizePow2)) 3546 return true; 3547 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3548 return TokError("unexpected token after expression in" 3549 " '.bundle_align_mode' directive"); 3550 else if (AlignSizePow2 < 0 || AlignSizePow2 > 30) 3551 return Error(ExprLoc, 3552 "invalid bundle alignment size (expected between 0 and 30)"); 3553 3554 Lex(); 3555 3556 // Because of AlignSizePow2's verified range we can safely truncate it to 3557 // unsigned. 3558 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 3559 return false; 3560 } 3561 3562 /// parseDirectiveBundleLock 3563 /// ::= {.bundle_lock} [align_to_end] 3564 bool AsmParser::parseDirectiveBundleLock() { 3565 checkForValidSection(); 3566 bool AlignToEnd = false; 3567 3568 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3569 StringRef Option; 3570 SMLoc Loc = getTok().getLoc(); 3571 const char *kInvalidOptionError = 3572 "invalid option for '.bundle_lock' directive"; 3573 3574 if (parseIdentifier(Option)) 3575 return Error(Loc, kInvalidOptionError); 3576 3577 if (Option != "align_to_end") 3578 return Error(Loc, kInvalidOptionError); 3579 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3580 return Error(Loc, 3581 "unexpected token after '.bundle_lock' directive option"); 3582 AlignToEnd = true; 3583 } 3584 3585 Lex(); 3586 3587 getStreamer().EmitBundleLock(AlignToEnd); 3588 return false; 3589 } 3590 3591 /// parseDirectiveBundleLock 3592 /// ::= {.bundle_lock} 3593 bool AsmParser::parseDirectiveBundleUnlock() { 3594 checkForValidSection(); 3595 3596 if (getLexer().isNot(AsmToken::EndOfStatement)) 3597 return TokError("unexpected token in '.bundle_unlock' directive"); 3598 Lex(); 3599 3600 getStreamer().EmitBundleUnlock(); 3601 return false; 3602 } 3603 3604 /// parseDirectiveSpace 3605 /// ::= (.skip | .space) expression [ , expression ] 3606 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 3607 checkForValidSection(); 3608 3609 int64_t NumBytes; 3610 if (parseAbsoluteExpression(NumBytes)) 3611 return true; 3612 3613 int64_t FillExpr = 0; 3614 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3615 if (getLexer().isNot(AsmToken::Comma)) 3616 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 3617 Lex(); 3618 3619 if (parseAbsoluteExpression(FillExpr)) 3620 return true; 3621 3622 if (getLexer().isNot(AsmToken::EndOfStatement)) 3623 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 3624 } 3625 3626 Lex(); 3627 3628 if (NumBytes <= 0) 3629 return TokError("invalid number of bytes in '" + Twine(IDVal) + 3630 "' directive"); 3631 3632 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 3633 getStreamer().EmitFill(NumBytes, FillExpr); 3634 3635 return false; 3636 } 3637 3638 /// parseDirectiveLEB128 3639 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 3640 bool AsmParser::parseDirectiveLEB128(bool Signed) { 3641 checkForValidSection(); 3642 const MCExpr *Value; 3643 3644 for (;;) { 3645 if (parseExpression(Value)) 3646 return true; 3647 3648 if (Signed) 3649 getStreamer().EmitSLEB128Value(Value); 3650 else 3651 getStreamer().EmitULEB128Value(Value); 3652 3653 if (getLexer().is(AsmToken::EndOfStatement)) 3654 break; 3655 3656 if (getLexer().isNot(AsmToken::Comma)) 3657 return TokError("unexpected token in directive"); 3658 Lex(); 3659 } 3660 3661 return false; 3662 } 3663 3664 /// parseDirectiveSymbolAttribute 3665 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 3666 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 3667 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3668 for (;;) { 3669 StringRef Name; 3670 SMLoc Loc = getTok().getLoc(); 3671 3672 if (parseIdentifier(Name)) 3673 return Error(Loc, "expected identifier in directive"); 3674 3675 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3676 3677 // Assembler local symbols don't make any sense here. Complain loudly. 3678 if (Sym->isTemporary()) 3679 return Error(Loc, "non-local symbol required in directive"); 3680 3681 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 3682 return Error(Loc, "unable to emit symbol attribute"); 3683 3684 if (getLexer().is(AsmToken::EndOfStatement)) 3685 break; 3686 3687 if (getLexer().isNot(AsmToken::Comma)) 3688 return TokError("unexpected token in directive"); 3689 Lex(); 3690 } 3691 } 3692 3693 Lex(); 3694 return false; 3695 } 3696 3697 /// parseDirectiveComm 3698 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 3699 bool AsmParser::parseDirectiveComm(bool IsLocal) { 3700 checkForValidSection(); 3701 3702 SMLoc IDLoc = getLexer().getLoc(); 3703 StringRef Name; 3704 if (parseIdentifier(Name)) 3705 return TokError("expected identifier in directive"); 3706 3707 // Handle the identifier as the key symbol. 3708 MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); 3709 3710 if (getLexer().isNot(AsmToken::Comma)) 3711 return TokError("unexpected token in directive"); 3712 Lex(); 3713 3714 int64_t Size; 3715 SMLoc SizeLoc = getLexer().getLoc(); 3716 if (parseAbsoluteExpression(Size)) 3717 return true; 3718 3719 int64_t Pow2Alignment = 0; 3720 SMLoc Pow2AlignmentLoc; 3721 if (getLexer().is(AsmToken::Comma)) { 3722 Lex(); 3723 Pow2AlignmentLoc = getLexer().getLoc(); 3724 if (parseAbsoluteExpression(Pow2Alignment)) 3725 return true; 3726 3727 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 3728 if (IsLocal && LCOMM == LCOMM::NoAlignment) 3729 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 3730 3731 // If this target takes alignments in bytes (not log) validate and convert. 3732 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 3733 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 3734 if (!isPowerOf2_64(Pow2Alignment)) 3735 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 3736 Pow2Alignment = Log2_64(Pow2Alignment); 3737 } 3738 } 3739 3740 if (getLexer().isNot(AsmToken::EndOfStatement)) 3741 return TokError("unexpected token in '.comm' or '.lcomm' directive"); 3742 3743 Lex(); 3744 3745 // NOTE: a size of zero for a .comm should create a undefined symbol 3746 // but a size of .lcomm creates a bss symbol of size zero. 3747 if (Size < 0) 3748 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 3749 "be less than zero"); 3750 3751 // NOTE: The alignment in the directive is a power of 2 value, the assembler 3752 // may internally end up wanting an alignment in bytes. 3753 // FIXME: Diagnose overflow. 3754 if (Pow2Alignment < 0) 3755 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 3756 "alignment, can't be less than zero"); 3757 3758 if (!Sym->isUndefined()) 3759 return Error(IDLoc, "invalid symbol redefinition"); 3760 3761 // Create the Symbol as a common or local common with Size and Pow2Alignment 3762 if (IsLocal) { 3763 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 3764 return false; 3765 } 3766 3767 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 3768 return false; 3769 } 3770 3771 /// parseDirectiveAbort 3772 /// ::= .abort [... message ...] 3773 bool AsmParser::parseDirectiveAbort() { 3774 // FIXME: Use loc from directive. 3775 SMLoc Loc = getLexer().getLoc(); 3776 3777 StringRef Str = parseStringToEndOfStatement(); 3778 if (getLexer().isNot(AsmToken::EndOfStatement)) 3779 return TokError("unexpected token in '.abort' directive"); 3780 3781 Lex(); 3782 3783 if (Str.empty()) 3784 Error(Loc, ".abort detected. Assembly stopping."); 3785 else 3786 Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 3787 // FIXME: Actually abort assembly here. 3788 3789 return false; 3790 } 3791 3792 /// parseDirectiveInclude 3793 /// ::= .include "filename" 3794 bool AsmParser::parseDirectiveInclude() { 3795 if (getLexer().isNot(AsmToken::String)) 3796 return TokError("expected string in '.include' directive"); 3797 3798 // Allow the strings to have escaped octal character sequence. 3799 std::string Filename; 3800 if (parseEscapedString(Filename)) 3801 return true; 3802 SMLoc IncludeLoc = getLexer().getLoc(); 3803 Lex(); 3804 3805 if (getLexer().isNot(AsmToken::EndOfStatement)) 3806 return TokError("unexpected token in '.include' directive"); 3807 3808 // Attempt to switch the lexer to the included file before consuming the end 3809 // of statement to avoid losing it when we switch. 3810 if (enterIncludeFile(Filename)) { 3811 Error(IncludeLoc, "Could not find include file '" + Filename + "'"); 3812 return true; 3813 } 3814 3815 return false; 3816 } 3817 3818 /// parseDirectiveIncbin 3819 /// ::= .incbin "filename" 3820 bool AsmParser::parseDirectiveIncbin() { 3821 if (getLexer().isNot(AsmToken::String)) 3822 return TokError("expected string in '.incbin' directive"); 3823 3824 // Allow the strings to have escaped octal character sequence. 3825 std::string Filename; 3826 if (parseEscapedString(Filename)) 3827 return true; 3828 SMLoc IncbinLoc = getLexer().getLoc(); 3829 Lex(); 3830 3831 if (getLexer().isNot(AsmToken::EndOfStatement)) 3832 return TokError("unexpected token in '.incbin' directive"); 3833 3834 // Attempt to process the included file. 3835 if (processIncbinFile(Filename)) { 3836 Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 3837 return true; 3838 } 3839 3840 return false; 3841 } 3842 3843 /// parseDirectiveIf 3844 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 3845 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 3846 TheCondStack.push_back(TheCondState); 3847 TheCondState.TheCond = AsmCond::IfCond; 3848 if (TheCondState.Ignore) { 3849 eatToEndOfStatement(); 3850 } else { 3851 int64_t ExprValue; 3852 if (parseAbsoluteExpression(ExprValue)) 3853 return true; 3854 3855 if (getLexer().isNot(AsmToken::EndOfStatement)) 3856 return TokError("unexpected token in '.if' directive"); 3857 3858 Lex(); 3859 3860 switch (DirKind) { 3861 default: 3862 llvm_unreachable("unsupported directive"); 3863 case DK_IF: 3864 case DK_IFNE: 3865 break; 3866 case DK_IFEQ: 3867 ExprValue = ExprValue == 0; 3868 break; 3869 case DK_IFGE: 3870 ExprValue = ExprValue >= 0; 3871 break; 3872 case DK_IFGT: 3873 ExprValue = ExprValue > 0; 3874 break; 3875 case DK_IFLE: 3876 ExprValue = ExprValue <= 0; 3877 break; 3878 case DK_IFLT: 3879 ExprValue = ExprValue < 0; 3880 break; 3881 } 3882 3883 TheCondState.CondMet = ExprValue; 3884 TheCondState.Ignore = !TheCondState.CondMet; 3885 } 3886 3887 return false; 3888 } 3889 3890 /// parseDirectiveIfb 3891 /// ::= .ifb string 3892 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 3893 TheCondStack.push_back(TheCondState); 3894 TheCondState.TheCond = AsmCond::IfCond; 3895 3896 if (TheCondState.Ignore) { 3897 eatToEndOfStatement(); 3898 } else { 3899 StringRef Str = parseStringToEndOfStatement(); 3900 3901 if (getLexer().isNot(AsmToken::EndOfStatement)) 3902 return TokError("unexpected token in '.ifb' directive"); 3903 3904 Lex(); 3905 3906 TheCondState.CondMet = ExpectBlank == Str.empty(); 3907 TheCondState.Ignore = !TheCondState.CondMet; 3908 } 3909 3910 return false; 3911 } 3912 3913 /// parseDirectiveIfc 3914 /// ::= .ifc string1, string2 3915 /// ::= .ifnc string1, string2 3916 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 3917 TheCondStack.push_back(TheCondState); 3918 TheCondState.TheCond = AsmCond::IfCond; 3919 3920 if (TheCondState.Ignore) { 3921 eatToEndOfStatement(); 3922 } else { 3923 StringRef Str1 = parseStringToComma(); 3924 3925 if (getLexer().isNot(AsmToken::Comma)) 3926 return TokError("unexpected token in '.ifc' directive"); 3927 3928 Lex(); 3929 3930 StringRef Str2 = parseStringToEndOfStatement(); 3931 3932 if (getLexer().isNot(AsmToken::EndOfStatement)) 3933 return TokError("unexpected token in '.ifc' directive"); 3934 3935 Lex(); 3936 3937 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 3938 TheCondState.Ignore = !TheCondState.CondMet; 3939 } 3940 3941 return false; 3942 } 3943 3944 /// parseDirectiveIfeqs 3945 /// ::= .ifeqs string1, string2 3946 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc) { 3947 if (Lexer.isNot(AsmToken::String)) { 3948 TokError("expected string parameter for '.ifeqs' directive"); 3949 eatToEndOfStatement(); 3950 return true; 3951 } 3952 3953 StringRef String1 = getTok().getStringContents(); 3954 Lex(); 3955 3956 if (Lexer.isNot(AsmToken::Comma)) { 3957 TokError("expected comma after first string for '.ifeqs' directive"); 3958 eatToEndOfStatement(); 3959 return true; 3960 } 3961 3962 Lex(); 3963 3964 if (Lexer.isNot(AsmToken::String)) { 3965 TokError("expected string parameter for '.ifeqs' directive"); 3966 eatToEndOfStatement(); 3967 return true; 3968 } 3969 3970 StringRef String2 = getTok().getStringContents(); 3971 Lex(); 3972 3973 TheCondStack.push_back(TheCondState); 3974 TheCondState.TheCond = AsmCond::IfCond; 3975 TheCondState.CondMet = String1 == String2; 3976 TheCondState.Ignore = !TheCondState.CondMet; 3977 3978 return false; 3979 } 3980 3981 /// parseDirectiveIfdef 3982 /// ::= .ifdef symbol 3983 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 3984 StringRef Name; 3985 TheCondStack.push_back(TheCondState); 3986 TheCondState.TheCond = AsmCond::IfCond; 3987 3988 if (TheCondState.Ignore) { 3989 eatToEndOfStatement(); 3990 } else { 3991 if (parseIdentifier(Name)) 3992 return TokError("expected identifier after '.ifdef'"); 3993 3994 Lex(); 3995 3996 MCSymbol *Sym = getContext().LookupSymbol(Name); 3997 3998 if (expect_defined) 3999 TheCondState.CondMet = (Sym && !Sym->isUndefined()); 4000 else 4001 TheCondState.CondMet = (!Sym || Sym->isUndefined()); 4002 TheCondState.Ignore = !TheCondState.CondMet; 4003 } 4004 4005 return false; 4006 } 4007 4008 /// parseDirectiveElseIf 4009 /// ::= .elseif expression 4010 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 4011 if (TheCondState.TheCond != AsmCond::IfCond && 4012 TheCondState.TheCond != AsmCond::ElseIfCond) 4013 Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " 4014 " an .elseif"); 4015 TheCondState.TheCond = AsmCond::ElseIfCond; 4016 4017 bool LastIgnoreState = false; 4018 if (!TheCondStack.empty()) 4019 LastIgnoreState = TheCondStack.back().Ignore; 4020 if (LastIgnoreState || TheCondState.CondMet) { 4021 TheCondState.Ignore = true; 4022 eatToEndOfStatement(); 4023 } else { 4024 int64_t ExprValue; 4025 if (parseAbsoluteExpression(ExprValue)) 4026 return true; 4027 4028 if (getLexer().isNot(AsmToken::EndOfStatement)) 4029 return TokError("unexpected token in '.elseif' directive"); 4030 4031 Lex(); 4032 TheCondState.CondMet = ExprValue; 4033 TheCondState.Ignore = !TheCondState.CondMet; 4034 } 4035 4036 return false; 4037 } 4038 4039 /// parseDirectiveElse 4040 /// ::= .else 4041 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4042 if (getLexer().isNot(AsmToken::EndOfStatement)) 4043 return TokError("unexpected token in '.else' directive"); 4044 4045 Lex(); 4046 4047 if (TheCondState.TheCond != AsmCond::IfCond && 4048 TheCondState.TheCond != AsmCond::ElseIfCond) 4049 Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " 4050 ".elseif"); 4051 TheCondState.TheCond = AsmCond::ElseCond; 4052 bool LastIgnoreState = false; 4053 if (!TheCondStack.empty()) 4054 LastIgnoreState = TheCondStack.back().Ignore; 4055 if (LastIgnoreState || TheCondState.CondMet) 4056 TheCondState.Ignore = true; 4057 else 4058 TheCondState.Ignore = false; 4059 4060 return false; 4061 } 4062 4063 /// parseDirectiveEnd 4064 /// ::= .end 4065 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4066 if (getLexer().isNot(AsmToken::EndOfStatement)) 4067 return TokError("unexpected token in '.end' directive"); 4068 4069 Lex(); 4070 4071 while (Lexer.isNot(AsmToken::Eof)) 4072 Lex(); 4073 4074 return false; 4075 } 4076 4077 /// parseDirectiveError 4078 /// ::= .err 4079 /// ::= .error [string] 4080 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 4081 if (!TheCondStack.empty()) { 4082 if (TheCondStack.back().Ignore) { 4083 eatToEndOfStatement(); 4084 return false; 4085 } 4086 } 4087 4088 if (!WithMessage) 4089 return Error(L, ".err encountered"); 4090 4091 StringRef Message = ".error directive invoked in source file"; 4092 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4093 if (Lexer.isNot(AsmToken::String)) { 4094 TokError(".error argument must be a string"); 4095 eatToEndOfStatement(); 4096 return true; 4097 } 4098 4099 Message = getTok().getStringContents(); 4100 Lex(); 4101 } 4102 4103 Error(L, Message); 4104 return true; 4105 } 4106 4107 /// parseDirectiveWarning 4108 /// ::= .warning [string] 4109 bool AsmParser::parseDirectiveWarning(SMLoc L) { 4110 if (!TheCondStack.empty()) { 4111 if (TheCondStack.back().Ignore) { 4112 eatToEndOfStatement(); 4113 return false; 4114 } 4115 } 4116 4117 StringRef Message = ".warning directive invoked in source file"; 4118 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4119 if (Lexer.isNot(AsmToken::String)) { 4120 TokError(".warning argument must be a string"); 4121 eatToEndOfStatement(); 4122 return true; 4123 } 4124 4125 Message = getTok().getStringContents(); 4126 Lex(); 4127 } 4128 4129 Warning(L, Message); 4130 return false; 4131 } 4132 4133 /// parseDirectiveEndIf 4134 /// ::= .endif 4135 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 4136 if (getLexer().isNot(AsmToken::EndOfStatement)) 4137 return TokError("unexpected token in '.endif' directive"); 4138 4139 Lex(); 4140 4141 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 4142 Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " 4143 ".else"); 4144 if (!TheCondStack.empty()) { 4145 TheCondState = TheCondStack.back(); 4146 TheCondStack.pop_back(); 4147 } 4148 4149 return false; 4150 } 4151 4152 void AsmParser::initializeDirectiveKindMap() { 4153 DirectiveKindMap[".set"] = DK_SET; 4154 DirectiveKindMap[".equ"] = DK_EQU; 4155 DirectiveKindMap[".equiv"] = DK_EQUIV; 4156 DirectiveKindMap[".ascii"] = DK_ASCII; 4157 DirectiveKindMap[".asciz"] = DK_ASCIZ; 4158 DirectiveKindMap[".string"] = DK_STRING; 4159 DirectiveKindMap[".byte"] = DK_BYTE; 4160 DirectiveKindMap[".short"] = DK_SHORT; 4161 DirectiveKindMap[".value"] = DK_VALUE; 4162 DirectiveKindMap[".2byte"] = DK_2BYTE; 4163 DirectiveKindMap[".long"] = DK_LONG; 4164 DirectiveKindMap[".int"] = DK_INT; 4165 DirectiveKindMap[".4byte"] = DK_4BYTE; 4166 DirectiveKindMap[".quad"] = DK_QUAD; 4167 DirectiveKindMap[".8byte"] = DK_8BYTE; 4168 DirectiveKindMap[".octa"] = DK_OCTA; 4169 DirectiveKindMap[".single"] = DK_SINGLE; 4170 DirectiveKindMap[".float"] = DK_FLOAT; 4171 DirectiveKindMap[".double"] = DK_DOUBLE; 4172 DirectiveKindMap[".align"] = DK_ALIGN; 4173 DirectiveKindMap[".align32"] = DK_ALIGN32; 4174 DirectiveKindMap[".balign"] = DK_BALIGN; 4175 DirectiveKindMap[".balignw"] = DK_BALIGNW; 4176 DirectiveKindMap[".balignl"] = DK_BALIGNL; 4177 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 4178 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 4179 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 4180 DirectiveKindMap[".org"] = DK_ORG; 4181 DirectiveKindMap[".fill"] = DK_FILL; 4182 DirectiveKindMap[".zero"] = DK_ZERO; 4183 DirectiveKindMap[".extern"] = DK_EXTERN; 4184 DirectiveKindMap[".globl"] = DK_GLOBL; 4185 DirectiveKindMap[".global"] = DK_GLOBAL; 4186 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 4187 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 4188 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 4189 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 4190 DirectiveKindMap[".reference"] = DK_REFERENCE; 4191 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 4192 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 4193 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 4194 DirectiveKindMap[".comm"] = DK_COMM; 4195 DirectiveKindMap[".common"] = DK_COMMON; 4196 DirectiveKindMap[".lcomm"] = DK_LCOMM; 4197 DirectiveKindMap[".abort"] = DK_ABORT; 4198 DirectiveKindMap[".include"] = DK_INCLUDE; 4199 DirectiveKindMap[".incbin"] = DK_INCBIN; 4200 DirectiveKindMap[".code16"] = DK_CODE16; 4201 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 4202 DirectiveKindMap[".rept"] = DK_REPT; 4203 DirectiveKindMap[".rep"] = DK_REPT; 4204 DirectiveKindMap[".irp"] = DK_IRP; 4205 DirectiveKindMap[".irpc"] = DK_IRPC; 4206 DirectiveKindMap[".endr"] = DK_ENDR; 4207 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 4208 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 4209 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 4210 DirectiveKindMap[".if"] = DK_IF; 4211 DirectiveKindMap[".ifeq"] = DK_IFEQ; 4212 DirectiveKindMap[".ifge"] = DK_IFGE; 4213 DirectiveKindMap[".ifgt"] = DK_IFGT; 4214 DirectiveKindMap[".ifle"] = DK_IFLE; 4215 DirectiveKindMap[".iflt"] = DK_IFLT; 4216 DirectiveKindMap[".ifne"] = DK_IFNE; 4217 DirectiveKindMap[".ifb"] = DK_IFB; 4218 DirectiveKindMap[".ifnb"] = DK_IFNB; 4219 DirectiveKindMap[".ifc"] = DK_IFC; 4220 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 4221 DirectiveKindMap[".ifnc"] = DK_IFNC; 4222 DirectiveKindMap[".ifdef"] = DK_IFDEF; 4223 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 4224 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 4225 DirectiveKindMap[".elseif"] = DK_ELSEIF; 4226 DirectiveKindMap[".else"] = DK_ELSE; 4227 DirectiveKindMap[".end"] = DK_END; 4228 DirectiveKindMap[".endif"] = DK_ENDIF; 4229 DirectiveKindMap[".skip"] = DK_SKIP; 4230 DirectiveKindMap[".space"] = DK_SPACE; 4231 DirectiveKindMap[".file"] = DK_FILE; 4232 DirectiveKindMap[".line"] = DK_LINE; 4233 DirectiveKindMap[".loc"] = DK_LOC; 4234 DirectiveKindMap[".stabs"] = DK_STABS; 4235 DirectiveKindMap[".sleb128"] = DK_SLEB128; 4236 DirectiveKindMap[".uleb128"] = DK_ULEB128; 4237 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 4238 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 4239 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 4240 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 4241 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 4242 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 4243 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 4244 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 4245 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 4246 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 4247 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 4248 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 4249 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 4250 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 4251 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 4252 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 4253 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 4254 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 4255 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 4256 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 4257 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 4258 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 4259 DirectiveKindMap[".macro"] = DK_MACRO; 4260 DirectiveKindMap[".exitm"] = DK_EXITM; 4261 DirectiveKindMap[".endm"] = DK_ENDM; 4262 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 4263 DirectiveKindMap[".purgem"] = DK_PURGEM; 4264 DirectiveKindMap[".err"] = DK_ERR; 4265 DirectiveKindMap[".error"] = DK_ERROR; 4266 DirectiveKindMap[".warning"] = DK_WARNING; 4267 } 4268 4269 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 4270 AsmToken EndToken, StartToken = getTok(); 4271 4272 unsigned NestLevel = 0; 4273 for (;;) { 4274 // Check whether we have reached the end of the file. 4275 if (getLexer().is(AsmToken::Eof)) { 4276 Error(DirectiveLoc, "no matching '.endr' in definition"); 4277 return nullptr; 4278 } 4279 4280 if (Lexer.is(AsmToken::Identifier) && 4281 (getTok().getIdentifier() == ".rept")) { 4282 ++NestLevel; 4283 } 4284 4285 // Otherwise, check whether we have reached the .endr. 4286 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 4287 if (NestLevel == 0) { 4288 EndToken = getTok(); 4289 Lex(); 4290 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4291 TokError("unexpected token in '.endr' directive"); 4292 return nullptr; 4293 } 4294 break; 4295 } 4296 --NestLevel; 4297 } 4298 4299 // Otherwise, scan till the end of the statement. 4300 eatToEndOfStatement(); 4301 } 4302 4303 const char *BodyStart = StartToken.getLoc().getPointer(); 4304 const char *BodyEnd = EndToken.getLoc().getPointer(); 4305 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4306 4307 // We Are Anonymous. 4308 MacroLikeBodies.push_back( 4309 MCAsmMacro(StringRef(), Body, MCAsmMacroParameters())); 4310 return &MacroLikeBodies.back(); 4311 } 4312 4313 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 4314 raw_svector_ostream &OS) { 4315 OS << ".endr\n"; 4316 4317 std::unique_ptr<MemoryBuffer> Instantiation = 4318 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 4319 4320 // Create the macro instantiation object and add to the current macro 4321 // instantiation stack. 4322 MacroInstantiation *MI = new MacroInstantiation( 4323 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 4324 ActiveMacros.push_back(MI); 4325 4326 // Jump to the macro instantiation and prime the lexer. 4327 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 4328 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 4329 Lex(); 4330 } 4331 4332 /// parseDirectiveRept 4333 /// ::= .rep | .rept count 4334 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 4335 const MCExpr *CountExpr; 4336 SMLoc CountLoc = getTok().getLoc(); 4337 if (parseExpression(CountExpr)) 4338 return true; 4339 4340 int64_t Count; 4341 if (!CountExpr->EvaluateAsAbsolute(Count)) { 4342 eatToEndOfStatement(); 4343 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 4344 } 4345 4346 if (Count < 0) 4347 return Error(CountLoc, "Count is negative"); 4348 4349 if (Lexer.isNot(AsmToken::EndOfStatement)) 4350 return TokError("unexpected token in '" + Dir + "' directive"); 4351 4352 // Eat the end of statement. 4353 Lex(); 4354 4355 // Lex the rept definition. 4356 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4357 if (!M) 4358 return true; 4359 4360 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4361 // to hold the macro body with substitutions. 4362 SmallString<256> Buf; 4363 raw_svector_ostream OS(Buf); 4364 while (Count--) { 4365 if (expandMacro(OS, M->Body, None, None, getTok().getLoc())) 4366 return true; 4367 } 4368 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4369 4370 return false; 4371 } 4372 4373 /// parseDirectiveIrp 4374 /// ::= .irp symbol,values 4375 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 4376 MCAsmMacroParameter Parameter; 4377 4378 if (parseIdentifier(Parameter.Name)) 4379 return TokError("expected identifier in '.irp' directive"); 4380 4381 if (Lexer.isNot(AsmToken::Comma)) 4382 return TokError("expected comma in '.irp' directive"); 4383 4384 Lex(); 4385 4386 MCAsmMacroArguments A; 4387 if (parseMacroArguments(nullptr, A)) 4388 return true; 4389 4390 // Eat the end of statement. 4391 Lex(); 4392 4393 // Lex the irp definition. 4394 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4395 if (!M) 4396 return true; 4397 4398 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4399 // to hold the macro body with substitutions. 4400 SmallString<256> Buf; 4401 raw_svector_ostream OS(Buf); 4402 4403 for (MCAsmMacroArguments::iterator i = A.begin(), e = A.end(); i != e; ++i) { 4404 if (expandMacro(OS, M->Body, Parameter, *i, getTok().getLoc())) 4405 return true; 4406 } 4407 4408 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4409 4410 return false; 4411 } 4412 4413 /// parseDirectiveIrpc 4414 /// ::= .irpc symbol,values 4415 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 4416 MCAsmMacroParameter Parameter; 4417 4418 if (parseIdentifier(Parameter.Name)) 4419 return TokError("expected identifier in '.irpc' directive"); 4420 4421 if (Lexer.isNot(AsmToken::Comma)) 4422 return TokError("expected comma in '.irpc' directive"); 4423 4424 Lex(); 4425 4426 MCAsmMacroArguments A; 4427 if (parseMacroArguments(nullptr, A)) 4428 return true; 4429 4430 if (A.size() != 1 || A.front().size() != 1) 4431 return TokError("unexpected token in '.irpc' directive"); 4432 4433 // Eat the end of statement. 4434 Lex(); 4435 4436 // Lex the irpc definition. 4437 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4438 if (!M) 4439 return true; 4440 4441 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4442 // to hold the macro body with substitutions. 4443 SmallString<256> Buf; 4444 raw_svector_ostream OS(Buf); 4445 4446 StringRef Values = A.front().front().getString(); 4447 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 4448 MCAsmMacroArgument Arg; 4449 Arg.push_back(AsmToken(AsmToken::Identifier, Values.slice(I, I + 1))); 4450 4451 if (expandMacro(OS, M->Body, Parameter, Arg, getTok().getLoc())) 4452 return true; 4453 } 4454 4455 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4456 4457 return false; 4458 } 4459 4460 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 4461 if (ActiveMacros.empty()) 4462 return TokError("unmatched '.endr' directive"); 4463 4464 // The only .repl that should get here are the ones created by 4465 // instantiateMacroLikeBody. 4466 assert(getLexer().is(AsmToken::EndOfStatement)); 4467 4468 handleMacroExit(); 4469 return false; 4470 } 4471 4472 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 4473 size_t Len) { 4474 const MCExpr *Value; 4475 SMLoc ExprLoc = getLexer().getLoc(); 4476 if (parseExpression(Value)) 4477 return true; 4478 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4479 if (!MCE) 4480 return Error(ExprLoc, "unexpected expression in _emit"); 4481 uint64_t IntValue = MCE->getValue(); 4482 if (!isUIntN(8, IntValue) && !isIntN(8, IntValue)) 4483 return Error(ExprLoc, "literal value out of range for directive"); 4484 4485 Info.AsmRewrites->push_back(AsmRewrite(AOK_Emit, IDLoc, Len)); 4486 return false; 4487 } 4488 4489 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 4490 const MCExpr *Value; 4491 SMLoc ExprLoc = getLexer().getLoc(); 4492 if (parseExpression(Value)) 4493 return true; 4494 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4495 if (!MCE) 4496 return Error(ExprLoc, "unexpected expression in align"); 4497 uint64_t IntValue = MCE->getValue(); 4498 if (!isPowerOf2_64(IntValue)) 4499 return Error(ExprLoc, "literal value not a power of two greater then zero"); 4500 4501 Info.AsmRewrites->push_back( 4502 AsmRewrite(AOK_Align, IDLoc, 5, Log2_64(IntValue))); 4503 return false; 4504 } 4505 4506 // We are comparing pointers, but the pointers are relative to a single string. 4507 // Thus, this should always be deterministic. 4508 static int rewritesSort(const AsmRewrite *AsmRewriteA, 4509 const AsmRewrite *AsmRewriteB) { 4510 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 4511 return -1; 4512 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 4513 return 1; 4514 4515 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 4516 // rewrite to the same location. Make sure the SizeDirective rewrite is 4517 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 4518 // ensures the sort algorithm is stable. 4519 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 4520 AsmRewritePrecedence[AsmRewriteB->Kind]) 4521 return -1; 4522 4523 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 4524 AsmRewritePrecedence[AsmRewriteB->Kind]) 4525 return 1; 4526 llvm_unreachable("Unstable rewrite sort."); 4527 } 4528 4529 bool AsmParser::parseMSInlineAsm( 4530 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 4531 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool> > &OpDecls, 4532 SmallVectorImpl<std::string> &Constraints, 4533 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 4534 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 4535 SmallVector<void *, 4> InputDecls; 4536 SmallVector<void *, 4> OutputDecls; 4537 SmallVector<bool, 4> InputDeclsAddressOf; 4538 SmallVector<bool, 4> OutputDeclsAddressOf; 4539 SmallVector<std::string, 4> InputConstraints; 4540 SmallVector<std::string, 4> OutputConstraints; 4541 SmallVector<unsigned, 4> ClobberRegs; 4542 4543 SmallVector<AsmRewrite, 4> AsmStrRewrites; 4544 4545 // Prime the lexer. 4546 Lex(); 4547 4548 // While we have input, parse each statement. 4549 unsigned InputIdx = 0; 4550 unsigned OutputIdx = 0; 4551 while (getLexer().isNot(AsmToken::Eof)) { 4552 ParseStatementInfo Info(&AsmStrRewrites); 4553 if (parseStatement(Info, &SI)) 4554 return true; 4555 4556 if (Info.ParseError) 4557 return true; 4558 4559 if (Info.Opcode == ~0U) 4560 continue; 4561 4562 const MCInstrDesc &Desc = MII->get(Info.Opcode); 4563 4564 // Build the list of clobbers, outputs and inputs. 4565 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 4566 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 4567 4568 // Immediate. 4569 if (Operand.isImm()) 4570 continue; 4571 4572 // Register operand. 4573 if (Operand.isReg() && !Operand.needAddressOf() && 4574 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 4575 unsigned NumDefs = Desc.getNumDefs(); 4576 // Clobber. 4577 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 4578 ClobberRegs.push_back(Operand.getReg()); 4579 continue; 4580 } 4581 4582 // Expr/Input or Output. 4583 StringRef SymName = Operand.getSymName(); 4584 if (SymName.empty()) 4585 continue; 4586 4587 void *OpDecl = Operand.getOpDecl(); 4588 if (!OpDecl) 4589 continue; 4590 4591 bool isOutput = (i == 1) && Desc.mayStore(); 4592 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 4593 if (isOutput) { 4594 ++InputIdx; 4595 OutputDecls.push_back(OpDecl); 4596 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 4597 OutputConstraints.push_back('=' + Operand.getConstraint().str()); 4598 AsmStrRewrites.push_back(AsmRewrite(AOK_Output, Start, SymName.size())); 4599 } else { 4600 InputDecls.push_back(OpDecl); 4601 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 4602 InputConstraints.push_back(Operand.getConstraint().str()); 4603 AsmStrRewrites.push_back(AsmRewrite(AOK_Input, Start, SymName.size())); 4604 } 4605 } 4606 4607 // Consider implicit defs to be clobbers. Think of cpuid and push. 4608 ArrayRef<uint16_t> ImpDefs(Desc.getImplicitDefs(), 4609 Desc.getNumImplicitDefs()); 4610 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 4611 } 4612 4613 // Set the number of Outputs and Inputs. 4614 NumOutputs = OutputDecls.size(); 4615 NumInputs = InputDecls.size(); 4616 4617 // Set the unique clobbers. 4618 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 4619 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 4620 ClobberRegs.end()); 4621 Clobbers.assign(ClobberRegs.size(), std::string()); 4622 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 4623 raw_string_ostream OS(Clobbers[I]); 4624 IP->printRegName(OS, ClobberRegs[I]); 4625 } 4626 4627 // Merge the various outputs and inputs. Output are expected first. 4628 if (NumOutputs || NumInputs) { 4629 unsigned NumExprs = NumOutputs + NumInputs; 4630 OpDecls.resize(NumExprs); 4631 Constraints.resize(NumExprs); 4632 for (unsigned i = 0; i < NumOutputs; ++i) { 4633 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 4634 Constraints[i] = OutputConstraints[i]; 4635 } 4636 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 4637 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 4638 Constraints[j] = InputConstraints[i]; 4639 } 4640 } 4641 4642 // Build the IR assembly string. 4643 std::string AsmStringIR; 4644 raw_string_ostream OS(AsmStringIR); 4645 StringRef ASMString = 4646 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 4647 const char *AsmStart = ASMString.begin(); 4648 const char *AsmEnd = ASMString.end(); 4649 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 4650 for (const AsmRewrite &AR : AsmStrRewrites) { 4651 AsmRewriteKind Kind = AR.Kind; 4652 if (Kind == AOK_Delete) 4653 continue; 4654 4655 const char *Loc = AR.Loc.getPointer(); 4656 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 4657 4658 // Emit everything up to the immediate/expression. 4659 if (unsigned Len = Loc - AsmStart) 4660 OS << StringRef(AsmStart, Len); 4661 4662 // Skip the original expression. 4663 if (Kind == AOK_Skip) { 4664 AsmStart = Loc + AR.Len; 4665 continue; 4666 } 4667 4668 unsigned AdditionalSkip = 0; 4669 // Rewrite expressions in $N notation. 4670 switch (Kind) { 4671 default: 4672 break; 4673 case AOK_Imm: 4674 OS << "$$" << AR.Val; 4675 break; 4676 case AOK_ImmPrefix: 4677 OS << "$$"; 4678 break; 4679 case AOK_Label: 4680 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 4681 break; 4682 case AOK_Input: 4683 OS << '$' << InputIdx++; 4684 break; 4685 case AOK_Output: 4686 OS << '$' << OutputIdx++; 4687 break; 4688 case AOK_SizeDirective: 4689 switch (AR.Val) { 4690 default: break; 4691 case 8: OS << "byte ptr "; break; 4692 case 16: OS << "word ptr "; break; 4693 case 32: OS << "dword ptr "; break; 4694 case 64: OS << "qword ptr "; break; 4695 case 80: OS << "xword ptr "; break; 4696 case 128: OS << "xmmword ptr "; break; 4697 case 256: OS << "ymmword ptr "; break; 4698 } 4699 break; 4700 case AOK_Emit: 4701 OS << ".byte"; 4702 break; 4703 case AOK_Align: { 4704 unsigned Val = AR.Val; 4705 OS << ".align " << Val; 4706 4707 // Skip the original immediate. 4708 assert(Val < 10 && "Expected alignment less then 2^10."); 4709 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 4710 break; 4711 } 4712 case AOK_DotOperator: 4713 // Insert the dot if the user omitted it. 4714 OS.flush(); 4715 if (AsmStringIR.back() != '.') 4716 OS << '.'; 4717 OS << AR.Val; 4718 break; 4719 } 4720 4721 // Skip the original expression. 4722 AsmStart = Loc + AR.Len + AdditionalSkip; 4723 } 4724 4725 // Emit the remainder of the asm string. 4726 if (AsmStart != AsmEnd) 4727 OS << StringRef(AsmStart, AsmEnd - AsmStart); 4728 4729 AsmString = OS.str(); 4730 return false; 4731 } 4732 4733 /// \brief Create an MCAsmParser instance. 4734 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 4735 MCStreamer &Out, const MCAsmInfo &MAI) { 4736 return new AsmParser(SM, C, Out, MAI); 4737 } 4738