1 //===- AsmParser.cpp - Parser for Assembly Files --------------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // This class implements the parser for assembly files. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "llvm/ADT/APFloat.h" 15 #include "llvm/ADT/STLExtras.h" 16 #include "llvm/ADT/SmallString.h" 17 #include "llvm/ADT/StringMap.h" 18 #include "llvm/ADT/Twine.h" 19 #include "llvm/MC/MCAsmInfo.h" 20 #include "llvm/MC/MCContext.h" 21 #include "llvm/MC/MCDwarf.h" 22 #include "llvm/MC/MCExpr.h" 23 #include "llvm/MC/MCInstPrinter.h" 24 #include "llvm/MC/MCInstrInfo.h" 25 #include "llvm/MC/MCObjectFileInfo.h" 26 #include "llvm/MC/MCParser/AsmCond.h" 27 #include "llvm/MC/MCParser/AsmLexer.h" 28 #include "llvm/MC/MCParser/MCAsmParser.h" 29 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 30 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 31 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 32 #include "llvm/MC/MCRegisterInfo.h" 33 #include "llvm/MC/MCSectionMachO.h" 34 #include "llvm/MC/MCStreamer.h" 35 #include "llvm/MC/MCSymbol.h" 36 #include "llvm/MC/MCValue.h" 37 #include "llvm/Support/CommandLine.h" 38 #include "llvm/Support/ErrorHandling.h" 39 #include "llvm/Support/MathExtras.h" 40 #include "llvm/Support/MemoryBuffer.h" 41 #include "llvm/Support/SourceMgr.h" 42 #include "llvm/Support/raw_ostream.h" 43 #include <cctype> 44 #include <deque> 45 #include <set> 46 #include <string> 47 #include <vector> 48 using namespace llvm; 49 50 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() {} 51 52 namespace { 53 /// \brief Helper types for tracking macro definitions. 54 typedef std::vector<AsmToken> MCAsmMacroArgument; 55 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 56 57 struct MCAsmMacroParameter { 58 StringRef Name; 59 MCAsmMacroArgument Value; 60 bool Required; 61 bool Vararg; 62 63 MCAsmMacroParameter() : Required(false), Vararg(false) {} 64 }; 65 66 typedef std::vector<MCAsmMacroParameter> MCAsmMacroParameters; 67 68 struct MCAsmMacro { 69 StringRef Name; 70 StringRef Body; 71 MCAsmMacroParameters Parameters; 72 73 public: 74 MCAsmMacro(StringRef N, StringRef B, MCAsmMacroParameters P) 75 : Name(N), Body(B), Parameters(std::move(P)) {} 76 }; 77 78 /// \brief Helper class for storing information about an active macro 79 /// instantiation. 80 struct MacroInstantiation { 81 /// The location of the instantiation. 82 SMLoc InstantiationLoc; 83 84 /// The buffer where parsing should resume upon instantiation completion. 85 int ExitBuffer; 86 87 /// The location where parsing should resume upon instantiation completion. 88 SMLoc ExitLoc; 89 90 /// The depth of TheCondStack at the start of the instantiation. 91 size_t CondStackDepth; 92 93 public: 94 MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth); 95 }; 96 97 struct ParseStatementInfo { 98 /// \brief The parsed operands from the last parsed statement. 99 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 100 101 /// \brief The opcode from the last parsed instruction. 102 unsigned Opcode; 103 104 /// \brief Was there an error parsing the inline assembly? 105 bool ParseError; 106 107 SmallVectorImpl<AsmRewrite> *AsmRewrites; 108 109 ParseStatementInfo() : Opcode(~0U), ParseError(false), AsmRewrites(nullptr) {} 110 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 111 : Opcode(~0), ParseError(false), AsmRewrites(rewrites) {} 112 }; 113 114 /// \brief The concrete assembly parser instance. 115 class AsmParser : public MCAsmParser { 116 AsmParser(const AsmParser &) = delete; 117 void operator=(const AsmParser &) = delete; 118 private: 119 AsmLexer Lexer; 120 MCContext &Ctx; 121 MCStreamer &Out; 122 const MCAsmInfo &MAI; 123 SourceMgr &SrcMgr; 124 SourceMgr::DiagHandlerTy SavedDiagHandler; 125 void *SavedDiagContext; 126 std::unique_ptr<MCAsmParserExtension> PlatformParser; 127 128 /// This is the current buffer index we're lexing from as managed by the 129 /// SourceMgr object. 130 unsigned CurBuffer; 131 132 AsmCond TheCondState; 133 std::vector<AsmCond> TheCondStack; 134 135 /// \brief maps directive names to handler methods in parser 136 /// extensions. Extensions register themselves in this map by calling 137 /// addDirectiveHandler. 138 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 139 140 /// \brief Map of currently defined macros. 141 StringMap<MCAsmMacro> MacroMap; 142 143 /// \brief Stack of active macro instantiations. 144 std::vector<MacroInstantiation*> ActiveMacros; 145 146 /// \brief List of bodies of anonymous macros. 147 std::deque<MCAsmMacro> MacroLikeBodies; 148 149 /// Boolean tracking whether macro substitution is enabled. 150 unsigned MacrosEnabledFlag : 1; 151 152 /// \brief Keeps track of how many .macro's have been instantiated. 153 unsigned NumOfMacroInstantiations; 154 155 /// Flag tracking whether any errors have been encountered. 156 unsigned HadError : 1; 157 158 /// The values from the last parsed cpp hash file line comment if any. 159 StringRef CppHashFilename; 160 int64_t CppHashLineNumber; 161 SMLoc CppHashLoc; 162 unsigned CppHashBuf; 163 /// When generating dwarf for assembly source files we need to calculate the 164 /// logical line number based on the last parsed cpp hash file line comment 165 /// and current line. Since this is slow and messes up the SourceMgr's 166 /// cache we save the last info we queried with SrcMgr.FindLineNumber(). 167 SMLoc LastQueryIDLoc; 168 unsigned LastQueryBuffer; 169 unsigned LastQueryLine; 170 171 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 172 unsigned AssemblerDialect; 173 174 /// \brief is Darwin compatibility enabled? 175 bool IsDarwin; 176 177 /// \brief Are we parsing ms-style inline assembly? 178 bool ParsingInlineAsm; 179 180 public: 181 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 182 const MCAsmInfo &MAI); 183 ~AsmParser() override; 184 185 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 186 187 void addDirectiveHandler(StringRef Directive, 188 ExtensionDirectiveHandler Handler) override { 189 ExtensionDirectiveMap[Directive] = Handler; 190 } 191 192 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 193 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 194 } 195 196 public: 197 /// @name MCAsmParser Interface 198 /// { 199 200 SourceMgr &getSourceManager() override { return SrcMgr; } 201 MCAsmLexer &getLexer() override { return Lexer; } 202 MCContext &getContext() override { return Ctx; } 203 MCStreamer &getStreamer() override { return Out; } 204 unsigned getAssemblerDialect() override { 205 if (AssemblerDialect == ~0U) 206 return MAI.getAssemblerDialect(); 207 else 208 return AssemblerDialect; 209 } 210 void setAssemblerDialect(unsigned i) override { 211 AssemblerDialect = i; 212 } 213 214 void Note(SMLoc L, const Twine &Msg, 215 ArrayRef<SMRange> Ranges = None) override; 216 bool Warning(SMLoc L, const Twine &Msg, 217 ArrayRef<SMRange> Ranges = None) override; 218 bool Error(SMLoc L, const Twine &Msg, 219 ArrayRef<SMRange> Ranges = None) override; 220 221 const AsmToken &Lex() override; 222 223 void setParsingInlineAsm(bool V) override { ParsingInlineAsm = V; } 224 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 225 226 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 227 unsigned &NumOutputs, unsigned &NumInputs, 228 SmallVectorImpl<std::pair<void *,bool> > &OpDecls, 229 SmallVectorImpl<std::string> &Constraints, 230 SmallVectorImpl<std::string> &Clobbers, 231 const MCInstrInfo *MII, const MCInstPrinter *IP, 232 MCAsmParserSemaCallback &SI) override; 233 234 bool parseExpression(const MCExpr *&Res); 235 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 236 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 237 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 238 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 239 SMLoc &EndLoc) override; 240 bool parseAbsoluteExpression(int64_t &Res) override; 241 242 /// \brief Parse an identifier or string (as a quoted identifier) 243 /// and set \p Res to the identifier contents. 244 bool parseIdentifier(StringRef &Res) override; 245 void eatToEndOfStatement() override; 246 247 void checkForValidSection() override; 248 /// } 249 250 private: 251 252 bool parseStatement(ParseStatementInfo &Info, 253 MCAsmParserSemaCallback *SI); 254 void eatToEndOfLine(); 255 bool parseCppHashLineFilenameComment(SMLoc L); 256 257 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 258 ArrayRef<MCAsmMacroParameter> Parameters); 259 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 260 ArrayRef<MCAsmMacroParameter> Parameters, 261 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 262 SMLoc L); 263 264 /// \brief Are macros enabled in the parser? 265 bool areMacrosEnabled() {return MacrosEnabledFlag;} 266 267 /// \brief Control a flag in the parser that enables or disables macros. 268 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 269 270 /// \brief Lookup a previously defined macro. 271 /// \param Name Macro name. 272 /// \returns Pointer to macro. NULL if no such macro was defined. 273 const MCAsmMacro* lookupMacro(StringRef Name); 274 275 /// \brief Define a new macro with the given name and information. 276 void defineMacro(StringRef Name, MCAsmMacro Macro); 277 278 /// \brief Undefine a macro. If no such macro was defined, it's a no-op. 279 void undefineMacro(StringRef Name); 280 281 /// \brief Are we inside a macro instantiation? 282 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 283 284 /// \brief Handle entry to macro instantiation. 285 /// 286 /// \param M The macro. 287 /// \param NameLoc Instantiation location. 288 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 289 290 /// \brief Handle exit from macro instantiation. 291 void handleMacroExit(); 292 293 /// \brief Extract AsmTokens for a macro argument. 294 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 295 296 /// \brief Parse all macro arguments for a given macro. 297 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 298 299 void printMacroInstantiations(); 300 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 301 ArrayRef<SMRange> Ranges = None) const { 302 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 303 } 304 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 305 306 /// \brief Enter the specified file. This returns true on failure. 307 bool enterIncludeFile(const std::string &Filename); 308 309 /// \brief Process the specified file for the .incbin directive. 310 /// This returns true on failure. 311 bool processIncbinFile(const std::string &Filename); 312 313 /// \brief Reset the current lexer position to that given by \p Loc. The 314 /// current token is not set; clients should ensure Lex() is called 315 /// subsequently. 316 /// 317 /// \param InBuffer If not 0, should be the known buffer id that contains the 318 /// location. 319 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 320 321 /// \brief Parse up to the end of statement and a return the contents from the 322 /// current token until the end of the statement; the current token on exit 323 /// will be either the EndOfStatement or EOF. 324 StringRef parseStringToEndOfStatement() override; 325 326 /// \brief Parse until the end of a statement or a comma is encountered, 327 /// return the contents from the current token up to the end or comma. 328 StringRef parseStringToComma(); 329 330 bool parseAssignment(StringRef Name, bool allow_redef, 331 bool NoDeadStrip = false); 332 333 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 334 MCBinaryExpr::Opcode &Kind); 335 336 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 337 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 338 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 339 340 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 341 342 // Generic (target and platform independent) directive parsing. 343 enum DirectiveKind { 344 DK_NO_DIRECTIVE, // Placeholder 345 DK_SET, DK_EQU, DK_EQUIV, DK_ASCII, DK_ASCIZ, DK_STRING, DK_BYTE, DK_SHORT, 346 DK_RELOC, 347 DK_VALUE, DK_2BYTE, DK_LONG, DK_INT, DK_4BYTE, DK_QUAD, DK_8BYTE, DK_OCTA, 348 DK_SINGLE, DK_FLOAT, DK_DOUBLE, DK_ALIGN, DK_ALIGN32, DK_BALIGN, DK_BALIGNW, 349 DK_BALIGNL, DK_P2ALIGN, DK_P2ALIGNW, DK_P2ALIGNL, DK_ORG, DK_FILL, DK_ENDR, 350 DK_BUNDLE_ALIGN_MODE, DK_BUNDLE_LOCK, DK_BUNDLE_UNLOCK, 351 DK_ZERO, DK_EXTERN, DK_GLOBL, DK_GLOBAL, 352 DK_LAZY_REFERENCE, DK_NO_DEAD_STRIP, DK_SYMBOL_RESOLVER, DK_PRIVATE_EXTERN, 353 DK_REFERENCE, DK_WEAK_DEFINITION, DK_WEAK_REFERENCE, 354 DK_WEAK_DEF_CAN_BE_HIDDEN, DK_COMM, DK_COMMON, DK_LCOMM, DK_ABORT, 355 DK_INCLUDE, DK_INCBIN, DK_CODE16, DK_CODE16GCC, DK_REPT, DK_IRP, DK_IRPC, 356 DK_IF, DK_IFEQ, DK_IFGE, DK_IFGT, DK_IFLE, DK_IFLT, DK_IFNE, DK_IFB, 357 DK_IFNB, DK_IFC, DK_IFEQS, DK_IFNC, DK_IFNES, DK_IFDEF, DK_IFNDEF, 358 DK_IFNOTDEF, DK_ELSEIF, DK_ELSE, DK_ENDIF, 359 DK_SPACE, DK_SKIP, DK_FILE, DK_LINE, DK_LOC, DK_STABS, 360 DK_CV_FILE, DK_CV_LOC, DK_CV_LINETABLE, DK_CV_INLINE_LINETABLE, 361 DK_CV_DEF_RANGE, DK_CV_STRINGTABLE, DK_CV_FILECHECKSUMS, 362 DK_CFI_SECTIONS, DK_CFI_STARTPROC, DK_CFI_ENDPROC, DK_CFI_DEF_CFA, 363 DK_CFI_DEF_CFA_OFFSET, DK_CFI_ADJUST_CFA_OFFSET, DK_CFI_DEF_CFA_REGISTER, 364 DK_CFI_OFFSET, DK_CFI_REL_OFFSET, DK_CFI_PERSONALITY, DK_CFI_LSDA, 365 DK_CFI_REMEMBER_STATE, DK_CFI_RESTORE_STATE, DK_CFI_SAME_VALUE, 366 DK_CFI_RESTORE, DK_CFI_ESCAPE, DK_CFI_SIGNAL_FRAME, DK_CFI_UNDEFINED, 367 DK_CFI_REGISTER, DK_CFI_WINDOW_SAVE, 368 DK_MACROS_ON, DK_MACROS_OFF, 369 DK_MACRO, DK_EXITM, DK_ENDM, DK_ENDMACRO, DK_PURGEM, 370 DK_SLEB128, DK_ULEB128, 371 DK_ERR, DK_ERROR, DK_WARNING, 372 DK_END 373 }; 374 375 /// \brief Maps directive name --> DirectiveKind enum, for 376 /// directives parsed by this class. 377 StringMap<DirectiveKind> DirectiveKindMap; 378 379 // ".ascii", ".asciz", ".string" 380 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 381 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 382 bool parseDirectiveValue(unsigned Size); // ".byte", ".long", ... 383 bool parseDirectiveOctaValue(); // ".octa" 384 bool parseDirectiveRealValue(const fltSemantics &); // ".single", ... 385 bool parseDirectiveFill(); // ".fill" 386 bool parseDirectiveZero(); // ".zero" 387 // ".set", ".equ", ".equiv" 388 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 389 bool parseDirectiveOrg(); // ".org" 390 // ".align{,32}", ".p2align{,w,l}" 391 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 392 393 // ".file", ".line", ".loc", ".stabs" 394 bool parseDirectiveFile(SMLoc DirectiveLoc); 395 bool parseDirectiveLine(); 396 bool parseDirectiveLoc(); 397 bool parseDirectiveStabs(); 398 399 // ".cv_file", ".cv_loc", ".cv_linetable", "cv_inline_linetable", 400 // ".cv_def_range" 401 bool parseDirectiveCVFile(); 402 bool parseDirectiveCVLoc(); 403 bool parseDirectiveCVLinetable(); 404 bool parseDirectiveCVInlineLinetable(); 405 bool parseDirectiveCVDefRange(); 406 bool parseDirectiveCVStringTable(); 407 bool parseDirectiveCVFileChecksums(); 408 409 // .cfi directives 410 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 411 bool parseDirectiveCFIWindowSave(); 412 bool parseDirectiveCFISections(); 413 bool parseDirectiveCFIStartProc(); 414 bool parseDirectiveCFIEndProc(); 415 bool parseDirectiveCFIDefCfaOffset(); 416 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 417 bool parseDirectiveCFIAdjustCfaOffset(); 418 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 419 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 420 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 421 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 422 bool parseDirectiveCFIRememberState(); 423 bool parseDirectiveCFIRestoreState(); 424 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 425 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 426 bool parseDirectiveCFIEscape(); 427 bool parseDirectiveCFISignalFrame(); 428 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 429 430 // macro directives 431 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 432 bool parseDirectiveExitMacro(StringRef Directive); 433 bool parseDirectiveEndMacro(StringRef Directive); 434 bool parseDirectiveMacro(SMLoc DirectiveLoc); 435 bool parseDirectiveMacrosOnOff(StringRef Directive); 436 437 // ".bundle_align_mode" 438 bool parseDirectiveBundleAlignMode(); 439 // ".bundle_lock" 440 bool parseDirectiveBundleLock(); 441 // ".bundle_unlock" 442 bool parseDirectiveBundleUnlock(); 443 444 // ".space", ".skip" 445 bool parseDirectiveSpace(StringRef IDVal); 446 447 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 448 bool parseDirectiveLEB128(bool Signed); 449 450 /// \brief Parse a directive like ".globl" which 451 /// accepts a single symbol (which should be a label or an external). 452 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 453 454 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 455 456 bool parseDirectiveAbort(); // ".abort" 457 bool parseDirectiveInclude(); // ".include" 458 bool parseDirectiveIncbin(); // ".incbin" 459 460 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 461 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 462 // ".ifb" or ".ifnb", depending on ExpectBlank. 463 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 464 // ".ifc" or ".ifnc", depending on ExpectEqual. 465 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 466 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 467 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 468 // ".ifdef" or ".ifndef", depending on expect_defined 469 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 470 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 471 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 472 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 473 bool parseEscapedString(std::string &Data) override; 474 475 const MCExpr *applyModifierToExpr(const MCExpr *E, 476 MCSymbolRefExpr::VariantKind Variant); 477 478 // Macro-like directives 479 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 480 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 481 raw_svector_ostream &OS); 482 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 483 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 484 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 485 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 486 487 // "_emit" or "__emit" 488 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 489 size_t Len); 490 491 // "align" 492 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 493 494 // "end" 495 bool parseDirectiveEnd(SMLoc DirectiveLoc); 496 497 // ".err" or ".error" 498 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 499 500 // ".warning" 501 bool parseDirectiveWarning(SMLoc DirectiveLoc); 502 503 void initializeDirectiveKindMap(); 504 }; 505 } 506 507 namespace llvm { 508 509 extern MCAsmParserExtension *createDarwinAsmParser(); 510 extern MCAsmParserExtension *createELFAsmParser(); 511 extern MCAsmParserExtension *createCOFFAsmParser(); 512 513 } 514 515 enum { DEFAULT_ADDRSPACE = 0 }; 516 517 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 518 const MCAsmInfo &MAI) 519 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 520 PlatformParser(nullptr), CurBuffer(SM.getMainFileID()), 521 MacrosEnabledFlag(true), HadError(false), CppHashLineNumber(0), 522 AssemblerDialect(~0U), IsDarwin(false), ParsingInlineAsm(false) { 523 // Save the old handler. 524 SavedDiagHandler = SrcMgr.getDiagHandler(); 525 SavedDiagContext = SrcMgr.getDiagContext(); 526 // Set our own handler which calls the saved handler. 527 SrcMgr.setDiagHandler(DiagHandler, this); 528 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 529 530 // Initialize the platform / file format parser. 531 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 532 case MCObjectFileInfo::IsCOFF: 533 PlatformParser.reset(createCOFFAsmParser()); 534 break; 535 case MCObjectFileInfo::IsMachO: 536 PlatformParser.reset(createDarwinAsmParser()); 537 IsDarwin = true; 538 break; 539 case MCObjectFileInfo::IsELF: 540 PlatformParser.reset(createELFAsmParser()); 541 break; 542 } 543 544 PlatformParser->Initialize(*this); 545 initializeDirectiveKindMap(); 546 547 NumOfMacroInstantiations = 0; 548 } 549 550 AsmParser::~AsmParser() { 551 assert((HadError || ActiveMacros.empty()) && 552 "Unexpected active macro instantiation!"); 553 } 554 555 void AsmParser::printMacroInstantiations() { 556 // Print the active macro instantiation stack. 557 for (std::vector<MacroInstantiation *>::const_reverse_iterator 558 it = ActiveMacros.rbegin(), 559 ie = ActiveMacros.rend(); 560 it != ie; ++it) 561 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 562 "while in macro instantiation"); 563 } 564 565 void AsmParser::Note(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 566 printMessage(L, SourceMgr::DK_Note, Msg, Ranges); 567 printMacroInstantiations(); 568 } 569 570 bool AsmParser::Warning(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 571 if(getTargetParser().getTargetOptions().MCNoWarn) 572 return false; 573 if (getTargetParser().getTargetOptions().MCFatalWarnings) 574 return Error(L, Msg, Ranges); 575 printMessage(L, SourceMgr::DK_Warning, Msg, Ranges); 576 printMacroInstantiations(); 577 return false; 578 } 579 580 bool AsmParser::Error(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges) { 581 HadError = true; 582 printMessage(L, SourceMgr::DK_Error, Msg, Ranges); 583 printMacroInstantiations(); 584 return true; 585 } 586 587 bool AsmParser::enterIncludeFile(const std::string &Filename) { 588 std::string IncludedFile; 589 unsigned NewBuf = 590 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 591 if (!NewBuf) 592 return true; 593 594 CurBuffer = NewBuf; 595 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 596 return false; 597 } 598 599 /// Process the specified .incbin file by searching for it in the include paths 600 /// then just emitting the byte contents of the file to the streamer. This 601 /// returns true on failure. 602 bool AsmParser::processIncbinFile(const std::string &Filename) { 603 std::string IncludedFile; 604 unsigned NewBuf = 605 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 606 if (!NewBuf) 607 return true; 608 609 // Pick up the bytes from the file and emit them. 610 getStreamer().EmitBytes(SrcMgr.getMemoryBuffer(NewBuf)->getBuffer()); 611 return false; 612 } 613 614 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 615 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 616 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 617 Loc.getPointer()); 618 } 619 620 const AsmToken &AsmParser::Lex() { 621 const AsmToken *tok = &Lexer.Lex(); 622 623 if (tok->is(AsmToken::Eof)) { 624 // If this is the end of an included file, pop the parent file off the 625 // include stack. 626 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 627 if (ParentIncludeLoc != SMLoc()) { 628 jumpToLoc(ParentIncludeLoc); 629 tok = &Lexer.Lex(); 630 } 631 } 632 633 if (tok->is(AsmToken::Error)) 634 Error(Lexer.getErrLoc(), Lexer.getErr()); 635 636 return *tok; 637 } 638 639 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 640 // Create the initial section, if requested. 641 if (!NoInitialTextSection) 642 Out.InitSections(false); 643 644 // Prime the lexer. 645 Lex(); 646 647 HadError = false; 648 AsmCond StartingCondState = TheCondState; 649 650 // If we are generating dwarf for assembly source files save the initial text 651 // section and generate a .file directive. 652 if (getContext().getGenDwarfForAssembly()) { 653 MCSection *Sec = getStreamer().getCurrentSection().first; 654 if (!Sec->getBeginSymbol()) { 655 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 656 getStreamer().EmitLabel(SectionStartSym); 657 Sec->setBeginSymbol(SectionStartSym); 658 } 659 bool InsertResult = getContext().addGenDwarfSection(Sec); 660 assert(InsertResult && ".text section should not have debug info yet"); 661 (void)InsertResult; 662 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 663 0, StringRef(), getContext().getMainFileName())); 664 } 665 666 // While we have input, parse each statement. 667 while (Lexer.isNot(AsmToken::Eof)) { 668 ParseStatementInfo Info; 669 if (!parseStatement(Info, nullptr)) 670 continue; 671 672 // We had an error, validate that one was emitted and recover by skipping to 673 // the next line. 674 assert(HadError && "Parse statement returned an error, but none emitted!"); 675 eatToEndOfStatement(); 676 } 677 678 if (TheCondState.TheCond != StartingCondState.TheCond || 679 TheCondState.Ignore != StartingCondState.Ignore) 680 return TokError("unmatched .ifs or .elses"); 681 682 // Check to see there are no empty DwarfFile slots. 683 const auto &LineTables = getContext().getMCDwarfLineTables(); 684 if (!LineTables.empty()) { 685 unsigned Index = 0; 686 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 687 if (File.Name.empty() && Index != 0) 688 TokError("unassigned file number: " + Twine(Index) + 689 " for .file directives"); 690 ++Index; 691 } 692 } 693 694 // Check to see that all assembler local symbols were actually defined. 695 // Targets that don't do subsections via symbols may not want this, though, 696 // so conservatively exclude them. Only do this if we're finalizing, though, 697 // as otherwise we won't necessarilly have seen everything yet. 698 if (!NoFinalize && MAI.hasSubsectionsViaSymbols()) { 699 for (const auto &TableEntry : getContext().getSymbols()) { 700 MCSymbol *Sym = TableEntry.getValue(); 701 // Variable symbols may not be marked as defined, so check those 702 // explicitly. If we know it's a variable, we have a definition for 703 // the purposes of this check. 704 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 705 // FIXME: We would really like to refer back to where the symbol was 706 // first referenced for a source location. We need to add something 707 // to track that. Currently, we just point to the end of the file. 708 return Error(getLexer().getLoc(), "assembler local symbol '" + 709 Sym->getName() + "' not defined"); 710 } 711 } 712 713 // Finalize the output stream if there are no errors and if the client wants 714 // us to. 715 if (!HadError && !NoFinalize) 716 Out.Finish(); 717 718 return HadError || getContext().hadError(); 719 } 720 721 void AsmParser::checkForValidSection() { 722 if (!ParsingInlineAsm && !getStreamer().getCurrentSection().first) { 723 TokError("expected section directive before assembly directive"); 724 Out.InitSections(false); 725 } 726 } 727 728 /// \brief Throw away the rest of the line for testing purposes. 729 void AsmParser::eatToEndOfStatement() { 730 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 731 Lex(); 732 733 // Eat EOL. 734 if (Lexer.is(AsmToken::EndOfStatement)) 735 Lex(); 736 } 737 738 StringRef AsmParser::parseStringToEndOfStatement() { 739 const char *Start = getTok().getLoc().getPointer(); 740 741 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 742 Lex(); 743 744 const char *End = getTok().getLoc().getPointer(); 745 return StringRef(Start, End - Start); 746 } 747 748 StringRef AsmParser::parseStringToComma() { 749 const char *Start = getTok().getLoc().getPointer(); 750 751 while (Lexer.isNot(AsmToken::EndOfStatement) && 752 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 753 Lex(); 754 755 const char *End = getTok().getLoc().getPointer(); 756 return StringRef(Start, End - Start); 757 } 758 759 /// \brief Parse a paren expression and return it. 760 /// NOTE: This assumes the leading '(' has already been consumed. 761 /// 762 /// parenexpr ::= expr) 763 /// 764 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 765 if (parseExpression(Res)) 766 return true; 767 if (Lexer.isNot(AsmToken::RParen)) 768 return TokError("expected ')' in parentheses expression"); 769 EndLoc = Lexer.getTok().getEndLoc(); 770 Lex(); 771 return false; 772 } 773 774 /// \brief Parse a bracket expression and return it. 775 /// NOTE: This assumes the leading '[' has already been consumed. 776 /// 777 /// bracketexpr ::= expr] 778 /// 779 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 780 if (parseExpression(Res)) 781 return true; 782 if (Lexer.isNot(AsmToken::RBrac)) 783 return TokError("expected ']' in brackets expression"); 784 EndLoc = Lexer.getTok().getEndLoc(); 785 Lex(); 786 return false; 787 } 788 789 /// \brief Parse a primary expression and return it. 790 /// primaryexpr ::= (parenexpr 791 /// primaryexpr ::= symbol 792 /// primaryexpr ::= number 793 /// primaryexpr ::= '.' 794 /// primaryexpr ::= ~,+,- primaryexpr 795 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 796 SMLoc FirstTokenLoc = getLexer().getLoc(); 797 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 798 switch (FirstTokenKind) { 799 default: 800 return TokError("unknown token in expression"); 801 // If we have an error assume that we've already handled it. 802 case AsmToken::Error: 803 return true; 804 case AsmToken::Exclaim: 805 Lex(); // Eat the operator. 806 if (parsePrimaryExpr(Res, EndLoc)) 807 return true; 808 Res = MCUnaryExpr::createLNot(Res, getContext()); 809 return false; 810 case AsmToken::Dollar: 811 case AsmToken::At: 812 case AsmToken::String: 813 case AsmToken::Identifier: { 814 StringRef Identifier; 815 if (parseIdentifier(Identifier)) { 816 if (FirstTokenKind == AsmToken::Dollar) { 817 if (Lexer.getMAI().getDollarIsPC()) { 818 // This is a '$' reference, which references the current PC. Emit a 819 // temporary label to the streamer and refer to it. 820 MCSymbol *Sym = Ctx.createTempSymbol(); 821 Out.EmitLabel(Sym); 822 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 823 getContext()); 824 EndLoc = FirstTokenLoc; 825 return false; 826 } 827 return Error(FirstTokenLoc, "invalid token in expression"); 828 } 829 } 830 // Parse symbol variant 831 std::pair<StringRef, StringRef> Split; 832 if (!MAI.useParensForSymbolVariant()) { 833 if (FirstTokenKind == AsmToken::String) { 834 if (Lexer.is(AsmToken::At)) { 835 Lexer.Lex(); // eat @ 836 SMLoc AtLoc = getLexer().getLoc(); 837 StringRef VName; 838 if (parseIdentifier(VName)) 839 return Error(AtLoc, "expected symbol variant after '@'"); 840 841 Split = std::make_pair(Identifier, VName); 842 } 843 } else { 844 Split = Identifier.split('@'); 845 } 846 } else if (Lexer.is(AsmToken::LParen)) { 847 Lexer.Lex(); // eat ( 848 StringRef VName; 849 parseIdentifier(VName); 850 if (Lexer.isNot(AsmToken::RParen)) { 851 return Error(Lexer.getTok().getLoc(), 852 "unexpected token in variant, expected ')'"); 853 } 854 Lexer.Lex(); // eat ) 855 Split = std::make_pair(Identifier, VName); 856 } 857 858 EndLoc = SMLoc::getFromPointer(Identifier.end()); 859 860 // This is a symbol reference. 861 StringRef SymbolName = Identifier; 862 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 863 864 // Lookup the symbol variant if used. 865 if (Split.second.size()) { 866 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 867 if (Variant != MCSymbolRefExpr::VK_Invalid) { 868 SymbolName = Split.first; 869 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 870 Variant = MCSymbolRefExpr::VK_None; 871 } else { 872 return Error(SMLoc::getFromPointer(Split.second.begin()), 873 "invalid variant '" + Split.second + "'"); 874 } 875 } 876 877 MCSymbol *Sym = getContext().getOrCreateSymbol(SymbolName); 878 879 // If this is an absolute variable reference, substitute it now to preserve 880 // semantics in the face of reassignment. 881 if (Sym->isVariable() && 882 isa<MCConstantExpr>(Sym->getVariableValue(/*SetUsed*/ false))) { 883 if (Variant) 884 return Error(EndLoc, "unexpected modifier on variable reference"); 885 886 Res = Sym->getVariableValue(/*SetUsed*/ false); 887 return false; 888 } 889 890 // Otherwise create a symbol ref. 891 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 892 return false; 893 } 894 case AsmToken::BigNum: 895 return TokError("literal value out of range for directive"); 896 case AsmToken::Integer: { 897 SMLoc Loc = getTok().getLoc(); 898 int64_t IntVal = getTok().getIntVal(); 899 Res = MCConstantExpr::create(IntVal, getContext()); 900 EndLoc = Lexer.getTok().getEndLoc(); 901 Lex(); // Eat token. 902 // Look for 'b' or 'f' following an Integer as a directional label 903 if (Lexer.getKind() == AsmToken::Identifier) { 904 StringRef IDVal = getTok().getString(); 905 // Lookup the symbol variant if used. 906 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 907 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 908 if (Split.first.size() != IDVal.size()) { 909 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 910 if (Variant == MCSymbolRefExpr::VK_Invalid) 911 return TokError("invalid variant '" + Split.second + "'"); 912 IDVal = Split.first; 913 } 914 if (IDVal == "f" || IDVal == "b") { 915 MCSymbol *Sym = 916 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 917 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 918 if (IDVal == "b" && Sym->isUndefined()) 919 return Error(Loc, "invalid reference to undefined symbol"); 920 EndLoc = Lexer.getTok().getEndLoc(); 921 Lex(); // Eat identifier. 922 } 923 } 924 return false; 925 } 926 case AsmToken::Real: { 927 APFloat RealVal(APFloat::IEEEdouble, getTok().getString()); 928 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 929 Res = MCConstantExpr::create(IntVal, getContext()); 930 EndLoc = Lexer.getTok().getEndLoc(); 931 Lex(); // Eat token. 932 return false; 933 } 934 case AsmToken::Dot: { 935 // This is a '.' reference, which references the current PC. Emit a 936 // temporary label to the streamer and refer to it. 937 MCSymbol *Sym = Ctx.createTempSymbol(); 938 Out.EmitLabel(Sym); 939 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 940 EndLoc = Lexer.getTok().getEndLoc(); 941 Lex(); // Eat identifier. 942 return false; 943 } 944 case AsmToken::LParen: 945 Lex(); // Eat the '('. 946 return parseParenExpr(Res, EndLoc); 947 case AsmToken::LBrac: 948 if (!PlatformParser->HasBracketExpressions()) 949 return TokError("brackets expression not supported on this target"); 950 Lex(); // Eat the '['. 951 return parseBracketExpr(Res, EndLoc); 952 case AsmToken::Minus: 953 Lex(); // Eat the operator. 954 if (parsePrimaryExpr(Res, EndLoc)) 955 return true; 956 Res = MCUnaryExpr::createMinus(Res, getContext()); 957 return false; 958 case AsmToken::Plus: 959 Lex(); // Eat the operator. 960 if (parsePrimaryExpr(Res, EndLoc)) 961 return true; 962 Res = MCUnaryExpr::createPlus(Res, getContext()); 963 return false; 964 case AsmToken::Tilde: 965 Lex(); // Eat the operator. 966 if (parsePrimaryExpr(Res, EndLoc)) 967 return true; 968 Res = MCUnaryExpr::createNot(Res, getContext()); 969 return false; 970 } 971 } 972 973 bool AsmParser::parseExpression(const MCExpr *&Res) { 974 SMLoc EndLoc; 975 return parseExpression(Res, EndLoc); 976 } 977 978 const MCExpr * 979 AsmParser::applyModifierToExpr(const MCExpr *E, 980 MCSymbolRefExpr::VariantKind Variant) { 981 // Ask the target implementation about this expression first. 982 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 983 if (NewE) 984 return NewE; 985 // Recurse over the given expression, rebuilding it to apply the given variant 986 // if there is exactly one symbol. 987 switch (E->getKind()) { 988 case MCExpr::Target: 989 case MCExpr::Constant: 990 return nullptr; 991 992 case MCExpr::SymbolRef: { 993 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 994 995 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 996 TokError("invalid variant on expression '" + getTok().getIdentifier() + 997 "' (already modified)"); 998 return E; 999 } 1000 1001 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1002 } 1003 1004 case MCExpr::Unary: { 1005 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1006 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1007 if (!Sub) 1008 return nullptr; 1009 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1010 } 1011 1012 case MCExpr::Binary: { 1013 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1014 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1015 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1016 1017 if (!LHS && !RHS) 1018 return nullptr; 1019 1020 if (!LHS) 1021 LHS = BE->getLHS(); 1022 if (!RHS) 1023 RHS = BE->getRHS(); 1024 1025 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1026 } 1027 } 1028 1029 llvm_unreachable("Invalid expression kind!"); 1030 } 1031 1032 /// \brief Parse an expression and return it. 1033 /// 1034 /// expr ::= expr &&,|| expr -> lowest. 1035 /// expr ::= expr |,^,&,! expr 1036 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1037 /// expr ::= expr <<,>> expr 1038 /// expr ::= expr +,- expr 1039 /// expr ::= expr *,/,% expr -> highest. 1040 /// expr ::= primaryexpr 1041 /// 1042 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1043 // Parse the expression. 1044 Res = nullptr; 1045 if (parsePrimaryExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc)) 1046 return true; 1047 1048 // As a special case, we support 'a op b @ modifier' by rewriting the 1049 // expression to include the modifier. This is inefficient, but in general we 1050 // expect users to use 'a@modifier op b'. 1051 if (Lexer.getKind() == AsmToken::At) { 1052 Lex(); 1053 1054 if (Lexer.isNot(AsmToken::Identifier)) 1055 return TokError("unexpected symbol modifier following '@'"); 1056 1057 MCSymbolRefExpr::VariantKind Variant = 1058 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1059 if (Variant == MCSymbolRefExpr::VK_Invalid) 1060 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1061 1062 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1063 if (!ModifiedRes) { 1064 return TokError("invalid modifier '" + getTok().getIdentifier() + 1065 "' (no symbols present)"); 1066 } 1067 1068 Res = ModifiedRes; 1069 Lex(); 1070 } 1071 1072 // Try to constant fold it up front, if possible. 1073 int64_t Value; 1074 if (Res->evaluateAsAbsolute(Value)) 1075 Res = MCConstantExpr::create(Value, getContext()); 1076 1077 return false; 1078 } 1079 1080 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1081 Res = nullptr; 1082 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1083 } 1084 1085 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1086 SMLoc &EndLoc) { 1087 if (parseParenExpr(Res, EndLoc)) 1088 return true; 1089 1090 for (; ParenDepth > 0; --ParenDepth) { 1091 if (parseBinOpRHS(1, Res, EndLoc)) 1092 return true; 1093 1094 // We don't Lex() the last RParen. 1095 // This is the same behavior as parseParenExpression(). 1096 if (ParenDepth - 1 > 0) { 1097 if (Lexer.isNot(AsmToken::RParen)) 1098 return TokError("expected ')' in parentheses expression"); 1099 EndLoc = Lexer.getTok().getEndLoc(); 1100 Lex(); 1101 } 1102 } 1103 return false; 1104 } 1105 1106 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1107 const MCExpr *Expr; 1108 1109 SMLoc StartLoc = Lexer.getLoc(); 1110 if (parseExpression(Expr)) 1111 return true; 1112 1113 if (!Expr->evaluateAsAbsolute(Res)) 1114 return Error(StartLoc, "expected absolute expression"); 1115 1116 return false; 1117 } 1118 1119 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1120 MCBinaryExpr::Opcode &Kind, 1121 bool ShouldUseLogicalShr) { 1122 switch (K) { 1123 default: 1124 return 0; // not a binop. 1125 1126 // Lowest Precedence: &&, || 1127 case AsmToken::AmpAmp: 1128 Kind = MCBinaryExpr::LAnd; 1129 return 1; 1130 case AsmToken::PipePipe: 1131 Kind = MCBinaryExpr::LOr; 1132 return 1; 1133 1134 // Low Precedence: |, &, ^ 1135 // 1136 // FIXME: gas seems to support '!' as an infix operator? 1137 case AsmToken::Pipe: 1138 Kind = MCBinaryExpr::Or; 1139 return 2; 1140 case AsmToken::Caret: 1141 Kind = MCBinaryExpr::Xor; 1142 return 2; 1143 case AsmToken::Amp: 1144 Kind = MCBinaryExpr::And; 1145 return 2; 1146 1147 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1148 case AsmToken::EqualEqual: 1149 Kind = MCBinaryExpr::EQ; 1150 return 3; 1151 case AsmToken::ExclaimEqual: 1152 case AsmToken::LessGreater: 1153 Kind = MCBinaryExpr::NE; 1154 return 3; 1155 case AsmToken::Less: 1156 Kind = MCBinaryExpr::LT; 1157 return 3; 1158 case AsmToken::LessEqual: 1159 Kind = MCBinaryExpr::LTE; 1160 return 3; 1161 case AsmToken::Greater: 1162 Kind = MCBinaryExpr::GT; 1163 return 3; 1164 case AsmToken::GreaterEqual: 1165 Kind = MCBinaryExpr::GTE; 1166 return 3; 1167 1168 // Intermediate Precedence: <<, >> 1169 case AsmToken::LessLess: 1170 Kind = MCBinaryExpr::Shl; 1171 return 4; 1172 case AsmToken::GreaterGreater: 1173 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1174 return 4; 1175 1176 // High Intermediate Precedence: +, - 1177 case AsmToken::Plus: 1178 Kind = MCBinaryExpr::Add; 1179 return 5; 1180 case AsmToken::Minus: 1181 Kind = MCBinaryExpr::Sub; 1182 return 5; 1183 1184 // Highest Precedence: *, /, % 1185 case AsmToken::Star: 1186 Kind = MCBinaryExpr::Mul; 1187 return 6; 1188 case AsmToken::Slash: 1189 Kind = MCBinaryExpr::Div; 1190 return 6; 1191 case AsmToken::Percent: 1192 Kind = MCBinaryExpr::Mod; 1193 return 6; 1194 } 1195 } 1196 1197 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1198 MCBinaryExpr::Opcode &Kind, 1199 bool ShouldUseLogicalShr) { 1200 switch (K) { 1201 default: 1202 return 0; // not a binop. 1203 1204 // Lowest Precedence: &&, || 1205 case AsmToken::AmpAmp: 1206 Kind = MCBinaryExpr::LAnd; 1207 return 2; 1208 case AsmToken::PipePipe: 1209 Kind = MCBinaryExpr::LOr; 1210 return 1; 1211 1212 // Low Precedence: ==, !=, <>, <, <=, >, >= 1213 case AsmToken::EqualEqual: 1214 Kind = MCBinaryExpr::EQ; 1215 return 3; 1216 case AsmToken::ExclaimEqual: 1217 case AsmToken::LessGreater: 1218 Kind = MCBinaryExpr::NE; 1219 return 3; 1220 case AsmToken::Less: 1221 Kind = MCBinaryExpr::LT; 1222 return 3; 1223 case AsmToken::LessEqual: 1224 Kind = MCBinaryExpr::LTE; 1225 return 3; 1226 case AsmToken::Greater: 1227 Kind = MCBinaryExpr::GT; 1228 return 3; 1229 case AsmToken::GreaterEqual: 1230 Kind = MCBinaryExpr::GTE; 1231 return 3; 1232 1233 // Low Intermediate Precedence: +, - 1234 case AsmToken::Plus: 1235 Kind = MCBinaryExpr::Add; 1236 return 4; 1237 case AsmToken::Minus: 1238 Kind = MCBinaryExpr::Sub; 1239 return 4; 1240 1241 // High Intermediate Precedence: |, &, ^ 1242 // 1243 // FIXME: gas seems to support '!' as an infix operator? 1244 case AsmToken::Pipe: 1245 Kind = MCBinaryExpr::Or; 1246 return 5; 1247 case AsmToken::Caret: 1248 Kind = MCBinaryExpr::Xor; 1249 return 5; 1250 case AsmToken::Amp: 1251 Kind = MCBinaryExpr::And; 1252 return 5; 1253 1254 // Highest Precedence: *, /, %, <<, >> 1255 case AsmToken::Star: 1256 Kind = MCBinaryExpr::Mul; 1257 return 6; 1258 case AsmToken::Slash: 1259 Kind = MCBinaryExpr::Div; 1260 return 6; 1261 case AsmToken::Percent: 1262 Kind = MCBinaryExpr::Mod; 1263 return 6; 1264 case AsmToken::LessLess: 1265 Kind = MCBinaryExpr::Shl; 1266 return 6; 1267 case AsmToken::GreaterGreater: 1268 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1269 return 6; 1270 } 1271 } 1272 1273 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1274 MCBinaryExpr::Opcode &Kind) { 1275 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1276 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1277 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1278 } 1279 1280 /// \brief Parse all binary operators with precedence >= 'Precedence'. 1281 /// Res contains the LHS of the expression on input. 1282 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1283 SMLoc &EndLoc) { 1284 while (1) { 1285 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1286 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1287 1288 // If the next token is lower precedence than we are allowed to eat, return 1289 // successfully with what we ate already. 1290 if (TokPrec < Precedence) 1291 return false; 1292 1293 Lex(); 1294 1295 // Eat the next primary expression. 1296 const MCExpr *RHS; 1297 if (parsePrimaryExpr(RHS, EndLoc)) 1298 return true; 1299 1300 // If BinOp binds less tightly with RHS than the operator after RHS, let 1301 // the pending operator take RHS as its LHS. 1302 MCBinaryExpr::Opcode Dummy; 1303 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1304 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1305 return true; 1306 1307 // Merge LHS and RHS according to operator. 1308 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext()); 1309 } 1310 } 1311 1312 /// ParseStatement: 1313 /// ::= EndOfStatement 1314 /// ::= Label* Directive ...Operands... EndOfStatement 1315 /// ::= Label* Identifier OperandList* EndOfStatement 1316 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1317 MCAsmParserSemaCallback *SI) { 1318 if (Lexer.is(AsmToken::EndOfStatement)) { 1319 Out.AddBlankLine(); 1320 Lex(); 1321 return false; 1322 } 1323 1324 // Statements always start with an identifier or are a full line comment. 1325 AsmToken ID = getTok(); 1326 SMLoc IDLoc = ID.getLoc(); 1327 StringRef IDVal; 1328 int64_t LocalLabelVal = -1; 1329 // A full line comment is a '#' as the first token. 1330 if (Lexer.is(AsmToken::Hash)) 1331 return parseCppHashLineFilenameComment(IDLoc); 1332 1333 // Allow an integer followed by a ':' as a directional local label. 1334 if (Lexer.is(AsmToken::Integer)) { 1335 LocalLabelVal = getTok().getIntVal(); 1336 if (LocalLabelVal < 0) { 1337 if (!TheCondState.Ignore) 1338 return TokError("unexpected token at start of statement"); 1339 IDVal = ""; 1340 } else { 1341 IDVal = getTok().getString(); 1342 Lex(); // Consume the integer token to be used as an identifier token. 1343 if (Lexer.getKind() != AsmToken::Colon) { 1344 if (!TheCondState.Ignore) 1345 return TokError("unexpected token at start of statement"); 1346 } 1347 } 1348 } else if (Lexer.is(AsmToken::Dot)) { 1349 // Treat '.' as a valid identifier in this context. 1350 Lex(); 1351 IDVal = "."; 1352 } else if (Lexer.is(AsmToken::LCurly)) { 1353 // Treat '{' as a valid identifier in this context. 1354 Lex(); 1355 IDVal = "{"; 1356 1357 } else if (Lexer.is(AsmToken::RCurly)) { 1358 // Treat '}' as a valid identifier in this context. 1359 Lex(); 1360 IDVal = "}"; 1361 } else if (parseIdentifier(IDVal)) { 1362 if (!TheCondState.Ignore) 1363 return TokError("unexpected token at start of statement"); 1364 IDVal = ""; 1365 } 1366 1367 // Handle conditional assembly here before checking for skipping. We 1368 // have to do this so that .endif isn't skipped in a ".if 0" block for 1369 // example. 1370 StringMap<DirectiveKind>::const_iterator DirKindIt = 1371 DirectiveKindMap.find(IDVal); 1372 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1373 ? DK_NO_DIRECTIVE 1374 : DirKindIt->getValue(); 1375 switch (DirKind) { 1376 default: 1377 break; 1378 case DK_IF: 1379 case DK_IFEQ: 1380 case DK_IFGE: 1381 case DK_IFGT: 1382 case DK_IFLE: 1383 case DK_IFLT: 1384 case DK_IFNE: 1385 return parseDirectiveIf(IDLoc, DirKind); 1386 case DK_IFB: 1387 return parseDirectiveIfb(IDLoc, true); 1388 case DK_IFNB: 1389 return parseDirectiveIfb(IDLoc, false); 1390 case DK_IFC: 1391 return parseDirectiveIfc(IDLoc, true); 1392 case DK_IFEQS: 1393 return parseDirectiveIfeqs(IDLoc, true); 1394 case DK_IFNC: 1395 return parseDirectiveIfc(IDLoc, false); 1396 case DK_IFNES: 1397 return parseDirectiveIfeqs(IDLoc, false); 1398 case DK_IFDEF: 1399 return parseDirectiveIfdef(IDLoc, true); 1400 case DK_IFNDEF: 1401 case DK_IFNOTDEF: 1402 return parseDirectiveIfdef(IDLoc, false); 1403 case DK_ELSEIF: 1404 return parseDirectiveElseIf(IDLoc); 1405 case DK_ELSE: 1406 return parseDirectiveElse(IDLoc); 1407 case DK_ENDIF: 1408 return parseDirectiveEndIf(IDLoc); 1409 } 1410 1411 // Ignore the statement if in the middle of inactive conditional 1412 // (e.g. ".if 0"). 1413 if (TheCondState.Ignore) { 1414 eatToEndOfStatement(); 1415 return false; 1416 } 1417 1418 // FIXME: Recurse on local labels? 1419 1420 // See what kind of statement we have. 1421 switch (Lexer.getKind()) { 1422 case AsmToken::Colon: { 1423 if (!getTargetParser().isLabel(ID)) 1424 break; 1425 checkForValidSection(); 1426 1427 // identifier ':' -> Label. 1428 Lex(); 1429 1430 // Diagnose attempt to use '.' as a label. 1431 if (IDVal == ".") 1432 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1433 1434 // Diagnose attempt to use a variable as a label. 1435 // 1436 // FIXME: Diagnostics. Note the location of the definition as a label. 1437 // FIXME: This doesn't diagnose assignment to a symbol which has been 1438 // implicitly marked as external. 1439 MCSymbol *Sym; 1440 if (LocalLabelVal == -1) { 1441 if (ParsingInlineAsm && SI) { 1442 StringRef RewrittenLabel = 1443 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1444 assert(RewrittenLabel.size() && 1445 "We should have an internal name here."); 1446 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1447 RewrittenLabel); 1448 IDVal = RewrittenLabel; 1449 } 1450 Sym = getContext().getOrCreateSymbol(IDVal); 1451 } else 1452 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1453 1454 Sym->redefineIfPossible(); 1455 1456 if (!Sym->isUndefined() || Sym->isVariable()) 1457 return Error(IDLoc, "invalid symbol redefinition"); 1458 1459 // Emit the label. 1460 if (!ParsingInlineAsm) 1461 Out.EmitLabel(Sym); 1462 1463 // If we are generating dwarf for assembly source files then gather the 1464 // info to make a dwarf label entry for this label if needed. 1465 if (getContext().getGenDwarfForAssembly()) 1466 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1467 IDLoc); 1468 1469 getTargetParser().onLabelParsed(Sym); 1470 1471 // Consume any end of statement token, if present, to avoid spurious 1472 // AddBlankLine calls(). 1473 if (Lexer.is(AsmToken::EndOfStatement)) { 1474 Lex(); 1475 if (Lexer.is(AsmToken::Eof)) 1476 return false; 1477 } 1478 1479 return false; 1480 } 1481 1482 case AsmToken::Equal: 1483 if (!getTargetParser().equalIsAsmAssignment()) 1484 break; 1485 // identifier '=' ... -> assignment statement 1486 Lex(); 1487 1488 return parseAssignment(IDVal, true); 1489 1490 default: // Normal instruction or directive. 1491 break; 1492 } 1493 1494 // If macros are enabled, check to see if this is a macro instantiation. 1495 if (areMacrosEnabled()) 1496 if (const MCAsmMacro *M = lookupMacro(IDVal)) { 1497 return handleMacroEntry(M, IDLoc); 1498 } 1499 1500 // Otherwise, we have a normal instruction or directive. 1501 1502 // Directives start with "." 1503 if (IDVal[0] == '.' && IDVal != ".") { 1504 // There are several entities interested in parsing directives: 1505 // 1506 // 1. The target-specific assembly parser. Some directives are target 1507 // specific or may potentially behave differently on certain targets. 1508 // 2. Asm parser extensions. For example, platform-specific parsers 1509 // (like the ELF parser) register themselves as extensions. 1510 // 3. The generic directive parser implemented by this class. These are 1511 // all the directives that behave in a target and platform independent 1512 // manner, or at least have a default behavior that's shared between 1513 // all targets and platforms. 1514 1515 // First query the target-specific parser. It will return 'true' if it 1516 // isn't interested in this directive. 1517 if (!getTargetParser().ParseDirective(ID)) 1518 return false; 1519 1520 // Next, check the extension directive map to see if any extension has 1521 // registered itself to parse this directive. 1522 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1523 ExtensionDirectiveMap.lookup(IDVal); 1524 if (Handler.first) 1525 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1526 1527 // Finally, if no one else is interested in this directive, it must be 1528 // generic and familiar to this class. 1529 switch (DirKind) { 1530 default: 1531 break; 1532 case DK_SET: 1533 case DK_EQU: 1534 return parseDirectiveSet(IDVal, true); 1535 case DK_EQUIV: 1536 return parseDirectiveSet(IDVal, false); 1537 case DK_ASCII: 1538 return parseDirectiveAscii(IDVal, false); 1539 case DK_ASCIZ: 1540 case DK_STRING: 1541 return parseDirectiveAscii(IDVal, true); 1542 case DK_BYTE: 1543 return parseDirectiveValue(1); 1544 case DK_SHORT: 1545 case DK_VALUE: 1546 case DK_2BYTE: 1547 return parseDirectiveValue(2); 1548 case DK_LONG: 1549 case DK_INT: 1550 case DK_4BYTE: 1551 return parseDirectiveValue(4); 1552 case DK_QUAD: 1553 case DK_8BYTE: 1554 return parseDirectiveValue(8); 1555 case DK_OCTA: 1556 return parseDirectiveOctaValue(); 1557 case DK_SINGLE: 1558 case DK_FLOAT: 1559 return parseDirectiveRealValue(APFloat::IEEEsingle); 1560 case DK_DOUBLE: 1561 return parseDirectiveRealValue(APFloat::IEEEdouble); 1562 case DK_ALIGN: { 1563 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1564 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1565 } 1566 case DK_ALIGN32: { 1567 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1568 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1569 } 1570 case DK_BALIGN: 1571 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1572 case DK_BALIGNW: 1573 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1574 case DK_BALIGNL: 1575 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1576 case DK_P2ALIGN: 1577 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1578 case DK_P2ALIGNW: 1579 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1580 case DK_P2ALIGNL: 1581 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1582 case DK_ORG: 1583 return parseDirectiveOrg(); 1584 case DK_FILL: 1585 return parseDirectiveFill(); 1586 case DK_ZERO: 1587 return parseDirectiveZero(); 1588 case DK_EXTERN: 1589 eatToEndOfStatement(); // .extern is the default, ignore it. 1590 return false; 1591 case DK_GLOBL: 1592 case DK_GLOBAL: 1593 return parseDirectiveSymbolAttribute(MCSA_Global); 1594 case DK_LAZY_REFERENCE: 1595 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1596 case DK_NO_DEAD_STRIP: 1597 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1598 case DK_SYMBOL_RESOLVER: 1599 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1600 case DK_PRIVATE_EXTERN: 1601 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1602 case DK_REFERENCE: 1603 return parseDirectiveSymbolAttribute(MCSA_Reference); 1604 case DK_WEAK_DEFINITION: 1605 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1606 case DK_WEAK_REFERENCE: 1607 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1608 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1609 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1610 case DK_COMM: 1611 case DK_COMMON: 1612 return parseDirectiveComm(/*IsLocal=*/false); 1613 case DK_LCOMM: 1614 return parseDirectiveComm(/*IsLocal=*/true); 1615 case DK_ABORT: 1616 return parseDirectiveAbort(); 1617 case DK_INCLUDE: 1618 return parseDirectiveInclude(); 1619 case DK_INCBIN: 1620 return parseDirectiveIncbin(); 1621 case DK_CODE16: 1622 case DK_CODE16GCC: 1623 return TokError(Twine(IDVal) + " not supported yet"); 1624 case DK_REPT: 1625 return parseDirectiveRept(IDLoc, IDVal); 1626 case DK_IRP: 1627 return parseDirectiveIrp(IDLoc); 1628 case DK_IRPC: 1629 return parseDirectiveIrpc(IDLoc); 1630 case DK_ENDR: 1631 return parseDirectiveEndr(IDLoc); 1632 case DK_BUNDLE_ALIGN_MODE: 1633 return parseDirectiveBundleAlignMode(); 1634 case DK_BUNDLE_LOCK: 1635 return parseDirectiveBundleLock(); 1636 case DK_BUNDLE_UNLOCK: 1637 return parseDirectiveBundleUnlock(); 1638 case DK_SLEB128: 1639 return parseDirectiveLEB128(true); 1640 case DK_ULEB128: 1641 return parseDirectiveLEB128(false); 1642 case DK_SPACE: 1643 case DK_SKIP: 1644 return parseDirectiveSpace(IDVal); 1645 case DK_FILE: 1646 return parseDirectiveFile(IDLoc); 1647 case DK_LINE: 1648 return parseDirectiveLine(); 1649 case DK_LOC: 1650 return parseDirectiveLoc(); 1651 case DK_STABS: 1652 return parseDirectiveStabs(); 1653 case DK_CV_FILE: 1654 return parseDirectiveCVFile(); 1655 case DK_CV_LOC: 1656 return parseDirectiveCVLoc(); 1657 case DK_CV_LINETABLE: 1658 return parseDirectiveCVLinetable(); 1659 case DK_CV_INLINE_LINETABLE: 1660 return parseDirectiveCVInlineLinetable(); 1661 case DK_CV_DEF_RANGE: 1662 return parseDirectiveCVDefRange(); 1663 case DK_CV_STRINGTABLE: 1664 return parseDirectiveCVStringTable(); 1665 case DK_CV_FILECHECKSUMS: 1666 return parseDirectiveCVFileChecksums(); 1667 case DK_CFI_SECTIONS: 1668 return parseDirectiveCFISections(); 1669 case DK_CFI_STARTPROC: 1670 return parseDirectiveCFIStartProc(); 1671 case DK_CFI_ENDPROC: 1672 return parseDirectiveCFIEndProc(); 1673 case DK_CFI_DEF_CFA: 1674 return parseDirectiveCFIDefCfa(IDLoc); 1675 case DK_CFI_DEF_CFA_OFFSET: 1676 return parseDirectiveCFIDefCfaOffset(); 1677 case DK_CFI_ADJUST_CFA_OFFSET: 1678 return parseDirectiveCFIAdjustCfaOffset(); 1679 case DK_CFI_DEF_CFA_REGISTER: 1680 return parseDirectiveCFIDefCfaRegister(IDLoc); 1681 case DK_CFI_OFFSET: 1682 return parseDirectiveCFIOffset(IDLoc); 1683 case DK_CFI_REL_OFFSET: 1684 return parseDirectiveCFIRelOffset(IDLoc); 1685 case DK_CFI_PERSONALITY: 1686 return parseDirectiveCFIPersonalityOrLsda(true); 1687 case DK_CFI_LSDA: 1688 return parseDirectiveCFIPersonalityOrLsda(false); 1689 case DK_CFI_REMEMBER_STATE: 1690 return parseDirectiveCFIRememberState(); 1691 case DK_CFI_RESTORE_STATE: 1692 return parseDirectiveCFIRestoreState(); 1693 case DK_CFI_SAME_VALUE: 1694 return parseDirectiveCFISameValue(IDLoc); 1695 case DK_CFI_RESTORE: 1696 return parseDirectiveCFIRestore(IDLoc); 1697 case DK_CFI_ESCAPE: 1698 return parseDirectiveCFIEscape(); 1699 case DK_CFI_SIGNAL_FRAME: 1700 return parseDirectiveCFISignalFrame(); 1701 case DK_CFI_UNDEFINED: 1702 return parseDirectiveCFIUndefined(IDLoc); 1703 case DK_CFI_REGISTER: 1704 return parseDirectiveCFIRegister(IDLoc); 1705 case DK_CFI_WINDOW_SAVE: 1706 return parseDirectiveCFIWindowSave(); 1707 case DK_MACROS_ON: 1708 case DK_MACROS_OFF: 1709 return parseDirectiveMacrosOnOff(IDVal); 1710 case DK_MACRO: 1711 return parseDirectiveMacro(IDLoc); 1712 case DK_EXITM: 1713 return parseDirectiveExitMacro(IDVal); 1714 case DK_ENDM: 1715 case DK_ENDMACRO: 1716 return parseDirectiveEndMacro(IDVal); 1717 case DK_PURGEM: 1718 return parseDirectivePurgeMacro(IDLoc); 1719 case DK_END: 1720 return parseDirectiveEnd(IDLoc); 1721 case DK_ERR: 1722 return parseDirectiveError(IDLoc, false); 1723 case DK_ERROR: 1724 return parseDirectiveError(IDLoc, true); 1725 case DK_WARNING: 1726 return parseDirectiveWarning(IDLoc); 1727 case DK_RELOC: 1728 return parseDirectiveReloc(IDLoc); 1729 } 1730 1731 return Error(IDLoc, "unknown directive"); 1732 } 1733 1734 // __asm _emit or __asm __emit 1735 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 1736 IDVal == "_EMIT" || IDVal == "__EMIT")) 1737 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 1738 1739 // __asm align 1740 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 1741 return parseDirectiveMSAlign(IDLoc, Info); 1742 1743 if (ParsingInlineAsm && (IDVal == "even")) 1744 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 1745 checkForValidSection(); 1746 1747 // Canonicalize the opcode to lower case. 1748 std::string OpcodeStr = IDVal.lower(); 1749 ParseInstructionInfo IInfo(Info.AsmRewrites); 1750 bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 1751 Info.ParsedOperands); 1752 Info.ParseError = HadError; 1753 1754 // Dump the parsed representation, if requested. 1755 if (getShowParsedOperands()) { 1756 SmallString<256> Str; 1757 raw_svector_ostream OS(Str); 1758 OS << "parsed instruction: ["; 1759 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 1760 if (i != 0) 1761 OS << ", "; 1762 Info.ParsedOperands[i]->print(OS); 1763 } 1764 OS << "]"; 1765 1766 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 1767 } 1768 1769 // If we are generating dwarf for the current section then generate a .loc 1770 // directive for the instruction. 1771 if (!HadError && getContext().getGenDwarfForAssembly() && 1772 getContext().getGenDwarfSectionSyms().count( 1773 getStreamer().getCurrentSection().first)) { 1774 unsigned Line; 1775 if (ActiveMacros.empty()) 1776 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 1777 else 1778 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 1779 ActiveMacros.front()->ExitBuffer); 1780 1781 // If we previously parsed a cpp hash file line comment then make sure the 1782 // current Dwarf File is for the CppHashFilename if not then emit the 1783 // Dwarf File table for it and adjust the line number for the .loc. 1784 if (CppHashFilename.size()) { 1785 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 1786 0, StringRef(), CppHashFilename); 1787 getContext().setGenDwarfFileNumber(FileNumber); 1788 1789 // Since SrcMgr.FindLineNumber() is slow and messes up the SourceMgr's 1790 // cache with the different Loc from the call above we save the last 1791 // info we queried here with SrcMgr.FindLineNumber(). 1792 unsigned CppHashLocLineNo; 1793 if (LastQueryIDLoc == CppHashLoc && LastQueryBuffer == CppHashBuf) 1794 CppHashLocLineNo = LastQueryLine; 1795 else { 1796 CppHashLocLineNo = SrcMgr.FindLineNumber(CppHashLoc, CppHashBuf); 1797 LastQueryLine = CppHashLocLineNo; 1798 LastQueryIDLoc = CppHashLoc; 1799 LastQueryBuffer = CppHashBuf; 1800 } 1801 Line = CppHashLineNumber - 1 + (Line - CppHashLocLineNo); 1802 } 1803 1804 getStreamer().EmitDwarfLocDirective( 1805 getContext().getGenDwarfFileNumber(), Line, 0, 1806 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 1807 StringRef()); 1808 } 1809 1810 // If parsing succeeded, match the instruction. 1811 if (!HadError) { 1812 uint64_t ErrorInfo; 1813 getTargetParser().MatchAndEmitInstruction(IDLoc, Info.Opcode, 1814 Info.ParsedOperands, Out, 1815 ErrorInfo, ParsingInlineAsm); 1816 } 1817 1818 // Don't skip the rest of the line, the instruction parser is responsible for 1819 // that. 1820 return false; 1821 } 1822 1823 /// eatToEndOfLine uses the Lexer to eat the characters to the end of the line 1824 /// since they may not be able to be tokenized to get to the end of line token. 1825 void AsmParser::eatToEndOfLine() { 1826 if (!Lexer.is(AsmToken::EndOfStatement)) 1827 Lexer.LexUntilEndOfLine(); 1828 // Eat EOL. 1829 Lex(); 1830 } 1831 1832 /// parseCppHashLineFilenameComment as this: 1833 /// ::= # number "filename" 1834 /// or just as a full line comment if it doesn't have a number and a string. 1835 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) { 1836 Lex(); // Eat the hash token. 1837 1838 if (getLexer().isNot(AsmToken::Integer)) { 1839 // Consume the line since in cases it is not a well-formed line directive, 1840 // as if were simply a full line comment. 1841 eatToEndOfLine(); 1842 return false; 1843 } 1844 1845 int64_t LineNumber = getTok().getIntVal(); 1846 Lex(); 1847 1848 if (getLexer().isNot(AsmToken::String)) { 1849 eatToEndOfLine(); 1850 return false; 1851 } 1852 1853 StringRef Filename = getTok().getString(); 1854 // Get rid of the enclosing quotes. 1855 Filename = Filename.substr(1, Filename.size() - 2); 1856 1857 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 1858 CppHashLoc = L; 1859 CppHashFilename = Filename; 1860 CppHashLineNumber = LineNumber; 1861 CppHashBuf = CurBuffer; 1862 1863 // Ignore any trailing characters, they're just comment. 1864 eatToEndOfLine(); 1865 return false; 1866 } 1867 1868 /// \brief will use the last parsed cpp hash line filename comment 1869 /// for the Filename and LineNo if any in the diagnostic. 1870 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 1871 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 1872 raw_ostream &OS = errs(); 1873 1874 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 1875 SMLoc DiagLoc = Diag.getLoc(); 1876 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1877 unsigned CppHashBuf = 1878 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashLoc); 1879 1880 // Like SourceMgr::printMessage() we need to print the include stack if any 1881 // before printing the message. 1882 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 1883 if (!Parser->SavedDiagHandler && DiagCurBuffer && 1884 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 1885 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 1886 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 1887 } 1888 1889 // If we have not parsed a cpp hash line filename comment or the source 1890 // manager changed or buffer changed (like in a nested include) then just 1891 // print the normal diagnostic using its Filename and LineNo. 1892 if (!Parser->CppHashLineNumber || &DiagSrcMgr != &Parser->SrcMgr || 1893 DiagBuf != CppHashBuf) { 1894 if (Parser->SavedDiagHandler) 1895 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 1896 else 1897 Diag.print(nullptr, OS); 1898 return; 1899 } 1900 1901 // Use the CppHashFilename and calculate a line number based on the 1902 // CppHashLoc and CppHashLineNumber relative to this Diag's SMLoc for 1903 // the diagnostic. 1904 const std::string &Filename = Parser->CppHashFilename; 1905 1906 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 1907 int CppHashLocLineNo = 1908 Parser->SrcMgr.FindLineNumber(Parser->CppHashLoc, CppHashBuf); 1909 int LineNo = 1910 Parser->CppHashLineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 1911 1912 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 1913 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 1914 Diag.getLineContents(), Diag.getRanges()); 1915 1916 if (Parser->SavedDiagHandler) 1917 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 1918 else 1919 NewDiag.print(nullptr, OS); 1920 } 1921 1922 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 1923 // difference being that that function accepts '@' as part of identifiers and 1924 // we can't do that. AsmLexer.cpp should probably be changed to handle 1925 // '@' as a special case when needed. 1926 static bool isIdentifierChar(char c) { 1927 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 1928 c == '.'; 1929 } 1930 1931 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 1932 ArrayRef<MCAsmMacroParameter> Parameters, 1933 ArrayRef<MCAsmMacroArgument> A, 1934 bool EnableAtPseudoVariable, SMLoc L) { 1935 unsigned NParameters = Parameters.size(); 1936 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 1937 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 1938 return Error(L, "Wrong number of arguments"); 1939 1940 // A macro without parameters is handled differently on Darwin: 1941 // gas accepts no arguments and does no substitutions 1942 while (!Body.empty()) { 1943 // Scan for the next substitution. 1944 std::size_t End = Body.size(), Pos = 0; 1945 for (; Pos != End; ++Pos) { 1946 // Check for a substitution or escape. 1947 if (IsDarwin && !NParameters) { 1948 // This macro has no parameters, look for $0, $1, etc. 1949 if (Body[Pos] != '$' || Pos + 1 == End) 1950 continue; 1951 1952 char Next = Body[Pos + 1]; 1953 if (Next == '$' || Next == 'n' || 1954 isdigit(static_cast<unsigned char>(Next))) 1955 break; 1956 } else { 1957 // This macro has parameters, look for \foo, \bar, etc. 1958 if (Body[Pos] == '\\' && Pos + 1 != End) 1959 break; 1960 } 1961 } 1962 1963 // Add the prefix. 1964 OS << Body.slice(0, Pos); 1965 1966 // Check if we reached the end. 1967 if (Pos == End) 1968 break; 1969 1970 if (IsDarwin && !NParameters) { 1971 switch (Body[Pos + 1]) { 1972 // $$ => $ 1973 case '$': 1974 OS << '$'; 1975 break; 1976 1977 // $n => number of arguments 1978 case 'n': 1979 OS << A.size(); 1980 break; 1981 1982 // $[0-9] => argument 1983 default: { 1984 // Missing arguments are ignored. 1985 unsigned Index = Body[Pos + 1] - '0'; 1986 if (Index >= A.size()) 1987 break; 1988 1989 // Otherwise substitute with the token values, with spaces eliminated. 1990 for (const AsmToken &Token : A[Index]) 1991 OS << Token.getString(); 1992 break; 1993 } 1994 } 1995 Pos += 2; 1996 } else { 1997 unsigned I = Pos + 1; 1998 1999 // Check for the \@ pseudo-variable. 2000 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2001 ++I; 2002 else 2003 while (isIdentifierChar(Body[I]) && I + 1 != End) 2004 ++I; 2005 2006 const char *Begin = Body.data() + Pos + 1; 2007 StringRef Argument(Begin, I - (Pos + 1)); 2008 unsigned Index = 0; 2009 2010 if (Argument == "@") { 2011 OS << NumOfMacroInstantiations; 2012 Pos += 2; 2013 } else { 2014 for (; Index < NParameters; ++Index) 2015 if (Parameters[Index].Name == Argument) 2016 break; 2017 2018 if (Index == NParameters) { 2019 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2020 Pos += 3; 2021 else { 2022 OS << '\\' << Argument; 2023 Pos = I; 2024 } 2025 } else { 2026 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2027 for (const AsmToken &Token : A[Index]) 2028 // We expect no quotes around the string's contents when 2029 // parsing for varargs. 2030 if (Token.getKind() != AsmToken::String || VarargParameter) 2031 OS << Token.getString(); 2032 else 2033 OS << Token.getStringContents(); 2034 2035 Pos += 1 + Argument.size(); 2036 } 2037 } 2038 } 2039 // Update the scan point. 2040 Body = Body.substr(Pos); 2041 } 2042 2043 return false; 2044 } 2045 2046 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 2047 size_t CondStackDepth) 2048 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 2049 CondStackDepth(CondStackDepth) {} 2050 2051 static bool isOperator(AsmToken::TokenKind kind) { 2052 switch (kind) { 2053 default: 2054 return false; 2055 case AsmToken::Plus: 2056 case AsmToken::Minus: 2057 case AsmToken::Tilde: 2058 case AsmToken::Slash: 2059 case AsmToken::Star: 2060 case AsmToken::Dot: 2061 case AsmToken::Equal: 2062 case AsmToken::EqualEqual: 2063 case AsmToken::Pipe: 2064 case AsmToken::PipePipe: 2065 case AsmToken::Caret: 2066 case AsmToken::Amp: 2067 case AsmToken::AmpAmp: 2068 case AsmToken::Exclaim: 2069 case AsmToken::ExclaimEqual: 2070 case AsmToken::Less: 2071 case AsmToken::LessEqual: 2072 case AsmToken::LessLess: 2073 case AsmToken::LessGreater: 2074 case AsmToken::Greater: 2075 case AsmToken::GreaterEqual: 2076 case AsmToken::GreaterGreater: 2077 return true; 2078 } 2079 } 2080 2081 namespace { 2082 class AsmLexerSkipSpaceRAII { 2083 public: 2084 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2085 Lexer.setSkipSpace(SkipSpace); 2086 } 2087 2088 ~AsmLexerSkipSpaceRAII() { 2089 Lexer.setSkipSpace(true); 2090 } 2091 2092 private: 2093 AsmLexer &Lexer; 2094 }; 2095 } 2096 2097 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2098 2099 if (Vararg) { 2100 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2101 StringRef Str = parseStringToEndOfStatement(); 2102 MA.emplace_back(AsmToken::String, Str); 2103 } 2104 return false; 2105 } 2106 2107 unsigned ParenLevel = 0; 2108 2109 // Darwin doesn't use spaces to delmit arguments. 2110 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2111 2112 bool SpaceEaten; 2113 2114 for (;;) { 2115 SpaceEaten = false; 2116 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2117 return TokError("unexpected token in macro instantiation"); 2118 2119 if (ParenLevel == 0) { 2120 2121 if (Lexer.is(AsmToken::Comma)) 2122 break; 2123 2124 if (Lexer.is(AsmToken::Space)) { 2125 SpaceEaten = true; 2126 Lex(); // Eat spaces 2127 } 2128 2129 // Spaces can delimit parameters, but could also be part an expression. 2130 // If the token after a space is an operator, add the token and the next 2131 // one into this argument 2132 if (!IsDarwin) { 2133 if (isOperator(Lexer.getKind())) { 2134 MA.push_back(getTok()); 2135 Lex(); 2136 2137 // Whitespace after an operator can be ignored. 2138 if (Lexer.is(AsmToken::Space)) 2139 Lex(); 2140 2141 continue; 2142 } 2143 } 2144 if (SpaceEaten) 2145 break; 2146 } 2147 2148 // handleMacroEntry relies on not advancing the lexer here 2149 // to be able to fill in the remaining default parameter values 2150 if (Lexer.is(AsmToken::EndOfStatement)) 2151 break; 2152 2153 // Adjust the current parentheses level. 2154 if (Lexer.is(AsmToken::LParen)) 2155 ++ParenLevel; 2156 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2157 --ParenLevel; 2158 2159 // Append the token to the current argument list. 2160 MA.push_back(getTok()); 2161 Lex(); 2162 } 2163 2164 if (ParenLevel != 0) 2165 return TokError("unbalanced parentheses in macro argument"); 2166 return false; 2167 } 2168 2169 // Parse the macro instantiation arguments. 2170 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2171 MCAsmMacroArguments &A) { 2172 const unsigned NParameters = M ? M->Parameters.size() : 0; 2173 bool NamedParametersFound = false; 2174 SmallVector<SMLoc, 4> FALocs; 2175 2176 A.resize(NParameters); 2177 FALocs.resize(NParameters); 2178 2179 // Parse two kinds of macro invocations: 2180 // - macros defined without any parameters accept an arbitrary number of them 2181 // - macros defined with parameters accept at most that many of them 2182 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2183 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2184 ++Parameter) { 2185 SMLoc IDLoc = Lexer.getLoc(); 2186 MCAsmMacroParameter FA; 2187 2188 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2189 if (parseIdentifier(FA.Name)) { 2190 Error(IDLoc, "invalid argument identifier for formal argument"); 2191 eatToEndOfStatement(); 2192 return true; 2193 } 2194 2195 if (!Lexer.is(AsmToken::Equal)) { 2196 TokError("expected '=' after formal parameter identifier"); 2197 eatToEndOfStatement(); 2198 return true; 2199 } 2200 Lex(); 2201 2202 NamedParametersFound = true; 2203 } 2204 2205 if (NamedParametersFound && FA.Name.empty()) { 2206 Error(IDLoc, "cannot mix positional and keyword arguments"); 2207 eatToEndOfStatement(); 2208 return true; 2209 } 2210 2211 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2212 if (parseMacroArgument(FA.Value, Vararg)) 2213 return true; 2214 2215 unsigned PI = Parameter; 2216 if (!FA.Name.empty()) { 2217 unsigned FAI = 0; 2218 for (FAI = 0; FAI < NParameters; ++FAI) 2219 if (M->Parameters[FAI].Name == FA.Name) 2220 break; 2221 2222 if (FAI >= NParameters) { 2223 assert(M && "expected macro to be defined"); 2224 Error(IDLoc, 2225 "parameter named '" + FA.Name + "' does not exist for macro '" + 2226 M->Name + "'"); 2227 return true; 2228 } 2229 PI = FAI; 2230 } 2231 2232 if (!FA.Value.empty()) { 2233 if (A.size() <= PI) 2234 A.resize(PI + 1); 2235 A[PI] = FA.Value; 2236 2237 if (FALocs.size() <= PI) 2238 FALocs.resize(PI + 1); 2239 2240 FALocs[PI] = Lexer.getLoc(); 2241 } 2242 2243 // At the end of the statement, fill in remaining arguments that have 2244 // default values. If there aren't any, then the next argument is 2245 // required but missing 2246 if (Lexer.is(AsmToken::EndOfStatement)) { 2247 bool Failure = false; 2248 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2249 if (A[FAI].empty()) { 2250 if (M->Parameters[FAI].Required) { 2251 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2252 "missing value for required parameter " 2253 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2254 Failure = true; 2255 } 2256 2257 if (!M->Parameters[FAI].Value.empty()) 2258 A[FAI] = M->Parameters[FAI].Value; 2259 } 2260 } 2261 return Failure; 2262 } 2263 2264 if (Lexer.is(AsmToken::Comma)) 2265 Lex(); 2266 } 2267 2268 return TokError("too many positional arguments"); 2269 } 2270 2271 const MCAsmMacro *AsmParser::lookupMacro(StringRef Name) { 2272 StringMap<MCAsmMacro>::iterator I = MacroMap.find(Name); 2273 return (I == MacroMap.end()) ? nullptr : &I->getValue(); 2274 } 2275 2276 void AsmParser::defineMacro(StringRef Name, MCAsmMacro Macro) { 2277 MacroMap.insert(std::make_pair(Name, std::move(Macro))); 2278 } 2279 2280 void AsmParser::undefineMacro(StringRef Name) { MacroMap.erase(Name); } 2281 2282 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2283 // Arbitrarily limit macro nesting depth, to match 'as'. We can eliminate 2284 // this, although we should protect against infinite loops. 2285 if (ActiveMacros.size() == 20) 2286 return TokError("macros cannot be nested more than 20 levels deep"); 2287 2288 MCAsmMacroArguments A; 2289 if (parseMacroArguments(M, A)) 2290 return true; 2291 2292 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2293 // to hold the macro body with substitutions. 2294 SmallString<256> Buf; 2295 StringRef Body = M->Body; 2296 raw_svector_ostream OS(Buf); 2297 2298 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2299 return true; 2300 2301 // We include the .endmacro in the buffer as our cue to exit the macro 2302 // instantiation. 2303 OS << ".endmacro\n"; 2304 2305 std::unique_ptr<MemoryBuffer> Instantiation = 2306 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2307 2308 // Create the macro instantiation object and add to the current macro 2309 // instantiation stack. 2310 MacroInstantiation *MI = new MacroInstantiation( 2311 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2312 ActiveMacros.push_back(MI); 2313 2314 ++NumOfMacroInstantiations; 2315 2316 // Jump to the macro instantiation and prime the lexer. 2317 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2318 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2319 Lex(); 2320 2321 return false; 2322 } 2323 2324 void AsmParser::handleMacroExit() { 2325 // Jump to the EndOfStatement we should return to, and consume it. 2326 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2327 Lex(); 2328 2329 // Pop the instantiation entry. 2330 delete ActiveMacros.back(); 2331 ActiveMacros.pop_back(); 2332 } 2333 2334 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2335 bool NoDeadStrip) { 2336 MCSymbol *Sym; 2337 const MCExpr *Value; 2338 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2339 Value)) 2340 return true; 2341 2342 if (!Sym) { 2343 // In the case where we parse an expression starting with a '.', we will 2344 // not generate an error, nor will we create a symbol. In this case we 2345 // should just return out. 2346 return false; 2347 } 2348 2349 // Do the assignment. 2350 Out.EmitAssignment(Sym, Value); 2351 if (NoDeadStrip) 2352 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2353 2354 return false; 2355 } 2356 2357 /// parseIdentifier: 2358 /// ::= identifier 2359 /// ::= string 2360 bool AsmParser::parseIdentifier(StringRef &Res) { 2361 // The assembler has relaxed rules for accepting identifiers, in particular we 2362 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2363 // separate tokens. At this level, we have already lexed so we cannot (currently) 2364 // handle this as a context dependent token, instead we detect adjacent tokens 2365 // and return the combined identifier. 2366 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2367 SMLoc PrefixLoc = getLexer().getLoc(); 2368 2369 // Consume the prefix character, and check for a following identifier. 2370 Lex(); 2371 if (Lexer.isNot(AsmToken::Identifier)) 2372 return true; 2373 2374 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2375 if (PrefixLoc.getPointer() + 1 != getTok().getLoc().getPointer()) 2376 return true; 2377 2378 // Construct the joined identifier and consume the token. 2379 Res = 2380 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2381 Lex(); 2382 return false; 2383 } 2384 2385 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2386 return true; 2387 2388 Res = getTok().getIdentifier(); 2389 2390 Lex(); // Consume the identifier token. 2391 2392 return false; 2393 } 2394 2395 /// parseDirectiveSet: 2396 /// ::= .equ identifier ',' expression 2397 /// ::= .equiv identifier ',' expression 2398 /// ::= .set identifier ',' expression 2399 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2400 StringRef Name; 2401 2402 if (parseIdentifier(Name)) 2403 return TokError("expected identifier after '" + Twine(IDVal) + "'"); 2404 2405 if (getLexer().isNot(AsmToken::Comma)) 2406 return TokError("unexpected token in '" + Twine(IDVal) + "'"); 2407 Lex(); 2408 2409 return parseAssignment(Name, allow_redef, true); 2410 } 2411 2412 bool AsmParser::parseEscapedString(std::string &Data) { 2413 assert(getLexer().is(AsmToken::String) && "Unexpected current token!"); 2414 2415 Data = ""; 2416 StringRef Str = getTok().getStringContents(); 2417 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2418 if (Str[i] != '\\') { 2419 Data += Str[i]; 2420 continue; 2421 } 2422 2423 // Recognize escaped characters. Note that this escape semantics currently 2424 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2425 ++i; 2426 if (i == e) 2427 return TokError("unexpected backslash at end of string"); 2428 2429 // Recognize octal sequences. 2430 if ((unsigned)(Str[i] - '0') <= 7) { 2431 // Consume up to three octal characters. 2432 unsigned Value = Str[i] - '0'; 2433 2434 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2435 ++i; 2436 Value = Value * 8 + (Str[i] - '0'); 2437 2438 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2439 ++i; 2440 Value = Value * 8 + (Str[i] - '0'); 2441 } 2442 } 2443 2444 if (Value > 255) 2445 return TokError("invalid octal escape sequence (out of range)"); 2446 2447 Data += (unsigned char)Value; 2448 continue; 2449 } 2450 2451 // Otherwise recognize individual escapes. 2452 switch (Str[i]) { 2453 default: 2454 // Just reject invalid escape sequences for now. 2455 return TokError("invalid escape sequence (unrecognized character)"); 2456 2457 case 'b': Data += '\b'; break; 2458 case 'f': Data += '\f'; break; 2459 case 'n': Data += '\n'; break; 2460 case 'r': Data += '\r'; break; 2461 case 't': Data += '\t'; break; 2462 case '"': Data += '"'; break; 2463 case '\\': Data += '\\'; break; 2464 } 2465 } 2466 2467 return false; 2468 } 2469 2470 /// parseDirectiveAscii: 2471 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2472 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2473 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2474 checkForValidSection(); 2475 2476 for (;;) { 2477 if (getLexer().isNot(AsmToken::String)) 2478 return TokError("expected string in '" + Twine(IDVal) + "' directive"); 2479 2480 std::string Data; 2481 if (parseEscapedString(Data)) 2482 return true; 2483 2484 getStreamer().EmitBytes(Data); 2485 if (ZeroTerminated) 2486 getStreamer().EmitBytes(StringRef("\0", 1)); 2487 2488 Lex(); 2489 2490 if (getLexer().is(AsmToken::EndOfStatement)) 2491 break; 2492 2493 if (getLexer().isNot(AsmToken::Comma)) 2494 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 2495 Lex(); 2496 } 2497 } 2498 2499 Lex(); 2500 return false; 2501 } 2502 2503 /// parseDirectiveReloc 2504 /// ::= .reloc expression , identifier [ , expression ] 2505 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 2506 const MCExpr *Offset; 2507 const MCExpr *Expr = nullptr; 2508 2509 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 2510 if (parseExpression(Offset)) 2511 return true; 2512 2513 // We can only deal with constant expressions at the moment. 2514 int64_t OffsetValue; 2515 if (!Offset->evaluateAsAbsolute(OffsetValue)) 2516 return Error(OffsetLoc, "expression is not a constant value"); 2517 2518 if (OffsetValue < 0) 2519 return Error(OffsetLoc, "expression is negative"); 2520 2521 if (Lexer.isNot(AsmToken::Comma)) 2522 return TokError("expected comma"); 2523 Lexer.Lex(); 2524 2525 if (Lexer.isNot(AsmToken::Identifier)) 2526 return TokError("expected relocation name"); 2527 SMLoc NameLoc = Lexer.getTok().getLoc(); 2528 StringRef Name = Lexer.getTok().getIdentifier(); 2529 Lexer.Lex(); 2530 2531 if (Lexer.is(AsmToken::Comma)) { 2532 Lexer.Lex(); 2533 SMLoc ExprLoc = Lexer.getLoc(); 2534 if (parseExpression(Expr)) 2535 return true; 2536 2537 MCValue Value; 2538 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 2539 return Error(ExprLoc, "expression must be relocatable"); 2540 } 2541 2542 if (Lexer.isNot(AsmToken::EndOfStatement)) 2543 return TokError("unexpected token in .reloc directive"); 2544 2545 if (getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc)) 2546 return Error(NameLoc, "unknown relocation name"); 2547 2548 return false; 2549 } 2550 2551 /// parseDirectiveValue 2552 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 2553 bool AsmParser::parseDirectiveValue(unsigned Size) { 2554 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2555 checkForValidSection(); 2556 2557 for (;;) { 2558 const MCExpr *Value; 2559 SMLoc ExprLoc = getLexer().getLoc(); 2560 if (parseExpression(Value)) 2561 return true; 2562 2563 // Special case constant expressions to match code generator. 2564 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2565 assert(Size <= 8 && "Invalid size"); 2566 uint64_t IntValue = MCE->getValue(); 2567 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2568 return Error(ExprLoc, "literal value out of range for directive"); 2569 getStreamer().EmitIntValue(IntValue, Size); 2570 } else 2571 getStreamer().EmitValue(Value, Size, ExprLoc); 2572 2573 if (getLexer().is(AsmToken::EndOfStatement)) 2574 break; 2575 2576 // FIXME: Improve diagnostic. 2577 if (getLexer().isNot(AsmToken::Comma)) 2578 return TokError("unexpected token in directive"); 2579 Lex(); 2580 } 2581 } 2582 2583 Lex(); 2584 return false; 2585 } 2586 2587 /// ParseDirectiveOctaValue 2588 /// ::= .octa [ hexconstant (, hexconstant)* ] 2589 bool AsmParser::parseDirectiveOctaValue() { 2590 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2591 checkForValidSection(); 2592 2593 for (;;) { 2594 if (Lexer.getKind() == AsmToken::Error) 2595 return true; 2596 if (Lexer.getKind() != AsmToken::Integer && 2597 Lexer.getKind() != AsmToken::BigNum) 2598 return TokError("unknown token in expression"); 2599 2600 SMLoc ExprLoc = getLexer().getLoc(); 2601 APInt IntValue = getTok().getAPIntVal(); 2602 Lex(); 2603 2604 uint64_t hi, lo; 2605 if (IntValue.isIntN(64)) { 2606 hi = 0; 2607 lo = IntValue.getZExtValue(); 2608 } else if (IntValue.isIntN(128)) { 2609 // It might actually have more than 128 bits, but the top ones are zero. 2610 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2611 lo = IntValue.getLoBits(64).getZExtValue(); 2612 } else 2613 return Error(ExprLoc, "literal value out of range for directive"); 2614 2615 if (MAI.isLittleEndian()) { 2616 getStreamer().EmitIntValue(lo, 8); 2617 getStreamer().EmitIntValue(hi, 8); 2618 } else { 2619 getStreamer().EmitIntValue(hi, 8); 2620 getStreamer().EmitIntValue(lo, 8); 2621 } 2622 2623 if (getLexer().is(AsmToken::EndOfStatement)) 2624 break; 2625 2626 // FIXME: Improve diagnostic. 2627 if (getLexer().isNot(AsmToken::Comma)) 2628 return TokError("unexpected token in directive"); 2629 Lex(); 2630 } 2631 } 2632 2633 Lex(); 2634 return false; 2635 } 2636 2637 /// parseDirectiveRealValue 2638 /// ::= (.single | .double) [ expression (, expression)* ] 2639 bool AsmParser::parseDirectiveRealValue(const fltSemantics &Semantics) { 2640 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2641 checkForValidSection(); 2642 2643 for (;;) { 2644 // We don't truly support arithmetic on floating point expressions, so we 2645 // have to manually parse unary prefixes. 2646 bool IsNeg = false; 2647 if (getLexer().is(AsmToken::Minus)) { 2648 Lex(); 2649 IsNeg = true; 2650 } else if (getLexer().is(AsmToken::Plus)) 2651 Lex(); 2652 2653 if (getLexer().isNot(AsmToken::Integer) && 2654 getLexer().isNot(AsmToken::Real) && 2655 getLexer().isNot(AsmToken::Identifier)) 2656 return TokError("unexpected token in directive"); 2657 2658 // Convert to an APFloat. 2659 APFloat Value(Semantics); 2660 StringRef IDVal = getTok().getString(); 2661 if (getLexer().is(AsmToken::Identifier)) { 2662 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2663 Value = APFloat::getInf(Semantics); 2664 else if (!IDVal.compare_lower("nan")) 2665 Value = APFloat::getNaN(Semantics, false, ~0); 2666 else 2667 return TokError("invalid floating point literal"); 2668 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 2669 APFloat::opInvalidOp) 2670 return TokError("invalid floating point literal"); 2671 if (IsNeg) 2672 Value.changeSign(); 2673 2674 // Consume the numeric token. 2675 Lex(); 2676 2677 // Emit the value as an integer. 2678 APInt AsInt = Value.bitcastToAPInt(); 2679 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 2680 AsInt.getBitWidth() / 8); 2681 2682 if (getLexer().is(AsmToken::EndOfStatement)) 2683 break; 2684 2685 if (getLexer().isNot(AsmToken::Comma)) 2686 return TokError("unexpected token in directive"); 2687 Lex(); 2688 } 2689 } 2690 2691 Lex(); 2692 return false; 2693 } 2694 2695 /// parseDirectiveZero 2696 /// ::= .zero expression 2697 bool AsmParser::parseDirectiveZero() { 2698 checkForValidSection(); 2699 2700 int64_t NumBytes; 2701 if (parseAbsoluteExpression(NumBytes)) 2702 return true; 2703 2704 int64_t Val = 0; 2705 if (getLexer().is(AsmToken::Comma)) { 2706 Lex(); 2707 if (parseAbsoluteExpression(Val)) 2708 return true; 2709 } 2710 2711 if (getLexer().isNot(AsmToken::EndOfStatement)) 2712 return TokError("unexpected token in '.zero' directive"); 2713 2714 Lex(); 2715 2716 getStreamer().EmitFill(NumBytes, Val); 2717 2718 return false; 2719 } 2720 2721 /// parseDirectiveFill 2722 /// ::= .fill expression [ , expression [ , expression ] ] 2723 bool AsmParser::parseDirectiveFill() { 2724 checkForValidSection(); 2725 2726 SMLoc RepeatLoc = getLexer().getLoc(); 2727 int64_t NumValues; 2728 if (parseAbsoluteExpression(NumValues)) 2729 return true; 2730 2731 if (NumValues < 0) { 2732 Warning(RepeatLoc, 2733 "'.fill' directive with negative repeat count has no effect"); 2734 NumValues = 0; 2735 } 2736 2737 int64_t FillSize = 1; 2738 int64_t FillExpr = 0; 2739 2740 SMLoc SizeLoc, ExprLoc; 2741 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2742 if (getLexer().isNot(AsmToken::Comma)) 2743 return TokError("unexpected token in '.fill' directive"); 2744 Lex(); 2745 2746 SizeLoc = getLexer().getLoc(); 2747 if (parseAbsoluteExpression(FillSize)) 2748 return true; 2749 2750 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2751 if (getLexer().isNot(AsmToken::Comma)) 2752 return TokError("unexpected token in '.fill' directive"); 2753 Lex(); 2754 2755 ExprLoc = getLexer().getLoc(); 2756 if (parseAbsoluteExpression(FillExpr)) 2757 return true; 2758 2759 if (getLexer().isNot(AsmToken::EndOfStatement)) 2760 return TokError("unexpected token in '.fill' directive"); 2761 2762 Lex(); 2763 } 2764 } 2765 2766 if (FillSize < 0) { 2767 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 2768 NumValues = 0; 2769 } 2770 if (FillSize > 8) { 2771 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 2772 FillSize = 8; 2773 } 2774 2775 if (!isUInt<32>(FillExpr) && FillSize > 4) 2776 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 2777 2778 if (NumValues > 0) { 2779 int64_t NonZeroFillSize = FillSize > 4 ? 4 : FillSize; 2780 FillExpr &= ~0ULL >> (64 - NonZeroFillSize * 8); 2781 for (uint64_t i = 0, e = NumValues; i != e; ++i) { 2782 getStreamer().EmitIntValue(FillExpr, NonZeroFillSize); 2783 if (NonZeroFillSize < FillSize) 2784 getStreamer().EmitIntValue(0, FillSize - NonZeroFillSize); 2785 } 2786 } 2787 2788 return false; 2789 } 2790 2791 /// parseDirectiveOrg 2792 /// ::= .org expression [ , expression ] 2793 bool AsmParser::parseDirectiveOrg() { 2794 checkForValidSection(); 2795 2796 const MCExpr *Offset; 2797 if (parseExpression(Offset)) 2798 return true; 2799 2800 // Parse optional fill expression. 2801 int64_t FillExpr = 0; 2802 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2803 if (getLexer().isNot(AsmToken::Comma)) 2804 return TokError("unexpected token in '.org' directive"); 2805 Lex(); 2806 2807 if (parseAbsoluteExpression(FillExpr)) 2808 return true; 2809 2810 if (getLexer().isNot(AsmToken::EndOfStatement)) 2811 return TokError("unexpected token in '.org' directive"); 2812 } 2813 2814 Lex(); 2815 getStreamer().emitValueToOffset(Offset, FillExpr); 2816 return false; 2817 } 2818 2819 /// parseDirectiveAlign 2820 /// ::= {.align, ...} expression [ , expression [ , expression ]] 2821 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 2822 checkForValidSection(); 2823 2824 SMLoc AlignmentLoc = getLexer().getLoc(); 2825 int64_t Alignment; 2826 if (parseAbsoluteExpression(Alignment)) 2827 return true; 2828 2829 SMLoc MaxBytesLoc; 2830 bool HasFillExpr = false; 2831 int64_t FillExpr = 0; 2832 int64_t MaxBytesToFill = 0; 2833 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2834 if (getLexer().isNot(AsmToken::Comma)) 2835 return TokError("unexpected token in directive"); 2836 Lex(); 2837 2838 // The fill expression can be omitted while specifying a maximum number of 2839 // alignment bytes, e.g: 2840 // .align 3,,4 2841 if (getLexer().isNot(AsmToken::Comma)) { 2842 HasFillExpr = true; 2843 if (parseAbsoluteExpression(FillExpr)) 2844 return true; 2845 } 2846 2847 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2848 if (getLexer().isNot(AsmToken::Comma)) 2849 return TokError("unexpected token in directive"); 2850 Lex(); 2851 2852 MaxBytesLoc = getLexer().getLoc(); 2853 if (parseAbsoluteExpression(MaxBytesToFill)) 2854 return true; 2855 2856 if (getLexer().isNot(AsmToken::EndOfStatement)) 2857 return TokError("unexpected token in directive"); 2858 } 2859 } 2860 2861 Lex(); 2862 2863 if (!HasFillExpr) 2864 FillExpr = 0; 2865 2866 // Compute alignment in bytes. 2867 if (IsPow2) { 2868 // FIXME: Diagnose overflow. 2869 if (Alignment >= 32) { 2870 Error(AlignmentLoc, "invalid alignment value"); 2871 Alignment = 31; 2872 } 2873 2874 Alignment = 1ULL << Alignment; 2875 } else { 2876 // Reject alignments that aren't either a power of two or zero, 2877 // for gas compatibility. Alignment of zero is silently rounded 2878 // up to one. 2879 if (Alignment == 0) 2880 Alignment = 1; 2881 if (!isPowerOf2_64(Alignment)) 2882 Error(AlignmentLoc, "alignment must be a power of 2"); 2883 } 2884 2885 // Diagnose non-sensical max bytes to align. 2886 if (MaxBytesLoc.isValid()) { 2887 if (MaxBytesToFill < 1) { 2888 Error(MaxBytesLoc, "alignment directive can never be satisfied in this " 2889 "many bytes, ignoring maximum bytes expression"); 2890 MaxBytesToFill = 0; 2891 } 2892 2893 if (MaxBytesToFill >= Alignment) { 2894 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 2895 "has no effect"); 2896 MaxBytesToFill = 0; 2897 } 2898 } 2899 2900 // Check whether we should use optimal code alignment for this .align 2901 // directive. 2902 const MCSection *Section = getStreamer().getCurrentSection().first; 2903 assert(Section && "must have section to emit alignment"); 2904 bool UseCodeAlign = Section->UseCodeAlign(); 2905 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 2906 ValueSize == 1 && UseCodeAlign) { 2907 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 2908 } else { 2909 // FIXME: Target specific behavior about how the "extra" bytes are filled. 2910 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 2911 MaxBytesToFill); 2912 } 2913 2914 return false; 2915 } 2916 2917 /// parseDirectiveFile 2918 /// ::= .file [number] filename 2919 /// ::= .file number directory filename 2920 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 2921 // FIXME: I'm not sure what this is. 2922 int64_t FileNumber = -1; 2923 SMLoc FileNumberLoc = getLexer().getLoc(); 2924 if (getLexer().is(AsmToken::Integer)) { 2925 FileNumber = getTok().getIntVal(); 2926 Lex(); 2927 2928 if (FileNumber < 1) 2929 return TokError("file number less than one"); 2930 } 2931 2932 if (getLexer().isNot(AsmToken::String)) 2933 return TokError("unexpected token in '.file' directive"); 2934 2935 // Usually the directory and filename together, otherwise just the directory. 2936 // Allow the strings to have escaped octal character sequence. 2937 std::string Path = getTok().getString(); 2938 if (parseEscapedString(Path)) 2939 return true; 2940 Lex(); 2941 2942 StringRef Directory; 2943 StringRef Filename; 2944 std::string FilenameData; 2945 if (getLexer().is(AsmToken::String)) { 2946 if (FileNumber == -1) 2947 return TokError("explicit path specified, but no file number"); 2948 if (parseEscapedString(FilenameData)) 2949 return true; 2950 Filename = FilenameData; 2951 Directory = Path; 2952 Lex(); 2953 } else { 2954 Filename = Path; 2955 } 2956 2957 if (getLexer().isNot(AsmToken::EndOfStatement)) 2958 return TokError("unexpected token in '.file' directive"); 2959 2960 if (FileNumber == -1) 2961 getStreamer().EmitFileDirective(Filename); 2962 else { 2963 if (getContext().getGenDwarfForAssembly()) 2964 Error(DirectiveLoc, 2965 "input can't have .file dwarf directives when -g is " 2966 "used to generate dwarf debug info for assembly code"); 2967 2968 if (getStreamer().EmitDwarfFileDirective(FileNumber, Directory, Filename) == 2969 0) 2970 Error(FileNumberLoc, "file number already allocated"); 2971 } 2972 2973 return false; 2974 } 2975 2976 /// parseDirectiveLine 2977 /// ::= .line [number] 2978 bool AsmParser::parseDirectiveLine() { 2979 if (getLexer().isNot(AsmToken::EndOfStatement)) { 2980 if (getLexer().isNot(AsmToken::Integer)) 2981 return TokError("unexpected token in '.line' directive"); 2982 2983 int64_t LineNumber = getTok().getIntVal(); 2984 (void)LineNumber; 2985 Lex(); 2986 2987 // FIXME: Do something with the .line. 2988 } 2989 2990 if (getLexer().isNot(AsmToken::EndOfStatement)) 2991 return TokError("unexpected token in '.line' directive"); 2992 2993 return false; 2994 } 2995 2996 /// parseDirectiveLoc 2997 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 2998 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 2999 /// The first number is a file number, must have been previously assigned with 3000 /// a .file directive, the second number is the line number and optionally the 3001 /// third number is a column position (zero if not specified). The remaining 3002 /// optional items are .loc sub-directives. 3003 bool AsmParser::parseDirectiveLoc() { 3004 if (getLexer().isNot(AsmToken::Integer)) 3005 return TokError("unexpected token in '.loc' directive"); 3006 int64_t FileNumber = getTok().getIntVal(); 3007 if (FileNumber < 1) 3008 return TokError("file number less than one in '.loc' directive"); 3009 if (!getContext().isValidDwarfFileNumber(FileNumber)) 3010 return TokError("unassigned file number in '.loc' directive"); 3011 Lex(); 3012 3013 int64_t LineNumber = 0; 3014 if (getLexer().is(AsmToken::Integer)) { 3015 LineNumber = getTok().getIntVal(); 3016 if (LineNumber < 0) 3017 return TokError("line number less than zero in '.loc' directive"); 3018 Lex(); 3019 } 3020 3021 int64_t ColumnPos = 0; 3022 if (getLexer().is(AsmToken::Integer)) { 3023 ColumnPos = getTok().getIntVal(); 3024 if (ColumnPos < 0) 3025 return TokError("column position less than zero in '.loc' directive"); 3026 Lex(); 3027 } 3028 3029 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3030 unsigned Isa = 0; 3031 int64_t Discriminator = 0; 3032 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3033 for (;;) { 3034 if (getLexer().is(AsmToken::EndOfStatement)) 3035 break; 3036 3037 StringRef Name; 3038 SMLoc Loc = getTok().getLoc(); 3039 if (parseIdentifier(Name)) 3040 return TokError("unexpected token in '.loc' directive"); 3041 3042 if (Name == "basic_block") 3043 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3044 else if (Name == "prologue_end") 3045 Flags |= DWARF2_FLAG_PROLOGUE_END; 3046 else if (Name == "epilogue_begin") 3047 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3048 else if (Name == "is_stmt") { 3049 Loc = getTok().getLoc(); 3050 const MCExpr *Value; 3051 if (parseExpression(Value)) 3052 return true; 3053 // The expression must be the constant 0 or 1. 3054 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3055 int Value = MCE->getValue(); 3056 if (Value == 0) 3057 Flags &= ~DWARF2_FLAG_IS_STMT; 3058 else if (Value == 1) 3059 Flags |= DWARF2_FLAG_IS_STMT; 3060 else 3061 return Error(Loc, "is_stmt value not 0 or 1"); 3062 } else { 3063 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3064 } 3065 } else if (Name == "isa") { 3066 Loc = getTok().getLoc(); 3067 const MCExpr *Value; 3068 if (parseExpression(Value)) 3069 return true; 3070 // The expression must be a constant greater or equal to 0. 3071 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3072 int Value = MCE->getValue(); 3073 if (Value < 0) 3074 return Error(Loc, "isa number less than zero"); 3075 Isa = Value; 3076 } else { 3077 return Error(Loc, "isa number not a constant value"); 3078 } 3079 } else if (Name == "discriminator") { 3080 if (parseAbsoluteExpression(Discriminator)) 3081 return true; 3082 } else { 3083 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3084 } 3085 3086 if (getLexer().is(AsmToken::EndOfStatement)) 3087 break; 3088 } 3089 } 3090 3091 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3092 Isa, Discriminator, StringRef()); 3093 3094 return false; 3095 } 3096 3097 /// parseDirectiveStabs 3098 /// ::= .stabs string, number, number, number 3099 bool AsmParser::parseDirectiveStabs() { 3100 return TokError("unsupported directive '.stabs'"); 3101 } 3102 3103 /// parseDirectiveCVFile 3104 /// ::= .cv_file number filename 3105 bool AsmParser::parseDirectiveCVFile() { 3106 SMLoc FileNumberLoc = getLexer().getLoc(); 3107 if (getLexer().isNot(AsmToken::Integer)) 3108 return TokError("expected file number in '.cv_file' directive"); 3109 3110 int64_t FileNumber = getTok().getIntVal(); 3111 Lex(); 3112 3113 if (FileNumber < 1) 3114 return TokError("file number less than one"); 3115 3116 if (getLexer().isNot(AsmToken::String)) 3117 return TokError("unexpected token in '.cv_file' directive"); 3118 3119 // Usually the directory and filename together, otherwise just the directory. 3120 // Allow the strings to have escaped octal character sequence. 3121 std::string Filename; 3122 if (parseEscapedString(Filename)) 3123 return true; 3124 Lex(); 3125 3126 if (getLexer().isNot(AsmToken::EndOfStatement)) 3127 return TokError("unexpected token in '.cv_file' directive"); 3128 3129 if (getStreamer().EmitCVFileDirective(FileNumber, Filename) == 0) 3130 Error(FileNumberLoc, "file number already allocated"); 3131 3132 return false; 3133 } 3134 3135 /// parseDirectiveCVLoc 3136 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3137 /// [is_stmt VALUE] 3138 /// The first number is a file number, must have been previously assigned with 3139 /// a .file directive, the second number is the line number and optionally the 3140 /// third number is a column position (zero if not specified). The remaining 3141 /// optional items are .loc sub-directives. 3142 bool AsmParser::parseDirectiveCVLoc() { 3143 if (getLexer().isNot(AsmToken::Integer)) 3144 return TokError("unexpected token in '.cv_loc' directive"); 3145 3146 int64_t FunctionId = getTok().getIntVal(); 3147 if (FunctionId < 0) 3148 return TokError("function id less than zero in '.cv_loc' directive"); 3149 Lex(); 3150 3151 int64_t FileNumber = getTok().getIntVal(); 3152 if (FileNumber < 1) 3153 return TokError("file number less than one in '.cv_loc' directive"); 3154 if (!getContext().isValidCVFileNumber(FileNumber)) 3155 return TokError("unassigned file number in '.cv_loc' directive"); 3156 Lex(); 3157 3158 int64_t LineNumber = 0; 3159 if (getLexer().is(AsmToken::Integer)) { 3160 LineNumber = getTok().getIntVal(); 3161 if (LineNumber < 0) 3162 return TokError("line number less than zero in '.cv_loc' directive"); 3163 Lex(); 3164 } 3165 3166 int64_t ColumnPos = 0; 3167 if (getLexer().is(AsmToken::Integer)) { 3168 ColumnPos = getTok().getIntVal(); 3169 if (ColumnPos < 0) 3170 return TokError("column position less than zero in '.cv_loc' directive"); 3171 Lex(); 3172 } 3173 3174 bool PrologueEnd = false; 3175 uint64_t IsStmt = 0; 3176 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3177 StringRef Name; 3178 SMLoc Loc = getTok().getLoc(); 3179 if (parseIdentifier(Name)) 3180 return TokError("unexpected token in '.cv_loc' directive"); 3181 3182 if (Name == "prologue_end") 3183 PrologueEnd = true; 3184 else if (Name == "is_stmt") { 3185 Loc = getTok().getLoc(); 3186 const MCExpr *Value; 3187 if (parseExpression(Value)) 3188 return true; 3189 // The expression must be the constant 0 or 1. 3190 IsStmt = ~0ULL; 3191 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3192 IsStmt = MCE->getValue(); 3193 3194 if (IsStmt > 1) 3195 return Error(Loc, "is_stmt value not 0 or 1"); 3196 } else { 3197 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3198 } 3199 } 3200 3201 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3202 ColumnPos, PrologueEnd, IsStmt, StringRef()); 3203 return false; 3204 } 3205 3206 /// parseDirectiveCVLinetable 3207 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3208 bool AsmParser::parseDirectiveCVLinetable() { 3209 int64_t FunctionId = getTok().getIntVal(); 3210 if (FunctionId < 0) 3211 return TokError("function id less than zero in '.cv_linetable' directive"); 3212 Lex(); 3213 3214 if (Lexer.isNot(AsmToken::Comma)) 3215 return TokError("unexpected token in '.cv_linetable' directive"); 3216 Lex(); 3217 3218 SMLoc Loc = getLexer().getLoc(); 3219 StringRef FnStartName; 3220 if (parseIdentifier(FnStartName)) 3221 return Error(Loc, "expected identifier in directive"); 3222 3223 if (Lexer.isNot(AsmToken::Comma)) 3224 return TokError("unexpected token in '.cv_linetable' directive"); 3225 Lex(); 3226 3227 Loc = getLexer().getLoc(); 3228 StringRef FnEndName; 3229 if (parseIdentifier(FnEndName)) 3230 return Error(Loc, "expected identifier in directive"); 3231 3232 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3233 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3234 3235 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3236 return false; 3237 } 3238 3239 /// parseDirectiveCVInlineLinetable 3240 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3241 /// ("contains" SecondaryFunctionId+)? 3242 bool AsmParser::parseDirectiveCVInlineLinetable() { 3243 int64_t PrimaryFunctionId = getTok().getIntVal(); 3244 if (PrimaryFunctionId < 0) 3245 return TokError( 3246 "function id less than zero in '.cv_inline_linetable' directive"); 3247 Lex(); 3248 3249 int64_t SourceFileId = getTok().getIntVal(); 3250 if (SourceFileId <= 0) 3251 return TokError( 3252 "File id less than zero in '.cv_inline_linetable' directive"); 3253 Lex(); 3254 3255 int64_t SourceLineNum = getTok().getIntVal(); 3256 if (SourceLineNum < 0) 3257 return TokError( 3258 "Line number less than zero in '.cv_inline_linetable' directive"); 3259 Lex(); 3260 3261 SMLoc Loc = getLexer().getLoc(); 3262 StringRef FnStartName; 3263 if (parseIdentifier(FnStartName)) 3264 return Error(Loc, "expected identifier in directive"); 3265 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3266 3267 Loc = getLexer().getLoc(); 3268 StringRef FnEndName; 3269 if (parseIdentifier(FnEndName)) 3270 return Error(Loc, "expected identifier in directive"); 3271 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3272 3273 SmallVector<unsigned, 8> SecondaryFunctionIds; 3274 if (getLexer().is(AsmToken::Identifier)) { 3275 if (getTok().getIdentifier() != "contains") 3276 return TokError( 3277 "unexpected identifier in '.cv_inline_linetable' directive"); 3278 Lex(); 3279 3280 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3281 int64_t SecondaryFunctionId = getTok().getIntVal(); 3282 if (SecondaryFunctionId < 0) 3283 return TokError( 3284 "function id less than zero in '.cv_inline_linetable' directive"); 3285 Lex(); 3286 3287 SecondaryFunctionIds.push_back(SecondaryFunctionId); 3288 } 3289 } 3290 3291 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3292 SourceLineNum, FnStartSym, 3293 FnEndSym, SecondaryFunctionIds); 3294 return false; 3295 } 3296 3297 /// parseDirectiveCVDefRange 3298 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3299 bool AsmParser::parseDirectiveCVDefRange() { 3300 SMLoc Loc; 3301 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3302 while (getLexer().is(AsmToken::Identifier)) { 3303 Loc = getLexer().getLoc(); 3304 StringRef GapStartName; 3305 if (parseIdentifier(GapStartName)) 3306 return Error(Loc, "expected identifier in directive"); 3307 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3308 3309 Loc = getLexer().getLoc(); 3310 StringRef GapEndName; 3311 if (parseIdentifier(GapEndName)) 3312 return Error(Loc, "expected identifier in directive"); 3313 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3314 3315 Ranges.push_back({GapStartSym, GapEndSym}); 3316 } 3317 3318 if (getLexer().isNot(AsmToken::Comma)) 3319 return TokError("unexpected token in directive"); 3320 Lex(); 3321 3322 std::string FixedSizePortion; 3323 if (parseEscapedString(FixedSizePortion)) 3324 return true; 3325 Lex(); 3326 3327 getStreamer().EmitCVDefRangeDirective(Ranges, FixedSizePortion); 3328 return false; 3329 } 3330 3331 /// parseDirectiveCVStringTable 3332 /// ::= .cv_stringtable 3333 bool AsmParser::parseDirectiveCVStringTable() { 3334 getStreamer().EmitCVStringTableDirective(); 3335 return false; 3336 } 3337 3338 /// parseDirectiveCVFileChecksums 3339 /// ::= .cv_filechecksums 3340 bool AsmParser::parseDirectiveCVFileChecksums() { 3341 getStreamer().EmitCVFileChecksumsDirective(); 3342 return false; 3343 } 3344 3345 /// parseDirectiveCFISections 3346 /// ::= .cfi_sections section [, section] 3347 bool AsmParser::parseDirectiveCFISections() { 3348 StringRef Name; 3349 bool EH = false; 3350 bool Debug = false; 3351 3352 if (parseIdentifier(Name)) 3353 return TokError("Expected an identifier"); 3354 3355 if (Name == ".eh_frame") 3356 EH = true; 3357 else if (Name == ".debug_frame") 3358 Debug = true; 3359 3360 if (getLexer().is(AsmToken::Comma)) { 3361 Lex(); 3362 3363 if (parseIdentifier(Name)) 3364 return TokError("Expected an identifier"); 3365 3366 if (Name == ".eh_frame") 3367 EH = true; 3368 else if (Name == ".debug_frame") 3369 Debug = true; 3370 } 3371 3372 getStreamer().EmitCFISections(EH, Debug); 3373 return false; 3374 } 3375 3376 /// parseDirectiveCFIStartProc 3377 /// ::= .cfi_startproc [simple] 3378 bool AsmParser::parseDirectiveCFIStartProc() { 3379 StringRef Simple; 3380 if (getLexer().isNot(AsmToken::EndOfStatement)) 3381 if (parseIdentifier(Simple) || Simple != "simple") 3382 return TokError("unexpected token in .cfi_startproc directive"); 3383 3384 getStreamer().EmitCFIStartProc(!Simple.empty()); 3385 return false; 3386 } 3387 3388 /// parseDirectiveCFIEndProc 3389 /// ::= .cfi_endproc 3390 bool AsmParser::parseDirectiveCFIEndProc() { 3391 getStreamer().EmitCFIEndProc(); 3392 return false; 3393 } 3394 3395 /// \brief parse register name or number. 3396 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 3397 SMLoc DirectiveLoc) { 3398 unsigned RegNo; 3399 3400 if (getLexer().isNot(AsmToken::Integer)) { 3401 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 3402 return true; 3403 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 3404 } else 3405 return parseAbsoluteExpression(Register); 3406 3407 return false; 3408 } 3409 3410 /// parseDirectiveCFIDefCfa 3411 /// ::= .cfi_def_cfa register, offset 3412 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3413 int64_t Register = 0; 3414 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3415 return true; 3416 3417 if (getLexer().isNot(AsmToken::Comma)) 3418 return TokError("unexpected token in directive"); 3419 Lex(); 3420 3421 int64_t Offset = 0; 3422 if (parseAbsoluteExpression(Offset)) 3423 return true; 3424 3425 getStreamer().EmitCFIDefCfa(Register, Offset); 3426 return false; 3427 } 3428 3429 /// parseDirectiveCFIDefCfaOffset 3430 /// ::= .cfi_def_cfa_offset offset 3431 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3432 int64_t Offset = 0; 3433 if (parseAbsoluteExpression(Offset)) 3434 return true; 3435 3436 getStreamer().EmitCFIDefCfaOffset(Offset); 3437 return false; 3438 } 3439 3440 /// parseDirectiveCFIRegister 3441 /// ::= .cfi_register register, register 3442 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3443 int64_t Register1 = 0; 3444 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc)) 3445 return true; 3446 3447 if (getLexer().isNot(AsmToken::Comma)) 3448 return TokError("unexpected token in directive"); 3449 Lex(); 3450 3451 int64_t Register2 = 0; 3452 if (parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3453 return true; 3454 3455 getStreamer().EmitCFIRegister(Register1, Register2); 3456 return false; 3457 } 3458 3459 /// parseDirectiveCFIWindowSave 3460 /// ::= .cfi_window_save 3461 bool AsmParser::parseDirectiveCFIWindowSave() { 3462 getStreamer().EmitCFIWindowSave(); 3463 return false; 3464 } 3465 3466 /// parseDirectiveCFIAdjustCfaOffset 3467 /// ::= .cfi_adjust_cfa_offset adjustment 3468 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 3469 int64_t Adjustment = 0; 3470 if (parseAbsoluteExpression(Adjustment)) 3471 return true; 3472 3473 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 3474 return false; 3475 } 3476 3477 /// parseDirectiveCFIDefCfaRegister 3478 /// ::= .cfi_def_cfa_register register 3479 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3480 int64_t Register = 0; 3481 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3482 return true; 3483 3484 getStreamer().EmitCFIDefCfaRegister(Register); 3485 return false; 3486 } 3487 3488 /// parseDirectiveCFIOffset 3489 /// ::= .cfi_offset register, offset 3490 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3491 int64_t Register = 0; 3492 int64_t Offset = 0; 3493 3494 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3495 return true; 3496 3497 if (getLexer().isNot(AsmToken::Comma)) 3498 return TokError("unexpected token in directive"); 3499 Lex(); 3500 3501 if (parseAbsoluteExpression(Offset)) 3502 return true; 3503 3504 getStreamer().EmitCFIOffset(Register, Offset); 3505 return false; 3506 } 3507 3508 /// parseDirectiveCFIRelOffset 3509 /// ::= .cfi_rel_offset register, offset 3510 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3511 int64_t Register = 0; 3512 3513 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3514 return true; 3515 3516 if (getLexer().isNot(AsmToken::Comma)) 3517 return TokError("unexpected token in directive"); 3518 Lex(); 3519 3520 int64_t Offset = 0; 3521 if (parseAbsoluteExpression(Offset)) 3522 return true; 3523 3524 getStreamer().EmitCFIRelOffset(Register, Offset); 3525 return false; 3526 } 3527 3528 static bool isValidEncoding(int64_t Encoding) { 3529 if (Encoding & ~0xff) 3530 return false; 3531 3532 if (Encoding == dwarf::DW_EH_PE_omit) 3533 return true; 3534 3535 const unsigned Format = Encoding & 0xf; 3536 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3537 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3538 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3539 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3540 return false; 3541 3542 const unsigned Application = Encoding & 0x70; 3543 if (Application != dwarf::DW_EH_PE_absptr && 3544 Application != dwarf::DW_EH_PE_pcrel) 3545 return false; 3546 3547 return true; 3548 } 3549 3550 /// parseDirectiveCFIPersonalityOrLsda 3551 /// IsPersonality true for cfi_personality, false for cfi_lsda 3552 /// ::= .cfi_personality encoding, [symbol_name] 3553 /// ::= .cfi_lsda encoding, [symbol_name] 3554 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3555 int64_t Encoding = 0; 3556 if (parseAbsoluteExpression(Encoding)) 3557 return true; 3558 if (Encoding == dwarf::DW_EH_PE_omit) 3559 return false; 3560 3561 if (!isValidEncoding(Encoding)) 3562 return TokError("unsupported encoding."); 3563 3564 if (getLexer().isNot(AsmToken::Comma)) 3565 return TokError("unexpected token in directive"); 3566 Lex(); 3567 3568 StringRef Name; 3569 if (parseIdentifier(Name)) 3570 return TokError("expected identifier in directive"); 3571 3572 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 3573 3574 if (IsPersonality) 3575 getStreamer().EmitCFIPersonality(Sym, Encoding); 3576 else 3577 getStreamer().EmitCFILsda(Sym, Encoding); 3578 return false; 3579 } 3580 3581 /// parseDirectiveCFIRememberState 3582 /// ::= .cfi_remember_state 3583 bool AsmParser::parseDirectiveCFIRememberState() { 3584 getStreamer().EmitCFIRememberState(); 3585 return false; 3586 } 3587 3588 /// parseDirectiveCFIRestoreState 3589 /// ::= .cfi_remember_state 3590 bool AsmParser::parseDirectiveCFIRestoreState() { 3591 getStreamer().EmitCFIRestoreState(); 3592 return false; 3593 } 3594 3595 /// parseDirectiveCFISameValue 3596 /// ::= .cfi_same_value register 3597 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3598 int64_t Register = 0; 3599 3600 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3601 return true; 3602 3603 getStreamer().EmitCFISameValue(Register); 3604 return false; 3605 } 3606 3607 /// parseDirectiveCFIRestore 3608 /// ::= .cfi_restore register 3609 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3610 int64_t Register = 0; 3611 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3612 return true; 3613 3614 getStreamer().EmitCFIRestore(Register); 3615 return false; 3616 } 3617 3618 /// parseDirectiveCFIEscape 3619 /// ::= .cfi_escape expression[,...] 3620 bool AsmParser::parseDirectiveCFIEscape() { 3621 std::string Values; 3622 int64_t CurrValue; 3623 if (parseAbsoluteExpression(CurrValue)) 3624 return true; 3625 3626 Values.push_back((uint8_t)CurrValue); 3627 3628 while (getLexer().is(AsmToken::Comma)) { 3629 Lex(); 3630 3631 if (parseAbsoluteExpression(CurrValue)) 3632 return true; 3633 3634 Values.push_back((uint8_t)CurrValue); 3635 } 3636 3637 getStreamer().EmitCFIEscape(Values); 3638 return false; 3639 } 3640 3641 /// parseDirectiveCFISignalFrame 3642 /// ::= .cfi_signal_frame 3643 bool AsmParser::parseDirectiveCFISignalFrame() { 3644 if (getLexer().isNot(AsmToken::EndOfStatement)) 3645 return Error(getLexer().getLoc(), 3646 "unexpected token in '.cfi_signal_frame'"); 3647 3648 getStreamer().EmitCFISignalFrame(); 3649 return false; 3650 } 3651 3652 /// parseDirectiveCFIUndefined 3653 /// ::= .cfi_undefined register 3654 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 3655 int64_t Register = 0; 3656 3657 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3658 return true; 3659 3660 getStreamer().EmitCFIUndefined(Register); 3661 return false; 3662 } 3663 3664 /// parseDirectiveMacrosOnOff 3665 /// ::= .macros_on 3666 /// ::= .macros_off 3667 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 3668 if (getLexer().isNot(AsmToken::EndOfStatement)) 3669 return Error(getLexer().getLoc(), 3670 "unexpected token in '" + Directive + "' directive"); 3671 3672 setMacrosEnabled(Directive == ".macros_on"); 3673 return false; 3674 } 3675 3676 /// parseDirectiveMacro 3677 /// ::= .macro name[,] [parameters] 3678 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 3679 StringRef Name; 3680 if (parseIdentifier(Name)) 3681 return TokError("expected identifier in '.macro' directive"); 3682 3683 if (getLexer().is(AsmToken::Comma)) 3684 Lex(); 3685 3686 MCAsmMacroParameters Parameters; 3687 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3688 3689 if (!Parameters.empty() && Parameters.back().Vararg) 3690 return Error(Lexer.getLoc(), 3691 "Vararg parameter '" + Parameters.back().Name + 3692 "' should be last one in the list of parameters."); 3693 3694 MCAsmMacroParameter Parameter; 3695 if (parseIdentifier(Parameter.Name)) 3696 return TokError("expected identifier in '.macro' directive"); 3697 3698 if (Lexer.is(AsmToken::Colon)) { 3699 Lex(); // consume ':' 3700 3701 SMLoc QualLoc; 3702 StringRef Qualifier; 3703 3704 QualLoc = Lexer.getLoc(); 3705 if (parseIdentifier(Qualifier)) 3706 return Error(QualLoc, "missing parameter qualifier for " 3707 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3708 3709 if (Qualifier == "req") 3710 Parameter.Required = true; 3711 else if (Qualifier == "vararg") 3712 Parameter.Vararg = true; 3713 else 3714 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 3715 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 3716 } 3717 3718 if (getLexer().is(AsmToken::Equal)) { 3719 Lex(); 3720 3721 SMLoc ParamLoc; 3722 3723 ParamLoc = Lexer.getLoc(); 3724 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 3725 return true; 3726 3727 if (Parameter.Required) 3728 Warning(ParamLoc, "pointless default value for required parameter " 3729 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3730 } 3731 3732 Parameters.push_back(std::move(Parameter)); 3733 3734 if (getLexer().is(AsmToken::Comma)) 3735 Lex(); 3736 } 3737 3738 // Eat the end of statement. 3739 Lex(); 3740 3741 AsmToken EndToken, StartToken = getTok(); 3742 unsigned MacroDepth = 0; 3743 3744 // Lex the macro definition. 3745 for (;;) { 3746 // Check whether we have reached the end of the file. 3747 if (getLexer().is(AsmToken::Eof)) 3748 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 3749 3750 // Otherwise, check whether we have reach the .endmacro. 3751 if (getLexer().is(AsmToken::Identifier)) { 3752 if (getTok().getIdentifier() == ".endm" || 3753 getTok().getIdentifier() == ".endmacro") { 3754 if (MacroDepth == 0) { // Outermost macro. 3755 EndToken = getTok(); 3756 Lex(); 3757 if (getLexer().isNot(AsmToken::EndOfStatement)) 3758 return TokError("unexpected token in '" + EndToken.getIdentifier() + 3759 "' directive"); 3760 break; 3761 } else { 3762 // Otherwise we just found the end of an inner macro. 3763 --MacroDepth; 3764 } 3765 } else if (getTok().getIdentifier() == ".macro") { 3766 // We allow nested macros. Those aren't instantiated until the outermost 3767 // macro is expanded so just ignore them for now. 3768 ++MacroDepth; 3769 } 3770 } 3771 3772 // Otherwise, scan til the end of the statement. 3773 eatToEndOfStatement(); 3774 } 3775 3776 if (lookupMacro(Name)) { 3777 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 3778 } 3779 3780 const char *BodyStart = StartToken.getLoc().getPointer(); 3781 const char *BodyEnd = EndToken.getLoc().getPointer(); 3782 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 3783 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 3784 defineMacro(Name, MCAsmMacro(Name, Body, std::move(Parameters))); 3785 return false; 3786 } 3787 3788 /// checkForBadMacro 3789 /// 3790 /// With the support added for named parameters there may be code out there that 3791 /// is transitioning from positional parameters. In versions of gas that did 3792 /// not support named parameters they would be ignored on the macro definition. 3793 /// But to support both styles of parameters this is not possible so if a macro 3794 /// definition has named parameters but does not use them and has what appears 3795 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 3796 /// warning that the positional parameter found in body which have no effect. 3797 /// Hoping the developer will either remove the named parameters from the macro 3798 /// definition so the positional parameters get used if that was what was 3799 /// intended or change the macro to use the named parameters. It is possible 3800 /// this warning will trigger when the none of the named parameters are used 3801 /// and the strings like $1 are infact to simply to be passed trough unchanged. 3802 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 3803 StringRef Body, 3804 ArrayRef<MCAsmMacroParameter> Parameters) { 3805 // If this macro is not defined with named parameters the warning we are 3806 // checking for here doesn't apply. 3807 unsigned NParameters = Parameters.size(); 3808 if (NParameters == 0) 3809 return; 3810 3811 bool NamedParametersFound = false; 3812 bool PositionalParametersFound = false; 3813 3814 // Look at the body of the macro for use of both the named parameters and what 3815 // are likely to be positional parameters. This is what expandMacro() is 3816 // doing when it finds the parameters in the body. 3817 while (!Body.empty()) { 3818 // Scan for the next possible parameter. 3819 std::size_t End = Body.size(), Pos = 0; 3820 for (; Pos != End; ++Pos) { 3821 // Check for a substitution or escape. 3822 // This macro is defined with parameters, look for \foo, \bar, etc. 3823 if (Body[Pos] == '\\' && Pos + 1 != End) 3824 break; 3825 3826 // This macro should have parameters, but look for $0, $1, ..., $n too. 3827 if (Body[Pos] != '$' || Pos + 1 == End) 3828 continue; 3829 char Next = Body[Pos + 1]; 3830 if (Next == '$' || Next == 'n' || 3831 isdigit(static_cast<unsigned char>(Next))) 3832 break; 3833 } 3834 3835 // Check if we reached the end. 3836 if (Pos == End) 3837 break; 3838 3839 if (Body[Pos] == '$') { 3840 switch (Body[Pos + 1]) { 3841 // $$ => $ 3842 case '$': 3843 break; 3844 3845 // $n => number of arguments 3846 case 'n': 3847 PositionalParametersFound = true; 3848 break; 3849 3850 // $[0-9] => argument 3851 default: { 3852 PositionalParametersFound = true; 3853 break; 3854 } 3855 } 3856 Pos += 2; 3857 } else { 3858 unsigned I = Pos + 1; 3859 while (isIdentifierChar(Body[I]) && I + 1 != End) 3860 ++I; 3861 3862 const char *Begin = Body.data() + Pos + 1; 3863 StringRef Argument(Begin, I - (Pos + 1)); 3864 unsigned Index = 0; 3865 for (; Index < NParameters; ++Index) 3866 if (Parameters[Index].Name == Argument) 3867 break; 3868 3869 if (Index == NParameters) { 3870 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 3871 Pos += 3; 3872 else { 3873 Pos = I; 3874 } 3875 } else { 3876 NamedParametersFound = true; 3877 Pos += 1 + Argument.size(); 3878 } 3879 } 3880 // Update the scan point. 3881 Body = Body.substr(Pos); 3882 } 3883 3884 if (!NamedParametersFound && PositionalParametersFound) 3885 Warning(DirectiveLoc, "macro defined with named parameters which are not " 3886 "used in macro body, possible positional parameter " 3887 "found in body which will have no effect"); 3888 } 3889 3890 /// parseDirectiveExitMacro 3891 /// ::= .exitm 3892 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 3893 if (getLexer().isNot(AsmToken::EndOfStatement)) 3894 return TokError("unexpected token in '" + Directive + "' directive"); 3895 3896 if (!isInsideMacroInstantiation()) 3897 return TokError("unexpected '" + Directive + "' in file, " 3898 "no current macro definition"); 3899 3900 // Exit all conditionals that are active in the current macro. 3901 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 3902 TheCondState = TheCondStack.back(); 3903 TheCondStack.pop_back(); 3904 } 3905 3906 handleMacroExit(); 3907 return false; 3908 } 3909 3910 /// parseDirectiveEndMacro 3911 /// ::= .endm 3912 /// ::= .endmacro 3913 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 3914 if (getLexer().isNot(AsmToken::EndOfStatement)) 3915 return TokError("unexpected token in '" + Directive + "' directive"); 3916 3917 // If we are inside a macro instantiation, terminate the current 3918 // instantiation. 3919 if (isInsideMacroInstantiation()) { 3920 handleMacroExit(); 3921 return false; 3922 } 3923 3924 // Otherwise, this .endmacro is a stray entry in the file; well formed 3925 // .endmacro directives are handled during the macro definition parsing. 3926 return TokError("unexpected '" + Directive + "' in file, " 3927 "no current macro definition"); 3928 } 3929 3930 /// parseDirectivePurgeMacro 3931 /// ::= .purgem 3932 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 3933 StringRef Name; 3934 if (parseIdentifier(Name)) 3935 return TokError("expected identifier in '.purgem' directive"); 3936 3937 if (getLexer().isNot(AsmToken::EndOfStatement)) 3938 return TokError("unexpected token in '.purgem' directive"); 3939 3940 if (!lookupMacro(Name)) 3941 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 3942 3943 undefineMacro(Name); 3944 return false; 3945 } 3946 3947 /// parseDirectiveBundleAlignMode 3948 /// ::= {.bundle_align_mode} expression 3949 bool AsmParser::parseDirectiveBundleAlignMode() { 3950 checkForValidSection(); 3951 3952 // Expect a single argument: an expression that evaluates to a constant 3953 // in the inclusive range 0-30. 3954 SMLoc ExprLoc = getLexer().getLoc(); 3955 int64_t AlignSizePow2; 3956 if (parseAbsoluteExpression(AlignSizePow2)) 3957 return true; 3958 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3959 return TokError("unexpected token after expression in" 3960 " '.bundle_align_mode' directive"); 3961 else if (AlignSizePow2 < 0 || AlignSizePow2 > 30) 3962 return Error(ExprLoc, 3963 "invalid bundle alignment size (expected between 0 and 30)"); 3964 3965 Lex(); 3966 3967 // Because of AlignSizePow2's verified range we can safely truncate it to 3968 // unsigned. 3969 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 3970 return false; 3971 } 3972 3973 /// parseDirectiveBundleLock 3974 /// ::= {.bundle_lock} [align_to_end] 3975 bool AsmParser::parseDirectiveBundleLock() { 3976 checkForValidSection(); 3977 bool AlignToEnd = false; 3978 3979 if (getLexer().isNot(AsmToken::EndOfStatement)) { 3980 StringRef Option; 3981 SMLoc Loc = getTok().getLoc(); 3982 const char *kInvalidOptionError = 3983 "invalid option for '.bundle_lock' directive"; 3984 3985 if (parseIdentifier(Option)) 3986 return Error(Loc, kInvalidOptionError); 3987 3988 if (Option != "align_to_end") 3989 return Error(Loc, kInvalidOptionError); 3990 else if (getLexer().isNot(AsmToken::EndOfStatement)) 3991 return Error(Loc, 3992 "unexpected token after '.bundle_lock' directive option"); 3993 AlignToEnd = true; 3994 } 3995 3996 Lex(); 3997 3998 getStreamer().EmitBundleLock(AlignToEnd); 3999 return false; 4000 } 4001 4002 /// parseDirectiveBundleLock 4003 /// ::= {.bundle_lock} 4004 bool AsmParser::parseDirectiveBundleUnlock() { 4005 checkForValidSection(); 4006 4007 if (getLexer().isNot(AsmToken::EndOfStatement)) 4008 return TokError("unexpected token in '.bundle_unlock' directive"); 4009 Lex(); 4010 4011 getStreamer().EmitBundleUnlock(); 4012 return false; 4013 } 4014 4015 /// parseDirectiveSpace 4016 /// ::= (.skip | .space) expression [ , expression ] 4017 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4018 checkForValidSection(); 4019 4020 int64_t NumBytes; 4021 if (parseAbsoluteExpression(NumBytes)) 4022 return true; 4023 4024 int64_t FillExpr = 0; 4025 if (getLexer().isNot(AsmToken::EndOfStatement)) { 4026 if (getLexer().isNot(AsmToken::Comma)) 4027 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 4028 Lex(); 4029 4030 if (parseAbsoluteExpression(FillExpr)) 4031 return true; 4032 4033 if (getLexer().isNot(AsmToken::EndOfStatement)) 4034 return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); 4035 } 4036 4037 Lex(); 4038 4039 if (NumBytes <= 0) 4040 return TokError("invalid number of bytes in '" + Twine(IDVal) + 4041 "' directive"); 4042 4043 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4044 getStreamer().EmitFill(NumBytes, FillExpr); 4045 4046 return false; 4047 } 4048 4049 /// parseDirectiveLEB128 4050 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4051 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4052 checkForValidSection(); 4053 const MCExpr *Value; 4054 4055 for (;;) { 4056 if (parseExpression(Value)) 4057 return true; 4058 4059 if (Signed) 4060 getStreamer().EmitSLEB128Value(Value); 4061 else 4062 getStreamer().EmitULEB128Value(Value); 4063 4064 if (getLexer().is(AsmToken::EndOfStatement)) 4065 break; 4066 4067 if (getLexer().isNot(AsmToken::Comma)) 4068 return TokError("unexpected token in directive"); 4069 Lex(); 4070 } 4071 4072 return false; 4073 } 4074 4075 /// parseDirectiveSymbolAttribute 4076 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4077 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4078 if (getLexer().isNot(AsmToken::EndOfStatement)) { 4079 for (;;) { 4080 StringRef Name; 4081 SMLoc Loc = getTok().getLoc(); 4082 4083 if (parseIdentifier(Name)) 4084 return Error(Loc, "expected identifier in directive"); 4085 4086 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4087 4088 // Assembler local symbols don't make any sense here. Complain loudly. 4089 if (Sym->isTemporary()) 4090 return Error(Loc, "non-local symbol required in directive"); 4091 4092 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 4093 return Error(Loc, "unable to emit symbol attribute"); 4094 4095 if (getLexer().is(AsmToken::EndOfStatement)) 4096 break; 4097 4098 if (getLexer().isNot(AsmToken::Comma)) 4099 return TokError("unexpected token in directive"); 4100 Lex(); 4101 } 4102 } 4103 4104 Lex(); 4105 return false; 4106 } 4107 4108 /// parseDirectiveComm 4109 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4110 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4111 checkForValidSection(); 4112 4113 SMLoc IDLoc = getLexer().getLoc(); 4114 StringRef Name; 4115 if (parseIdentifier(Name)) 4116 return TokError("expected identifier in directive"); 4117 4118 // Handle the identifier as the key symbol. 4119 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4120 4121 if (getLexer().isNot(AsmToken::Comma)) 4122 return TokError("unexpected token in directive"); 4123 Lex(); 4124 4125 int64_t Size; 4126 SMLoc SizeLoc = getLexer().getLoc(); 4127 if (parseAbsoluteExpression(Size)) 4128 return true; 4129 4130 int64_t Pow2Alignment = 0; 4131 SMLoc Pow2AlignmentLoc; 4132 if (getLexer().is(AsmToken::Comma)) { 4133 Lex(); 4134 Pow2AlignmentLoc = getLexer().getLoc(); 4135 if (parseAbsoluteExpression(Pow2Alignment)) 4136 return true; 4137 4138 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4139 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4140 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4141 4142 // If this target takes alignments in bytes (not log) validate and convert. 4143 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4144 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4145 if (!isPowerOf2_64(Pow2Alignment)) 4146 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4147 Pow2Alignment = Log2_64(Pow2Alignment); 4148 } 4149 } 4150 4151 if (getLexer().isNot(AsmToken::EndOfStatement)) 4152 return TokError("unexpected token in '.comm' or '.lcomm' directive"); 4153 4154 Lex(); 4155 4156 // NOTE: a size of zero for a .comm should create a undefined symbol 4157 // but a size of .lcomm creates a bss symbol of size zero. 4158 if (Size < 0) 4159 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4160 "be less than zero"); 4161 4162 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4163 // may internally end up wanting an alignment in bytes. 4164 // FIXME: Diagnose overflow. 4165 if (Pow2Alignment < 0) 4166 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4167 "alignment, can't be less than zero"); 4168 4169 if (!Sym->isUndefined()) 4170 return Error(IDLoc, "invalid symbol redefinition"); 4171 4172 // Create the Symbol as a common or local common with Size and Pow2Alignment 4173 if (IsLocal) { 4174 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4175 return false; 4176 } 4177 4178 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4179 return false; 4180 } 4181 4182 /// parseDirectiveAbort 4183 /// ::= .abort [... message ...] 4184 bool AsmParser::parseDirectiveAbort() { 4185 // FIXME: Use loc from directive. 4186 SMLoc Loc = getLexer().getLoc(); 4187 4188 StringRef Str = parseStringToEndOfStatement(); 4189 if (getLexer().isNot(AsmToken::EndOfStatement)) 4190 return TokError("unexpected token in '.abort' directive"); 4191 4192 Lex(); 4193 4194 if (Str.empty()) 4195 Error(Loc, ".abort detected. Assembly stopping."); 4196 else 4197 Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4198 // FIXME: Actually abort assembly here. 4199 4200 return false; 4201 } 4202 4203 /// parseDirectiveInclude 4204 /// ::= .include "filename" 4205 bool AsmParser::parseDirectiveInclude() { 4206 if (getLexer().isNot(AsmToken::String)) 4207 return TokError("expected string in '.include' directive"); 4208 4209 // Allow the strings to have escaped octal character sequence. 4210 std::string Filename; 4211 if (parseEscapedString(Filename)) 4212 return true; 4213 SMLoc IncludeLoc = getLexer().getLoc(); 4214 Lex(); 4215 4216 if (getLexer().isNot(AsmToken::EndOfStatement)) 4217 return TokError("unexpected token in '.include' directive"); 4218 4219 // Attempt to switch the lexer to the included file before consuming the end 4220 // of statement to avoid losing it when we switch. 4221 if (enterIncludeFile(Filename)) { 4222 Error(IncludeLoc, "Could not find include file '" + Filename + "'"); 4223 return true; 4224 } 4225 4226 return false; 4227 } 4228 4229 /// parseDirectiveIncbin 4230 /// ::= .incbin "filename" 4231 bool AsmParser::parseDirectiveIncbin() { 4232 if (getLexer().isNot(AsmToken::String)) 4233 return TokError("expected string in '.incbin' directive"); 4234 4235 // Allow the strings to have escaped octal character sequence. 4236 std::string Filename; 4237 if (parseEscapedString(Filename)) 4238 return true; 4239 SMLoc IncbinLoc = getLexer().getLoc(); 4240 Lex(); 4241 4242 if (getLexer().isNot(AsmToken::EndOfStatement)) 4243 return TokError("unexpected token in '.incbin' directive"); 4244 4245 // Attempt to process the included file. 4246 if (processIncbinFile(Filename)) { 4247 Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 4248 return true; 4249 } 4250 4251 return false; 4252 } 4253 4254 /// parseDirectiveIf 4255 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 4256 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 4257 TheCondStack.push_back(TheCondState); 4258 TheCondState.TheCond = AsmCond::IfCond; 4259 if (TheCondState.Ignore) { 4260 eatToEndOfStatement(); 4261 } else { 4262 int64_t ExprValue; 4263 if (parseAbsoluteExpression(ExprValue)) 4264 return true; 4265 4266 if (getLexer().isNot(AsmToken::EndOfStatement)) 4267 return TokError("unexpected token in '.if' directive"); 4268 4269 Lex(); 4270 4271 switch (DirKind) { 4272 default: 4273 llvm_unreachable("unsupported directive"); 4274 case DK_IF: 4275 case DK_IFNE: 4276 break; 4277 case DK_IFEQ: 4278 ExprValue = ExprValue == 0; 4279 break; 4280 case DK_IFGE: 4281 ExprValue = ExprValue >= 0; 4282 break; 4283 case DK_IFGT: 4284 ExprValue = ExprValue > 0; 4285 break; 4286 case DK_IFLE: 4287 ExprValue = ExprValue <= 0; 4288 break; 4289 case DK_IFLT: 4290 ExprValue = ExprValue < 0; 4291 break; 4292 } 4293 4294 TheCondState.CondMet = ExprValue; 4295 TheCondState.Ignore = !TheCondState.CondMet; 4296 } 4297 4298 return false; 4299 } 4300 4301 /// parseDirectiveIfb 4302 /// ::= .ifb string 4303 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4304 TheCondStack.push_back(TheCondState); 4305 TheCondState.TheCond = AsmCond::IfCond; 4306 4307 if (TheCondState.Ignore) { 4308 eatToEndOfStatement(); 4309 } else { 4310 StringRef Str = parseStringToEndOfStatement(); 4311 4312 if (getLexer().isNot(AsmToken::EndOfStatement)) 4313 return TokError("unexpected token in '.ifb' directive"); 4314 4315 Lex(); 4316 4317 TheCondState.CondMet = ExpectBlank == Str.empty(); 4318 TheCondState.Ignore = !TheCondState.CondMet; 4319 } 4320 4321 return false; 4322 } 4323 4324 /// parseDirectiveIfc 4325 /// ::= .ifc string1, string2 4326 /// ::= .ifnc string1, string2 4327 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 4328 TheCondStack.push_back(TheCondState); 4329 TheCondState.TheCond = AsmCond::IfCond; 4330 4331 if (TheCondState.Ignore) { 4332 eatToEndOfStatement(); 4333 } else { 4334 StringRef Str1 = parseStringToComma(); 4335 4336 if (getLexer().isNot(AsmToken::Comma)) 4337 return TokError("unexpected token in '.ifc' directive"); 4338 4339 Lex(); 4340 4341 StringRef Str2 = parseStringToEndOfStatement(); 4342 4343 if (getLexer().isNot(AsmToken::EndOfStatement)) 4344 return TokError("unexpected token in '.ifc' directive"); 4345 4346 Lex(); 4347 4348 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 4349 TheCondState.Ignore = !TheCondState.CondMet; 4350 } 4351 4352 return false; 4353 } 4354 4355 /// parseDirectiveIfeqs 4356 /// ::= .ifeqs string1, string2 4357 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 4358 if (Lexer.isNot(AsmToken::String)) { 4359 if (ExpectEqual) 4360 TokError("expected string parameter for '.ifeqs' directive"); 4361 else 4362 TokError("expected string parameter for '.ifnes' directive"); 4363 eatToEndOfStatement(); 4364 return true; 4365 } 4366 4367 StringRef String1 = getTok().getStringContents(); 4368 Lex(); 4369 4370 if (Lexer.isNot(AsmToken::Comma)) { 4371 if (ExpectEqual) 4372 TokError("expected comma after first string for '.ifeqs' directive"); 4373 else 4374 TokError("expected comma after first string for '.ifnes' directive"); 4375 eatToEndOfStatement(); 4376 return true; 4377 } 4378 4379 Lex(); 4380 4381 if (Lexer.isNot(AsmToken::String)) { 4382 if (ExpectEqual) 4383 TokError("expected string parameter for '.ifeqs' directive"); 4384 else 4385 TokError("expected string parameter for '.ifnes' directive"); 4386 eatToEndOfStatement(); 4387 return true; 4388 } 4389 4390 StringRef String2 = getTok().getStringContents(); 4391 Lex(); 4392 4393 TheCondStack.push_back(TheCondState); 4394 TheCondState.TheCond = AsmCond::IfCond; 4395 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4396 TheCondState.Ignore = !TheCondState.CondMet; 4397 4398 return false; 4399 } 4400 4401 /// parseDirectiveIfdef 4402 /// ::= .ifdef symbol 4403 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 4404 StringRef Name; 4405 TheCondStack.push_back(TheCondState); 4406 TheCondState.TheCond = AsmCond::IfCond; 4407 4408 if (TheCondState.Ignore) { 4409 eatToEndOfStatement(); 4410 } else { 4411 if (parseIdentifier(Name)) 4412 return TokError("expected identifier after '.ifdef'"); 4413 4414 Lex(); 4415 4416 MCSymbol *Sym = getContext().lookupSymbol(Name); 4417 4418 if (expect_defined) 4419 TheCondState.CondMet = (Sym && !Sym->isUndefined()); 4420 else 4421 TheCondState.CondMet = (!Sym || Sym->isUndefined()); 4422 TheCondState.Ignore = !TheCondState.CondMet; 4423 } 4424 4425 return false; 4426 } 4427 4428 /// parseDirectiveElseIf 4429 /// ::= .elseif expression 4430 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 4431 if (TheCondState.TheCond != AsmCond::IfCond && 4432 TheCondState.TheCond != AsmCond::ElseIfCond) 4433 Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " 4434 " an .elseif"); 4435 TheCondState.TheCond = AsmCond::ElseIfCond; 4436 4437 bool LastIgnoreState = false; 4438 if (!TheCondStack.empty()) 4439 LastIgnoreState = TheCondStack.back().Ignore; 4440 if (LastIgnoreState || TheCondState.CondMet) { 4441 TheCondState.Ignore = true; 4442 eatToEndOfStatement(); 4443 } else { 4444 int64_t ExprValue; 4445 if (parseAbsoluteExpression(ExprValue)) 4446 return true; 4447 4448 if (getLexer().isNot(AsmToken::EndOfStatement)) 4449 return TokError("unexpected token in '.elseif' directive"); 4450 4451 Lex(); 4452 TheCondState.CondMet = ExprValue; 4453 TheCondState.Ignore = !TheCondState.CondMet; 4454 } 4455 4456 return false; 4457 } 4458 4459 /// parseDirectiveElse 4460 /// ::= .else 4461 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4462 if (getLexer().isNot(AsmToken::EndOfStatement)) 4463 return TokError("unexpected token in '.else' directive"); 4464 4465 Lex(); 4466 4467 if (TheCondState.TheCond != AsmCond::IfCond && 4468 TheCondState.TheCond != AsmCond::ElseIfCond) 4469 Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " 4470 ".elseif"); 4471 TheCondState.TheCond = AsmCond::ElseCond; 4472 bool LastIgnoreState = false; 4473 if (!TheCondStack.empty()) 4474 LastIgnoreState = TheCondStack.back().Ignore; 4475 if (LastIgnoreState || TheCondState.CondMet) 4476 TheCondState.Ignore = true; 4477 else 4478 TheCondState.Ignore = false; 4479 4480 return false; 4481 } 4482 4483 /// parseDirectiveEnd 4484 /// ::= .end 4485 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4486 if (getLexer().isNot(AsmToken::EndOfStatement)) 4487 return TokError("unexpected token in '.end' directive"); 4488 4489 Lex(); 4490 4491 while (Lexer.isNot(AsmToken::Eof)) 4492 Lex(); 4493 4494 return false; 4495 } 4496 4497 /// parseDirectiveError 4498 /// ::= .err 4499 /// ::= .error [string] 4500 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 4501 if (!TheCondStack.empty()) { 4502 if (TheCondStack.back().Ignore) { 4503 eatToEndOfStatement(); 4504 return false; 4505 } 4506 } 4507 4508 if (!WithMessage) 4509 return Error(L, ".err encountered"); 4510 4511 StringRef Message = ".error directive invoked in source file"; 4512 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4513 if (Lexer.isNot(AsmToken::String)) { 4514 TokError(".error argument must be a string"); 4515 eatToEndOfStatement(); 4516 return true; 4517 } 4518 4519 Message = getTok().getStringContents(); 4520 Lex(); 4521 } 4522 4523 Error(L, Message); 4524 return true; 4525 } 4526 4527 /// parseDirectiveWarning 4528 /// ::= .warning [string] 4529 bool AsmParser::parseDirectiveWarning(SMLoc L) { 4530 if (!TheCondStack.empty()) { 4531 if (TheCondStack.back().Ignore) { 4532 eatToEndOfStatement(); 4533 return false; 4534 } 4535 } 4536 4537 StringRef Message = ".warning directive invoked in source file"; 4538 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4539 if (Lexer.isNot(AsmToken::String)) { 4540 TokError(".warning argument must be a string"); 4541 eatToEndOfStatement(); 4542 return true; 4543 } 4544 4545 Message = getTok().getStringContents(); 4546 Lex(); 4547 } 4548 4549 Warning(L, Message); 4550 return false; 4551 } 4552 4553 /// parseDirectiveEndIf 4554 /// ::= .endif 4555 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 4556 if (getLexer().isNot(AsmToken::EndOfStatement)) 4557 return TokError("unexpected token in '.endif' directive"); 4558 4559 Lex(); 4560 4561 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 4562 Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " 4563 ".else"); 4564 if (!TheCondStack.empty()) { 4565 TheCondState = TheCondStack.back(); 4566 TheCondStack.pop_back(); 4567 } 4568 4569 return false; 4570 } 4571 4572 void AsmParser::initializeDirectiveKindMap() { 4573 DirectiveKindMap[".set"] = DK_SET; 4574 DirectiveKindMap[".equ"] = DK_EQU; 4575 DirectiveKindMap[".equiv"] = DK_EQUIV; 4576 DirectiveKindMap[".ascii"] = DK_ASCII; 4577 DirectiveKindMap[".asciz"] = DK_ASCIZ; 4578 DirectiveKindMap[".string"] = DK_STRING; 4579 DirectiveKindMap[".byte"] = DK_BYTE; 4580 DirectiveKindMap[".short"] = DK_SHORT; 4581 DirectiveKindMap[".value"] = DK_VALUE; 4582 DirectiveKindMap[".2byte"] = DK_2BYTE; 4583 DirectiveKindMap[".long"] = DK_LONG; 4584 DirectiveKindMap[".int"] = DK_INT; 4585 DirectiveKindMap[".4byte"] = DK_4BYTE; 4586 DirectiveKindMap[".quad"] = DK_QUAD; 4587 DirectiveKindMap[".8byte"] = DK_8BYTE; 4588 DirectiveKindMap[".octa"] = DK_OCTA; 4589 DirectiveKindMap[".single"] = DK_SINGLE; 4590 DirectiveKindMap[".float"] = DK_FLOAT; 4591 DirectiveKindMap[".double"] = DK_DOUBLE; 4592 DirectiveKindMap[".align"] = DK_ALIGN; 4593 DirectiveKindMap[".align32"] = DK_ALIGN32; 4594 DirectiveKindMap[".balign"] = DK_BALIGN; 4595 DirectiveKindMap[".balignw"] = DK_BALIGNW; 4596 DirectiveKindMap[".balignl"] = DK_BALIGNL; 4597 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 4598 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 4599 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 4600 DirectiveKindMap[".org"] = DK_ORG; 4601 DirectiveKindMap[".fill"] = DK_FILL; 4602 DirectiveKindMap[".zero"] = DK_ZERO; 4603 DirectiveKindMap[".extern"] = DK_EXTERN; 4604 DirectiveKindMap[".globl"] = DK_GLOBL; 4605 DirectiveKindMap[".global"] = DK_GLOBAL; 4606 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 4607 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 4608 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 4609 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 4610 DirectiveKindMap[".reference"] = DK_REFERENCE; 4611 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 4612 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 4613 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 4614 DirectiveKindMap[".comm"] = DK_COMM; 4615 DirectiveKindMap[".common"] = DK_COMMON; 4616 DirectiveKindMap[".lcomm"] = DK_LCOMM; 4617 DirectiveKindMap[".abort"] = DK_ABORT; 4618 DirectiveKindMap[".include"] = DK_INCLUDE; 4619 DirectiveKindMap[".incbin"] = DK_INCBIN; 4620 DirectiveKindMap[".code16"] = DK_CODE16; 4621 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 4622 DirectiveKindMap[".rept"] = DK_REPT; 4623 DirectiveKindMap[".rep"] = DK_REPT; 4624 DirectiveKindMap[".irp"] = DK_IRP; 4625 DirectiveKindMap[".irpc"] = DK_IRPC; 4626 DirectiveKindMap[".endr"] = DK_ENDR; 4627 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 4628 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 4629 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 4630 DirectiveKindMap[".if"] = DK_IF; 4631 DirectiveKindMap[".ifeq"] = DK_IFEQ; 4632 DirectiveKindMap[".ifge"] = DK_IFGE; 4633 DirectiveKindMap[".ifgt"] = DK_IFGT; 4634 DirectiveKindMap[".ifle"] = DK_IFLE; 4635 DirectiveKindMap[".iflt"] = DK_IFLT; 4636 DirectiveKindMap[".ifne"] = DK_IFNE; 4637 DirectiveKindMap[".ifb"] = DK_IFB; 4638 DirectiveKindMap[".ifnb"] = DK_IFNB; 4639 DirectiveKindMap[".ifc"] = DK_IFC; 4640 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 4641 DirectiveKindMap[".ifnc"] = DK_IFNC; 4642 DirectiveKindMap[".ifnes"] = DK_IFNES; 4643 DirectiveKindMap[".ifdef"] = DK_IFDEF; 4644 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 4645 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 4646 DirectiveKindMap[".elseif"] = DK_ELSEIF; 4647 DirectiveKindMap[".else"] = DK_ELSE; 4648 DirectiveKindMap[".end"] = DK_END; 4649 DirectiveKindMap[".endif"] = DK_ENDIF; 4650 DirectiveKindMap[".skip"] = DK_SKIP; 4651 DirectiveKindMap[".space"] = DK_SPACE; 4652 DirectiveKindMap[".file"] = DK_FILE; 4653 DirectiveKindMap[".line"] = DK_LINE; 4654 DirectiveKindMap[".loc"] = DK_LOC; 4655 DirectiveKindMap[".stabs"] = DK_STABS; 4656 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 4657 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 4658 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 4659 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 4660 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 4661 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 4662 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 4663 DirectiveKindMap[".sleb128"] = DK_SLEB128; 4664 DirectiveKindMap[".uleb128"] = DK_ULEB128; 4665 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 4666 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 4667 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 4668 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 4669 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 4670 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 4671 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 4672 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 4673 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 4674 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 4675 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 4676 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 4677 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 4678 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 4679 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 4680 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 4681 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 4682 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 4683 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 4684 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 4685 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 4686 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 4687 DirectiveKindMap[".macro"] = DK_MACRO; 4688 DirectiveKindMap[".exitm"] = DK_EXITM; 4689 DirectiveKindMap[".endm"] = DK_ENDM; 4690 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 4691 DirectiveKindMap[".purgem"] = DK_PURGEM; 4692 DirectiveKindMap[".err"] = DK_ERR; 4693 DirectiveKindMap[".error"] = DK_ERROR; 4694 DirectiveKindMap[".warning"] = DK_WARNING; 4695 DirectiveKindMap[".reloc"] = DK_RELOC; 4696 } 4697 4698 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 4699 AsmToken EndToken, StartToken = getTok(); 4700 4701 unsigned NestLevel = 0; 4702 for (;;) { 4703 // Check whether we have reached the end of the file. 4704 if (getLexer().is(AsmToken::Eof)) { 4705 Error(DirectiveLoc, "no matching '.endr' in definition"); 4706 return nullptr; 4707 } 4708 4709 if (Lexer.is(AsmToken::Identifier) && 4710 (getTok().getIdentifier() == ".rept")) { 4711 ++NestLevel; 4712 } 4713 4714 // Otherwise, check whether we have reached the .endr. 4715 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 4716 if (NestLevel == 0) { 4717 EndToken = getTok(); 4718 Lex(); 4719 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4720 TokError("unexpected token in '.endr' directive"); 4721 return nullptr; 4722 } 4723 break; 4724 } 4725 --NestLevel; 4726 } 4727 4728 // Otherwise, scan till the end of the statement. 4729 eatToEndOfStatement(); 4730 } 4731 4732 const char *BodyStart = StartToken.getLoc().getPointer(); 4733 const char *BodyEnd = EndToken.getLoc().getPointer(); 4734 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4735 4736 // We Are Anonymous. 4737 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 4738 return &MacroLikeBodies.back(); 4739 } 4740 4741 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 4742 raw_svector_ostream &OS) { 4743 OS << ".endr\n"; 4744 4745 std::unique_ptr<MemoryBuffer> Instantiation = 4746 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 4747 4748 // Create the macro instantiation object and add to the current macro 4749 // instantiation stack. 4750 MacroInstantiation *MI = new MacroInstantiation( 4751 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 4752 ActiveMacros.push_back(MI); 4753 4754 // Jump to the macro instantiation and prime the lexer. 4755 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 4756 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 4757 Lex(); 4758 } 4759 4760 /// parseDirectiveRept 4761 /// ::= .rep | .rept count 4762 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 4763 const MCExpr *CountExpr; 4764 SMLoc CountLoc = getTok().getLoc(); 4765 if (parseExpression(CountExpr)) 4766 return true; 4767 4768 int64_t Count; 4769 if (!CountExpr->evaluateAsAbsolute(Count)) { 4770 eatToEndOfStatement(); 4771 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 4772 } 4773 4774 if (Count < 0) 4775 return Error(CountLoc, "Count is negative"); 4776 4777 if (Lexer.isNot(AsmToken::EndOfStatement)) 4778 return TokError("unexpected token in '" + Dir + "' directive"); 4779 4780 // Eat the end of statement. 4781 Lex(); 4782 4783 // Lex the rept definition. 4784 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4785 if (!M) 4786 return true; 4787 4788 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4789 // to hold the macro body with substitutions. 4790 SmallString<256> Buf; 4791 raw_svector_ostream OS(Buf); 4792 while (Count--) { 4793 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 4794 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 4795 return true; 4796 } 4797 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4798 4799 return false; 4800 } 4801 4802 /// parseDirectiveIrp 4803 /// ::= .irp symbol,values 4804 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 4805 MCAsmMacroParameter Parameter; 4806 4807 if (parseIdentifier(Parameter.Name)) 4808 return TokError("expected identifier in '.irp' directive"); 4809 4810 if (Lexer.isNot(AsmToken::Comma)) 4811 return TokError("expected comma in '.irp' directive"); 4812 4813 Lex(); 4814 4815 MCAsmMacroArguments A; 4816 if (parseMacroArguments(nullptr, A)) 4817 return true; 4818 4819 // Eat the end of statement. 4820 Lex(); 4821 4822 // Lex the irp definition. 4823 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4824 if (!M) 4825 return true; 4826 4827 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4828 // to hold the macro body with substitutions. 4829 SmallString<256> Buf; 4830 raw_svector_ostream OS(Buf); 4831 4832 for (const MCAsmMacroArgument &Arg : A) { 4833 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 4834 // This is undocumented, but GAS seems to support it. 4835 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 4836 return true; 4837 } 4838 4839 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4840 4841 return false; 4842 } 4843 4844 /// parseDirectiveIrpc 4845 /// ::= .irpc symbol,values 4846 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 4847 MCAsmMacroParameter Parameter; 4848 4849 if (parseIdentifier(Parameter.Name)) 4850 return TokError("expected identifier in '.irpc' directive"); 4851 4852 if (Lexer.isNot(AsmToken::Comma)) 4853 return TokError("expected comma in '.irpc' directive"); 4854 4855 Lex(); 4856 4857 MCAsmMacroArguments A; 4858 if (parseMacroArguments(nullptr, A)) 4859 return true; 4860 4861 if (A.size() != 1 || A.front().size() != 1) 4862 return TokError("unexpected token in '.irpc' directive"); 4863 4864 // Eat the end of statement. 4865 Lex(); 4866 4867 // Lex the irpc definition. 4868 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 4869 if (!M) 4870 return true; 4871 4872 // Macro instantiation is lexical, unfortunately. We construct a new buffer 4873 // to hold the macro body with substitutions. 4874 SmallString<256> Buf; 4875 raw_svector_ostream OS(Buf); 4876 4877 StringRef Values = A.front().front().getString(); 4878 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 4879 MCAsmMacroArgument Arg; 4880 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 4881 4882 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 4883 // This is undocumented, but GAS seems to support it. 4884 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 4885 return true; 4886 } 4887 4888 instantiateMacroLikeBody(M, DirectiveLoc, OS); 4889 4890 return false; 4891 } 4892 4893 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 4894 if (ActiveMacros.empty()) 4895 return TokError("unmatched '.endr' directive"); 4896 4897 // The only .repl that should get here are the ones created by 4898 // instantiateMacroLikeBody. 4899 assert(getLexer().is(AsmToken::EndOfStatement)); 4900 4901 handleMacroExit(); 4902 return false; 4903 } 4904 4905 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 4906 size_t Len) { 4907 const MCExpr *Value; 4908 SMLoc ExprLoc = getLexer().getLoc(); 4909 if (parseExpression(Value)) 4910 return true; 4911 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4912 if (!MCE) 4913 return Error(ExprLoc, "unexpected expression in _emit"); 4914 uint64_t IntValue = MCE->getValue(); 4915 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 4916 return Error(ExprLoc, "literal value out of range for directive"); 4917 4918 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 4919 return false; 4920 } 4921 4922 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 4923 const MCExpr *Value; 4924 SMLoc ExprLoc = getLexer().getLoc(); 4925 if (parseExpression(Value)) 4926 return true; 4927 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 4928 if (!MCE) 4929 return Error(ExprLoc, "unexpected expression in align"); 4930 uint64_t IntValue = MCE->getValue(); 4931 if (!isPowerOf2_64(IntValue)) 4932 return Error(ExprLoc, "literal value not a power of two greater then zero"); 4933 4934 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 4935 return false; 4936 } 4937 4938 // We are comparing pointers, but the pointers are relative to a single string. 4939 // Thus, this should always be deterministic. 4940 static int rewritesSort(const AsmRewrite *AsmRewriteA, 4941 const AsmRewrite *AsmRewriteB) { 4942 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 4943 return -1; 4944 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 4945 return 1; 4946 4947 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 4948 // rewrite to the same location. Make sure the SizeDirective rewrite is 4949 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 4950 // ensures the sort algorithm is stable. 4951 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 4952 AsmRewritePrecedence[AsmRewriteB->Kind]) 4953 return -1; 4954 4955 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 4956 AsmRewritePrecedence[AsmRewriteB->Kind]) 4957 return 1; 4958 llvm_unreachable("Unstable rewrite sort."); 4959 } 4960 4961 bool AsmParser::parseMSInlineAsm( 4962 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 4963 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool> > &OpDecls, 4964 SmallVectorImpl<std::string> &Constraints, 4965 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 4966 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 4967 SmallVector<void *, 4> InputDecls; 4968 SmallVector<void *, 4> OutputDecls; 4969 SmallVector<bool, 4> InputDeclsAddressOf; 4970 SmallVector<bool, 4> OutputDeclsAddressOf; 4971 SmallVector<std::string, 4> InputConstraints; 4972 SmallVector<std::string, 4> OutputConstraints; 4973 SmallVector<unsigned, 4> ClobberRegs; 4974 4975 SmallVector<AsmRewrite, 4> AsmStrRewrites; 4976 4977 // Prime the lexer. 4978 Lex(); 4979 4980 // While we have input, parse each statement. 4981 unsigned InputIdx = 0; 4982 unsigned OutputIdx = 0; 4983 while (getLexer().isNot(AsmToken::Eof)) { 4984 ParseStatementInfo Info(&AsmStrRewrites); 4985 if (parseStatement(Info, &SI)) 4986 return true; 4987 4988 if (Info.ParseError) 4989 return true; 4990 4991 if (Info.Opcode == ~0U) 4992 continue; 4993 4994 const MCInstrDesc &Desc = MII->get(Info.Opcode); 4995 4996 // Build the list of clobbers, outputs and inputs. 4997 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 4998 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 4999 5000 // Immediate. 5001 if (Operand.isImm()) 5002 continue; 5003 5004 // Register operand. 5005 if (Operand.isReg() && !Operand.needAddressOf() && 5006 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5007 unsigned NumDefs = Desc.getNumDefs(); 5008 // Clobber. 5009 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5010 ClobberRegs.push_back(Operand.getReg()); 5011 continue; 5012 } 5013 5014 // Expr/Input or Output. 5015 StringRef SymName = Operand.getSymName(); 5016 if (SymName.empty()) 5017 continue; 5018 5019 void *OpDecl = Operand.getOpDecl(); 5020 if (!OpDecl) 5021 continue; 5022 5023 bool isOutput = (i == 1) && Desc.mayStore(); 5024 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5025 if (isOutput) { 5026 ++InputIdx; 5027 OutputDecls.push_back(OpDecl); 5028 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5029 OutputConstraints.push_back(("=" + Operand.getConstraint()).str()); 5030 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5031 } else { 5032 InputDecls.push_back(OpDecl); 5033 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5034 InputConstraints.push_back(Operand.getConstraint().str()); 5035 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5036 } 5037 } 5038 5039 // Consider implicit defs to be clobbers. Think of cpuid and push. 5040 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5041 Desc.getNumImplicitDefs()); 5042 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5043 } 5044 5045 // Set the number of Outputs and Inputs. 5046 NumOutputs = OutputDecls.size(); 5047 NumInputs = InputDecls.size(); 5048 5049 // Set the unique clobbers. 5050 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5051 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5052 ClobberRegs.end()); 5053 Clobbers.assign(ClobberRegs.size(), std::string()); 5054 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5055 raw_string_ostream OS(Clobbers[I]); 5056 IP->printRegName(OS, ClobberRegs[I]); 5057 } 5058 5059 // Merge the various outputs and inputs. Output are expected first. 5060 if (NumOutputs || NumInputs) { 5061 unsigned NumExprs = NumOutputs + NumInputs; 5062 OpDecls.resize(NumExprs); 5063 Constraints.resize(NumExprs); 5064 for (unsigned i = 0; i < NumOutputs; ++i) { 5065 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5066 Constraints[i] = OutputConstraints[i]; 5067 } 5068 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5069 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5070 Constraints[j] = InputConstraints[i]; 5071 } 5072 } 5073 5074 // Build the IR assembly string. 5075 std::string AsmStringIR; 5076 raw_string_ostream OS(AsmStringIR); 5077 StringRef ASMString = 5078 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5079 const char *AsmStart = ASMString.begin(); 5080 const char *AsmEnd = ASMString.end(); 5081 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5082 for (const AsmRewrite &AR : AsmStrRewrites) { 5083 AsmRewriteKind Kind = AR.Kind; 5084 if (Kind == AOK_Delete) 5085 continue; 5086 5087 const char *Loc = AR.Loc.getPointer(); 5088 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5089 5090 // Emit everything up to the immediate/expression. 5091 if (unsigned Len = Loc - AsmStart) 5092 OS << StringRef(AsmStart, Len); 5093 5094 // Skip the original expression. 5095 if (Kind == AOK_Skip) { 5096 AsmStart = Loc + AR.Len; 5097 continue; 5098 } 5099 5100 unsigned AdditionalSkip = 0; 5101 // Rewrite expressions in $N notation. 5102 switch (Kind) { 5103 default: 5104 break; 5105 case AOK_Imm: 5106 OS << "$$" << AR.Val; 5107 break; 5108 case AOK_ImmPrefix: 5109 OS << "$$"; 5110 break; 5111 case AOK_Label: 5112 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5113 break; 5114 case AOK_Input: 5115 OS << '$' << InputIdx++; 5116 break; 5117 case AOK_Output: 5118 OS << '$' << OutputIdx++; 5119 break; 5120 case AOK_SizeDirective: 5121 switch (AR.Val) { 5122 default: break; 5123 case 8: OS << "byte ptr "; break; 5124 case 16: OS << "word ptr "; break; 5125 case 32: OS << "dword ptr "; break; 5126 case 64: OS << "qword ptr "; break; 5127 case 80: OS << "xword ptr "; break; 5128 case 128: OS << "xmmword ptr "; break; 5129 case 256: OS << "ymmword ptr "; break; 5130 } 5131 break; 5132 case AOK_Emit: 5133 OS << ".byte"; 5134 break; 5135 case AOK_Align: { 5136 // MS alignment directives are measured in bytes. If the native assembler 5137 // measures alignment in bytes, we can pass it straight through. 5138 OS << ".align"; 5139 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5140 break; 5141 5142 // Alignment is in log2 form, so print that instead and skip the original 5143 // immediate. 5144 unsigned Val = AR.Val; 5145 OS << ' ' << Val; 5146 assert(Val < 10 && "Expected alignment less then 2^10."); 5147 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5148 break; 5149 } 5150 case AOK_EVEN: 5151 OS << ".even"; 5152 break; 5153 case AOK_DotOperator: 5154 // Insert the dot if the user omitted it. 5155 OS.flush(); 5156 if (AsmStringIR.back() != '.') 5157 OS << '.'; 5158 OS << AR.Val; 5159 break; 5160 } 5161 5162 // Skip the original expression. 5163 AsmStart = Loc + AR.Len + AdditionalSkip; 5164 } 5165 5166 // Emit the remainder of the asm string. 5167 if (AsmStart != AsmEnd) 5168 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5169 5170 AsmString = OS.str(); 5171 return false; 5172 } 5173 5174 namespace llvm { 5175 namespace MCParserUtils { 5176 5177 /// Returns whether the given symbol is used anywhere in the given expression, 5178 /// or subexpressions. 5179 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 5180 switch (Value->getKind()) { 5181 case MCExpr::Binary: { 5182 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 5183 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 5184 isSymbolUsedInExpression(Sym, BE->getRHS()); 5185 } 5186 case MCExpr::Target: 5187 case MCExpr::Constant: 5188 return false; 5189 case MCExpr::SymbolRef: { 5190 const MCSymbol &S = 5191 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 5192 if (S.isVariable()) 5193 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 5194 return &S == Sym; 5195 } 5196 case MCExpr::Unary: 5197 return isSymbolUsedInExpression( 5198 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 5199 } 5200 5201 llvm_unreachable("Unknown expr kind!"); 5202 } 5203 5204 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 5205 MCAsmParser &Parser, MCSymbol *&Sym, 5206 const MCExpr *&Value) { 5207 MCAsmLexer &Lexer = Parser.getLexer(); 5208 5209 // FIXME: Use better location, we should use proper tokens. 5210 SMLoc EqualLoc = Lexer.getLoc(); 5211 5212 if (Parser.parseExpression(Value)) { 5213 Parser.TokError("missing expression"); 5214 Parser.eatToEndOfStatement(); 5215 return true; 5216 } 5217 5218 // Note: we don't count b as used in "a = b". This is to allow 5219 // a = b 5220 // b = c 5221 5222 if (Lexer.isNot(AsmToken::EndOfStatement)) 5223 return Parser.TokError("unexpected token in assignment"); 5224 5225 // Eat the end of statement marker. 5226 Parser.Lex(); 5227 5228 // Validate that the LHS is allowed to be a variable (either it has not been 5229 // used as a symbol, or it is an absolute symbol). 5230 Sym = Parser.getContext().lookupSymbol(Name); 5231 if (Sym) { 5232 // Diagnose assignment to a label. 5233 // 5234 // FIXME: Diagnostics. Note the location of the definition as a label. 5235 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 5236 if (isSymbolUsedInExpression(Sym, Value)) 5237 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 5238 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 5239 !Sym->isVariable()) 5240 ; // Allow redefinitions of undefined symbols only used in directives. 5241 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 5242 ; // Allow redefinitions of variables that haven't yet been used. 5243 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 5244 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 5245 else if (!Sym->isVariable()) 5246 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 5247 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 5248 return Parser.Error(EqualLoc, 5249 "invalid reassignment of non-absolute variable '" + 5250 Name + "'"); 5251 } else if (Name == ".") { 5252 Parser.getStreamer().emitValueToOffset(Value, 0); 5253 return false; 5254 } else 5255 Sym = Parser.getContext().getOrCreateSymbol(Name); 5256 5257 Sym->setRedefinable(allow_redef); 5258 5259 return false; 5260 } 5261 5262 } // namespace MCParserUtils 5263 } // namespace llvm 5264 5265 /// \brief Create an MCAsmParser instance. 5266 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 5267 MCStreamer &Out, const MCAsmInfo &MAI) { 5268 return new AsmParser(SM, C, Out, MAI); 5269 } 5270