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/APInt.h" 16 #include "llvm/ADT/ArrayRef.h" 17 #include "llvm/ADT/None.h" 18 #include "llvm/ADT/STLExtras.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/SmallVector.h" 21 #include "llvm/ADT/StringMap.h" 22 #include "llvm/ADT/StringRef.h" 23 #include "llvm/ADT/Twine.h" 24 #include "llvm/BinaryFormat/Dwarf.h" 25 #include "llvm/MC/MCAsmInfo.h" 26 #include "llvm/MC/MCCodeView.h" 27 #include "llvm/MC/MCContext.h" 28 #include "llvm/MC/MCDirectives.h" 29 #include "llvm/MC/MCDwarf.h" 30 #include "llvm/MC/MCExpr.h" 31 #include "llvm/MC/MCInstPrinter.h" 32 #include "llvm/MC/MCInstrDesc.h" 33 #include "llvm/MC/MCInstrInfo.h" 34 #include "llvm/MC/MCObjectFileInfo.h" 35 #include "llvm/MC/MCParser/AsmCond.h" 36 #include "llvm/MC/MCParser/AsmLexer.h" 37 #include "llvm/MC/MCParser/MCAsmLexer.h" 38 #include "llvm/MC/MCParser/MCAsmParser.h" 39 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 40 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 41 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 42 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 43 #include "llvm/MC/MCRegisterInfo.h" 44 #include "llvm/MC/MCSection.h" 45 #include "llvm/MC/MCStreamer.h" 46 #include "llvm/MC/MCSymbol.h" 47 #include "llvm/MC/MCTargetOptions.h" 48 #include "llvm/MC/MCValue.h" 49 #include "llvm/Support/Casting.h" 50 #include "llvm/Support/CommandLine.h" 51 #include "llvm/Support/ErrorHandling.h" 52 #include "llvm/Support/MathExtras.h" 53 #include "llvm/Support/MemoryBuffer.h" 54 #include "llvm/Support/SMLoc.h" 55 #include "llvm/Support/SourceMgr.h" 56 #include "llvm/Support/raw_ostream.h" 57 #include <algorithm> 58 #include <cassert> 59 #include <cctype> 60 #include <climits> 61 #include <cstddef> 62 #include <cstdint> 63 #include <deque> 64 #include <memory> 65 #include <sstream> 66 #include <string> 67 #include <tuple> 68 #include <utility> 69 #include <vector> 70 71 using namespace llvm; 72 73 MCAsmParserSemaCallback::~MCAsmParserSemaCallback() = default; 74 75 static cl::opt<unsigned> AsmMacroMaxNestingDepth( 76 "asm-macro-max-nesting-depth", cl::init(20), cl::Hidden, 77 cl::desc("The maximum nesting depth allowed for assembly macros.")); 78 79 namespace { 80 81 /// \brief Helper types for tracking macro definitions. 82 typedef std::vector<AsmToken> MCAsmMacroArgument; 83 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 84 85 struct MCAsmMacroParameter { 86 StringRef Name; 87 MCAsmMacroArgument Value; 88 bool Required = false; 89 bool Vararg = false; 90 91 MCAsmMacroParameter() = default; 92 }; 93 94 typedef std::vector<MCAsmMacroParameter> MCAsmMacroParameters; 95 96 struct MCAsmMacro { 97 StringRef Name; 98 StringRef Body; 99 MCAsmMacroParameters Parameters; 100 101 public: 102 MCAsmMacro(StringRef N, StringRef B, MCAsmMacroParameters P) 103 : Name(N), Body(B), Parameters(std::move(P)) {} 104 }; 105 106 /// \brief Helper class for storing information about an active macro 107 /// instantiation. 108 struct MacroInstantiation { 109 /// The location of the instantiation. 110 SMLoc InstantiationLoc; 111 112 /// The buffer where parsing should resume upon instantiation completion. 113 int ExitBuffer; 114 115 /// The location where parsing should resume upon instantiation completion. 116 SMLoc ExitLoc; 117 118 /// The depth of TheCondStack at the start of the instantiation. 119 size_t CondStackDepth; 120 121 public: 122 MacroInstantiation(SMLoc IL, int EB, SMLoc EL, size_t CondStackDepth); 123 }; 124 125 struct ParseStatementInfo { 126 /// \brief The parsed operands from the last parsed statement. 127 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 128 129 /// \brief The opcode from the last parsed instruction. 130 unsigned Opcode = ~0U; 131 132 /// \brief Was there an error parsing the inline assembly? 133 bool ParseError = false; 134 135 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 136 137 ParseStatementInfo() = delete; 138 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 139 : AsmRewrites(rewrites) {} 140 }; 141 142 /// \brief The concrete assembly parser instance. 143 class AsmParser : public MCAsmParser { 144 private: 145 AsmLexer Lexer; 146 MCContext &Ctx; 147 MCStreamer &Out; 148 const MCAsmInfo &MAI; 149 SourceMgr &SrcMgr; 150 SourceMgr::DiagHandlerTy SavedDiagHandler; 151 void *SavedDiagContext; 152 std::unique_ptr<MCAsmParserExtension> PlatformParser; 153 154 /// This is the current buffer index we're lexing from as managed by the 155 /// SourceMgr object. 156 unsigned CurBuffer; 157 158 AsmCond TheCondState; 159 std::vector<AsmCond> TheCondStack; 160 161 /// \brief maps directive names to handler methods in parser 162 /// extensions. Extensions register themselves in this map by calling 163 /// addDirectiveHandler. 164 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 165 166 /// \brief Map of currently defined macros. 167 StringMap<MCAsmMacro> MacroMap; 168 169 /// \brief Stack of active macro instantiations. 170 std::vector<MacroInstantiation*> ActiveMacros; 171 172 /// \brief List of bodies of anonymous macros. 173 std::deque<MCAsmMacro> MacroLikeBodies; 174 175 /// Boolean tracking whether macro substitution is enabled. 176 unsigned MacrosEnabledFlag : 1; 177 178 /// \brief Keeps track of how many .macro's have been instantiated. 179 unsigned NumOfMacroInstantiations; 180 181 /// The values from the last parsed cpp hash file line comment if any. 182 struct CppHashInfoTy { 183 StringRef Filename; 184 int64_t LineNumber = 0; 185 SMLoc Loc; 186 unsigned Buf = 0; 187 }; 188 CppHashInfoTy CppHashInfo; 189 190 /// \brief List of forward directional labels for diagnosis at the end. 191 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 192 193 /// When generating dwarf for assembly source files we need to calculate the 194 /// logical line number based on the last parsed cpp hash file line comment 195 /// and current line. Since this is slow and messes up the SourceMgr's 196 /// cache we save the last info we queried with SrcMgr.FindLineNumber(). 197 SMLoc LastQueryIDLoc; 198 unsigned LastQueryBuffer; 199 unsigned LastQueryLine; 200 201 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 202 unsigned AssemblerDialect = ~0U; 203 204 /// \brief is Darwin compatibility enabled? 205 bool IsDarwin = false; 206 207 /// \brief Are we parsing ms-style inline assembly? 208 bool ParsingInlineAsm = false; 209 210 public: 211 AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 212 const MCAsmInfo &MAI, unsigned CB); 213 AsmParser(const AsmParser &) = delete; 214 AsmParser &operator=(const AsmParser &) = delete; 215 ~AsmParser() override; 216 217 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 218 219 void addDirectiveHandler(StringRef Directive, 220 ExtensionDirectiveHandler Handler) override { 221 ExtensionDirectiveMap[Directive] = Handler; 222 } 223 224 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 225 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 226 } 227 228 /// @name MCAsmParser Interface 229 /// { 230 231 SourceMgr &getSourceManager() override { return SrcMgr; } 232 MCAsmLexer &getLexer() override { return Lexer; } 233 MCContext &getContext() override { return Ctx; } 234 MCStreamer &getStreamer() override { return Out; } 235 236 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 237 238 unsigned getAssemblerDialect() override { 239 if (AssemblerDialect == ~0U) 240 return MAI.getAssemblerDialect(); 241 else 242 return AssemblerDialect; 243 } 244 void setAssemblerDialect(unsigned i) override { 245 AssemblerDialect = i; 246 } 247 248 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 249 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 250 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 251 252 const AsmToken &Lex() override; 253 254 void setParsingInlineAsm(bool V) override { 255 ParsingInlineAsm = V; 256 Lexer.setParsingMSInlineAsm(V); 257 } 258 bool isParsingInlineAsm() override { return ParsingInlineAsm; } 259 260 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 261 unsigned &NumOutputs, unsigned &NumInputs, 262 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 263 SmallVectorImpl<std::string> &Constraints, 264 SmallVectorImpl<std::string> &Clobbers, 265 const MCInstrInfo *MII, const MCInstPrinter *IP, 266 MCAsmParserSemaCallback &SI) override; 267 268 bool parseExpression(const MCExpr *&Res); 269 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 270 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 271 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 272 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 273 SMLoc &EndLoc) override; 274 bool parseAbsoluteExpression(int64_t &Res) override; 275 276 /// \brief Parse a floating point expression using the float \p Semantics 277 /// and set \p Res to the value. 278 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 279 280 /// \brief Parse an identifier or string (as a quoted identifier) 281 /// and set \p Res to the identifier contents. 282 bool parseIdentifier(StringRef &Res) override; 283 void eatToEndOfStatement() override; 284 285 bool checkForValidSection() override; 286 287 /// } 288 289 private: 290 bool isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc); 291 void altMacroString(StringRef AltMacroStr, std::string &Res); 292 bool parseStatement(ParseStatementInfo &Info, 293 MCAsmParserSemaCallback *SI); 294 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 295 bool parseCppHashLineFilenameComment(SMLoc L); 296 297 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 298 ArrayRef<MCAsmMacroParameter> Parameters); 299 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 300 ArrayRef<MCAsmMacroParameter> Parameters, 301 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 302 SMLoc L); 303 304 /// \brief Are macros enabled in the parser? 305 bool areMacrosEnabled() {return MacrosEnabledFlag;} 306 307 /// \brief Control a flag in the parser that enables or disables macros. 308 void setMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} 309 310 /// \brief Lookup a previously defined macro. 311 /// \param Name Macro name. 312 /// \returns Pointer to macro. NULL if no such macro was defined. 313 const MCAsmMacro* lookupMacro(StringRef Name); 314 315 /// \brief Define a new macro with the given name and information. 316 void defineMacro(StringRef Name, MCAsmMacro Macro); 317 318 /// \brief Undefine a macro. If no such macro was defined, it's a no-op. 319 void undefineMacro(StringRef Name); 320 321 /// \brief Are we inside a macro instantiation? 322 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 323 324 /// \brief Handle entry to macro instantiation. 325 /// 326 /// \param M The macro. 327 /// \param NameLoc Instantiation location. 328 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 329 330 /// \brief Handle exit from macro instantiation. 331 void handleMacroExit(); 332 333 /// \brief Extract AsmTokens for a macro argument. 334 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 335 336 /// \brief Parse all macro arguments for a given macro. 337 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 338 339 void printMacroInstantiations(); 340 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 341 SMRange Range = None) const { 342 ArrayRef<SMRange> Ranges(Range); 343 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 344 } 345 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 346 347 /// \brief Enter the specified file. This returns true on failure. 348 bool enterIncludeFile(const std::string &Filename); 349 350 /// \brief Process the specified file for the .incbin directive. 351 /// This returns true on failure. 352 bool processIncbinFile(const std::string &Filename, int64_t Skip = 0, 353 const MCExpr *Count = nullptr, SMLoc Loc = SMLoc()); 354 355 /// \brief Reset the current lexer position to that given by \p Loc. The 356 /// current token is not set; clients should ensure Lex() is called 357 /// subsequently. 358 /// 359 /// \param InBuffer If not 0, should be the known buffer id that contains the 360 /// location. 361 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 362 363 /// \brief Parse up to the end of statement and a return the contents from the 364 /// current token until the end of the statement; the current token on exit 365 /// will be either the EndOfStatement or EOF. 366 StringRef parseStringToEndOfStatement() override; 367 368 /// \brief Parse until the end of a statement or a comma is encountered, 369 /// return the contents from the current token up to the end or comma. 370 StringRef parseStringToComma(); 371 372 bool parseAssignment(StringRef Name, bool allow_redef, 373 bool NoDeadStrip = false); 374 375 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 376 MCBinaryExpr::Opcode &Kind); 377 378 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 379 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 380 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 381 382 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 383 384 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 385 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 386 387 // Generic (target and platform independent) directive parsing. 388 enum DirectiveKind { 389 DK_NO_DIRECTIVE, // Placeholder 390 DK_SET, DK_EQU, DK_EQUIV, DK_ASCII, DK_ASCIZ, DK_STRING, DK_BYTE, DK_SHORT, 391 DK_RELOC, 392 DK_VALUE, DK_2BYTE, DK_LONG, DK_INT, DK_4BYTE, DK_QUAD, DK_8BYTE, DK_OCTA, 393 DK_DC, DK_DC_A, DK_DC_B, DK_DC_D, DK_DC_L, DK_DC_S, DK_DC_W, DK_DC_X, 394 DK_DCB, DK_DCB_B, DK_DCB_D, DK_DCB_L, DK_DCB_S, DK_DCB_W, DK_DCB_X, 395 DK_DS, DK_DS_B, DK_DS_D, DK_DS_L, DK_DS_P, DK_DS_S, DK_DS_W, DK_DS_X, 396 DK_SINGLE, DK_FLOAT, DK_DOUBLE, DK_ALIGN, DK_ALIGN32, DK_BALIGN, DK_BALIGNW, 397 DK_BALIGNL, DK_P2ALIGN, DK_P2ALIGNW, DK_P2ALIGNL, DK_ORG, DK_FILL, DK_ENDR, 398 DK_BUNDLE_ALIGN_MODE, DK_BUNDLE_LOCK, DK_BUNDLE_UNLOCK, 399 DK_ZERO, DK_EXTERN, DK_GLOBL, DK_GLOBAL, 400 DK_LAZY_REFERENCE, DK_NO_DEAD_STRIP, DK_SYMBOL_RESOLVER, 401 DK_PRIVATE_EXTERN, DK_REFERENCE, DK_WEAK_DEFINITION, DK_WEAK_REFERENCE, 402 DK_WEAK_DEF_CAN_BE_HIDDEN, DK_COMM, DK_COMMON, DK_LCOMM, DK_ABORT, 403 DK_INCLUDE, DK_INCBIN, DK_CODE16, DK_CODE16GCC, DK_REPT, DK_IRP, DK_IRPC, 404 DK_IF, DK_IFEQ, DK_IFGE, DK_IFGT, DK_IFLE, DK_IFLT, DK_IFNE, DK_IFB, 405 DK_IFNB, DK_IFC, DK_IFEQS, DK_IFNC, DK_IFNES, DK_IFDEF, DK_IFNDEF, 406 DK_IFNOTDEF, DK_ELSEIF, DK_ELSE, DK_ENDIF, 407 DK_SPACE, DK_SKIP, DK_FILE, DK_LINE, DK_LOC, DK_STABS, 408 DK_CV_FILE, DK_CV_FUNC_ID, DK_CV_INLINE_SITE_ID, DK_CV_LOC, DK_CV_LINETABLE, 409 DK_CV_INLINE_LINETABLE, DK_CV_DEF_RANGE, DK_CV_STRINGTABLE, 410 DK_CV_FILECHECKSUMS, 411 DK_CFI_SECTIONS, DK_CFI_STARTPROC, DK_CFI_ENDPROC, DK_CFI_DEF_CFA, 412 DK_CFI_DEF_CFA_OFFSET, DK_CFI_ADJUST_CFA_OFFSET, DK_CFI_DEF_CFA_REGISTER, 413 DK_CFI_OFFSET, DK_CFI_REL_OFFSET, DK_CFI_PERSONALITY, DK_CFI_LSDA, 414 DK_CFI_REMEMBER_STATE, DK_CFI_RESTORE_STATE, DK_CFI_SAME_VALUE, 415 DK_CFI_RESTORE, DK_CFI_ESCAPE, DK_CFI_SIGNAL_FRAME, DK_CFI_UNDEFINED, 416 DK_CFI_REGISTER, DK_CFI_WINDOW_SAVE, 417 DK_MACROS_ON, DK_MACROS_OFF, DK_ALTMACRO, DK_NOALTMACRO, 418 DK_MACRO, DK_EXITM, DK_ENDM, DK_ENDMACRO, DK_PURGEM, 419 DK_SLEB128, DK_ULEB128, 420 DK_ERR, DK_ERROR, DK_WARNING, 421 DK_END 422 }; 423 424 /// \brief Maps directive name --> DirectiveKind enum, for 425 /// directives parsed by this class. 426 StringMap<DirectiveKind> DirectiveKindMap; 427 428 // ".ascii", ".asciz", ".string" 429 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 430 bool parseDirectiveReloc(SMLoc DirectiveLoc); // ".reloc" 431 bool parseDirectiveValue(StringRef IDVal, 432 unsigned Size); // ".byte", ".long", ... 433 bool parseDirectiveOctaValue(StringRef IDVal); // ".octa", ... 434 bool parseDirectiveRealValue(StringRef IDVal, 435 const fltSemantics &); // ".single", ... 436 bool parseDirectiveFill(); // ".fill" 437 bool parseDirectiveZero(); // ".zero" 438 // ".set", ".equ", ".equiv" 439 bool parseDirectiveSet(StringRef IDVal, bool allow_redef); 440 bool parseDirectiveOrg(); // ".org" 441 // ".align{,32}", ".p2align{,w,l}" 442 bool parseDirectiveAlign(bool IsPow2, unsigned ValueSize); 443 444 // ".file", ".line", ".loc", ".stabs" 445 bool parseDirectiveFile(SMLoc DirectiveLoc); 446 bool parseDirectiveLine(); 447 bool parseDirectiveLoc(); 448 bool parseDirectiveStabs(); 449 450 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 451 // ".cv_inline_linetable", ".cv_def_range" 452 bool parseDirectiveCVFile(); 453 bool parseDirectiveCVFuncId(); 454 bool parseDirectiveCVInlineSiteId(); 455 bool parseDirectiveCVLoc(); 456 bool parseDirectiveCVLinetable(); 457 bool parseDirectiveCVInlineLinetable(); 458 bool parseDirectiveCVDefRange(); 459 bool parseDirectiveCVStringTable(); 460 bool parseDirectiveCVFileChecksums(); 461 462 // .cfi directives 463 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 464 bool parseDirectiveCFIWindowSave(); 465 bool parseDirectiveCFISections(); 466 bool parseDirectiveCFIStartProc(); 467 bool parseDirectiveCFIEndProc(); 468 bool parseDirectiveCFIDefCfaOffset(); 469 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 470 bool parseDirectiveCFIAdjustCfaOffset(); 471 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 472 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 473 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 474 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 475 bool parseDirectiveCFIRememberState(); 476 bool parseDirectiveCFIRestoreState(); 477 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 478 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 479 bool parseDirectiveCFIEscape(); 480 bool parseDirectiveCFISignalFrame(); 481 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 482 483 // macro directives 484 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 485 bool parseDirectiveExitMacro(StringRef Directive); 486 bool parseDirectiveEndMacro(StringRef Directive); 487 bool parseDirectiveMacro(SMLoc DirectiveLoc); 488 bool parseDirectiveMacrosOnOff(StringRef Directive); 489 // alternate macro mode directives 490 bool parseDirectiveAltmacro(StringRef Directive); 491 // ".bundle_align_mode" 492 bool parseDirectiveBundleAlignMode(); 493 // ".bundle_lock" 494 bool parseDirectiveBundleLock(); 495 // ".bundle_unlock" 496 bool parseDirectiveBundleUnlock(); 497 498 // ".space", ".skip" 499 bool parseDirectiveSpace(StringRef IDVal); 500 501 // ".dcb" 502 bool parseDirectiveDCB(StringRef IDVal, unsigned Size); 503 bool parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &); 504 // ".ds" 505 bool parseDirectiveDS(StringRef IDVal, unsigned Size); 506 507 // .sleb128 (Signed=true) and .uleb128 (Signed=false) 508 bool parseDirectiveLEB128(bool Signed); 509 510 /// \brief Parse a directive like ".globl" which 511 /// accepts a single symbol (which should be a label or an external). 512 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 513 514 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 515 516 bool parseDirectiveAbort(); // ".abort" 517 bool parseDirectiveInclude(); // ".include" 518 bool parseDirectiveIncbin(); // ".incbin" 519 520 // ".if", ".ifeq", ".ifge", ".ifgt" , ".ifle", ".iflt" or ".ifne" 521 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 522 // ".ifb" or ".ifnb", depending on ExpectBlank. 523 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 524 // ".ifc" or ".ifnc", depending on ExpectEqual. 525 bool parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual); 526 // ".ifeqs" or ".ifnes", depending on ExpectEqual. 527 bool parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual); 528 // ".ifdef" or ".ifndef", depending on expect_defined 529 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 530 bool parseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" 531 bool parseDirectiveElse(SMLoc DirectiveLoc); // ".else" 532 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // .endif 533 bool parseEscapedString(std::string &Data) override; 534 535 const MCExpr *applyModifierToExpr(const MCExpr *E, 536 MCSymbolRefExpr::VariantKind Variant); 537 538 // Macro-like directives 539 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 540 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 541 raw_svector_ostream &OS); 542 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 543 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 544 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 545 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 546 547 // "_emit" or "__emit" 548 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 549 size_t Len); 550 551 // "align" 552 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 553 554 // "end" 555 bool parseDirectiveEnd(SMLoc DirectiveLoc); 556 557 // ".err" or ".error" 558 bool parseDirectiveError(SMLoc DirectiveLoc, bool WithMessage); 559 560 // ".warning" 561 bool parseDirectiveWarning(SMLoc DirectiveLoc); 562 563 void initializeDirectiveKindMap(); 564 }; 565 566 } // end anonymous namespace 567 568 namespace llvm { 569 570 extern MCAsmParserExtension *createDarwinAsmParser(); 571 extern MCAsmParserExtension *createELFAsmParser(); 572 extern MCAsmParserExtension *createCOFFAsmParser(); 573 574 } // end namespace llvm 575 576 enum { DEFAULT_ADDRSPACE = 0 }; 577 578 AsmParser::AsmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 579 const MCAsmInfo &MAI, unsigned CB = 0) 580 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 581 CurBuffer(CB ? CB : SM.getMainFileID()), MacrosEnabledFlag(true) { 582 HadError = false; 583 // Save the old handler. 584 SavedDiagHandler = SrcMgr.getDiagHandler(); 585 SavedDiagContext = SrcMgr.getDiagContext(); 586 // Set our own handler which calls the saved handler. 587 SrcMgr.setDiagHandler(DiagHandler, this); 588 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 589 590 // Initialize the platform / file format parser. 591 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 592 case MCObjectFileInfo::IsCOFF: 593 PlatformParser.reset(createCOFFAsmParser()); 594 break; 595 case MCObjectFileInfo::IsMachO: 596 PlatformParser.reset(createDarwinAsmParser()); 597 IsDarwin = true; 598 break; 599 case MCObjectFileInfo::IsELF: 600 PlatformParser.reset(createELFAsmParser()); 601 break; 602 case MCObjectFileInfo::IsWasm: 603 llvm_unreachable("Wasm parsing not supported yet"); 604 break; 605 } 606 607 PlatformParser->Initialize(*this); 608 initializeDirectiveKindMap(); 609 610 NumOfMacroInstantiations = 0; 611 } 612 613 AsmParser::~AsmParser() { 614 assert((HadError || ActiveMacros.empty()) && 615 "Unexpected active macro instantiation!"); 616 617 // Restore the saved diagnostics handler and context for use during 618 // finalization. 619 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 620 } 621 622 void AsmParser::printMacroInstantiations() { 623 // Print the active macro instantiation stack. 624 for (std::vector<MacroInstantiation *>::const_reverse_iterator 625 it = ActiveMacros.rbegin(), 626 ie = ActiveMacros.rend(); 627 it != ie; ++it) 628 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 629 "while in macro instantiation"); 630 } 631 632 void AsmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 633 printPendingErrors(); 634 printMessage(L, SourceMgr::DK_Note, Msg, Range); 635 printMacroInstantiations(); 636 } 637 638 bool AsmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 639 if(getTargetParser().getTargetOptions().MCNoWarn) 640 return false; 641 if (getTargetParser().getTargetOptions().MCFatalWarnings) 642 return Error(L, Msg, Range); 643 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 644 printMacroInstantiations(); 645 return false; 646 } 647 648 bool AsmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 649 HadError = true; 650 printMessage(L, SourceMgr::DK_Error, Msg, Range); 651 printMacroInstantiations(); 652 return true; 653 } 654 655 bool AsmParser::enterIncludeFile(const std::string &Filename) { 656 std::string IncludedFile; 657 unsigned NewBuf = 658 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 659 if (!NewBuf) 660 return true; 661 662 CurBuffer = NewBuf; 663 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 664 return false; 665 } 666 667 /// Process the specified .incbin file by searching for it in the include paths 668 /// then just emitting the byte contents of the file to the streamer. This 669 /// returns true on failure. 670 bool AsmParser::processIncbinFile(const std::string &Filename, int64_t Skip, 671 const MCExpr *Count, SMLoc Loc) { 672 std::string IncludedFile; 673 unsigned NewBuf = 674 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 675 if (!NewBuf) 676 return true; 677 678 // Pick up the bytes from the file and emit them. 679 StringRef Bytes = SrcMgr.getMemoryBuffer(NewBuf)->getBuffer(); 680 Bytes = Bytes.drop_front(Skip); 681 if (Count) { 682 int64_t Res; 683 if (!Count->evaluateAsAbsolute(Res)) 684 return Error(Loc, "expected absolute expression"); 685 if (Res < 0) 686 return Warning(Loc, "negative count has no effect"); 687 Bytes = Bytes.take_front(Res); 688 } 689 getStreamer().EmitBytes(Bytes); 690 return false; 691 } 692 693 void AsmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 694 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 695 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 696 Loc.getPointer()); 697 } 698 699 const AsmToken &AsmParser::Lex() { 700 if (Lexer.getTok().is(AsmToken::Error)) 701 Error(Lexer.getErrLoc(), Lexer.getErr()); 702 703 // if it's a end of statement with a comment in it 704 if (getTok().is(AsmToken::EndOfStatement)) { 705 // if this is a line comment output it. 706 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 707 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 708 Out.addExplicitComment(Twine(getTok().getString())); 709 } 710 711 const AsmToken *tok = &Lexer.Lex(); 712 713 // Parse comments here to be deferred until end of next statement. 714 while (tok->is(AsmToken::Comment)) { 715 if (MAI.preserveAsmComments()) 716 Out.addExplicitComment(Twine(tok->getString())); 717 tok = &Lexer.Lex(); 718 } 719 720 if (tok->is(AsmToken::Eof)) { 721 // If this is the end of an included file, pop the parent file off the 722 // include stack. 723 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 724 if (ParentIncludeLoc != SMLoc()) { 725 jumpToLoc(ParentIncludeLoc); 726 return Lex(); 727 } 728 } 729 730 return *tok; 731 } 732 733 bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 734 // Create the initial section, if requested. 735 if (!NoInitialTextSection) 736 Out.InitSections(false); 737 738 // Prime the lexer. 739 Lex(); 740 741 HadError = false; 742 AsmCond StartingCondState = TheCondState; 743 SmallVector<AsmRewrite, 4> AsmStrRewrites; 744 745 // If we are generating dwarf for assembly source files save the initial text 746 // section and generate a .file directive. 747 if (getContext().getGenDwarfForAssembly()) { 748 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 749 if (!Sec->getBeginSymbol()) { 750 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 751 getStreamer().EmitLabel(SectionStartSym); 752 Sec->setBeginSymbol(SectionStartSym); 753 } 754 bool InsertResult = getContext().addGenDwarfSection(Sec); 755 assert(InsertResult && ".text section should not have debug info yet"); 756 (void)InsertResult; 757 getContext().setGenDwarfFileNumber(getStreamer().EmitDwarfFileDirective( 758 0, StringRef(), getContext().getMainFileName())); 759 } 760 761 // While we have input, parse each statement. 762 while (Lexer.isNot(AsmToken::Eof)) { 763 ParseStatementInfo Info(&AsmStrRewrites); 764 if (!parseStatement(Info, nullptr)) 765 continue; 766 767 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 768 // for printing ErrMsg via Lex() only if no (presumably better) parser error 769 // exists. 770 if (!hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 771 Lex(); 772 } 773 774 // parseStatement returned true so may need to emit an error. 775 printPendingErrors(); 776 777 // Skipping to the next line if needed. 778 if (!getLexer().isAtStartOfStatement()) 779 eatToEndOfStatement(); 780 } 781 782 // All errors should have been emitted. 783 assert(!hasPendingError() && "unexpected error from parseStatement"); 784 785 getTargetParser().flushPendingInstructions(getStreamer()); 786 787 if (TheCondState.TheCond != StartingCondState.TheCond || 788 TheCondState.Ignore != StartingCondState.Ignore) 789 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 790 // Check to see there are no empty DwarfFile slots. 791 const auto &LineTables = getContext().getMCDwarfLineTables(); 792 if (!LineTables.empty()) { 793 unsigned Index = 0; 794 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 795 if (File.Name.empty() && Index != 0) 796 printError(getTok().getLoc(), "unassigned file number: " + 797 Twine(Index) + 798 " for .file directives"); 799 ++Index; 800 } 801 } 802 803 // Check to see that all assembler local symbols were actually defined. 804 // Targets that don't do subsections via symbols may not want this, though, 805 // so conservatively exclude them. Only do this if we're finalizing, though, 806 // as otherwise we won't necessarilly have seen everything yet. 807 if (!NoFinalize) { 808 if (MAI.hasSubsectionsViaSymbols()) { 809 for (const auto &TableEntry : getContext().getSymbols()) { 810 MCSymbol *Sym = TableEntry.getValue(); 811 // Variable symbols may not be marked as defined, so check those 812 // explicitly. If we know it's a variable, we have a definition for 813 // the purposes of this check. 814 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 815 // FIXME: We would really like to refer back to where the symbol was 816 // first referenced for a source location. We need to add something 817 // to track that. Currently, we just point to the end of the file. 818 printError(getTok().getLoc(), "assembler local symbol '" + 819 Sym->getName() + "' not defined"); 820 } 821 } 822 823 // Temporary symbols like the ones for directional jumps don't go in the 824 // symbol table. They also need to be diagnosed in all (final) cases. 825 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 826 if (std::get<2>(LocSym)->isUndefined()) { 827 // Reset the state of any "# line file" directives we've seen to the 828 // context as it was at the diagnostic site. 829 CppHashInfo = std::get<1>(LocSym); 830 printError(std::get<0>(LocSym), "directional label undefined"); 831 } 832 } 833 } 834 835 // Finalize the output stream if there are no errors and if the client wants 836 // us to. 837 if (!HadError && !NoFinalize) 838 Out.Finish(); 839 840 return HadError || getContext().hadError(); 841 } 842 843 bool AsmParser::checkForValidSection() { 844 if (!ParsingInlineAsm && !getStreamer().getCurrentSectionOnly()) { 845 Out.InitSections(false); 846 return Error(getTok().getLoc(), 847 "expected section directive before assembly directive"); 848 } 849 return false; 850 } 851 852 /// \brief Throw away the rest of the line for testing purposes. 853 void AsmParser::eatToEndOfStatement() { 854 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 855 Lexer.Lex(); 856 857 // Eat EOL. 858 if (Lexer.is(AsmToken::EndOfStatement)) 859 Lexer.Lex(); 860 } 861 862 StringRef AsmParser::parseStringToEndOfStatement() { 863 const char *Start = getTok().getLoc().getPointer(); 864 865 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 866 Lexer.Lex(); 867 868 const char *End = getTok().getLoc().getPointer(); 869 return StringRef(Start, End - Start); 870 } 871 872 StringRef AsmParser::parseStringToComma() { 873 const char *Start = getTok().getLoc().getPointer(); 874 875 while (Lexer.isNot(AsmToken::EndOfStatement) && 876 Lexer.isNot(AsmToken::Comma) && Lexer.isNot(AsmToken::Eof)) 877 Lexer.Lex(); 878 879 const char *End = getTok().getLoc().getPointer(); 880 return StringRef(Start, End - Start); 881 } 882 883 /// \brief Parse a paren expression and return it. 884 /// NOTE: This assumes the leading '(' has already been consumed. 885 /// 886 /// parenexpr ::= expr) 887 /// 888 bool AsmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 889 if (parseExpression(Res)) 890 return true; 891 if (Lexer.isNot(AsmToken::RParen)) 892 return TokError("expected ')' in parentheses expression"); 893 EndLoc = Lexer.getTok().getEndLoc(); 894 Lex(); 895 return false; 896 } 897 898 /// \brief Parse a bracket expression and return it. 899 /// NOTE: This assumes the leading '[' has already been consumed. 900 /// 901 /// bracketexpr ::= expr] 902 /// 903 bool AsmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 904 if (parseExpression(Res)) 905 return true; 906 EndLoc = getTok().getEndLoc(); 907 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 908 return true; 909 return false; 910 } 911 912 /// \brief Parse a primary expression and return it. 913 /// primaryexpr ::= (parenexpr 914 /// primaryexpr ::= symbol 915 /// primaryexpr ::= number 916 /// primaryexpr ::= '.' 917 /// primaryexpr ::= ~,+,- primaryexpr 918 bool AsmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 919 SMLoc FirstTokenLoc = getLexer().getLoc(); 920 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 921 switch (FirstTokenKind) { 922 default: 923 return TokError("unknown token in expression"); 924 // If we have an error assume that we've already handled it. 925 case AsmToken::Error: 926 return true; 927 case AsmToken::Exclaim: 928 Lex(); // Eat the operator. 929 if (parsePrimaryExpr(Res, EndLoc)) 930 return true; 931 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 932 return false; 933 case AsmToken::Dollar: 934 case AsmToken::At: 935 case AsmToken::String: 936 case AsmToken::Identifier: { 937 StringRef Identifier; 938 if (parseIdentifier(Identifier)) { 939 // We may have failed but $ may be a valid token. 940 if (getTok().is(AsmToken::Dollar)) { 941 if (Lexer.getMAI().getDollarIsPC()) { 942 Lex(); 943 // This is a '$' reference, which references the current PC. Emit a 944 // temporary label to the streamer and refer to it. 945 MCSymbol *Sym = Ctx.createTempSymbol(); 946 Out.EmitLabel(Sym); 947 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 948 getContext()); 949 EndLoc = FirstTokenLoc; 950 return false; 951 } 952 return Error(FirstTokenLoc, "invalid token in expression"); 953 } 954 } 955 // Parse symbol variant 956 std::pair<StringRef, StringRef> Split; 957 if (!MAI.useParensForSymbolVariant()) { 958 if (FirstTokenKind == AsmToken::String) { 959 if (Lexer.is(AsmToken::At)) { 960 Lex(); // eat @ 961 SMLoc AtLoc = getLexer().getLoc(); 962 StringRef VName; 963 if (parseIdentifier(VName)) 964 return Error(AtLoc, "expected symbol variant after '@'"); 965 966 Split = std::make_pair(Identifier, VName); 967 } 968 } else { 969 Split = Identifier.split('@'); 970 } 971 } else if (Lexer.is(AsmToken::LParen)) { 972 Lex(); // eat '('. 973 StringRef VName; 974 parseIdentifier(VName); 975 // eat ')'. 976 if (parseToken(AsmToken::RParen, 977 "unexpected token in variant, expected ')'")) 978 return true; 979 Split = std::make_pair(Identifier, VName); 980 } 981 982 EndLoc = SMLoc::getFromPointer(Identifier.end()); 983 984 // This is a symbol reference. 985 StringRef SymbolName = Identifier; 986 if (SymbolName.empty()) 987 return true; 988 989 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 990 991 // Lookup the symbol variant if used. 992 if (!Split.second.empty()) { 993 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 994 if (Variant != MCSymbolRefExpr::VK_Invalid) { 995 SymbolName = Split.first; 996 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 997 Variant = MCSymbolRefExpr::VK_None; 998 } else { 999 return Error(SMLoc::getFromPointer(Split.second.begin()), 1000 "invalid variant '" + Split.second + "'"); 1001 } 1002 } 1003 1004 MCSymbol *Sym = getContext().getOrCreateSymbol(SymbolName); 1005 1006 // If this is an absolute variable reference, substitute it now to preserve 1007 // semantics in the face of reassignment. 1008 if (Sym->isVariable() && 1009 isa<MCConstantExpr>(Sym->getVariableValue(/*SetUsed*/ false))) { 1010 if (Variant) 1011 return Error(EndLoc, "unexpected modifier on variable reference"); 1012 1013 Res = Sym->getVariableValue(/*SetUsed*/ false); 1014 return false; 1015 } 1016 1017 // Otherwise create a symbol ref. 1018 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1019 return false; 1020 } 1021 case AsmToken::BigNum: 1022 return TokError("literal value out of range for directive"); 1023 case AsmToken::Integer: { 1024 SMLoc Loc = getTok().getLoc(); 1025 int64_t IntVal = getTok().getIntVal(); 1026 Res = MCConstantExpr::create(IntVal, getContext()); 1027 EndLoc = Lexer.getTok().getEndLoc(); 1028 Lex(); // Eat token. 1029 // Look for 'b' or 'f' following an Integer as a directional label 1030 if (Lexer.getKind() == AsmToken::Identifier) { 1031 StringRef IDVal = getTok().getString(); 1032 // Lookup the symbol variant if used. 1033 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1034 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1035 if (Split.first.size() != IDVal.size()) { 1036 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1037 if (Variant == MCSymbolRefExpr::VK_Invalid) 1038 return TokError("invalid variant '" + Split.second + "'"); 1039 IDVal = Split.first; 1040 } 1041 if (IDVal == "f" || IDVal == "b") { 1042 MCSymbol *Sym = 1043 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1044 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1045 if (IDVal == "b" && Sym->isUndefined()) 1046 return Error(Loc, "directional label undefined"); 1047 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1048 EndLoc = Lexer.getTok().getEndLoc(); 1049 Lex(); // Eat identifier. 1050 } 1051 } 1052 return false; 1053 } 1054 case AsmToken::Real: { 1055 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1056 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1057 Res = MCConstantExpr::create(IntVal, getContext()); 1058 EndLoc = Lexer.getTok().getEndLoc(); 1059 Lex(); // Eat token. 1060 return false; 1061 } 1062 case AsmToken::Dot: { 1063 // This is a '.' reference, which references the current PC. Emit a 1064 // temporary label to the streamer and refer to it. 1065 MCSymbol *Sym = Ctx.createTempSymbol(); 1066 Out.EmitLabel(Sym); 1067 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1068 EndLoc = Lexer.getTok().getEndLoc(); 1069 Lex(); // Eat identifier. 1070 return false; 1071 } 1072 case AsmToken::LParen: 1073 Lex(); // Eat the '('. 1074 return parseParenExpr(Res, EndLoc); 1075 case AsmToken::LBrac: 1076 if (!PlatformParser->HasBracketExpressions()) 1077 return TokError("brackets expression not supported on this target"); 1078 Lex(); // Eat the '['. 1079 return parseBracketExpr(Res, EndLoc); 1080 case AsmToken::Minus: 1081 Lex(); // Eat the operator. 1082 if (parsePrimaryExpr(Res, EndLoc)) 1083 return true; 1084 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1085 return false; 1086 case AsmToken::Plus: 1087 Lex(); // Eat the operator. 1088 if (parsePrimaryExpr(Res, EndLoc)) 1089 return true; 1090 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1091 return false; 1092 case AsmToken::Tilde: 1093 Lex(); // Eat the operator. 1094 if (parsePrimaryExpr(Res, EndLoc)) 1095 return true; 1096 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1097 return false; 1098 // MIPS unary expression operators. The lexer won't generate these tokens if 1099 // MCAsmInfo::HasMipsExpressions is false for the target. 1100 case AsmToken::PercentCall16: 1101 case AsmToken::PercentCall_Hi: 1102 case AsmToken::PercentCall_Lo: 1103 case AsmToken::PercentDtprel_Hi: 1104 case AsmToken::PercentDtprel_Lo: 1105 case AsmToken::PercentGot: 1106 case AsmToken::PercentGot_Disp: 1107 case AsmToken::PercentGot_Hi: 1108 case AsmToken::PercentGot_Lo: 1109 case AsmToken::PercentGot_Ofst: 1110 case AsmToken::PercentGot_Page: 1111 case AsmToken::PercentGottprel: 1112 case AsmToken::PercentGp_Rel: 1113 case AsmToken::PercentHi: 1114 case AsmToken::PercentHigher: 1115 case AsmToken::PercentHighest: 1116 case AsmToken::PercentLo: 1117 case AsmToken::PercentNeg: 1118 case AsmToken::PercentPcrel_Hi: 1119 case AsmToken::PercentPcrel_Lo: 1120 case AsmToken::PercentTlsgd: 1121 case AsmToken::PercentTlsldm: 1122 case AsmToken::PercentTprel_Hi: 1123 case AsmToken::PercentTprel_Lo: 1124 Lex(); // Eat the operator. 1125 if (Lexer.isNot(AsmToken::LParen)) 1126 return TokError("expected '(' after operator"); 1127 Lex(); // Eat the operator. 1128 if (parseExpression(Res, EndLoc)) 1129 return true; 1130 if (Lexer.isNot(AsmToken::RParen)) 1131 return TokError("expected ')'"); 1132 Lex(); // Eat the operator. 1133 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1134 return !Res; 1135 } 1136 } 1137 1138 bool AsmParser::parseExpression(const MCExpr *&Res) { 1139 SMLoc EndLoc; 1140 return parseExpression(Res, EndLoc); 1141 } 1142 1143 const MCExpr * 1144 AsmParser::applyModifierToExpr(const MCExpr *E, 1145 MCSymbolRefExpr::VariantKind Variant) { 1146 // Ask the target implementation about this expression first. 1147 const MCExpr *NewE = getTargetParser().applyModifierToExpr(E, Variant, Ctx); 1148 if (NewE) 1149 return NewE; 1150 // Recurse over the given expression, rebuilding it to apply the given variant 1151 // if there is exactly one symbol. 1152 switch (E->getKind()) { 1153 case MCExpr::Target: 1154 case MCExpr::Constant: 1155 return nullptr; 1156 1157 case MCExpr::SymbolRef: { 1158 const MCSymbolRefExpr *SRE = cast<MCSymbolRefExpr>(E); 1159 1160 if (SRE->getKind() != MCSymbolRefExpr::VK_None) { 1161 TokError("invalid variant on expression '" + getTok().getIdentifier() + 1162 "' (already modified)"); 1163 return E; 1164 } 1165 1166 return MCSymbolRefExpr::create(&SRE->getSymbol(), Variant, getContext()); 1167 } 1168 1169 case MCExpr::Unary: { 1170 const MCUnaryExpr *UE = cast<MCUnaryExpr>(E); 1171 const MCExpr *Sub = applyModifierToExpr(UE->getSubExpr(), Variant); 1172 if (!Sub) 1173 return nullptr; 1174 return MCUnaryExpr::create(UE->getOpcode(), Sub, getContext()); 1175 } 1176 1177 case MCExpr::Binary: { 1178 const MCBinaryExpr *BE = cast<MCBinaryExpr>(E); 1179 const MCExpr *LHS = applyModifierToExpr(BE->getLHS(), Variant); 1180 const MCExpr *RHS = applyModifierToExpr(BE->getRHS(), Variant); 1181 1182 if (!LHS && !RHS) 1183 return nullptr; 1184 1185 if (!LHS) 1186 LHS = BE->getLHS(); 1187 if (!RHS) 1188 RHS = BE->getRHS(); 1189 1190 return MCBinaryExpr::create(BE->getOpcode(), LHS, RHS, getContext()); 1191 } 1192 } 1193 1194 llvm_unreachable("Invalid expression kind!"); 1195 } 1196 1197 /// This function checks if the next token is <string> type or arithmetic. 1198 /// string that begin with character '<' must end with character '>'. 1199 /// otherwise it is arithmetics. 1200 /// If the function returns a 'true' value, 1201 /// the End argument will be filled with the last location pointed to the '>' 1202 /// character. 1203 1204 /// There is a gap between the AltMacro's documentation and the single quote implementation. 1205 /// GCC does not fully support this feature and so we will not support it. 1206 /// TODO: Adding single quote as a string. 1207 bool AsmParser::isAltmacroString(SMLoc &StrLoc, SMLoc &EndLoc) { 1208 assert((StrLoc.getPointer() != NULL) && 1209 "Argument to the function cannot be a NULL value"); 1210 const char *CharPtr = StrLoc.getPointer(); 1211 while ((*CharPtr != '>') && (*CharPtr != '\n') && 1212 (*CharPtr != '\r') && (*CharPtr != '\0')){ 1213 if(*CharPtr == '!') 1214 CharPtr++; 1215 CharPtr++; 1216 } 1217 if (*CharPtr == '>') { 1218 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1219 return true; 1220 } 1221 return false; 1222 } 1223 1224 /// \brief creating a string without the escape characters '!'. 1225 void AsmParser::altMacroString(StringRef AltMacroStr,std::string &Res) { 1226 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1227 if (AltMacroStr[Pos] == '!') 1228 Pos++; 1229 Res += AltMacroStr[Pos]; 1230 } 1231 } 1232 1233 /// \brief Parse an expression and return it. 1234 /// 1235 /// expr ::= expr &&,|| expr -> lowest. 1236 /// expr ::= expr |,^,&,! expr 1237 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1238 /// expr ::= expr <<,>> expr 1239 /// expr ::= expr +,- expr 1240 /// expr ::= expr *,/,% expr -> highest. 1241 /// expr ::= primaryexpr 1242 /// 1243 bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1244 // Parse the expression. 1245 Res = nullptr; 1246 if (parsePrimaryExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc)) 1247 return true; 1248 1249 // As a special case, we support 'a op b @ modifier' by rewriting the 1250 // expression to include the modifier. This is inefficient, but in general we 1251 // expect users to use 'a@modifier op b'. 1252 if (Lexer.getKind() == AsmToken::At) { 1253 Lex(); 1254 1255 if (Lexer.isNot(AsmToken::Identifier)) 1256 return TokError("unexpected symbol modifier following '@'"); 1257 1258 MCSymbolRefExpr::VariantKind Variant = 1259 MCSymbolRefExpr::getVariantKindForName(getTok().getIdentifier()); 1260 if (Variant == MCSymbolRefExpr::VK_Invalid) 1261 return TokError("invalid variant '" + getTok().getIdentifier() + "'"); 1262 1263 const MCExpr *ModifiedRes = applyModifierToExpr(Res, Variant); 1264 if (!ModifiedRes) { 1265 return TokError("invalid modifier '" + getTok().getIdentifier() + 1266 "' (no symbols present)"); 1267 } 1268 1269 Res = ModifiedRes; 1270 Lex(); 1271 } 1272 1273 // Try to constant fold it up front, if possible. 1274 int64_t Value; 1275 if (Res->evaluateAsAbsolute(Value)) 1276 Res = MCConstantExpr::create(Value, getContext()); 1277 1278 return false; 1279 } 1280 1281 bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1282 Res = nullptr; 1283 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1284 } 1285 1286 bool AsmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1287 SMLoc &EndLoc) { 1288 if (parseParenExpr(Res, EndLoc)) 1289 return true; 1290 1291 for (; ParenDepth > 0; --ParenDepth) { 1292 if (parseBinOpRHS(1, Res, EndLoc)) 1293 return true; 1294 1295 // We don't Lex() the last RParen. 1296 // This is the same behavior as parseParenExpression(). 1297 if (ParenDepth - 1 > 0) { 1298 EndLoc = getTok().getEndLoc(); 1299 if (parseToken(AsmToken::RParen, 1300 "expected ')' in parentheses expression")) 1301 return true; 1302 } 1303 } 1304 return false; 1305 } 1306 1307 bool AsmParser::parseAbsoluteExpression(int64_t &Res) { 1308 const MCExpr *Expr; 1309 1310 SMLoc StartLoc = Lexer.getLoc(); 1311 if (parseExpression(Expr)) 1312 return true; 1313 1314 if (!Expr->evaluateAsAbsolute(Res)) 1315 return Error(StartLoc, "expected absolute expression"); 1316 1317 return false; 1318 } 1319 1320 static unsigned getDarwinBinOpPrecedence(AsmToken::TokenKind K, 1321 MCBinaryExpr::Opcode &Kind, 1322 bool ShouldUseLogicalShr) { 1323 switch (K) { 1324 default: 1325 return 0; // not a binop. 1326 1327 // Lowest Precedence: &&, || 1328 case AsmToken::AmpAmp: 1329 Kind = MCBinaryExpr::LAnd; 1330 return 1; 1331 case AsmToken::PipePipe: 1332 Kind = MCBinaryExpr::LOr; 1333 return 1; 1334 1335 // Low Precedence: |, &, ^ 1336 // 1337 // FIXME: gas seems to support '!' as an infix operator? 1338 case AsmToken::Pipe: 1339 Kind = MCBinaryExpr::Or; 1340 return 2; 1341 case AsmToken::Caret: 1342 Kind = MCBinaryExpr::Xor; 1343 return 2; 1344 case AsmToken::Amp: 1345 Kind = MCBinaryExpr::And; 1346 return 2; 1347 1348 // Low Intermediate Precedence: ==, !=, <>, <, <=, >, >= 1349 case AsmToken::EqualEqual: 1350 Kind = MCBinaryExpr::EQ; 1351 return 3; 1352 case AsmToken::ExclaimEqual: 1353 case AsmToken::LessGreater: 1354 Kind = MCBinaryExpr::NE; 1355 return 3; 1356 case AsmToken::Less: 1357 Kind = MCBinaryExpr::LT; 1358 return 3; 1359 case AsmToken::LessEqual: 1360 Kind = MCBinaryExpr::LTE; 1361 return 3; 1362 case AsmToken::Greater: 1363 Kind = MCBinaryExpr::GT; 1364 return 3; 1365 case AsmToken::GreaterEqual: 1366 Kind = MCBinaryExpr::GTE; 1367 return 3; 1368 1369 // Intermediate Precedence: <<, >> 1370 case AsmToken::LessLess: 1371 Kind = MCBinaryExpr::Shl; 1372 return 4; 1373 case AsmToken::GreaterGreater: 1374 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1375 return 4; 1376 1377 // High Intermediate Precedence: +, - 1378 case AsmToken::Plus: 1379 Kind = MCBinaryExpr::Add; 1380 return 5; 1381 case AsmToken::Minus: 1382 Kind = MCBinaryExpr::Sub; 1383 return 5; 1384 1385 // Highest Precedence: *, /, % 1386 case AsmToken::Star: 1387 Kind = MCBinaryExpr::Mul; 1388 return 6; 1389 case AsmToken::Slash: 1390 Kind = MCBinaryExpr::Div; 1391 return 6; 1392 case AsmToken::Percent: 1393 Kind = MCBinaryExpr::Mod; 1394 return 6; 1395 } 1396 } 1397 1398 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1399 MCBinaryExpr::Opcode &Kind, 1400 bool ShouldUseLogicalShr) { 1401 switch (K) { 1402 default: 1403 return 0; // not a binop. 1404 1405 // Lowest Precedence: &&, || 1406 case AsmToken::AmpAmp: 1407 Kind = MCBinaryExpr::LAnd; 1408 return 2; 1409 case AsmToken::PipePipe: 1410 Kind = MCBinaryExpr::LOr; 1411 return 1; 1412 1413 // Low Precedence: ==, !=, <>, <, <=, >, >= 1414 case AsmToken::EqualEqual: 1415 Kind = MCBinaryExpr::EQ; 1416 return 3; 1417 case AsmToken::ExclaimEqual: 1418 case AsmToken::LessGreater: 1419 Kind = MCBinaryExpr::NE; 1420 return 3; 1421 case AsmToken::Less: 1422 Kind = MCBinaryExpr::LT; 1423 return 3; 1424 case AsmToken::LessEqual: 1425 Kind = MCBinaryExpr::LTE; 1426 return 3; 1427 case AsmToken::Greater: 1428 Kind = MCBinaryExpr::GT; 1429 return 3; 1430 case AsmToken::GreaterEqual: 1431 Kind = MCBinaryExpr::GTE; 1432 return 3; 1433 1434 // Low Intermediate Precedence: +, - 1435 case AsmToken::Plus: 1436 Kind = MCBinaryExpr::Add; 1437 return 4; 1438 case AsmToken::Minus: 1439 Kind = MCBinaryExpr::Sub; 1440 return 4; 1441 1442 // High Intermediate Precedence: |, &, ^ 1443 // 1444 // FIXME: gas seems to support '!' as an infix operator? 1445 case AsmToken::Pipe: 1446 Kind = MCBinaryExpr::Or; 1447 return 5; 1448 case AsmToken::Caret: 1449 Kind = MCBinaryExpr::Xor; 1450 return 5; 1451 case AsmToken::Amp: 1452 Kind = MCBinaryExpr::And; 1453 return 5; 1454 1455 // Highest Precedence: *, /, %, <<, >> 1456 case AsmToken::Star: 1457 Kind = MCBinaryExpr::Mul; 1458 return 6; 1459 case AsmToken::Slash: 1460 Kind = MCBinaryExpr::Div; 1461 return 6; 1462 case AsmToken::Percent: 1463 Kind = MCBinaryExpr::Mod; 1464 return 6; 1465 case AsmToken::LessLess: 1466 Kind = MCBinaryExpr::Shl; 1467 return 6; 1468 case AsmToken::GreaterGreater: 1469 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1470 return 6; 1471 } 1472 } 1473 1474 unsigned AsmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1475 MCBinaryExpr::Opcode &Kind) { 1476 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1477 return IsDarwin ? getDarwinBinOpPrecedence(K, Kind, ShouldUseLogicalShr) 1478 : getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1479 } 1480 1481 /// \brief Parse all binary operators with precedence >= 'Precedence'. 1482 /// Res contains the LHS of the expression on input. 1483 bool AsmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1484 SMLoc &EndLoc) { 1485 SMLoc StartLoc = Lexer.getLoc(); 1486 while (true) { 1487 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1488 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1489 1490 // If the next token is lower precedence than we are allowed to eat, return 1491 // successfully with what we ate already. 1492 if (TokPrec < Precedence) 1493 return false; 1494 1495 Lex(); 1496 1497 // Eat the next primary expression. 1498 const MCExpr *RHS; 1499 if (parsePrimaryExpr(RHS, EndLoc)) 1500 return true; 1501 1502 // If BinOp binds less tightly with RHS than the operator after RHS, let 1503 // the pending operator take RHS as its LHS. 1504 MCBinaryExpr::Opcode Dummy; 1505 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1506 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1507 return true; 1508 1509 // Merge LHS and RHS according to operator. 1510 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1511 } 1512 } 1513 1514 /// ParseStatement: 1515 /// ::= EndOfStatement 1516 /// ::= Label* Directive ...Operands... EndOfStatement 1517 /// ::= Label* Identifier OperandList* EndOfStatement 1518 bool AsmParser::parseStatement(ParseStatementInfo &Info, 1519 MCAsmParserSemaCallback *SI) { 1520 assert(!hasPendingError() && "parseStatement started with pending error"); 1521 // Eat initial spaces and comments 1522 while (Lexer.is(AsmToken::Space)) 1523 Lex(); 1524 if (Lexer.is(AsmToken::EndOfStatement)) { 1525 // if this is a line comment we can drop it safely 1526 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1527 getTok().getString().front() == '\n') 1528 Out.AddBlankLine(); 1529 Lex(); 1530 return false; 1531 } 1532 if (Lexer.is(AsmToken::Hash)) { 1533 // Seeing a hash here means that it was an end-of-line comment in 1534 // an asm syntax where hash's are not comment and the previous 1535 // statement parser did not check the end of statement. Relex as 1536 // EndOfStatement. 1537 StringRef CommentStr = parseStringToEndOfStatement(); 1538 Lexer.Lex(); 1539 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1540 return false; 1541 } 1542 // Statements always start with an identifier. 1543 AsmToken ID = getTok(); 1544 SMLoc IDLoc = ID.getLoc(); 1545 StringRef IDVal; 1546 int64_t LocalLabelVal = -1; 1547 if (Lexer.is(AsmToken::HashDirective)) 1548 return parseCppHashLineFilenameComment(IDLoc); 1549 // Allow an integer followed by a ':' as a directional local label. 1550 if (Lexer.is(AsmToken::Integer)) { 1551 LocalLabelVal = getTok().getIntVal(); 1552 if (LocalLabelVal < 0) { 1553 if (!TheCondState.Ignore) { 1554 Lex(); // always eat a token 1555 return Error(IDLoc, "unexpected token at start of statement"); 1556 } 1557 IDVal = ""; 1558 } else { 1559 IDVal = getTok().getString(); 1560 Lex(); // Consume the integer token to be used as an identifier token. 1561 if (Lexer.getKind() != AsmToken::Colon) { 1562 if (!TheCondState.Ignore) { 1563 Lex(); // always eat a token 1564 return Error(IDLoc, "unexpected token at start of statement"); 1565 } 1566 } 1567 } 1568 } else if (Lexer.is(AsmToken::Dot)) { 1569 // Treat '.' as a valid identifier in this context. 1570 Lex(); 1571 IDVal = "."; 1572 } else if (Lexer.is(AsmToken::LCurly)) { 1573 // Treat '{' as a valid identifier in this context. 1574 Lex(); 1575 IDVal = "{"; 1576 1577 } else if (Lexer.is(AsmToken::RCurly)) { 1578 // Treat '}' as a valid identifier in this context. 1579 Lex(); 1580 IDVal = "}"; 1581 } else if (parseIdentifier(IDVal)) { 1582 if (!TheCondState.Ignore) { 1583 Lex(); // always eat a token 1584 return Error(IDLoc, "unexpected token at start of statement"); 1585 } 1586 IDVal = ""; 1587 } 1588 1589 // Handle conditional assembly here before checking for skipping. We 1590 // have to do this so that .endif isn't skipped in a ".if 0" block for 1591 // example. 1592 StringMap<DirectiveKind>::const_iterator DirKindIt = 1593 DirectiveKindMap.find(IDVal); 1594 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1595 ? DK_NO_DIRECTIVE 1596 : DirKindIt->getValue(); 1597 switch (DirKind) { 1598 default: 1599 break; 1600 case DK_IF: 1601 case DK_IFEQ: 1602 case DK_IFGE: 1603 case DK_IFGT: 1604 case DK_IFLE: 1605 case DK_IFLT: 1606 case DK_IFNE: 1607 return parseDirectiveIf(IDLoc, DirKind); 1608 case DK_IFB: 1609 return parseDirectiveIfb(IDLoc, true); 1610 case DK_IFNB: 1611 return parseDirectiveIfb(IDLoc, false); 1612 case DK_IFC: 1613 return parseDirectiveIfc(IDLoc, true); 1614 case DK_IFEQS: 1615 return parseDirectiveIfeqs(IDLoc, true); 1616 case DK_IFNC: 1617 return parseDirectiveIfc(IDLoc, false); 1618 case DK_IFNES: 1619 return parseDirectiveIfeqs(IDLoc, false); 1620 case DK_IFDEF: 1621 return parseDirectiveIfdef(IDLoc, true); 1622 case DK_IFNDEF: 1623 case DK_IFNOTDEF: 1624 return parseDirectiveIfdef(IDLoc, false); 1625 case DK_ELSEIF: 1626 return parseDirectiveElseIf(IDLoc); 1627 case DK_ELSE: 1628 return parseDirectiveElse(IDLoc); 1629 case DK_ENDIF: 1630 return parseDirectiveEndIf(IDLoc); 1631 } 1632 1633 // Ignore the statement if in the middle of inactive conditional 1634 // (e.g. ".if 0"). 1635 if (TheCondState.Ignore) { 1636 eatToEndOfStatement(); 1637 return false; 1638 } 1639 1640 // FIXME: Recurse on local labels? 1641 1642 // See what kind of statement we have. 1643 switch (Lexer.getKind()) { 1644 case AsmToken::Colon: { 1645 if (!getTargetParser().isLabel(ID)) 1646 break; 1647 if (checkForValidSection()) 1648 return true; 1649 1650 // identifier ':' -> Label. 1651 Lex(); 1652 1653 // Diagnose attempt to use '.' as a label. 1654 if (IDVal == ".") 1655 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1656 1657 // Diagnose attempt to use a variable as a label. 1658 // 1659 // FIXME: Diagnostics. Note the location of the definition as a label. 1660 // FIXME: This doesn't diagnose assignment to a symbol which has been 1661 // implicitly marked as external. 1662 MCSymbol *Sym; 1663 if (LocalLabelVal == -1) { 1664 if (ParsingInlineAsm && SI) { 1665 StringRef RewrittenLabel = 1666 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1667 assert(!RewrittenLabel.empty() && 1668 "We should have an internal name here."); 1669 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1670 RewrittenLabel); 1671 IDVal = RewrittenLabel; 1672 } 1673 Sym = getContext().getOrCreateSymbol(IDVal); 1674 } else 1675 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1676 // End of Labels should be treated as end of line for lexing 1677 // purposes but that information is not available to the Lexer who 1678 // does not understand Labels. This may cause us to see a Hash 1679 // here instead of a preprocessor line comment. 1680 if (getTok().is(AsmToken::Hash)) { 1681 StringRef CommentStr = parseStringToEndOfStatement(); 1682 Lexer.Lex(); 1683 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1684 } 1685 1686 // Consume any end of statement token, if present, to avoid spurious 1687 // AddBlankLine calls(). 1688 if (getTok().is(AsmToken::EndOfStatement)) { 1689 Lex(); 1690 } 1691 1692 // Emit the label. 1693 if (!getTargetParser().isParsingInlineAsm()) 1694 Out.EmitLabel(Sym, IDLoc); 1695 1696 // If we are generating dwarf for assembly source files then gather the 1697 // info to make a dwarf label entry for this label if needed. 1698 if (getContext().getGenDwarfForAssembly()) 1699 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1700 IDLoc); 1701 1702 getTargetParser().onLabelParsed(Sym); 1703 1704 return false; 1705 } 1706 1707 case AsmToken::Equal: 1708 if (!getTargetParser().equalIsAsmAssignment()) 1709 break; 1710 // identifier '=' ... -> assignment statement 1711 Lex(); 1712 1713 return parseAssignment(IDVal, true); 1714 1715 default: // Normal instruction or directive. 1716 break; 1717 } 1718 1719 // If macros are enabled, check to see if this is a macro instantiation. 1720 if (areMacrosEnabled()) 1721 if (const MCAsmMacro *M = lookupMacro(IDVal)) { 1722 return handleMacroEntry(M, IDLoc); 1723 } 1724 1725 // Otherwise, we have a normal instruction or directive. 1726 1727 // Directives start with "." 1728 if (IDVal[0] == '.' && IDVal != ".") { 1729 // There are several entities interested in parsing directives: 1730 // 1731 // 1. The target-specific assembly parser. Some directives are target 1732 // specific or may potentially behave differently on certain targets. 1733 // 2. Asm parser extensions. For example, platform-specific parsers 1734 // (like the ELF parser) register themselves as extensions. 1735 // 3. The generic directive parser implemented by this class. These are 1736 // all the directives that behave in a target and platform independent 1737 // manner, or at least have a default behavior that's shared between 1738 // all targets and platforms. 1739 1740 getTargetParser().flushPendingInstructions(getStreamer()); 1741 1742 SMLoc StartTokLoc = getTok().getLoc(); 1743 bool TPDirectiveReturn = getTargetParser().ParseDirective(ID); 1744 1745 if (hasPendingError()) 1746 return true; 1747 // Currently the return value should be true if we are 1748 // uninterested but as this is at odds with the standard parsing 1749 // convention (return true = error) we have instances of a parsed 1750 // directive that fails returning true as an error. Catch these 1751 // cases as best as possible errors here. 1752 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1753 return true; 1754 // Return if we did some parsing or believe we succeeded. 1755 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1756 return false; 1757 1758 // Next, check the extension directive map to see if any extension has 1759 // registered itself to parse this directive. 1760 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1761 ExtensionDirectiveMap.lookup(IDVal); 1762 if (Handler.first) 1763 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1764 1765 // Finally, if no one else is interested in this directive, it must be 1766 // generic and familiar to this class. 1767 switch (DirKind) { 1768 default: 1769 break; 1770 case DK_SET: 1771 case DK_EQU: 1772 return parseDirectiveSet(IDVal, true); 1773 case DK_EQUIV: 1774 return parseDirectiveSet(IDVal, false); 1775 case DK_ASCII: 1776 return parseDirectiveAscii(IDVal, false); 1777 case DK_ASCIZ: 1778 case DK_STRING: 1779 return parseDirectiveAscii(IDVal, true); 1780 case DK_BYTE: 1781 case DK_DC_B: 1782 return parseDirectiveValue(IDVal, 1); 1783 case DK_DC: 1784 case DK_DC_W: 1785 case DK_SHORT: 1786 case DK_VALUE: 1787 case DK_2BYTE: 1788 return parseDirectiveValue(IDVal, 2); 1789 case DK_LONG: 1790 case DK_INT: 1791 case DK_4BYTE: 1792 case DK_DC_L: 1793 return parseDirectiveValue(IDVal, 4); 1794 case DK_QUAD: 1795 case DK_8BYTE: 1796 return parseDirectiveValue(IDVal, 8); 1797 case DK_DC_A: 1798 return parseDirectiveValue( 1799 IDVal, getContext().getAsmInfo()->getCodePointerSize()); 1800 case DK_OCTA: 1801 return parseDirectiveOctaValue(IDVal); 1802 case DK_SINGLE: 1803 case DK_FLOAT: 1804 case DK_DC_S: 1805 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1806 case DK_DOUBLE: 1807 case DK_DC_D: 1808 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1809 case DK_ALIGN: { 1810 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1811 return parseDirectiveAlign(IsPow2, /*ExprSize=*/1); 1812 } 1813 case DK_ALIGN32: { 1814 bool IsPow2 = !getContext().getAsmInfo()->getAlignmentIsInBytes(); 1815 return parseDirectiveAlign(IsPow2, /*ExprSize=*/4); 1816 } 1817 case DK_BALIGN: 1818 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); 1819 case DK_BALIGNW: 1820 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); 1821 case DK_BALIGNL: 1822 return parseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); 1823 case DK_P2ALIGN: 1824 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); 1825 case DK_P2ALIGNW: 1826 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); 1827 case DK_P2ALIGNL: 1828 return parseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); 1829 case DK_ORG: 1830 return parseDirectiveOrg(); 1831 case DK_FILL: 1832 return parseDirectiveFill(); 1833 case DK_ZERO: 1834 return parseDirectiveZero(); 1835 case DK_EXTERN: 1836 eatToEndOfStatement(); // .extern is the default, ignore it. 1837 return false; 1838 case DK_GLOBL: 1839 case DK_GLOBAL: 1840 return parseDirectiveSymbolAttribute(MCSA_Global); 1841 case DK_LAZY_REFERENCE: 1842 return parseDirectiveSymbolAttribute(MCSA_LazyReference); 1843 case DK_NO_DEAD_STRIP: 1844 return parseDirectiveSymbolAttribute(MCSA_NoDeadStrip); 1845 case DK_SYMBOL_RESOLVER: 1846 return parseDirectiveSymbolAttribute(MCSA_SymbolResolver); 1847 case DK_PRIVATE_EXTERN: 1848 return parseDirectiveSymbolAttribute(MCSA_PrivateExtern); 1849 case DK_REFERENCE: 1850 return parseDirectiveSymbolAttribute(MCSA_Reference); 1851 case DK_WEAK_DEFINITION: 1852 return parseDirectiveSymbolAttribute(MCSA_WeakDefinition); 1853 case DK_WEAK_REFERENCE: 1854 return parseDirectiveSymbolAttribute(MCSA_WeakReference); 1855 case DK_WEAK_DEF_CAN_BE_HIDDEN: 1856 return parseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); 1857 case DK_COMM: 1858 case DK_COMMON: 1859 return parseDirectiveComm(/*IsLocal=*/false); 1860 case DK_LCOMM: 1861 return parseDirectiveComm(/*IsLocal=*/true); 1862 case DK_ABORT: 1863 return parseDirectiveAbort(); 1864 case DK_INCLUDE: 1865 return parseDirectiveInclude(); 1866 case DK_INCBIN: 1867 return parseDirectiveIncbin(); 1868 case DK_CODE16: 1869 case DK_CODE16GCC: 1870 return TokError(Twine(IDVal) + 1871 " not currently supported for this target"); 1872 case DK_REPT: 1873 return parseDirectiveRept(IDLoc, IDVal); 1874 case DK_IRP: 1875 return parseDirectiveIrp(IDLoc); 1876 case DK_IRPC: 1877 return parseDirectiveIrpc(IDLoc); 1878 case DK_ENDR: 1879 return parseDirectiveEndr(IDLoc); 1880 case DK_BUNDLE_ALIGN_MODE: 1881 return parseDirectiveBundleAlignMode(); 1882 case DK_BUNDLE_LOCK: 1883 return parseDirectiveBundleLock(); 1884 case DK_BUNDLE_UNLOCK: 1885 return parseDirectiveBundleUnlock(); 1886 case DK_SLEB128: 1887 return parseDirectiveLEB128(true); 1888 case DK_ULEB128: 1889 return parseDirectiveLEB128(false); 1890 case DK_SPACE: 1891 case DK_SKIP: 1892 return parseDirectiveSpace(IDVal); 1893 case DK_FILE: 1894 return parseDirectiveFile(IDLoc); 1895 case DK_LINE: 1896 return parseDirectiveLine(); 1897 case DK_LOC: 1898 return parseDirectiveLoc(); 1899 case DK_STABS: 1900 return parseDirectiveStabs(); 1901 case DK_CV_FILE: 1902 return parseDirectiveCVFile(); 1903 case DK_CV_FUNC_ID: 1904 return parseDirectiveCVFuncId(); 1905 case DK_CV_INLINE_SITE_ID: 1906 return parseDirectiveCVInlineSiteId(); 1907 case DK_CV_LOC: 1908 return parseDirectiveCVLoc(); 1909 case DK_CV_LINETABLE: 1910 return parseDirectiveCVLinetable(); 1911 case DK_CV_INLINE_LINETABLE: 1912 return parseDirectiveCVInlineLinetable(); 1913 case DK_CV_DEF_RANGE: 1914 return parseDirectiveCVDefRange(); 1915 case DK_CV_STRINGTABLE: 1916 return parseDirectiveCVStringTable(); 1917 case DK_CV_FILECHECKSUMS: 1918 return parseDirectiveCVFileChecksums(); 1919 case DK_CFI_SECTIONS: 1920 return parseDirectiveCFISections(); 1921 case DK_CFI_STARTPROC: 1922 return parseDirectiveCFIStartProc(); 1923 case DK_CFI_ENDPROC: 1924 return parseDirectiveCFIEndProc(); 1925 case DK_CFI_DEF_CFA: 1926 return parseDirectiveCFIDefCfa(IDLoc); 1927 case DK_CFI_DEF_CFA_OFFSET: 1928 return parseDirectiveCFIDefCfaOffset(); 1929 case DK_CFI_ADJUST_CFA_OFFSET: 1930 return parseDirectiveCFIAdjustCfaOffset(); 1931 case DK_CFI_DEF_CFA_REGISTER: 1932 return parseDirectiveCFIDefCfaRegister(IDLoc); 1933 case DK_CFI_OFFSET: 1934 return parseDirectiveCFIOffset(IDLoc); 1935 case DK_CFI_REL_OFFSET: 1936 return parseDirectiveCFIRelOffset(IDLoc); 1937 case DK_CFI_PERSONALITY: 1938 return parseDirectiveCFIPersonalityOrLsda(true); 1939 case DK_CFI_LSDA: 1940 return parseDirectiveCFIPersonalityOrLsda(false); 1941 case DK_CFI_REMEMBER_STATE: 1942 return parseDirectiveCFIRememberState(); 1943 case DK_CFI_RESTORE_STATE: 1944 return parseDirectiveCFIRestoreState(); 1945 case DK_CFI_SAME_VALUE: 1946 return parseDirectiveCFISameValue(IDLoc); 1947 case DK_CFI_RESTORE: 1948 return parseDirectiveCFIRestore(IDLoc); 1949 case DK_CFI_ESCAPE: 1950 return parseDirectiveCFIEscape(); 1951 case DK_CFI_SIGNAL_FRAME: 1952 return parseDirectiveCFISignalFrame(); 1953 case DK_CFI_UNDEFINED: 1954 return parseDirectiveCFIUndefined(IDLoc); 1955 case DK_CFI_REGISTER: 1956 return parseDirectiveCFIRegister(IDLoc); 1957 case DK_CFI_WINDOW_SAVE: 1958 return parseDirectiveCFIWindowSave(); 1959 case DK_MACROS_ON: 1960 case DK_MACROS_OFF: 1961 return parseDirectiveMacrosOnOff(IDVal); 1962 case DK_MACRO: 1963 return parseDirectiveMacro(IDLoc); 1964 case DK_ALTMACRO: 1965 case DK_NOALTMACRO: 1966 return parseDirectiveAltmacro(IDVal); 1967 case DK_EXITM: 1968 return parseDirectiveExitMacro(IDVal); 1969 case DK_ENDM: 1970 case DK_ENDMACRO: 1971 return parseDirectiveEndMacro(IDVal); 1972 case DK_PURGEM: 1973 return parseDirectivePurgeMacro(IDLoc); 1974 case DK_END: 1975 return parseDirectiveEnd(IDLoc); 1976 case DK_ERR: 1977 return parseDirectiveError(IDLoc, false); 1978 case DK_ERROR: 1979 return parseDirectiveError(IDLoc, true); 1980 case DK_WARNING: 1981 return parseDirectiveWarning(IDLoc); 1982 case DK_RELOC: 1983 return parseDirectiveReloc(IDLoc); 1984 case DK_DCB: 1985 case DK_DCB_W: 1986 return parseDirectiveDCB(IDVal, 2); 1987 case DK_DCB_B: 1988 return parseDirectiveDCB(IDVal, 1); 1989 case DK_DCB_D: 1990 return parseDirectiveRealDCB(IDVal, APFloat::IEEEdouble()); 1991 case DK_DCB_L: 1992 return parseDirectiveDCB(IDVal, 4); 1993 case DK_DCB_S: 1994 return parseDirectiveRealDCB(IDVal, APFloat::IEEEsingle()); 1995 case DK_DC_X: 1996 case DK_DCB_X: 1997 return TokError(Twine(IDVal) + 1998 " not currently supported for this target"); 1999 case DK_DS: 2000 case DK_DS_W: 2001 return parseDirectiveDS(IDVal, 2); 2002 case DK_DS_B: 2003 return parseDirectiveDS(IDVal, 1); 2004 case DK_DS_D: 2005 return parseDirectiveDS(IDVal, 8); 2006 case DK_DS_L: 2007 case DK_DS_S: 2008 return parseDirectiveDS(IDVal, 4); 2009 case DK_DS_P: 2010 case DK_DS_X: 2011 return parseDirectiveDS(IDVal, 12); 2012 } 2013 2014 return Error(IDLoc, "unknown directive"); 2015 } 2016 2017 // __asm _emit or __asm __emit 2018 if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 2019 IDVal == "_EMIT" || IDVal == "__EMIT")) 2020 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 2021 2022 // __asm align 2023 if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 2024 return parseDirectiveMSAlign(IDLoc, Info); 2025 2026 if (ParsingInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 2027 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 2028 if (checkForValidSection()) 2029 return true; 2030 2031 // Canonicalize the opcode to lower case. 2032 std::string OpcodeStr = IDVal.lower(); 2033 ParseInstructionInfo IInfo(Info.AsmRewrites); 2034 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 2035 Info.ParsedOperands); 2036 Info.ParseError = ParseHadError; 2037 2038 // Dump the parsed representation, if requested. 2039 if (getShowParsedOperands()) { 2040 SmallString<256> Str; 2041 raw_svector_ostream OS(Str); 2042 OS << "parsed instruction: ["; 2043 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 2044 if (i != 0) 2045 OS << ", "; 2046 Info.ParsedOperands[i]->print(OS); 2047 } 2048 OS << "]"; 2049 2050 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 2051 } 2052 2053 // Fail even if ParseInstruction erroneously returns false. 2054 if (hasPendingError() || ParseHadError) 2055 return true; 2056 2057 // If we are generating dwarf for the current section then generate a .loc 2058 // directive for the instruction. 2059 if (!ParseHadError && getContext().getGenDwarfForAssembly() && 2060 getContext().getGenDwarfSectionSyms().count( 2061 getStreamer().getCurrentSectionOnly())) { 2062 unsigned Line; 2063 if (ActiveMacros.empty()) 2064 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 2065 else 2066 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 2067 ActiveMacros.front()->ExitBuffer); 2068 2069 // If we previously parsed a cpp hash file line comment then make sure the 2070 // current Dwarf File is for the CppHashFilename if not then emit the 2071 // Dwarf File table for it and adjust the line number for the .loc. 2072 if (!CppHashInfo.Filename.empty()) { 2073 unsigned FileNumber = getStreamer().EmitDwarfFileDirective( 2074 0, StringRef(), CppHashInfo.Filename); 2075 getContext().setGenDwarfFileNumber(FileNumber); 2076 2077 // Since SrcMgr.FindLineNumber() is slow and messes up the SourceMgr's 2078 // cache with the different Loc from the call above we save the last 2079 // info we queried here with SrcMgr.FindLineNumber(). 2080 unsigned CppHashLocLineNo; 2081 if (LastQueryIDLoc == CppHashInfo.Loc && 2082 LastQueryBuffer == CppHashInfo.Buf) 2083 CppHashLocLineNo = LastQueryLine; 2084 else { 2085 CppHashLocLineNo = 2086 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2087 LastQueryLine = CppHashLocLineNo; 2088 LastQueryIDLoc = CppHashInfo.Loc; 2089 LastQueryBuffer = CppHashInfo.Buf; 2090 } 2091 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2092 } 2093 2094 getStreamer().EmitDwarfLocDirective( 2095 getContext().getGenDwarfFileNumber(), Line, 0, 2096 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2097 StringRef()); 2098 } 2099 2100 // If parsing succeeded, match the instruction. 2101 if (!ParseHadError) { 2102 uint64_t ErrorInfo; 2103 if (getTargetParser().MatchAndEmitInstruction( 2104 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2105 getTargetParser().isParsingInlineAsm())) 2106 return true; 2107 } 2108 return false; 2109 } 2110 2111 // Parse and erase curly braces marking block start/end 2112 bool 2113 AsmParser::parseCurlyBlockScope(SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2114 // Identify curly brace marking block start/end 2115 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2116 return false; 2117 2118 SMLoc StartLoc = Lexer.getLoc(); 2119 Lex(); // Eat the brace 2120 if (Lexer.is(AsmToken::EndOfStatement)) 2121 Lex(); // Eat EndOfStatement following the brace 2122 2123 // Erase the block start/end brace from the output asm string 2124 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2125 StartLoc.getPointer()); 2126 return true; 2127 } 2128 2129 /// parseCppHashLineFilenameComment as this: 2130 /// ::= # number "filename" 2131 bool AsmParser::parseCppHashLineFilenameComment(SMLoc L) { 2132 Lex(); // Eat the hash token. 2133 // Lexer only ever emits HashDirective if it fully formed if it's 2134 // done the checking already so this is an internal error. 2135 assert(getTok().is(AsmToken::Integer) && 2136 "Lexing Cpp line comment: Expected Integer"); 2137 int64_t LineNumber = getTok().getIntVal(); 2138 Lex(); 2139 assert(getTok().is(AsmToken::String) && 2140 "Lexing Cpp line comment: Expected String"); 2141 StringRef Filename = getTok().getString(); 2142 Lex(); 2143 2144 // Get rid of the enclosing quotes. 2145 Filename = Filename.substr(1, Filename.size() - 2); 2146 2147 // Save the SMLoc, Filename and LineNumber for later use by diagnostics. 2148 CppHashInfo.Loc = L; 2149 CppHashInfo.Filename = Filename; 2150 CppHashInfo.LineNumber = LineNumber; 2151 CppHashInfo.Buf = CurBuffer; 2152 return false; 2153 } 2154 2155 /// \brief will use the last parsed cpp hash line filename comment 2156 /// for the Filename and LineNo if any in the diagnostic. 2157 void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2158 const AsmParser *Parser = static_cast<const AsmParser *>(Context); 2159 raw_ostream &OS = errs(); 2160 2161 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2162 SMLoc DiagLoc = Diag.getLoc(); 2163 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2164 unsigned CppHashBuf = 2165 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2166 2167 // Like SourceMgr::printMessage() we need to print the include stack if any 2168 // before printing the message. 2169 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2170 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2171 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2172 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2173 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2174 } 2175 2176 // If we have not parsed a cpp hash line filename comment or the source 2177 // manager changed or buffer changed (like in a nested include) then just 2178 // print the normal diagnostic using its Filename and LineNo. 2179 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2180 DiagBuf != CppHashBuf) { 2181 if (Parser->SavedDiagHandler) 2182 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2183 else 2184 Diag.print(nullptr, OS); 2185 return; 2186 } 2187 2188 // Use the CppHashFilename and calculate a line number based on the 2189 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2190 // for the diagnostic. 2191 const std::string &Filename = Parser->CppHashInfo.Filename; 2192 2193 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2194 int CppHashLocLineNo = 2195 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2196 int LineNo = 2197 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2198 2199 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2200 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2201 Diag.getLineContents(), Diag.getRanges()); 2202 2203 if (Parser->SavedDiagHandler) 2204 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2205 else 2206 NewDiag.print(nullptr, OS); 2207 } 2208 2209 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2210 // difference being that that function accepts '@' as part of identifiers and 2211 // we can't do that. AsmLexer.cpp should probably be changed to handle 2212 // '@' as a special case when needed. 2213 static bool isIdentifierChar(char c) { 2214 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2215 c == '.'; 2216 } 2217 2218 bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2219 ArrayRef<MCAsmMacroParameter> Parameters, 2220 ArrayRef<MCAsmMacroArgument> A, 2221 bool EnableAtPseudoVariable, SMLoc L) { 2222 unsigned NParameters = Parameters.size(); 2223 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2224 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2225 return Error(L, "Wrong number of arguments"); 2226 2227 // A macro without parameters is handled differently on Darwin: 2228 // gas accepts no arguments and does no substitutions 2229 while (!Body.empty()) { 2230 // Scan for the next substitution. 2231 std::size_t End = Body.size(), Pos = 0; 2232 for (; Pos != End; ++Pos) { 2233 // Check for a substitution or escape. 2234 if (IsDarwin && !NParameters) { 2235 // This macro has no parameters, look for $0, $1, etc. 2236 if (Body[Pos] != '$' || Pos + 1 == End) 2237 continue; 2238 2239 char Next = Body[Pos + 1]; 2240 if (Next == '$' || Next == 'n' || 2241 isdigit(static_cast<unsigned char>(Next))) 2242 break; 2243 } else { 2244 // This macro has parameters, look for \foo, \bar, etc. 2245 if (Body[Pos] == '\\' && Pos + 1 != End) 2246 break; 2247 } 2248 } 2249 2250 // Add the prefix. 2251 OS << Body.slice(0, Pos); 2252 2253 // Check if we reached the end. 2254 if (Pos == End) 2255 break; 2256 2257 if (IsDarwin && !NParameters) { 2258 switch (Body[Pos + 1]) { 2259 // $$ => $ 2260 case '$': 2261 OS << '$'; 2262 break; 2263 2264 // $n => number of arguments 2265 case 'n': 2266 OS << A.size(); 2267 break; 2268 2269 // $[0-9] => argument 2270 default: { 2271 // Missing arguments are ignored. 2272 unsigned Index = Body[Pos + 1] - '0'; 2273 if (Index >= A.size()) 2274 break; 2275 2276 // Otherwise substitute with the token values, with spaces eliminated. 2277 for (const AsmToken &Token : A[Index]) 2278 OS << Token.getString(); 2279 break; 2280 } 2281 } 2282 Pos += 2; 2283 } else { 2284 unsigned I = Pos + 1; 2285 2286 // Check for the \@ pseudo-variable. 2287 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2288 ++I; 2289 else 2290 while (isIdentifierChar(Body[I]) && I + 1 != End) 2291 ++I; 2292 2293 const char *Begin = Body.data() + Pos + 1; 2294 StringRef Argument(Begin, I - (Pos + 1)); 2295 unsigned Index = 0; 2296 2297 if (Argument == "@") { 2298 OS << NumOfMacroInstantiations; 2299 Pos += 2; 2300 } else { 2301 for (; Index < NParameters; ++Index) 2302 if (Parameters[Index].Name == Argument) 2303 break; 2304 2305 if (Index == NParameters) { 2306 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2307 Pos += 3; 2308 else { 2309 OS << '\\' << Argument; 2310 Pos = I; 2311 } 2312 } else { 2313 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2314 for (const AsmToken &Token : A[Index]) 2315 // For altmacro mode, you can write '%expr'. 2316 // The prefix '%' evaluates the expression 'expr' 2317 // and uses the result as a string (e.g. replace %(1+2) with the string "3"). 2318 // Here, we identify the integer token which is the result of the 2319 // absolute expression evaluation and replace it with its string representation. 2320 if ((Lexer.IsaAltMacroMode()) && 2321 (*(Token.getString().begin()) == '%') && Token.is(AsmToken::Integer)) 2322 // Emit an integer value to the buffer. 2323 OS << Token.getIntVal(); 2324 // Only Token that was validated as a string and begins with '<' 2325 // is considered altMacroString!!! 2326 else if ((Lexer.IsaAltMacroMode()) && 2327 (*(Token.getString().begin()) == '<') && 2328 Token.is(AsmToken::String)) { 2329 std::string Res; 2330 altMacroString(Token.getStringContents(), Res); 2331 OS << Res; 2332 } 2333 // We expect no quotes around the string's contents when 2334 // parsing for varargs. 2335 else if (Token.isNot(AsmToken::String) || VarargParameter) 2336 OS << Token.getString(); 2337 else 2338 OS << Token.getStringContents(); 2339 2340 Pos += 1 + Argument.size(); 2341 } 2342 } 2343 } 2344 // Update the scan point. 2345 Body = Body.substr(Pos); 2346 } 2347 2348 return false; 2349 } 2350 2351 MacroInstantiation::MacroInstantiation(SMLoc IL, int EB, SMLoc EL, 2352 size_t CondStackDepth) 2353 : InstantiationLoc(IL), ExitBuffer(EB), ExitLoc(EL), 2354 CondStackDepth(CondStackDepth) {} 2355 2356 static bool isOperator(AsmToken::TokenKind kind) { 2357 switch (kind) { 2358 default: 2359 return false; 2360 case AsmToken::Plus: 2361 case AsmToken::Minus: 2362 case AsmToken::Tilde: 2363 case AsmToken::Slash: 2364 case AsmToken::Star: 2365 case AsmToken::Dot: 2366 case AsmToken::Equal: 2367 case AsmToken::EqualEqual: 2368 case AsmToken::Pipe: 2369 case AsmToken::PipePipe: 2370 case AsmToken::Caret: 2371 case AsmToken::Amp: 2372 case AsmToken::AmpAmp: 2373 case AsmToken::Exclaim: 2374 case AsmToken::ExclaimEqual: 2375 case AsmToken::Less: 2376 case AsmToken::LessEqual: 2377 case AsmToken::LessLess: 2378 case AsmToken::LessGreater: 2379 case AsmToken::Greater: 2380 case AsmToken::GreaterEqual: 2381 case AsmToken::GreaterGreater: 2382 return true; 2383 } 2384 } 2385 2386 namespace { 2387 2388 class AsmLexerSkipSpaceRAII { 2389 public: 2390 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2391 Lexer.setSkipSpace(SkipSpace); 2392 } 2393 2394 ~AsmLexerSkipSpaceRAII() { 2395 Lexer.setSkipSpace(true); 2396 } 2397 2398 private: 2399 AsmLexer &Lexer; 2400 }; 2401 2402 } // end anonymous namespace 2403 2404 bool AsmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2405 2406 if (Vararg) { 2407 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2408 StringRef Str = parseStringToEndOfStatement(); 2409 MA.emplace_back(AsmToken::String, Str); 2410 } 2411 return false; 2412 } 2413 2414 unsigned ParenLevel = 0; 2415 2416 // Darwin doesn't use spaces to delmit arguments. 2417 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2418 2419 bool SpaceEaten; 2420 2421 while (true) { 2422 SpaceEaten = false; 2423 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2424 return TokError("unexpected token in macro instantiation"); 2425 2426 if (ParenLevel == 0) { 2427 2428 if (Lexer.is(AsmToken::Comma)) 2429 break; 2430 2431 if (Lexer.is(AsmToken::Space)) { 2432 SpaceEaten = true; 2433 Lexer.Lex(); // Eat spaces 2434 } 2435 2436 // Spaces can delimit parameters, but could also be part an expression. 2437 // If the token after a space is an operator, add the token and the next 2438 // one into this argument 2439 if (!IsDarwin) { 2440 if (isOperator(Lexer.getKind())) { 2441 MA.push_back(getTok()); 2442 Lexer.Lex(); 2443 2444 // Whitespace after an operator can be ignored. 2445 if (Lexer.is(AsmToken::Space)) 2446 Lexer.Lex(); 2447 2448 continue; 2449 } 2450 } 2451 if (SpaceEaten) 2452 break; 2453 } 2454 2455 // handleMacroEntry relies on not advancing the lexer here 2456 // to be able to fill in the remaining default parameter values 2457 if (Lexer.is(AsmToken::EndOfStatement)) 2458 break; 2459 2460 // Adjust the current parentheses level. 2461 if (Lexer.is(AsmToken::LParen)) 2462 ++ParenLevel; 2463 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2464 --ParenLevel; 2465 2466 // Append the token to the current argument list. 2467 MA.push_back(getTok()); 2468 Lexer.Lex(); 2469 } 2470 2471 if (ParenLevel != 0) 2472 return TokError("unbalanced parentheses in macro argument"); 2473 return false; 2474 } 2475 2476 // Parse the macro instantiation arguments. 2477 bool AsmParser::parseMacroArguments(const MCAsmMacro *M, 2478 MCAsmMacroArguments &A) { 2479 const unsigned NParameters = M ? M->Parameters.size() : 0; 2480 bool NamedParametersFound = false; 2481 SmallVector<SMLoc, 4> FALocs; 2482 2483 A.resize(NParameters); 2484 FALocs.resize(NParameters); 2485 2486 // Parse two kinds of macro invocations: 2487 // - macros defined without any parameters accept an arbitrary number of them 2488 // - macros defined with parameters accept at most that many of them 2489 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2490 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2491 ++Parameter) { 2492 SMLoc IDLoc = Lexer.getLoc(); 2493 MCAsmMacroParameter FA; 2494 2495 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2496 if (parseIdentifier(FA.Name)) 2497 return Error(IDLoc, "invalid argument identifier for formal argument"); 2498 2499 if (Lexer.isNot(AsmToken::Equal)) 2500 return TokError("expected '=' after formal parameter identifier"); 2501 2502 Lex(); 2503 2504 NamedParametersFound = true; 2505 } 2506 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2507 2508 if (NamedParametersFound && FA.Name.empty()) 2509 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2510 2511 SMLoc StrLoc = Lexer.getLoc(); 2512 SMLoc EndLoc; 2513 if (Lexer.IsaAltMacroMode() && Lexer.is(AsmToken::Percent)) { 2514 const MCExpr *AbsoluteExp; 2515 int64_t Value; 2516 /// Eat '%' 2517 Lex(); 2518 if (parseExpression(AbsoluteExp, EndLoc)) 2519 return false; 2520 if (!AbsoluteExp->evaluateAsAbsolute(Value)) 2521 return Error(StrLoc, "expected absolute expression"); 2522 const char *StrChar = StrLoc.getPointer(); 2523 const char *EndChar = EndLoc.getPointer(); 2524 AsmToken newToken(AsmToken::Integer, StringRef(StrChar , EndChar - StrChar), Value); 2525 FA.Value.push_back(newToken); 2526 } else if (Lexer.IsaAltMacroMode() && Lexer.is(AsmToken::Less) && 2527 isAltmacroString(StrLoc, EndLoc)) { 2528 const char *StrChar = StrLoc.getPointer(); 2529 const char *EndChar = EndLoc.getPointer(); 2530 jumpToLoc(EndLoc, CurBuffer); 2531 /// Eat from '<' to '>' 2532 Lex(); 2533 AsmToken newToken(AsmToken::String, StringRef(StrChar, EndChar - StrChar)); 2534 FA.Value.push_back(newToken); 2535 } else if(parseMacroArgument(FA.Value, Vararg)) 2536 return true; 2537 2538 unsigned PI = Parameter; 2539 if (!FA.Name.empty()) { 2540 unsigned FAI = 0; 2541 for (FAI = 0; FAI < NParameters; ++FAI) 2542 if (M->Parameters[FAI].Name == FA.Name) 2543 break; 2544 2545 if (FAI >= NParameters) { 2546 assert(M && "expected macro to be defined"); 2547 return Error(IDLoc, "parameter named '" + FA.Name + 2548 "' does not exist for macro '" + M->Name + "'"); 2549 } 2550 PI = FAI; 2551 } 2552 2553 if (!FA.Value.empty()) { 2554 if (A.size() <= PI) 2555 A.resize(PI + 1); 2556 A[PI] = FA.Value; 2557 2558 if (FALocs.size() <= PI) 2559 FALocs.resize(PI + 1); 2560 2561 FALocs[PI] = Lexer.getLoc(); 2562 } 2563 2564 // At the end of the statement, fill in remaining arguments that have 2565 // default values. If there aren't any, then the next argument is 2566 // required but missing 2567 if (Lexer.is(AsmToken::EndOfStatement)) { 2568 bool Failure = false; 2569 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2570 if (A[FAI].empty()) { 2571 if (M->Parameters[FAI].Required) { 2572 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2573 "missing value for required parameter " 2574 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2575 Failure = true; 2576 } 2577 2578 if (!M->Parameters[FAI].Value.empty()) 2579 A[FAI] = M->Parameters[FAI].Value; 2580 } 2581 } 2582 return Failure; 2583 } 2584 2585 if (Lexer.is(AsmToken::Comma)) 2586 Lex(); 2587 } 2588 2589 return TokError("too many positional arguments"); 2590 } 2591 2592 const MCAsmMacro *AsmParser::lookupMacro(StringRef Name) { 2593 StringMap<MCAsmMacro>::iterator I = MacroMap.find(Name); 2594 return (I == MacroMap.end()) ? nullptr : &I->getValue(); 2595 } 2596 2597 void AsmParser::defineMacro(StringRef Name, MCAsmMacro Macro) { 2598 MacroMap.insert(std::make_pair(Name, std::move(Macro))); 2599 } 2600 2601 void AsmParser::undefineMacro(StringRef Name) { MacroMap.erase(Name); } 2602 2603 bool AsmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2604 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2605 // eliminate this, although we should protect against infinite loops. 2606 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2607 if (ActiveMacros.size() == MaxNestingDepth) { 2608 std::ostringstream MaxNestingDepthError; 2609 MaxNestingDepthError << "macros cannot be nested more than " 2610 << MaxNestingDepth << " levels deep." 2611 << " Use -asm-macro-max-nesting-depth to increase " 2612 "this limit."; 2613 return TokError(MaxNestingDepthError.str()); 2614 } 2615 2616 MCAsmMacroArguments A; 2617 if (parseMacroArguments(M, A)) 2618 return true; 2619 2620 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2621 // to hold the macro body with substitutions. 2622 SmallString<256> Buf; 2623 StringRef Body = M->Body; 2624 raw_svector_ostream OS(Buf); 2625 2626 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2627 return true; 2628 2629 // We include the .endmacro in the buffer as our cue to exit the macro 2630 // instantiation. 2631 OS << ".endmacro\n"; 2632 2633 std::unique_ptr<MemoryBuffer> Instantiation = 2634 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2635 2636 // Create the macro instantiation object and add to the current macro 2637 // instantiation stack. 2638 MacroInstantiation *MI = new MacroInstantiation( 2639 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 2640 ActiveMacros.push_back(MI); 2641 2642 ++NumOfMacroInstantiations; 2643 2644 // Jump to the macro instantiation and prime the lexer. 2645 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2646 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2647 Lex(); 2648 2649 return false; 2650 } 2651 2652 void AsmParser::handleMacroExit() { 2653 // Jump to the EndOfStatement we should return to, and consume it. 2654 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2655 Lex(); 2656 2657 // Pop the instantiation entry. 2658 delete ActiveMacros.back(); 2659 ActiveMacros.pop_back(); 2660 } 2661 2662 bool AsmParser::parseAssignment(StringRef Name, bool allow_redef, 2663 bool NoDeadStrip) { 2664 MCSymbol *Sym; 2665 const MCExpr *Value; 2666 if (MCParserUtils::parseAssignmentExpression(Name, allow_redef, *this, Sym, 2667 Value)) 2668 return true; 2669 2670 if (!Sym) { 2671 // In the case where we parse an expression starting with a '.', we will 2672 // not generate an error, nor will we create a symbol. In this case we 2673 // should just return out. 2674 return false; 2675 } 2676 2677 // Do the assignment. 2678 Out.EmitAssignment(Sym, Value); 2679 if (NoDeadStrip) 2680 Out.EmitSymbolAttribute(Sym, MCSA_NoDeadStrip); 2681 2682 return false; 2683 } 2684 2685 /// parseIdentifier: 2686 /// ::= identifier 2687 /// ::= string 2688 bool AsmParser::parseIdentifier(StringRef &Res) { 2689 // The assembler has relaxed rules for accepting identifiers, in particular we 2690 // allow things like '.globl $foo' and '.def @feat.00', which would normally be 2691 // separate tokens. At this level, we have already lexed so we cannot (currently) 2692 // handle this as a context dependent token, instead we detect adjacent tokens 2693 // and return the combined identifier. 2694 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2695 SMLoc PrefixLoc = getLexer().getLoc(); 2696 2697 // Consume the prefix character, and check for a following identifier. 2698 2699 AsmToken Buf[1]; 2700 Lexer.peekTokens(Buf, false); 2701 2702 if (Buf[0].isNot(AsmToken::Identifier)) 2703 return true; 2704 2705 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2706 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2707 return true; 2708 2709 // eat $ or @ 2710 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2711 // Construct the joined identifier and consume the token. 2712 Res = 2713 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2714 Lex(); // Parser Lex to maintain invariants. 2715 return false; 2716 } 2717 2718 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2719 return true; 2720 2721 Res = getTok().getIdentifier(); 2722 2723 Lex(); // Consume the identifier token. 2724 2725 return false; 2726 } 2727 2728 /// parseDirectiveSet: 2729 /// ::= .equ identifier ',' expression 2730 /// ::= .equiv identifier ',' expression 2731 /// ::= .set identifier ',' expression 2732 bool AsmParser::parseDirectiveSet(StringRef IDVal, bool allow_redef) { 2733 StringRef Name; 2734 if (check(parseIdentifier(Name), "expected identifier") || 2735 parseToken(AsmToken::Comma) || parseAssignment(Name, allow_redef, true)) 2736 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2737 return false; 2738 } 2739 2740 bool AsmParser::parseEscapedString(std::string &Data) { 2741 if (check(getTok().isNot(AsmToken::String), "expected string")) 2742 return true; 2743 2744 Data = ""; 2745 StringRef Str = getTok().getStringContents(); 2746 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2747 if (Str[i] != '\\') { 2748 Data += Str[i]; 2749 continue; 2750 } 2751 2752 // Recognize escaped characters. Note that this escape semantics currently 2753 // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes. 2754 ++i; 2755 if (i == e) 2756 return TokError("unexpected backslash at end of string"); 2757 2758 // Recognize octal sequences. 2759 if ((unsigned)(Str[i] - '0') <= 7) { 2760 // Consume up to three octal characters. 2761 unsigned Value = Str[i] - '0'; 2762 2763 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2764 ++i; 2765 Value = Value * 8 + (Str[i] - '0'); 2766 2767 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2768 ++i; 2769 Value = Value * 8 + (Str[i] - '0'); 2770 } 2771 } 2772 2773 if (Value > 255) 2774 return TokError("invalid octal escape sequence (out of range)"); 2775 2776 Data += (unsigned char)Value; 2777 continue; 2778 } 2779 2780 // Otherwise recognize individual escapes. 2781 switch (Str[i]) { 2782 default: 2783 // Just reject invalid escape sequences for now. 2784 return TokError("invalid escape sequence (unrecognized character)"); 2785 2786 case 'b': Data += '\b'; break; 2787 case 'f': Data += '\f'; break; 2788 case 'n': Data += '\n'; break; 2789 case 'r': Data += '\r'; break; 2790 case 't': Data += '\t'; break; 2791 case '"': Data += '"'; break; 2792 case '\\': Data += '\\'; break; 2793 } 2794 } 2795 2796 Lex(); 2797 return false; 2798 } 2799 2800 /// parseDirectiveAscii: 2801 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2802 bool AsmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2803 auto parseOp = [&]() -> bool { 2804 std::string Data; 2805 if (checkForValidSection() || parseEscapedString(Data)) 2806 return true; 2807 getStreamer().EmitBytes(Data); 2808 if (ZeroTerminated) 2809 getStreamer().EmitBytes(StringRef("\0", 1)); 2810 return false; 2811 }; 2812 2813 if (parseMany(parseOp)) 2814 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2815 return false; 2816 } 2817 2818 /// parseDirectiveReloc 2819 /// ::= .reloc expression , identifier [ , expression ] 2820 bool AsmParser::parseDirectiveReloc(SMLoc DirectiveLoc) { 2821 const MCExpr *Offset; 2822 const MCExpr *Expr = nullptr; 2823 2824 SMLoc OffsetLoc = Lexer.getTok().getLoc(); 2825 int64_t OffsetValue; 2826 // We can only deal with constant expressions at the moment. 2827 2828 if (parseExpression(Offset)) 2829 return true; 2830 2831 if (check(!Offset->evaluateAsAbsolute(OffsetValue), OffsetLoc, 2832 "expression is not a constant value") || 2833 check(OffsetValue < 0, OffsetLoc, "expression is negative") || 2834 parseToken(AsmToken::Comma, "expected comma") || 2835 check(getTok().isNot(AsmToken::Identifier), "expected relocation name")) 2836 return true; 2837 2838 SMLoc NameLoc = Lexer.getTok().getLoc(); 2839 StringRef Name = Lexer.getTok().getIdentifier(); 2840 Lex(); 2841 2842 if (Lexer.is(AsmToken::Comma)) { 2843 Lex(); 2844 SMLoc ExprLoc = Lexer.getLoc(); 2845 if (parseExpression(Expr)) 2846 return true; 2847 2848 MCValue Value; 2849 if (!Expr->evaluateAsRelocatable(Value, nullptr, nullptr)) 2850 return Error(ExprLoc, "expression must be relocatable"); 2851 } 2852 2853 if (parseToken(AsmToken::EndOfStatement, 2854 "unexpected token in .reloc directive")) 2855 return true; 2856 2857 if (getStreamer().EmitRelocDirective(*Offset, Name, Expr, DirectiveLoc)) 2858 return Error(NameLoc, "unknown relocation name"); 2859 2860 return false; 2861 } 2862 2863 /// parseDirectiveValue 2864 /// ::= (.byte | .short | ... ) [ expression (, expression)* ] 2865 bool AsmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 2866 auto parseOp = [&]() -> bool { 2867 const MCExpr *Value; 2868 SMLoc ExprLoc = getLexer().getLoc(); 2869 if (checkForValidSection() || parseExpression(Value)) 2870 return true; 2871 // Special case constant expressions to match code generator. 2872 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2873 assert(Size <= 8 && "Invalid size"); 2874 uint64_t IntValue = MCE->getValue(); 2875 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2876 return Error(ExprLoc, "out of range literal value"); 2877 getStreamer().EmitIntValue(IntValue, Size); 2878 } else 2879 getStreamer().EmitValue(Value, Size, ExprLoc); 2880 return false; 2881 }; 2882 2883 if (parseMany(parseOp)) 2884 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2885 return false; 2886 } 2887 2888 /// ParseDirectiveOctaValue 2889 /// ::= .octa [ hexconstant (, hexconstant)* ] 2890 2891 bool AsmParser::parseDirectiveOctaValue(StringRef IDVal) { 2892 auto parseOp = [&]() -> bool { 2893 if (checkForValidSection()) 2894 return true; 2895 if (getTok().isNot(AsmToken::Integer) && getTok().isNot(AsmToken::BigNum)) 2896 return TokError("unknown token in expression"); 2897 SMLoc ExprLoc = getTok().getLoc(); 2898 APInt IntValue = getTok().getAPIntVal(); 2899 uint64_t hi, lo; 2900 Lex(); 2901 if (!IntValue.isIntN(128)) 2902 return Error(ExprLoc, "out of range literal value"); 2903 if (!IntValue.isIntN(64)) { 2904 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2905 lo = IntValue.getLoBits(64).getZExtValue(); 2906 } else { 2907 hi = 0; 2908 lo = IntValue.getZExtValue(); 2909 } 2910 if (MAI.isLittleEndian()) { 2911 getStreamer().EmitIntValue(lo, 8); 2912 getStreamer().EmitIntValue(hi, 8); 2913 } else { 2914 getStreamer().EmitIntValue(hi, 8); 2915 getStreamer().EmitIntValue(lo, 8); 2916 } 2917 return false; 2918 }; 2919 2920 if (parseMany(parseOp)) 2921 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2922 return false; 2923 } 2924 2925 bool AsmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 2926 // We don't truly support arithmetic on floating point expressions, so we 2927 // have to manually parse unary prefixes. 2928 bool IsNeg = false; 2929 if (getLexer().is(AsmToken::Minus)) { 2930 Lexer.Lex(); 2931 IsNeg = true; 2932 } else if (getLexer().is(AsmToken::Plus)) 2933 Lexer.Lex(); 2934 2935 if (Lexer.is(AsmToken::Error)) 2936 return TokError(Lexer.getErr()); 2937 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 2938 Lexer.isNot(AsmToken::Identifier)) 2939 return TokError("unexpected token in directive"); 2940 2941 // Convert to an APFloat. 2942 APFloat Value(Semantics); 2943 StringRef IDVal = getTok().getString(); 2944 if (getLexer().is(AsmToken::Identifier)) { 2945 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2946 Value = APFloat::getInf(Semantics); 2947 else if (!IDVal.compare_lower("nan")) 2948 Value = APFloat::getNaN(Semantics, false, ~0); 2949 else 2950 return TokError("invalid floating point literal"); 2951 } else if (Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) == 2952 APFloat::opInvalidOp) 2953 return TokError("invalid floating point literal"); 2954 if (IsNeg) 2955 Value.changeSign(); 2956 2957 // Consume the numeric token. 2958 Lex(); 2959 2960 Res = Value.bitcastToAPInt(); 2961 2962 return false; 2963 } 2964 2965 /// parseDirectiveRealValue 2966 /// ::= (.single | .double) [ expression (, expression)* ] 2967 bool AsmParser::parseDirectiveRealValue(StringRef IDVal, 2968 const fltSemantics &Semantics) { 2969 auto parseOp = [&]() -> bool { 2970 APInt AsInt; 2971 if (checkForValidSection() || parseRealValue(Semantics, AsInt)) 2972 return true; 2973 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 2974 AsInt.getBitWidth() / 8); 2975 return false; 2976 }; 2977 2978 if (parseMany(parseOp)) 2979 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2980 return false; 2981 } 2982 2983 /// parseDirectiveZero 2984 /// ::= .zero expression 2985 bool AsmParser::parseDirectiveZero() { 2986 SMLoc NumBytesLoc = Lexer.getLoc(); 2987 const MCExpr *NumBytes; 2988 if (checkForValidSection() || parseExpression(NumBytes)) 2989 return true; 2990 2991 int64_t Val = 0; 2992 if (getLexer().is(AsmToken::Comma)) { 2993 Lex(); 2994 if (parseAbsoluteExpression(Val)) 2995 return true; 2996 } 2997 2998 if (parseToken(AsmToken::EndOfStatement, 2999 "unexpected token in '.zero' directive")) 3000 return true; 3001 getStreamer().emitFill(*NumBytes, Val, NumBytesLoc); 3002 3003 return false; 3004 } 3005 3006 /// parseDirectiveFill 3007 /// ::= .fill expression [ , expression [ , expression ] ] 3008 bool AsmParser::parseDirectiveFill() { 3009 SMLoc NumValuesLoc = Lexer.getLoc(); 3010 const MCExpr *NumValues; 3011 if (checkForValidSection() || parseExpression(NumValues)) 3012 return true; 3013 3014 int64_t FillSize = 1; 3015 int64_t FillExpr = 0; 3016 3017 SMLoc SizeLoc, ExprLoc; 3018 3019 if (parseOptionalToken(AsmToken::Comma)) { 3020 SizeLoc = getTok().getLoc(); 3021 if (parseAbsoluteExpression(FillSize)) 3022 return true; 3023 if (parseOptionalToken(AsmToken::Comma)) { 3024 ExprLoc = getTok().getLoc(); 3025 if (parseAbsoluteExpression(FillExpr)) 3026 return true; 3027 } 3028 } 3029 if (parseToken(AsmToken::EndOfStatement, 3030 "unexpected token in '.fill' directive")) 3031 return true; 3032 3033 if (FillSize < 0) { 3034 Warning(SizeLoc, "'.fill' directive with negative size has no effect"); 3035 return false; 3036 } 3037 if (FillSize > 8) { 3038 Warning(SizeLoc, "'.fill' directive with size greater than 8 has been truncated to 8"); 3039 FillSize = 8; 3040 } 3041 3042 if (!isUInt<32>(FillExpr) && FillSize > 4) 3043 Warning(ExprLoc, "'.fill' directive pattern has been truncated to 32-bits"); 3044 3045 getStreamer().emitFill(*NumValues, FillSize, FillExpr, NumValuesLoc); 3046 3047 return false; 3048 } 3049 3050 /// parseDirectiveOrg 3051 /// ::= .org expression [ , expression ] 3052 bool AsmParser::parseDirectiveOrg() { 3053 const MCExpr *Offset; 3054 SMLoc OffsetLoc = Lexer.getLoc(); 3055 if (checkForValidSection() || parseExpression(Offset)) 3056 return true; 3057 3058 // Parse optional fill expression. 3059 int64_t FillExpr = 0; 3060 if (parseOptionalToken(AsmToken::Comma)) 3061 if (parseAbsoluteExpression(FillExpr)) 3062 return addErrorSuffix(" in '.org' directive"); 3063 if (parseToken(AsmToken::EndOfStatement)) 3064 return addErrorSuffix(" in '.org' directive"); 3065 3066 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3067 return false; 3068 } 3069 3070 /// parseDirectiveAlign 3071 /// ::= {.align, ...} expression [ , expression [ , expression ]] 3072 bool AsmParser::parseDirectiveAlign(bool IsPow2, unsigned ValueSize) { 3073 SMLoc AlignmentLoc = getLexer().getLoc(); 3074 int64_t Alignment; 3075 SMLoc MaxBytesLoc; 3076 bool HasFillExpr = false; 3077 int64_t FillExpr = 0; 3078 int64_t MaxBytesToFill = 0; 3079 3080 auto parseAlign = [&]() -> bool { 3081 if (checkForValidSection() || parseAbsoluteExpression(Alignment)) 3082 return true; 3083 if (parseOptionalToken(AsmToken::Comma)) { 3084 // The fill expression can be omitted while specifying a maximum number of 3085 // alignment bytes, e.g: 3086 // .align 3,,4 3087 if (getTok().isNot(AsmToken::Comma)) { 3088 HasFillExpr = true; 3089 if (parseAbsoluteExpression(FillExpr)) 3090 return true; 3091 } 3092 if (parseOptionalToken(AsmToken::Comma)) 3093 if (parseTokenLoc(MaxBytesLoc) || 3094 parseAbsoluteExpression(MaxBytesToFill)) 3095 return true; 3096 } 3097 return parseToken(AsmToken::EndOfStatement); 3098 }; 3099 3100 if (parseAlign()) 3101 return addErrorSuffix(" in directive"); 3102 3103 // Always emit an alignment here even if we thrown an error. 3104 bool ReturnVal = false; 3105 3106 // Compute alignment in bytes. 3107 if (IsPow2) { 3108 // FIXME: Diagnose overflow. 3109 if (Alignment >= 32) { 3110 ReturnVal |= Error(AlignmentLoc, "invalid alignment value"); 3111 Alignment = 31; 3112 } 3113 3114 Alignment = 1ULL << Alignment; 3115 } else { 3116 // Reject alignments that aren't either a power of two or zero, 3117 // for gas compatibility. Alignment of zero is silently rounded 3118 // up to one. 3119 if (Alignment == 0) 3120 Alignment = 1; 3121 if (!isPowerOf2_64(Alignment)) 3122 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3123 } 3124 3125 // Diagnose non-sensical max bytes to align. 3126 if (MaxBytesLoc.isValid()) { 3127 if (MaxBytesToFill < 1) { 3128 ReturnVal |= Error(MaxBytesLoc, 3129 "alignment directive can never be satisfied in this " 3130 "many bytes, ignoring maximum bytes expression"); 3131 MaxBytesToFill = 0; 3132 } 3133 3134 if (MaxBytesToFill >= Alignment) { 3135 Warning(MaxBytesLoc, "maximum bytes expression exceeds alignment and " 3136 "has no effect"); 3137 MaxBytesToFill = 0; 3138 } 3139 } 3140 3141 // Check whether we should use optimal code alignment for this .align 3142 // directive. 3143 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3144 assert(Section && "must have section to emit alignment"); 3145 bool UseCodeAlign = Section->UseCodeAlign(); 3146 if ((!HasFillExpr || Lexer.getMAI().getTextAlignFillValue() == FillExpr) && 3147 ValueSize == 1 && UseCodeAlign) { 3148 getStreamer().EmitCodeAlignment(Alignment, MaxBytesToFill); 3149 } else { 3150 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3151 getStreamer().EmitValueToAlignment(Alignment, FillExpr, ValueSize, 3152 MaxBytesToFill); 3153 } 3154 3155 return ReturnVal; 3156 } 3157 3158 /// parseDirectiveFile 3159 /// ::= .file [number] filename 3160 /// ::= .file number directory filename 3161 bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3162 // FIXME: I'm not sure what this is. 3163 int64_t FileNumber = -1; 3164 SMLoc FileNumberLoc = getLexer().getLoc(); 3165 if (getLexer().is(AsmToken::Integer)) { 3166 FileNumber = getTok().getIntVal(); 3167 Lex(); 3168 3169 if (FileNumber < 1) 3170 return TokError("file number less than one"); 3171 } 3172 3173 std::string Path = getTok().getString(); 3174 3175 // Usually the directory and filename together, otherwise just the directory. 3176 // Allow the strings to have escaped octal character sequence. 3177 if (check(getTok().isNot(AsmToken::String), 3178 "unexpected token in '.file' directive") || 3179 parseEscapedString(Path)) 3180 return true; 3181 3182 StringRef Directory; 3183 StringRef Filename; 3184 std::string FilenameData; 3185 if (getLexer().is(AsmToken::String)) { 3186 if (check(FileNumber == -1, 3187 "explicit path specified, but no file number") || 3188 parseEscapedString(FilenameData)) 3189 return true; 3190 Filename = FilenameData; 3191 Directory = Path; 3192 } else { 3193 Filename = Path; 3194 } 3195 3196 if (parseToken(AsmToken::EndOfStatement, 3197 "unexpected token in '.file' directive")) 3198 return true; 3199 3200 if (FileNumber == -1) 3201 getStreamer().EmitFileDirective(Filename); 3202 else { 3203 // If there is -g option as well as debug info from directive file, 3204 // we turn off -g option, directly use the existing debug info instead. 3205 if (getContext().getGenDwarfForAssembly()) 3206 getContext().setGenDwarfForAssembly(false); 3207 else if (getStreamer().EmitDwarfFileDirective(FileNumber, Directory, Filename) == 3208 0) 3209 return Error(FileNumberLoc, "file number already allocated"); 3210 } 3211 3212 return false; 3213 } 3214 3215 /// parseDirectiveLine 3216 /// ::= .line [number] 3217 bool AsmParser::parseDirectiveLine() { 3218 int64_t LineNumber; 3219 if (getLexer().is(AsmToken::Integer)) { 3220 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3221 return true; 3222 (void)LineNumber; 3223 // FIXME: Do something with the .line. 3224 } 3225 if (parseToken(AsmToken::EndOfStatement, 3226 "unexpected token in '.line' directive")) 3227 return true; 3228 3229 return false; 3230 } 3231 3232 /// parseDirectiveLoc 3233 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3234 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3235 /// The first number is a file number, must have been previously assigned with 3236 /// a .file directive, the second number is the line number and optionally the 3237 /// third number is a column position (zero if not specified). The remaining 3238 /// optional items are .loc sub-directives. 3239 bool AsmParser::parseDirectiveLoc() { 3240 int64_t FileNumber = 0, LineNumber = 0; 3241 SMLoc Loc = getTok().getLoc(); 3242 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3243 check(FileNumber < 1, Loc, 3244 "file number less than one in '.loc' directive") || 3245 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3246 "unassigned file number in '.loc' directive")) 3247 return true; 3248 3249 // optional 3250 if (getLexer().is(AsmToken::Integer)) { 3251 LineNumber = getTok().getIntVal(); 3252 if (LineNumber < 0) 3253 return TokError("line number less than zero in '.loc' directive"); 3254 Lex(); 3255 } 3256 3257 int64_t ColumnPos = 0; 3258 if (getLexer().is(AsmToken::Integer)) { 3259 ColumnPos = getTok().getIntVal(); 3260 if (ColumnPos < 0) 3261 return TokError("column position less than zero in '.loc' directive"); 3262 Lex(); 3263 } 3264 3265 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3266 unsigned Isa = 0; 3267 int64_t Discriminator = 0; 3268 3269 auto parseLocOp = [&]() -> bool { 3270 StringRef Name; 3271 SMLoc Loc = getTok().getLoc(); 3272 if (parseIdentifier(Name)) 3273 return TokError("unexpected token in '.loc' directive"); 3274 3275 if (Name == "basic_block") 3276 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3277 else if (Name == "prologue_end") 3278 Flags |= DWARF2_FLAG_PROLOGUE_END; 3279 else if (Name == "epilogue_begin") 3280 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3281 else if (Name == "is_stmt") { 3282 Loc = getTok().getLoc(); 3283 const MCExpr *Value; 3284 if (parseExpression(Value)) 3285 return true; 3286 // The expression must be the constant 0 or 1. 3287 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3288 int Value = MCE->getValue(); 3289 if (Value == 0) 3290 Flags &= ~DWARF2_FLAG_IS_STMT; 3291 else if (Value == 1) 3292 Flags |= DWARF2_FLAG_IS_STMT; 3293 else 3294 return Error(Loc, "is_stmt value not 0 or 1"); 3295 } else { 3296 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3297 } 3298 } else if (Name == "isa") { 3299 Loc = getTok().getLoc(); 3300 const MCExpr *Value; 3301 if (parseExpression(Value)) 3302 return true; 3303 // The expression must be a constant greater or equal to 0. 3304 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3305 int Value = MCE->getValue(); 3306 if (Value < 0) 3307 return Error(Loc, "isa number less than zero"); 3308 Isa = Value; 3309 } else { 3310 return Error(Loc, "isa number not a constant value"); 3311 } 3312 } else if (Name == "discriminator") { 3313 if (parseAbsoluteExpression(Discriminator)) 3314 return true; 3315 } else { 3316 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3317 } 3318 return false; 3319 }; 3320 3321 if (parseMany(parseLocOp, false /*hasComma*/)) 3322 return true; 3323 3324 getStreamer().EmitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3325 Isa, Discriminator, StringRef()); 3326 3327 return false; 3328 } 3329 3330 /// parseDirectiveStabs 3331 /// ::= .stabs string, number, number, number 3332 bool AsmParser::parseDirectiveStabs() { 3333 return TokError("unsupported directive '.stabs'"); 3334 } 3335 3336 /// parseDirectiveCVFile 3337 /// ::= .cv_file number filename 3338 bool AsmParser::parseDirectiveCVFile() { 3339 SMLoc FileNumberLoc = getTok().getLoc(); 3340 int64_t FileNumber; 3341 std::string Filename; 3342 3343 if (parseIntToken(FileNumber, 3344 "expected file number in '.cv_file' directive") || 3345 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3346 check(getTok().isNot(AsmToken::String), 3347 "unexpected token in '.cv_file' directive") || 3348 // Usually directory and filename are together, otherwise just 3349 // directory. Allow the strings to have escaped octal character sequence. 3350 parseEscapedString(Filename) || 3351 parseToken(AsmToken::EndOfStatement, 3352 "unexpected token in '.cv_file' directive")) 3353 return true; 3354 3355 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename)) 3356 return Error(FileNumberLoc, "file number already allocated"); 3357 3358 return false; 3359 } 3360 3361 bool AsmParser::parseCVFunctionId(int64_t &FunctionId, 3362 StringRef DirectiveName) { 3363 SMLoc Loc; 3364 return parseTokenLoc(Loc) || 3365 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3366 "' directive") || 3367 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3368 "expected function id within range [0, UINT_MAX)"); 3369 } 3370 3371 bool AsmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3372 SMLoc Loc; 3373 return parseTokenLoc(Loc) || 3374 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3375 "' directive") || 3376 check(FileNumber < 1, Loc, "file number less than one in '" + 3377 DirectiveName + "' directive") || 3378 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3379 "unassigned file number in '" + DirectiveName + "' directive"); 3380 } 3381 3382 /// parseDirectiveCVFuncId 3383 /// ::= .cv_func_id FunctionId 3384 /// 3385 /// Introduces a function ID that can be used with .cv_loc. 3386 bool AsmParser::parseDirectiveCVFuncId() { 3387 SMLoc FunctionIdLoc = getTok().getLoc(); 3388 int64_t FunctionId; 3389 3390 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3391 parseToken(AsmToken::EndOfStatement, 3392 "unexpected token in '.cv_func_id' directive")) 3393 return true; 3394 3395 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3396 return Error(FunctionIdLoc, "function id already allocated"); 3397 3398 return false; 3399 } 3400 3401 /// parseDirectiveCVInlineSiteId 3402 /// ::= .cv_inline_site_id FunctionId 3403 /// "within" IAFunc 3404 /// "inlined_at" IAFile IALine [IACol] 3405 /// 3406 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3407 /// at" source location information for use in the line table of the caller, 3408 /// whether the caller is a real function or another inlined call site. 3409 bool AsmParser::parseDirectiveCVInlineSiteId() { 3410 SMLoc FunctionIdLoc = getTok().getLoc(); 3411 int64_t FunctionId; 3412 int64_t IAFunc; 3413 int64_t IAFile; 3414 int64_t IALine; 3415 int64_t IACol = 0; 3416 3417 // FunctionId 3418 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3419 return true; 3420 3421 // "within" 3422 if (check((getLexer().isNot(AsmToken::Identifier) || 3423 getTok().getIdentifier() != "within"), 3424 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3425 return true; 3426 Lex(); 3427 3428 // IAFunc 3429 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3430 return true; 3431 3432 // "inlined_at" 3433 if (check((getLexer().isNot(AsmToken::Identifier) || 3434 getTok().getIdentifier() != "inlined_at"), 3435 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3436 "directive") ) 3437 return true; 3438 Lex(); 3439 3440 // IAFile IALine 3441 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3442 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3443 return true; 3444 3445 // [IACol] 3446 if (getLexer().is(AsmToken::Integer)) { 3447 IACol = getTok().getIntVal(); 3448 Lex(); 3449 } 3450 3451 if (parseToken(AsmToken::EndOfStatement, 3452 "unexpected token in '.cv_inline_site_id' directive")) 3453 return true; 3454 3455 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3456 IALine, IACol, FunctionIdLoc)) 3457 return Error(FunctionIdLoc, "function id already allocated"); 3458 3459 return false; 3460 } 3461 3462 /// parseDirectiveCVLoc 3463 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3464 /// [is_stmt VALUE] 3465 /// The first number is a file number, must have been previously assigned with 3466 /// a .file directive, the second number is the line number and optionally the 3467 /// third number is a column position (zero if not specified). The remaining 3468 /// optional items are .loc sub-directives. 3469 bool AsmParser::parseDirectiveCVLoc() { 3470 SMLoc DirectiveLoc = getTok().getLoc(); 3471 SMLoc Loc; 3472 int64_t FunctionId, FileNumber; 3473 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3474 parseCVFileId(FileNumber, ".cv_loc")) 3475 return true; 3476 3477 int64_t LineNumber = 0; 3478 if (getLexer().is(AsmToken::Integer)) { 3479 LineNumber = getTok().getIntVal(); 3480 if (LineNumber < 0) 3481 return TokError("line number less than zero in '.cv_loc' directive"); 3482 Lex(); 3483 } 3484 3485 int64_t ColumnPos = 0; 3486 if (getLexer().is(AsmToken::Integer)) { 3487 ColumnPos = getTok().getIntVal(); 3488 if (ColumnPos < 0) 3489 return TokError("column position less than zero in '.cv_loc' directive"); 3490 Lex(); 3491 } 3492 3493 bool PrologueEnd = false; 3494 uint64_t IsStmt = 0; 3495 3496 auto parseOp = [&]() -> bool { 3497 StringRef Name; 3498 SMLoc Loc = getTok().getLoc(); 3499 if (parseIdentifier(Name)) 3500 return TokError("unexpected token in '.cv_loc' directive"); 3501 if (Name == "prologue_end") 3502 PrologueEnd = true; 3503 else if (Name == "is_stmt") { 3504 Loc = getTok().getLoc(); 3505 const MCExpr *Value; 3506 if (parseExpression(Value)) 3507 return true; 3508 // The expression must be the constant 0 or 1. 3509 IsStmt = ~0ULL; 3510 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3511 IsStmt = MCE->getValue(); 3512 3513 if (IsStmt > 1) 3514 return Error(Loc, "is_stmt value not 0 or 1"); 3515 } else { 3516 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3517 } 3518 return false; 3519 }; 3520 3521 if (parseMany(parseOp, false /*hasComma*/)) 3522 return true; 3523 3524 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3525 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3526 DirectiveLoc); 3527 return false; 3528 } 3529 3530 /// parseDirectiveCVLinetable 3531 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3532 bool AsmParser::parseDirectiveCVLinetable() { 3533 int64_t FunctionId; 3534 StringRef FnStartName, FnEndName; 3535 SMLoc Loc = getTok().getLoc(); 3536 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3537 parseToken(AsmToken::Comma, 3538 "unexpected token in '.cv_linetable' directive") || 3539 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3540 "expected identifier in directive") || 3541 parseToken(AsmToken::Comma, 3542 "unexpected token in '.cv_linetable' directive") || 3543 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3544 "expected identifier in directive")) 3545 return true; 3546 3547 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3548 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3549 3550 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3551 return false; 3552 } 3553 3554 /// parseDirectiveCVInlineLinetable 3555 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3556 bool AsmParser::parseDirectiveCVInlineLinetable() { 3557 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3558 StringRef FnStartName, FnEndName; 3559 SMLoc Loc = getTok().getLoc(); 3560 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3561 parseTokenLoc(Loc) || 3562 parseIntToken( 3563 SourceFileId, 3564 "expected SourceField in '.cv_inline_linetable' directive") || 3565 check(SourceFileId <= 0, Loc, 3566 "File id less than zero in '.cv_inline_linetable' directive") || 3567 parseTokenLoc(Loc) || 3568 parseIntToken( 3569 SourceLineNum, 3570 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3571 check(SourceLineNum < 0, Loc, 3572 "Line number less than zero in '.cv_inline_linetable' directive") || 3573 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3574 "expected identifier in directive") || 3575 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3576 "expected identifier in directive")) 3577 return true; 3578 3579 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3580 return true; 3581 3582 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3583 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3584 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3585 SourceLineNum, FnStartSym, 3586 FnEndSym); 3587 return false; 3588 } 3589 3590 /// parseDirectiveCVDefRange 3591 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3592 bool AsmParser::parseDirectiveCVDefRange() { 3593 SMLoc Loc; 3594 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3595 while (getLexer().is(AsmToken::Identifier)) { 3596 Loc = getLexer().getLoc(); 3597 StringRef GapStartName; 3598 if (parseIdentifier(GapStartName)) 3599 return Error(Loc, "expected identifier in directive"); 3600 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3601 3602 Loc = getLexer().getLoc(); 3603 StringRef GapEndName; 3604 if (parseIdentifier(GapEndName)) 3605 return Error(Loc, "expected identifier in directive"); 3606 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3607 3608 Ranges.push_back({GapStartSym, GapEndSym}); 3609 } 3610 3611 std::string FixedSizePortion; 3612 if (parseToken(AsmToken::Comma, "unexpected token in directive") || 3613 parseEscapedString(FixedSizePortion)) 3614 return true; 3615 3616 getStreamer().EmitCVDefRangeDirective(Ranges, FixedSizePortion); 3617 return false; 3618 } 3619 3620 /// parseDirectiveCVStringTable 3621 /// ::= .cv_stringtable 3622 bool AsmParser::parseDirectiveCVStringTable() { 3623 getStreamer().EmitCVStringTableDirective(); 3624 return false; 3625 } 3626 3627 /// parseDirectiveCVFileChecksums 3628 /// ::= .cv_filechecksums 3629 bool AsmParser::parseDirectiveCVFileChecksums() { 3630 getStreamer().EmitCVFileChecksumsDirective(); 3631 return false; 3632 } 3633 3634 /// parseDirectiveCFISections 3635 /// ::= .cfi_sections section [, section] 3636 bool AsmParser::parseDirectiveCFISections() { 3637 StringRef Name; 3638 bool EH = false; 3639 bool Debug = false; 3640 3641 if (parseIdentifier(Name)) 3642 return TokError("Expected an identifier"); 3643 3644 if (Name == ".eh_frame") 3645 EH = true; 3646 else if (Name == ".debug_frame") 3647 Debug = true; 3648 3649 if (getLexer().is(AsmToken::Comma)) { 3650 Lex(); 3651 3652 if (parseIdentifier(Name)) 3653 return TokError("Expected an identifier"); 3654 3655 if (Name == ".eh_frame") 3656 EH = true; 3657 else if (Name == ".debug_frame") 3658 Debug = true; 3659 } 3660 3661 getStreamer().EmitCFISections(EH, Debug); 3662 return false; 3663 } 3664 3665 /// parseDirectiveCFIStartProc 3666 /// ::= .cfi_startproc [simple] 3667 bool AsmParser::parseDirectiveCFIStartProc() { 3668 StringRef Simple; 3669 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3670 if (check(parseIdentifier(Simple) || Simple != "simple", 3671 "unexpected token") || 3672 parseToken(AsmToken::EndOfStatement)) 3673 return addErrorSuffix(" in '.cfi_startproc' directive"); 3674 } 3675 3676 getStreamer().EmitCFIStartProc(!Simple.empty()); 3677 return false; 3678 } 3679 3680 /// parseDirectiveCFIEndProc 3681 /// ::= .cfi_endproc 3682 bool AsmParser::parseDirectiveCFIEndProc() { 3683 getStreamer().EmitCFIEndProc(); 3684 return false; 3685 } 3686 3687 /// \brief parse register name or number. 3688 bool AsmParser::parseRegisterOrRegisterNumber(int64_t &Register, 3689 SMLoc DirectiveLoc) { 3690 unsigned RegNo; 3691 3692 if (getLexer().isNot(AsmToken::Integer)) { 3693 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 3694 return true; 3695 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 3696 } else 3697 return parseAbsoluteExpression(Register); 3698 3699 return false; 3700 } 3701 3702 /// parseDirectiveCFIDefCfa 3703 /// ::= .cfi_def_cfa register, offset 3704 bool AsmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3705 int64_t Register = 0, Offset = 0; 3706 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3707 parseToken(AsmToken::Comma, "unexpected token in directive") || 3708 parseAbsoluteExpression(Offset)) 3709 return true; 3710 3711 getStreamer().EmitCFIDefCfa(Register, Offset); 3712 return false; 3713 } 3714 3715 /// parseDirectiveCFIDefCfaOffset 3716 /// ::= .cfi_def_cfa_offset offset 3717 bool AsmParser::parseDirectiveCFIDefCfaOffset() { 3718 int64_t Offset = 0; 3719 if (parseAbsoluteExpression(Offset)) 3720 return true; 3721 3722 getStreamer().EmitCFIDefCfaOffset(Offset); 3723 return false; 3724 } 3725 3726 /// parseDirectiveCFIRegister 3727 /// ::= .cfi_register register, register 3728 bool AsmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3729 int64_t Register1 = 0, Register2 = 0; 3730 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 3731 parseToken(AsmToken::Comma, "unexpected token in directive") || 3732 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3733 return true; 3734 3735 getStreamer().EmitCFIRegister(Register1, Register2); 3736 return false; 3737 } 3738 3739 /// parseDirectiveCFIWindowSave 3740 /// ::= .cfi_window_save 3741 bool AsmParser::parseDirectiveCFIWindowSave() { 3742 getStreamer().EmitCFIWindowSave(); 3743 return false; 3744 } 3745 3746 /// parseDirectiveCFIAdjustCfaOffset 3747 /// ::= .cfi_adjust_cfa_offset adjustment 3748 bool AsmParser::parseDirectiveCFIAdjustCfaOffset() { 3749 int64_t Adjustment = 0; 3750 if (parseAbsoluteExpression(Adjustment)) 3751 return true; 3752 3753 getStreamer().EmitCFIAdjustCfaOffset(Adjustment); 3754 return false; 3755 } 3756 3757 /// parseDirectiveCFIDefCfaRegister 3758 /// ::= .cfi_def_cfa_register register 3759 bool AsmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3760 int64_t Register = 0; 3761 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3762 return true; 3763 3764 getStreamer().EmitCFIDefCfaRegister(Register); 3765 return false; 3766 } 3767 3768 /// parseDirectiveCFIOffset 3769 /// ::= .cfi_offset register, offset 3770 bool AsmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3771 int64_t Register = 0; 3772 int64_t Offset = 0; 3773 3774 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3775 parseToken(AsmToken::Comma, "unexpected token in directive") || 3776 parseAbsoluteExpression(Offset)) 3777 return true; 3778 3779 getStreamer().EmitCFIOffset(Register, Offset); 3780 return false; 3781 } 3782 3783 /// parseDirectiveCFIRelOffset 3784 /// ::= .cfi_rel_offset register, offset 3785 bool AsmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3786 int64_t Register = 0, Offset = 0; 3787 3788 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3789 parseToken(AsmToken::Comma, "unexpected token in directive") || 3790 parseAbsoluteExpression(Offset)) 3791 return true; 3792 3793 getStreamer().EmitCFIRelOffset(Register, Offset); 3794 return false; 3795 } 3796 3797 static bool isValidEncoding(int64_t Encoding) { 3798 if (Encoding & ~0xff) 3799 return false; 3800 3801 if (Encoding == dwarf::DW_EH_PE_omit) 3802 return true; 3803 3804 const unsigned Format = Encoding & 0xf; 3805 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3806 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3807 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3808 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3809 return false; 3810 3811 const unsigned Application = Encoding & 0x70; 3812 if (Application != dwarf::DW_EH_PE_absptr && 3813 Application != dwarf::DW_EH_PE_pcrel) 3814 return false; 3815 3816 return true; 3817 } 3818 3819 /// parseDirectiveCFIPersonalityOrLsda 3820 /// IsPersonality true for cfi_personality, false for cfi_lsda 3821 /// ::= .cfi_personality encoding, [symbol_name] 3822 /// ::= .cfi_lsda encoding, [symbol_name] 3823 bool AsmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3824 int64_t Encoding = 0; 3825 if (parseAbsoluteExpression(Encoding)) 3826 return true; 3827 if (Encoding == dwarf::DW_EH_PE_omit) 3828 return false; 3829 3830 StringRef Name; 3831 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 3832 parseToken(AsmToken::Comma, "unexpected token in directive") || 3833 check(parseIdentifier(Name), "expected identifier in directive")) 3834 return true; 3835 3836 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 3837 3838 if (IsPersonality) 3839 getStreamer().EmitCFIPersonality(Sym, Encoding); 3840 else 3841 getStreamer().EmitCFILsda(Sym, Encoding); 3842 return false; 3843 } 3844 3845 /// parseDirectiveCFIRememberState 3846 /// ::= .cfi_remember_state 3847 bool AsmParser::parseDirectiveCFIRememberState() { 3848 getStreamer().EmitCFIRememberState(); 3849 return false; 3850 } 3851 3852 /// parseDirectiveCFIRestoreState 3853 /// ::= .cfi_remember_state 3854 bool AsmParser::parseDirectiveCFIRestoreState() { 3855 getStreamer().EmitCFIRestoreState(); 3856 return false; 3857 } 3858 3859 /// parseDirectiveCFISameValue 3860 /// ::= .cfi_same_value register 3861 bool AsmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3862 int64_t Register = 0; 3863 3864 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3865 return true; 3866 3867 getStreamer().EmitCFISameValue(Register); 3868 return false; 3869 } 3870 3871 /// parseDirectiveCFIRestore 3872 /// ::= .cfi_restore register 3873 bool AsmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3874 int64_t Register = 0; 3875 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3876 return true; 3877 3878 getStreamer().EmitCFIRestore(Register); 3879 return false; 3880 } 3881 3882 /// parseDirectiveCFIEscape 3883 /// ::= .cfi_escape expression[,...] 3884 bool AsmParser::parseDirectiveCFIEscape() { 3885 std::string Values; 3886 int64_t CurrValue; 3887 if (parseAbsoluteExpression(CurrValue)) 3888 return true; 3889 3890 Values.push_back((uint8_t)CurrValue); 3891 3892 while (getLexer().is(AsmToken::Comma)) { 3893 Lex(); 3894 3895 if (parseAbsoluteExpression(CurrValue)) 3896 return true; 3897 3898 Values.push_back((uint8_t)CurrValue); 3899 } 3900 3901 getStreamer().EmitCFIEscape(Values); 3902 return false; 3903 } 3904 3905 /// parseDirectiveCFISignalFrame 3906 /// ::= .cfi_signal_frame 3907 bool AsmParser::parseDirectiveCFISignalFrame() { 3908 if (parseToken(AsmToken::EndOfStatement, 3909 "unexpected token in '.cfi_signal_frame'")) 3910 return true; 3911 3912 getStreamer().EmitCFISignalFrame(); 3913 return false; 3914 } 3915 3916 /// parseDirectiveCFIUndefined 3917 /// ::= .cfi_undefined register 3918 bool AsmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 3919 int64_t Register = 0; 3920 3921 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3922 return true; 3923 3924 getStreamer().EmitCFIUndefined(Register); 3925 return false; 3926 } 3927 3928 /// parseDirectiveAltmacro 3929 /// ::= .altmacro 3930 /// ::= .noaltmacro 3931 bool AsmParser::parseDirectiveAltmacro(StringRef Directive) { 3932 if (getLexer().isNot(AsmToken::EndOfStatement)) 3933 return TokError("unexpected token in '" + Directive + "' directive"); 3934 if (Directive == ".altmacro") 3935 getLexer().SetAltMacroMode(true); 3936 else 3937 getLexer().SetAltMacroMode(false); 3938 return false; 3939 } 3940 3941 /// parseDirectiveMacrosOnOff 3942 /// ::= .macros_on 3943 /// ::= .macros_off 3944 bool AsmParser::parseDirectiveMacrosOnOff(StringRef Directive) { 3945 if (parseToken(AsmToken::EndOfStatement, 3946 "unexpected token in '" + Directive + "' directive")) 3947 return true; 3948 3949 setMacrosEnabled(Directive == ".macros_on"); 3950 return false; 3951 } 3952 3953 /// parseDirectiveMacro 3954 /// ::= .macro name[,] [parameters] 3955 bool AsmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 3956 StringRef Name; 3957 if (parseIdentifier(Name)) 3958 return TokError("expected identifier in '.macro' directive"); 3959 3960 if (getLexer().is(AsmToken::Comma)) 3961 Lex(); 3962 3963 MCAsmMacroParameters Parameters; 3964 while (getLexer().isNot(AsmToken::EndOfStatement)) { 3965 3966 if (!Parameters.empty() && Parameters.back().Vararg) 3967 return Error(Lexer.getLoc(), 3968 "Vararg parameter '" + Parameters.back().Name + 3969 "' should be last one in the list of parameters."); 3970 3971 MCAsmMacroParameter Parameter; 3972 if (parseIdentifier(Parameter.Name)) 3973 return TokError("expected identifier in '.macro' directive"); 3974 3975 // Emit an error if two (or more) named parameters share the same name 3976 for (const MCAsmMacroParameter& CurrParam : Parameters) 3977 if (CurrParam.Name.equals(Parameter.Name)) 3978 return TokError("macro '" + Name + "' has multiple parameters" 3979 " named '" + Parameter.Name + "'"); 3980 3981 if (Lexer.is(AsmToken::Colon)) { 3982 Lex(); // consume ':' 3983 3984 SMLoc QualLoc; 3985 StringRef Qualifier; 3986 3987 QualLoc = Lexer.getLoc(); 3988 if (parseIdentifier(Qualifier)) 3989 return Error(QualLoc, "missing parameter qualifier for " 3990 "'" + Parameter.Name + "' in macro '" + Name + "'"); 3991 3992 if (Qualifier == "req") 3993 Parameter.Required = true; 3994 else if (Qualifier == "vararg") 3995 Parameter.Vararg = true; 3996 else 3997 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 3998 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 3999 } 4000 4001 if (getLexer().is(AsmToken::Equal)) { 4002 Lex(); 4003 4004 SMLoc ParamLoc; 4005 4006 ParamLoc = Lexer.getLoc(); 4007 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4008 return true; 4009 4010 if (Parameter.Required) 4011 Warning(ParamLoc, "pointless default value for required parameter " 4012 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4013 } 4014 4015 Parameters.push_back(std::move(Parameter)); 4016 4017 if (getLexer().is(AsmToken::Comma)) 4018 Lex(); 4019 } 4020 4021 // Eat just the end of statement. 4022 Lexer.Lex(); 4023 4024 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4025 AsmToken EndToken, StartToken = getTok(); 4026 unsigned MacroDepth = 0; 4027 // Lex the macro definition. 4028 while (true) { 4029 // Ignore Lexing errors in macros. 4030 while (Lexer.is(AsmToken::Error)) { 4031 Lexer.Lex(); 4032 } 4033 4034 // Check whether we have reached the end of the file. 4035 if (getLexer().is(AsmToken::Eof)) 4036 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4037 4038 // Otherwise, check whether we have reach the .endmacro. 4039 if (getLexer().is(AsmToken::Identifier)) { 4040 if (getTok().getIdentifier() == ".endm" || 4041 getTok().getIdentifier() == ".endmacro") { 4042 if (MacroDepth == 0) { // Outermost macro. 4043 EndToken = getTok(); 4044 Lexer.Lex(); 4045 if (getLexer().isNot(AsmToken::EndOfStatement)) 4046 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4047 "' directive"); 4048 break; 4049 } else { 4050 // Otherwise we just found the end of an inner macro. 4051 --MacroDepth; 4052 } 4053 } else if (getTok().getIdentifier() == ".macro") { 4054 // We allow nested macros. Those aren't instantiated until the outermost 4055 // macro is expanded so just ignore them for now. 4056 ++MacroDepth; 4057 } 4058 } 4059 4060 // Otherwise, scan til the end of the statement. 4061 eatToEndOfStatement(); 4062 } 4063 4064 if (lookupMacro(Name)) { 4065 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4066 } 4067 4068 const char *BodyStart = StartToken.getLoc().getPointer(); 4069 const char *BodyEnd = EndToken.getLoc().getPointer(); 4070 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4071 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4072 defineMacro(Name, MCAsmMacro(Name, Body, std::move(Parameters))); 4073 return false; 4074 } 4075 4076 /// checkForBadMacro 4077 /// 4078 /// With the support added for named parameters there may be code out there that 4079 /// is transitioning from positional parameters. In versions of gas that did 4080 /// not support named parameters they would be ignored on the macro definition. 4081 /// But to support both styles of parameters this is not possible so if a macro 4082 /// definition has named parameters but does not use them and has what appears 4083 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4084 /// warning that the positional parameter found in body which have no effect. 4085 /// Hoping the developer will either remove the named parameters from the macro 4086 /// definition so the positional parameters get used if that was what was 4087 /// intended or change the macro to use the named parameters. It is possible 4088 /// this warning will trigger when the none of the named parameters are used 4089 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4090 void AsmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4091 StringRef Body, 4092 ArrayRef<MCAsmMacroParameter> Parameters) { 4093 // If this macro is not defined with named parameters the warning we are 4094 // checking for here doesn't apply. 4095 unsigned NParameters = Parameters.size(); 4096 if (NParameters == 0) 4097 return; 4098 4099 bool NamedParametersFound = false; 4100 bool PositionalParametersFound = false; 4101 4102 // Look at the body of the macro for use of both the named parameters and what 4103 // are likely to be positional parameters. This is what expandMacro() is 4104 // doing when it finds the parameters in the body. 4105 while (!Body.empty()) { 4106 // Scan for the next possible parameter. 4107 std::size_t End = Body.size(), Pos = 0; 4108 for (; Pos != End; ++Pos) { 4109 // Check for a substitution or escape. 4110 // This macro is defined with parameters, look for \foo, \bar, etc. 4111 if (Body[Pos] == '\\' && Pos + 1 != End) 4112 break; 4113 4114 // This macro should have parameters, but look for $0, $1, ..., $n too. 4115 if (Body[Pos] != '$' || Pos + 1 == End) 4116 continue; 4117 char Next = Body[Pos + 1]; 4118 if (Next == '$' || Next == 'n' || 4119 isdigit(static_cast<unsigned char>(Next))) 4120 break; 4121 } 4122 4123 // Check if we reached the end. 4124 if (Pos == End) 4125 break; 4126 4127 if (Body[Pos] == '$') { 4128 switch (Body[Pos + 1]) { 4129 // $$ => $ 4130 case '$': 4131 break; 4132 4133 // $n => number of arguments 4134 case 'n': 4135 PositionalParametersFound = true; 4136 break; 4137 4138 // $[0-9] => argument 4139 default: { 4140 PositionalParametersFound = true; 4141 break; 4142 } 4143 } 4144 Pos += 2; 4145 } else { 4146 unsigned I = Pos + 1; 4147 while (isIdentifierChar(Body[I]) && I + 1 != End) 4148 ++I; 4149 4150 const char *Begin = Body.data() + Pos + 1; 4151 StringRef Argument(Begin, I - (Pos + 1)); 4152 unsigned Index = 0; 4153 for (; Index < NParameters; ++Index) 4154 if (Parameters[Index].Name == Argument) 4155 break; 4156 4157 if (Index == NParameters) { 4158 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4159 Pos += 3; 4160 else { 4161 Pos = I; 4162 } 4163 } else { 4164 NamedParametersFound = true; 4165 Pos += 1 + Argument.size(); 4166 } 4167 } 4168 // Update the scan point. 4169 Body = Body.substr(Pos); 4170 } 4171 4172 if (!NamedParametersFound && PositionalParametersFound) 4173 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4174 "used in macro body, possible positional parameter " 4175 "found in body which will have no effect"); 4176 } 4177 4178 /// parseDirectiveExitMacro 4179 /// ::= .exitm 4180 bool AsmParser::parseDirectiveExitMacro(StringRef Directive) { 4181 if (parseToken(AsmToken::EndOfStatement, 4182 "unexpected token in '" + Directive + "' directive")) 4183 return true; 4184 4185 if (!isInsideMacroInstantiation()) 4186 return TokError("unexpected '" + Directive + "' in file, " 4187 "no current macro definition"); 4188 4189 // Exit all conditionals that are active in the current macro. 4190 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4191 TheCondState = TheCondStack.back(); 4192 TheCondStack.pop_back(); 4193 } 4194 4195 handleMacroExit(); 4196 return false; 4197 } 4198 4199 /// parseDirectiveEndMacro 4200 /// ::= .endm 4201 /// ::= .endmacro 4202 bool AsmParser::parseDirectiveEndMacro(StringRef Directive) { 4203 if (getLexer().isNot(AsmToken::EndOfStatement)) 4204 return TokError("unexpected token in '" + Directive + "' directive"); 4205 4206 // If we are inside a macro instantiation, terminate the current 4207 // instantiation. 4208 if (isInsideMacroInstantiation()) { 4209 handleMacroExit(); 4210 return false; 4211 } 4212 4213 // Otherwise, this .endmacro is a stray entry in the file; well formed 4214 // .endmacro directives are handled during the macro definition parsing. 4215 return TokError("unexpected '" + Directive + "' in file, " 4216 "no current macro definition"); 4217 } 4218 4219 /// parseDirectivePurgeMacro 4220 /// ::= .purgem 4221 bool AsmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4222 StringRef Name; 4223 SMLoc Loc; 4224 if (parseTokenLoc(Loc) || 4225 check(parseIdentifier(Name), Loc, 4226 "expected identifier in '.purgem' directive") || 4227 parseToken(AsmToken::EndOfStatement, 4228 "unexpected token in '.purgem' directive")) 4229 return true; 4230 4231 if (!lookupMacro(Name)) 4232 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4233 4234 undefineMacro(Name); 4235 return false; 4236 } 4237 4238 /// parseDirectiveBundleAlignMode 4239 /// ::= {.bundle_align_mode} expression 4240 bool AsmParser::parseDirectiveBundleAlignMode() { 4241 // Expect a single argument: an expression that evaluates to a constant 4242 // in the inclusive range 0-30. 4243 SMLoc ExprLoc = getLexer().getLoc(); 4244 int64_t AlignSizePow2; 4245 if (checkForValidSection() || parseAbsoluteExpression(AlignSizePow2) || 4246 parseToken(AsmToken::EndOfStatement, "unexpected token after expression " 4247 "in '.bundle_align_mode' " 4248 "directive") || 4249 check(AlignSizePow2 < 0 || AlignSizePow2 > 30, ExprLoc, 4250 "invalid bundle alignment size (expected between 0 and 30)")) 4251 return true; 4252 4253 // Because of AlignSizePow2's verified range we can safely truncate it to 4254 // unsigned. 4255 getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); 4256 return false; 4257 } 4258 4259 /// parseDirectiveBundleLock 4260 /// ::= {.bundle_lock} [align_to_end] 4261 bool AsmParser::parseDirectiveBundleLock() { 4262 if (checkForValidSection()) 4263 return true; 4264 bool AlignToEnd = false; 4265 4266 StringRef Option; 4267 SMLoc Loc = getTok().getLoc(); 4268 const char *kInvalidOptionError = 4269 "invalid option for '.bundle_lock' directive"; 4270 4271 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4272 if (check(parseIdentifier(Option), Loc, kInvalidOptionError) || 4273 check(Option != "align_to_end", Loc, kInvalidOptionError) || 4274 parseToken(AsmToken::EndOfStatement, 4275 "unexpected token after '.bundle_lock' directive option")) 4276 return true; 4277 AlignToEnd = true; 4278 } 4279 4280 getStreamer().EmitBundleLock(AlignToEnd); 4281 return false; 4282 } 4283 4284 /// parseDirectiveBundleLock 4285 /// ::= {.bundle_lock} 4286 bool AsmParser::parseDirectiveBundleUnlock() { 4287 if (checkForValidSection() || 4288 parseToken(AsmToken::EndOfStatement, 4289 "unexpected token in '.bundle_unlock' directive")) 4290 return true; 4291 4292 getStreamer().EmitBundleUnlock(); 4293 return false; 4294 } 4295 4296 /// parseDirectiveSpace 4297 /// ::= (.skip | .space) expression [ , expression ] 4298 bool AsmParser::parseDirectiveSpace(StringRef IDVal) { 4299 SMLoc NumBytesLoc = Lexer.getLoc(); 4300 const MCExpr *NumBytes; 4301 if (checkForValidSection() || parseExpression(NumBytes)) 4302 return true; 4303 4304 int64_t FillExpr = 0; 4305 if (parseOptionalToken(AsmToken::Comma)) 4306 if (parseAbsoluteExpression(FillExpr)) 4307 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4308 if (parseToken(AsmToken::EndOfStatement)) 4309 return addErrorSuffix("in '" + Twine(IDVal) + "' directive"); 4310 4311 // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. 4312 getStreamer().emitFill(*NumBytes, FillExpr, NumBytesLoc); 4313 4314 return false; 4315 } 4316 4317 /// parseDirectiveDCB 4318 /// ::= .dcb.{b, l, w} expression, expression 4319 bool AsmParser::parseDirectiveDCB(StringRef IDVal, unsigned Size) { 4320 SMLoc NumValuesLoc = Lexer.getLoc(); 4321 int64_t NumValues; 4322 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4323 return true; 4324 4325 if (NumValues < 0) { 4326 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4327 return false; 4328 } 4329 4330 if (parseToken(AsmToken::Comma, 4331 "unexpected token in '" + Twine(IDVal) + "' directive")) 4332 return true; 4333 4334 const MCExpr *Value; 4335 SMLoc ExprLoc = getLexer().getLoc(); 4336 if (parseExpression(Value)) 4337 return true; 4338 4339 // Special case constant expressions to match code generator. 4340 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 4341 assert(Size <= 8 && "Invalid size"); 4342 uint64_t IntValue = MCE->getValue(); 4343 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 4344 return Error(ExprLoc, "literal value out of range for directive"); 4345 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4346 getStreamer().EmitIntValue(IntValue, Size); 4347 } else { 4348 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4349 getStreamer().EmitValue(Value, Size, ExprLoc); 4350 } 4351 4352 if (parseToken(AsmToken::EndOfStatement, 4353 "unexpected token in '" + Twine(IDVal) + "' directive")) 4354 return true; 4355 4356 return false; 4357 } 4358 4359 /// parseDirectiveRealDCB 4360 /// ::= .dcb.{d, s} expression, expression 4361 bool AsmParser::parseDirectiveRealDCB(StringRef IDVal, const fltSemantics &Semantics) { 4362 SMLoc NumValuesLoc = Lexer.getLoc(); 4363 int64_t NumValues; 4364 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4365 return true; 4366 4367 if (NumValues < 0) { 4368 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4369 return false; 4370 } 4371 4372 if (parseToken(AsmToken::Comma, 4373 "unexpected token in '" + Twine(IDVal) + "' directive")) 4374 return true; 4375 4376 APInt AsInt; 4377 if (parseRealValue(Semantics, AsInt)) 4378 return true; 4379 4380 if (parseToken(AsmToken::EndOfStatement, 4381 "unexpected token in '" + Twine(IDVal) + "' directive")) 4382 return true; 4383 4384 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4385 getStreamer().EmitIntValue(AsInt.getLimitedValue(), 4386 AsInt.getBitWidth() / 8); 4387 4388 return false; 4389 } 4390 4391 /// parseDirectiveDS 4392 /// ::= .ds.{b, d, l, p, s, w, x} expression 4393 bool AsmParser::parseDirectiveDS(StringRef IDVal, unsigned Size) { 4394 SMLoc NumValuesLoc = Lexer.getLoc(); 4395 int64_t NumValues; 4396 if (checkForValidSection() || parseAbsoluteExpression(NumValues)) 4397 return true; 4398 4399 if (NumValues < 0) { 4400 Warning(NumValuesLoc, "'" + Twine(IDVal) + "' directive with negative repeat count has no effect"); 4401 return false; 4402 } 4403 4404 if (parseToken(AsmToken::EndOfStatement, 4405 "unexpected token in '" + Twine(IDVal) + "' directive")) 4406 return true; 4407 4408 for (uint64_t i = 0, e = NumValues; i != e; ++i) 4409 getStreamer().emitFill(Size, 0); 4410 4411 return false; 4412 } 4413 4414 /// parseDirectiveLEB128 4415 /// ::= (.sleb128 | .uleb128) [ expression (, expression)* ] 4416 bool AsmParser::parseDirectiveLEB128(bool Signed) { 4417 if (checkForValidSection()) 4418 return true; 4419 4420 auto parseOp = [&]() -> bool { 4421 const MCExpr *Value; 4422 if (parseExpression(Value)) 4423 return true; 4424 if (Signed) 4425 getStreamer().EmitSLEB128Value(Value); 4426 else 4427 getStreamer().EmitULEB128Value(Value); 4428 return false; 4429 }; 4430 4431 if (parseMany(parseOp)) 4432 return addErrorSuffix(" in directive"); 4433 4434 return false; 4435 } 4436 4437 /// parseDirectiveSymbolAttribute 4438 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4439 bool AsmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4440 auto parseOp = [&]() -> bool { 4441 StringRef Name; 4442 SMLoc Loc = getTok().getLoc(); 4443 if (parseIdentifier(Name)) 4444 return Error(Loc, "expected identifier"); 4445 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4446 4447 // Assembler local symbols don't make any sense here. Complain loudly. 4448 if (Sym->isTemporary()) 4449 return Error(Loc, "non-local symbol required"); 4450 4451 if (!getStreamer().EmitSymbolAttribute(Sym, Attr)) 4452 return Error(Loc, "unable to emit symbol attribute"); 4453 return false; 4454 }; 4455 4456 if (parseMany(parseOp)) 4457 return addErrorSuffix(" in directive"); 4458 return false; 4459 } 4460 4461 /// parseDirectiveComm 4462 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4463 bool AsmParser::parseDirectiveComm(bool IsLocal) { 4464 if (checkForValidSection()) 4465 return true; 4466 4467 SMLoc IDLoc = getLexer().getLoc(); 4468 StringRef Name; 4469 if (parseIdentifier(Name)) 4470 return TokError("expected identifier in directive"); 4471 4472 // Handle the identifier as the key symbol. 4473 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4474 4475 if (getLexer().isNot(AsmToken::Comma)) 4476 return TokError("unexpected token in directive"); 4477 Lex(); 4478 4479 int64_t Size; 4480 SMLoc SizeLoc = getLexer().getLoc(); 4481 if (parseAbsoluteExpression(Size)) 4482 return true; 4483 4484 int64_t Pow2Alignment = 0; 4485 SMLoc Pow2AlignmentLoc; 4486 if (getLexer().is(AsmToken::Comma)) { 4487 Lex(); 4488 Pow2AlignmentLoc = getLexer().getLoc(); 4489 if (parseAbsoluteExpression(Pow2Alignment)) 4490 return true; 4491 4492 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4493 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4494 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4495 4496 // If this target takes alignments in bytes (not log) validate and convert. 4497 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4498 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4499 if (!isPowerOf2_64(Pow2Alignment)) 4500 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4501 Pow2Alignment = Log2_64(Pow2Alignment); 4502 } 4503 } 4504 4505 if (parseToken(AsmToken::EndOfStatement, 4506 "unexpected token in '.comm' or '.lcomm' directive")) 4507 return true; 4508 4509 // NOTE: a size of zero for a .comm should create a undefined symbol 4510 // but a size of .lcomm creates a bss symbol of size zero. 4511 if (Size < 0) 4512 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4513 "be less than zero"); 4514 4515 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4516 // may internally end up wanting an alignment in bytes. 4517 // FIXME: Diagnose overflow. 4518 if (Pow2Alignment < 0) 4519 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4520 "alignment, can't be less than zero"); 4521 4522 Sym->redefineIfPossible(); 4523 if (!Sym->isUndefined()) 4524 return Error(IDLoc, "invalid symbol redefinition"); 4525 4526 // Create the Symbol as a common or local common with Size and Pow2Alignment 4527 if (IsLocal) { 4528 getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4529 return false; 4530 } 4531 4532 getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4533 return false; 4534 } 4535 4536 /// parseDirectiveAbort 4537 /// ::= .abort [... message ...] 4538 bool AsmParser::parseDirectiveAbort() { 4539 // FIXME: Use loc from directive. 4540 SMLoc Loc = getLexer().getLoc(); 4541 4542 StringRef Str = parseStringToEndOfStatement(); 4543 if (parseToken(AsmToken::EndOfStatement, 4544 "unexpected token in '.abort' directive")) 4545 return true; 4546 4547 if (Str.empty()) 4548 return Error(Loc, ".abort detected. Assembly stopping."); 4549 else 4550 return Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); 4551 // FIXME: Actually abort assembly here. 4552 4553 return false; 4554 } 4555 4556 /// parseDirectiveInclude 4557 /// ::= .include "filename" 4558 bool AsmParser::parseDirectiveInclude() { 4559 // Allow the strings to have escaped octal character sequence. 4560 std::string Filename; 4561 SMLoc IncludeLoc = getTok().getLoc(); 4562 4563 if (check(getTok().isNot(AsmToken::String), 4564 "expected string in '.include' directive") || 4565 parseEscapedString(Filename) || 4566 check(getTok().isNot(AsmToken::EndOfStatement), 4567 "unexpected token in '.include' directive") || 4568 // Attempt to switch the lexer to the included file before consuming the 4569 // end of statement to avoid losing it when we switch. 4570 check(enterIncludeFile(Filename), IncludeLoc, 4571 "Could not find include file '" + Filename + "'")) 4572 return true; 4573 4574 return false; 4575 } 4576 4577 /// parseDirectiveIncbin 4578 /// ::= .incbin "filename" [ , skip [ , count ] ] 4579 bool AsmParser::parseDirectiveIncbin() { 4580 // Allow the strings to have escaped octal character sequence. 4581 std::string Filename; 4582 SMLoc IncbinLoc = getTok().getLoc(); 4583 if (check(getTok().isNot(AsmToken::String), 4584 "expected string in '.incbin' directive") || 4585 parseEscapedString(Filename)) 4586 return true; 4587 4588 int64_t Skip = 0; 4589 const MCExpr *Count = nullptr; 4590 SMLoc SkipLoc, CountLoc; 4591 if (parseOptionalToken(AsmToken::Comma)) { 4592 // The skip expression can be omitted while specifying the count, e.g: 4593 // .incbin "filename",,4 4594 if (getTok().isNot(AsmToken::Comma)) { 4595 if (parseTokenLoc(SkipLoc) || parseAbsoluteExpression(Skip)) 4596 return true; 4597 } 4598 if (parseOptionalToken(AsmToken::Comma)) { 4599 CountLoc = getTok().getLoc(); 4600 if (parseExpression(Count)) 4601 return true; 4602 } 4603 } 4604 4605 if (parseToken(AsmToken::EndOfStatement, 4606 "unexpected token in '.incbin' directive")) 4607 return true; 4608 4609 if (check(Skip < 0, SkipLoc, "skip is negative")) 4610 return true; 4611 4612 // Attempt to process the included file. 4613 if (processIncbinFile(Filename, Skip, Count, CountLoc)) 4614 return Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); 4615 return false; 4616 } 4617 4618 /// parseDirectiveIf 4619 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 4620 bool AsmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 4621 TheCondStack.push_back(TheCondState); 4622 TheCondState.TheCond = AsmCond::IfCond; 4623 if (TheCondState.Ignore) { 4624 eatToEndOfStatement(); 4625 } else { 4626 int64_t ExprValue; 4627 if (parseAbsoluteExpression(ExprValue) || 4628 parseToken(AsmToken::EndOfStatement, 4629 "unexpected token in '.if' directive")) 4630 return true; 4631 4632 switch (DirKind) { 4633 default: 4634 llvm_unreachable("unsupported directive"); 4635 case DK_IF: 4636 case DK_IFNE: 4637 break; 4638 case DK_IFEQ: 4639 ExprValue = ExprValue == 0; 4640 break; 4641 case DK_IFGE: 4642 ExprValue = ExprValue >= 0; 4643 break; 4644 case DK_IFGT: 4645 ExprValue = ExprValue > 0; 4646 break; 4647 case DK_IFLE: 4648 ExprValue = ExprValue <= 0; 4649 break; 4650 case DK_IFLT: 4651 ExprValue = ExprValue < 0; 4652 break; 4653 } 4654 4655 TheCondState.CondMet = ExprValue; 4656 TheCondState.Ignore = !TheCondState.CondMet; 4657 } 4658 4659 return false; 4660 } 4661 4662 /// parseDirectiveIfb 4663 /// ::= .ifb string 4664 bool AsmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4665 TheCondStack.push_back(TheCondState); 4666 TheCondState.TheCond = AsmCond::IfCond; 4667 4668 if (TheCondState.Ignore) { 4669 eatToEndOfStatement(); 4670 } else { 4671 StringRef Str = parseStringToEndOfStatement(); 4672 4673 if (parseToken(AsmToken::EndOfStatement, 4674 "unexpected token in '.ifb' directive")) 4675 return true; 4676 4677 TheCondState.CondMet = ExpectBlank == Str.empty(); 4678 TheCondState.Ignore = !TheCondState.CondMet; 4679 } 4680 4681 return false; 4682 } 4683 4684 /// parseDirectiveIfc 4685 /// ::= .ifc string1, string2 4686 /// ::= .ifnc string1, string2 4687 bool AsmParser::parseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { 4688 TheCondStack.push_back(TheCondState); 4689 TheCondState.TheCond = AsmCond::IfCond; 4690 4691 if (TheCondState.Ignore) { 4692 eatToEndOfStatement(); 4693 } else { 4694 StringRef Str1 = parseStringToComma(); 4695 4696 if (parseToken(AsmToken::Comma, "unexpected token in '.ifc' directive")) 4697 return true; 4698 4699 StringRef Str2 = parseStringToEndOfStatement(); 4700 4701 if (parseToken(AsmToken::EndOfStatement, 4702 "unexpected token in '.ifc' directive")) 4703 return true; 4704 4705 TheCondState.CondMet = ExpectEqual == (Str1.trim() == Str2.trim()); 4706 TheCondState.Ignore = !TheCondState.CondMet; 4707 } 4708 4709 return false; 4710 } 4711 4712 /// parseDirectiveIfeqs 4713 /// ::= .ifeqs string1, string2 4714 bool AsmParser::parseDirectiveIfeqs(SMLoc DirectiveLoc, bool ExpectEqual) { 4715 if (Lexer.isNot(AsmToken::String)) { 4716 if (ExpectEqual) 4717 return TokError("expected string parameter for '.ifeqs' directive"); 4718 return TokError("expected string parameter for '.ifnes' directive"); 4719 } 4720 4721 StringRef String1 = getTok().getStringContents(); 4722 Lex(); 4723 4724 if (Lexer.isNot(AsmToken::Comma)) { 4725 if (ExpectEqual) 4726 return TokError( 4727 "expected comma after first string for '.ifeqs' directive"); 4728 return TokError("expected comma after first string for '.ifnes' directive"); 4729 } 4730 4731 Lex(); 4732 4733 if (Lexer.isNot(AsmToken::String)) { 4734 if (ExpectEqual) 4735 return TokError("expected string parameter for '.ifeqs' directive"); 4736 return TokError("expected string parameter for '.ifnes' directive"); 4737 } 4738 4739 StringRef String2 = getTok().getStringContents(); 4740 Lex(); 4741 4742 TheCondStack.push_back(TheCondState); 4743 TheCondState.TheCond = AsmCond::IfCond; 4744 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4745 TheCondState.Ignore = !TheCondState.CondMet; 4746 4747 return false; 4748 } 4749 4750 /// parseDirectiveIfdef 4751 /// ::= .ifdef symbol 4752 bool AsmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 4753 StringRef Name; 4754 TheCondStack.push_back(TheCondState); 4755 TheCondState.TheCond = AsmCond::IfCond; 4756 4757 if (TheCondState.Ignore) { 4758 eatToEndOfStatement(); 4759 } else { 4760 if (check(parseIdentifier(Name), "expected identifier after '.ifdef'") || 4761 parseToken(AsmToken::EndOfStatement, "unexpected token in '.ifdef'")) 4762 return true; 4763 4764 MCSymbol *Sym = getContext().lookupSymbol(Name); 4765 4766 if (expect_defined) 4767 TheCondState.CondMet = (Sym && !Sym->isUndefined()); 4768 else 4769 TheCondState.CondMet = (!Sym || Sym->isUndefined()); 4770 TheCondState.Ignore = !TheCondState.CondMet; 4771 } 4772 4773 return false; 4774 } 4775 4776 /// parseDirectiveElseIf 4777 /// ::= .elseif expression 4778 bool AsmParser::parseDirectiveElseIf(SMLoc DirectiveLoc) { 4779 if (TheCondState.TheCond != AsmCond::IfCond && 4780 TheCondState.TheCond != AsmCond::ElseIfCond) 4781 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 4782 " .if or an .elseif"); 4783 TheCondState.TheCond = AsmCond::ElseIfCond; 4784 4785 bool LastIgnoreState = false; 4786 if (!TheCondStack.empty()) 4787 LastIgnoreState = TheCondStack.back().Ignore; 4788 if (LastIgnoreState || TheCondState.CondMet) { 4789 TheCondState.Ignore = true; 4790 eatToEndOfStatement(); 4791 } else { 4792 int64_t ExprValue; 4793 if (parseAbsoluteExpression(ExprValue)) 4794 return true; 4795 4796 if (parseToken(AsmToken::EndOfStatement, 4797 "unexpected token in '.elseif' directive")) 4798 return true; 4799 4800 TheCondState.CondMet = ExprValue; 4801 TheCondState.Ignore = !TheCondState.CondMet; 4802 } 4803 4804 return false; 4805 } 4806 4807 /// parseDirectiveElse 4808 /// ::= .else 4809 bool AsmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4810 if (parseToken(AsmToken::EndOfStatement, 4811 "unexpected token in '.else' directive")) 4812 return true; 4813 4814 if (TheCondState.TheCond != AsmCond::IfCond && 4815 TheCondState.TheCond != AsmCond::ElseIfCond) 4816 return Error(DirectiveLoc, "Encountered a .else that doesn't follow " 4817 " an .if or an .elseif"); 4818 TheCondState.TheCond = AsmCond::ElseCond; 4819 bool LastIgnoreState = false; 4820 if (!TheCondStack.empty()) 4821 LastIgnoreState = TheCondStack.back().Ignore; 4822 if (LastIgnoreState || TheCondState.CondMet) 4823 TheCondState.Ignore = true; 4824 else 4825 TheCondState.Ignore = false; 4826 4827 return false; 4828 } 4829 4830 /// parseDirectiveEnd 4831 /// ::= .end 4832 bool AsmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4833 if (parseToken(AsmToken::EndOfStatement, 4834 "unexpected token in '.end' directive")) 4835 return true; 4836 4837 while (Lexer.isNot(AsmToken::Eof)) 4838 Lexer.Lex(); 4839 4840 return false; 4841 } 4842 4843 /// parseDirectiveError 4844 /// ::= .err 4845 /// ::= .error [string] 4846 bool AsmParser::parseDirectiveError(SMLoc L, bool WithMessage) { 4847 if (!TheCondStack.empty()) { 4848 if (TheCondStack.back().Ignore) { 4849 eatToEndOfStatement(); 4850 return false; 4851 } 4852 } 4853 4854 if (!WithMessage) 4855 return Error(L, ".err encountered"); 4856 4857 StringRef Message = ".error directive invoked in source file"; 4858 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4859 if (Lexer.isNot(AsmToken::String)) 4860 return TokError(".error argument must be a string"); 4861 4862 Message = getTok().getStringContents(); 4863 Lex(); 4864 } 4865 4866 return Error(L, Message); 4867 } 4868 4869 /// parseDirectiveWarning 4870 /// ::= .warning [string] 4871 bool AsmParser::parseDirectiveWarning(SMLoc L) { 4872 if (!TheCondStack.empty()) { 4873 if (TheCondStack.back().Ignore) { 4874 eatToEndOfStatement(); 4875 return false; 4876 } 4877 } 4878 4879 StringRef Message = ".warning directive invoked in source file"; 4880 4881 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 4882 if (Lexer.isNot(AsmToken::String)) 4883 return TokError(".warning argument must be a string"); 4884 4885 Message = getTok().getStringContents(); 4886 Lex(); 4887 if (parseToken(AsmToken::EndOfStatement, 4888 "expected end of statement in '.warning' directive")) 4889 return true; 4890 } 4891 4892 return Warning(L, Message); 4893 } 4894 4895 /// parseDirectiveEndIf 4896 /// ::= .endif 4897 bool AsmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 4898 if (parseToken(AsmToken::EndOfStatement, 4899 "unexpected token in '.endif' directive")) 4900 return true; 4901 4902 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 4903 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 4904 "an .if or .else"); 4905 if (!TheCondStack.empty()) { 4906 TheCondState = TheCondStack.back(); 4907 TheCondStack.pop_back(); 4908 } 4909 4910 return false; 4911 } 4912 4913 void AsmParser::initializeDirectiveKindMap() { 4914 DirectiveKindMap[".set"] = DK_SET; 4915 DirectiveKindMap[".equ"] = DK_EQU; 4916 DirectiveKindMap[".equiv"] = DK_EQUIV; 4917 DirectiveKindMap[".ascii"] = DK_ASCII; 4918 DirectiveKindMap[".asciz"] = DK_ASCIZ; 4919 DirectiveKindMap[".string"] = DK_STRING; 4920 DirectiveKindMap[".byte"] = DK_BYTE; 4921 DirectiveKindMap[".short"] = DK_SHORT; 4922 DirectiveKindMap[".value"] = DK_VALUE; 4923 DirectiveKindMap[".2byte"] = DK_2BYTE; 4924 DirectiveKindMap[".long"] = DK_LONG; 4925 DirectiveKindMap[".int"] = DK_INT; 4926 DirectiveKindMap[".4byte"] = DK_4BYTE; 4927 DirectiveKindMap[".quad"] = DK_QUAD; 4928 DirectiveKindMap[".8byte"] = DK_8BYTE; 4929 DirectiveKindMap[".octa"] = DK_OCTA; 4930 DirectiveKindMap[".single"] = DK_SINGLE; 4931 DirectiveKindMap[".float"] = DK_FLOAT; 4932 DirectiveKindMap[".double"] = DK_DOUBLE; 4933 DirectiveKindMap[".align"] = DK_ALIGN; 4934 DirectiveKindMap[".align32"] = DK_ALIGN32; 4935 DirectiveKindMap[".balign"] = DK_BALIGN; 4936 DirectiveKindMap[".balignw"] = DK_BALIGNW; 4937 DirectiveKindMap[".balignl"] = DK_BALIGNL; 4938 DirectiveKindMap[".p2align"] = DK_P2ALIGN; 4939 DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; 4940 DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; 4941 DirectiveKindMap[".org"] = DK_ORG; 4942 DirectiveKindMap[".fill"] = DK_FILL; 4943 DirectiveKindMap[".zero"] = DK_ZERO; 4944 DirectiveKindMap[".extern"] = DK_EXTERN; 4945 DirectiveKindMap[".globl"] = DK_GLOBL; 4946 DirectiveKindMap[".global"] = DK_GLOBAL; 4947 DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; 4948 DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; 4949 DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; 4950 DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; 4951 DirectiveKindMap[".reference"] = DK_REFERENCE; 4952 DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; 4953 DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; 4954 DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; 4955 DirectiveKindMap[".comm"] = DK_COMM; 4956 DirectiveKindMap[".common"] = DK_COMMON; 4957 DirectiveKindMap[".lcomm"] = DK_LCOMM; 4958 DirectiveKindMap[".abort"] = DK_ABORT; 4959 DirectiveKindMap[".include"] = DK_INCLUDE; 4960 DirectiveKindMap[".incbin"] = DK_INCBIN; 4961 DirectiveKindMap[".code16"] = DK_CODE16; 4962 DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; 4963 DirectiveKindMap[".rept"] = DK_REPT; 4964 DirectiveKindMap[".rep"] = DK_REPT; 4965 DirectiveKindMap[".irp"] = DK_IRP; 4966 DirectiveKindMap[".irpc"] = DK_IRPC; 4967 DirectiveKindMap[".endr"] = DK_ENDR; 4968 DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; 4969 DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; 4970 DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; 4971 DirectiveKindMap[".if"] = DK_IF; 4972 DirectiveKindMap[".ifeq"] = DK_IFEQ; 4973 DirectiveKindMap[".ifge"] = DK_IFGE; 4974 DirectiveKindMap[".ifgt"] = DK_IFGT; 4975 DirectiveKindMap[".ifle"] = DK_IFLE; 4976 DirectiveKindMap[".iflt"] = DK_IFLT; 4977 DirectiveKindMap[".ifne"] = DK_IFNE; 4978 DirectiveKindMap[".ifb"] = DK_IFB; 4979 DirectiveKindMap[".ifnb"] = DK_IFNB; 4980 DirectiveKindMap[".ifc"] = DK_IFC; 4981 DirectiveKindMap[".ifeqs"] = DK_IFEQS; 4982 DirectiveKindMap[".ifnc"] = DK_IFNC; 4983 DirectiveKindMap[".ifnes"] = DK_IFNES; 4984 DirectiveKindMap[".ifdef"] = DK_IFDEF; 4985 DirectiveKindMap[".ifndef"] = DK_IFNDEF; 4986 DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; 4987 DirectiveKindMap[".elseif"] = DK_ELSEIF; 4988 DirectiveKindMap[".else"] = DK_ELSE; 4989 DirectiveKindMap[".end"] = DK_END; 4990 DirectiveKindMap[".endif"] = DK_ENDIF; 4991 DirectiveKindMap[".skip"] = DK_SKIP; 4992 DirectiveKindMap[".space"] = DK_SPACE; 4993 DirectiveKindMap[".file"] = DK_FILE; 4994 DirectiveKindMap[".line"] = DK_LINE; 4995 DirectiveKindMap[".loc"] = DK_LOC; 4996 DirectiveKindMap[".stabs"] = DK_STABS; 4997 DirectiveKindMap[".cv_file"] = DK_CV_FILE; 4998 DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 4999 DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5000 DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5001 DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5002 DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5003 DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5004 DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5005 DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5006 DirectiveKindMap[".sleb128"] = DK_SLEB128; 5007 DirectiveKindMap[".uleb128"] = DK_ULEB128; 5008 DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5009 DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5010 DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5011 DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5012 DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5013 DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5014 DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5015 DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5016 DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5017 DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5018 DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5019 DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5020 DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5021 DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5022 DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5023 DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5024 DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5025 DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5026 DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5027 DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5028 DirectiveKindMap[".macros_on"] = DK_MACROS_ON; 5029 DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; 5030 DirectiveKindMap[".macro"] = DK_MACRO; 5031 DirectiveKindMap[".exitm"] = DK_EXITM; 5032 DirectiveKindMap[".endm"] = DK_ENDM; 5033 DirectiveKindMap[".endmacro"] = DK_ENDMACRO; 5034 DirectiveKindMap[".purgem"] = DK_PURGEM; 5035 DirectiveKindMap[".err"] = DK_ERR; 5036 DirectiveKindMap[".error"] = DK_ERROR; 5037 DirectiveKindMap[".warning"] = DK_WARNING; 5038 DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5039 DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5040 DirectiveKindMap[".reloc"] = DK_RELOC; 5041 DirectiveKindMap[".dc"] = DK_DC; 5042 DirectiveKindMap[".dc.a"] = DK_DC_A; 5043 DirectiveKindMap[".dc.b"] = DK_DC_B; 5044 DirectiveKindMap[".dc.d"] = DK_DC_D; 5045 DirectiveKindMap[".dc.l"] = DK_DC_L; 5046 DirectiveKindMap[".dc.s"] = DK_DC_S; 5047 DirectiveKindMap[".dc.w"] = DK_DC_W; 5048 DirectiveKindMap[".dc.x"] = DK_DC_X; 5049 DirectiveKindMap[".dcb"] = DK_DCB; 5050 DirectiveKindMap[".dcb.b"] = DK_DCB_B; 5051 DirectiveKindMap[".dcb.d"] = DK_DCB_D; 5052 DirectiveKindMap[".dcb.l"] = DK_DCB_L; 5053 DirectiveKindMap[".dcb.s"] = DK_DCB_S; 5054 DirectiveKindMap[".dcb.w"] = DK_DCB_W; 5055 DirectiveKindMap[".dcb.x"] = DK_DCB_X; 5056 DirectiveKindMap[".ds"] = DK_DS; 5057 DirectiveKindMap[".ds.b"] = DK_DS_B; 5058 DirectiveKindMap[".ds.d"] = DK_DS_D; 5059 DirectiveKindMap[".ds.l"] = DK_DS_L; 5060 DirectiveKindMap[".ds.p"] = DK_DS_P; 5061 DirectiveKindMap[".ds.s"] = DK_DS_S; 5062 DirectiveKindMap[".ds.w"] = DK_DS_W; 5063 DirectiveKindMap[".ds.x"] = DK_DS_X; 5064 } 5065 5066 MCAsmMacro *AsmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5067 AsmToken EndToken, StartToken = getTok(); 5068 5069 unsigned NestLevel = 0; 5070 while (true) { 5071 // Check whether we have reached the end of the file. 5072 if (getLexer().is(AsmToken::Eof)) { 5073 printError(DirectiveLoc, "no matching '.endr' in definition"); 5074 return nullptr; 5075 } 5076 5077 if (Lexer.is(AsmToken::Identifier) && 5078 (getTok().getIdentifier() == ".rept" || 5079 getTok().getIdentifier() == ".irp" || 5080 getTok().getIdentifier() == ".irpc")) { 5081 ++NestLevel; 5082 } 5083 5084 // Otherwise, check whether we have reached the .endr. 5085 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5086 if (NestLevel == 0) { 5087 EndToken = getTok(); 5088 Lex(); 5089 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5090 printError(getTok().getLoc(), 5091 "unexpected token in '.endr' directive"); 5092 return nullptr; 5093 } 5094 break; 5095 } 5096 --NestLevel; 5097 } 5098 5099 // Otherwise, scan till the end of the statement. 5100 eatToEndOfStatement(); 5101 } 5102 5103 const char *BodyStart = StartToken.getLoc().getPointer(); 5104 const char *BodyEnd = EndToken.getLoc().getPointer(); 5105 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5106 5107 // We Are Anonymous. 5108 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5109 return &MacroLikeBodies.back(); 5110 } 5111 5112 void AsmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5113 raw_svector_ostream &OS) { 5114 OS << ".endr\n"; 5115 5116 std::unique_ptr<MemoryBuffer> Instantiation = 5117 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5118 5119 // Create the macro instantiation object and add to the current macro 5120 // instantiation stack. 5121 MacroInstantiation *MI = new MacroInstantiation( 5122 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()); 5123 ActiveMacros.push_back(MI); 5124 5125 // Jump to the macro instantiation and prime the lexer. 5126 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5127 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5128 Lex(); 5129 } 5130 5131 /// parseDirectiveRept 5132 /// ::= .rep | .rept count 5133 bool AsmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5134 const MCExpr *CountExpr; 5135 SMLoc CountLoc = getTok().getLoc(); 5136 if (parseExpression(CountExpr)) 5137 return true; 5138 5139 int64_t Count; 5140 if (!CountExpr->evaluateAsAbsolute(Count)) { 5141 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5142 } 5143 5144 if (check(Count < 0, CountLoc, "Count is negative") || 5145 parseToken(AsmToken::EndOfStatement, 5146 "unexpected token in '" + Dir + "' directive")) 5147 return true; 5148 5149 // Lex the rept definition. 5150 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5151 if (!M) 5152 return true; 5153 5154 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5155 // to hold the macro body with substitutions. 5156 SmallString<256> Buf; 5157 raw_svector_ostream OS(Buf); 5158 while (Count--) { 5159 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5160 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5161 return true; 5162 } 5163 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5164 5165 return false; 5166 } 5167 5168 /// parseDirectiveIrp 5169 /// ::= .irp symbol,values 5170 bool AsmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5171 MCAsmMacroParameter Parameter; 5172 MCAsmMacroArguments A; 5173 if (check(parseIdentifier(Parameter.Name), 5174 "expected identifier in '.irp' directive") || 5175 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5176 parseMacroArguments(nullptr, A) || 5177 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5178 return true; 5179 5180 // Lex the irp definition. 5181 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5182 if (!M) 5183 return true; 5184 5185 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5186 // to hold the macro body with substitutions. 5187 SmallString<256> Buf; 5188 raw_svector_ostream OS(Buf); 5189 5190 for (const MCAsmMacroArgument &Arg : A) { 5191 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5192 // This is undocumented, but GAS seems to support it. 5193 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5194 return true; 5195 } 5196 5197 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5198 5199 return false; 5200 } 5201 5202 /// parseDirectiveIrpc 5203 /// ::= .irpc symbol,values 5204 bool AsmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5205 MCAsmMacroParameter Parameter; 5206 MCAsmMacroArguments A; 5207 5208 if (check(parseIdentifier(Parameter.Name), 5209 "expected identifier in '.irpc' directive") || 5210 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5211 parseMacroArguments(nullptr, A)) 5212 return true; 5213 5214 if (A.size() != 1 || A.front().size() != 1) 5215 return TokError("unexpected token in '.irpc' directive"); 5216 5217 // Eat the end of statement. 5218 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5219 return true; 5220 5221 // Lex the irpc definition. 5222 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5223 if (!M) 5224 return true; 5225 5226 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5227 // to hold the macro body with substitutions. 5228 SmallString<256> Buf; 5229 raw_svector_ostream OS(Buf); 5230 5231 StringRef Values = A.front().front().getString(); 5232 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5233 MCAsmMacroArgument Arg; 5234 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5235 5236 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5237 // This is undocumented, but GAS seems to support it. 5238 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5239 return true; 5240 } 5241 5242 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5243 5244 return false; 5245 } 5246 5247 bool AsmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5248 if (ActiveMacros.empty()) 5249 return TokError("unmatched '.endr' directive"); 5250 5251 // The only .repl that should get here are the ones created by 5252 // instantiateMacroLikeBody. 5253 assert(getLexer().is(AsmToken::EndOfStatement)); 5254 5255 handleMacroExit(); 5256 return false; 5257 } 5258 5259 bool AsmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5260 size_t Len) { 5261 const MCExpr *Value; 5262 SMLoc ExprLoc = getLexer().getLoc(); 5263 if (parseExpression(Value)) 5264 return true; 5265 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5266 if (!MCE) 5267 return Error(ExprLoc, "unexpected expression in _emit"); 5268 uint64_t IntValue = MCE->getValue(); 5269 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5270 return Error(ExprLoc, "literal value out of range for directive"); 5271 5272 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5273 return false; 5274 } 5275 5276 bool AsmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5277 const MCExpr *Value; 5278 SMLoc ExprLoc = getLexer().getLoc(); 5279 if (parseExpression(Value)) 5280 return true; 5281 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5282 if (!MCE) 5283 return Error(ExprLoc, "unexpected expression in align"); 5284 uint64_t IntValue = MCE->getValue(); 5285 if (!isPowerOf2_64(IntValue)) 5286 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5287 5288 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5289 return false; 5290 } 5291 5292 // We are comparing pointers, but the pointers are relative to a single string. 5293 // Thus, this should always be deterministic. 5294 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5295 const AsmRewrite *AsmRewriteB) { 5296 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5297 return -1; 5298 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5299 return 1; 5300 5301 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5302 // rewrite to the same location. Make sure the SizeDirective rewrite is 5303 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5304 // ensures the sort algorithm is stable. 5305 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5306 AsmRewritePrecedence[AsmRewriteB->Kind]) 5307 return -1; 5308 5309 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5310 AsmRewritePrecedence[AsmRewriteB->Kind]) 5311 return 1; 5312 llvm_unreachable("Unstable rewrite sort."); 5313 } 5314 5315 bool AsmParser::parseMSInlineAsm( 5316 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5317 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5318 SmallVectorImpl<std::string> &Constraints, 5319 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5320 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5321 SmallVector<void *, 4> InputDecls; 5322 SmallVector<void *, 4> OutputDecls; 5323 SmallVector<bool, 4> InputDeclsAddressOf; 5324 SmallVector<bool, 4> OutputDeclsAddressOf; 5325 SmallVector<std::string, 4> InputConstraints; 5326 SmallVector<std::string, 4> OutputConstraints; 5327 SmallVector<unsigned, 4> ClobberRegs; 5328 5329 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5330 5331 // Prime the lexer. 5332 Lex(); 5333 5334 // While we have input, parse each statement. 5335 unsigned InputIdx = 0; 5336 unsigned OutputIdx = 0; 5337 while (getLexer().isNot(AsmToken::Eof)) { 5338 // Parse curly braces marking block start/end 5339 if (parseCurlyBlockScope(AsmStrRewrites)) 5340 continue; 5341 5342 ParseStatementInfo Info(&AsmStrRewrites); 5343 bool StatementErr = parseStatement(Info, &SI); 5344 5345 if (StatementErr || Info.ParseError) { 5346 // Emit pending errors if any exist. 5347 printPendingErrors(); 5348 return true; 5349 } 5350 5351 // No pending error should exist here. 5352 assert(!hasPendingError() && "unexpected error from parseStatement"); 5353 5354 if (Info.Opcode == ~0U) 5355 continue; 5356 5357 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5358 5359 // Build the list of clobbers, outputs and inputs. 5360 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5361 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5362 5363 // Immediate. 5364 if (Operand.isImm()) 5365 continue; 5366 5367 // Register operand. 5368 if (Operand.isReg() && !Operand.needAddressOf() && 5369 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5370 unsigned NumDefs = Desc.getNumDefs(); 5371 // Clobber. 5372 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5373 ClobberRegs.push_back(Operand.getReg()); 5374 continue; 5375 } 5376 5377 // Expr/Input or Output. 5378 StringRef SymName = Operand.getSymName(); 5379 if (SymName.empty()) 5380 continue; 5381 5382 void *OpDecl = Operand.getOpDecl(); 5383 if (!OpDecl) 5384 continue; 5385 5386 bool isOutput = (i == 1) && Desc.mayStore(); 5387 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5388 if (isOutput) { 5389 ++InputIdx; 5390 OutputDecls.push_back(OpDecl); 5391 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5392 OutputConstraints.push_back(("=" + Operand.getConstraint()).str()); 5393 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5394 } else { 5395 InputDecls.push_back(OpDecl); 5396 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5397 InputConstraints.push_back(Operand.getConstraint().str()); 5398 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5399 } 5400 } 5401 5402 // Consider implicit defs to be clobbers. Think of cpuid and push. 5403 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5404 Desc.getNumImplicitDefs()); 5405 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5406 } 5407 5408 // Set the number of Outputs and Inputs. 5409 NumOutputs = OutputDecls.size(); 5410 NumInputs = InputDecls.size(); 5411 5412 // Set the unique clobbers. 5413 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5414 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5415 ClobberRegs.end()); 5416 Clobbers.assign(ClobberRegs.size(), std::string()); 5417 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5418 raw_string_ostream OS(Clobbers[I]); 5419 IP->printRegName(OS, ClobberRegs[I]); 5420 } 5421 5422 // Merge the various outputs and inputs. Output are expected first. 5423 if (NumOutputs || NumInputs) { 5424 unsigned NumExprs = NumOutputs + NumInputs; 5425 OpDecls.resize(NumExprs); 5426 Constraints.resize(NumExprs); 5427 for (unsigned i = 0; i < NumOutputs; ++i) { 5428 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5429 Constraints[i] = OutputConstraints[i]; 5430 } 5431 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5432 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5433 Constraints[j] = InputConstraints[i]; 5434 } 5435 } 5436 5437 // Build the IR assembly string. 5438 std::string AsmStringIR; 5439 raw_string_ostream OS(AsmStringIR); 5440 StringRef ASMString = 5441 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5442 const char *AsmStart = ASMString.begin(); 5443 const char *AsmEnd = ASMString.end(); 5444 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5445 for (const AsmRewrite &AR : AsmStrRewrites) { 5446 AsmRewriteKind Kind = AR.Kind; 5447 if (Kind == AOK_Delete) 5448 continue; 5449 5450 const char *Loc = AR.Loc.getPointer(); 5451 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5452 5453 // Emit everything up to the immediate/expression. 5454 if (unsigned Len = Loc - AsmStart) 5455 OS << StringRef(AsmStart, Len); 5456 5457 // Skip the original expression. 5458 if (Kind == AOK_Skip) { 5459 AsmStart = Loc + AR.Len; 5460 continue; 5461 } 5462 5463 unsigned AdditionalSkip = 0; 5464 // Rewrite expressions in $N notation. 5465 switch (Kind) { 5466 default: 5467 break; 5468 case AOK_Imm: 5469 OS << "$$" << AR.Val; 5470 break; 5471 case AOK_ImmPrefix: 5472 OS << "$$"; 5473 break; 5474 case AOK_Label: 5475 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5476 break; 5477 case AOK_Input: 5478 OS << '$' << InputIdx++; 5479 break; 5480 case AOK_Output: 5481 OS << '$' << OutputIdx++; 5482 break; 5483 case AOK_SizeDirective: 5484 switch (AR.Val) { 5485 default: break; 5486 case 8: OS << "byte ptr "; break; 5487 case 16: OS << "word ptr "; break; 5488 case 32: OS << "dword ptr "; break; 5489 case 64: OS << "qword ptr "; break; 5490 case 80: OS << "xword ptr "; break; 5491 case 128: OS << "xmmword ptr "; break; 5492 case 256: OS << "ymmword ptr "; break; 5493 } 5494 break; 5495 case AOK_Emit: 5496 OS << ".byte"; 5497 break; 5498 case AOK_Align: { 5499 // MS alignment directives are measured in bytes. If the native assembler 5500 // measures alignment in bytes, we can pass it straight through. 5501 OS << ".align"; 5502 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5503 break; 5504 5505 // Alignment is in log2 form, so print that instead and skip the original 5506 // immediate. 5507 unsigned Val = AR.Val; 5508 OS << ' ' << Val; 5509 assert(Val < 10 && "Expected alignment less then 2^10."); 5510 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5511 break; 5512 } 5513 case AOK_EVEN: 5514 OS << ".even"; 5515 break; 5516 case AOK_DotOperator: 5517 // Insert the dot if the user omitted it. 5518 OS.flush(); 5519 if (AsmStringIR.back() != '.') 5520 OS << '.'; 5521 OS << AR.Val; 5522 break; 5523 case AOK_EndOfStatement: 5524 OS << "\n\t"; 5525 break; 5526 } 5527 5528 // Skip the original expression. 5529 AsmStart = Loc + AR.Len + AdditionalSkip; 5530 } 5531 5532 // Emit the remainder of the asm string. 5533 if (AsmStart != AsmEnd) 5534 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5535 5536 AsmString = OS.str(); 5537 return false; 5538 } 5539 5540 namespace llvm { 5541 namespace MCParserUtils { 5542 5543 /// Returns whether the given symbol is used anywhere in the given expression, 5544 /// or subexpressions. 5545 static bool isSymbolUsedInExpression(const MCSymbol *Sym, const MCExpr *Value) { 5546 switch (Value->getKind()) { 5547 case MCExpr::Binary: { 5548 const MCBinaryExpr *BE = static_cast<const MCBinaryExpr *>(Value); 5549 return isSymbolUsedInExpression(Sym, BE->getLHS()) || 5550 isSymbolUsedInExpression(Sym, BE->getRHS()); 5551 } 5552 case MCExpr::Target: 5553 case MCExpr::Constant: 5554 return false; 5555 case MCExpr::SymbolRef: { 5556 const MCSymbol &S = 5557 static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); 5558 if (S.isVariable()) 5559 return isSymbolUsedInExpression(Sym, S.getVariableValue()); 5560 return &S == Sym; 5561 } 5562 case MCExpr::Unary: 5563 return isSymbolUsedInExpression( 5564 Sym, static_cast<const MCUnaryExpr *>(Value)->getSubExpr()); 5565 } 5566 5567 llvm_unreachable("Unknown expr kind!"); 5568 } 5569 5570 bool parseAssignmentExpression(StringRef Name, bool allow_redef, 5571 MCAsmParser &Parser, MCSymbol *&Sym, 5572 const MCExpr *&Value) { 5573 5574 // FIXME: Use better location, we should use proper tokens. 5575 SMLoc EqualLoc = Parser.getTok().getLoc(); 5576 5577 if (Parser.parseExpression(Value)) { 5578 return Parser.TokError("missing expression"); 5579 } 5580 5581 // Note: we don't count b as used in "a = b". This is to allow 5582 // a = b 5583 // b = c 5584 5585 if (Parser.parseToken(AsmToken::EndOfStatement)) 5586 return true; 5587 5588 // Validate that the LHS is allowed to be a variable (either it has not been 5589 // used as a symbol, or it is an absolute symbol). 5590 Sym = Parser.getContext().lookupSymbol(Name); 5591 if (Sym) { 5592 // Diagnose assignment to a label. 5593 // 5594 // FIXME: Diagnostics. Note the location of the definition as a label. 5595 // FIXME: Diagnose assignment to protected identifier (e.g., register name). 5596 if (isSymbolUsedInExpression(Sym, Value)) 5597 return Parser.Error(EqualLoc, "Recursive use of '" + Name + "'"); 5598 else if (Sym->isUndefined(/*SetUsed*/ false) && !Sym->isUsed() && 5599 !Sym->isVariable()) 5600 ; // Allow redefinitions of undefined symbols only used in directives. 5601 else if (Sym->isVariable() && !Sym->isUsed() && allow_redef) 5602 ; // Allow redefinitions of variables that haven't yet been used. 5603 else if (!Sym->isUndefined() && (!Sym->isVariable() || !allow_redef)) 5604 return Parser.Error(EqualLoc, "redefinition of '" + Name + "'"); 5605 else if (!Sym->isVariable()) 5606 return Parser.Error(EqualLoc, "invalid assignment to '" + Name + "'"); 5607 else if (!isa<MCConstantExpr>(Sym->getVariableValue())) 5608 return Parser.Error(EqualLoc, 5609 "invalid reassignment of non-absolute variable '" + 5610 Name + "'"); 5611 } else if (Name == ".") { 5612 Parser.getStreamer().emitValueToOffset(Value, 0, EqualLoc); 5613 return false; 5614 } else 5615 Sym = Parser.getContext().getOrCreateSymbol(Name); 5616 5617 Sym->setRedefinable(allow_redef); 5618 5619 return false; 5620 } 5621 5622 } // end namespace MCParserUtils 5623 } // end namespace llvm 5624 5625 /// \brief Create an MCAsmParser instance. 5626 MCAsmParser *llvm::createMCAsmParser(SourceMgr &SM, MCContext &C, 5627 MCStreamer &Out, const MCAsmInfo &MAI, 5628 unsigned CB) { 5629 return new AsmParser(SM, C, Out, MAI, CB); 5630 } 5631