1 //===- AsmParser.cpp - Parser for Assembly Files --------------------------===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 // 9 // This class implements the parser for assembly files. 10 // 11 //===----------------------------------------------------------------------===// 12 13 #include "llvm/ADT/APFloat.h" 14 #include "llvm/ADT/APInt.h" 15 #include "llvm/ADT/ArrayRef.h" 16 #include "llvm/ADT/None.h" 17 #include "llvm/ADT/STLExtras.h" 18 #include "llvm/ADT/SmallString.h" 19 #include "llvm/ADT/SmallVector.h" 20 #include "llvm/ADT/StringExtras.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/DebugInfo/CodeView/SymbolRecord.h" 26 #include "llvm/MC/MCAsmInfo.h" 27 #include "llvm/MC/MCCodeView.h" 28 #include "llvm/MC/MCContext.h" 29 #include "llvm/MC/MCDirectives.h" 30 #include "llvm/MC/MCDwarf.h" 31 #include "llvm/MC/MCExpr.h" 32 #include "llvm/MC/MCInstPrinter.h" 33 #include "llvm/MC/MCInstrDesc.h" 34 #include "llvm/MC/MCInstrInfo.h" 35 #include "llvm/MC/MCObjectFileInfo.h" 36 #include "llvm/MC/MCParser/AsmCond.h" 37 #include "llvm/MC/MCParser/AsmLexer.h" 38 #include "llvm/MC/MCParser/MCAsmLexer.h" 39 #include "llvm/MC/MCParser/MCAsmParser.h" 40 #include "llvm/MC/MCParser/MCAsmParserExtension.h" 41 #include "llvm/MC/MCParser/MCAsmParserUtils.h" 42 #include "llvm/MC/MCParser/MCParsedAsmOperand.h" 43 #include "llvm/MC/MCParser/MCTargetAsmParser.h" 44 #include "llvm/MC/MCRegisterInfo.h" 45 #include "llvm/MC/MCSection.h" 46 #include "llvm/MC/MCStreamer.h" 47 #include "llvm/MC/MCSymbol.h" 48 #include "llvm/MC/MCTargetOptions.h" 49 #include "llvm/MC/MCValue.h" 50 #include "llvm/Support/Casting.h" 51 #include "llvm/Support/CommandLine.h" 52 #include "llvm/Support/ErrorHandling.h" 53 #include "llvm/Support/MD5.h" 54 #include "llvm/Support/MathExtras.h" 55 #include "llvm/Support/MemoryBuffer.h" 56 #include "llvm/Support/SMLoc.h" 57 #include "llvm/Support/SourceMgr.h" 58 #include "llvm/Support/raw_ostream.h" 59 #include <algorithm> 60 #include <cassert> 61 #include <cctype> 62 #include <climits> 63 #include <cstddef> 64 #include <cstdint> 65 #include <deque> 66 #include <memory> 67 #include <sstream> 68 #include <string> 69 #include <tuple> 70 #include <utility> 71 #include <vector> 72 73 using namespace llvm; 74 75 extern cl::opt<unsigned> AsmMacroMaxNestingDepth; 76 77 namespace { 78 79 /// Helper types for tracking macro definitions. 80 typedef std::vector<AsmToken> MCAsmMacroArgument; 81 typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; 82 83 /// Helper class for storing information about an active macro 84 /// instantiation. 85 struct MacroInstantiation { 86 /// The location of the instantiation. 87 SMLoc InstantiationLoc; 88 89 /// The buffer where parsing should resume upon instantiation completion. 90 unsigned ExitBuffer; 91 92 /// The location where parsing should resume upon instantiation completion. 93 SMLoc ExitLoc; 94 95 /// The depth of TheCondStack at the start of the instantiation. 96 size_t CondStackDepth; 97 }; 98 99 struct ParseStatementInfo { 100 /// The parsed operands from the last parsed statement. 101 SmallVector<std::unique_ptr<MCParsedAsmOperand>, 8> ParsedOperands; 102 103 /// The opcode from the last parsed instruction. 104 unsigned Opcode = ~0U; 105 106 /// Was there an error parsing the inline assembly? 107 bool ParseError = false; 108 109 SmallVectorImpl<AsmRewrite> *AsmRewrites = nullptr; 110 111 ParseStatementInfo() = delete; 112 ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) 113 : AsmRewrites(rewrites) {} 114 }; 115 116 /// The concrete assembly parser instance. 117 // Note that this is a full MCAsmParser, not an MCAsmParserExtension! 118 // It's a peer of AsmParser, not of COFFAsmParser, WasmAsmParser, etc. 119 class MasmParser : public MCAsmParser { 120 private: 121 AsmLexer Lexer; 122 MCContext &Ctx; 123 MCStreamer &Out; 124 const MCAsmInfo &MAI; 125 SourceMgr &SrcMgr; 126 SourceMgr::DiagHandlerTy SavedDiagHandler; 127 void *SavedDiagContext; 128 std::unique_ptr<MCAsmParserExtension> PlatformParser; 129 130 /// This is the current buffer index we're lexing from as managed by the 131 /// SourceMgr object. 132 unsigned CurBuffer; 133 134 AsmCond TheCondState; 135 std::vector<AsmCond> TheCondStack; 136 137 /// maps directive names to handler methods in parser 138 /// extensions. Extensions register themselves in this map by calling 139 /// addDirectiveHandler. 140 StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; 141 142 /// maps assembly-time variable names to variables 143 struct Variable { 144 StringRef Name; 145 bool Redefinable = true; 146 bool IsText = false; 147 int64_t NumericValue = 0; 148 std::string TextValue; 149 }; 150 StringMap<Variable> Variables; 151 152 /// Stack of active macro instantiations. 153 std::vector<MacroInstantiation*> ActiveMacros; 154 155 /// List of bodies of anonymous macros. 156 std::deque<MCAsmMacro> MacroLikeBodies; 157 158 /// Keeps track of how many .macro's have been instantiated. 159 unsigned NumOfMacroInstantiations; 160 161 /// The values from the last parsed cpp hash file line comment if any. 162 struct CppHashInfoTy { 163 StringRef Filename; 164 int64_t LineNumber; 165 SMLoc Loc; 166 unsigned Buf; 167 CppHashInfoTy() : Filename(), LineNumber(0), Loc(), Buf(0) {} 168 }; 169 CppHashInfoTy CppHashInfo; 170 171 /// The filename from the first cpp hash file line comment, if any. 172 StringRef FirstCppHashFilename; 173 174 /// List of forward directional labels for diagnosis at the end. 175 SmallVector<std::tuple<SMLoc, CppHashInfoTy, MCSymbol *>, 4> DirLabels; 176 177 /// AssemblerDialect. ~OU means unset value and use value provided by MAI. 178 /// Defaults to 1U, meaning Intel. 179 unsigned AssemblerDialect = 1U; 180 181 /// is Darwin compatibility enabled? 182 bool IsDarwin = false; 183 184 /// Are we parsing ms-style inline assembly? 185 bool ParsingMSInlineAsm = false; 186 187 /// Did we already inform the user about inconsistent MD5 usage? 188 bool ReportedInconsistentMD5 = false; 189 190 // Is alt macro mode enabled. 191 bool AltMacroMode = false; 192 193 public: 194 MasmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 195 const MCAsmInfo &MAI, unsigned CB); 196 MasmParser(const MasmParser &) = delete; 197 MasmParser &operator=(const MasmParser &) = delete; 198 ~MasmParser() override; 199 200 bool Run(bool NoInitialTextSection, bool NoFinalize = false) override; 201 202 void addDirectiveHandler(StringRef Directive, 203 ExtensionDirectiveHandler Handler) override { 204 ExtensionDirectiveMap[Directive] = Handler; 205 if (DirectiveKindMap.find(Directive) == DirectiveKindMap.end()) { 206 DirectiveKindMap[Directive] = DK_HANDLER_DIRECTIVE; 207 } 208 } 209 210 void addAliasForDirective(StringRef Directive, StringRef Alias) override { 211 DirectiveKindMap[Directive] = DirectiveKindMap[Alias]; 212 } 213 214 /// @name MCAsmParser Interface 215 /// { 216 217 SourceMgr &getSourceManager() override { return SrcMgr; } 218 MCAsmLexer &getLexer() override { return Lexer; } 219 MCContext &getContext() override { return Ctx; } 220 MCStreamer &getStreamer() override { return Out; } 221 222 CodeViewContext &getCVContext() { return Ctx.getCVContext(); } 223 224 unsigned getAssemblerDialect() override { 225 if (AssemblerDialect == ~0U) 226 return MAI.getAssemblerDialect(); 227 else 228 return AssemblerDialect; 229 } 230 void setAssemblerDialect(unsigned i) override { 231 AssemblerDialect = i; 232 } 233 234 void Note(SMLoc L, const Twine &Msg, SMRange Range = None) override; 235 bool Warning(SMLoc L, const Twine &Msg, SMRange Range = None) override; 236 bool printError(SMLoc L, const Twine &Msg, SMRange Range = None) override; 237 238 const AsmToken &Lex() override; 239 240 void setParsingMSInlineAsm(bool V) override { 241 ParsingMSInlineAsm = V; 242 // When parsing MS inline asm, we must lex 0b1101 and 0ABCH as binary and 243 // hex integer literals. 244 Lexer.setLexMasmIntegers(V); 245 } 246 bool isParsingMSInlineAsm() override { return ParsingMSInlineAsm; } 247 248 bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, 249 unsigned &NumOutputs, unsigned &NumInputs, 250 SmallVectorImpl<std::pair<void *,bool>> &OpDecls, 251 SmallVectorImpl<std::string> &Constraints, 252 SmallVectorImpl<std::string> &Clobbers, 253 const MCInstrInfo *MII, const MCInstPrinter *IP, 254 MCAsmParserSemaCallback &SI) override; 255 256 bool parseExpression(const MCExpr *&Res); 257 bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 258 bool parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) override; 259 bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) override; 260 bool parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 261 SMLoc &EndLoc) override; 262 bool parseAbsoluteExpression(int64_t &Res) override; 263 264 /// Parse a floating point expression using the float \p Semantics 265 /// and set \p Res to the value. 266 bool parseRealValue(const fltSemantics &Semantics, APInt &Res); 267 268 /// Parse an identifier or string (as a quoted identifier) 269 /// and set \p Res to the identifier contents. 270 bool parseIdentifier(StringRef &Res) override; 271 void eatToEndOfStatement() override; 272 273 bool checkForValidSection() override; 274 275 /// } 276 277 private: 278 bool parseStatement(ParseStatementInfo &Info, 279 MCAsmParserSemaCallback *SI); 280 bool parseCurlyBlockScope(SmallVectorImpl<AsmRewrite>& AsmStrRewrites); 281 bool parseCppHashLineFilenameComment(SMLoc L); 282 283 void checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, 284 ArrayRef<MCAsmMacroParameter> Parameters); 285 bool expandMacro(raw_svector_ostream &OS, StringRef Body, 286 ArrayRef<MCAsmMacroParameter> Parameters, 287 ArrayRef<MCAsmMacroArgument> A, bool EnableAtPseudoVariable, 288 SMLoc L); 289 290 /// Are we inside a macro instantiation? 291 bool isInsideMacroInstantiation() {return !ActiveMacros.empty();} 292 293 /// Handle entry to macro instantiation. 294 /// 295 /// \param M The macro. 296 /// \param NameLoc Instantiation location. 297 bool handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); 298 299 /// Handle exit from macro instantiation. 300 void handleMacroExit(); 301 302 /// Extract AsmTokens for a macro argument. 303 bool parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg); 304 305 /// Parse all macro arguments for a given macro. 306 bool parseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); 307 308 void printMacroInstantiations(); 309 void printMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, 310 SMRange Range = None) const { 311 ArrayRef<SMRange> Ranges(Range); 312 SrcMgr.PrintMessage(Loc, Kind, Msg, Ranges); 313 } 314 static void DiagHandler(const SMDiagnostic &Diag, void *Context); 315 316 /// Should we emit DWARF describing this assembler source? (Returns false if 317 /// the source has .file directives, which means we don't want to generate 318 /// info describing the assembler source itself.) 319 bool enabledGenDwarfForAssembly(); 320 321 /// Enter the specified file. This returns true on failure. 322 bool enterIncludeFile(const std::string &Filename); 323 324 /// Reset the current lexer position to that given by \p Loc. The 325 /// current token is not set; clients should ensure Lex() is called 326 /// subsequently. 327 /// 328 /// \param InBuffer If not 0, should be the known buffer id that contains the 329 /// location. 330 void jumpToLoc(SMLoc Loc, unsigned InBuffer = 0); 331 332 /// Parse up to the end of statement and a return the contents from the 333 /// current token until the end of the statement; the current token on exit 334 /// will be either the EndOfStatement or EOF. 335 StringRef parseStringToEndOfStatement() override; 336 337 bool parseTextItem(std::string &Data); 338 339 unsigned getBinOpPrecedence(AsmToken::TokenKind K, 340 MCBinaryExpr::Opcode &Kind); 341 342 bool parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, SMLoc &EndLoc); 343 bool parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); 344 bool parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); 345 346 bool parseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); 347 348 bool parseCVFunctionId(int64_t &FunctionId, StringRef DirectiveName); 349 bool parseCVFileId(int64_t &FileId, StringRef DirectiveName); 350 351 // Generic (target and platform independent) directive parsing. 352 enum DirectiveKind { 353 DK_NO_DIRECTIVE, // Placeholder 354 DK_HANDLER_DIRECTIVE, 355 DK_ASSIGN, 356 DK_EQU, 357 DK_TEXTEQU, 358 DK_ASCII, 359 DK_ASCIZ, 360 DK_STRING, 361 DK_BYTE, 362 DK_SBYTE, 363 DK_WORD, 364 DK_SWORD, 365 DK_DWORD, 366 DK_SDWORD, 367 DK_FWORD, 368 DK_QWORD, 369 DK_SQWORD, 370 DK_DB, 371 DK_DD, 372 DK_DQ, 373 DK_DW, 374 DK_REAL4, 375 DK_REAL8, 376 DK_ALIGN, 377 DK_ORG, 378 DK_ENDR, 379 DK_EXTERN, 380 DK_PUBLIC, 381 DK_COMM, 382 DK_COMMENT, 383 DK_INCLUDE, 384 DK_REPT, 385 DK_IRP, 386 DK_IRPC, 387 DK_IF, 388 DK_IFE, 389 DK_IFB, 390 DK_IFNB, 391 DK_IFDEF, 392 DK_IFNDEF, 393 DK_IFDIF, 394 DK_IFDIFI, 395 DK_IFIDN, 396 DK_IFIDNI, 397 DK_ELSEIF, 398 DK_ELSEIFE, 399 DK_ELSEIFB, 400 DK_ELSEIFNB, 401 DK_ELSEIFDEF, 402 DK_ELSEIFNDEF, 403 DK_ELSEIFDIF, 404 DK_ELSEIFDIFI, 405 DK_ELSEIFIDN, 406 DK_ELSEIFIDNI, 407 DK_ELSE, 408 DK_ENDIF, 409 DK_FILE, 410 DK_LINE, 411 DK_LOC, 412 DK_STABS, 413 DK_CV_FILE, 414 DK_CV_FUNC_ID, 415 DK_CV_INLINE_SITE_ID, 416 DK_CV_LOC, 417 DK_CV_LINETABLE, 418 DK_CV_INLINE_LINETABLE, 419 DK_CV_DEF_RANGE, 420 DK_CV_STRINGTABLE, 421 DK_CV_STRING, 422 DK_CV_FILECHECKSUMS, 423 DK_CV_FILECHECKSUM_OFFSET, 424 DK_CV_FPO_DATA, 425 DK_CFI_SECTIONS, 426 DK_CFI_STARTPROC, 427 DK_CFI_ENDPROC, 428 DK_CFI_DEF_CFA, 429 DK_CFI_DEF_CFA_OFFSET, 430 DK_CFI_ADJUST_CFA_OFFSET, 431 DK_CFI_DEF_CFA_REGISTER, 432 DK_CFI_OFFSET, 433 DK_CFI_REL_OFFSET, 434 DK_CFI_PERSONALITY, 435 DK_CFI_LSDA, 436 DK_CFI_REMEMBER_STATE, 437 DK_CFI_RESTORE_STATE, 438 DK_CFI_SAME_VALUE, 439 DK_CFI_RESTORE, 440 DK_CFI_ESCAPE, 441 DK_CFI_RETURN_COLUMN, 442 DK_CFI_SIGNAL_FRAME, 443 DK_CFI_UNDEFINED, 444 DK_CFI_REGISTER, 445 DK_CFI_WINDOW_SAVE, 446 DK_CFI_B_KEY_FRAME, 447 DK_ALTMACRO, 448 DK_NOALTMACRO, 449 DK_MACRO, 450 DK_EXITM, 451 DK_ENDM, 452 DK_PURGEM, 453 DK_ERR, 454 DK_ERRB, 455 DK_ERRNB, 456 DK_ERRDEF, 457 DK_ERRNDEF, 458 DK_ERRDIF, 459 DK_ERRDIFI, 460 DK_ERRIDN, 461 DK_ERRIDNI, 462 DK_ERRE, 463 DK_ERRNZ, 464 DK_ECHO, 465 DK_END 466 }; 467 468 /// Maps directive name --> DirectiveKind enum, for 469 /// directives parsed by this class. 470 StringMap<DirectiveKind> DirectiveKindMap; 471 472 // Codeview def_range type parsing. 473 enum CVDefRangeType { 474 CVDR_DEFRANGE = 0, // Placeholder 475 CVDR_DEFRANGE_REGISTER, 476 CVDR_DEFRANGE_FRAMEPOINTER_REL, 477 CVDR_DEFRANGE_SUBFIELD_REGISTER, 478 CVDR_DEFRANGE_REGISTER_REL 479 }; 480 481 /// Maps Codeview def_range types --> CVDefRangeType enum, for 482 /// Codeview def_range types parsed by this class. 483 StringMap<CVDefRangeType> CVDefRangeTypeMap; 484 485 bool parseInitValue(unsigned Size); 486 487 // ".ascii", ".asciz", ".string" 488 bool parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); 489 490 // "byte", "word", ... 491 bool parseScalarInstList(unsigned Size, 492 SmallVectorImpl<const MCExpr *> &Values); 493 bool parseDirectiveValue(StringRef IDVal, unsigned Size); 494 bool parseDirectiveNamedValue(StringRef IDVal, unsigned Size, StringRef Name, 495 SMLoc NameLoc); 496 497 // "real4", "real8" 498 bool parseDirectiveRealValue(StringRef IDVal, const fltSemantics &Semantics); 499 bool parseRealInstList(const fltSemantics &Semantics, 500 SmallVectorImpl<APInt> &Values); 501 bool parseDirectiveNamedRealValue(StringRef IDVal, 502 const fltSemantics &Semantics, 503 StringRef Name, SMLoc NameLoc); 504 505 // "=", "equ", "textequ" 506 bool parseDirectiveEquate(StringRef IDVal, StringRef Name, 507 DirectiveKind DirKind); 508 509 bool parseDirectiveOrg(); // ".org" 510 bool parseDirectiveAlign(); // "align" 511 512 // ".file", ".line", ".loc", ".stabs" 513 bool parseDirectiveFile(SMLoc DirectiveLoc); 514 bool parseDirectiveLine(); 515 bool parseDirectiveLoc(); 516 bool parseDirectiveStabs(); 517 518 // ".cv_file", ".cv_func_id", ".cv_inline_site_id", ".cv_loc", ".cv_linetable", 519 // ".cv_inline_linetable", ".cv_def_range", ".cv_string" 520 bool parseDirectiveCVFile(); 521 bool parseDirectiveCVFuncId(); 522 bool parseDirectiveCVInlineSiteId(); 523 bool parseDirectiveCVLoc(); 524 bool parseDirectiveCVLinetable(); 525 bool parseDirectiveCVInlineLinetable(); 526 bool parseDirectiveCVDefRange(); 527 bool parseDirectiveCVString(); 528 bool parseDirectiveCVStringTable(); 529 bool parseDirectiveCVFileChecksums(); 530 bool parseDirectiveCVFileChecksumOffset(); 531 bool parseDirectiveCVFPOData(); 532 533 // .cfi directives 534 bool parseDirectiveCFIRegister(SMLoc DirectiveLoc); 535 bool parseDirectiveCFIWindowSave(); 536 bool parseDirectiveCFISections(); 537 bool parseDirectiveCFIStartProc(); 538 bool parseDirectiveCFIEndProc(); 539 bool parseDirectiveCFIDefCfaOffset(); 540 bool parseDirectiveCFIDefCfa(SMLoc DirectiveLoc); 541 bool parseDirectiveCFIAdjustCfaOffset(); 542 bool parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); 543 bool parseDirectiveCFIOffset(SMLoc DirectiveLoc); 544 bool parseDirectiveCFIRelOffset(SMLoc DirectiveLoc); 545 bool parseDirectiveCFIPersonalityOrLsda(bool IsPersonality); 546 bool parseDirectiveCFIRememberState(); 547 bool parseDirectiveCFIRestoreState(); 548 bool parseDirectiveCFISameValue(SMLoc DirectiveLoc); 549 bool parseDirectiveCFIRestore(SMLoc DirectiveLoc); 550 bool parseDirectiveCFIEscape(); 551 bool parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc); 552 bool parseDirectiveCFISignalFrame(); 553 bool parseDirectiveCFIUndefined(SMLoc DirectiveLoc); 554 555 // macro directives 556 bool parseDirectivePurgeMacro(SMLoc DirectiveLoc); 557 bool parseDirectiveExitMacro(StringRef Directive); 558 bool parseDirectiveEndMacro(StringRef Directive); 559 bool parseDirectiveMacro(SMLoc DirectiveLoc); 560 // alternate macro mode directives 561 bool parseDirectiveAltmacro(StringRef Directive); 562 563 /// Parse a directive like ".globl" which 564 /// accepts a single symbol (which should be a label or an external). 565 bool parseDirectiveSymbolAttribute(MCSymbolAttr Attr); 566 567 bool parseDirectiveComm(bool IsLocal); // ".comm" and ".lcomm" 568 569 bool parseDirectiveComment(SMLoc DirectiveLoc); // "comment" 570 571 bool parseDirectiveInclude(); // "include" 572 573 // "if" or "ife" 574 bool parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 575 // "ifb" or "ifnb", depending on ExpectBlank. 576 bool parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank); 577 // "ifidn", "ifdif", "ifidni", or "ifdifi", depending on ExpectEqual and 578 // CaseInsensitive. 579 bool parseDirectiveIfidn(SMLoc DirectiveLoc, bool ExpectEqual, 580 bool CaseInsensitive); 581 // "ifdef" or "ifndef", depending on expect_defined 582 bool parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined); 583 // "elseif" or "elseife" 584 bool parseDirectiveElseIf(SMLoc DirectiveLoc, DirectiveKind DirKind); 585 // "elseifb" or "elseifnb", depending on ExpectBlank. 586 bool parseDirectiveElseIfb(SMLoc DirectiveLoc, bool ExpectBlank); 587 // ".elseifdef" or ".elseifndef", depending on expect_defined 588 bool parseDirectiveElseIfdef(SMLoc DirectiveLoc, bool expect_defined); 589 // "elseifidn", "elseifdif", "elseifidni", or "elseifdifi", depending on 590 // ExpectEqual and CaseInsensitive. 591 bool parseDirectiveElseIfidn(SMLoc DirectiveLoc, bool ExpectEqual, 592 bool CaseInsensitive); 593 bool parseDirectiveElse(SMLoc DirectiveLoc); // "else" 594 bool parseDirectiveEndIf(SMLoc DirectiveLoc); // "endif" 595 bool parseEscapedString(std::string &Data) override; 596 bool parseAngleBracketString(std::string &Data) override; 597 598 // Macro-like directives 599 MCAsmMacro *parseMacroLikeBody(SMLoc DirectiveLoc); 600 void instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 601 raw_svector_ostream &OS); 602 bool parseDirectiveRept(SMLoc DirectiveLoc, StringRef Directive); 603 bool parseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" 604 bool parseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" 605 bool parseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" 606 607 // "_emit" or "__emit" 608 bool parseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, 609 size_t Len); 610 611 // "align" 612 bool parseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); 613 614 // "end" 615 bool parseDirectiveEnd(SMLoc DirectiveLoc); 616 617 // ".err" 618 bool parseDirectiveError(SMLoc DirectiveLoc); 619 // ".errb" or ".errnb", depending on ExpectBlank. 620 bool parseDirectiveErrorIfb(SMLoc DirectiveLoc, bool ExpectBlank); 621 // ".errdef" or ".errndef", depending on ExpectBlank. 622 bool parseDirectiveErrorIfdef(SMLoc DirectiveLoc, bool ExpectDefined); 623 // ".erridn", ".errdif", ".erridni", or ".errdifi", depending on ExpectEqual 624 // and CaseInsensitive. 625 bool parseDirectiveErrorIfidn(SMLoc DirectiveLoc, bool ExpectEqual, 626 bool CaseInsensitive); 627 // ".erre" or ".errnz", depending on ExpectZero. 628 bool parseDirectiveErrorIfe(SMLoc DirectiveLoc, bool ExpectZero); 629 630 // "echo" 631 bool parseDirectiveEcho(); 632 633 void initializeDirectiveKindMap(); 634 void initializeCVDefRangeTypeMap(); 635 }; 636 637 } // end anonymous namespace 638 639 namespace llvm { 640 641 extern MCAsmParserExtension *createCOFFMasmParser(); 642 643 } // end namespace llvm 644 645 enum { DEFAULT_ADDRSPACE = 0 }; 646 647 MasmParser::MasmParser(SourceMgr &SM, MCContext &Ctx, MCStreamer &Out, 648 const MCAsmInfo &MAI, unsigned CB = 0) 649 : Lexer(MAI), Ctx(Ctx), Out(Out), MAI(MAI), SrcMgr(SM), 650 CurBuffer(CB ? CB : SM.getMainFileID()) { 651 HadError = false; 652 // Save the old handler. 653 SavedDiagHandler = SrcMgr.getDiagHandler(); 654 SavedDiagContext = SrcMgr.getDiagContext(); 655 // Set our own handler which calls the saved handler. 656 SrcMgr.setDiagHandler(DiagHandler, this); 657 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 658 659 // Initialize the platform / file format parser. 660 switch (Ctx.getObjectFileInfo()->getObjectFileType()) { 661 case MCObjectFileInfo::IsCOFF: 662 PlatformParser.reset(createCOFFMasmParser()); 663 break; 664 default: 665 report_fatal_error("llvm-ml currently supports only COFF output."); 666 break; 667 } 668 669 initializeDirectiveKindMap(); 670 PlatformParser->Initialize(*this); 671 initializeCVDefRangeTypeMap(); 672 673 NumOfMacroInstantiations = 0; 674 } 675 676 MasmParser::~MasmParser() { 677 assert((HadError || ActiveMacros.empty()) && 678 "Unexpected active macro instantiation!"); 679 680 // Restore the saved diagnostics handler and context for use during 681 // finalization. 682 SrcMgr.setDiagHandler(SavedDiagHandler, SavedDiagContext); 683 } 684 685 void MasmParser::printMacroInstantiations() { 686 // Print the active macro instantiation stack. 687 for (std::vector<MacroInstantiation *>::const_reverse_iterator 688 it = ActiveMacros.rbegin(), 689 ie = ActiveMacros.rend(); 690 it != ie; ++it) 691 printMessage((*it)->InstantiationLoc, SourceMgr::DK_Note, 692 "while in macro instantiation"); 693 } 694 695 void MasmParser::Note(SMLoc L, const Twine &Msg, SMRange Range) { 696 printPendingErrors(); 697 printMessage(L, SourceMgr::DK_Note, Msg, Range); 698 printMacroInstantiations(); 699 } 700 701 bool MasmParser::Warning(SMLoc L, const Twine &Msg, SMRange Range) { 702 if (getTargetParser().getTargetOptions().MCNoWarn) 703 return false; 704 if (getTargetParser().getTargetOptions().MCFatalWarnings) 705 return Error(L, Msg, Range); 706 printMessage(L, SourceMgr::DK_Warning, Msg, Range); 707 printMacroInstantiations(); 708 return false; 709 } 710 711 bool MasmParser::printError(SMLoc L, const Twine &Msg, SMRange Range) { 712 HadError = true; 713 printMessage(L, SourceMgr::DK_Error, Msg, Range); 714 printMacroInstantiations(); 715 return true; 716 } 717 718 bool MasmParser::enterIncludeFile(const std::string &Filename) { 719 std::string IncludedFile; 720 unsigned NewBuf = 721 SrcMgr.AddIncludeFile(Filename, Lexer.getLoc(), IncludedFile); 722 if (!NewBuf) 723 return true; 724 725 CurBuffer = NewBuf; 726 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 727 return false; 728 } 729 730 void MasmParser::jumpToLoc(SMLoc Loc, unsigned InBuffer) { 731 CurBuffer = InBuffer ? InBuffer : SrcMgr.FindBufferContainingLoc(Loc); 732 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), 733 Loc.getPointer()); 734 } 735 736 const AsmToken &MasmParser::Lex() { 737 if (Lexer.getTok().is(AsmToken::Error)) 738 Error(Lexer.getErrLoc(), Lexer.getErr()); 739 740 // if it's a end of statement with a comment in it 741 if (getTok().is(AsmToken::EndOfStatement)) { 742 // if this is a line comment output it. 743 if (!getTok().getString().empty() && getTok().getString().front() != '\n' && 744 getTok().getString().front() != '\r' && MAI.preserveAsmComments()) 745 Out.addExplicitComment(Twine(getTok().getString())); 746 } 747 748 const AsmToken *tok = &Lexer.Lex(); 749 750 while (tok->is(AsmToken::Identifier)) { 751 auto it = Variables.find(tok->getIdentifier()); 752 if (it != Variables.end() && it->second.IsText) { 753 std::unique_ptr<MemoryBuffer> Instantiation = 754 MemoryBuffer::getMemBufferCopy(it->second.TextValue, 755 "<instantiation>"); 756 757 // Jump to the macro instantiation and prime the lexer. 758 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), 759 getTok().getEndLoc()); 760 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer(), nullptr, 761 /*EndStatementAtEOF=*/false); 762 tok = &Lexer.Lex(); 763 } else { 764 break; 765 } 766 } 767 768 // Parse comments here to be deferred until end of next statement. 769 while (tok->is(AsmToken::Comment)) { 770 if (MAI.preserveAsmComments()) 771 Out.addExplicitComment(Twine(tok->getString())); 772 tok = &Lexer.Lex(); 773 } 774 775 if (tok->is(AsmToken::Eof)) { 776 // If this is the end of an included file, pop the parent file off the 777 // include stack. 778 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer); 779 if (ParentIncludeLoc != SMLoc()) { 780 jumpToLoc(ParentIncludeLoc); 781 return Lex(); 782 } 783 } 784 785 return *tok; 786 } 787 788 bool MasmParser::enabledGenDwarfForAssembly() { 789 // Check whether the user specified -g. 790 if (!getContext().getGenDwarfForAssembly()) 791 return false; 792 // If we haven't encountered any .file directives (which would imply that 793 // the assembler source was produced with debug info already) then emit one 794 // describing the assembler source file itself. 795 if (getContext().getGenDwarfFileNumber() == 0) { 796 // Use the first #line directive for this, if any. It's preprocessed, so 797 // there is no checksum, and of course no source directive. 798 if (!FirstCppHashFilename.empty()) 799 getContext().setMCLineTableRootFile(/*CUID=*/0, 800 getContext().getCompilationDir(), 801 FirstCppHashFilename, 802 /*Cksum=*/None, /*Source=*/None); 803 const MCDwarfFile &RootFile = 804 getContext().getMCDwarfLineTable(/*CUID=*/0).getRootFile(); 805 getContext().setGenDwarfFileNumber(getStreamer().emitDwarfFileDirective( 806 /*CUID=*/0, getContext().getCompilationDir(), RootFile.Name, 807 RootFile.Checksum, RootFile.Source)); 808 } 809 return true; 810 } 811 812 bool MasmParser::Run(bool NoInitialTextSection, bool NoFinalize) { 813 // Create the initial section, if requested. 814 if (!NoInitialTextSection) 815 Out.InitSections(false); 816 817 // Prime the lexer. 818 Lex(); 819 820 HadError = false; 821 AsmCond StartingCondState = TheCondState; 822 SmallVector<AsmRewrite, 4> AsmStrRewrites; 823 824 // If we are generating dwarf for assembly source files save the initial text 825 // section. (Don't use enabledGenDwarfForAssembly() here, as we aren't 826 // emitting any actual debug info yet and haven't had a chance to parse any 827 // embedded .file directives.) 828 if (getContext().getGenDwarfForAssembly()) { 829 MCSection *Sec = getStreamer().getCurrentSectionOnly(); 830 if (!Sec->getBeginSymbol()) { 831 MCSymbol *SectionStartSym = getContext().createTempSymbol(); 832 getStreamer().emitLabel(SectionStartSym); 833 Sec->setBeginSymbol(SectionStartSym); 834 } 835 bool InsertResult = getContext().addGenDwarfSection(Sec); 836 assert(InsertResult && ".text section should not have debug info yet"); 837 (void)InsertResult; 838 } 839 840 // While we have input, parse each statement. 841 while (Lexer.isNot(AsmToken::Eof)) { 842 ParseStatementInfo Info(&AsmStrRewrites); 843 bool Parsed = parseStatement(Info, nullptr); 844 845 // If we have a Lexer Error we are on an Error Token. Load in Lexer Error 846 // for printing ErrMsg via Lex() only if no (presumably better) parser error 847 // exists. 848 if (Parsed && !hasPendingError() && Lexer.getTok().is(AsmToken::Error)) { 849 Lex(); 850 } 851 852 // parseStatement returned true so may need to emit an error. 853 printPendingErrors(); 854 855 // Skipping to the next line if needed. 856 if (Parsed && !getLexer().isAtStartOfStatement()) 857 eatToEndOfStatement(); 858 } 859 860 getTargetParser().onEndOfFile(); 861 printPendingErrors(); 862 863 // All errors should have been emitted. 864 assert(!hasPendingError() && "unexpected error from parseStatement"); 865 866 getTargetParser().flushPendingInstructions(getStreamer()); 867 868 if (TheCondState.TheCond != StartingCondState.TheCond || 869 TheCondState.Ignore != StartingCondState.Ignore) 870 printError(getTok().getLoc(), "unmatched .ifs or .elses"); 871 // Check to see there are no empty DwarfFile slots. 872 const auto &LineTables = getContext().getMCDwarfLineTables(); 873 if (!LineTables.empty()) { 874 unsigned Index = 0; 875 for (const auto &File : LineTables.begin()->second.getMCDwarfFiles()) { 876 if (File.Name.empty() && Index != 0) 877 printError(getTok().getLoc(), "unassigned file number: " + 878 Twine(Index) + 879 " for .file directives"); 880 ++Index; 881 } 882 } 883 884 // Check to see that all assembler local symbols were actually defined. 885 // Targets that don't do subsections via symbols may not want this, though, 886 // so conservatively exclude them. Only do this if we're finalizing, though, 887 // as otherwise we won't necessarilly have seen everything yet. 888 if (!NoFinalize) { 889 if (MAI.hasSubsectionsViaSymbols()) { 890 for (const auto &TableEntry : getContext().getSymbols()) { 891 MCSymbol *Sym = TableEntry.getValue(); 892 // Variable symbols may not be marked as defined, so check those 893 // explicitly. If we know it's a variable, we have a definition for 894 // the purposes of this check. 895 if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined()) 896 // FIXME: We would really like to refer back to where the symbol was 897 // first referenced for a source location. We need to add something 898 // to track that. Currently, we just point to the end of the file. 899 printError(getTok().getLoc(), "assembler local symbol '" + 900 Sym->getName() + "' not defined"); 901 } 902 } 903 904 // Temporary symbols like the ones for directional jumps don't go in the 905 // symbol table. They also need to be diagnosed in all (final) cases. 906 for (std::tuple<SMLoc, CppHashInfoTy, MCSymbol *> &LocSym : DirLabels) { 907 if (std::get<2>(LocSym)->isUndefined()) { 908 // Reset the state of any "# line file" directives we've seen to the 909 // context as it was at the diagnostic site. 910 CppHashInfo = std::get<1>(LocSym); 911 printError(std::get<0>(LocSym), "directional label undefined"); 912 } 913 } 914 } 915 916 // Finalize the output stream if there are no errors and if the client wants 917 // us to. 918 if (!HadError && !NoFinalize) 919 Out.Finish(); 920 921 return HadError || getContext().hadError(); 922 } 923 924 bool MasmParser::checkForValidSection() { 925 if (!ParsingMSInlineAsm && !getStreamer().getCurrentSectionOnly()) { 926 Out.InitSections(false); 927 return Error(getTok().getLoc(), 928 "expected section directive before assembly directive"); 929 } 930 return false; 931 } 932 933 /// Throw away the rest of the line for testing purposes. 934 void MasmParser::eatToEndOfStatement() { 935 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 936 Lexer.Lex(); 937 938 // Eat EOL. 939 if (Lexer.is(AsmToken::EndOfStatement)) 940 Lexer.Lex(); 941 } 942 943 StringRef MasmParser::parseStringToEndOfStatement() { 944 const char *Start = getTok().getLoc().getPointer(); 945 946 while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) 947 Lexer.Lex(); 948 949 const char *End = getTok().getLoc().getPointer(); 950 return StringRef(Start, End - Start); 951 } 952 953 /// Parse a paren expression and return it. 954 /// NOTE: This assumes the leading '(' has already been consumed. 955 /// 956 /// parenexpr ::= expr) 957 /// 958 bool MasmParser::parseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { 959 if (parseExpression(Res)) 960 return true; 961 if (Lexer.isNot(AsmToken::RParen)) 962 return TokError("expected ')' in parentheses expression"); 963 EndLoc = Lexer.getTok().getEndLoc(); 964 Lex(); 965 return false; 966 } 967 968 /// Parse a bracket expression and return it. 969 /// NOTE: This assumes the leading '[' has already been consumed. 970 /// 971 /// bracketexpr ::= expr] 972 /// 973 bool MasmParser::parseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { 974 if (parseExpression(Res)) 975 return true; 976 EndLoc = getTok().getEndLoc(); 977 if (parseToken(AsmToken::RBrac, "expected ']' in brackets expression")) 978 return true; 979 return false; 980 } 981 982 /// Parse a primary expression and return it. 983 /// primaryexpr ::= (parenexpr 984 /// primaryexpr ::= symbol 985 /// primaryexpr ::= number 986 /// primaryexpr ::= '.' 987 /// primaryexpr ::= ~,+,- primaryexpr 988 bool MasmParser::parsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { 989 SMLoc FirstTokenLoc = getLexer().getLoc(); 990 AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); 991 switch (FirstTokenKind) { 992 default: 993 return TokError("unknown token in expression"); 994 // If we have an error assume that we've already handled it. 995 case AsmToken::Error: 996 return true; 997 case AsmToken::Exclaim: 998 Lex(); // Eat the operator. 999 if (parsePrimaryExpr(Res, EndLoc)) 1000 return true; 1001 Res = MCUnaryExpr::createLNot(Res, getContext(), FirstTokenLoc); 1002 return false; 1003 case AsmToken::Dollar: 1004 case AsmToken::At: 1005 case AsmToken::String: 1006 case AsmToken::Identifier: { 1007 StringRef Identifier; 1008 if (parseIdentifier(Identifier)) { 1009 // We may have failed but $ may be a valid token. 1010 if (getTok().is(AsmToken::Dollar)) { 1011 if (Lexer.getMAI().getDollarIsPC()) { 1012 Lex(); 1013 // This is a '$' reference, which references the current PC. Emit a 1014 // temporary label to the streamer and refer to it. 1015 MCSymbol *Sym = Ctx.createTempSymbol(); 1016 Out.emitLabel(Sym); 1017 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, 1018 getContext()); 1019 EndLoc = FirstTokenLoc; 1020 return false; 1021 } 1022 return Error(FirstTokenLoc, "invalid token in expression"); 1023 } 1024 } 1025 // Parse symbol variant 1026 std::pair<StringRef, StringRef> Split; 1027 if (!MAI.useParensForSymbolVariant()) { 1028 if (FirstTokenKind == AsmToken::String) { 1029 if (Lexer.is(AsmToken::At)) { 1030 Lex(); // eat @ 1031 SMLoc AtLoc = getLexer().getLoc(); 1032 StringRef VName; 1033 if (parseIdentifier(VName)) 1034 return Error(AtLoc, "expected symbol variant after '@'"); 1035 1036 Split = std::make_pair(Identifier, VName); 1037 } 1038 } else { 1039 Split = Identifier.split('@'); 1040 } 1041 } else if (Lexer.is(AsmToken::LParen)) { 1042 Lex(); // eat '('. 1043 StringRef VName; 1044 parseIdentifier(VName); 1045 // eat ')'. 1046 if (parseToken(AsmToken::RParen, 1047 "unexpected token in variant, expected ')'")) 1048 return true; 1049 Split = std::make_pair(Identifier, VName); 1050 } 1051 1052 EndLoc = SMLoc::getFromPointer(Identifier.end()); 1053 1054 // This is a symbol reference. 1055 StringRef SymbolName = Identifier; 1056 if (SymbolName.empty()) 1057 return Error(getLexer().getLoc(), "expected a symbol reference"); 1058 1059 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1060 1061 // Lookup the symbol variant if used. 1062 if (!Split.second.empty()) { 1063 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1064 if (Variant != MCSymbolRefExpr::VK_Invalid) { 1065 SymbolName = Split.first; 1066 } else if (MAI.doesAllowAtInName() && !MAI.useParensForSymbolVariant()) { 1067 Variant = MCSymbolRefExpr::VK_None; 1068 } else { 1069 return Error(SMLoc::getFromPointer(Split.second.begin()), 1070 "invalid variant '" + Split.second + "'"); 1071 } 1072 } 1073 1074 MCSymbol *Sym = getContext().getInlineAsmLabel(SymbolName); 1075 if (!Sym) 1076 Sym = getContext().getOrCreateSymbol(SymbolName); 1077 1078 // If this is an absolute variable reference, substitute it now to preserve 1079 // semantics in the face of reassignment. 1080 if (Sym->isVariable()) { 1081 auto V = Sym->getVariableValue(/*SetUsed*/ false); 1082 bool DoInline = isa<MCConstantExpr>(V) && !Variant; 1083 if (auto TV = dyn_cast<MCTargetExpr>(V)) 1084 DoInline = TV->inlineAssignedExpr(); 1085 if (DoInline) { 1086 if (Variant) 1087 return Error(EndLoc, "unexpected modifier on variable reference"); 1088 Res = Sym->getVariableValue(/*SetUsed*/ false); 1089 return false; 1090 } 1091 } 1092 1093 // Otherwise create a symbol ref. 1094 Res = MCSymbolRefExpr::create(Sym, Variant, getContext(), FirstTokenLoc); 1095 return false; 1096 } 1097 case AsmToken::BigNum: 1098 return TokError("literal value out of range for directive"); 1099 case AsmToken::Integer: { 1100 SMLoc Loc = getTok().getLoc(); 1101 int64_t IntVal = getTok().getIntVal(); 1102 Res = MCConstantExpr::create(IntVal, getContext()); 1103 EndLoc = Lexer.getTok().getEndLoc(); 1104 Lex(); // Eat token. 1105 // Look for 'b' or 'f' following an Integer as a directional label 1106 if (Lexer.getKind() == AsmToken::Identifier) { 1107 StringRef IDVal = getTok().getString(); 1108 // Lookup the symbol variant if used. 1109 std::pair<StringRef, StringRef> Split = IDVal.split('@'); 1110 MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; 1111 if (Split.first.size() != IDVal.size()) { 1112 Variant = MCSymbolRefExpr::getVariantKindForName(Split.second); 1113 if (Variant == MCSymbolRefExpr::VK_Invalid) 1114 return TokError("invalid variant '" + Split.second + "'"); 1115 IDVal = Split.first; 1116 } 1117 if (IDVal == "f" || IDVal == "b") { 1118 MCSymbol *Sym = 1119 Ctx.getDirectionalLocalSymbol(IntVal, IDVal == "b"); 1120 Res = MCSymbolRefExpr::create(Sym, Variant, getContext()); 1121 if (IDVal == "b" && Sym->isUndefined()) 1122 return Error(Loc, "directional label undefined"); 1123 DirLabels.push_back(std::make_tuple(Loc, CppHashInfo, Sym)); 1124 EndLoc = Lexer.getTok().getEndLoc(); 1125 Lex(); // Eat identifier. 1126 } 1127 } 1128 return false; 1129 } 1130 case AsmToken::Real: { 1131 APFloat RealVal(APFloat::IEEEdouble(), getTok().getString()); 1132 uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); 1133 Res = MCConstantExpr::create(IntVal, getContext()); 1134 EndLoc = Lexer.getTok().getEndLoc(); 1135 Lex(); // Eat token. 1136 return false; 1137 } 1138 case AsmToken::Dot: { 1139 // This is a '.' reference, which references the current PC. Emit a 1140 // temporary label to the streamer and refer to it. 1141 MCSymbol *Sym = Ctx.createTempSymbol(); 1142 Out.emitLabel(Sym); 1143 Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext()); 1144 EndLoc = Lexer.getTok().getEndLoc(); 1145 Lex(); // Eat identifier. 1146 return false; 1147 } 1148 case AsmToken::LParen: 1149 Lex(); // Eat the '('. 1150 return parseParenExpr(Res, EndLoc); 1151 case AsmToken::LBrac: 1152 if (!PlatformParser->HasBracketExpressions()) 1153 return TokError("brackets expression not supported on this target"); 1154 Lex(); // Eat the '['. 1155 return parseBracketExpr(Res, EndLoc); 1156 case AsmToken::Minus: 1157 Lex(); // Eat the operator. 1158 if (parsePrimaryExpr(Res, EndLoc)) 1159 return true; 1160 Res = MCUnaryExpr::createMinus(Res, getContext(), FirstTokenLoc); 1161 return false; 1162 case AsmToken::Plus: 1163 Lex(); // Eat the operator. 1164 if (parsePrimaryExpr(Res, EndLoc)) 1165 return true; 1166 Res = MCUnaryExpr::createPlus(Res, getContext(), FirstTokenLoc); 1167 return false; 1168 case AsmToken::Tilde: 1169 Lex(); // Eat the operator. 1170 if (parsePrimaryExpr(Res, EndLoc)) 1171 return true; 1172 Res = MCUnaryExpr::createNot(Res, getContext(), FirstTokenLoc); 1173 return false; 1174 // MIPS unary expression operators. The lexer won't generate these tokens if 1175 // MCAsmInfo::HasMipsExpressions is false for the target. 1176 case AsmToken::PercentCall16: 1177 case AsmToken::PercentCall_Hi: 1178 case AsmToken::PercentCall_Lo: 1179 case AsmToken::PercentDtprel_Hi: 1180 case AsmToken::PercentDtprel_Lo: 1181 case AsmToken::PercentGot: 1182 case AsmToken::PercentGot_Disp: 1183 case AsmToken::PercentGot_Hi: 1184 case AsmToken::PercentGot_Lo: 1185 case AsmToken::PercentGot_Ofst: 1186 case AsmToken::PercentGot_Page: 1187 case AsmToken::PercentGottprel: 1188 case AsmToken::PercentGp_Rel: 1189 case AsmToken::PercentHi: 1190 case AsmToken::PercentHigher: 1191 case AsmToken::PercentHighest: 1192 case AsmToken::PercentLo: 1193 case AsmToken::PercentNeg: 1194 case AsmToken::PercentPcrel_Hi: 1195 case AsmToken::PercentPcrel_Lo: 1196 case AsmToken::PercentTlsgd: 1197 case AsmToken::PercentTlsldm: 1198 case AsmToken::PercentTprel_Hi: 1199 case AsmToken::PercentTprel_Lo: 1200 Lex(); // Eat the operator. 1201 if (Lexer.isNot(AsmToken::LParen)) 1202 return TokError("expected '(' after operator"); 1203 Lex(); // Eat the operator. 1204 if (parseExpression(Res, EndLoc)) 1205 return true; 1206 if (Lexer.isNot(AsmToken::RParen)) 1207 return TokError("expected ')'"); 1208 Lex(); // Eat the operator. 1209 Res = getTargetParser().createTargetUnaryExpr(Res, FirstTokenKind, Ctx); 1210 return !Res; 1211 } 1212 } 1213 1214 bool MasmParser::parseExpression(const MCExpr *&Res) { 1215 SMLoc EndLoc; 1216 return parseExpression(Res, EndLoc); 1217 } 1218 1219 /// This function checks if the next token is <string> type or arithmetic. 1220 /// string that begin with character '<' must end with character '>'. 1221 /// otherwise it is arithmetics. 1222 /// If the function returns a 'true' value, 1223 /// the End argument will be filled with the last location pointed to the '>' 1224 /// character. 1225 1226 /// There is a gap between the AltMacro's documentation and the single quote 1227 /// implementation. GCC does not fully support this feature and so we will not 1228 /// support it. 1229 /// TODO: Adding single quote as a string. 1230 static bool isAngleBracketString(SMLoc &StrLoc, SMLoc &EndLoc) { 1231 assert((StrLoc.getPointer() != nullptr) && 1232 "Argument to the function cannot be a NULL value"); 1233 const char *CharPtr = StrLoc.getPointer(); 1234 while ((*CharPtr != '>') && (*CharPtr != '\n') && (*CharPtr != '\r') && 1235 (*CharPtr != '\0')) { 1236 if (*CharPtr == '!') 1237 CharPtr++; 1238 CharPtr++; 1239 } 1240 if (*CharPtr == '>') { 1241 EndLoc = StrLoc.getFromPointer(CharPtr + 1); 1242 return true; 1243 } 1244 return false; 1245 } 1246 1247 /// creating a string without the escape characters '!'. 1248 static std::string angleBracketString(StringRef AltMacroStr) { 1249 std::string Res; 1250 for (size_t Pos = 0; Pos < AltMacroStr.size(); Pos++) { 1251 if (AltMacroStr[Pos] == '!') 1252 Pos++; 1253 Res += AltMacroStr[Pos]; 1254 } 1255 return Res; 1256 } 1257 1258 /// Parse an expression and return it. 1259 /// 1260 /// expr ::= expr &&,|| expr -> lowest. 1261 /// expr ::= expr |,^,&,! expr 1262 /// expr ::= expr ==,!=,<>,<,<=,>,>= expr 1263 /// expr ::= expr <<,>> expr 1264 /// expr ::= expr +,- expr 1265 /// expr ::= expr *,/,% expr -> highest. 1266 /// expr ::= primaryexpr 1267 /// 1268 bool MasmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1269 // Parse the expression. 1270 Res = nullptr; 1271 if (getTargetParser().parsePrimaryExpr(Res, EndLoc) || 1272 parseBinOpRHS(1, Res, EndLoc)) 1273 return true; 1274 1275 // Try to constant fold it up front, if possible. Do not exploit 1276 // assembler here. 1277 int64_t Value; 1278 if (Res->evaluateAsAbsolute(Value)) 1279 Res = MCConstantExpr::create(Value, getContext()); 1280 1281 return false; 1282 } 1283 1284 bool MasmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { 1285 Res = nullptr; 1286 return parseParenExpr(Res, EndLoc) || parseBinOpRHS(1, Res, EndLoc); 1287 } 1288 1289 bool MasmParser::parseParenExprOfDepth(unsigned ParenDepth, const MCExpr *&Res, 1290 SMLoc &EndLoc) { 1291 if (parseParenExpr(Res, EndLoc)) 1292 return true; 1293 1294 for (; ParenDepth > 0; --ParenDepth) { 1295 if (parseBinOpRHS(1, Res, EndLoc)) 1296 return true; 1297 1298 // We don't Lex() the last RParen. 1299 // This is the same behavior as parseParenExpression(). 1300 if (ParenDepth - 1 > 0) { 1301 EndLoc = getTok().getEndLoc(); 1302 if (parseToken(AsmToken::RParen, 1303 "expected ')' in parentheses expression")) 1304 return true; 1305 } 1306 } 1307 return false; 1308 } 1309 1310 bool MasmParser::parseAbsoluteExpression(int64_t &Res) { 1311 const MCExpr *Expr; 1312 1313 SMLoc StartLoc = Lexer.getLoc(); 1314 if (parseExpression(Expr)) 1315 return true; 1316 1317 if (!Expr->evaluateAsAbsolute(Res, getStreamer().getAssemblerPtr())) 1318 return Error(StartLoc, "expected absolute expression"); 1319 1320 return false; 1321 } 1322 1323 static unsigned getGNUBinOpPrecedence(AsmToken::TokenKind K, 1324 MCBinaryExpr::Opcode &Kind, 1325 bool ShouldUseLogicalShr) { 1326 switch (K) { 1327 default: 1328 return 0; // not a binop. 1329 1330 // Lowest Precedence: &&, || 1331 case AsmToken::AmpAmp: 1332 Kind = MCBinaryExpr::LAnd; 1333 return 2; 1334 case AsmToken::PipePipe: 1335 Kind = MCBinaryExpr::LOr; 1336 return 1; 1337 1338 // Low Precedence: ==, !=, <>, <, <=, >, >= 1339 case AsmToken::EqualEqual: 1340 Kind = MCBinaryExpr::EQ; 1341 return 3; 1342 case AsmToken::ExclaimEqual: 1343 case AsmToken::LessGreater: 1344 Kind = MCBinaryExpr::NE; 1345 return 3; 1346 case AsmToken::Less: 1347 Kind = MCBinaryExpr::LT; 1348 return 3; 1349 case AsmToken::LessEqual: 1350 Kind = MCBinaryExpr::LTE; 1351 return 3; 1352 case AsmToken::Greater: 1353 Kind = MCBinaryExpr::GT; 1354 return 3; 1355 case AsmToken::GreaterEqual: 1356 Kind = MCBinaryExpr::GTE; 1357 return 3; 1358 1359 // Low Intermediate Precedence: +, - 1360 case AsmToken::Plus: 1361 Kind = MCBinaryExpr::Add; 1362 return 4; 1363 case AsmToken::Minus: 1364 Kind = MCBinaryExpr::Sub; 1365 return 4; 1366 1367 // High Intermediate Precedence: |, &, ^ 1368 // 1369 // FIXME: gas seems to support '!' as an infix operator? 1370 case AsmToken::Pipe: 1371 Kind = MCBinaryExpr::Or; 1372 return 5; 1373 case AsmToken::Caret: 1374 Kind = MCBinaryExpr::Xor; 1375 return 5; 1376 case AsmToken::Amp: 1377 Kind = MCBinaryExpr::And; 1378 return 5; 1379 1380 // Highest Precedence: *, /, %, <<, >> 1381 case AsmToken::Star: 1382 Kind = MCBinaryExpr::Mul; 1383 return 6; 1384 case AsmToken::Slash: 1385 Kind = MCBinaryExpr::Div; 1386 return 6; 1387 case AsmToken::Percent: 1388 Kind = MCBinaryExpr::Mod; 1389 return 6; 1390 case AsmToken::LessLess: 1391 Kind = MCBinaryExpr::Shl; 1392 return 6; 1393 case AsmToken::GreaterGreater: 1394 Kind = ShouldUseLogicalShr ? MCBinaryExpr::LShr : MCBinaryExpr::AShr; 1395 return 6; 1396 } 1397 } 1398 1399 unsigned MasmParser::getBinOpPrecedence(AsmToken::TokenKind K, 1400 MCBinaryExpr::Opcode &Kind) { 1401 bool ShouldUseLogicalShr = MAI.shouldUseLogicalShr(); 1402 return getGNUBinOpPrecedence(K, Kind, ShouldUseLogicalShr); 1403 } 1404 1405 /// Parse all binary operators with precedence >= 'Precedence'. 1406 /// Res contains the LHS of the expression on input. 1407 bool MasmParser::parseBinOpRHS(unsigned Precedence, const MCExpr *&Res, 1408 SMLoc &EndLoc) { 1409 SMLoc StartLoc = Lexer.getLoc(); 1410 while (true) { 1411 MCBinaryExpr::Opcode Kind = MCBinaryExpr::Add; 1412 unsigned TokPrec = getBinOpPrecedence(Lexer.getKind(), Kind); 1413 1414 // If the next token is lower precedence than we are allowed to eat, return 1415 // successfully with what we ate already. 1416 if (TokPrec < Precedence) 1417 return false; 1418 1419 Lex(); 1420 1421 // Eat the next primary expression. 1422 const MCExpr *RHS; 1423 if (getTargetParser().parsePrimaryExpr(RHS, EndLoc)) 1424 return true; 1425 1426 // If BinOp binds less tightly with RHS than the operator after RHS, let 1427 // the pending operator take RHS as its LHS. 1428 MCBinaryExpr::Opcode Dummy; 1429 unsigned NextTokPrec = getBinOpPrecedence(Lexer.getKind(), Dummy); 1430 if (TokPrec < NextTokPrec && parseBinOpRHS(TokPrec + 1, RHS, EndLoc)) 1431 return true; 1432 1433 // Merge LHS and RHS according to operator. 1434 Res = MCBinaryExpr::create(Kind, Res, RHS, getContext(), StartLoc); 1435 } 1436 } 1437 1438 /// ParseStatement: 1439 /// ::= EndOfStatement 1440 /// ::= Label* Directive ...Operands... EndOfStatement 1441 /// ::= Label* Identifier OperandList* EndOfStatement 1442 bool MasmParser::parseStatement(ParseStatementInfo &Info, 1443 MCAsmParserSemaCallback *SI) { 1444 assert(!hasPendingError() && "parseStatement started with pending error"); 1445 // Eat initial spaces and comments 1446 while (Lexer.is(AsmToken::Space)) 1447 Lex(); 1448 if (Lexer.is(AsmToken::EndOfStatement)) { 1449 // if this is a line comment we can drop it safely 1450 if (getTok().getString().empty() || getTok().getString().front() == '\r' || 1451 getTok().getString().front() == '\n') 1452 Out.AddBlankLine(); 1453 Lex(); 1454 return false; 1455 } 1456 // Statements always start with an identifier, unless we're dealing with a 1457 // processor directive (.386, .686, etc.) that lexes as a real. 1458 AsmToken ID = getTok(); 1459 SMLoc IDLoc = ID.getLoc(); 1460 StringRef IDVal; 1461 int64_t LocalLabelVal = -1; 1462 if (Lexer.is(AsmToken::HashDirective)) 1463 return parseCppHashLineFilenameComment(IDLoc); 1464 // Allow an integer followed by a ':' as a directional local label. 1465 if (Lexer.is(AsmToken::Integer)) { 1466 LocalLabelVal = getTok().getIntVal(); 1467 if (LocalLabelVal < 0) { 1468 if (!TheCondState.Ignore) { 1469 Lex(); // always eat a token 1470 return Error(IDLoc, "unexpected token at start of statement"); 1471 } 1472 IDVal = ""; 1473 } else { 1474 IDVal = getTok().getString(); 1475 Lex(); // Consume the integer token to be used as an identifier token. 1476 if (Lexer.getKind() != AsmToken::Colon) { 1477 if (!TheCondState.Ignore) { 1478 Lex(); // always eat a token 1479 return Error(IDLoc, "unexpected token at start of statement"); 1480 } 1481 } 1482 } 1483 } else if (Lexer.is(AsmToken::Dot)) { 1484 // Treat '.' as a valid identifier in this context. 1485 Lex(); 1486 IDVal = "."; 1487 } else if (Lexer.is(AsmToken::LCurly)) { 1488 // Treat '{' as a valid identifier in this context. 1489 Lex(); 1490 IDVal = "{"; 1491 1492 } else if (Lexer.is(AsmToken::RCurly)) { 1493 // Treat '}' as a valid identifier in this context. 1494 Lex(); 1495 IDVal = "}"; 1496 } else if (Lexer.is(AsmToken::Star) && 1497 getTargetParser().starIsStartOfStatement()) { 1498 // Accept '*' as a valid start of statement. 1499 Lex(); 1500 IDVal = "*"; 1501 } else if (Lexer.is(AsmToken::Real)) { 1502 // Treat ".<number>" as a valid identifier in this context. 1503 IDVal = getTok().getString(); 1504 Lex(); // always eat a token 1505 if (!IDVal.startswith(".")) 1506 return Error(IDLoc, "unexpected token at start of statement"); 1507 } else if (parseIdentifier(IDVal)) { 1508 if (!TheCondState.Ignore) { 1509 Lex(); // always eat a token 1510 return Error(IDLoc, "unexpected token at start of statement"); 1511 } 1512 IDVal = ""; 1513 } 1514 1515 // Handle conditional assembly here before checking for skipping. We 1516 // have to do this so that .endif isn't skipped in a ".if 0" block for 1517 // example. 1518 StringMap<DirectiveKind>::const_iterator DirKindIt = 1519 DirectiveKindMap.find(IDVal.lower()); 1520 DirectiveKind DirKind = (DirKindIt == DirectiveKindMap.end()) 1521 ? DK_NO_DIRECTIVE 1522 : DirKindIt->getValue(); 1523 switch (DirKind) { 1524 default: 1525 break; 1526 case DK_IF: 1527 case DK_IFE: 1528 return parseDirectiveIf(IDLoc, DirKind); 1529 case DK_IFB: 1530 return parseDirectiveIfb(IDLoc, true); 1531 case DK_IFNB: 1532 return parseDirectiveIfb(IDLoc, false); 1533 case DK_IFDEF: 1534 return parseDirectiveIfdef(IDLoc, true); 1535 case DK_IFNDEF: 1536 return parseDirectiveIfdef(IDLoc, false); 1537 case DK_IFDIF: 1538 return parseDirectiveIfidn(IDLoc, /*ExpectEqual=*/false, 1539 /*CaseInsensitive=*/false); 1540 case DK_IFDIFI: 1541 return parseDirectiveIfidn(IDLoc, /*ExpectEqual=*/false, 1542 /*CaseInsensitive=*/true); 1543 case DK_IFIDN: 1544 return parseDirectiveIfidn(IDLoc, /*ExpectEqual=*/true, 1545 /*CaseInsensitive=*/false); 1546 case DK_IFIDNI: 1547 return parseDirectiveIfidn(IDLoc, /*ExpectEqual=*/true, 1548 /*CaseInsensitive=*/true); 1549 case DK_ELSEIF: 1550 case DK_ELSEIFE: 1551 return parseDirectiveElseIf(IDLoc, DirKind); 1552 case DK_ELSEIFB: 1553 return parseDirectiveElseIfb(IDLoc, true); 1554 case DK_ELSEIFNB: 1555 return parseDirectiveElseIfb(IDLoc, false); 1556 case DK_ELSEIFDEF: 1557 return parseDirectiveElseIfdef(IDLoc, true); 1558 case DK_ELSEIFNDEF: 1559 return parseDirectiveElseIfdef(IDLoc, false); 1560 case DK_ELSEIFDIF: 1561 return parseDirectiveElseIfidn(IDLoc, /*ExpectEqual=*/false, 1562 /*CaseInsensitive=*/false); 1563 case DK_ELSEIFDIFI: 1564 return parseDirectiveElseIfidn(IDLoc, /*ExpectEqual=*/false, 1565 /*CaseInsensitive=*/true); 1566 case DK_ELSEIFIDN: 1567 return parseDirectiveElseIfidn(IDLoc, /*ExpectEqual=*/true, 1568 /*CaseInsensitive=*/false); 1569 case DK_ELSEIFIDNI: 1570 return parseDirectiveElseIfidn(IDLoc, /*ExpectEqual=*/true, 1571 /*CaseInsensitive=*/true); 1572 case DK_ELSE: 1573 return parseDirectiveElse(IDLoc); 1574 case DK_ENDIF: 1575 return parseDirectiveEndIf(IDLoc); 1576 } 1577 1578 // Ignore the statement if in the middle of inactive conditional 1579 // (e.g. ".if 0"). 1580 if (TheCondState.Ignore) { 1581 eatToEndOfStatement(); 1582 return false; 1583 } 1584 1585 // FIXME: Recurse on local labels? 1586 1587 // See what kind of statement we have. 1588 switch (Lexer.getKind()) { 1589 case AsmToken::Colon: { 1590 if (!getTargetParser().isLabel(ID)) 1591 break; 1592 if (checkForValidSection()) 1593 return true; 1594 1595 // identifier ':' -> Label. 1596 Lex(); 1597 1598 // Diagnose attempt to use '.' as a label. 1599 if (IDVal == ".") 1600 return Error(IDLoc, "invalid use of pseudo-symbol '.' as a label"); 1601 1602 // Diagnose attempt to use a variable as a label. 1603 // 1604 // FIXME: Diagnostics. Note the location of the definition as a label. 1605 // FIXME: This doesn't diagnose assignment to a symbol which has been 1606 // implicitly marked as external. 1607 MCSymbol *Sym; 1608 if (LocalLabelVal == -1) { 1609 if (ParsingMSInlineAsm && SI) { 1610 StringRef RewrittenLabel = 1611 SI->LookupInlineAsmLabel(IDVal, getSourceManager(), IDLoc, true); 1612 assert(!RewrittenLabel.empty() && 1613 "We should have an internal name here."); 1614 Info.AsmRewrites->emplace_back(AOK_Label, IDLoc, IDVal.size(), 1615 RewrittenLabel); 1616 IDVal = RewrittenLabel; 1617 } 1618 Sym = getContext().getOrCreateSymbol(IDVal); 1619 } else 1620 Sym = Ctx.createDirectionalLocalSymbol(LocalLabelVal); 1621 // End of Labels should be treated as end of line for lexing 1622 // purposes but that information is not available to the Lexer who 1623 // does not understand Labels. This may cause us to see a Hash 1624 // here instead of a preprocessor line comment. 1625 if (getTok().is(AsmToken::Hash)) { 1626 StringRef CommentStr = parseStringToEndOfStatement(); 1627 Lexer.Lex(); 1628 Lexer.UnLex(AsmToken(AsmToken::EndOfStatement, CommentStr)); 1629 } 1630 1631 // Consume any end of statement token, if present, to avoid spurious 1632 // AddBlankLine calls(). 1633 if (getTok().is(AsmToken::EndOfStatement)) { 1634 Lex(); 1635 } 1636 1637 getTargetParser().doBeforeLabelEmit(Sym); 1638 1639 // Emit the label. 1640 if (!getTargetParser().isParsingMSInlineAsm()) 1641 Out.emitLabel(Sym, IDLoc); 1642 1643 // If we are generating dwarf for assembly source files then gather the 1644 // info to make a dwarf label entry for this label if needed. 1645 if (enabledGenDwarfForAssembly()) 1646 MCGenDwarfLabelEntry::Make(Sym, &getStreamer(), getSourceManager(), 1647 IDLoc); 1648 1649 getTargetParser().onLabelParsed(Sym); 1650 1651 return false; 1652 } 1653 1654 default: // Normal instruction or directive. 1655 break; 1656 } 1657 1658 // If macros are enabled, check to see if this is a macro instantiation. 1659 if (const MCAsmMacro *M = getContext().lookupMacro(IDVal)) { 1660 return handleMacroEntry(M, IDLoc); 1661 } 1662 1663 // Otherwise, we have a normal instruction or directive. 1664 1665 if (DirKind != DK_NO_DIRECTIVE) { 1666 // There are several entities interested in parsing directives: 1667 // 1668 // 1. Asm parser extensions. For example, platform-specific parsers 1669 // (like the ELF parser) register themselves as extensions. 1670 // 2. The target-specific assembly parser. Some directives are target 1671 // specific or may potentially behave differently on certain targets. 1672 // 3. The generic directive parser implemented by this class. These are 1673 // all the directives that behave in a target and platform independent 1674 // manner, or at least have a default behavior that's shared between 1675 // all targets and platforms. 1676 1677 getTargetParser().flushPendingInstructions(getStreamer()); 1678 1679 // First, check the extension directive map to see if any extension has 1680 // registered itself to parse this directive. 1681 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1682 ExtensionDirectiveMap.lookup(IDVal.lower()); 1683 if (Handler.first) 1684 return (*Handler.second)(Handler.first, IDVal, IDLoc); 1685 1686 // Next, let the target-specific assembly parser try. 1687 SMLoc StartTokLoc = getTok().getLoc(); 1688 bool TPDirectiveReturn = 1689 ID.is(AsmToken::Identifier) && getTargetParser().ParseDirective(ID); 1690 1691 if (hasPendingError()) 1692 return true; 1693 // Currently the return value should be true if we are 1694 // uninterested but as this is at odds with the standard parsing 1695 // convention (return true = error) we have instances of a parsed 1696 // directive that fails returning true as an error. Catch these 1697 // cases as best as possible errors here. 1698 if (TPDirectiveReturn && StartTokLoc != getTok().getLoc()) 1699 return true; 1700 // Return if we did some parsing or believe we succeeded. 1701 if (!TPDirectiveReturn || StartTokLoc != getTok().getLoc()) 1702 return false; 1703 1704 // Finally, if no one else is interested in this directive, it must be 1705 // generic and familiar to this class. 1706 switch (DirKind) { 1707 default: 1708 break; 1709 case DK_ASCII: 1710 return parseDirectiveAscii(IDVal, false); 1711 case DK_ASCIZ: 1712 case DK_STRING: 1713 return parseDirectiveAscii(IDVal, true); 1714 case DK_BYTE: 1715 case DK_SBYTE: 1716 case DK_DB: 1717 return parseDirectiveValue(IDVal, 1); 1718 case DK_WORD: 1719 case DK_SWORD: 1720 case DK_DW: 1721 return parseDirectiveValue(IDVal, 2); 1722 case DK_DWORD: 1723 case DK_SDWORD: 1724 case DK_DD: 1725 return parseDirectiveValue(IDVal, 4); 1726 case DK_FWORD: 1727 return parseDirectiveValue(IDVal, 6); 1728 case DK_QWORD: 1729 case DK_SQWORD: 1730 case DK_DQ: 1731 return parseDirectiveValue(IDVal, 8); 1732 case DK_REAL4: 1733 return parseDirectiveRealValue(IDVal, APFloat::IEEEsingle()); 1734 case DK_REAL8: 1735 return parseDirectiveRealValue(IDVal, APFloat::IEEEdouble()); 1736 case DK_ALIGN: 1737 return parseDirectiveAlign(); 1738 case DK_ORG: 1739 return parseDirectiveOrg(); 1740 case DK_EXTERN: 1741 eatToEndOfStatement(); // .extern is the default, ignore it. 1742 return false; 1743 case DK_PUBLIC: 1744 return parseDirectiveSymbolAttribute(MCSA_Global); 1745 case DK_COMM: 1746 return parseDirectiveComm(/*IsLocal=*/false); 1747 case DK_COMMENT: 1748 return parseDirectiveComment(IDLoc); 1749 case DK_INCLUDE: 1750 return parseDirectiveInclude(); 1751 case DK_REPT: 1752 return parseDirectiveRept(IDLoc, IDVal); 1753 case DK_IRP: 1754 return parseDirectiveIrp(IDLoc); 1755 case DK_IRPC: 1756 return parseDirectiveIrpc(IDLoc); 1757 case DK_ENDR: 1758 return parseDirectiveEndr(IDLoc); 1759 case DK_FILE: 1760 return parseDirectiveFile(IDLoc); 1761 case DK_LINE: 1762 return parseDirectiveLine(); 1763 case DK_LOC: 1764 return parseDirectiveLoc(); 1765 case DK_STABS: 1766 return parseDirectiveStabs(); 1767 case DK_CV_FILE: 1768 return parseDirectiveCVFile(); 1769 case DK_CV_FUNC_ID: 1770 return parseDirectiveCVFuncId(); 1771 case DK_CV_INLINE_SITE_ID: 1772 return parseDirectiveCVInlineSiteId(); 1773 case DK_CV_LOC: 1774 return parseDirectiveCVLoc(); 1775 case DK_CV_LINETABLE: 1776 return parseDirectiveCVLinetable(); 1777 case DK_CV_INLINE_LINETABLE: 1778 return parseDirectiveCVInlineLinetable(); 1779 case DK_CV_DEF_RANGE: 1780 return parseDirectiveCVDefRange(); 1781 case DK_CV_STRING: 1782 return parseDirectiveCVString(); 1783 case DK_CV_STRINGTABLE: 1784 return parseDirectiveCVStringTable(); 1785 case DK_CV_FILECHECKSUMS: 1786 return parseDirectiveCVFileChecksums(); 1787 case DK_CV_FILECHECKSUM_OFFSET: 1788 return parseDirectiveCVFileChecksumOffset(); 1789 case DK_CV_FPO_DATA: 1790 return parseDirectiveCVFPOData(); 1791 case DK_CFI_SECTIONS: 1792 return parseDirectiveCFISections(); 1793 case DK_CFI_STARTPROC: 1794 return parseDirectiveCFIStartProc(); 1795 case DK_CFI_ENDPROC: 1796 return parseDirectiveCFIEndProc(); 1797 case DK_CFI_DEF_CFA: 1798 return parseDirectiveCFIDefCfa(IDLoc); 1799 case DK_CFI_DEF_CFA_OFFSET: 1800 return parseDirectiveCFIDefCfaOffset(); 1801 case DK_CFI_ADJUST_CFA_OFFSET: 1802 return parseDirectiveCFIAdjustCfaOffset(); 1803 case DK_CFI_DEF_CFA_REGISTER: 1804 return parseDirectiveCFIDefCfaRegister(IDLoc); 1805 case DK_CFI_OFFSET: 1806 return parseDirectiveCFIOffset(IDLoc); 1807 case DK_CFI_REL_OFFSET: 1808 return parseDirectiveCFIRelOffset(IDLoc); 1809 case DK_CFI_PERSONALITY: 1810 return parseDirectiveCFIPersonalityOrLsda(true); 1811 case DK_CFI_LSDA: 1812 return parseDirectiveCFIPersonalityOrLsda(false); 1813 case DK_CFI_REMEMBER_STATE: 1814 return parseDirectiveCFIRememberState(); 1815 case DK_CFI_RESTORE_STATE: 1816 return parseDirectiveCFIRestoreState(); 1817 case DK_CFI_SAME_VALUE: 1818 return parseDirectiveCFISameValue(IDLoc); 1819 case DK_CFI_RESTORE: 1820 return parseDirectiveCFIRestore(IDLoc); 1821 case DK_CFI_ESCAPE: 1822 return parseDirectiveCFIEscape(); 1823 case DK_CFI_RETURN_COLUMN: 1824 return parseDirectiveCFIReturnColumn(IDLoc); 1825 case DK_CFI_SIGNAL_FRAME: 1826 return parseDirectiveCFISignalFrame(); 1827 case DK_CFI_UNDEFINED: 1828 return parseDirectiveCFIUndefined(IDLoc); 1829 case DK_CFI_REGISTER: 1830 return parseDirectiveCFIRegister(IDLoc); 1831 case DK_CFI_WINDOW_SAVE: 1832 return parseDirectiveCFIWindowSave(); 1833 case DK_MACRO: 1834 return parseDirectiveMacro(IDLoc); 1835 case DK_ALTMACRO: 1836 case DK_NOALTMACRO: 1837 return parseDirectiveAltmacro(IDVal); 1838 case DK_EXITM: 1839 return parseDirectiveExitMacro(IDVal); 1840 case DK_ENDM: 1841 return parseDirectiveEndMacro(IDVal); 1842 case DK_PURGEM: 1843 return parseDirectivePurgeMacro(IDLoc); 1844 case DK_END: 1845 return parseDirectiveEnd(IDLoc); 1846 case DK_ERR: 1847 return parseDirectiveError(IDLoc); 1848 case DK_ERRB: 1849 return parseDirectiveErrorIfb(IDLoc, true); 1850 case DK_ERRNB: 1851 return parseDirectiveErrorIfb(IDLoc, false); 1852 case DK_ERRDEF: 1853 return parseDirectiveErrorIfdef(IDLoc, true); 1854 case DK_ERRNDEF: 1855 return parseDirectiveErrorIfdef(IDLoc, false); 1856 case DK_ERRDIF: 1857 return parseDirectiveErrorIfidn(IDLoc, /*ExpectEqual=*/false, 1858 /*CaseInsensitive=*/false); 1859 case DK_ERRDIFI: 1860 return parseDirectiveErrorIfidn(IDLoc, /*ExpectEqual=*/false, 1861 /*CaseInsensitive=*/true); 1862 case DK_ERRIDN: 1863 return parseDirectiveErrorIfidn(IDLoc, /*ExpectEqual=*/true, 1864 /*CaseInsensitive=*/false); 1865 case DK_ERRIDNI: 1866 return parseDirectiveErrorIfidn(IDLoc, /*ExpectEqual=*/true, 1867 /*CaseInsensitive=*/true); 1868 case DK_ERRE: 1869 return parseDirectiveErrorIfe(IDLoc, true); 1870 case DK_ERRNZ: 1871 return parseDirectiveErrorIfe(IDLoc, false); 1872 case DK_ECHO: 1873 return parseDirectiveEcho(); 1874 } 1875 1876 return Error(IDLoc, "unknown directive"); 1877 } 1878 1879 // Non-conditional Microsoft directives sometimes follow their first argument. 1880 const AsmToken nextTok = getTok(); 1881 const StringRef nextVal = nextTok.getString(); 1882 const SMLoc nextLoc = nextTok.getLoc(); 1883 1884 // There are several entities interested in parsing infix directives: 1885 // 1886 // 1. Asm parser extensions. For example, platform-specific parsers 1887 // (like the ELF parser) register themselves as extensions. 1888 // 2. The generic directive parser implemented by this class. These are 1889 // all the directives that behave in a target and platform independent 1890 // manner, or at least have a default behavior that's shared between 1891 // all targets and platforms. 1892 1893 getTargetParser().flushPendingInstructions(getStreamer()); 1894 1895 // First, check the extension directive map to see if any extension has 1896 // registered itself to parse this directive. 1897 std::pair<MCAsmParserExtension *, DirectiveHandler> Handler = 1898 ExtensionDirectiveMap.lookup(nextVal.lower()); 1899 if (Handler.first) { 1900 Lex(); 1901 Lexer.UnLex(ID); 1902 return (*Handler.second)(Handler.first, nextVal, nextLoc); 1903 } 1904 1905 // Finally, if no one else is interested in this directive, it must be 1906 // generic and familiar to this class. 1907 DirKindIt = DirectiveKindMap.find(nextVal.lower()); 1908 DirKind = (DirKindIt == DirectiveKindMap.end()) 1909 ? DK_NO_DIRECTIVE 1910 : DirKindIt->getValue(); 1911 switch (DirKind) { 1912 default: 1913 break; 1914 case DK_ASSIGN: 1915 case DK_EQU: 1916 case DK_TEXTEQU: 1917 Lex(); 1918 return parseDirectiveEquate(nextVal, IDVal, DirKind); 1919 case DK_BYTE: 1920 case DK_DB: 1921 Lex(); 1922 return parseDirectiveNamedValue(nextVal, 1, IDVal, IDLoc); 1923 case DK_WORD: 1924 case DK_DW: 1925 Lex(); 1926 return parseDirectiveNamedValue(nextVal, 2, IDVal, IDLoc); 1927 case DK_DWORD: 1928 case DK_DD: 1929 Lex(); 1930 return parseDirectiveNamedValue(nextVal, 4, IDVal, IDLoc); 1931 case DK_FWORD: 1932 Lex(); 1933 return parseDirectiveNamedValue(nextVal, 6, IDVal, IDLoc); 1934 case DK_QWORD: 1935 case DK_DQ: 1936 Lex(); 1937 return parseDirectiveNamedValue(nextVal, 8, IDVal, IDLoc); 1938 case DK_REAL4: 1939 Lex(); 1940 return parseDirectiveNamedRealValue(nextVal, APFloat::IEEEsingle(), IDVal, 1941 IDLoc); 1942 case DK_REAL8: 1943 Lex(); 1944 return parseDirectiveNamedRealValue(nextVal, APFloat::IEEEdouble(), IDVal, 1945 IDLoc); 1946 } 1947 1948 // __asm _emit or __asm __emit 1949 if (ParsingMSInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || 1950 IDVal == "_EMIT" || IDVal == "__EMIT")) 1951 return parseDirectiveMSEmit(IDLoc, Info, IDVal.size()); 1952 1953 // __asm align 1954 if (ParsingMSInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) 1955 return parseDirectiveMSAlign(IDLoc, Info); 1956 1957 if (ParsingMSInlineAsm && (IDVal == "even" || IDVal == "EVEN")) 1958 Info.AsmRewrites->emplace_back(AOK_EVEN, IDLoc, 4); 1959 if (checkForValidSection()) 1960 return true; 1961 1962 // Canonicalize the opcode to lower case. 1963 std::string OpcodeStr = IDVal.lower(); 1964 ParseInstructionInfo IInfo(Info.AsmRewrites); 1965 bool ParseHadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, ID, 1966 Info.ParsedOperands); 1967 Info.ParseError = ParseHadError; 1968 1969 // Dump the parsed representation, if requested. 1970 if (getShowParsedOperands()) { 1971 SmallString<256> Str; 1972 raw_svector_ostream OS(Str); 1973 OS << "parsed instruction: ["; 1974 for (unsigned i = 0; i != Info.ParsedOperands.size(); ++i) { 1975 if (i != 0) 1976 OS << ", "; 1977 Info.ParsedOperands[i]->print(OS); 1978 } 1979 OS << "]"; 1980 1981 printMessage(IDLoc, SourceMgr::DK_Note, OS.str()); 1982 } 1983 1984 // Fail even if ParseInstruction erroneously returns false. 1985 if (hasPendingError() || ParseHadError) 1986 return true; 1987 1988 // If we are generating dwarf for the current section then generate a .loc 1989 // directive for the instruction. 1990 if (!ParseHadError && enabledGenDwarfForAssembly() && 1991 getContext().getGenDwarfSectionSyms().count( 1992 getStreamer().getCurrentSectionOnly())) { 1993 unsigned Line; 1994 if (ActiveMacros.empty()) 1995 Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); 1996 else 1997 Line = SrcMgr.FindLineNumber(ActiveMacros.front()->InstantiationLoc, 1998 ActiveMacros.front()->ExitBuffer); 1999 2000 // If we previously parsed a cpp hash file line comment then make sure the 2001 // current Dwarf File is for the CppHashFilename if not then emit the 2002 // Dwarf File table for it and adjust the line number for the .loc. 2003 if (!CppHashInfo.Filename.empty()) { 2004 unsigned FileNumber = getStreamer().emitDwarfFileDirective( 2005 0, StringRef(), CppHashInfo.Filename); 2006 getContext().setGenDwarfFileNumber(FileNumber); 2007 2008 unsigned CppHashLocLineNo = 2009 SrcMgr.FindLineNumber(CppHashInfo.Loc, CppHashInfo.Buf); 2010 Line = CppHashInfo.LineNumber - 1 + (Line - CppHashLocLineNo); 2011 } 2012 2013 getStreamer().emitDwarfLocDirective( 2014 getContext().getGenDwarfFileNumber(), Line, 0, 2015 DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0, 0, 0, 2016 StringRef()); 2017 } 2018 2019 // If parsing succeeded, match the instruction. 2020 if (!ParseHadError) { 2021 uint64_t ErrorInfo; 2022 if (getTargetParser().MatchAndEmitInstruction( 2023 IDLoc, Info.Opcode, Info.ParsedOperands, Out, ErrorInfo, 2024 getTargetParser().isParsingMSInlineAsm())) 2025 return true; 2026 } 2027 return false; 2028 } 2029 2030 // Parse and erase curly braces marking block start/end 2031 bool MasmParser::parseCurlyBlockScope( 2032 SmallVectorImpl<AsmRewrite> &AsmStrRewrites) { 2033 // Identify curly brace marking block start/end 2034 if (Lexer.isNot(AsmToken::LCurly) && Lexer.isNot(AsmToken::RCurly)) 2035 return false; 2036 2037 SMLoc StartLoc = Lexer.getLoc(); 2038 Lex(); // Eat the brace 2039 if (Lexer.is(AsmToken::EndOfStatement)) 2040 Lex(); // Eat EndOfStatement following the brace 2041 2042 // Erase the block start/end brace from the output asm string 2043 AsmStrRewrites.emplace_back(AOK_Skip, StartLoc, Lexer.getLoc().getPointer() - 2044 StartLoc.getPointer()); 2045 return true; 2046 } 2047 2048 /// parseCppHashLineFilenameComment as this: 2049 /// ::= # number "filename" 2050 bool MasmParser::parseCppHashLineFilenameComment(SMLoc L) { 2051 Lex(); // Eat the hash token. 2052 // Lexer only ever emits HashDirective if it fully formed if it's 2053 // done the checking already so this is an internal error. 2054 assert(getTok().is(AsmToken::Integer) && 2055 "Lexing Cpp line comment: Expected Integer"); 2056 int64_t LineNumber = getTok().getIntVal(); 2057 Lex(); 2058 assert(getTok().is(AsmToken::String) && 2059 "Lexing Cpp line comment: Expected String"); 2060 StringRef Filename = getTok().getString(); 2061 Lex(); 2062 2063 // Get rid of the enclosing quotes. 2064 Filename = Filename.substr(1, Filename.size() - 2); 2065 2066 // Save the SMLoc, Filename and LineNumber for later use by diagnostics 2067 // and possibly DWARF file info. 2068 CppHashInfo.Loc = L; 2069 CppHashInfo.Filename = Filename; 2070 CppHashInfo.LineNumber = LineNumber; 2071 CppHashInfo.Buf = CurBuffer; 2072 if (FirstCppHashFilename.empty()) 2073 FirstCppHashFilename = Filename; 2074 return false; 2075 } 2076 2077 /// will use the last parsed cpp hash line filename comment 2078 /// for the Filename and LineNo if any in the diagnostic. 2079 void MasmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { 2080 const MasmParser *Parser = static_cast<const MasmParser *>(Context); 2081 raw_ostream &OS = errs(); 2082 2083 const SourceMgr &DiagSrcMgr = *Diag.getSourceMgr(); 2084 SMLoc DiagLoc = Diag.getLoc(); 2085 unsigned DiagBuf = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2086 unsigned CppHashBuf = 2087 Parser->SrcMgr.FindBufferContainingLoc(Parser->CppHashInfo.Loc); 2088 2089 // Like SourceMgr::printMessage() we need to print the include stack if any 2090 // before printing the message. 2091 unsigned DiagCurBuffer = DiagSrcMgr.FindBufferContainingLoc(DiagLoc); 2092 if (!Parser->SavedDiagHandler && DiagCurBuffer && 2093 DiagCurBuffer != DiagSrcMgr.getMainFileID()) { 2094 SMLoc ParentIncludeLoc = DiagSrcMgr.getParentIncludeLoc(DiagCurBuffer); 2095 DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); 2096 } 2097 2098 // If we have not parsed a cpp hash line filename comment or the source 2099 // manager changed or buffer changed (like in a nested include) then just 2100 // print the normal diagnostic using its Filename and LineNo. 2101 if (!Parser->CppHashInfo.LineNumber || &DiagSrcMgr != &Parser->SrcMgr || 2102 DiagBuf != CppHashBuf) { 2103 if (Parser->SavedDiagHandler) 2104 Parser->SavedDiagHandler(Diag, Parser->SavedDiagContext); 2105 else 2106 Diag.print(nullptr, OS); 2107 return; 2108 } 2109 2110 // Use the CppHashFilename and calculate a line number based on the 2111 // CppHashInfo.Loc and CppHashInfo.LineNumber relative to this Diag's SMLoc 2112 // for the diagnostic. 2113 const std::string &Filename = std::string(Parser->CppHashInfo.Filename); 2114 2115 int DiagLocLineNo = DiagSrcMgr.FindLineNumber(DiagLoc, DiagBuf); 2116 int CppHashLocLineNo = 2117 Parser->SrcMgr.FindLineNumber(Parser->CppHashInfo.Loc, CppHashBuf); 2118 int LineNo = 2119 Parser->CppHashInfo.LineNumber - 1 + (DiagLocLineNo - CppHashLocLineNo); 2120 2121 SMDiagnostic NewDiag(*Diag.getSourceMgr(), Diag.getLoc(), Filename, LineNo, 2122 Diag.getColumnNo(), Diag.getKind(), Diag.getMessage(), 2123 Diag.getLineContents(), Diag.getRanges()); 2124 2125 if (Parser->SavedDiagHandler) 2126 Parser->SavedDiagHandler(NewDiag, Parser->SavedDiagContext); 2127 else 2128 NewDiag.print(nullptr, OS); 2129 } 2130 2131 // FIXME: This is mostly duplicated from the function in AsmLexer.cpp. The 2132 // difference being that that function accepts '@' as part of identifiers and 2133 // we can't do that. AsmLexer.cpp should probably be changed to handle 2134 // '@' as a special case when needed. 2135 static bool isIdentifierChar(char c) { 2136 return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || 2137 c == '.'; 2138 } 2139 2140 bool MasmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, 2141 ArrayRef<MCAsmMacroParameter> Parameters, 2142 ArrayRef<MCAsmMacroArgument> A, 2143 bool EnableAtPseudoVariable, SMLoc L) { 2144 unsigned NParameters = Parameters.size(); 2145 bool HasVararg = NParameters ? Parameters.back().Vararg : false; 2146 if ((!IsDarwin || NParameters != 0) && NParameters != A.size()) 2147 return Error(L, "Wrong number of arguments"); 2148 2149 // A macro without parameters is handled differently on Darwin: 2150 // gas accepts no arguments and does no substitutions 2151 while (!Body.empty()) { 2152 // Scan for the next substitution. 2153 std::size_t End = Body.size(), Pos = 0; 2154 for (; Pos != End; ++Pos) { 2155 // Check for a substitution or escape. 2156 if (IsDarwin && !NParameters) { 2157 // This macro has no parameters, look for $0, $1, etc. 2158 if (Body[Pos] != '$' || Pos + 1 == End) 2159 continue; 2160 2161 char Next = Body[Pos + 1]; 2162 if (Next == '$' || Next == 'n' || 2163 isdigit(static_cast<unsigned char>(Next))) 2164 break; 2165 } else { 2166 // This macro has parameters, look for \foo, \bar, etc. 2167 if (Body[Pos] == '\\' && Pos + 1 != End) 2168 break; 2169 } 2170 } 2171 2172 // Add the prefix. 2173 OS << Body.slice(0, Pos); 2174 2175 // Check if we reached the end. 2176 if (Pos == End) 2177 break; 2178 2179 if (IsDarwin && !NParameters) { 2180 switch (Body[Pos + 1]) { 2181 // $$ => $ 2182 case '$': 2183 OS << '$'; 2184 break; 2185 2186 // $n => number of arguments 2187 case 'n': 2188 OS << A.size(); 2189 break; 2190 2191 // $[0-9] => argument 2192 default: { 2193 // Missing arguments are ignored. 2194 unsigned Index = Body[Pos + 1] - '0'; 2195 if (Index >= A.size()) 2196 break; 2197 2198 // Otherwise substitute with the token values, with spaces eliminated. 2199 for (const AsmToken &Token : A[Index]) 2200 OS << Token.getString(); 2201 break; 2202 } 2203 } 2204 Pos += 2; 2205 } else { 2206 unsigned I = Pos + 1; 2207 2208 // Check for the \@ pseudo-variable. 2209 if (EnableAtPseudoVariable && Body[I] == '@' && I + 1 != End) 2210 ++I; 2211 else 2212 while (isIdentifierChar(Body[I]) && I + 1 != End) 2213 ++I; 2214 2215 const char *Begin = Body.data() + Pos + 1; 2216 StringRef Argument(Begin, I - (Pos + 1)); 2217 unsigned Index = 0; 2218 2219 if (Argument == "@") { 2220 OS << NumOfMacroInstantiations; 2221 Pos += 2; 2222 } else { 2223 for (; Index < NParameters; ++Index) 2224 if (Parameters[Index].Name == Argument) 2225 break; 2226 2227 if (Index == NParameters) { 2228 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 2229 Pos += 3; 2230 else { 2231 OS << '\\' << Argument; 2232 Pos = I; 2233 } 2234 } else { 2235 bool VarargParameter = HasVararg && Index == (NParameters - 1); 2236 for (const AsmToken &Token : A[Index]) 2237 // For altmacro mode, you can write '%expr'. 2238 // The prefix '%' evaluates the expression 'expr' 2239 // and uses the result as a string (e.g. replace %(1+2) with the 2240 // string "3"). 2241 // Here, we identify the integer token which is the result of the 2242 // absolute expression evaluation and replace it with its string 2243 // representation. 2244 if (AltMacroMode && Token.getString().front() == '%' && 2245 Token.is(AsmToken::Integer)) 2246 // Emit an integer value to the buffer. 2247 OS << Token.getIntVal(); 2248 // Only Token that was validated as a string and begins with '<' 2249 // is considered altMacroString!!! 2250 else if (AltMacroMode && Token.getString().front() == '<' && 2251 Token.is(AsmToken::String)) { 2252 OS << angleBracketString(Token.getStringContents()); 2253 } 2254 // We expect no quotes around the string's contents when 2255 // parsing for varargs. 2256 else if (Token.isNot(AsmToken::String) || VarargParameter) 2257 OS << Token.getString(); 2258 else 2259 OS << Token.getStringContents(); 2260 2261 Pos += 1 + Argument.size(); 2262 } 2263 } 2264 } 2265 // Update the scan point. 2266 Body = Body.substr(Pos); 2267 } 2268 2269 return false; 2270 } 2271 2272 static bool isOperator(AsmToken::TokenKind kind) { 2273 switch (kind) { 2274 default: 2275 return false; 2276 case AsmToken::Plus: 2277 case AsmToken::Minus: 2278 case AsmToken::Tilde: 2279 case AsmToken::Slash: 2280 case AsmToken::Star: 2281 case AsmToken::Dot: 2282 case AsmToken::Equal: 2283 case AsmToken::EqualEqual: 2284 case AsmToken::Pipe: 2285 case AsmToken::PipePipe: 2286 case AsmToken::Caret: 2287 case AsmToken::Amp: 2288 case AsmToken::AmpAmp: 2289 case AsmToken::Exclaim: 2290 case AsmToken::ExclaimEqual: 2291 case AsmToken::Less: 2292 case AsmToken::LessEqual: 2293 case AsmToken::LessLess: 2294 case AsmToken::LessGreater: 2295 case AsmToken::Greater: 2296 case AsmToken::GreaterEqual: 2297 case AsmToken::GreaterGreater: 2298 return true; 2299 } 2300 } 2301 2302 namespace { 2303 2304 class AsmLexerSkipSpaceRAII { 2305 public: 2306 AsmLexerSkipSpaceRAII(AsmLexer &Lexer, bool SkipSpace) : Lexer(Lexer) { 2307 Lexer.setSkipSpace(SkipSpace); 2308 } 2309 2310 ~AsmLexerSkipSpaceRAII() { 2311 Lexer.setSkipSpace(true); 2312 } 2313 2314 private: 2315 AsmLexer &Lexer; 2316 }; 2317 2318 } // end anonymous namespace 2319 2320 bool MasmParser::parseMacroArgument(MCAsmMacroArgument &MA, bool Vararg) { 2321 2322 if (Vararg) { 2323 if (Lexer.isNot(AsmToken::EndOfStatement)) { 2324 StringRef Str = parseStringToEndOfStatement(); 2325 MA.emplace_back(AsmToken::String, Str); 2326 } 2327 return false; 2328 } 2329 2330 unsigned ParenLevel = 0; 2331 2332 // Darwin doesn't use spaces to delmit arguments. 2333 AsmLexerSkipSpaceRAII ScopedSkipSpace(Lexer, IsDarwin); 2334 2335 bool SpaceEaten; 2336 2337 while (true) { 2338 SpaceEaten = false; 2339 if (Lexer.is(AsmToken::Eof) || Lexer.is(AsmToken::Equal)) 2340 return TokError("unexpected token in macro instantiation"); 2341 2342 if (ParenLevel == 0) { 2343 2344 if (Lexer.is(AsmToken::Comma)) 2345 break; 2346 2347 if (Lexer.is(AsmToken::Space)) { 2348 SpaceEaten = true; 2349 Lexer.Lex(); // Eat spaces 2350 } 2351 2352 // Spaces can delimit parameters, but could also be part an expression. 2353 // If the token after a space is an operator, add the token and the next 2354 // one into this argument 2355 if (!IsDarwin) { 2356 if (isOperator(Lexer.getKind())) { 2357 MA.push_back(getTok()); 2358 Lexer.Lex(); 2359 2360 // Whitespace after an operator can be ignored. 2361 if (Lexer.is(AsmToken::Space)) 2362 Lexer.Lex(); 2363 2364 continue; 2365 } 2366 } 2367 if (SpaceEaten) 2368 break; 2369 } 2370 2371 // handleMacroEntry relies on not advancing the lexer here 2372 // to be able to fill in the remaining default parameter values 2373 if (Lexer.is(AsmToken::EndOfStatement)) 2374 break; 2375 2376 // Adjust the current parentheses level. 2377 if (Lexer.is(AsmToken::LParen)) 2378 ++ParenLevel; 2379 else if (Lexer.is(AsmToken::RParen) && ParenLevel) 2380 --ParenLevel; 2381 2382 // Append the token to the current argument list. 2383 MA.push_back(getTok()); 2384 Lexer.Lex(); 2385 } 2386 2387 if (ParenLevel != 0) 2388 return TokError("unbalanced parentheses in macro argument"); 2389 return false; 2390 } 2391 2392 // Parse the macro instantiation arguments. 2393 bool MasmParser::parseMacroArguments(const MCAsmMacro *M, 2394 MCAsmMacroArguments &A) { 2395 const unsigned NParameters = M ? M->Parameters.size() : 0; 2396 bool NamedParametersFound = false; 2397 SmallVector<SMLoc, 4> FALocs; 2398 2399 A.resize(NParameters); 2400 FALocs.resize(NParameters); 2401 2402 // Parse two kinds of macro invocations: 2403 // - macros defined without any parameters accept an arbitrary number of them 2404 // - macros defined with parameters accept at most that many of them 2405 bool HasVararg = NParameters ? M->Parameters.back().Vararg : false; 2406 for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; 2407 ++Parameter) { 2408 SMLoc IDLoc = Lexer.getLoc(); 2409 MCAsmMacroParameter FA; 2410 2411 if (Lexer.is(AsmToken::Identifier) && Lexer.peekTok().is(AsmToken::Equal)) { 2412 if (parseIdentifier(FA.Name)) 2413 return Error(IDLoc, "invalid argument identifier for formal argument"); 2414 2415 if (Lexer.isNot(AsmToken::Equal)) 2416 return TokError("expected '=' after formal parameter identifier"); 2417 2418 Lex(); 2419 2420 NamedParametersFound = true; 2421 } 2422 bool Vararg = HasVararg && Parameter == (NParameters - 1); 2423 2424 if (NamedParametersFound && FA.Name.empty()) 2425 return Error(IDLoc, "cannot mix positional and keyword arguments"); 2426 2427 SMLoc StrLoc = Lexer.getLoc(); 2428 SMLoc EndLoc; 2429 if (AltMacroMode && Lexer.is(AsmToken::Percent)) { 2430 const MCExpr *AbsoluteExp; 2431 int64_t Value; 2432 /// Eat '%' 2433 Lex(); 2434 if (parseExpression(AbsoluteExp, EndLoc)) 2435 return false; 2436 if (!AbsoluteExp->evaluateAsAbsolute(Value, 2437 getStreamer().getAssemblerPtr())) 2438 return Error(StrLoc, "expected absolute expression"); 2439 const char *StrChar = StrLoc.getPointer(); 2440 const char *EndChar = EndLoc.getPointer(); 2441 AsmToken newToken(AsmToken::Integer, 2442 StringRef(StrChar, EndChar - StrChar), Value); 2443 FA.Value.push_back(newToken); 2444 } else if (AltMacroMode && Lexer.is(AsmToken::Less) && 2445 isAngleBracketString(StrLoc, EndLoc)) { 2446 const char *StrChar = StrLoc.getPointer(); 2447 const char *EndChar = EndLoc.getPointer(); 2448 jumpToLoc(EndLoc, CurBuffer); 2449 /// Eat from '<' to '>' 2450 Lex(); 2451 AsmToken newToken(AsmToken::String, 2452 StringRef(StrChar, EndChar - StrChar)); 2453 FA.Value.push_back(newToken); 2454 } else if(parseMacroArgument(FA.Value, Vararg)) 2455 return true; 2456 2457 unsigned PI = Parameter; 2458 if (!FA.Name.empty()) { 2459 unsigned FAI = 0; 2460 for (FAI = 0; FAI < NParameters; ++FAI) 2461 if (M->Parameters[FAI].Name == FA.Name) 2462 break; 2463 2464 if (FAI >= NParameters) { 2465 assert(M && "expected macro to be defined"); 2466 return Error(IDLoc, "parameter named '" + FA.Name + 2467 "' does not exist for macro '" + M->Name + "'"); 2468 } 2469 PI = FAI; 2470 } 2471 2472 if (!FA.Value.empty()) { 2473 if (A.size() <= PI) 2474 A.resize(PI + 1); 2475 A[PI] = FA.Value; 2476 2477 if (FALocs.size() <= PI) 2478 FALocs.resize(PI + 1); 2479 2480 FALocs[PI] = Lexer.getLoc(); 2481 } 2482 2483 // At the end of the statement, fill in remaining arguments that have 2484 // default values. If there aren't any, then the next argument is 2485 // required but missing 2486 if (Lexer.is(AsmToken::EndOfStatement)) { 2487 bool Failure = false; 2488 for (unsigned FAI = 0; FAI < NParameters; ++FAI) { 2489 if (A[FAI].empty()) { 2490 if (M->Parameters[FAI].Required) { 2491 Error(FALocs[FAI].isValid() ? FALocs[FAI] : Lexer.getLoc(), 2492 "missing value for required parameter " 2493 "'" + M->Parameters[FAI].Name + "' in macro '" + M->Name + "'"); 2494 Failure = true; 2495 } 2496 2497 if (!M->Parameters[FAI].Value.empty()) 2498 A[FAI] = M->Parameters[FAI].Value; 2499 } 2500 } 2501 return Failure; 2502 } 2503 2504 if (Lexer.is(AsmToken::Comma)) 2505 Lex(); 2506 } 2507 2508 return TokError("too many positional arguments"); 2509 } 2510 2511 bool MasmParser::handleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { 2512 // Arbitrarily limit macro nesting depth (default matches 'as'). We can 2513 // eliminate this, although we should protect against infinite loops. 2514 unsigned MaxNestingDepth = AsmMacroMaxNestingDepth; 2515 if (ActiveMacros.size() == MaxNestingDepth) { 2516 std::ostringstream MaxNestingDepthError; 2517 MaxNestingDepthError << "macros cannot be nested more than " 2518 << MaxNestingDepth << " levels deep." 2519 << " Use -asm-macro-max-nesting-depth to increase " 2520 "this limit."; 2521 return TokError(MaxNestingDepthError.str()); 2522 } 2523 2524 MCAsmMacroArguments A; 2525 if (parseMacroArguments(M, A)) 2526 return true; 2527 2528 // Macro instantiation is lexical, unfortunately. We construct a new buffer 2529 // to hold the macro body with substitutions. 2530 SmallString<256> Buf; 2531 StringRef Body = M->Body; 2532 raw_svector_ostream OS(Buf); 2533 2534 if (expandMacro(OS, Body, M->Parameters, A, true, getTok().getLoc())) 2535 return true; 2536 2537 // We include the .endmacro in the buffer as our cue to exit the macro 2538 // instantiation. 2539 OS << ".endmacro\n"; 2540 2541 std::unique_ptr<MemoryBuffer> Instantiation = 2542 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 2543 2544 // Create the macro instantiation object and add to the current macro 2545 // instantiation stack. 2546 MacroInstantiation *MI = new MacroInstantiation{ 2547 NameLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 2548 ActiveMacros.push_back(MI); 2549 2550 ++NumOfMacroInstantiations; 2551 2552 // Jump to the macro instantiation and prime the lexer. 2553 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 2554 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 2555 Lex(); 2556 2557 return false; 2558 } 2559 2560 void MasmParser::handleMacroExit() { 2561 // Jump to the EndOfStatement we should return to, and consume it. 2562 jumpToLoc(ActiveMacros.back()->ExitLoc, ActiveMacros.back()->ExitBuffer); 2563 Lex(); 2564 2565 // Pop the instantiation entry. 2566 delete ActiveMacros.back(); 2567 ActiveMacros.pop_back(); 2568 } 2569 2570 /// parseIdentifier: 2571 /// ::= identifier 2572 /// ::= string 2573 bool MasmParser::parseIdentifier(StringRef &Res) { 2574 // The assembler has relaxed rules for accepting identifiers, in particular we 2575 // allow things like '.globl $foo' and '.def @feat.00', which would normally 2576 // be separate tokens. At this level, we have already lexed so we cannot 2577 // (currently) handle this as a context dependent token, instead we detect 2578 // adjacent tokens and return the combined identifier. 2579 if (Lexer.is(AsmToken::Dollar) || Lexer.is(AsmToken::At)) { 2580 SMLoc PrefixLoc = getLexer().getLoc(); 2581 2582 // Consume the prefix character, and check for a following identifier. 2583 2584 AsmToken Buf[1]; 2585 Lexer.peekTokens(Buf, false); 2586 2587 if (Buf[0].isNot(AsmToken::Identifier)) 2588 return true; 2589 2590 // We have a '$' or '@' followed by an identifier, make sure they are adjacent. 2591 if (PrefixLoc.getPointer() + 1 != Buf[0].getLoc().getPointer()) 2592 return true; 2593 2594 // eat $ or @ 2595 Lexer.Lex(); // Lexer's Lex guarantees consecutive token. 2596 // Construct the joined identifier and consume the token. 2597 Res = 2598 StringRef(PrefixLoc.getPointer(), getTok().getIdentifier().size() + 1); 2599 Lex(); // Parser Lex to maintain invariants. 2600 return false; 2601 } 2602 2603 if (Lexer.isNot(AsmToken::Identifier) && Lexer.isNot(AsmToken::String)) 2604 return true; 2605 2606 Res = getTok().getIdentifier(); 2607 2608 Lex(); // Consume the identifier token. 2609 2610 return false; 2611 } 2612 2613 /// parseDirectiveEquate: 2614 /// ::= name "=" expression 2615 /// | name "equ" expression (not redefinable) 2616 /// | name "equ" text-list 2617 /// | name "textequ" text-list 2618 bool MasmParser::parseDirectiveEquate(StringRef IDVal, StringRef Name, 2619 DirectiveKind DirKind) { 2620 Variable &Var = Variables[Name]; 2621 if (Var.Name.empty()) { 2622 Var.Name = Name; 2623 } else if (!Var.Redefinable) { 2624 return TokError("invalid variable redefinition"); 2625 } 2626 Var.Redefinable = (DirKind != DK_EQU); 2627 2628 if (DirKind == DK_EQU || DirKind == DK_TEXTEQU) { 2629 // "equ" and "textequ" both allow text expressions. 2630 std::string Value; 2631 if (!parseTextItem(Value)) { 2632 Var.IsText = true; 2633 Var.TextValue = Value; 2634 2635 // Accept a text-list, not just one text-item 2636 auto parseItem = [&]() -> bool { 2637 if (parseTextItem(Value)) 2638 return true; 2639 Var.TextValue += Value; 2640 return false; 2641 }; 2642 if (parseOptionalToken(AsmToken::Comma) && parseMany(parseItem)) 2643 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2644 2645 return false; 2646 } 2647 } 2648 if (DirKind == DK_TEXTEQU) 2649 return TokError("expected <text> in '" + Twine(IDVal) + "' directive"); 2650 2651 // Parse as expression assignment 2652 const MCExpr *Expr; 2653 SMLoc EndLoc, StartLoc = Lexer.getLoc(); 2654 if (parseExpression(Expr, EndLoc)) 2655 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2656 if (Expr->evaluateAsAbsolute(Var.NumericValue, 2657 getStreamer().getAssemblerPtr())) 2658 return false; 2659 2660 // Not an absolute expression; define as a text replacement. 2661 Var.IsText = true; 2662 Var.TextValue = StringRef(StartLoc.getPointer(), 2663 EndLoc.getPointer() - StartLoc.getPointer()).str(); 2664 return false; 2665 } 2666 2667 bool MasmParser::parseEscapedString(std::string &Data) { 2668 if (check(getTok().isNot(AsmToken::String), "expected string")) 2669 return true; 2670 2671 Data = ""; 2672 StringRef Str = getTok().getStringContents(); 2673 for (unsigned i = 0, e = Str.size(); i != e; ++i) { 2674 if (Str[i] != '\\') { 2675 Data += Str[i]; 2676 continue; 2677 } 2678 2679 // Recognize escaped characters. Note that this escape semantics currently 2680 // loosely follows Darwin 'as'. 2681 ++i; 2682 if (i == e) 2683 return TokError("unexpected backslash at end of string"); 2684 2685 // Recognize hex sequences similarly to GNU 'as'. 2686 if (Str[i] == 'x' || Str[i] == 'X') { 2687 size_t length = Str.size(); 2688 if (i + 1 >= length || !isHexDigit(Str[i + 1])) 2689 return TokError("invalid hexadecimal escape sequence"); 2690 2691 // Consume hex characters. GNU 'as' reads all hexadecimal characters and 2692 // then truncates to the lower 16 bits. Seems reasonable. 2693 unsigned Value = 0; 2694 while (i + 1 < length && isHexDigit(Str[i + 1])) 2695 Value = Value * 16 + hexDigitValue(Str[++i]); 2696 2697 Data += (unsigned char)(Value & 0xFF); 2698 continue; 2699 } 2700 2701 // Recognize octal sequences. 2702 if ((unsigned)(Str[i] - '0') <= 7) { 2703 // Consume up to three octal characters. 2704 unsigned Value = Str[i] - '0'; 2705 2706 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2707 ++i; 2708 Value = Value * 8 + (Str[i] - '0'); 2709 2710 if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) { 2711 ++i; 2712 Value = Value * 8 + (Str[i] - '0'); 2713 } 2714 } 2715 2716 if (Value > 255) 2717 return TokError("invalid octal escape sequence (out of range)"); 2718 2719 Data += (unsigned char)Value; 2720 continue; 2721 } 2722 2723 // Otherwise recognize individual escapes. 2724 switch (Str[i]) { 2725 default: 2726 // Just reject invalid escape sequences for now. 2727 return TokError("invalid escape sequence (unrecognized character)"); 2728 2729 case 'b': Data += '\b'; break; 2730 case 'f': Data += '\f'; break; 2731 case 'n': Data += '\n'; break; 2732 case 'r': Data += '\r'; break; 2733 case 't': Data += '\t'; break; 2734 case '"': Data += '"'; break; 2735 case '\\': Data += '\\'; break; 2736 } 2737 } 2738 2739 Lex(); 2740 return false; 2741 } 2742 2743 bool MasmParser::parseAngleBracketString(std::string &Data) { 2744 SMLoc EndLoc, StartLoc = getTok().getLoc(); 2745 if (isAngleBracketString(StartLoc, EndLoc)) { 2746 const char *StartChar = StartLoc.getPointer() + 1; 2747 const char *EndChar = EndLoc.getPointer() - 1; 2748 jumpToLoc(EndLoc, CurBuffer); 2749 /// Eat from '<' to '>' 2750 Lex(); 2751 2752 Data = angleBracketString(StringRef(StartChar, EndChar - StartChar)); 2753 return false; 2754 } 2755 return true; 2756 } 2757 2758 /// textItem ::= textLiteral | textMacroID | % constExpr 2759 bool MasmParser::parseTextItem(std::string &Data) { 2760 // TODO(epastor): Support textMacroID and % expansion of expressions. 2761 return parseAngleBracketString(Data); 2762 } 2763 2764 /// parseDirectiveAscii: 2765 /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] 2766 bool MasmParser::parseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { 2767 auto parseOp = [&]() -> bool { 2768 std::string Data; 2769 if (checkForValidSection() || parseEscapedString(Data)) 2770 return true; 2771 getStreamer().emitBytes(Data); 2772 if (ZeroTerminated) 2773 getStreamer().emitBytes(StringRef("\0", 1)); 2774 return false; 2775 }; 2776 2777 if (parseMany(parseOp)) 2778 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2779 return false; 2780 } 2781 2782 bool MasmParser::parseScalarInstList(unsigned Size, 2783 SmallVectorImpl<const MCExpr *> &Values) { 2784 do { 2785 if (getTok().is(AsmToken::String)) { 2786 StringRef Value = getTok().getStringContents(); 2787 if (Size == 1) { 2788 // Treat each character as an initializer. 2789 for (const char CharVal : Value) 2790 Values.push_back(MCConstantExpr::create(CharVal, getContext())); 2791 } else { 2792 // Treat the string as an initial value in big-endian representation. 2793 if (Value.size() > Size) 2794 return Error(getTok().getLoc(), "out of range literal value"); 2795 2796 uint64_t IntValue = 0; 2797 for (const unsigned char CharVal : Value.bytes()) 2798 IntValue = (IntValue << 8) | CharVal; 2799 Values.push_back(MCConstantExpr::create(IntValue, getContext())); 2800 } 2801 Lex(); 2802 } else { 2803 const MCExpr *Value; 2804 if (checkForValidSection() || parseExpression(Value)) 2805 return true; 2806 if (getTok().is(AsmToken::Identifier) && 2807 getTok().getString().equals_lower("dup")) { 2808 Lex(); // eat 'dup' 2809 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 2810 if (!MCE) 2811 return Error(Value->getLoc(), 2812 "cannot repeat value a non-constant number of times"); 2813 const int64_t Repetitions = MCE->getValue(); 2814 if (Repetitions < 0) 2815 return Error(Value->getLoc(), 2816 "cannot repeat value a negative number of times"); 2817 2818 SmallVector<const MCExpr *, 1> DuplicatedValues; 2819 if (parseToken(AsmToken::LParen, 2820 "parentheses required for 'dup' contents") || 2821 parseScalarInstList(Size, DuplicatedValues) || 2822 parseToken(AsmToken::RParen, "unmatched parentheses")) 2823 return true; 2824 2825 for (int i = 0; i < Repetitions; ++i) 2826 Values.append(DuplicatedValues.begin(), DuplicatedValues.end()); 2827 } else { 2828 Values.push_back(Value); 2829 } 2830 } 2831 2832 // Continue if we see a comma. (Also, allow line continuation.) 2833 } while (parseOptionalToken(AsmToken::Comma) && 2834 (getTok().isNot(AsmToken::EndOfStatement) || 2835 !parseToken(AsmToken::EndOfStatement))); 2836 2837 return false; 2838 } 2839 2840 /// parseDirectiveValue 2841 /// ::= (byte | word | ... ) [ expression (, expression)* ] 2842 bool MasmParser::parseDirectiveValue(StringRef IDVal, unsigned Size) { 2843 SmallVector<const MCExpr *, 1> Values; 2844 if (parseScalarInstList(Size, Values)) 2845 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2846 2847 for (const MCExpr *Value : Values) { 2848 // Special case constant expressions to match code generator. 2849 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 2850 assert(Size <= 8 && "Invalid size"); 2851 int64_t IntValue = MCE->getValue(); 2852 if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue)) 2853 return Error(MCE->getLoc(), "out of range literal value"); 2854 getStreamer().emitIntValue(IntValue, Size); 2855 } else { 2856 const MCSymbolRefExpr *MSE = dyn_cast<MCSymbolRefExpr>(Value); 2857 if (MSE && MSE->getSymbol().getName() == "?") { 2858 // ? initializer; treat as 0. 2859 getStreamer().emitIntValue(0, Size); 2860 } else { 2861 getStreamer().emitValue(Value, Size, Value->getLoc()); 2862 } 2863 } 2864 } 2865 return false; 2866 } 2867 2868 /// parseDirectiveNamedValue 2869 /// ::= name (byte | word | ... ) [ expression (, expression)* ] 2870 bool MasmParser::parseDirectiveNamedValue(StringRef IDVal, unsigned Size, 2871 StringRef Name, SMLoc NameLoc) { 2872 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 2873 getStreamer().emitLabel(Sym); 2874 return parseDirectiveValue(IDVal, Size); 2875 } 2876 2877 static bool parseHexOcta(MasmParser &Asm, uint64_t &hi, uint64_t &lo) { 2878 if (Asm.getTok().isNot(AsmToken::Integer) && 2879 Asm.getTok().isNot(AsmToken::BigNum)) 2880 return Asm.TokError("unknown token in expression"); 2881 SMLoc ExprLoc = Asm.getTok().getLoc(); 2882 APInt IntValue = Asm.getTok().getAPIntVal(); 2883 Asm.Lex(); 2884 if (!IntValue.isIntN(128)) 2885 return Asm.Error(ExprLoc, "out of range literal value"); 2886 if (!IntValue.isIntN(64)) { 2887 hi = IntValue.getHiBits(IntValue.getBitWidth() - 64).getZExtValue(); 2888 lo = IntValue.getLoBits(64).getZExtValue(); 2889 } else { 2890 hi = 0; 2891 lo = IntValue.getZExtValue(); 2892 } 2893 return false; 2894 } 2895 2896 bool MasmParser::parseRealValue(const fltSemantics &Semantics, APInt &Res) { 2897 // We don't truly support arithmetic on floating point expressions, so we 2898 // have to manually parse unary prefixes. 2899 bool IsNeg = false; 2900 if (getLexer().is(AsmToken::Minus)) { 2901 Lexer.Lex(); 2902 IsNeg = true; 2903 } else if (getLexer().is(AsmToken::Plus)) 2904 Lexer.Lex(); 2905 2906 if (Lexer.is(AsmToken::Error)) 2907 return TokError(Lexer.getErr()); 2908 if (Lexer.isNot(AsmToken::Integer) && Lexer.isNot(AsmToken::Real) && 2909 Lexer.isNot(AsmToken::Identifier)) 2910 return TokError("unexpected token in directive"); 2911 2912 // Convert to an APFloat. 2913 APFloat Value(Semantics); 2914 StringRef IDVal = getTok().getString(); 2915 if (getLexer().is(AsmToken::Identifier)) { 2916 if (!IDVal.compare_lower("infinity") || !IDVal.compare_lower("inf")) 2917 Value = APFloat::getInf(Semantics); 2918 else if (!IDVal.compare_lower("nan")) 2919 Value = APFloat::getNaN(Semantics, false, ~0); 2920 else 2921 return TokError("invalid floating point literal"); 2922 } else if (errorToBool( 2923 Value.convertFromString(IDVal, APFloat::rmNearestTiesToEven) 2924 .takeError())) 2925 return TokError("invalid floating point literal"); 2926 if (IsNeg) 2927 Value.changeSign(); 2928 2929 // Consume the numeric token. 2930 Lex(); 2931 2932 Res = Value.bitcastToAPInt(); 2933 2934 return false; 2935 } 2936 2937 bool MasmParser::parseRealInstList(const fltSemantics &Semantics, 2938 SmallVectorImpl<APInt> &ValuesAsInt) { 2939 do { 2940 const AsmToken NextTok = Lexer.peekTok(); 2941 if (NextTok.is(AsmToken::Identifier) && 2942 NextTok.getString().equals_lower("dup")) { 2943 const MCExpr *Value; 2944 if (parseExpression(Value) || parseToken(AsmToken::Identifier)) 2945 return true; 2946 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 2947 if (!MCE) 2948 return Error(Value->getLoc(), 2949 "cannot repeat value a non-constant number of times"); 2950 const int64_t Repetitions = MCE->getValue(); 2951 if (Repetitions < 0) 2952 return Error(Value->getLoc(), 2953 "cannot repeat value a negative number of times"); 2954 2955 SmallVector<APInt, 1> DuplicatedValues; 2956 if (parseToken(AsmToken::LParen, 2957 "parentheses required for 'dup' contents") || 2958 parseRealInstList(Semantics, DuplicatedValues) || 2959 parseToken(AsmToken::RParen, "unmatched parentheses")) 2960 return true; 2961 2962 for (int i = 0; i < Repetitions; ++i) 2963 ValuesAsInt.append(DuplicatedValues.begin(), DuplicatedValues.end()); 2964 } else { 2965 APInt AsInt; 2966 if (parseRealValue(Semantics, AsInt)) 2967 return true; 2968 ValuesAsInt.push_back(AsInt); 2969 } 2970 // Continue if we see a comma. (Also, allow line continuation.) 2971 } while (parseOptionalToken(AsmToken::Comma) && 2972 (getTok().isNot(AsmToken::EndOfStatement) || 2973 !parseToken(AsmToken::EndOfStatement))); 2974 2975 return false; 2976 } 2977 2978 /// parseDirectiveRealValue 2979 /// ::= (real4 | real8) [ expression (, expression)* ] 2980 bool MasmParser::parseDirectiveRealValue(StringRef IDVal, 2981 const fltSemantics &Semantics) { 2982 if (checkForValidSection()) 2983 return true; 2984 2985 SmallVector<APInt, 1> ValuesAsInt; 2986 if (parseRealInstList(Semantics, ValuesAsInt)) 2987 return addErrorSuffix(" in '" + Twine(IDVal) + "' directive"); 2988 2989 for (const APInt &AsInt : ValuesAsInt) { 2990 getStreamer().emitIntValue(AsInt.getLimitedValue(), 2991 AsInt.getBitWidth() / 8); 2992 } 2993 return false; 2994 } 2995 2996 /// parseDirectiveNamedRealValue 2997 /// ::= name (real4 | real8) [ expression (, expression)* ] 2998 bool MasmParser::parseDirectiveNamedRealValue(StringRef IDVal, 2999 const fltSemantics &Semantics, 3000 StringRef Name, SMLoc NameLoc) { 3001 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 3002 getStreamer().emitLabel(Sym); 3003 return parseDirectiveRealValue(IDVal, Semantics); 3004 } 3005 3006 /// parseDirectiveOrg 3007 /// ::= .org expression [ , expression ] 3008 bool MasmParser::parseDirectiveOrg() { 3009 const MCExpr *Offset; 3010 SMLoc OffsetLoc = Lexer.getLoc(); 3011 if (checkForValidSection() || parseExpression(Offset)) 3012 return true; 3013 3014 // Parse optional fill expression. 3015 int64_t FillExpr = 0; 3016 if (parseOptionalToken(AsmToken::Comma)) 3017 if (parseAbsoluteExpression(FillExpr)) 3018 return addErrorSuffix(" in '.org' directive"); 3019 if (parseToken(AsmToken::EndOfStatement)) 3020 return addErrorSuffix(" in '.org' directive"); 3021 3022 getStreamer().emitValueToOffset(Offset, FillExpr, OffsetLoc); 3023 return false; 3024 } 3025 3026 /// parseDirectiveAlign 3027 /// ::= align expression 3028 bool MasmParser::parseDirectiveAlign() { 3029 SMLoc AlignmentLoc = getLexer().getLoc(); 3030 int64_t Alignment; 3031 3032 if (checkForValidSection()) 3033 return addErrorSuffix(" in align directive"); 3034 // Ignore empty 'align' directives 3035 if (getTok().is(AsmToken::EndOfStatement)) { 3036 Warning(AlignmentLoc, "align directive with no operand is ignored"); 3037 return parseToken(AsmToken::EndOfStatement); 3038 } 3039 if (parseAbsoluteExpression(Alignment) || 3040 parseToken(AsmToken::EndOfStatement)) 3041 return addErrorSuffix(" in align directive"); 3042 3043 // Always emit an alignment here even if we thrown an error. 3044 bool ReturnVal = false; 3045 3046 // Reject alignments that aren't either a power of two or zero, 3047 // for gas compatibility. Alignment of zero is silently rounded 3048 // up to one. 3049 if (Alignment == 0) 3050 Alignment = 1; 3051 if (!isPowerOf2_64(Alignment)) 3052 ReturnVal |= Error(AlignmentLoc, "alignment must be a power of 2"); 3053 3054 // Check whether we should use optimal code alignment for this align 3055 // directive. 3056 const MCSection *Section = getStreamer().getCurrentSectionOnly(); 3057 assert(Section && "must have section to emit alignment"); 3058 if (Section->UseCodeAlign()) { 3059 getStreamer().emitCodeAlignment(Alignment, /*MaxBytesToEmit=*/0); 3060 } else { 3061 // FIXME: Target specific behavior about how the "extra" bytes are filled. 3062 getStreamer().emitValueToAlignment(Alignment, /*Value=*/0, /*ValueSize=*/1, 3063 /*MaxBytesToEmit=*/0); 3064 } 3065 3066 return ReturnVal; 3067 } 3068 3069 /// parseDirectiveFile 3070 /// ::= .file filename 3071 /// ::= .file number [directory] filename [md5 checksum] [source source-text] 3072 bool MasmParser::parseDirectiveFile(SMLoc DirectiveLoc) { 3073 // FIXME: I'm not sure what this is. 3074 int64_t FileNumber = -1; 3075 if (getLexer().is(AsmToken::Integer)) { 3076 FileNumber = getTok().getIntVal(); 3077 Lex(); 3078 3079 if (FileNumber < 0) 3080 return TokError("negative file number"); 3081 } 3082 3083 std::string Path; 3084 3085 // Usually the directory and filename together, otherwise just the directory. 3086 // Allow the strings to have escaped octal character sequence. 3087 if (check(getTok().isNot(AsmToken::String), 3088 "unexpected token in '.file' directive") || 3089 parseEscapedString(Path)) 3090 return true; 3091 3092 StringRef Directory; 3093 StringRef Filename; 3094 std::string FilenameData; 3095 if (getLexer().is(AsmToken::String)) { 3096 if (check(FileNumber == -1, 3097 "explicit path specified, but no file number") || 3098 parseEscapedString(FilenameData)) 3099 return true; 3100 Filename = FilenameData; 3101 Directory = Path; 3102 } else { 3103 Filename = Path; 3104 } 3105 3106 uint64_t MD5Hi, MD5Lo; 3107 bool HasMD5 = false; 3108 3109 Optional<StringRef> Source; 3110 bool HasSource = false; 3111 std::string SourceString; 3112 3113 while (!parseOptionalToken(AsmToken::EndOfStatement)) { 3114 StringRef Keyword; 3115 if (check(getTok().isNot(AsmToken::Identifier), 3116 "unexpected token in '.file' directive") || 3117 parseIdentifier(Keyword)) 3118 return true; 3119 if (Keyword == "md5") { 3120 HasMD5 = true; 3121 if (check(FileNumber == -1, 3122 "MD5 checksum specified, but no file number") || 3123 parseHexOcta(*this, MD5Hi, MD5Lo)) 3124 return true; 3125 } else if (Keyword == "source") { 3126 HasSource = true; 3127 if (check(FileNumber == -1, 3128 "source specified, but no file number") || 3129 check(getTok().isNot(AsmToken::String), 3130 "unexpected token in '.file' directive") || 3131 parseEscapedString(SourceString)) 3132 return true; 3133 } else { 3134 return TokError("unexpected token in '.file' directive"); 3135 } 3136 } 3137 3138 if (FileNumber == -1) { 3139 // Ignore the directive if there is no number and the target doesn't support 3140 // numberless .file directives. This allows some portability of assembler 3141 // between different object file formats. 3142 if (getContext().getAsmInfo()->hasSingleParameterDotFile()) 3143 getStreamer().emitFileDirective(Filename); 3144 } else { 3145 // In case there is a -g option as well as debug info from directive .file, 3146 // we turn off the -g option, directly use the existing debug info instead. 3147 // Throw away any implicit file table for the assembler source. 3148 if (Ctx.getGenDwarfForAssembly()) { 3149 Ctx.getMCDwarfLineTable(0).resetFileTable(); 3150 Ctx.setGenDwarfForAssembly(false); 3151 } 3152 3153 Optional<MD5::MD5Result> CKMem; 3154 if (HasMD5) { 3155 MD5::MD5Result Sum; 3156 for (unsigned i = 0; i != 8; ++i) { 3157 Sum.Bytes[i] = uint8_t(MD5Hi >> ((7 - i) * 8)); 3158 Sum.Bytes[i + 8] = uint8_t(MD5Lo >> ((7 - i) * 8)); 3159 } 3160 CKMem = Sum; 3161 } 3162 if (HasSource) { 3163 char *SourceBuf = static_cast<char *>(Ctx.allocate(SourceString.size())); 3164 memcpy(SourceBuf, SourceString.data(), SourceString.size()); 3165 Source = StringRef(SourceBuf, SourceString.size()); 3166 } 3167 if (FileNumber == 0) { 3168 if (Ctx.getDwarfVersion() < 5) 3169 return Warning(DirectiveLoc, "file 0 not supported prior to DWARF-5"); 3170 getStreamer().emitDwarfFile0Directive(Directory, Filename, CKMem, Source); 3171 } else { 3172 Expected<unsigned> FileNumOrErr = getStreamer().tryEmitDwarfFileDirective( 3173 FileNumber, Directory, Filename, CKMem, Source); 3174 if (!FileNumOrErr) 3175 return Error(DirectiveLoc, toString(FileNumOrErr.takeError())); 3176 } 3177 // Alert the user if there are some .file directives with MD5 and some not. 3178 // But only do that once. 3179 if (!ReportedInconsistentMD5 && !Ctx.isDwarfMD5UsageConsistent(0)) { 3180 ReportedInconsistentMD5 = true; 3181 return Warning(DirectiveLoc, "inconsistent use of MD5 checksums"); 3182 } 3183 } 3184 3185 return false; 3186 } 3187 3188 /// parseDirectiveLine 3189 /// ::= .line [number] 3190 bool MasmParser::parseDirectiveLine() { 3191 int64_t LineNumber; 3192 if (getLexer().is(AsmToken::Integer)) { 3193 if (parseIntToken(LineNumber, "unexpected token in '.line' directive")) 3194 return true; 3195 (void)LineNumber; 3196 // FIXME: Do something with the .line. 3197 } 3198 if (parseToken(AsmToken::EndOfStatement, 3199 "unexpected token in '.line' directive")) 3200 return true; 3201 3202 return false; 3203 } 3204 3205 /// parseDirectiveLoc 3206 /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] 3207 /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] 3208 /// The first number is a file number, must have been previously assigned with 3209 /// a .file directive, the second number is the line number and optionally the 3210 /// third number is a column position (zero if not specified). The remaining 3211 /// optional items are .loc sub-directives. 3212 bool MasmParser::parseDirectiveLoc() { 3213 int64_t FileNumber = 0, LineNumber = 0; 3214 SMLoc Loc = getTok().getLoc(); 3215 if (parseIntToken(FileNumber, "unexpected token in '.loc' directive") || 3216 check(FileNumber < 1 && Ctx.getDwarfVersion() < 5, Loc, 3217 "file number less than one in '.loc' directive") || 3218 check(!getContext().isValidDwarfFileNumber(FileNumber), Loc, 3219 "unassigned file number in '.loc' directive")) 3220 return true; 3221 3222 // optional 3223 if (getLexer().is(AsmToken::Integer)) { 3224 LineNumber = getTok().getIntVal(); 3225 if (LineNumber < 0) 3226 return TokError("line number less than zero in '.loc' directive"); 3227 Lex(); 3228 } 3229 3230 int64_t ColumnPos = 0; 3231 if (getLexer().is(AsmToken::Integer)) { 3232 ColumnPos = getTok().getIntVal(); 3233 if (ColumnPos < 0) 3234 return TokError("column position less than zero in '.loc' directive"); 3235 Lex(); 3236 } 3237 3238 unsigned Flags = DWARF2_LINE_DEFAULT_IS_STMT ? DWARF2_FLAG_IS_STMT : 0; 3239 unsigned Isa = 0; 3240 int64_t Discriminator = 0; 3241 3242 auto parseLocOp = [&]() -> bool { 3243 StringRef Name; 3244 SMLoc Loc = getTok().getLoc(); 3245 if (parseIdentifier(Name)) 3246 return TokError("unexpected token in '.loc' directive"); 3247 3248 if (Name == "basic_block") 3249 Flags |= DWARF2_FLAG_BASIC_BLOCK; 3250 else if (Name == "prologue_end") 3251 Flags |= DWARF2_FLAG_PROLOGUE_END; 3252 else if (Name == "epilogue_begin") 3253 Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; 3254 else if (Name == "is_stmt") { 3255 Loc = getTok().getLoc(); 3256 const MCExpr *Value; 3257 if (parseExpression(Value)) 3258 return true; 3259 // The expression must be the constant 0 or 1. 3260 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3261 int Value = MCE->getValue(); 3262 if (Value == 0) 3263 Flags &= ~DWARF2_FLAG_IS_STMT; 3264 else if (Value == 1) 3265 Flags |= DWARF2_FLAG_IS_STMT; 3266 else 3267 return Error(Loc, "is_stmt value not 0 or 1"); 3268 } else { 3269 return Error(Loc, "is_stmt value not the constant value of 0 or 1"); 3270 } 3271 } else if (Name == "isa") { 3272 Loc = getTok().getLoc(); 3273 const MCExpr *Value; 3274 if (parseExpression(Value)) 3275 return true; 3276 // The expression must be a constant greater or equal to 0. 3277 if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { 3278 int Value = MCE->getValue(); 3279 if (Value < 0) 3280 return Error(Loc, "isa number less than zero"); 3281 Isa = Value; 3282 } else { 3283 return Error(Loc, "isa number not a constant value"); 3284 } 3285 } else if (Name == "discriminator") { 3286 if (parseAbsoluteExpression(Discriminator)) 3287 return true; 3288 } else { 3289 return Error(Loc, "unknown sub-directive in '.loc' directive"); 3290 } 3291 return false; 3292 }; 3293 3294 if (parseMany(parseLocOp, false /*hasComma*/)) 3295 return true; 3296 3297 getStreamer().emitDwarfLocDirective(FileNumber, LineNumber, ColumnPos, Flags, 3298 Isa, Discriminator, StringRef()); 3299 3300 return false; 3301 } 3302 3303 /// parseDirectiveStabs 3304 /// ::= .stabs string, number, number, number 3305 bool MasmParser::parseDirectiveStabs() { 3306 return TokError("unsupported directive '.stabs'"); 3307 } 3308 3309 /// parseDirectiveCVFile 3310 /// ::= .cv_file number filename [checksum] [checksumkind] 3311 bool MasmParser::parseDirectiveCVFile() { 3312 SMLoc FileNumberLoc = getTok().getLoc(); 3313 int64_t FileNumber; 3314 std::string Filename; 3315 std::string Checksum; 3316 int64_t ChecksumKind = 0; 3317 3318 if (parseIntToken(FileNumber, 3319 "expected file number in '.cv_file' directive") || 3320 check(FileNumber < 1, FileNumberLoc, "file number less than one") || 3321 check(getTok().isNot(AsmToken::String), 3322 "unexpected token in '.cv_file' directive") || 3323 parseEscapedString(Filename)) 3324 return true; 3325 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3326 if (check(getTok().isNot(AsmToken::String), 3327 "unexpected token in '.cv_file' directive") || 3328 parseEscapedString(Checksum) || 3329 parseIntToken(ChecksumKind, 3330 "expected checksum kind in '.cv_file' directive") || 3331 parseToken(AsmToken::EndOfStatement, 3332 "unexpected token in '.cv_file' directive")) 3333 return true; 3334 } 3335 3336 Checksum = fromHex(Checksum); 3337 void *CKMem = Ctx.allocate(Checksum.size(), 1); 3338 memcpy(CKMem, Checksum.data(), Checksum.size()); 3339 ArrayRef<uint8_t> ChecksumAsBytes(reinterpret_cast<const uint8_t *>(CKMem), 3340 Checksum.size()); 3341 3342 if (!getStreamer().EmitCVFileDirective(FileNumber, Filename, ChecksumAsBytes, 3343 static_cast<uint8_t>(ChecksumKind))) 3344 return Error(FileNumberLoc, "file number already allocated"); 3345 3346 return false; 3347 } 3348 3349 bool MasmParser::parseCVFunctionId(int64_t &FunctionId, 3350 StringRef DirectiveName) { 3351 SMLoc Loc; 3352 return parseTokenLoc(Loc) || 3353 parseIntToken(FunctionId, "expected function id in '" + DirectiveName + 3354 "' directive") || 3355 check(FunctionId < 0 || FunctionId >= UINT_MAX, Loc, 3356 "expected function id within range [0, UINT_MAX)"); 3357 } 3358 3359 bool MasmParser::parseCVFileId(int64_t &FileNumber, StringRef DirectiveName) { 3360 SMLoc Loc; 3361 return parseTokenLoc(Loc) || 3362 parseIntToken(FileNumber, "expected integer in '" + DirectiveName + 3363 "' directive") || 3364 check(FileNumber < 1, Loc, "file number less than one in '" + 3365 DirectiveName + "' directive") || 3366 check(!getCVContext().isValidFileNumber(FileNumber), Loc, 3367 "unassigned file number in '" + DirectiveName + "' directive"); 3368 } 3369 3370 /// parseDirectiveCVFuncId 3371 /// ::= .cv_func_id FunctionId 3372 /// 3373 /// Introduces a function ID that can be used with .cv_loc. 3374 bool MasmParser::parseDirectiveCVFuncId() { 3375 SMLoc FunctionIdLoc = getTok().getLoc(); 3376 int64_t FunctionId; 3377 3378 if (parseCVFunctionId(FunctionId, ".cv_func_id") || 3379 parseToken(AsmToken::EndOfStatement, 3380 "unexpected token in '.cv_func_id' directive")) 3381 return true; 3382 3383 if (!getStreamer().EmitCVFuncIdDirective(FunctionId)) 3384 return Error(FunctionIdLoc, "function id already allocated"); 3385 3386 return false; 3387 } 3388 3389 /// parseDirectiveCVInlineSiteId 3390 /// ::= .cv_inline_site_id FunctionId 3391 /// "within" IAFunc 3392 /// "inlined_at" IAFile IALine [IACol] 3393 /// 3394 /// Introduces a function ID that can be used with .cv_loc. Includes "inlined 3395 /// at" source location information for use in the line table of the caller, 3396 /// whether the caller is a real function or another inlined call site. 3397 bool MasmParser::parseDirectiveCVInlineSiteId() { 3398 SMLoc FunctionIdLoc = getTok().getLoc(); 3399 int64_t FunctionId; 3400 int64_t IAFunc; 3401 int64_t IAFile; 3402 int64_t IALine; 3403 int64_t IACol = 0; 3404 3405 // FunctionId 3406 if (parseCVFunctionId(FunctionId, ".cv_inline_site_id")) 3407 return true; 3408 3409 // "within" 3410 if (check((getLexer().isNot(AsmToken::Identifier) || 3411 getTok().getIdentifier() != "within"), 3412 "expected 'within' identifier in '.cv_inline_site_id' directive")) 3413 return true; 3414 Lex(); 3415 3416 // IAFunc 3417 if (parseCVFunctionId(IAFunc, ".cv_inline_site_id")) 3418 return true; 3419 3420 // "inlined_at" 3421 if (check((getLexer().isNot(AsmToken::Identifier) || 3422 getTok().getIdentifier() != "inlined_at"), 3423 "expected 'inlined_at' identifier in '.cv_inline_site_id' " 3424 "directive") ) 3425 return true; 3426 Lex(); 3427 3428 // IAFile IALine 3429 if (parseCVFileId(IAFile, ".cv_inline_site_id") || 3430 parseIntToken(IALine, "expected line number after 'inlined_at'")) 3431 return true; 3432 3433 // [IACol] 3434 if (getLexer().is(AsmToken::Integer)) { 3435 IACol = getTok().getIntVal(); 3436 Lex(); 3437 } 3438 3439 if (parseToken(AsmToken::EndOfStatement, 3440 "unexpected token in '.cv_inline_site_id' directive")) 3441 return true; 3442 3443 if (!getStreamer().EmitCVInlineSiteIdDirective(FunctionId, IAFunc, IAFile, 3444 IALine, IACol, FunctionIdLoc)) 3445 return Error(FunctionIdLoc, "function id already allocated"); 3446 3447 return false; 3448 } 3449 3450 /// parseDirectiveCVLoc 3451 /// ::= .cv_loc FunctionId FileNumber [LineNumber] [ColumnPos] [prologue_end] 3452 /// [is_stmt VALUE] 3453 /// The first number is a file number, must have been previously assigned with 3454 /// a .file directive, the second number is the line number and optionally the 3455 /// third number is a column position (zero if not specified). The remaining 3456 /// optional items are .loc sub-directives. 3457 bool MasmParser::parseDirectiveCVLoc() { 3458 SMLoc DirectiveLoc = getTok().getLoc(); 3459 int64_t FunctionId, FileNumber; 3460 if (parseCVFunctionId(FunctionId, ".cv_loc") || 3461 parseCVFileId(FileNumber, ".cv_loc")) 3462 return true; 3463 3464 int64_t LineNumber = 0; 3465 if (getLexer().is(AsmToken::Integer)) { 3466 LineNumber = getTok().getIntVal(); 3467 if (LineNumber < 0) 3468 return TokError("line number less than zero in '.cv_loc' directive"); 3469 Lex(); 3470 } 3471 3472 int64_t ColumnPos = 0; 3473 if (getLexer().is(AsmToken::Integer)) { 3474 ColumnPos = getTok().getIntVal(); 3475 if (ColumnPos < 0) 3476 return TokError("column position less than zero in '.cv_loc' directive"); 3477 Lex(); 3478 } 3479 3480 bool PrologueEnd = false; 3481 uint64_t IsStmt = 0; 3482 3483 auto parseOp = [&]() -> bool { 3484 StringRef Name; 3485 SMLoc Loc = getTok().getLoc(); 3486 if (parseIdentifier(Name)) 3487 return TokError("unexpected token in '.cv_loc' directive"); 3488 if (Name == "prologue_end") 3489 PrologueEnd = true; 3490 else if (Name == "is_stmt") { 3491 Loc = getTok().getLoc(); 3492 const MCExpr *Value; 3493 if (parseExpression(Value)) 3494 return true; 3495 // The expression must be the constant 0 or 1. 3496 IsStmt = ~0ULL; 3497 if (const auto *MCE = dyn_cast<MCConstantExpr>(Value)) 3498 IsStmt = MCE->getValue(); 3499 3500 if (IsStmt > 1) 3501 return Error(Loc, "is_stmt value not 0 or 1"); 3502 } else { 3503 return Error(Loc, "unknown sub-directive in '.cv_loc' directive"); 3504 } 3505 return false; 3506 }; 3507 3508 if (parseMany(parseOp, false /*hasComma*/)) 3509 return true; 3510 3511 getStreamer().EmitCVLocDirective(FunctionId, FileNumber, LineNumber, 3512 ColumnPos, PrologueEnd, IsStmt, StringRef(), 3513 DirectiveLoc); 3514 return false; 3515 } 3516 3517 /// parseDirectiveCVLinetable 3518 /// ::= .cv_linetable FunctionId, FnStart, FnEnd 3519 bool MasmParser::parseDirectiveCVLinetable() { 3520 int64_t FunctionId; 3521 StringRef FnStartName, FnEndName; 3522 SMLoc Loc = getTok().getLoc(); 3523 if (parseCVFunctionId(FunctionId, ".cv_linetable") || 3524 parseToken(AsmToken::Comma, 3525 "unexpected token in '.cv_linetable' directive") || 3526 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3527 "expected identifier in directive") || 3528 parseToken(AsmToken::Comma, 3529 "unexpected token in '.cv_linetable' directive") || 3530 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3531 "expected identifier in directive")) 3532 return true; 3533 3534 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3535 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3536 3537 getStreamer().EmitCVLinetableDirective(FunctionId, FnStartSym, FnEndSym); 3538 return false; 3539 } 3540 3541 /// parseDirectiveCVInlineLinetable 3542 /// ::= .cv_inline_linetable PrimaryFunctionId FileId LineNum FnStart FnEnd 3543 bool MasmParser::parseDirectiveCVInlineLinetable() { 3544 int64_t PrimaryFunctionId, SourceFileId, SourceLineNum; 3545 StringRef FnStartName, FnEndName; 3546 SMLoc Loc = getTok().getLoc(); 3547 if (parseCVFunctionId(PrimaryFunctionId, ".cv_inline_linetable") || 3548 parseTokenLoc(Loc) || 3549 parseIntToken( 3550 SourceFileId, 3551 "expected SourceField in '.cv_inline_linetable' directive") || 3552 check(SourceFileId <= 0, Loc, 3553 "File id less than zero in '.cv_inline_linetable' directive") || 3554 parseTokenLoc(Loc) || 3555 parseIntToken( 3556 SourceLineNum, 3557 "expected SourceLineNum in '.cv_inline_linetable' directive") || 3558 check(SourceLineNum < 0, Loc, 3559 "Line number less than zero in '.cv_inline_linetable' directive") || 3560 parseTokenLoc(Loc) || check(parseIdentifier(FnStartName), Loc, 3561 "expected identifier in directive") || 3562 parseTokenLoc(Loc) || check(parseIdentifier(FnEndName), Loc, 3563 "expected identifier in directive")) 3564 return true; 3565 3566 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3567 return true; 3568 3569 MCSymbol *FnStartSym = getContext().getOrCreateSymbol(FnStartName); 3570 MCSymbol *FnEndSym = getContext().getOrCreateSymbol(FnEndName); 3571 getStreamer().EmitCVInlineLinetableDirective(PrimaryFunctionId, SourceFileId, 3572 SourceLineNum, FnStartSym, 3573 FnEndSym); 3574 return false; 3575 } 3576 3577 void MasmParser::initializeCVDefRangeTypeMap() { 3578 CVDefRangeTypeMap["reg"] = CVDR_DEFRANGE_REGISTER; 3579 CVDefRangeTypeMap["frame_ptr_rel"] = CVDR_DEFRANGE_FRAMEPOINTER_REL; 3580 CVDefRangeTypeMap["subfield_reg"] = CVDR_DEFRANGE_SUBFIELD_REGISTER; 3581 CVDefRangeTypeMap["reg_rel"] = CVDR_DEFRANGE_REGISTER_REL; 3582 } 3583 3584 /// parseDirectiveCVDefRange 3585 /// ::= .cv_def_range RangeStart RangeEnd (GapStart GapEnd)*, bytes* 3586 bool MasmParser::parseDirectiveCVDefRange() { 3587 SMLoc Loc; 3588 std::vector<std::pair<const MCSymbol *, const MCSymbol *>> Ranges; 3589 while (getLexer().is(AsmToken::Identifier)) { 3590 Loc = getLexer().getLoc(); 3591 StringRef GapStartName; 3592 if (parseIdentifier(GapStartName)) 3593 return Error(Loc, "expected identifier in directive"); 3594 MCSymbol *GapStartSym = getContext().getOrCreateSymbol(GapStartName); 3595 3596 Loc = getLexer().getLoc(); 3597 StringRef GapEndName; 3598 if (parseIdentifier(GapEndName)) 3599 return Error(Loc, "expected identifier in directive"); 3600 MCSymbol *GapEndSym = getContext().getOrCreateSymbol(GapEndName); 3601 3602 Ranges.push_back({GapStartSym, GapEndSym}); 3603 } 3604 3605 StringRef CVDefRangeTypeStr; 3606 if (parseToken( 3607 AsmToken::Comma, 3608 "expected comma before def_range type in .cv_def_range directive") || 3609 parseIdentifier(CVDefRangeTypeStr)) 3610 return Error(Loc, "expected def_range type in directive"); 3611 3612 StringMap<CVDefRangeType>::const_iterator CVTypeIt = 3613 CVDefRangeTypeMap.find(CVDefRangeTypeStr); 3614 CVDefRangeType CVDRType = (CVTypeIt == CVDefRangeTypeMap.end()) 3615 ? CVDR_DEFRANGE 3616 : CVTypeIt->getValue(); 3617 switch (CVDRType) { 3618 case CVDR_DEFRANGE_REGISTER: { 3619 int64_t DRRegister; 3620 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3621 ".cv_def_range directive") || 3622 parseAbsoluteExpression(DRRegister)) 3623 return Error(Loc, "expected register number"); 3624 3625 codeview::DefRangeRegisterHeader DRHdr; 3626 DRHdr.Register = DRRegister; 3627 DRHdr.MayHaveNoName = 0; 3628 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3629 break; 3630 } 3631 case CVDR_DEFRANGE_FRAMEPOINTER_REL: { 3632 int64_t DROffset; 3633 if (parseToken(AsmToken::Comma, 3634 "expected comma before offset in .cv_def_range directive") || 3635 parseAbsoluteExpression(DROffset)) 3636 return Error(Loc, "expected offset value"); 3637 3638 codeview::DefRangeFramePointerRelHeader DRHdr; 3639 DRHdr.Offset = DROffset; 3640 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3641 break; 3642 } 3643 case CVDR_DEFRANGE_SUBFIELD_REGISTER: { 3644 int64_t DRRegister; 3645 int64_t DROffsetInParent; 3646 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3647 ".cv_def_range directive") || 3648 parseAbsoluteExpression(DRRegister)) 3649 return Error(Loc, "expected register number"); 3650 if (parseToken(AsmToken::Comma, 3651 "expected comma before offset in .cv_def_range directive") || 3652 parseAbsoluteExpression(DROffsetInParent)) 3653 return Error(Loc, "expected offset value"); 3654 3655 codeview::DefRangeSubfieldRegisterHeader DRHdr; 3656 DRHdr.Register = DRRegister; 3657 DRHdr.MayHaveNoName = 0; 3658 DRHdr.OffsetInParent = DROffsetInParent; 3659 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3660 break; 3661 } 3662 case CVDR_DEFRANGE_REGISTER_REL: { 3663 int64_t DRRegister; 3664 int64_t DRFlags; 3665 int64_t DRBasePointerOffset; 3666 if (parseToken(AsmToken::Comma, "expected comma before register number in " 3667 ".cv_def_range directive") || 3668 parseAbsoluteExpression(DRRegister)) 3669 return Error(Loc, "expected register value"); 3670 if (parseToken( 3671 AsmToken::Comma, 3672 "expected comma before flag value in .cv_def_range directive") || 3673 parseAbsoluteExpression(DRFlags)) 3674 return Error(Loc, "expected flag value"); 3675 if (parseToken(AsmToken::Comma, "expected comma before base pointer offset " 3676 "in .cv_def_range directive") || 3677 parseAbsoluteExpression(DRBasePointerOffset)) 3678 return Error(Loc, "expected base pointer offset value"); 3679 3680 codeview::DefRangeRegisterRelHeader DRHdr; 3681 DRHdr.Register = DRRegister; 3682 DRHdr.Flags = DRFlags; 3683 DRHdr.BasePointerOffset = DRBasePointerOffset; 3684 getStreamer().EmitCVDefRangeDirective(Ranges, DRHdr); 3685 break; 3686 } 3687 default: 3688 return Error(Loc, "unexpected def_range type in .cv_def_range directive"); 3689 } 3690 return true; 3691 } 3692 3693 /// parseDirectiveCVString 3694 /// ::= .cv_stringtable "string" 3695 bool MasmParser::parseDirectiveCVString() { 3696 std::string Data; 3697 if (checkForValidSection() || parseEscapedString(Data)) 3698 return addErrorSuffix(" in '.cv_string' directive"); 3699 3700 // Put the string in the table and emit the offset. 3701 std::pair<StringRef, unsigned> Insertion = 3702 getCVContext().addToStringTable(Data); 3703 getStreamer().emitIntValue(Insertion.second, 4); 3704 return false; 3705 } 3706 3707 /// parseDirectiveCVStringTable 3708 /// ::= .cv_stringtable 3709 bool MasmParser::parseDirectiveCVStringTable() { 3710 getStreamer().EmitCVStringTableDirective(); 3711 return false; 3712 } 3713 3714 /// parseDirectiveCVFileChecksums 3715 /// ::= .cv_filechecksums 3716 bool MasmParser::parseDirectiveCVFileChecksums() { 3717 getStreamer().EmitCVFileChecksumsDirective(); 3718 return false; 3719 } 3720 3721 /// parseDirectiveCVFileChecksumOffset 3722 /// ::= .cv_filechecksumoffset fileno 3723 bool MasmParser::parseDirectiveCVFileChecksumOffset() { 3724 int64_t FileNo; 3725 if (parseIntToken(FileNo, "expected identifier in directive")) 3726 return true; 3727 if (parseToken(AsmToken::EndOfStatement, "Expected End of Statement")) 3728 return true; 3729 getStreamer().EmitCVFileChecksumOffsetDirective(FileNo); 3730 return false; 3731 } 3732 3733 /// parseDirectiveCVFPOData 3734 /// ::= .cv_fpo_data procsym 3735 bool MasmParser::parseDirectiveCVFPOData() { 3736 SMLoc DirLoc = getLexer().getLoc(); 3737 StringRef ProcName; 3738 if (parseIdentifier(ProcName)) 3739 return TokError("expected symbol name"); 3740 if (parseEOL("unexpected tokens")) 3741 return addErrorSuffix(" in '.cv_fpo_data' directive"); 3742 MCSymbol *ProcSym = getContext().getOrCreateSymbol(ProcName); 3743 getStreamer().EmitCVFPOData(ProcSym, DirLoc); 3744 return false; 3745 } 3746 3747 /// parseDirectiveCFISections 3748 /// ::= .cfi_sections section [, section] 3749 bool MasmParser::parseDirectiveCFISections() { 3750 StringRef Name; 3751 bool EH = false; 3752 bool Debug = false; 3753 3754 if (parseIdentifier(Name)) 3755 return TokError("Expected an identifier"); 3756 3757 if (Name == ".eh_frame") 3758 EH = true; 3759 else if (Name == ".debug_frame") 3760 Debug = true; 3761 3762 if (getLexer().is(AsmToken::Comma)) { 3763 Lex(); 3764 3765 if (parseIdentifier(Name)) 3766 return TokError("Expected an identifier"); 3767 3768 if (Name == ".eh_frame") 3769 EH = true; 3770 else if (Name == ".debug_frame") 3771 Debug = true; 3772 } 3773 3774 getStreamer().emitCFISections(EH, Debug); 3775 return false; 3776 } 3777 3778 /// parseDirectiveCFIStartProc 3779 /// ::= .cfi_startproc [simple] 3780 bool MasmParser::parseDirectiveCFIStartProc() { 3781 StringRef Simple; 3782 if (!parseOptionalToken(AsmToken::EndOfStatement)) { 3783 if (check(parseIdentifier(Simple) || Simple != "simple", 3784 "unexpected token") || 3785 parseToken(AsmToken::EndOfStatement)) 3786 return addErrorSuffix(" in '.cfi_startproc' directive"); 3787 } 3788 3789 // TODO(kristina): Deal with a corner case of incorrect diagnostic context 3790 // being produced if this directive is emitted as part of preprocessor macro 3791 // expansion which can *ONLY* happen if Clang's cc1as is the API consumer. 3792 // Tools like llvm-mc on the other hand are not affected by it, and report 3793 // correct context information. 3794 getStreamer().emitCFIStartProc(!Simple.empty(), Lexer.getLoc()); 3795 return false; 3796 } 3797 3798 /// parseDirectiveCFIEndProc 3799 /// ::= .cfi_endproc 3800 bool MasmParser::parseDirectiveCFIEndProc() { 3801 getStreamer().emitCFIEndProc(); 3802 return false; 3803 } 3804 3805 /// parse register name or number. 3806 bool MasmParser::parseRegisterOrRegisterNumber(int64_t &Register, 3807 SMLoc DirectiveLoc) { 3808 unsigned RegNo; 3809 3810 if (getLexer().isNot(AsmToken::Integer)) { 3811 if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) 3812 return true; 3813 Register = getContext().getRegisterInfo()->getDwarfRegNum(RegNo, true); 3814 } else 3815 return parseAbsoluteExpression(Register); 3816 3817 return false; 3818 } 3819 3820 /// parseDirectiveCFIDefCfa 3821 /// ::= .cfi_def_cfa register, offset 3822 bool MasmParser::parseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { 3823 int64_t Register = 0, Offset = 0; 3824 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3825 parseToken(AsmToken::Comma, "unexpected token in directive") || 3826 parseAbsoluteExpression(Offset)) 3827 return true; 3828 3829 getStreamer().emitCFIDefCfa(Register, Offset); 3830 return false; 3831 } 3832 3833 /// parseDirectiveCFIDefCfaOffset 3834 /// ::= .cfi_def_cfa_offset offset 3835 bool MasmParser::parseDirectiveCFIDefCfaOffset() { 3836 int64_t Offset = 0; 3837 if (parseAbsoluteExpression(Offset)) 3838 return true; 3839 3840 getStreamer().emitCFIDefCfaOffset(Offset); 3841 return false; 3842 } 3843 3844 /// parseDirectiveCFIRegister 3845 /// ::= .cfi_register register, register 3846 bool MasmParser::parseDirectiveCFIRegister(SMLoc DirectiveLoc) { 3847 int64_t Register1 = 0, Register2 = 0; 3848 if (parseRegisterOrRegisterNumber(Register1, DirectiveLoc) || 3849 parseToken(AsmToken::Comma, "unexpected token in directive") || 3850 parseRegisterOrRegisterNumber(Register2, DirectiveLoc)) 3851 return true; 3852 3853 getStreamer().emitCFIRegister(Register1, Register2); 3854 return false; 3855 } 3856 3857 /// parseDirectiveCFIWindowSave 3858 /// ::= .cfi_window_save 3859 bool MasmParser::parseDirectiveCFIWindowSave() { 3860 getStreamer().emitCFIWindowSave(); 3861 return false; 3862 } 3863 3864 /// parseDirectiveCFIAdjustCfaOffset 3865 /// ::= .cfi_adjust_cfa_offset adjustment 3866 bool MasmParser::parseDirectiveCFIAdjustCfaOffset() { 3867 int64_t Adjustment = 0; 3868 if (parseAbsoluteExpression(Adjustment)) 3869 return true; 3870 3871 getStreamer().emitCFIAdjustCfaOffset(Adjustment); 3872 return false; 3873 } 3874 3875 /// parseDirectiveCFIDefCfaRegister 3876 /// ::= .cfi_def_cfa_register register 3877 bool MasmParser::parseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { 3878 int64_t Register = 0; 3879 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3880 return true; 3881 3882 getStreamer().emitCFIDefCfaRegister(Register); 3883 return false; 3884 } 3885 3886 /// parseDirectiveCFIOffset 3887 /// ::= .cfi_offset register, offset 3888 bool MasmParser::parseDirectiveCFIOffset(SMLoc DirectiveLoc) { 3889 int64_t Register = 0; 3890 int64_t Offset = 0; 3891 3892 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3893 parseToken(AsmToken::Comma, "unexpected token in directive") || 3894 parseAbsoluteExpression(Offset)) 3895 return true; 3896 3897 getStreamer().emitCFIOffset(Register, Offset); 3898 return false; 3899 } 3900 3901 /// parseDirectiveCFIRelOffset 3902 /// ::= .cfi_rel_offset register, offset 3903 bool MasmParser::parseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { 3904 int64_t Register = 0, Offset = 0; 3905 3906 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc) || 3907 parseToken(AsmToken::Comma, "unexpected token in directive") || 3908 parseAbsoluteExpression(Offset)) 3909 return true; 3910 3911 getStreamer().emitCFIRelOffset(Register, Offset); 3912 return false; 3913 } 3914 3915 static bool isValidEncoding(int64_t Encoding) { 3916 if (Encoding & ~0xff) 3917 return false; 3918 3919 if (Encoding == dwarf::DW_EH_PE_omit) 3920 return true; 3921 3922 const unsigned Format = Encoding & 0xf; 3923 if (Format != dwarf::DW_EH_PE_absptr && Format != dwarf::DW_EH_PE_udata2 && 3924 Format != dwarf::DW_EH_PE_udata4 && Format != dwarf::DW_EH_PE_udata8 && 3925 Format != dwarf::DW_EH_PE_sdata2 && Format != dwarf::DW_EH_PE_sdata4 && 3926 Format != dwarf::DW_EH_PE_sdata8 && Format != dwarf::DW_EH_PE_signed) 3927 return false; 3928 3929 const unsigned Application = Encoding & 0x70; 3930 if (Application != dwarf::DW_EH_PE_absptr && 3931 Application != dwarf::DW_EH_PE_pcrel) 3932 return false; 3933 3934 return true; 3935 } 3936 3937 /// parseDirectiveCFIPersonalityOrLsda 3938 /// IsPersonality true for cfi_personality, false for cfi_lsda 3939 /// ::= .cfi_personality encoding, [symbol_name] 3940 /// ::= .cfi_lsda encoding, [symbol_name] 3941 bool MasmParser::parseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { 3942 int64_t Encoding = 0; 3943 if (parseAbsoluteExpression(Encoding)) 3944 return true; 3945 if (Encoding == dwarf::DW_EH_PE_omit) 3946 return false; 3947 3948 StringRef Name; 3949 if (check(!isValidEncoding(Encoding), "unsupported encoding.") || 3950 parseToken(AsmToken::Comma, "unexpected token in directive") || 3951 check(parseIdentifier(Name), "expected identifier in directive")) 3952 return true; 3953 3954 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 3955 3956 if (IsPersonality) 3957 getStreamer().emitCFIPersonality(Sym, Encoding); 3958 else 3959 getStreamer().emitCFILsda(Sym, Encoding); 3960 return false; 3961 } 3962 3963 /// parseDirectiveCFIRememberState 3964 /// ::= .cfi_remember_state 3965 bool MasmParser::parseDirectiveCFIRememberState() { 3966 getStreamer().emitCFIRememberState(); 3967 return false; 3968 } 3969 3970 /// parseDirectiveCFIRestoreState 3971 /// ::= .cfi_remember_state 3972 bool MasmParser::parseDirectiveCFIRestoreState() { 3973 getStreamer().emitCFIRestoreState(); 3974 return false; 3975 } 3976 3977 /// parseDirectiveCFISameValue 3978 /// ::= .cfi_same_value register 3979 bool MasmParser::parseDirectiveCFISameValue(SMLoc DirectiveLoc) { 3980 int64_t Register = 0; 3981 3982 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3983 return true; 3984 3985 getStreamer().emitCFISameValue(Register); 3986 return false; 3987 } 3988 3989 /// parseDirectiveCFIRestore 3990 /// ::= .cfi_restore register 3991 bool MasmParser::parseDirectiveCFIRestore(SMLoc DirectiveLoc) { 3992 int64_t Register = 0; 3993 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 3994 return true; 3995 3996 getStreamer().emitCFIRestore(Register); 3997 return false; 3998 } 3999 4000 /// parseDirectiveCFIEscape 4001 /// ::= .cfi_escape expression[,...] 4002 bool MasmParser::parseDirectiveCFIEscape() { 4003 std::string Values; 4004 int64_t CurrValue; 4005 if (parseAbsoluteExpression(CurrValue)) 4006 return true; 4007 4008 Values.push_back((uint8_t)CurrValue); 4009 4010 while (getLexer().is(AsmToken::Comma)) { 4011 Lex(); 4012 4013 if (parseAbsoluteExpression(CurrValue)) 4014 return true; 4015 4016 Values.push_back((uint8_t)CurrValue); 4017 } 4018 4019 getStreamer().emitCFIEscape(Values); 4020 return false; 4021 } 4022 4023 /// parseDirectiveCFIReturnColumn 4024 /// ::= .cfi_return_column register 4025 bool MasmParser::parseDirectiveCFIReturnColumn(SMLoc DirectiveLoc) { 4026 int64_t Register = 0; 4027 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4028 return true; 4029 getStreamer().emitCFIReturnColumn(Register); 4030 return false; 4031 } 4032 4033 /// parseDirectiveCFISignalFrame 4034 /// ::= .cfi_signal_frame 4035 bool MasmParser::parseDirectiveCFISignalFrame() { 4036 if (parseToken(AsmToken::EndOfStatement, 4037 "unexpected token in '.cfi_signal_frame'")) 4038 return true; 4039 4040 getStreamer().emitCFISignalFrame(); 4041 return false; 4042 } 4043 4044 /// parseDirectiveCFIUndefined 4045 /// ::= .cfi_undefined register 4046 bool MasmParser::parseDirectiveCFIUndefined(SMLoc DirectiveLoc) { 4047 int64_t Register = 0; 4048 4049 if (parseRegisterOrRegisterNumber(Register, DirectiveLoc)) 4050 return true; 4051 4052 getStreamer().emitCFIUndefined(Register); 4053 return false; 4054 } 4055 4056 /// parseDirectiveAltmacro 4057 /// ::= .altmacro 4058 /// ::= .noaltmacro 4059 bool MasmParser::parseDirectiveAltmacro(StringRef Directive) { 4060 if (getLexer().isNot(AsmToken::EndOfStatement)) 4061 return TokError("unexpected token in '" + Directive + "' directive"); 4062 AltMacroMode = (Directive == ".altmacro"); 4063 return false; 4064 } 4065 4066 /// parseDirectiveMacro 4067 /// ::= .macro name[,] [parameters] 4068 bool MasmParser::parseDirectiveMacro(SMLoc DirectiveLoc) { 4069 StringRef Name; 4070 if (parseIdentifier(Name)) 4071 return TokError("expected identifier in '.macro' directive"); 4072 4073 if (getLexer().is(AsmToken::Comma)) 4074 Lex(); 4075 4076 MCAsmMacroParameters Parameters; 4077 while (getLexer().isNot(AsmToken::EndOfStatement)) { 4078 4079 if (!Parameters.empty() && Parameters.back().Vararg) 4080 return Error(Lexer.getLoc(), 4081 "Vararg parameter '" + Parameters.back().Name + 4082 "' should be last one in the list of parameters."); 4083 4084 MCAsmMacroParameter Parameter; 4085 if (parseIdentifier(Parameter.Name)) 4086 return TokError("expected identifier in '.macro' directive"); 4087 4088 // Emit an error if two (or more) named parameters share the same name 4089 for (const MCAsmMacroParameter& CurrParam : Parameters) 4090 if (CurrParam.Name.equals(Parameter.Name)) 4091 return TokError("macro '" + Name + "' has multiple parameters" 4092 " named '" + Parameter.Name + "'"); 4093 4094 if (Lexer.is(AsmToken::Colon)) { 4095 Lex(); // consume ':' 4096 4097 SMLoc QualLoc; 4098 StringRef Qualifier; 4099 4100 QualLoc = Lexer.getLoc(); 4101 if (parseIdentifier(Qualifier)) 4102 return Error(QualLoc, "missing parameter qualifier for " 4103 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4104 4105 if (Qualifier == "req") 4106 Parameter.Required = true; 4107 else if (Qualifier == "vararg") 4108 Parameter.Vararg = true; 4109 else 4110 return Error(QualLoc, Qualifier + " is not a valid parameter qualifier " 4111 "for '" + Parameter.Name + "' in macro '" + Name + "'"); 4112 } 4113 4114 if (getLexer().is(AsmToken::Equal)) { 4115 Lex(); 4116 4117 SMLoc ParamLoc; 4118 4119 ParamLoc = Lexer.getLoc(); 4120 if (parseMacroArgument(Parameter.Value, /*Vararg=*/false )) 4121 return true; 4122 4123 if (Parameter.Required) 4124 Warning(ParamLoc, "pointless default value for required parameter " 4125 "'" + Parameter.Name + "' in macro '" + Name + "'"); 4126 } 4127 4128 Parameters.push_back(std::move(Parameter)); 4129 4130 if (getLexer().is(AsmToken::Comma)) 4131 Lex(); 4132 } 4133 4134 // Eat just the end of statement. 4135 Lexer.Lex(); 4136 4137 // Consuming deferred text, so use Lexer.Lex to ignore Lexing Errors 4138 AsmToken EndToken, StartToken = getTok(); 4139 unsigned MacroDepth = 0; 4140 // Lex the macro definition. 4141 while (true) { 4142 // Ignore Lexing errors in macros. 4143 while (Lexer.is(AsmToken::Error)) { 4144 Lexer.Lex(); 4145 } 4146 4147 // Check whether we have reached the end of the file. 4148 if (getLexer().is(AsmToken::Eof)) 4149 return Error(DirectiveLoc, "no matching '.endmacro' in definition"); 4150 4151 // Otherwise, check whether we have reach the .endmacro. 4152 if (getLexer().is(AsmToken::Identifier)) { 4153 if (getTok().getIdentifier() == ".endm" || 4154 getTok().getIdentifier() == ".endmacro") { 4155 if (MacroDepth == 0) { // Outermost macro. 4156 EndToken = getTok(); 4157 Lexer.Lex(); 4158 if (getLexer().isNot(AsmToken::EndOfStatement)) 4159 return TokError("unexpected token in '" + EndToken.getIdentifier() + 4160 "' directive"); 4161 break; 4162 } else { 4163 // Otherwise we just found the end of an inner macro. 4164 --MacroDepth; 4165 } 4166 } else if (getTok().getIdentifier() == ".macro") { 4167 // We allow nested macros. Those aren't instantiated until the outermost 4168 // macro is expanded so just ignore them for now. 4169 ++MacroDepth; 4170 } 4171 } 4172 4173 // Otherwise, scan til the end of the statement. 4174 eatToEndOfStatement(); 4175 } 4176 4177 if (getContext().lookupMacro(Name)) { 4178 return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); 4179 } 4180 4181 const char *BodyStart = StartToken.getLoc().getPointer(); 4182 const char *BodyEnd = EndToken.getLoc().getPointer(); 4183 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 4184 checkForBadMacro(DirectiveLoc, Name, Body, Parameters); 4185 MCAsmMacro Macro(Name, Body, std::move(Parameters)); 4186 DEBUG_WITH_TYPE("asm-macros", dbgs() << "Defining new macro:\n"; 4187 Macro.dump()); 4188 getContext().defineMacro(Name, std::move(Macro)); 4189 return false; 4190 } 4191 4192 /// checkForBadMacro 4193 /// 4194 /// With the support added for named parameters there may be code out there that 4195 /// is transitioning from positional parameters. In versions of gas that did 4196 /// not support named parameters they would be ignored on the macro definition. 4197 /// But to support both styles of parameters this is not possible so if a macro 4198 /// definition has named parameters but does not use them and has what appears 4199 /// to be positional parameters, strings like $1, $2, ... and $n, then issue a 4200 /// warning that the positional parameter found in body which have no effect. 4201 /// Hoping the developer will either remove the named parameters from the macro 4202 /// definition so the positional parameters get used if that was what was 4203 /// intended or change the macro to use the named parameters. It is possible 4204 /// this warning will trigger when the none of the named parameters are used 4205 /// and the strings like $1 are infact to simply to be passed trough unchanged. 4206 void MasmParser::checkForBadMacro(SMLoc DirectiveLoc, StringRef Name, 4207 StringRef Body, 4208 ArrayRef<MCAsmMacroParameter> Parameters) { 4209 // If this macro is not defined with named parameters the warning we are 4210 // checking for here doesn't apply. 4211 unsigned NParameters = Parameters.size(); 4212 if (NParameters == 0) 4213 return; 4214 4215 bool NamedParametersFound = false; 4216 bool PositionalParametersFound = false; 4217 4218 // Look at the body of the macro for use of both the named parameters and what 4219 // are likely to be positional parameters. This is what expandMacro() is 4220 // doing when it finds the parameters in the body. 4221 while (!Body.empty()) { 4222 // Scan for the next possible parameter. 4223 std::size_t End = Body.size(), Pos = 0; 4224 for (; Pos != End; ++Pos) { 4225 // Check for a substitution or escape. 4226 // This macro is defined with parameters, look for \foo, \bar, etc. 4227 if (Body[Pos] == '\\' && Pos + 1 != End) 4228 break; 4229 4230 // This macro should have parameters, but look for $0, $1, ..., $n too. 4231 if (Body[Pos] != '$' || Pos + 1 == End) 4232 continue; 4233 char Next = Body[Pos + 1]; 4234 if (Next == '$' || Next == 'n' || 4235 isdigit(static_cast<unsigned char>(Next))) 4236 break; 4237 } 4238 4239 // Check if we reached the end. 4240 if (Pos == End) 4241 break; 4242 4243 if (Body[Pos] == '$') { 4244 switch (Body[Pos + 1]) { 4245 // $$ => $ 4246 case '$': 4247 break; 4248 4249 // $n => number of arguments 4250 case 'n': 4251 PositionalParametersFound = true; 4252 break; 4253 4254 // $[0-9] => argument 4255 default: { 4256 PositionalParametersFound = true; 4257 break; 4258 } 4259 } 4260 Pos += 2; 4261 } else { 4262 unsigned I = Pos + 1; 4263 while (isIdentifierChar(Body[I]) && I + 1 != End) 4264 ++I; 4265 4266 const char *Begin = Body.data() + Pos + 1; 4267 StringRef Argument(Begin, I - (Pos + 1)); 4268 unsigned Index = 0; 4269 for (; Index < NParameters; ++Index) 4270 if (Parameters[Index].Name == Argument) 4271 break; 4272 4273 if (Index == NParameters) { 4274 if (Body[Pos + 1] == '(' && Body[Pos + 2] == ')') 4275 Pos += 3; 4276 else { 4277 Pos = I; 4278 } 4279 } else { 4280 NamedParametersFound = true; 4281 Pos += 1 + Argument.size(); 4282 } 4283 } 4284 // Update the scan point. 4285 Body = Body.substr(Pos); 4286 } 4287 4288 if (!NamedParametersFound && PositionalParametersFound) 4289 Warning(DirectiveLoc, "macro defined with named parameters which are not " 4290 "used in macro body, possible positional parameter " 4291 "found in body which will have no effect"); 4292 } 4293 4294 /// parseDirectiveExitMacro 4295 /// ::= .exitm 4296 bool MasmParser::parseDirectiveExitMacro(StringRef Directive) { 4297 if (parseToken(AsmToken::EndOfStatement, 4298 "unexpected token in '" + Directive + "' directive")) 4299 return true; 4300 4301 if (!isInsideMacroInstantiation()) 4302 return TokError("unexpected '" + Directive + "' in file, " 4303 "no current macro definition"); 4304 4305 // Exit all conditionals that are active in the current macro. 4306 while (TheCondStack.size() != ActiveMacros.back()->CondStackDepth) { 4307 TheCondState = TheCondStack.back(); 4308 TheCondStack.pop_back(); 4309 } 4310 4311 handleMacroExit(); 4312 return false; 4313 } 4314 4315 /// parseDirectiveEndMacro 4316 /// ::= .endm 4317 /// ::= .endmacro 4318 bool MasmParser::parseDirectiveEndMacro(StringRef Directive) { 4319 if (getLexer().isNot(AsmToken::EndOfStatement)) 4320 return TokError("unexpected token in '" + Directive + "' directive"); 4321 4322 // If we are inside a macro instantiation, terminate the current 4323 // instantiation. 4324 if (isInsideMacroInstantiation()) { 4325 handleMacroExit(); 4326 return false; 4327 } 4328 4329 // Otherwise, this .endmacro is a stray entry in the file; well formed 4330 // .endmacro directives are handled during the macro definition parsing. 4331 return TokError("unexpected '" + Directive + "' in file, " 4332 "no current macro definition"); 4333 } 4334 4335 /// parseDirectivePurgeMacro 4336 /// ::= .purgem 4337 bool MasmParser::parseDirectivePurgeMacro(SMLoc DirectiveLoc) { 4338 StringRef Name; 4339 SMLoc Loc; 4340 if (parseTokenLoc(Loc) || 4341 check(parseIdentifier(Name), Loc, 4342 "expected identifier in '.purgem' directive") || 4343 parseToken(AsmToken::EndOfStatement, 4344 "unexpected token in '.purgem' directive")) 4345 return true; 4346 4347 if (!getContext().lookupMacro(Name)) 4348 return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); 4349 4350 getContext().undefineMacro(Name); 4351 DEBUG_WITH_TYPE("asm-macros", dbgs() 4352 << "Un-defining macro: " << Name << "\n"); 4353 return false; 4354 } 4355 4356 /// parseDirectiveSymbolAttribute 4357 /// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] 4358 bool MasmParser::parseDirectiveSymbolAttribute(MCSymbolAttr Attr) { 4359 auto parseOp = [&]() -> bool { 4360 StringRef Name; 4361 SMLoc Loc = getTok().getLoc(); 4362 if (parseIdentifier(Name)) 4363 return Error(Loc, "expected identifier"); 4364 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4365 4366 // Assembler local symbols don't make any sense here. Complain loudly. 4367 if (Sym->isTemporary()) 4368 return Error(Loc, "non-local symbol required"); 4369 4370 if (!getStreamer().emitSymbolAttribute(Sym, Attr)) 4371 return Error(Loc, "unable to emit symbol attribute"); 4372 return false; 4373 }; 4374 4375 if (parseMany(parseOp)) 4376 return addErrorSuffix(" in directive"); 4377 return false; 4378 } 4379 4380 /// parseDirectiveComm 4381 /// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] 4382 bool MasmParser::parseDirectiveComm(bool IsLocal) { 4383 if (checkForValidSection()) 4384 return true; 4385 4386 SMLoc IDLoc = getLexer().getLoc(); 4387 StringRef Name; 4388 if (parseIdentifier(Name)) 4389 return TokError("expected identifier in directive"); 4390 4391 // Handle the identifier as the key symbol. 4392 MCSymbol *Sym = getContext().getOrCreateSymbol(Name); 4393 4394 if (getLexer().isNot(AsmToken::Comma)) 4395 return TokError("unexpected token in directive"); 4396 Lex(); 4397 4398 int64_t Size; 4399 SMLoc SizeLoc = getLexer().getLoc(); 4400 if (parseAbsoluteExpression(Size)) 4401 return true; 4402 4403 int64_t Pow2Alignment = 0; 4404 SMLoc Pow2AlignmentLoc; 4405 if (getLexer().is(AsmToken::Comma)) { 4406 Lex(); 4407 Pow2AlignmentLoc = getLexer().getLoc(); 4408 if (parseAbsoluteExpression(Pow2Alignment)) 4409 return true; 4410 4411 LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); 4412 if (IsLocal && LCOMM == LCOMM::NoAlignment) 4413 return Error(Pow2AlignmentLoc, "alignment not supported on this target"); 4414 4415 // If this target takes alignments in bytes (not log) validate and convert. 4416 if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || 4417 (IsLocal && LCOMM == LCOMM::ByteAlignment)) { 4418 if (!isPowerOf2_64(Pow2Alignment)) 4419 return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); 4420 Pow2Alignment = Log2_64(Pow2Alignment); 4421 } 4422 } 4423 4424 if (parseToken(AsmToken::EndOfStatement, 4425 "unexpected token in '.comm' or '.lcomm' directive")) 4426 return true; 4427 4428 // NOTE: a size of zero for a .comm should create a undefined symbol 4429 // but a size of .lcomm creates a bss symbol of size zero. 4430 if (Size < 0) 4431 return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " 4432 "be less than zero"); 4433 4434 // NOTE: The alignment in the directive is a power of 2 value, the assembler 4435 // may internally end up wanting an alignment in bytes. 4436 // FIXME: Diagnose overflow. 4437 if (Pow2Alignment < 0) 4438 return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " 4439 "alignment, can't be less than zero"); 4440 4441 Sym->redefineIfPossible(); 4442 if (!Sym->isUndefined()) 4443 return Error(IDLoc, "invalid symbol redefinition"); 4444 4445 // Create the Symbol as a common or local common with Size and Pow2Alignment 4446 if (IsLocal) { 4447 getStreamer().emitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4448 return false; 4449 } 4450 4451 getStreamer().emitCommonSymbol(Sym, Size, 1 << Pow2Alignment); 4452 return false; 4453 } 4454 4455 /// parseDirectiveComment 4456 /// ::= comment delimiter [[text]] 4457 /// [[text]] 4458 /// [[text]] delimiter [[text]] 4459 bool MasmParser::parseDirectiveComment(SMLoc DirectiveLoc) { 4460 StringRef FirstLine = parseStringToEndOfStatement(); 4461 size_t DelimiterEnd = FirstLine.find_first_of("\b\t\v\f\r\x1A "); 4462 StringRef Delimiter = FirstLine.take_front(DelimiterEnd); 4463 if (Delimiter.empty()) 4464 return Error(DirectiveLoc, "no delimiter in 'comment' directive"); 4465 do { 4466 if (getTok().is(AsmToken::Eof)) 4467 return Error(DirectiveLoc, "unmatched delimiter in 'comment' directive"); 4468 Lex(); // eat end of statement 4469 } while (!parseStringToEndOfStatement().contains(Delimiter)); 4470 return parseToken(AsmToken::EndOfStatement, 4471 "unexpected token in 'comment' directive"); 4472 } 4473 4474 /// parseDirectiveInclude 4475 /// ::= include <filename> 4476 /// | include filename 4477 bool MasmParser::parseDirectiveInclude() { 4478 // Allow the strings to have escaped octal character sequence. 4479 std::string Filename; 4480 SMLoc IncludeLoc = getTok().getLoc(); 4481 4482 if (!parseAngleBracketString(Filename)) 4483 Filename = parseStringToEndOfStatement().str(); 4484 if (check(!Filename.empty(), "missing filename in 'include' directive") || 4485 check(getTok().isNot(AsmToken::EndOfStatement), 4486 "unexpected token in 'include' directive") || 4487 // Attempt to switch the lexer to the included file before consuming the 4488 // end of statement to avoid losing it when we switch. 4489 check(enterIncludeFile(Filename), IncludeLoc, 4490 "Could not find include file '" + Filename + "'")) 4491 return true; 4492 4493 return false; 4494 } 4495 4496 /// parseDirectiveIf 4497 /// ::= .if{,eq,ge,gt,le,lt,ne} expression 4498 bool MasmParser::parseDirectiveIf(SMLoc DirectiveLoc, DirectiveKind DirKind) { 4499 TheCondStack.push_back(TheCondState); 4500 TheCondState.TheCond = AsmCond::IfCond; 4501 if (TheCondState.Ignore) { 4502 eatToEndOfStatement(); 4503 } else { 4504 int64_t ExprValue; 4505 if (parseAbsoluteExpression(ExprValue) || 4506 parseToken(AsmToken::EndOfStatement, 4507 "unexpected token in '.if' directive")) 4508 return true; 4509 4510 switch (DirKind) { 4511 default: 4512 llvm_unreachable("unsupported directive"); 4513 case DK_IF: 4514 break; 4515 case DK_IFE: 4516 ExprValue = ExprValue == 0; 4517 break; 4518 } 4519 4520 TheCondState.CondMet = ExprValue; 4521 TheCondState.Ignore = !TheCondState.CondMet; 4522 } 4523 4524 return false; 4525 } 4526 4527 /// parseDirectiveIfb 4528 /// ::= .ifb string 4529 bool MasmParser::parseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4530 TheCondStack.push_back(TheCondState); 4531 TheCondState.TheCond = AsmCond::IfCond; 4532 4533 if (TheCondState.Ignore) { 4534 eatToEndOfStatement(); 4535 } else { 4536 std::string Str; 4537 if (parseTextItem(Str)) 4538 return TokError("expected string parameter for 'ifb' directive"); 4539 4540 if (parseToken(AsmToken::EndOfStatement, 4541 "unexpected token in 'ifb' directive")) 4542 return true; 4543 4544 TheCondState.CondMet = ExpectBlank == Str.empty(); 4545 TheCondState.Ignore = !TheCondState.CondMet; 4546 } 4547 4548 return false; 4549 } 4550 4551 /// parseDirectiveIfidn 4552 /// ::= ifidn string1, string2 4553 bool MasmParser::parseDirectiveIfidn(SMLoc DirectiveLoc, bool ExpectEqual, bool CaseInsensitive) { 4554 std::string String1, String2; 4555 4556 if (parseTextItem(String1)) { 4557 if (ExpectEqual) 4558 return TokError("expected string parameter for 'ifidn' directive"); 4559 return TokError("expected string parameter for 'ifdif' directive"); 4560 } 4561 4562 if (Lexer.isNot(AsmToken::Comma)) { 4563 if (ExpectEqual) 4564 return TokError( 4565 "expected comma after first string for 'ifidn' directive"); 4566 return TokError("expected comma after first string for 'ifdif' directive"); 4567 } 4568 Lex(); 4569 4570 if (parseTextItem(String2)) { 4571 if (ExpectEqual) 4572 return TokError("expected string parameter for 'ifidn' directive"); 4573 return TokError("expected string parameter for 'ifdif' directive"); 4574 } 4575 4576 TheCondStack.push_back(TheCondState); 4577 TheCondState.TheCond = AsmCond::IfCond; 4578 if (CaseInsensitive) 4579 TheCondState.CondMet = 4580 ExpectEqual == (StringRef(String1).equals_lower(String2)); 4581 else 4582 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4583 TheCondState.Ignore = !TheCondState.CondMet; 4584 4585 return false; 4586 } 4587 4588 /// parseDirectiveIfdef 4589 /// ::= ifdef symbol 4590 /// | ifdef variable 4591 bool MasmParser::parseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { 4592 TheCondStack.push_back(TheCondState); 4593 TheCondState.TheCond = AsmCond::IfCond; 4594 4595 if (TheCondState.Ignore) { 4596 eatToEndOfStatement(); 4597 } else { 4598 bool is_defined = false; 4599 unsigned RegNo; 4600 SMLoc StartLoc, EndLoc; 4601 is_defined = (getTargetParser().tryParseRegister( 4602 RegNo, StartLoc, EndLoc) == MatchOperand_Success); 4603 if (!is_defined) { 4604 StringRef Name; 4605 if (check(parseIdentifier(Name), "expected identifier after 'ifdef'") || 4606 parseToken(AsmToken::EndOfStatement, "unexpected token in 'ifdef'")) 4607 return true; 4608 4609 if (Variables.find(Name) != Variables.end()) { 4610 is_defined = true; 4611 } else { 4612 MCSymbol *Sym = getContext().lookupSymbol(Name); 4613 is_defined = (Sym && !Sym->isUndefined(false)); 4614 } 4615 } 4616 4617 TheCondState.CondMet = (is_defined == expect_defined); 4618 TheCondState.Ignore = !TheCondState.CondMet; 4619 } 4620 4621 return false; 4622 } 4623 4624 /// parseDirectiveElseIf 4625 /// ::= elseif expression 4626 bool MasmParser::parseDirectiveElseIf(SMLoc DirectiveLoc, 4627 DirectiveKind DirKind) { 4628 if (TheCondState.TheCond != AsmCond::IfCond && 4629 TheCondState.TheCond != AsmCond::ElseIfCond) 4630 return Error(DirectiveLoc, "Encountered a .elseif that doesn't follow an" 4631 " .if or an .elseif"); 4632 TheCondState.TheCond = AsmCond::ElseIfCond; 4633 4634 bool LastIgnoreState = false; 4635 if (!TheCondStack.empty()) 4636 LastIgnoreState = TheCondStack.back().Ignore; 4637 if (LastIgnoreState || TheCondState.CondMet) { 4638 TheCondState.Ignore = true; 4639 eatToEndOfStatement(); 4640 } else { 4641 int64_t ExprValue; 4642 if (parseAbsoluteExpression(ExprValue)) 4643 return true; 4644 4645 if (parseToken(AsmToken::EndOfStatement, 4646 "unexpected token in '.elseif' directive")) 4647 return true; 4648 4649 switch (DirKind) { 4650 default: 4651 llvm_unreachable("unsupported directive"); 4652 case DK_ELSEIF: 4653 break; 4654 case DK_ELSEIFE: 4655 ExprValue = ExprValue == 0; 4656 break; 4657 } 4658 4659 TheCondState.CondMet = ExprValue; 4660 TheCondState.Ignore = !TheCondState.CondMet; 4661 } 4662 4663 return false; 4664 } 4665 4666 /// parseDirectiveElseIfb 4667 /// ::= elseifb expression 4668 bool MasmParser::parseDirectiveElseIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4669 if (TheCondState.TheCond != AsmCond::IfCond && 4670 TheCondState.TheCond != AsmCond::ElseIfCond) 4671 return Error(DirectiveLoc, "Encountered an elseif that doesn't follow an" 4672 " if or an elseif"); 4673 TheCondState.TheCond = AsmCond::ElseIfCond; 4674 4675 bool LastIgnoreState = false; 4676 if (!TheCondStack.empty()) 4677 LastIgnoreState = TheCondStack.back().Ignore; 4678 if (LastIgnoreState || TheCondState.CondMet) { 4679 TheCondState.Ignore = true; 4680 eatToEndOfStatement(); 4681 } else { 4682 std::string Str; 4683 if (parseTextItem(Str)) 4684 return TokError("expected string parameter for 'elseifb' directive"); 4685 4686 if (parseToken(AsmToken::EndOfStatement, 4687 "unexpected token in 'elseifb' directive")) 4688 return true; 4689 4690 TheCondState.CondMet = ExpectBlank == Str.empty(); 4691 TheCondState.Ignore = !TheCondState.CondMet; 4692 } 4693 4694 return false; 4695 } 4696 4697 /// parseDirectiveElseIfdef 4698 /// ::= elseifdef symbol 4699 /// | elseifdef variable 4700 bool MasmParser::parseDirectiveElseIfdef(SMLoc DirectiveLoc, 4701 bool expect_defined) { 4702 if (TheCondState.TheCond != AsmCond::IfCond && 4703 TheCondState.TheCond != AsmCond::ElseIfCond) 4704 return Error(DirectiveLoc, "Encountered an elseif that doesn't follow an" 4705 " if or an elseif"); 4706 TheCondState.TheCond = AsmCond::ElseIfCond; 4707 4708 bool LastIgnoreState = false; 4709 if (!TheCondStack.empty()) 4710 LastIgnoreState = TheCondStack.back().Ignore; 4711 if (LastIgnoreState || TheCondState.CondMet) { 4712 TheCondState.Ignore = true; 4713 eatToEndOfStatement(); 4714 } else { 4715 bool is_defined = false; 4716 unsigned RegNo; 4717 SMLoc StartLoc, EndLoc; 4718 is_defined = (getTargetParser().tryParseRegister(RegNo, StartLoc, EndLoc) == 4719 MatchOperand_Success); 4720 if (!is_defined) { 4721 StringRef Name; 4722 if (check(parseIdentifier(Name), 4723 "expected identifier after 'elseifdef'") || 4724 parseToken(AsmToken::EndOfStatement, 4725 "unexpected token in 'elseifdef'")) 4726 return true; 4727 4728 if (Variables.find(Name) != Variables.end()) { 4729 is_defined = true; 4730 } else { 4731 MCSymbol *Sym = getContext().lookupSymbol(Name); 4732 is_defined = (Sym && !Sym->isUndefined(false)); 4733 } 4734 } 4735 4736 TheCondState.CondMet = (is_defined == expect_defined); 4737 TheCondState.Ignore = !TheCondState.CondMet; 4738 } 4739 4740 return false; 4741 } 4742 4743 /// parseDirectiveElseIfidn 4744 /// ::= elseifidn string1, string2 4745 bool MasmParser::parseDirectiveElseIfidn(SMLoc DirectiveLoc, bool ExpectEqual, 4746 bool CaseInsensitive) { 4747 if (TheCondState.TheCond != AsmCond::IfCond && 4748 TheCondState.TheCond != AsmCond::ElseIfCond) 4749 return Error(DirectiveLoc, "Encountered an elseif that doesn't follow an" 4750 " if or an elseif"); 4751 TheCondState.TheCond = AsmCond::ElseIfCond; 4752 4753 bool LastIgnoreState = false; 4754 if (!TheCondStack.empty()) 4755 LastIgnoreState = TheCondStack.back().Ignore; 4756 if (LastIgnoreState || TheCondState.CondMet) { 4757 TheCondState.Ignore = true; 4758 eatToEndOfStatement(); 4759 } else { 4760 std::string String1, String2; 4761 4762 if (parseTextItem(String1)) { 4763 if (ExpectEqual) 4764 return TokError("expected string parameter for 'elseifidn' directive"); 4765 return TokError("expected string parameter for 'elseifdif' directive"); 4766 } 4767 4768 if (Lexer.isNot(AsmToken::Comma)) { 4769 if (ExpectEqual) 4770 return TokError( 4771 "expected comma after first string for 'elseifidn' directive"); 4772 return TokError( 4773 "expected comma after first string for 'elseifdif' directive"); 4774 } 4775 Lex(); 4776 4777 if (parseTextItem(String2)) { 4778 if (ExpectEqual) 4779 return TokError("expected string parameter for 'elseifidn' directive"); 4780 return TokError("expected string parameter for 'elseifdif' directive"); 4781 } 4782 4783 if (CaseInsensitive) 4784 TheCondState.CondMet = 4785 ExpectEqual == (StringRef(String1).equals_lower(String2)); 4786 else 4787 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4788 TheCondState.Ignore = !TheCondState.CondMet; 4789 } 4790 4791 return false; 4792 } 4793 4794 /// parseDirectiveElse 4795 /// ::= else 4796 bool MasmParser::parseDirectiveElse(SMLoc DirectiveLoc) { 4797 if (parseToken(AsmToken::EndOfStatement, 4798 "unexpected token in 'else' directive")) 4799 return true; 4800 4801 if (TheCondState.TheCond != AsmCond::IfCond && 4802 TheCondState.TheCond != AsmCond::ElseIfCond) 4803 return Error(DirectiveLoc, "Encountered an else that doesn't follow an if" 4804 " or an elseif"); 4805 TheCondState.TheCond = AsmCond::ElseCond; 4806 bool LastIgnoreState = false; 4807 if (!TheCondStack.empty()) 4808 LastIgnoreState = TheCondStack.back().Ignore; 4809 if (LastIgnoreState || TheCondState.CondMet) 4810 TheCondState.Ignore = true; 4811 else 4812 TheCondState.Ignore = false; 4813 4814 return false; 4815 } 4816 4817 /// parseDirectiveEnd 4818 /// ::= end 4819 bool MasmParser::parseDirectiveEnd(SMLoc DirectiveLoc) { 4820 if (parseToken(AsmToken::EndOfStatement, 4821 "unexpected token in 'end' directive")) 4822 return true; 4823 4824 while (Lexer.isNot(AsmToken::Eof)) 4825 Lexer.Lex(); 4826 4827 return false; 4828 } 4829 4830 /// parseDirectiveError 4831 /// ::= .err [message] 4832 bool MasmParser::parseDirectiveError(SMLoc DirectiveLoc) { 4833 if (!TheCondStack.empty()) { 4834 if (TheCondStack.back().Ignore) { 4835 eatToEndOfStatement(); 4836 return false; 4837 } 4838 } 4839 4840 StringRef Message = ".err directive invoked in source file"; 4841 if (Lexer.isNot(AsmToken::EndOfStatement)) 4842 Message = parseStringToEndOfStatement(); 4843 Lex(); 4844 4845 return Error(DirectiveLoc, Message); 4846 } 4847 4848 /// parseDirectiveErrorIfb 4849 /// ::= .errb textitem[, message] 4850 bool MasmParser::parseDirectiveErrorIfb(SMLoc DirectiveLoc, bool ExpectBlank) { 4851 if (!TheCondStack.empty()) { 4852 if (TheCondStack.back().Ignore) { 4853 eatToEndOfStatement(); 4854 return false; 4855 } 4856 } 4857 4858 std::string Text; 4859 if (parseTextItem(Text)) 4860 return Error(getTok().getLoc(), "missing text item in '.errb' directive"); 4861 4862 StringRef Message = ".errb directive invoked in source file"; 4863 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4864 if (parseToken(AsmToken::Comma)) 4865 return addErrorSuffix(" in '.errb' directive"); 4866 Message = parseStringToEndOfStatement(); 4867 } 4868 Lex(); 4869 4870 if (Text.empty() == ExpectBlank) 4871 return Error(DirectiveLoc, Message); 4872 return false; 4873 } 4874 4875 /// parseDirectiveErrorIfdef 4876 /// ::= .errdef name[, message] 4877 bool MasmParser::parseDirectiveErrorIfdef(SMLoc DirectiveLoc, 4878 bool ExpectDefined) { 4879 if (!TheCondStack.empty()) { 4880 if (TheCondStack.back().Ignore) { 4881 eatToEndOfStatement(); 4882 return false; 4883 } 4884 } 4885 4886 bool IsDefined = false; 4887 unsigned RegNo; 4888 SMLoc StartLoc, EndLoc; 4889 IsDefined = (getTargetParser().tryParseRegister(RegNo, StartLoc, EndLoc) == 4890 MatchOperand_Success); 4891 if (!IsDefined) { 4892 StringRef Name; 4893 if (check(parseIdentifier(Name), "expected identifier after '.errdef'")) 4894 return true; 4895 4896 if (Variables.find(Name) != Variables.end()) { 4897 IsDefined = true; 4898 } else { 4899 MCSymbol *Sym = getContext().lookupSymbol(Name); 4900 IsDefined = (Sym && !Sym->isUndefined(false)); 4901 } 4902 } 4903 4904 StringRef Message = ".errdef directive invoked in source file"; 4905 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4906 if (parseToken(AsmToken::Comma)) 4907 return addErrorSuffix(" in '.errdef' directive"); 4908 Message = parseStringToEndOfStatement(); 4909 } 4910 Lex(); 4911 4912 if (IsDefined == ExpectDefined) 4913 return Error(DirectiveLoc, Message); 4914 return false; 4915 } 4916 4917 /// parseDirectiveErrorIfidn 4918 /// ::= .erridn textitem1, textitem2[, message] 4919 bool MasmParser::parseDirectiveErrorIfidn(SMLoc DirectiveLoc, bool ExpectEqual, 4920 bool CaseInsensitive) { 4921 if (!TheCondStack.empty()) { 4922 if (TheCondStack.back().Ignore) { 4923 eatToEndOfStatement(); 4924 return false; 4925 } 4926 } 4927 4928 std::string String1, String2; 4929 4930 if (parseTextItem(String1)) { 4931 if (ExpectEqual) 4932 return TokError("expected string parameter for '.erridn' directive"); 4933 return TokError("expected string parameter for '.errdif' directive"); 4934 } 4935 4936 if (Lexer.isNot(AsmToken::Comma)) { 4937 if (ExpectEqual) 4938 return TokError( 4939 "expected comma after first string for '.erridn' directive"); 4940 return TokError( 4941 "expected comma after first string for '.errdif' directive"); 4942 } 4943 Lex(); 4944 4945 if (parseTextItem(String2)) { 4946 if (ExpectEqual) 4947 return TokError("expected string parameter for '.erridn' directive"); 4948 return TokError("expected string parameter for '.errdif' directive"); 4949 } 4950 4951 StringRef Message; 4952 if (ExpectEqual) 4953 Message = ".erridn directive invoked in source file"; 4954 else 4955 Message = ".errdif directive invoked in source file"; 4956 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4957 if (parseToken(AsmToken::Comma)) 4958 return addErrorSuffix(" in '.erridn' directive"); 4959 Message = parseStringToEndOfStatement(); 4960 } 4961 Lex(); 4962 4963 if (CaseInsensitive) 4964 TheCondState.CondMet = 4965 ExpectEqual == (StringRef(String1).equals_lower(String2)); 4966 else 4967 TheCondState.CondMet = ExpectEqual == (String1 == String2); 4968 TheCondState.Ignore = !TheCondState.CondMet; 4969 4970 if ((CaseInsensitive && 4971 ExpectEqual == StringRef(String1).equals_lower(String2)) || 4972 (ExpectEqual == (String1 == String2))) 4973 return Error(DirectiveLoc, Message); 4974 return false; 4975 } 4976 4977 /// parseDirectiveErrorIfe 4978 /// ::= .erre expression[, message] 4979 bool MasmParser::parseDirectiveErrorIfe(SMLoc DirectiveLoc, bool ExpectZero) { 4980 if (!TheCondStack.empty()) { 4981 if (TheCondStack.back().Ignore) { 4982 eatToEndOfStatement(); 4983 return false; 4984 } 4985 } 4986 4987 int64_t ExprValue; 4988 if (parseAbsoluteExpression(ExprValue)) 4989 return addErrorSuffix(" in '.erre' directive"); 4990 4991 StringRef Message = ".erre directive invoked in source file"; 4992 if (Lexer.isNot(AsmToken::EndOfStatement)) { 4993 if (parseToken(AsmToken::Comma)) 4994 return addErrorSuffix(" in '.erre' directive"); 4995 Message = parseStringToEndOfStatement(); 4996 } 4997 Lex(); 4998 4999 if ((ExprValue == 0) == ExpectZero) 5000 return Error(DirectiveLoc, Message); 5001 return false; 5002 } 5003 5004 /// parseDirectiveEndIf 5005 /// ::= .endif 5006 bool MasmParser::parseDirectiveEndIf(SMLoc DirectiveLoc) { 5007 if (parseToken(AsmToken::EndOfStatement, 5008 "unexpected token in '.endif' directive")) 5009 return true; 5010 5011 if ((TheCondState.TheCond == AsmCond::NoCond) || TheCondStack.empty()) 5012 return Error(DirectiveLoc, "Encountered a .endif that doesn't follow " 5013 "an .if or .else"); 5014 if (!TheCondStack.empty()) { 5015 TheCondState = TheCondStack.back(); 5016 TheCondStack.pop_back(); 5017 } 5018 5019 return false; 5020 } 5021 5022 void MasmParser::initializeDirectiveKindMap() { 5023 DirectiveKindMap["="] = DK_ASSIGN; 5024 DirectiveKindMap["equ"] = DK_EQU; 5025 DirectiveKindMap["textequ"] = DK_TEXTEQU; 5026 // DirectiveKindMap[".ascii"] = DK_ASCII; 5027 // DirectiveKindMap[".asciz"] = DK_ASCIZ; 5028 // DirectiveKindMap[".string"] = DK_STRING; 5029 DirectiveKindMap["byte"] = DK_BYTE; 5030 DirectiveKindMap["sbyte"] = DK_SBYTE; 5031 DirectiveKindMap["word"] = DK_WORD; 5032 DirectiveKindMap["sword"] = DK_SWORD; 5033 DirectiveKindMap["dword"] = DK_DWORD; 5034 DirectiveKindMap["sdword"] = DK_SDWORD; 5035 DirectiveKindMap["fword"] = DK_FWORD; 5036 DirectiveKindMap["qword"] = DK_QWORD; 5037 DirectiveKindMap["sqword"] = DK_SQWORD; 5038 DirectiveKindMap["real4"] = DK_REAL4; 5039 DirectiveKindMap["real8"] = DK_REAL8; 5040 DirectiveKindMap["align"] = DK_ALIGN; 5041 // DirectiveKindMap[".org"] = DK_ORG; 5042 DirectiveKindMap["extern"] = DK_EXTERN; 5043 DirectiveKindMap["public"] = DK_PUBLIC; 5044 // DirectiveKindMap[".comm"] = DK_COMM; 5045 DirectiveKindMap["comment"] = DK_COMMENT; 5046 DirectiveKindMap["include"] = DK_INCLUDE; 5047 // DirectiveKindMap[".rept"] = DK_REPT; 5048 // DirectiveKindMap[".rep"] = DK_REPT; 5049 // DirectiveKindMap[".irp"] = DK_IRP; 5050 // DirectiveKindMap[".irpc"] = DK_IRPC; 5051 // DirectiveKindMap[".endr"] = DK_ENDR; 5052 DirectiveKindMap["if"] = DK_IF; 5053 DirectiveKindMap["ife"] = DK_IFE; 5054 DirectiveKindMap["ifb"] = DK_IFB; 5055 DirectiveKindMap["ifnb"] = DK_IFNB; 5056 DirectiveKindMap["ifdef"] = DK_IFDEF; 5057 DirectiveKindMap["ifndef"] = DK_IFNDEF; 5058 DirectiveKindMap["ifdif"] = DK_IFDIF; 5059 DirectiveKindMap["ifdifi"] = DK_IFDIFI; 5060 DirectiveKindMap["ifidn"] = DK_IFIDN; 5061 DirectiveKindMap["ifidni"] = DK_IFIDNI; 5062 DirectiveKindMap["elseif"] = DK_ELSEIF; 5063 DirectiveKindMap["elseifdef"] = DK_ELSEIFDEF; 5064 DirectiveKindMap["elseifndef"] = DK_ELSEIFNDEF; 5065 DirectiveKindMap["elseifdif"] = DK_ELSEIFDIF; 5066 DirectiveKindMap["elseifidn"] = DK_ELSEIFIDN; 5067 DirectiveKindMap["else"] = DK_ELSE; 5068 DirectiveKindMap["end"] = DK_END; 5069 DirectiveKindMap["endif"] = DK_ENDIF; 5070 // DirectiveKindMap[".file"] = DK_FILE; 5071 // DirectiveKindMap[".line"] = DK_LINE; 5072 // DirectiveKindMap[".loc"] = DK_LOC; 5073 // DirectiveKindMap[".stabs"] = DK_STABS; 5074 // DirectiveKindMap[".cv_file"] = DK_CV_FILE; 5075 // DirectiveKindMap[".cv_func_id"] = DK_CV_FUNC_ID; 5076 // DirectiveKindMap[".cv_loc"] = DK_CV_LOC; 5077 // DirectiveKindMap[".cv_linetable"] = DK_CV_LINETABLE; 5078 // DirectiveKindMap[".cv_inline_linetable"] = DK_CV_INLINE_LINETABLE; 5079 // DirectiveKindMap[".cv_inline_site_id"] = DK_CV_INLINE_SITE_ID; 5080 // DirectiveKindMap[".cv_def_range"] = DK_CV_DEF_RANGE; 5081 // DirectiveKindMap[".cv_string"] = DK_CV_STRING; 5082 // DirectiveKindMap[".cv_stringtable"] = DK_CV_STRINGTABLE; 5083 // DirectiveKindMap[".cv_filechecksums"] = DK_CV_FILECHECKSUMS; 5084 // DirectiveKindMap[".cv_filechecksumoffset"] = DK_CV_FILECHECKSUM_OFFSET; 5085 // DirectiveKindMap[".cv_fpo_data"] = DK_CV_FPO_DATA; 5086 // DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; 5087 // DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; 5088 // DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; 5089 // DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; 5090 // DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; 5091 // DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; 5092 // DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; 5093 // DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; 5094 // DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; 5095 // DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; 5096 // DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; 5097 // DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; 5098 // DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; 5099 // DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; 5100 // DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; 5101 // DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; 5102 // DirectiveKindMap[".cfi_return_column"] = DK_CFI_RETURN_COLUMN; 5103 // DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; 5104 // DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; 5105 // DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; 5106 // DirectiveKindMap[".cfi_window_save"] = DK_CFI_WINDOW_SAVE; 5107 // DirectiveKindMap[".cfi_b_key_frame"] = DK_CFI_B_KEY_FRAME; 5108 // DirectiveKindMap[".macro"] = DK_MACRO; 5109 // DirectiveKindMap[".exitm"] = DK_EXITM; 5110 // DirectiveKindMap[".endm"] = DK_ENDM; 5111 // DirectiveKindMap[".purgem"] = DK_PURGEM; 5112 DirectiveKindMap[".err"] = DK_ERR; 5113 DirectiveKindMap[".errb"] = DK_ERRB; 5114 DirectiveKindMap[".errnb"] = DK_ERRNB; 5115 DirectiveKindMap[".errdef"] = DK_ERRDEF; 5116 DirectiveKindMap[".errndef"] = DK_ERRNDEF; 5117 DirectiveKindMap[".errdif"] = DK_ERRDIF; 5118 DirectiveKindMap[".errdifi"] = DK_ERRDIFI; 5119 DirectiveKindMap[".erridn"] = DK_ERRIDN; 5120 DirectiveKindMap[".erridni"] = DK_ERRIDNI; 5121 DirectiveKindMap[".erre"] = DK_ERRE; 5122 DirectiveKindMap[".errnz"] = DK_ERRNZ; 5123 // DirectiveKindMap[".altmacro"] = DK_ALTMACRO; 5124 // DirectiveKindMap[".noaltmacro"] = DK_NOALTMACRO; 5125 DirectiveKindMap["db"] = DK_DB; 5126 DirectiveKindMap["dd"] = DK_DD; 5127 DirectiveKindMap["dq"] = DK_DQ; 5128 DirectiveKindMap["dw"] = DK_DW; 5129 DirectiveKindMap["echo"] = DK_ECHO; 5130 } 5131 5132 MCAsmMacro *MasmParser::parseMacroLikeBody(SMLoc DirectiveLoc) { 5133 AsmToken EndToken, StartToken = getTok(); 5134 5135 unsigned NestLevel = 0; 5136 while (true) { 5137 // Check whether we have reached the end of the file. 5138 if (getLexer().is(AsmToken::Eof)) { 5139 printError(DirectiveLoc, "no matching '.endr' in definition"); 5140 return nullptr; 5141 } 5142 5143 if (Lexer.is(AsmToken::Identifier) && 5144 (getTok().getIdentifier() == ".rep" || 5145 getTok().getIdentifier() == ".rept" || 5146 getTok().getIdentifier() == ".irp" || 5147 getTok().getIdentifier() == ".irpc")) { 5148 ++NestLevel; 5149 } 5150 5151 // Otherwise, check whether we have reached the .endr. 5152 if (Lexer.is(AsmToken::Identifier) && getTok().getIdentifier() == ".endr") { 5153 if (NestLevel == 0) { 5154 EndToken = getTok(); 5155 Lex(); 5156 if (Lexer.isNot(AsmToken::EndOfStatement)) { 5157 printError(getTok().getLoc(), 5158 "unexpected token in '.endr' directive"); 5159 return nullptr; 5160 } 5161 break; 5162 } 5163 --NestLevel; 5164 } 5165 5166 // Otherwise, scan till the end of the statement. 5167 eatToEndOfStatement(); 5168 } 5169 5170 const char *BodyStart = StartToken.getLoc().getPointer(); 5171 const char *BodyEnd = EndToken.getLoc().getPointer(); 5172 StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); 5173 5174 // We Are Anonymous. 5175 MacroLikeBodies.emplace_back(StringRef(), Body, MCAsmMacroParameters()); 5176 return &MacroLikeBodies.back(); 5177 } 5178 5179 void MasmParser::instantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, 5180 raw_svector_ostream &OS) { 5181 OS << ".endr\n"; 5182 5183 std::unique_ptr<MemoryBuffer> Instantiation = 5184 MemoryBuffer::getMemBufferCopy(OS.str(), "<instantiation>"); 5185 5186 // Create the macro instantiation object and add to the current macro 5187 // instantiation stack. 5188 MacroInstantiation *MI = new MacroInstantiation{ 5189 DirectiveLoc, CurBuffer, getTok().getLoc(), TheCondStack.size()}; 5190 ActiveMacros.push_back(MI); 5191 5192 // Jump to the macro instantiation and prime the lexer. 5193 CurBuffer = SrcMgr.AddNewSourceBuffer(std::move(Instantiation), SMLoc()); 5194 Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)->getBuffer()); 5195 Lex(); 5196 } 5197 5198 /// parseDirectiveRept 5199 /// ::= .rep | .rept count 5200 bool MasmParser::parseDirectiveRept(SMLoc DirectiveLoc, StringRef Dir) { 5201 const MCExpr *CountExpr; 5202 SMLoc CountLoc = getTok().getLoc(); 5203 if (parseExpression(CountExpr)) 5204 return true; 5205 5206 int64_t Count; 5207 if (!CountExpr->evaluateAsAbsolute(Count, getStreamer().getAssemblerPtr())) { 5208 return Error(CountLoc, "unexpected token in '" + Dir + "' directive"); 5209 } 5210 5211 if (check(Count < 0, CountLoc, "Count is negative") || 5212 parseToken(AsmToken::EndOfStatement, 5213 "unexpected token in '" + Dir + "' directive")) 5214 return true; 5215 5216 // Lex the rept definition. 5217 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5218 if (!M) 5219 return true; 5220 5221 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5222 // to hold the macro body with substitutions. 5223 SmallString<256> Buf; 5224 raw_svector_ostream OS(Buf); 5225 while (Count--) { 5226 // Note that the AtPseudoVariable is disabled for instantiations of .rep(t). 5227 if (expandMacro(OS, M->Body, None, None, false, getTok().getLoc())) 5228 return true; 5229 } 5230 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5231 5232 return false; 5233 } 5234 5235 /// parseDirectiveIrp 5236 /// ::= .irp symbol,values 5237 bool MasmParser::parseDirectiveIrp(SMLoc DirectiveLoc) { 5238 MCAsmMacroParameter Parameter; 5239 MCAsmMacroArguments A; 5240 if (check(parseIdentifier(Parameter.Name), 5241 "expected identifier in '.irp' directive") || 5242 parseToken(AsmToken::Comma, "expected comma in '.irp' directive") || 5243 parseMacroArguments(nullptr, A) || 5244 parseToken(AsmToken::EndOfStatement, "expected End of Statement")) 5245 return true; 5246 5247 // Lex the irp definition. 5248 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5249 if (!M) 5250 return true; 5251 5252 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5253 // to hold the macro body with substitutions. 5254 SmallString<256> Buf; 5255 raw_svector_ostream OS(Buf); 5256 5257 for (const MCAsmMacroArgument &Arg : A) { 5258 // Note that the AtPseudoVariable is enabled for instantiations of .irp. 5259 // This is undocumented, but GAS seems to support it. 5260 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5261 return true; 5262 } 5263 5264 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5265 5266 return false; 5267 } 5268 5269 /// parseDirectiveIrpc 5270 /// ::= .irpc symbol,values 5271 bool MasmParser::parseDirectiveIrpc(SMLoc DirectiveLoc) { 5272 MCAsmMacroParameter Parameter; 5273 MCAsmMacroArguments A; 5274 5275 if (check(parseIdentifier(Parameter.Name), 5276 "expected identifier in '.irpc' directive") || 5277 parseToken(AsmToken::Comma, "expected comma in '.irpc' directive") || 5278 parseMacroArguments(nullptr, A)) 5279 return true; 5280 5281 if (A.size() != 1 || A.front().size() != 1) 5282 return TokError("unexpected token in '.irpc' directive"); 5283 5284 // Eat the end of statement. 5285 if (parseToken(AsmToken::EndOfStatement, "expected end of statement")) 5286 return true; 5287 5288 // Lex the irpc definition. 5289 MCAsmMacro *M = parseMacroLikeBody(DirectiveLoc); 5290 if (!M) 5291 return true; 5292 5293 // Macro instantiation is lexical, unfortunately. We construct a new buffer 5294 // to hold the macro body with substitutions. 5295 SmallString<256> Buf; 5296 raw_svector_ostream OS(Buf); 5297 5298 StringRef Values = A.front().front().getString(); 5299 for (std::size_t I = 0, End = Values.size(); I != End; ++I) { 5300 MCAsmMacroArgument Arg; 5301 Arg.emplace_back(AsmToken::Identifier, Values.slice(I, I + 1)); 5302 5303 // Note that the AtPseudoVariable is enabled for instantiations of .irpc. 5304 // This is undocumented, but GAS seems to support it. 5305 if (expandMacro(OS, M->Body, Parameter, Arg, true, getTok().getLoc())) 5306 return true; 5307 } 5308 5309 instantiateMacroLikeBody(M, DirectiveLoc, OS); 5310 5311 return false; 5312 } 5313 5314 bool MasmParser::parseDirectiveEndr(SMLoc DirectiveLoc) { 5315 if (ActiveMacros.empty()) 5316 return TokError("unmatched '.endr' directive"); 5317 5318 // The only .repl that should get here are the ones created by 5319 // instantiateMacroLikeBody. 5320 assert(getLexer().is(AsmToken::EndOfStatement)); 5321 5322 handleMacroExit(); 5323 return false; 5324 } 5325 5326 bool MasmParser::parseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, 5327 size_t Len) { 5328 const MCExpr *Value; 5329 SMLoc ExprLoc = getLexer().getLoc(); 5330 if (parseExpression(Value)) 5331 return true; 5332 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5333 if (!MCE) 5334 return Error(ExprLoc, "unexpected expression in _emit"); 5335 uint64_t IntValue = MCE->getValue(); 5336 if (!isUInt<8>(IntValue) && !isInt<8>(IntValue)) 5337 return Error(ExprLoc, "literal value out of range for directive"); 5338 5339 Info.AsmRewrites->emplace_back(AOK_Emit, IDLoc, Len); 5340 return false; 5341 } 5342 5343 bool MasmParser::parseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { 5344 const MCExpr *Value; 5345 SMLoc ExprLoc = getLexer().getLoc(); 5346 if (parseExpression(Value)) 5347 return true; 5348 const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); 5349 if (!MCE) 5350 return Error(ExprLoc, "unexpected expression in align"); 5351 uint64_t IntValue = MCE->getValue(); 5352 if (!isPowerOf2_64(IntValue)) 5353 return Error(ExprLoc, "literal value not a power of two greater then zero"); 5354 5355 Info.AsmRewrites->emplace_back(AOK_Align, IDLoc, 5, Log2_64(IntValue)); 5356 return false; 5357 } 5358 5359 bool MasmParser::parseDirectiveEcho() { 5360 StringRef Message = parseStringToEndOfStatement(); 5361 Lex(); // eat end of statement 5362 llvm::outs() << Message << '\n'; 5363 return false; 5364 } 5365 5366 // We are comparing pointers, but the pointers are relative to a single string. 5367 // Thus, this should always be deterministic. 5368 static int rewritesSort(const AsmRewrite *AsmRewriteA, 5369 const AsmRewrite *AsmRewriteB) { 5370 if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) 5371 return -1; 5372 if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) 5373 return 1; 5374 5375 // It's possible to have a SizeDirective, Imm/ImmPrefix and an Input/Output 5376 // rewrite to the same location. Make sure the SizeDirective rewrite is 5377 // performed first, then the Imm/ImmPrefix and finally the Input/Output. This 5378 // ensures the sort algorithm is stable. 5379 if (AsmRewritePrecedence[AsmRewriteA->Kind] > 5380 AsmRewritePrecedence[AsmRewriteB->Kind]) 5381 return -1; 5382 5383 if (AsmRewritePrecedence[AsmRewriteA->Kind] < 5384 AsmRewritePrecedence[AsmRewriteB->Kind]) 5385 return 1; 5386 llvm_unreachable("Unstable rewrite sort."); 5387 } 5388 5389 bool MasmParser::parseMSInlineAsm( 5390 void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, 5391 unsigned &NumInputs, SmallVectorImpl<std::pair<void *, bool>> &OpDecls, 5392 SmallVectorImpl<std::string> &Constraints, 5393 SmallVectorImpl<std::string> &Clobbers, const MCInstrInfo *MII, 5394 const MCInstPrinter *IP, MCAsmParserSemaCallback &SI) { 5395 SmallVector<void *, 4> InputDecls; 5396 SmallVector<void *, 4> OutputDecls; 5397 SmallVector<bool, 4> InputDeclsAddressOf; 5398 SmallVector<bool, 4> OutputDeclsAddressOf; 5399 SmallVector<std::string, 4> InputConstraints; 5400 SmallVector<std::string, 4> OutputConstraints; 5401 SmallVector<unsigned, 4> ClobberRegs; 5402 5403 SmallVector<AsmRewrite, 4> AsmStrRewrites; 5404 5405 // Prime the lexer. 5406 Lex(); 5407 5408 // While we have input, parse each statement. 5409 unsigned InputIdx = 0; 5410 unsigned OutputIdx = 0; 5411 while (getLexer().isNot(AsmToken::Eof)) { 5412 // Parse curly braces marking block start/end 5413 if (parseCurlyBlockScope(AsmStrRewrites)) 5414 continue; 5415 5416 ParseStatementInfo Info(&AsmStrRewrites); 5417 bool StatementErr = parseStatement(Info, &SI); 5418 5419 if (StatementErr || Info.ParseError) { 5420 // Emit pending errors if any exist. 5421 printPendingErrors(); 5422 return true; 5423 } 5424 5425 // No pending error should exist here. 5426 assert(!hasPendingError() && "unexpected error from parseStatement"); 5427 5428 if (Info.Opcode == ~0U) 5429 continue; 5430 5431 const MCInstrDesc &Desc = MII->get(Info.Opcode); 5432 5433 // Build the list of clobbers, outputs and inputs. 5434 for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { 5435 MCParsedAsmOperand &Operand = *Info.ParsedOperands[i]; 5436 5437 // Register operand. 5438 if (Operand.isReg() && !Operand.needAddressOf() && 5439 !getTargetParser().OmitRegisterFromClobberLists(Operand.getReg())) { 5440 unsigned NumDefs = Desc.getNumDefs(); 5441 // Clobber. 5442 if (NumDefs && Operand.getMCOperandNum() < NumDefs) 5443 ClobberRegs.push_back(Operand.getReg()); 5444 continue; 5445 } 5446 5447 // Expr/Input or Output. 5448 StringRef SymName = Operand.getSymName(); 5449 if (SymName.empty()) 5450 continue; 5451 5452 void *OpDecl = Operand.getOpDecl(); 5453 if (!OpDecl) 5454 continue; 5455 5456 StringRef Constraint = Operand.getConstraint(); 5457 if (Operand.isImm()) { 5458 // Offset as immediate 5459 if (Operand.isOffsetOfLocal()) 5460 Constraint = "r"; 5461 else 5462 Constraint = "i"; 5463 } 5464 5465 bool isOutput = (i == 1) && Desc.mayStore(); 5466 SMLoc Start = SMLoc::getFromPointer(SymName.data()); 5467 if (isOutput) { 5468 ++InputIdx; 5469 OutputDecls.push_back(OpDecl); 5470 OutputDeclsAddressOf.push_back(Operand.needAddressOf()); 5471 OutputConstraints.push_back(("=" + Constraint).str()); 5472 AsmStrRewrites.emplace_back(AOK_Output, Start, SymName.size()); 5473 } else { 5474 InputDecls.push_back(OpDecl); 5475 InputDeclsAddressOf.push_back(Operand.needAddressOf()); 5476 InputConstraints.push_back(Constraint.str()); 5477 if (Desc.OpInfo[i - 1].isBranchTarget()) 5478 AsmStrRewrites.emplace_back(AOK_CallInput, Start, SymName.size()); 5479 else 5480 AsmStrRewrites.emplace_back(AOK_Input, Start, SymName.size()); 5481 } 5482 } 5483 5484 // Consider implicit defs to be clobbers. Think of cpuid and push. 5485 ArrayRef<MCPhysReg> ImpDefs(Desc.getImplicitDefs(), 5486 Desc.getNumImplicitDefs()); 5487 ClobberRegs.insert(ClobberRegs.end(), ImpDefs.begin(), ImpDefs.end()); 5488 } 5489 5490 // Set the number of Outputs and Inputs. 5491 NumOutputs = OutputDecls.size(); 5492 NumInputs = InputDecls.size(); 5493 5494 // Set the unique clobbers. 5495 array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); 5496 ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), 5497 ClobberRegs.end()); 5498 Clobbers.assign(ClobberRegs.size(), std::string()); 5499 for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { 5500 raw_string_ostream OS(Clobbers[I]); 5501 IP->printRegName(OS, ClobberRegs[I]); 5502 } 5503 5504 // Merge the various outputs and inputs. Output are expected first. 5505 if (NumOutputs || NumInputs) { 5506 unsigned NumExprs = NumOutputs + NumInputs; 5507 OpDecls.resize(NumExprs); 5508 Constraints.resize(NumExprs); 5509 for (unsigned i = 0; i < NumOutputs; ++i) { 5510 OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); 5511 Constraints[i] = OutputConstraints[i]; 5512 } 5513 for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { 5514 OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); 5515 Constraints[j] = InputConstraints[i]; 5516 } 5517 } 5518 5519 // Build the IR assembly string. 5520 std::string AsmStringIR; 5521 raw_string_ostream OS(AsmStringIR); 5522 StringRef ASMString = 5523 SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer(); 5524 const char *AsmStart = ASMString.begin(); 5525 const char *AsmEnd = ASMString.end(); 5526 array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), rewritesSort); 5527 for (auto it = AsmStrRewrites.begin(); it != AsmStrRewrites.end(); ++it) { 5528 const AsmRewrite &AR = *it; 5529 // Check if this has already been covered by another rewrite... 5530 if (AR.Done) 5531 continue; 5532 AsmRewriteKind Kind = AR.Kind; 5533 5534 const char *Loc = AR.Loc.getPointer(); 5535 assert(Loc >= AsmStart && "Expected Loc to be at or after Start!"); 5536 5537 // Emit everything up to the immediate/expression. 5538 if (unsigned Len = Loc - AsmStart) 5539 OS << StringRef(AsmStart, Len); 5540 5541 // Skip the original expression. 5542 if (Kind == AOK_Skip) { 5543 AsmStart = Loc + AR.Len; 5544 continue; 5545 } 5546 5547 unsigned AdditionalSkip = 0; 5548 // Rewrite expressions in $N notation. 5549 switch (Kind) { 5550 default: 5551 break; 5552 case AOK_IntelExpr: 5553 assert(AR.IntelExp.isValid() && "cannot write invalid intel expression"); 5554 if (AR.IntelExp.NeedBracs) 5555 OS << "["; 5556 if (AR.IntelExp.hasBaseReg()) 5557 OS << AR.IntelExp.BaseReg; 5558 if (AR.IntelExp.hasIndexReg()) 5559 OS << (AR.IntelExp.hasBaseReg() ? " + " : "") 5560 << AR.IntelExp.IndexReg; 5561 if (AR.IntelExp.Scale > 1) 5562 OS << " * $$" << AR.IntelExp.Scale; 5563 if (AR.IntelExp.hasOffset()) { 5564 if (AR.IntelExp.hasRegs()) 5565 OS << " + "; 5566 // Fuse this rewrite with a rewrite of the offset name, if present. 5567 StringRef OffsetName = AR.IntelExp.OffsetName; 5568 SMLoc OffsetLoc = SMLoc::getFromPointer(AR.IntelExp.OffsetName.data()); 5569 size_t OffsetLen = OffsetName.size(); 5570 auto rewrite_it = std::find_if( 5571 it, AsmStrRewrites.end(), [&](const AsmRewrite &FusingAR) { 5572 return FusingAR.Loc == OffsetLoc && FusingAR.Len == OffsetLen && 5573 (FusingAR.Kind == AOK_Input || 5574 FusingAR.Kind == AOK_CallInput); 5575 }); 5576 if (rewrite_it == AsmStrRewrites.end()) { 5577 OS << "offset " << OffsetName; 5578 } else if (rewrite_it->Kind == AOK_CallInput) { 5579 OS << "${" << InputIdx++ << ":P}"; 5580 rewrite_it->Done = true; 5581 } else { 5582 OS << '$' << InputIdx++; 5583 rewrite_it->Done = true; 5584 } 5585 } 5586 if (AR.IntelExp.Imm || AR.IntelExp.emitImm()) 5587 OS << (AR.IntelExp.emitImm() ? "$$" : " + $$") << AR.IntelExp.Imm; 5588 if (AR.IntelExp.NeedBracs) 5589 OS << "]"; 5590 break; 5591 case AOK_Label: 5592 OS << Ctx.getAsmInfo()->getPrivateLabelPrefix() << AR.Label; 5593 break; 5594 case AOK_Input: 5595 OS << '$' << InputIdx++; 5596 break; 5597 case AOK_CallInput: 5598 OS << "${" << InputIdx++ << ":P}"; 5599 break; 5600 case AOK_Output: 5601 OS << '$' << OutputIdx++; 5602 break; 5603 case AOK_SizeDirective: 5604 switch (AR.Val) { 5605 default: break; 5606 case 8: OS << "byte ptr "; break; 5607 case 16: OS << "word ptr "; break; 5608 case 32: OS << "dword ptr "; break; 5609 case 64: OS << "qword ptr "; break; 5610 case 80: OS << "xword ptr "; break; 5611 case 128: OS << "xmmword ptr "; break; 5612 case 256: OS << "ymmword ptr "; break; 5613 } 5614 break; 5615 case AOK_Emit: 5616 OS << ".byte"; 5617 break; 5618 case AOK_Align: { 5619 // MS alignment directives are measured in bytes. If the native assembler 5620 // measures alignment in bytes, we can pass it straight through. 5621 OS << ".align"; 5622 if (getContext().getAsmInfo()->getAlignmentIsInBytes()) 5623 break; 5624 5625 // Alignment is in log2 form, so print that instead and skip the original 5626 // immediate. 5627 unsigned Val = AR.Val; 5628 OS << ' ' << Val; 5629 assert(Val < 10 && "Expected alignment less then 2^10."); 5630 AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; 5631 break; 5632 } 5633 case AOK_EVEN: 5634 OS << ".even"; 5635 break; 5636 case AOK_EndOfStatement: 5637 OS << "\n\t"; 5638 break; 5639 } 5640 5641 // Skip the original expression. 5642 AsmStart = Loc + AR.Len + AdditionalSkip; 5643 } 5644 5645 // Emit the remainder of the asm string. 5646 if (AsmStart != AsmEnd) 5647 OS << StringRef(AsmStart, AsmEnd - AsmStart); 5648 5649 AsmString = OS.str(); 5650 return false; 5651 } 5652 5653 /// Create an MCAsmParser instance. 5654 MCAsmParser *llvm::createMCMasmParser(SourceMgr &SM, MCContext &C, 5655 MCStreamer &Out, const MCAsmInfo &MAI, 5656 unsigned CB) { 5657 return new MasmParser(SM, C, Out, MAI, CB); 5658 } 5659